Where There is Fire There is Haze: The Economic...

Where There is Fire There is HazeThe Economic and Political Causes of

Indonesiarsquos Forest Fireslowast

James Macdonalddagger and Russell TothDagger

The University of Sydney

May 2018

Abstract

In Indonesia annual forest fires between July and November raise substantial healthenvironmental and economic costs for Indonesia and its neighbours In spite of substantialmedia and policy interest in the fires there is little systematic quantitative analysis ofthe economic and political forces behind them We utilize data on fires rainfall forestsand agricultural conditions and prices to analyze the role of two prospective channelsweak governance leading to underenforcement of laws against burning and global demandfor palm oil First we leverage plausibly exogenous variation in governance quality(through splits of administrative districts) that may alter local government capacity andincentives While important prior work (Burgess et al 2012) has shown that this can bea mechanism increasing overall deforestation primarily through illegal logging forestburning is arguably a significantly more publicly visible form of illegal activity so mayinteract differently with governance dynamics We show that an extra district per provinceleads to a 3-117 increase in the number of fires However this effect is concentratedimmediately after the time of a district split and in newly-created districts suggestingthe primary mechanisms are temporary factors such as weakened governance capacity innewly-created districts Furthermore we present evidence that fires account for slightlyunder half of all deforestation in Indonesia Second we interact global palm oil prices witha measure of a districtrsquos suitability for conversion to palm oil showing that global palmoil demand substantially increases fire activity particularly in the areas most suitable forconversion to palm oil These results provide insights toward efforts to predict and preventthe fires and deforestation more broadly

lowastFirst version Nov 4 2017 This version May 16 2018 We thank Teevrat Garg Tom Pepinsky Daniel Suryadarma DavidUbilava and Jordi Vidal-Robert for useful feedback Johannes Pirker and Evan Kresch for generously sharing data andKyle Navis and Sanji Pallegedara Dewage for assistance with ArcGIS All errors are our owndaggerjmac9695unisydneyeduauDaggerCorresponding author russelltothsydneyeduau Room 488 Merewether H04 School of Economics The University ofSydney 2006 Australia

1 Introduction

Each year between July and November equatorial Asia experiences a dense smokey haze as Indonesiarsquosforests are illegally burnt to clear land for agriculture The fires cause numerous complications forIndonesia and neighboring Malaysia and Singapore and have become a contentious internationalrelations issue in the region The haze causes significant respiratory issues in affected populations(Frankenberg Mckee and Thomas 2005) Studies have estimated a 12 percent increase in infantmortality in regions exposed to the haze in 1997 and over 100000 premature deaths in IndonesiaMalaysia and Singapore during 2015 a year of particularly intense fire activity (Jaychandran 2009Koplitz et al 2016) The fires also have substantial environmental consequences especially when theburning occurs on peatland a dense forest that emits substantial amounts of carbon when burnt (Pageet al 2002 Alisjahbana and Busch 2017) During 2015 Indonesia emitted more carbon than the entireEuropean Union and ranked third for total carbon pollution globally (Huijen et al 2016 Edwards andHeiduk 2015) Forest fires contribute an overwhelming amount of this pollution accounting for 69percent of Indonesiarsquos carbon emissions over the period 1989-2008 (Carson et al 2012) The economicconsequences are also severe with estimates suggesting that the fires in 2015 cost the Indonesianeconomy $161 billion USD equivalent to 19 percent of GDP a figure which fails to consider additionaladverse impacts on Singapore and Malaysia (World Bank 2016) This paper provides systematicquantitative evidence on some of the political and economic forces behind these annual fire events

The forest fires are considered to primarily be the result of slash and burn land conversionpractices generally to convert land to palm oil (Dauvergne 1998 Dennis et al 2005 Purnomo etal 2017) Palm oil is Indonesiarsquos biggest agricultural export and has experienced rapid growth sincethe turn of the century with a deliberate government goal to increase production of palm oil as apoverty reduction strategy (Gatto et al 2017 Edwards 2017 Edwards and Heiduk 2015) Palmoil concessions covered under 4 million hectares before 2000 and over 12 million hectares by 2014(Edwards 2017) Most of the expansion of palm oil has come from converting forest-covered land topalm plantations (Furumo and Aide 2017)

This paper provides causal estimates motivated by two of the main hypothesized channels behindthe fires (1) the role of weak governance and the (lack of) incentives for local government officials toenforce laws outlawing the fires and (2) the returns from land conversion to palm oil driven by globalpalm oil demand To do so we combine a number of remotely-sensed and administrative data sourcescovering the early 2000s to the mid-2010s on fires rainfall forest cover administrative boundariesagricultural conditions (the suitability of local regions for palm oil conversion) and agriculturalprices of the two fastest-growing crops in terms of land area palm oil and rubber In supplementalexploratory analysis to assess the role of the fires in overall deforestation in Indonesia we also utilizedata on deforestation Because of the relatively unavailability and distinct unreliability of surveyor administrative data on fires and deforestation we instead rely on remotely-sensed data on theseoutcomes which has come available in recent years

Despite successive central government attempts to reduce the prevalence of fires including President

1

Yudhoyonorsquos public war on haze in 2006 the fires remain a major issue in Indonesia (Edwardsand Heiduk 2015) Existing studies have found links between weak governance and deforestationin Indonesia (eg Burgess et al 2012 and Smith et al 2003) but little economic research hassystematically quantified the role of governance quality specifically in forest fires To provide evidenceon this channel we leverage the approach of Burgess et al (2012) who hypothesize that changingadministrative boundaries beginning in the early 2000s provide plausible exogenous variation in thequality of governance Specifically after the fall of President Suharto in 1998 Indonesia began aprocess of decentralization that delegated significant powers to political districts and allowed districtsto seek approval to split into separate administrative units As a result with brief interruption in themid-2000s the number of districts increased from 289 to 408 between 2002 and 2013

Motivated by a model of Cournot competition between district officials to accept bribes to overlookillegal deforestation Burgess et al (2012) focus on increases in the number of districts per province4

over time as a source of increasing competition in a provincersquos market for bribes which should lowerthe bribe price and increase illegal deforestation Burgess et al (2012) also consider the possibilitythat newly-created districts will suffer from low capacity including in their ability to enforce lawsagainst illegal logging during the transition period as newly-created district builds up administrativecapacity In spite of the best intentions the new government may simply lack the capacity to carryout law enforcement particularly in far-flung rural areas In the case of Burgess et al (2012) thebalance of the evidence is in favour of the corruption channel We implement the same empiricalstrategy leveraging district splits also considering both the corruption and weakened capacity channelsHowever it is plausible to see different response dynamics when focusing on fires as they are plausiblya much more publicly visible form of illegal activity than logging with different public accountabilitydynamics Smoke plumes are visible over long distances and flames may also be visible particularlyat night Meanwhile illegal logging occurs in the forest local citizens may not know which forests aredesignated for (legal) logging or not and while the local government is likely able to closely monitorlog shipments through roadblocks along the major intradistrict and interdistrict roads the averagecitizen may not have much awareness of these shipments

Our main finding from this part of the analysis is that the creation of a new district in a givenprovince leads to an increase in the number of fires of 3 to 117 percent This then raises the question ofwhether the corruption channel or the weakened capacity channel is driving these effects We find thatthe largest effect of district splits is concentrated in the year directly following a split and furthermorethat when we estimate effects separately for parent and child districts after a split we find no effectsin the former and strong effects in the latter Both of these findings are the opposite of what Burgess etal (2012) estimate with deforestation as an outcome and most consistent with the weakened capacitychannel We also find differences in spending patterns between existing and newly-created districts

A natural follow-up question is to what extent the fires account for overall deforestation as firesare just one among a number of channels for deforestation including legal and illegal logging Theimportance of each of these channels has important implications for policy efforts to combat defor-

4Alesina et al (2016) leverage the same administrative changes to provide variation in the ethnic composition of districts

2

estation in one of the worldrsquos three largest stands of tropical forest as different forms of deforestationrequire different approaches to enforcement In a simple correlation exercise with province and yearfixed effects we show that variation in fires accounts for just under 50 of deforestation in Indonesiasuggesting that the fires are an important channel for deforestation

In the second part of the analysis we consider the demand side of fire activity focusing on the role ofglobal demand for palm oil as proxied through global palm oil prices To do so we leverage time seriesvariation in global palm oil prices interacted with cross-sectional variation in a measure of a districtrsquosbio-physical suitability for conversion to palm oil which is constructed based on slow-changing orunchanging environmental factors such as soil climate and topography While Indonesia is the worldrsquoslargest producer of palm oil we argue that concerns about reverse causality from palm oil prices to firesare minimized under our empirical strategy as it takes 3 years for a palm tree to mature and producefruit and harvested palm that might have been affected by smoke will take months to transit from alocal plantation to global markets where prices are set Using a rich distributed lag specification weshow that higher global palm oil prices lead to substantially greater fire activity in districts most suitableto palm oil however with a complex pattern of lag effects To interpret these effects we run simplesimulations on the net effect of changing palm oil prices showing that a 10 percent increase in palmoil prices in the lead-up to the traditional burning season in a district with average suitability for palmconversion would generate a 129 percent increase in monthly district fire activity This confirms theimportant the role of global palm oil demand in driving illegal behaviour that is complex and expensiveto regulate

While there is a large body of research documenting the health environmental and economicimpacts of forest fires and the haze to date there has been a significant gap in broad-scale researchquantifying the causes of the fires This study to the best of our knowledge contributes for the first timesystematic quantitative evidence on the role of weak governance and global palm oil price dynamics incausing forest fires across Indonesia The prior literature has largely been drawn from smaller samplesFor example Purnomo et al (2017) draw on a survey of 131 respondents in four districts in Riauprovince to show that local elites protected by district governments through patronage networks areactive in the organisation of the fires to convert land and receive most of the gains They estimate that68 percent of the economic gain from local fire activity goes to local elites while only 22 percent goesto the individuals burning the land RCA and UNICEF (2016) argue that locals show little concern forthe legal penalties for starting fires and that there is significant confusion about who starts large fires

A larger literature addresses the causes of deforestation in Indonesia and tends to focus on corruptbehaviour by government officials responsible for enforcing laws against illegal logging BeyondBurgess et al (2012) and Alesina et al (2016) which both conclude that the primary channel oftransmission of district splits to deforestation is corruption by local government officials rather thanweakened administrative capacity Other studies linking Indonesian government officials and illegalforest activities particularly illegal logging include Smith et al (2003) Palmer (2001) and Scotlandet al (2000) as well as various NGO reports from numerous institutions (Sundstroumlm 2016) We arehowever not aware of prior literature providing systematic quantitative evidence on the link (or lack

3

thereof) between corruption and the fires However forest burning is arguably a significantly morepublicly visible form of illegal activity so may interact differently with governance dynamics thanillegal logging An important stream of the literature on corruption focuses on the role of public visibilityand transparency in preventing corrupt activity (Olken and Pande 2012) For example in the context ofIndonesia Olken (2007) shows that enhanced public accountability through increasing participation inpublic meetings for large local infrastructure projects has little effect on reducing corruption howeverBanerjee et al (forthcoming) show that a more decentralized information treatment leads to significantreductions in leakage from a large government program Fires due to their widespread visibility arearguably a highly publicly visible form of illegal activity which may limit the ability of governmentofficials to intentionally overlook them in return for bribes Hence our contrasting results adds to theliterature on governance and corruption particularly in regards to natural resource degradation It alsoadds to the growing literature on the role of decentralisation in environmental outcomes

Finally we provide new evidence on a previously understudied issue in the broader literature onpalm oil and economic development the role of demand-side factors in causing the fires The firesare perhaps the key method for land conversion to palm oil as in spite of its illegality (raising risksof punishment) the method carries much lower cost than other methods such as mechanized landclearing This contributes to the broader literature reviewed by Ferretti-Gallon and Busch (2014) whothrough an extensive meta-analysis of 117 studies considering the contributing factors to deforestationfind that rates of deforestation are higher when the economic returns to converting the land are higherSurprisingly we are not aware of systematic quantitative evidence on the role of palm oil pricesspecifically in motivating forest conversion Of the evidence that does exist Wheeler et al (2013)consider the role of palm oil prices in logging to deforest land finding that forest clearing increases withthe expected return to clearing relative to alternatives but they do not consider burning as a specificchannel of deforestation nor do they comment on the role of palm oil expansion and fires This studyalso contributes to the recent literature showing that palm oil can be an important channel of economicdevelopment and poverty reduction (Edwards 2017 Gatto et al 2017 Cahyadia and Waibel 2013)though the largest gains tend to be skewed towards elites (Cahyadia and Waibel 2013 World Bank2016) This literature notes that a substantial benefit of palm oil over other crops is that it requiresminimal ongoing maintenance effort after the 3-year initial period before maturity of the palm oil treesallowing rural populations the potential to simultaneously earn other revenues We contribute to thisliterature by providing insights on how the dynamics of global palm oil prices motivate land conversionthrough forest burning as potential producers consider various economic opportunities

This paper proceeds as follows First we provide a simple framework for understanding whatmay cause fires in Section 2 We then outline the various data sources in Section 3 the empiricalmethodology in Section 4 and the empirical results in Section 5 Section 6 provides concludingremarks

4

2 Why Do People Burn the Forest

This section provides background for understanding what motivates people to light fires and outlinesthe testable predictions that we test the later sections of the paper We hypothesize that the agentsstarting the fires weigh the costs of illegally lighting fires against the benefits of the conversion of landvia fire primarily to palm oil This assumption is supported by surveys of the causes of the fires thatgenerally suggest that citizens have little concern for the penalties from fires but are well aware ofthe gains (see Purnomo et al 2017 RCA and UNICEF 2017) In an environment where an agent isconsidering starting a fire to convert land they likely consider (1) the possibility of punishment and theassociated fine or judicial punishment which may in some cases be non-existent if government officialsare willing to overlook illegal activity or unable to enforce laws and regulations against burning and(2) the net benefit from land conversion in this case the yield and output price of the crop the landis converted to weighed against production and other costs Other factors influencing the decisionto convert land via fire include climatic factors such as rainfall which is known to have a substantialimpact on the variation in fires across years (Page et al 2002 Huijnen et al 2016) The decisionto light fires is therefore likely a function of the return to cleared land the probability of judicialpunishment or associated costs to avoid judicial punishment and climatic conditions It is likely thatthe decision to light a fire to clear land is less likely for a higher risk of punishment and more likely thehigher the return to cleared land This paper seeks to explicitly test these two implications

21 Administrative changes in Post-Suharto Indonesia

Agentsrsquo beliefs about probability of punishment are not directly observable but changing administrativeboundaries provide a credible source of variation in governance quality at a local level allowing us totest the role of governance in allowing fires Following the fall of President Suharto in 1998 Indonesiaundertook a significant political restructuring that included the delegation of significant fiscal andadministrative powers to political districts (kabupaten) one administrative level below provinces (Paland Wahhaj 2017)5 These powers included control of the enforcement of forest policy and brought anincreased share of the national fiscal budget (Burgess et al 2012)

In addition to the decentralization of powers and fiscal resources to the existing set of districts thegovernment also allowing districts to apply to split into separate administrative units In the wake ofthis policy a significant increase in the number of districts occurred starting in 2002 generating newopportunities for corruption and rent seeking behavior by local elites due the proliferation of politicalpower (Alesina et al 2016) There was an increase in the number of districts from 289 to 347 between2002 and 2005 when there was a temporary suspension of the district splitting process and then andincrease to 408 between 2007 and 2013 These district splits varied across time and locations Figure1 plots the number of district splits by island over the sample period Figure 2 displays the numberof district splits by province and Table A2 records the number of districts by province in 2002 and

5District splits also lead to the creation of cities (kota) but for this analysis we focus on areas with arable land and so do notconsider new kota as district splits

5

2015 The process of district splitting creates valid quasi-experimental variation exploited to considerthe role of governance in outcomes under the identifying assumption that the splits are not directlymotivated by factors correlated with the outcome variable in this case fires In Section 4 we defend thisassumption

Both Burgess et al (2012) and Alesina et al (2016) show that the rate of deforestation inIndonesia changes with the establishment of new districts and both are examples of a growingliterature considering the impact of decentralisation on a range of outcomes based on quasi-experimentalapproaches6 Here we leverage a similar identification strategy to Burgess et al (2012) extending uponthe methodology and considering the more publicly observable fire activity rather than deforestationas our outcome of interest

Figure 1 District Splits Per Island (2002-2015)

Hypothesis 1 District splits lead to increase in fire activity at district and province level

There are a number of scenarios under which provinces seeing one or more district splits couldexperience an increase in fire activity We focus on two in particular that have been emphasized in

6Eg Jia and Nie (2017) assess decentralisation of mining regulation in China and Martinez-Bravo Mukherjee and Stegmann(2017) study the differential influence of legacy mayors chosen by Suhartorsquos central government and newly elected mayorson Indonesiarsquos transition to democracy

6

Figure 2 District Splits Per Province (2002-2015)

prior literature one in which governments want to enforce legislation and reduce fires in their districtbut lack the resources or capacity to do so and another in which district splits lead to an increase incorruption among officials charged with enforcing policies against forest burning for example becausean increase in the number of districts increases competition for bribes (Burgess et al 2012)

We explore the potential implications of these channels briefly in the following

Hypothesis 2 If district splits generate increased fire activity primarily through weakened gov-ernance capacity then effects will be short-lived and concentrated in newly-created districts

District governments as part of the decentralisation process gained power to enforce numerous piecesof legislation and despite forest fires being made illegal in national legislation district governmentspolice them (Edwards and Heiduk 2015) When a new district is formed if the new district governmentlacks the initial capacity to police fires due to the delay developing appropriate enforcement capacityitrsquos likely that the rate of fires will increase in the short-run If enforcement is primarily due to lack ofcapacity it is likely that most of the increase in fires occurs in the years immediately following a districtsplit because over time as the government strengthens and is stricter on fire activity the probability ofgetting caught is also likely to rise In line with this channel we would also expect that district splitswould be much more likely to lead to an increase in fires in newly-created districts (child districts)rather than the part of the district that keeps the district capital (parent districts)

7

Hypothesis 3 If district splits generate increased fire activity primarily through increased cor-ruption then effects will be persistent and largely similar in continuing and newly-created dis-tricts

If district governments are willing to allow fire activity in return for bribes or some other transfer adistrict split may also increase fire activity7 If there is a regional or national market for land conversionan additional player in the market is likely to induce greater competition This effect is modelled byBurgess et al (2012) in a Cournot competition framework in which an additional market player (anewly created district in a given province) generates more competition in the market for bribes leadingto a reduction in the bribe price and hence greater deforestation Itrsquos plausible that the market for landconversion via fire could operate in a similar manner If this is the case we may see a permanent increasein the number of fires in provinces with more district splits relative to other provinces and increasesin fires even in districts that do not split One force that might oppose this channel is that fires areplausibly much more publicly visible method of deforestation in comparison to illegal logging Henceit is possible that these effects would be muted for fire burning due to greater public accountabilitybetter preventing such illegal activity from occurring

22 Indonesiarsquos palm oil industry

Indonesiarsquos relationship with palm oil has strengthened significantly over the past 20 years Growthin land under palm has expanded rapidly in this period growing 230 percent between 2000 and 2014Figure 3 plots data from the FAO database on agricultural outputs of key Indonesian exports overthe period 1990 to 2014 Palm oil growth substantially exceeds all other crop types This followsdeliberate government strategies to expand palm oil as a means of poverty reduction (Dennis et al2005 Purnomo et al 2017) As a result of this expansion palm oil has been Indonesiarsquos largestagricultural export over the last two decades and Indonesia now supplies nearly half of the $40 BillionUSD industry (Edwards 2017 FAO 2017)

Hypothesis 4 The number of fires will increase with the global palm oil price

The decision to convert land is ultimately driven by the predicted returns to that land once convertedExpectations about the return to the land will be formed from the expected yield of the land and theexpected output price of the crop functions of the suitability of the land and belief formation aboutthe price time series based on past prices Because palm oil is the main crop driving conversion andbecause the government has targeted increases in the amount of land under palm it is the main cropthat we consider

We briefly outline the palm oil production process in the following

7This is a hypothesis suggested in numerous press articles including one report of extensive links between fire activitydistrict politicians and large commercial palm oil plantations (Mongabay 2017)

8

Figure 3 Land under certain crops (1990-2014)

221 The Characteristics of Palm Oil

Palm oil grows best in tropical conditions with moderate rainfall levels and warm temperatures ona range of soils (Pirker et al 2016) The suitability of many regions of Indonesia for palm oil hasled to large losses of diverse tropical forest for palm oil plantations (Pacheco et al 2017)8 Onceestablished palm plantations are economically viable for 25-30 years and they return output throughoutthe year (Pacheco et al 2017) The decision to convert land to palm oil in Indonesia is driven by bothsmall-holders and large corporations9 The conversion decisions by large corporations may differ tothose of small-holders if they have different knowledge and understanding of the market for palm oilLarge companies likely expand land in an organised fashion planning investments across their businessinto new land for palm oil in advance while small-holders may be more likely to respond to currentcircumstances We are able to investigate the relationships between price dynamics and forest burningin our estimation through different lag specifications

8For a visualisation of regions suitable for palm oil see Figure 59Data from the Indonesia Ministry of Agriculture indicates roughly half of palm oil is privately-owned by large corporations40 percent is small-holder owned and 10 percent is state-owned

9

3 Data

This section outlines the main data set used in this analysis Most importantly it provides detailson the remote sensing data on fire signatures and how we use those to define fires the district-levelmeasure of suitability for palm oil production and data on the construction of the global real palmoil price variable We utilize remote sensing data to detect fires in light of the unreliability and lackof coverage of alternative data sources such as surveys and administrative data sources Finally itprovides comments on the remaining data sources summarised in Table A1

31 Active Fire Data

Fires our primary outcome variable of interest are recorded using remote sensing data from the Mod-erate Resolution Spectroradiometer (MODIS) instruments from two NASA satellites Terra (launchedFebruary 2000) and Aqua (launched May 2002) (Oom and Periera 2013) These satellites give dailyrecords of active fires in 1 km2 pixels detected via an algorithm measuring heat and brightness (Giglio2015) MODIS Collection 6 is an updated fire detection algorithm introduced in 2015 and backdatedto 2000 improving the detection of large fires impeded by haze and reducing incidence of errorsdetecting small fires a critical improvement for the case of Indonesia (Giglio Schroeder and Justice2016) While detection of fires is not perfect a global commission rate of 12 percent suggests thatrelatively few fire points are false detections and the Collection 6 algorithm outperforms previousMODIS algorithms in this regard (Giglio Schroeder and Justice 2016)10 Satellite data is likely to bemore reliable than government reported estimates or crowd sourced data and as such even with a slightbias in detection it is likely to be a more a more reflective indicator of true fire activity than any othercurrently-available measure Our primary data set contains geographic information on the locationand timing of fire detections at the resolution of 1 km2 pixels the brightness and fire radiative power(FRP) of the detected fire and a confidence measure on a scale of 0-100 A visualisation of the firesdata by districts is given in Figure 4 It suggests that there is quite substantial variation in fire activityacross provinces and districts over the sample period with fire activity most prevalent in the regionsof central-north and southern Sumatra most of Borneo (Indonesian Kalimantan) and southwest andsouthern Papua Unsurprisingly we see very little fire activity in Java which is heavily populated andhad been largely deforested prior to the sample period11

Sheldon and Sankaran (2017) use the active fire data to consider the impact of Indonesian forestfires on health outcomes in Singapore Their analysis focuses on the fire radiative power (FRP) ofdetected forest fires rather than a count variable which is used in this analysis FRP detects the heat ofthe fire and is therefore relevant to an application considering the amount of smoke produced by a fireIn the current case the detection of a fire is relevant regardless of the radiative power and so a simple

10The global commission rate measures false fire detections by comparing fire detections with high resolution reference mapsIf no fire activity or burn scar is detected on the high-resolution reference maps a fire is declared a false alarm (GiglioSchoreder and Justice 2016)

11For a reference map see Figure A1

10

count variable of detected fires is preferred Another major application of the data is the PulseLabJakarta Haze Gazer Pulselab developed a system to evaluate and understand Indonesian forest fires andtheir role in Indonesian Haze12 We use PulseLabrsquos preferred confidence level of fire likelihood for ourmain specification although results can be shown to be robust to changes in confidence levels13

Figure 4 Fires by District (2000-2015)

32 Palm Oil Suitability Index

To capture district-level palm oil suitability we utilize the FAO Global Agro-Ecological Zone (GAEZ)14

data set This dataset estimates the suitability of a district for palm oil conversion based on factors thatare fixed prior to our study period hence removing concerns about reverse causality from fire-relatedoutcomes to factors determining the index From here onwards we refer to this measure as the palm oilsuitability index

33 Price Data

The main global market for palm oil is in Malaysia Given that Indonesia contributes nearly 50 percentof the global market aggregate production levels in Indonesia are also likely to dictate global pricemovements particularly in the Malaysian market of which it largely supplies (FAO 2017) To considerprice dynamics we therefore use the Palm Oil Futures Prices from New York denominated in USdollars rather Malaysian market prices that are likely to be influenced by factors influencing Indonesian

12Knowledge of the PulseLab Haze Gazer comes from meetings with PulseLab in their Jakarta office in July 2017 and fromPulseLab annual reports from 2015 and 2016

13The confidence levels measures the confidence in fire detection based on an algorithm using the brightness and FRP ofthe fire detection PulseLab indicated during meetings in July 2017 that the 80 percent confidence level appears to captureknown fires without capturing false fires with relatively high accuracy so that is our preferred specification

14GAEZ data set httpwwwfaoorgnrgaezen

11

Figure 5 Palm Oil Suitability

supply and potentially fires Specifically we use the Monthly Crude Palm Oil Futures End of DaySettlement Prices on the Chicago Mercantile Exchange sourced from the World Bank To remove theimpact of inflation we deflate the prices using the US GDP Implicit Price Deflator from the Bureau ofEconomic Analysis as per Wheeler et al (2013) to construct real palm oil prices We also considerthe Malaysian market price converted to Indonesian Rupiah and Rubber Prices from the SingaporeCommodity Exchange More details on all the price variables are included in Table A1 in the AppendixA plot of the real palm oil price index between 2001 and 2015 is included in Appendix Figure A2 and aplot of the co-movement between the NY market prices and the Malaysian market prices for palm oilare shown in Figure A3 It suggests that the prices in New York generally reflect the Malaysian marketprices converted to Rupiah

34 Additional Data

The active fire data and palm oil suitability is merged with Indonesian district and province boundariesusing 2013 borders for the entirety of the analysis15 This allows an understanding of fire activityin 2015 administrative units because there was no changes to districts or provinces after 2013 Tocontrol for rainfall we construct a province level rainfall measure from the NOAA NCEP CPC PRECLmonthly rainfall data set constructed by Chen et al (2002) averaging rainfall observations acrossprovinces to form annual average rainfall per province for the period 2002-2015 Administrative datais sourced from Statistics Indonesia (Badan Pusat Statistik) and a range of district level controls aresourced from the Indonesia Database for Policy and Economic Research (DAPOER) made available bythe World Bank All geographic data was merged to province and district boundaries using ArcGIS andQGIS and economic data was then matched using province or district codes A table summarizing

15We use official 2013 level 1 and level 2 administrative boundaries made available by the United Nations World FoodProgram httpsgeonodewfporglayersgeonode3Aidn_bnd_a2_2013_bps_a

12

each data used is contained in Table A1 in the Appendix

4 Methodology

This section details the empirical approach used to test each of the four hypotheses outlined in Section2 First we present the results on the effect of district splits on fire activity We then provide furtherevidence on the mechanism behind the district split result Second we present our analysis of the effectof global palm oil prices on fires

41 Weak Governance and Fires the Effect of District Splits

To understand the role of the strength of governance in preventing fires we take advantage of waves ofchanging administrative boundaries that occurred following the political decentralisation after the endof Indonesian President Suhartorsquos reign in 1998 This follows the identification strategy of Burgesset al (2012) and Alesina et al (2016) who leverage district splits as a source of quasi-experimentalvariation in governance to study impacts on deforestation As discussed in detail in Section 2 wehypothesize similar potential mechanisms for the relationship between strength of governance and fireswhile noting that deforestation by burning may be more publicly visible than illegal logging whichmay lead to differential impacts from variation in governance

Hence our empirical approach is similar to Burgess et al (2012) regressing the number of fires(rather than a measure of deforestation) per province per year on the number of districts per provinceHowever the specific regression model differs due to differences in the structure of the outcome variableBurgess et al (2012) measure deforestation through a relatively sparse count variable with frequentzeros (8 percent of observations) the number of pixels deforested per district in a given year which isbest modeled through a quasi-Poisson maximum likelihood count model16 The Poisson count modelisnrsquot suitable for the fires data because the variance of the fires variable substantially exceeds the mean(mean = 1000 variance = 400000) and therefore the data is over-dispersed (Cameron and Trivedi2013) In addition zero values are less frequent (less than 1 percent of cases and there are no cases inthe main estimation sample) Hence we present our main estimates from a negative binomial countmodel that accounts for the over-dispersion better than a Poisson model Our base negative binomialspecified is specified in Equation (1)

Firespit = β0 + β1NumberDispit + Pp + Y eart + εpit (1)

where Firespit is a count of fires by province NumberDispit is the number of districts per provinceper year Pp is a province fixed effect and Yeart is an island by year fixed effect17 We estimate twoextensions to this model First we add the log of rainfall by province-year Second we replace year

16Burgess et al (2012) show that their results are quantitatively similar with a fixed effects specification just less precise17This is the closest possible equivalent to the model run by Burgess et al (2012) but we include rainfall because it has

important implications for fires across years

13

fixed effects with island-by-year fixed effects according to the major islands and a linear trend in forestcover constructed by interacting the baseline value of forest cover by district with a year dummy Inall models we specify province fixed effects according to the province codes in effect at our base yearSimilarly to Burgess et al (2012) we also cluster standard errors at the level of these province codes

In addition we model the number of fires as an approximation to a continuous variable with thelog of fires as our primary dependent variable with estimation following a more traditional panel dataapproach

We estimate Equation (2)

lnFirespit = β0 + β1NumberDispit + γprimeXpi + δprimeZpit + Yt + Ii + εpit (2)

where Firespit is the number of fires in province p in island i in year t and NumberDispit measuresthe number of districts in province p in year t Xpi records time-invariant province characteristics theprovincersquos suitability for palm oil constructed from the GAEZ suitability measure and the log of forestcover in the year 200018 Zpit captures the average province annual rainfall Yt is a year fixed effectand Ii is an island fixed effect εpt is a robust random error term clustered at province level

Equation (2) is estimated using a generalised least squares random effects model based on the resultsof a Hausman test (1978) which indicates that random effects are the appropriate panel estimationapproach failing to reject the null hypothesis that fixed effects is most suitable for all models run forequation (2) (p-values range between 01211-08688) Each model using random effects reports theHausman test χ2 statistics and the p-value of the test in the tables The random effects model also allowsboth between and within effects incorporating the role of competition across provinces19 Furtherrandom effect models allow greater flexibility for the extended models used below and also allow forthe inclusion of time-invariant controls initial forest cover and palm-oil suitability at province levelwhich is valuable information lost in a fixed effect model (Wooldridge 2015) All specifications alsoreport cluster robust standard errors at province level according to the baseline province allocation

As an additional specification and an extension on the prior literature we stratify the effects of adistrict split over time To do so we consider the number of splits in each province per year rather thanthe level number of districts in the province This allows a greater understanding of the time dynamicswhich aids in uncovering whether weak capacity or corruption are the channels causing fire activityfollowing a district split For this we estimate Equation (3)

lnFirespit = β0 + β1minus4Splitspit [minus2 +2 ] + γprimeXpi + δprimeZpit + Yt + Ii + εpit (3)

where Splitspit [minus2 +2 ] are the two year lead one year lead current year and one and two year lagsof the number of splits that occurred in a given province All other variables are previously definedEquation (3) is also estimated by random effects as a robustness check as we fail to reject the null of

18For more details on the palm oil suitability measure see Section 319It is possible that not only districts compete to allow fire activity but also provinces and the random effects estimator would

best incorporate this effect because it captures both between and within effects

14

the Hausman test for all specifications with p-values ranging from 01941-09998 As in Equation (2)all standard errors are robust and clustered at province level

Considering district-level fire activity Equation (4) uses an indicator for whether a district hassplit This allows determination of the impact of a district split at district level across two differentspecifications first to consider the impact for the newly created child district (the part of the districtnot inheriting the district capital) and second to consider the impact of a split on the parent district(the part of the district inheriting the district capital) Our specific is as in Equation (4)

lnFiresdpt = β0 + β1minus6Splitdpt [minus1 +4 ] + γprimeXdp + δprimeZdpt + Pp + Yt + εdpt (4)

where Splitdpt [minus1 +4 ] is = 1 if district d split from another district in year t The results from theHausman test strongly reject the null so for Equation (4) we use a fixed effects estimator When we runthe fixed effects estimator we only include forest cover and omit palm oil suitability from Xdp and weuse a district fixed effect rather than the province fixed effect Pp Zdpt contains annual province rainfalland the unemployment rate Yt is a year fixed effect and εdpt is a robust standard error clustered atdistrict level

411 Exogeneity of District Splits

In our panel data setting what matters is not that district splits occur but rather timing where theinter-annual variation in the timing of district splits provides potential exogenous variation in thecreation of new administrative units Causal interpretations based on the district splits relies on thekey identifying assumption that the factors causing district splits are not a function of the processesdriving fire activity We provide a number of pieces of evidence corroborating this assumption in ourcontext First as discussed in Burgess et al (2012) the bureaucratic process of district splitting isquite idiosyncratic and cumbersome with a series of formal procedures and agreements requiringfinal ratification by the national parliament through passing a national law The timeline for theseprocedures can be highly variable Furthermore the moratorium on splits from 2004 to 2007 providefurther plausibly exogenous delay in some splits Second Burgess et al (2012) conduct a number oftests showing that the timing of splits is not associated with pretrends in deforestation the yeara district split is uncorrelated with factors such as population area oil and gas revenues share ofland that is forested or the pre-period rate of deforestation and that neither district corruption (asmeasured by the share of missing rice from a public distribution program see Olken 2006) nor the voteshare of the former party of Suharto (Golkar) is correlated with the year when a district splits

To show that the timing of district splits are uncorrelated with a range of factors that may influencethe rate of fire activity we regress the number of district splits in each year on a range of potentialcovariates that might drive district splits and simultaneously be related to ongoing fire activity Theresults in Table A3 show that the number of district splits by province are uncorrelated with rainfallforest cover in the year 2000 palm oil suitability and average annual province level unemploymentNationally district splits are uncorrelated with the central bank interest rate and the current and lagged

15

real palm oil prices This supports Burgess et al (2012)rsquos finding that district splits are uncorrelatedwith a range of factors related to deforestation including district corruption and political alliances toSuhartorsquos former party before 2000

Figure 6 Forest Cover (2000)

412 Samples

The main results are estimated for all Indonesian provinces excluding Jakarta which has no districtsplit variation nor any recorded fire activity We also report results for a sample excluding the entiretyof the island of Java the Maluku Islands and the Sunda Islands The primary rationale for using areduced sample is evident in Figure 6 which we use to display primary forest cover data from the year2000 It is evident that there is little forest cover on these excluded islands (for a reference map seeFigure A1) Furthermore while the Maluku islands have a reasonable amount of forest cover they haveminimal palm oil production The reduced estimation sample therefore includes the islands of SumatraKalimantan Sulawesi and Papua which all have significant forest cover in the year 2000 These islandsalso account for the overwhelming majority of palm oil production and as seen in Figure 5 exhibit themost land most suitable for palm production20

42 Global Palm Oil Prices and Fires

To formally assess the role of demand-side factors in fire activity we utilize global palm oil prices asa source of temporal variation in incentives to burn forest for palm oil production While the use ofglobal prices means that claims of exogeneity are more plausible such a measure does not provide anycross-sectional variation Hence our explanatory variable in this part of the analysis interacts the globalpalm oil price with our district-level measure of suitability for conversion to palm oil a measure whichas noted in Section 3 relies on inputs that are invariant over time Increases in palm oil prices are likely

20In 2010 55 percent of palm oil production occurred in Sumatra 42 percent in Kalimantan 2 percent in Sulawesi and 1percent in Papua (Ministry of Agriculture Data from DAPOER 2015)

16

to increase fire activity in provinces that are more suitable for palm oil production Hence for this set ofanalysis the most suitable geographic unit is the district so we will have larger sample sizes having thedistrict as the geographic unit of analysis

The specification of our main equation is given in Equation (5)

lnFiresdt = β0 + β1Suitd lowast lnPricet + β2Suitd + β3lnPricet +Mt + Yt +Dd + εdt (5)

where Firesdt are monthly fires in district d Suitd is district average palm oil suitability Pricet isthe real palm oil price term as defined in Section 33 and Mt Yt and Dd are month year and districtfixed effects respectively The interaction term can be considered a differences-in-differences estimatorand is the main estimator of interest εdt is a robust standard error clustered at district level Becausewe now consider monthly data we exclude the non-forested islands for all specifications because thereare much higher incidences of zero counts in the data for these islands Similarly to Wheeler et al(2013) who note the strength of random effects in their analysis of the effects of palm oil prices ondeforestation this model is also estimated using random effects validated again by strong Hausmantest results We experiment with numerous lag specifications

From the perspective of causal identification one may be concerned about reverse causality underwhich fire activity negatively impacts current palm oil production (eg as smoke has adverse effects onthe productivity of palm oil plants) and hence impacts prices which could in turn drive fire activityWith over 50 percent of global palm oil production located in Indonesia and Malaysia fires and haze inIndonesia may impact palm oil production processes regionally and hence lead to lower global supplyor a lower quality of supply21 However we argue that this concern is largely mitigated due to timingFirst as we look at contemporaneous fire outcomes and palm oil prices are likely based on productionyields from prior months due to harvest and transport times this influence is likely to be low Secondlythe decision to burn forest to plant palm oil is likely based on expectations formed about future palm oilreturns These are likely to be based on a series of palm earnings realisations over time mitigating theinfluence of the current or most recent price realisations Using lagged rather than contemporaneousprices further mitigates this issue

As noted the agentrsquos decision to start fires to convert land to palm oil is likely based on theirexpectations about future returns from the conversion Because of the time it takes to develop palm oilplantations there is likely a complex process of belief formation that occurs over a matter of months oryears For this reason we investigate numerous lag specifications Because the overwhelming majorityof fires occur between July and November as evident in Figure A4 we exclude all other months for thepalm oil price results We therefore have estimates that show the impact of prices on fires occurringbetween July and November the time during which majority of land conversions occur

21Caliman and Southworth (1998) show that haze had a negative impact on palm oil extraction rates

17

5 Results

This section first reports results on the effects of strength of governance on fires then proceeds to reportthe results on the effects of global palm oil prices on fires

51 Weak Governance and Fires The Effect of District Splits

511 Main Results

The results for Equation (1) are reported below in Table 1 The coefficient on the number of districtsterm suggest a statistically significant increase in fire activity for an additional district of 3 to 117percent Columns 1-3 exclude the non-forested islands while columns 3-6 only exclude Jakarta Asnoted columns 2-3 and 5-6 include a control for the log of rainfall while columns 3 and 6 include alinear forest trend and island-by-year fixed effects22 The inclusion if IslandYear fixed effects alsoallows conditions in different islands to vary across time The results also indicate that consistent withexpectations fires are decreasing in rainfall

Table 1 Number of Districts Per Province

Excludes Non-Forest Islands Only Jakarta

(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532

(0223) (0230) (0175) (0212)Palm Oil Suitability 00004 00005 -00000 00005

(00001) (00001) (00000) (00000)Province 1990 FE Yes Yes Yes Yes Yes YesYear FE Yes Yes No Yes Yes NoIsland by Year No No Yes No No YesLinear Forest Trend No No Yes No No YesNumber of Clusters 17 23 17 27 27 27Number of Observations 322 322 322 462 458 458

All estimates from negative binomial count data model Dependent variable is a count of the number of fires per province peryear Standard errors clustered by 1990 province boundaries in parentheses Columns 1-3 exclude Java the Maluku Islands andthe Lesser Sunda Islands and all columns exclude Jakartalowast plt01 lowastlowast plt005 lowast lowast lowast plt001

The results for the random effects panel model in Equation (2) are reported in Table The Hausmantest results strongly confirm the suitability of the random effects estimation23 The dependent variableis the log of the number of fires with -01 inserted for zeros and missing observations We findqualitatively consistent results with the negative binomial specification

22Results are virtually unchanged for the inclusion of a quadratic trend23For completeness we also include fixed effects estimates in the appendix

18

To understand the impacts of district splits over time the results from Equation (3) and Equation (4)are included in Table 2 and Table 3 respectively The results in Table 2 indicate a 89 to 176 percentincrease in fires in the year a split is announced and a 5 to 65 percent increase in fires the year the splitis implemented with the latter only significant in our richest specifications in columns 3 and 6 Thereare no effects detectable in subsequent years at a 5 significance threshold however it is notable thatthe 2-year lag of a split has a qualitatively negative coefficient in one case significant at 10

The results in Table 3 consider district-level fire activity so that we can distinguish effects betweenparent and child districts Columns 1-3 report results for the child district newly established followinga split and columns 4-6 report results for the parent districts that lose territory to the child districtbut retain the prior district capital (and hence are assumed to retain much of the previous districtgovernment apparatus) The Hausman test results strongly reject the null hypothesis and so the resultsfor Equation (4) are estimated using fixed effects The results are generally consistent for randomeffects specifications although less precise There is a statistically significant 15-23 percent increasein fires in the child district the year following the split with no immediate effect in the parent districtAgain this is suggestive of weakened capacity to prevent fires in the child district following the splitwhile the parent district may have better resources following the split as noted by Pal and Wahhaj(2017) Interestingly there are statistically significant decreases in fire activity 2-3 years followingthe split for the parent district which could again reinforce the theory of capacity if governmentsstrengthen control over a reduced territory in the years following the split

Overall these findings are more consistent with a theory of weakened governance capacity as themain explanation for our results with newly-established district governments lacking the resources orcapability to prevent fires in the short-run while setting up a new district It is also possible that themechanical effect of a split reducing the size of a district eases the burden of enforcement In any casethe results are strongly inconsistent with a mechanism of district splits generating a persistent increasein corrupt behaviour to allow for fires

512 Robustness of the District Split Results

To show that the district split results are not driven by construction of our data we vary the sampleby changing the confidence and FRP percentiles and report the results in Table A4 Panel A reportsinformation on the fires for the varied sample Panel B reports estimates for column 4 of Table 2 andPanel C reports estimates from column 3 of Table The results appear to grow in magnitude withmore accurate fire detection data but are relatively unchanged to changes in FRP This suggests thatthese results are unaffected by focusing on larger stronger fires The use of 80 percent confidencelevel in the fire detection data hence seems like a reasonable middle ground The negative binomialmodel results in Table are further evidence of the reliability of the district splits results with aconsistent 5 percent rise in province level fire activity established using a count model relative to themain specifications with a log dependent variable

19

Table 2 Number of Splits per Province per Year


(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189

(0221) (0216) (0227) (0169) (0174) (0681)Palm Oil Suitability -00002 -00000 00005 -00002 -00000 -00003

(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819



513 What Drives the District Split Result

The above results are most consistent with a model of weakened governance capacity of newly createddistrict governments as opposed to alternative theories in particular that of increased corruptiondue to heightened competition for bribes amongst a slightly larger set of district governments in agiven province To further corroborate these findings this section provides additional analysis on themechanisms driving the result for the effects of district splits on fires

Evidence from Government FinancesTable 4 reports district spending patterns as a percentage of district budgets for a range of categories

The average for each measure over 2001 and 2010 is constructed and a means test is undertakenbetween split districts (that were newly created over the period 2001-2010) and non-split districts(including parent districts) New districts spend lower percentages of their budget on education healthhousing the environment One possible explanation would be that newly created district governmentsmust spend more on administration and infrastructure than their counterparts due to lower levels of

20

Table 3 The Effect of District Splits Across Time on District Fire Activity for Child and Parent Districts

Excludes Child Parent

(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)

Non-Forested Yes No No Yes No NoYear FE Yes Yes Yes Yes Yes YesDistrict FE Yes Yes Yes Yes Yes YesLinear Forest Trend No No Yes No No YesR2 00143 00181 000442 00139 00190 000504Number of Districts 386 258 255 386 258 255Number of Observations 5387 3612 2873 5387 3612 2873

Cluster-robust standard errors in parentheses The dependent variable is the log of annual number of fires in the province Allcolumns run a fixed-effects panel regression Split is a dummy = 1 if a district splits in a given year Child districts are thenewly created districts parent districts retain the existing capital following the split The linear forest trend interacts the forestcover in the year 2000 with a linear time trend Columns 1 and 4 exclude Jakarta and Columns 2 3 5 and 6 exclude Java theLesser-Sunda islands and the Maluku Islands The bottom panel contains results for a Hausman (1978) test for random effectsagainst a null of fixed effectslowast plt01 lowastlowast plt005 lowast lowast lowast plt001

administrative infrastructure following the split and this would be consistent with a theory of weakenedcapacity of new districts This is disputed by Sjahir Kis-Katos and Schulze (2014) who argue thathigher administrative spending is characteristic of poor governance and not explained by districtsplits Another potential explanation is that newly elected governments must generate revenue streamsand do so via misappropriation of administrative and infrastructure spending (as argued by SjahirKis-Katos and Schulze (2014)) Using data from the Indonesian Village census by PODEs (via theDAPOER database) Table 4 also shows that newly created districts have 23 percent less properlysurfaced village roads than their counterparts Olken (2007) notes that village road contracts are asource of district level corruption and despite larger infrastructure spending budgets across the samplelower rates of road coverage in 2010 may suggest some level of appropriation These results are only

21

Table 4 Comparing Split and Non-Split Districts Finances

No Split Split Difference p-valueDistrict Government Spending () of Budget (2001-2010)Education 3311 2165 -1146 0000General Administration 3274 4098 8233 0000Infrastructure 1394 187 475 0000Health 827 714 -113 0000Agriculture 386 487 101 0000Economy 236 224 -012 0103Housing 221 107 -114 0000Environment 16 122 -038 0000Social Protection 067 08 0124 0000Tourism 056 0512 -0044 0369Crime 055 078 022 0000Religious Spending 013 006 -007 0000

Villages with Asphalted Roads () (2010) 6751 4406 -2344 0000

Split refers to newly created districts over the period 2001-2010 Data from the Ministry of Finance andPODES via the DAPOER Database (see Table A1 for more information)

indicative and certainly donrsquot suggest that district governments act corruptly nor do they suggest thatspending patterns are indicative of weak capacity following but they do show that non-split and splitdistricts have differences in spending patterns following their split

Evidence from Palm Oil Production To investigate if fire activity is linked to both palm oil anddistrict splits data on palm oil production at district level was accessed from the Ministry of Agriculture(via DAPOER) Unfortunately these data were of poor quality with multiple missing years and somedistricts having unreasonable falls then rises in palm oil production sometimes in excess of 1000percent across years for a given district For this reason it is not possible to use these data for acomprehensive analysis Instead using data from 2010 that generally appears complete we conducta means test on the percentage of palm oil ownership by type in 2010 and the associated palm oilyields The results in Table 5 indicate that newly created district governments have 9 percent moreprivate sector palm oil If private sector firms are more likely to burn large sections of land to clear itas indicated by numerous sources but disputed by others this may contribute to a rise in fires that wesee24 However given the limitation of the data we are unable to comment on whether or not palm oilcompanies act in this way There are no significant differences in yields although small holders have 8percent lower yields in newly created districts

The Confusion Over Land Initially following a district split we may expect that agents may takeadvantage over the confusion of the new district border There are also comments in the literature aboutthe confusion over land ownership rights between agents (see Purnomo et al 2017 RCA and UNICEF2017) While we canrsquot directly comment on land rights and the confusion surrounding fires we areable to explore if fires increase along borders following a district split One illustrative example is the

24Purnomo et al (2017) outline the confusion over who is responsible for lighting fires on palm oil concessions

22

Table 5 Comparing Split and Non-Split Districts Palm Oil Production

No Split Split Difference p-valuePercentage of Palm Oil Land by Ownership (2010)Private Firms 2206 3126 921 0000Small Holders 686 6219 -641 0011State Owned Firms 608 318 -291 0005

Mean Yield (KgHa) by Ownership (2001-2010)Private Firms 333122 34057 7448 0608Small Holders 24942 227836 -21581 0006State Owned Firms 335792 320362 -15431 0498

Split refers to newly created districts over the period 2001-2010 Data from theIndonesian Ministry of Agriculture via the DAPOER Database (see Table A1 formore information)

case of Ogan Komering Ulu shown in Figure 7 that split into three separate districts in 2003 Panel Ashows the district In 2002 A B and C all belonged to one district In 2003 a split occurred formingOgan Komering Ulu (A) Ogan Komering Ulu Timur (B) and Ogan Komering Selatan (C) Panels B Cand D show fires in 2002 2003 and 2004 respectively indicating the year before the year of and theyear after the split There doesnrsquot appear to be an increase in fire activity along the border regions ofthe newly created districts and fire activity appears to be fairly scattered This pattern is evident acrossthe fires data and fires appear evenly dispersed within the borders of all districts rather than confinedto the borders Instead fire activity is often likely to clearly belong to certain districts and agents donrsquotappear to take advantage of the confusion surrounding district borders

514 Fires or Other Forms of Deforestation

The above analysis raises the question of how important fires are as an instrument of deforestation inIndonesia Burgess et al (2012) provide important evidence to corroborate the idea that illegal loggingis an important channel for deforestation We conduct additional analysis to explore the extent of theimportance of fires as a channel of deforestation Notably a significant amount of deforestation occursin regions such as Riau province along the northern coast of Sumatra which are well-known to also behome to a significant amount of fire activity To do so we utilize province level data from 2002-2008on deforestation from Burgess et al (2012) and consider the correlation between deforestation andfires with the following equation

Deforestationpt = β0 + β1Firespt + Pp + Yy + εpt (6)

where Deforestationpt is a count of the number of pixels deforested in province p in year t Firesptis a count of the number of fires per province and Pp and Yy are province and year fixed effects Themodel is run using OLS Because province boundaries changed after 2008 we revert our fires data to2008 boundaries the boundaries used in Burgess et al (2012)

23

Figure 7 Fire Activity in Ogan Komering

The results in Table 6 suggest a strong correlation between deforestation and fires We vary thesample by island and then by confidence level of fires detection The R2 term suggests that we accountfor roughly 60 percent of the variation in deforestation with fires alone

We can use a back-of-envelope calculation with these estimates to assess the role of fires indeforestation Taking the estimates in column (2) which covers all forested islands we see that theaverage fires causes a loss of 294 pixels of forest in the average province and the average year Themean number of fires per province is 1079 which would account for about 3172 pixels lost per year outof an average of 6674 pixels deforested per year Based on these estimates fires account for just underhalf (475) of deforestation in Indonesia We see variable impacts when considering particularlyprovinces similar magnitude in Sumatra smaller impacts in Kalimantan larger impacts in Sulawesiand slightly larger impacts on Papua The last two columns show the extent to which the estimates varyif we vary the confidence level of the fires index

24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25


This section considers the demand-side channel for fire activity proxying palm oil demand throughglobal palm oil prices First we discuss the correlation between fires and the palm suitability measurefrom the GAEZ data then present results for Equation (5)

To show the rationale behind considering palm oil suitability in our interaction term in Equation(5) we generate the correlation between our palm suitability measure and fire activity A 20 percentincrease in palm suitability is correlated with a 58-84 percent increase in provincial fire activity and a26-33 percent increase in district fire activity on average Moreover the R2 term suggests that we canaccount for 7-23 percent of the variation in fires across provinces and districts from palm oil suitabilityalone

With this in mind we now consider the impact of palm oil prices on fire activity Table 7 presentsestimates of the impact of palm oil prices interacted with palm oil suitability on fires using quarterly lagsfrom 3-12 months The interaction of the palm oil suitability index and the global palm oil price allowsthe palm oil price to have a heterogeneous effect on fires in different districts given their suitabilityColumns 1-4 consider real palm oil prices To provide a point of comparison columns 5-6 considerreal rubber prices as rubber is the second-fastest growing agricultural crop by land area Column 3controls for rubber prices in the palm oil equation and column 6 controls for palm oil prices in therubber equation Doing so incorporates controls for the relative price fluctuations of Indonesiarsquos twofastest growing crops by land area (as evident in Figure 3) and hence the two crops most driving theconversion decisions Column 4 incorporates a linear time trend allowing forest cover in 2000 to havea differential effect across time The interaction terms are all significant indicating that there doesappear to be an important interaction between suitability of a district global prices and fire activityThere is also as expected a clear difference between the impact of palm oil prices and rubber pricesWe also show results for monthly lags from 1-6 months in Table and find similar results

These results clearly suggest that global palm oil prices have a significant role in the decision toconvert forests for palm oil and these decisions are taken in a context with dynamic belief formationHowever the individual coefficient estimates can be difficult to interpret individually with both largepositive and negative coefficients That these effects are estimated even in the presence of time periodfixed effects is particularly difficult to interpret In order to better collectively interpret these estimateswe construct a measure for the marginal effect of various price increases By subtracting the marginalimpact of an increased price from the marginal impact of the average price on fire activity we are ableto better understand how the estimates in Table 7 and Table relate to fire activity The magnitudescomputed give the percentage increase in fires in the average district for a rise in palm oil prices Usingthe coefficients from the results of Equation (5) found in Tables 7 and the expected impact of anincrease in price is given by Equation (7)25

= (NewPricelowastAveSuitlowastβ1+NewPricelowastβ2)minus (AvePricelowastAveSuitlowastβ1+AvePricelowastβ2) (7)

25Because all other terms do not change they drop out and we are only left with the terms in Equation (7)

26

Table 7 Palm Oil Price Increases and District Fires (Quarterly Lags)

Palm Oil Prices Rubber Prices

(1) (2) (3) (4) (5) (6)Suitability x Price[t-3] 1560 1560 1561 1545 0875 0878

(0338) (0341) (0341) (0344) (0296) (0296)Suitability x Price[t-6] -1062 -1067 -1051 -1039 1704 1703

(0321) (0325) (0324) (0327) (0326) (0327)Suitability x Price[t-9] -1957 -1955 -1977 -2012 -4785 -4792


(0225) (0227) (0227) (0225) (0258) (0259)Suitability Index 2009 -0208 -0155 -0179 -0400 -0334

(0483) (0343) (0345) (0346) (00818) (00866)Log Real Palm Oil[t-3] -0863 -0864 00231 00311 1167

(0249) (0251) (0279) (0280) (0144)Log Real Palm Oil[t-6] 0589 0594 1478 1491 0716


(0274) (0276) (0278) (0280) (0165)Log Real Palm Oil[t-12] -0701 -0704 -0554 -0584 0416

(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703

(0115) (0116) (0176) (0181)Month FE Yes Yes Yes Yes Yes YesYear FE Yes Yes Yes Yes Yes YesDistrict FE No Yes Yes Yes Yes YesLinear Time Trend No No No Yes No NoR2 0209 0555 0559 0560 0562 0564Number of Districts 256 256 256 253 256 256Number of Observations 12951 12951 12951 12832 12951 12951

Cluster-robust standard errors in parentheses The dependent variable is the log of annual number of fires in the district Allcolumns run a random-effects panel regression Column 4 controls for a linear time trend by interacting log forest cover atdistrict level with a linear time trend Columns 1-4 use real palm oil prices in the suitability x price interaction while columns5-6 use real rubber priceslowast plt01 lowastlowast plt005 lowast lowast lowast plt001

where NewPrice is the average price plus the chosen increase in prices AveSuit is the average palm oilsuitability for all districts (= 0724) and Average Price is the average of the Log of the Real Palm OilPrice (= 181) We can also alter the palm oil suitability average to show the effect for districts more orless suitable for palm oil

27

Table 8 Magnitude of Impact for Increased Palm Oil Prices on Fires (Quarterly Lags)

Increase in Prices 05 Std Dev 1 Std Dev 10 PercentMonth Lag

3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516

Average Effect 0130 0259 0129

Column 2 shows the marginal effect of a 05 standard deviation increase inreal palm oil prices column 3 shows a 1 standard deviation in prices andcolumn 4 shows a 10 percent increase in prices Calculations of marginalimpacts are made from column 4 in Table 7

The magnitudes of price increases for the quarterly lag model over twelve months are included inTable 8 Column 2 shows the increase for a 05 standard deviation increase in real palm oil prices whilecolumns 3 and 4 show a 1 standard deviation and 10 percent increase in prices respectively Giventhe large fluctuation of the global real palm oil price with large rapid swings common (as evident inFigure A2) price rises of this magnitude are realistic The average log price is 181 and the standarddeviation is 041 so column 2 represents a 22 percent increase while column 3 represents a 45 percentincrease in prices

The results show that rising real palm oil prices have a large positive effect on the rate of fires Theaverage marginal effect of a rise in global palm oil prices of 10 percent at a quarterly lag within twelvemonths of the fire season is a 129 percent rise in monthly fire activity in the average district Table considers monthly lags for the six months leading into the fire season Again there appears to bea strong effect of global palm oil prices on fire activity For a 10 percent increase in monthly laggedprices on average there is a 7 percent increase in monthly fires We also consider the net effect ofa consistent rise in prices across the year or six months In Table the net effect suggests that ifpalm oil prices rose at 10 percent per quarter leading into fire season that we would expect to see a 35percent increase in monthly fire activity for the average district Table 9 suggests that if monthly pricesrose consistently at 10 percent per month for the six months leading into the fire season that we wouldexpect to see a 43 percent increase in monthly fire activity in the average district

We also construct the same magnitude measure for rubber prices to consider the impact of rubberprices on fire activity Using Column 5 in Tables 7 and we summarise the impact of rubber pricesfinding that increased rubber prices have no impact on fire activity and that if price rises are largeenough that there can be negative effects on fires Table A6 and Table A7 show that the expectedimpact of a 05 standard deviation increase in rubber prices leads to a 10 to 16 percent decrease in fireactivity in the average district A 10 percent increase in rubber prices has no effect on fire activity inboth the quarterly and monthly models This wouldnrsquot be explained by rubber plantations not using firefor conversion as generally all land conversion in Indonesia occurs via fire When the rubber pricerises agents may delay their decision to convert land to palm oil until they sure of the long-term price

28

Table 9 Magnitude of Impact for Increased Palm Oil Prices on Fires (Monthly Lags)

Increase in Prices 05 Std Dev 1 Std Dev 10 PercentMonths Lag

1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317

Net Effect 0478 0955 0422Average Effect 0080 0159 0070

Column 2 shows the marginal effect of a 05 standard deviation increase inreal palm oil prices column 3 shows a 1 standard deviation in prices andcolumn 4 shows a 10 percent increase in prices Calculations of marginalimpacts made from column 3 in Table A5

movement Another potential explanation is that palm oil is a more attractive crop than rubber becauseit returns output within four years of planting while rubber takes seven to ten years (Feintrenie etal 2010) and agents may be converting rubber plantations to palm oil but delay doing so when therelative price of rubber is high The model controlling for palm oil prices may capture this due to thepositive correlation between the two series26

These computed magnitudes suggest that the conversion to palm oil is a significant contributingfactor to annual district fire activity This validates the hypothesized link between palm oil and forestfires from numerous authors including Dauvergne (1998) Dennis et al (2005) and Purnomo et al(2017) Further while our regression approach is likely to overfit the existing data within sample ourestimates suggest that it would possible to develop an appropriate model of palm oil prices and districtmonthly fire activity that could be used for active prediction of fire hot spots across years

6 Conclusion

A significant literature documents that the fires plaguing southeast Asia each year have major envi-ronmental health and economic implications However to date there has been a lack of systematicquantitative research into the factors contributing to the fires We consider two potential channels Firstwe consider the role of weak governance in preventing fires Taking advantage of exogenous variationin the creation of new administrative districts we show that the creation of a new district leads to a3-117 percent increase in fire activity at province level These results complement the findings ofBurgess et al (2012) who provide evidence that some district officials compete to accept bribes foroverlooking illegal logging with district splits intensifying this competition In the context of firesthe evidence is more consistent with the mechanism of reduced governance capacity of newly-createddistrict governments following a district split increases in fires are concentrated around the time of a

26The correlation between palm oil and rubber prices is 086

29

split and in newly-created (rather than continuing) districts We can rationalize these differences byobserving that fires are a much more publicly-visible and unambiguous form of illegal activity thanillegal logging Furthermore we show that variation in fire activity explains a little under half of alldeforestation in Indonesia

Using global palm oil prices and a geo-specific palm oil suitability measure we establish a causallink between global palm oil demand as proxied by prices and fires A simple simulation exercisebased on our linear model shows that for the average district an increase in palm oil prices of 10 percentin the lead up to fire season will lead to a 129 percent increase in monthly fire activity We show thatpalm oil prices contribute positively to fire activity across annual and six-monthly lags leading into thefire season This analysis suggests the possibility of developing a quantitative tool to rapidly predictfires across space and time More generally we have shown that demand-side factors have significantimplications for fire activity while failing to find a similar link for Indonesiarsquos second fastest growingcrop by land area rubber

More broadly these results highlight two key factors that contribute to differences in fires acrossdistricts and years Rainfall and hence the cycles of El Nino and La Nina are also a major contributingfactor to annual fire activity as is the level of forest cover This paper also shows that land more suitableto palm is more likely to be burnt The results suggest that improved governance has a key role to playin fire prevention and the Indonesian government must promote sustainable methods for conversion topalm oil as they seek to meet their palm oil production objectives

This study has important implications for the design of policies intended to reduce the prevalenceof fires The central government of Indonesia aims to double palm oil production between 2015 and2020 so it is important to consider the impact palm oil may be having on fires (Edwards and Heiduk2015) Our analysis provides evidence on a couple of key channels behind the fires and shows that theyplay a significant overall role in causing deforestation in Indonesia In particular the results suggestthat in order to reduce fires the government should pay particular attention to governance capacity inidentifying monitoring and enforcing anti-burning laws Our analysis also suggests the possibility ofconstructing a predictive model to forecast fire activity across Indonesia to coordinate fire preventionactivities combining indicators that are cheap and publicly-available with minimal time lag such asglobal palm oil prices rainfall and measures of suitability for palm oil conversion Such a model couldbe augmented with richer data such as on local palm oil prices This raises the potential of developingmethods to target prevention and enforcement activities where new fires are most likely to occur

30

ReferencesAlesina A Gennaioli C and Lovo S (2016) Public Goods and Ethnic Diversity Evidence fromDeforestation in Indonesia (No w20504) National Bureau of Economic Research

Alisjahbana A and Busch J (2017) Forestry Forest Fires and Climate Change in Indonesia Bulletinof Indonesian Economic Studies 53(3) 111-36

Banerjee A Hanna R Kyle J Olken B and Sumarto S (Forthcoming) Tangible Informationand Citizen Empowerment Identification Cards and Food Subsidy Programs in Indonesia Journal ofPolitical Economy

Bell A and Jones K (2015) Explaining fixed effects Random effects modeling of time-seriescross-sectional and panel data Political Science Research and Methods 3(1) 133-153

Blumenfeld J (2016) Near Real-Time VIIRS Products Now Available through LANCE NASA EOS-DIS Accessed online October 30 Retrieved from httpsearthdatanasagovearth-observation-datanear-real-timeviirs-nrt-products-available

Burgess R Hansen M Olken B A Potapov P and Sieber S (2012) The political economy ofdeforestation in the tropics The Quarterly Journal of Economics 127(4) 1707-1754

Cahyadi E R and Waibel H (2013) Is contract farming in the Indonesian oil palm industry pro-poorJournal of Southeast Asian Economies 30(1) 62-76

Caliman J P and Southworth A (1998) Effect of drought and haze on the performance of oil palmIn 1998 International Oil Palm Conference Bali (Indonesia) 23-25 Sep 1998 Puslit Kelapa Sawit

Cameron A C and Trivedi P K (2013) Count panel data Oxford Handbook of Panel Data OxfordUniversity Press Oxford

Chander P (2017) A Political Economy Analysis of the Southeast Asian Haze and Some SolutionsRSIS Working Paper No 303

Chen M Xie P Janowiak J E and Arkin P A (2002) Global land precipitation A 50-yr monthlyanalysis based on gauge observations Journal of Hydrometeorology 3(3) 249-266

Dauvergne P (1998) ldquoThe political economy of Indonesiarsquos 1997 forest firesrdquo Australian Journal ofInternational Affairs 52(1) 13 ndash 17

Dennis R A Mayer J Applegate G Chokkalingam U Colfer C J P Kurniawan I LachowskiH Maus P Permana RP Ruchiat Y and Stolle F (2005) Fire people and pixels linking socialscience and remote sensing to understand underlying causes and impacts of fires in Indonesia HumanEcology 33(4) 465-504

Edwards S A and Heiduk F (2015) Hazy days Forest fires and the politics of environmentalsecurity in Indonesia Journal of Current Southeast Asian Affairs 34(3) 65-94

Edwards R (2017) Tropical Oil Crops and Rural Poverty Working Paper April 2017

Feintrenie L Chong W K and Levang P (2010) Why do farmers prefer oil palm Lessons learnt

31

from Bungo district Indonesia Small-scale Forestry 9(3) 379-396

Ferretti-Gallon K and Busch J (2014) What drives deforestation and what stops it A meta-analysisof spatially explicit econometric studies

Fisman R (2001) Estimating the value of political connections The American Economic Review91(4) 1095-1102

Frankenberg E McKee D and Thomas D (2005) Health consequences of forest fires in IndonesiaDemography 42(1) 109-129

Furumo P R and Aide T M (2017) Characterizing commercial oil palm expansion in Latin Americaland use change and trade Environmental Research Letters 12(2) 024008

Gatto M Wollni M Asnawi R and Qaim M (2017) Oil palm boom contract farming and ruraleconomic development Village-level evidence from Indonesia World Development 95 127-140

Giglio L (2015) MODIS Collection 6 Active Fire Product Userrsquos Guide Revision A Unpublishedmanuscript Department of Geographical Sciences University of Maryland httpslpdaacusgsgovsitesdefaultfilespublicproduct_documentationmod14_user_guidepdf

Giglio L Schroeder W and Justice C O (2016) The collection 6 MODIS active fire detectionalgorithm and fire products Remote Sensing of Environment 178 31-41

Hausman J A (1978) Specification tests in econometrics Econometrica Journal of the EconometricSociety 1251-1271

Huijnen V Wooster M J Kaiser J W Gaveau D L A Flemming J Parrington MInness AMurdiyarso D Main B and Van Weele M (2016) Fire carbon emissions over maritime southeastAsia in 2015 largest since 1997 Scientific Reports 6 26886

Jayachandran S (2009) Air quality and early-life mortality evidence from Indonesiarsquos wildfiresJournal of Human Resources 44(4) 916-954

Jia R and Nie H (2017) Decentralization Collusion and Coal Mine Deaths Review of Economicsand Statistics 99(1) 105-118

Koplitz S N Mickley L J Marlier M E Buonocore J J Kim P S Liu T Sulprizio MPDeFries RS Jacob DJ Schwartz J and Pongsiri M (2016) Public health impacts of the severehaze in Equatorial Asia in SeptemberndashOctober 2015 demonstration of a new framework for informingfire management strategies to reduce downwind smoke exposure Environmental Research Letters11(9) 094023

Koyuncu C and Yilmaz R (2009) The impact of corruption on deforestation a cross-countryevidence The Journal of Developing Areas 42(2) 213-222

Martinez-Bravo M Mukherjee P and Stegmann A (2016) The Non-Democratic Roots of EliteCapture Evidence from Soeharto Mayors in Indonesia Centro de Estudios Monetarios y Financieros(CEMFI) Working Paper 1601

Mendes C M and Junior S P (2012) Deforestation economic growth and corruption a nonpara-

32

metric analysis on the case of Amazon forest Applied Economics Letters 19(13) 1285-1291

Mongabay (2017) The palm oil fiefdom A politician in Borneo turned his district into a sea of oilpalm Did it benefit the people who elected him or the members of his family Mongabay October 102017 Retrieved from httpsnewsmongabaycom201710the-palm-oil-fiefdom

Mosnier A Boere E Reumann A Yowargana P Pirker J Havlik P and Pacheco P (2017)Palm oil and likely futures Assessing the potential impacts of zero deforestation commitments and amoratorium on large-scale oil palm plantations in Indonesia CIFOR Infobrief no 177 p 8p) Centerfor International Forestry Research (CIFOR) Bogor Indonesia

Olken B A (2006) Corruption and the costs of redistribution Micro evidence from IndonesiaJournal of Public Economics 90(4) 853-870

Olken B A (2007) Monitoring corruption evidence from a field experiment in Indonesia Journal ofPolitical Economy 115(2) 200-249

Olken B A and Pande R (2012) Corruption in developing countries Annual Review of Economics4(1) 479-509

Oom D and Pereira J M (2013) Exploratory spatial data analysis of global MODIS active fire dataInternational Journal of Applied Earth Observation and Geoinformation 21 326-340

Page S E Siegert F Rieley J O Boehm H D V Jaya A and Limin S (2002) The amount ofcarbon released from peat and forest fires in Indonesia during 1997 Nature 420(6911) 61-65

Pal S and Wahhaj Z (2017) Fiscal decentralisation local institutions and public good provisionevidence from Indonesia Journal of Comparative Economics 45(2) 383-409

Palmer C (2001) The extent and causes of illegal logging an analysis of a major cause of tropicaldeforestation in Indonesia httpdiscoveryuclacuk17588117588pdf

Pierskalla J H (2016) Splitting the difference The politics of district creation in Indonesia Compar-ative Politics 48(2) 249-268

Pirker J Mosnier A Kraxner F Havliacutek P and Obersteiner M (2016) What are the limits to oilpalm expansion Global Environmental Change 40 73-81

Purnomo H Shantiko B Sitorus S Gunawan H Achdiawan R Kartodihardjo H and DewayaniA A (2017) Fire economy and actor network of forest and land fires in Indonesia Forest Policy andEconomics 78 21-31

Scotland N Smith J Lisa H Hiller M Jarvis B Kaiser C Leighton M Paulson L PollardE Ratnasari D and Ravanell R (2000 August) Indonesia country paper on illegal logging In WorldBankndashWWF Workshop on Control of Illegal Logging in East Asia August (Vol 28)

Sheldon T L and Sankaran C (2017) The Impact of Indonesian Forest Fires on SingaporeanPollution and Health American Economic Review 107(5) 526-529

Sjahrir B S Kis-Katos K and Schulze G G (2014) Administrative overspending in Indonesiandistricts The role of local politics World Development 59 166-183

33

Smith J Obidzinski K Subarudi S and Suramenggala I (2003) Illegal logging collusive cor-ruption and fragmented governments in Kalimantan Indonesia International Forestry Review 5(3)293-302

Sundstroumlm A (2016) Understanding illegality and corruption in forest governance Journal of Envi-ronmental Management 181 779-790

The Reality Check Approach+ Team (RCA) and UNICEF (2016) Reality check Approach Perspec-tives of People Affected by Haze from Peatland and Forest Fire Jakarta The Palladium Group andUNICEF

Transparency International (2017) Corruption Perceptions Index 2016 Transparency InternationalAccessed Online May 21 2017 httpswwwtransparencyorg

Wooldridge J M (2015) Introductory econometrics A modern approach Nelson Education

World Bank (2016) The Cost of Fire An Economic Analysis of Indonesiarsquos 2015 Fire Crisis WorldBank Group Jakarta Indonesia February 2016

34

A Additional Tables and Figures

Figure A1 Context Map

Figure A2 Real Palm Oil Price (2001-2015)

35

Figure A3 Co-movement of Rupiah and USD prices

Figure A4 Number of Fires By Month (2002-2015)

36

Table A1 Additional Data Sources

Product Notes SourceRainfall Monthly mean rainfall at 05 degree resolu-

tion measured in mm per dayNOAA NCEP CPC PRECLChen et al (2002)

District Splits Induced from BPS village codes 1998-2014and verified using the World Bank DAPOERCross-Walk

BPS (2014) httpswwwbpsgoid DAPOER (World Bank2015) httpsdataworldbankorgdata-catalog

Forest Cover Primary forest cover across Indonesia in theyear 2000

Global Forest Watch Hansen(2006)

Palm Oil Suitabil-ity

High resolution global and Indonesian spe-cific Palm Suitability index

Pirker et al (2016) Mosnier etal (2017)

USD Palm Prices Monthly Crude Palm Oil Futures End of DaySettlement Prices on the Chicago MercantileExchange Prices in USD per Metric Ton

World Bank (2017) httpsdataworldbankorgdata-catalogcommodity-price-data

USA GDP Im-plicit Price Defla-tor

Quarterly USA GDP Implicit Price deflatorused to convert prices to real terms

Bureau of Economic Analysis(2017) httpsfredstlouisfedorgseriesGDPDEF

Palm Oil Produc-tion Data

Annual district and province level palm oilproduction statistics by type of palm palmplantation owner and yields

Indonesian Ministry of Agri-culture via DAPOER (WorldBank 2015) httpsdataworldbankorgdata-catalog

Interbank InterestRates

Annual Average 3-Month or 90-day Inter-bank Rates for Indonesia

OECD (2017) httpsdataoecdorg

District Govern-ment Budgets

National District spending figures Used tocompute percentages of annual budgets

Indonesian Ministry of Financevia DAPOER (World Bank2015) httpsdataworldbankorgdata-catalog

Village RoadData

Percentage of villages per district with As-phalted roads

PODES Data via DAPOER(World Bank) httpsdataworldbankorgdata-catalog

Unemployment Annual province unemployment rate perprovince

BPS (2017)httpswwwbpsgoid

Rupiah PalmPrices

Crude Palm Oil Futures End of Day Settle-ment Prices on Malaysian Palm Oil Mar-ket Converted to Indonesian Rupiah perMetric Ton httpwwwindexmundicomcommoditiescommodity=palm-oil

World Bank (2017) viahttpsdataworldbankorgdata-catalogcommodity-price-data

USD RubberPrices

Monthly Singapore Commodity ExchangeNo3 Rubber Smoked Sheets 1st Contract inUS cents per pound deflated using the USAGDP Implicit Price Deflator

IMF (2017) via httpswwwquandlcomdataODAPRUBB_USD-Rubber-Price

Area of Har-vested Crops

Total area of production of a range of Cropsfor Indonesia over the period 1990-2015

FAO (2017) httpwwwfaoorgfaostatendataQC

37

Table A2 Summary Statistics

Province Kabupaten Kabupaten District Fires Rainfall Palm Unemployment2002 2015 Splits Suitability

Sumatera 81 119 38 147470 8788 337 810Aceh 16 18 2 3332 7569 179 969Riau 12 10 1 59055 8652 430 883Jambi 9 9 0 17162 8597 352 571Bengkulu 3 9 6 1310 9342 252 532Lampung 8 13 5 3699 8137 346 690Kepulauan Riau 0 5 2 713 8784 360 864Kepulauan Bangka Belitung 2 6 4 4010 8530 445 590Sumatera Utara (North) 13 25 12 7427 9193 240 891Sumatera Selatan (South) 6 12 6 47236 8758 364 768Sumatera Barat (West) 12 12 0 3526 10351 209 902Java 82 84 2 4827 8012 251 917DKI Jakarta 0 0 0 0 8139 238 1177Banten 4 4 0 115 9433 338 1354Daerah Istimewa Yogyakarta 4 4 0 40 7403 269 522Jawa Timur (East) 29 29 0 3440 6806 201 604Jawa Barat (West) 16 18 2 715 8270 202 1152Jawa Tengah (Centre) 29 29 0 517 8021 256 692Sunda Islands 27 37 10 10456 5657 059 462Bali 8 8 0 36 6286 126 357Nusa Tenggara Timur (East) 13 21 8 8103 5826 023 376Nusa Tenggara Barat (West) 6 8 2 2317 4944 027 652Kalimantan 30 46 16 146960 9130 298 749Kalimantan Utara (North) 0 4 0 2897 8899 147 1014Kalimantan Selatan (South) 9 11 2 9803 7892 334 631Kalimantan Timur (East) 8 6 2 14431 8802 311 1027Kalimantan Barat (West) 8 12 4 48727 10960 341 592Kalimantan Tengah (Centre) 5 13 8 71102 9096 358 479Sulawesi 38 66 29 16632 7356 158 797Gorontalo 2 5 3 826 5281 125 752Sulawesi Utara (North) 3 11 8 991 9463 173 1045Sulawesi Selatan (South) 19 21 2 6444 8125 173 1015Sulawesi Barat (West) 3 5 3 1060 9419 140 707Sulawesi Tengah (Centre) 7 12 5 3526 5418 107 594Sulawesi Tenggara 4 12 8 3785 6433 226 666Maluku Islands 6 17 11 3852 7444 231 908Maluku 4 9 5 3018 7977 225 1096Maluku Utara (North) 2 8 6 834 6911 237 721Papua 12 38 26 12104 10167 273 605Papua 9 28 19 11585 10160 269 498Papua Barat (West) 3 10 7 519 10174 277 712

Islands in Bold Fires are the number of fires between 2002-2015 Rainfall is average annual rainfall in mm Palm Suitability is the averagepalm oil suitability by province and Unemployment is the annual average unemployment rate from 2002-2015

38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39

Table A4 Robustness checks by varying sample details

Confidence Fire Radiative Power

Percentile 80 60 70 90 100 75 95

(1) (2) (3) (4) (5) (6) (7)Panel A Fires DataNumber of Fires 347103 770610 562652 177362 79315 232567 46573Average FRP 7404795 461101 5604213 1032906 1363001 1099337 2716709Average Brightness 3381676 327967 3319394 3459732 3515538 3426847 3659683Panel B Equation (2)Districts 00490 00272 00358 00783 00797 00563 00516

(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117

(0117) (0108) (0110) (0146) (0163) (0134) (0137)Province Suitability 1266 1059 1126 1479 1533 1287 1322

(0160) (0176) (0181) (0226) (0254) (0211) (0227)Island FE Yes Yes Yes Yes Yes Yes YesDistrict FE Yes Yes Yes Yes Yes Yes YesR2 0788 0795 0793 0751 0716 0766 0737N 322 322 322 302 295 304 282Panel C Equation (3)Splits [t-1] 00480 00486 00494 00403 00177 00506 00147

(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219


(0163) (0180) (0189) (0254) (0274) (0231) (0235)Island FE Yes Yes Yes Yes Yes Yes YesDistrict FE Yes Yes Yes Yes Yes Yes YesR2 0738 0770 0760 0672 0642 0708 0714N 322 322 322 302 295 304 282

Cluster-robust standard errors in parentheses The dependent variable is the log of annual province fire activity Columns 1-5 vary theconfidence level of the fire detection while columns 6 and 7 vary the sample based on the 75th and 95th percentile of the Fire Radiative Powermeasure respectively Panel A includes details about the number of fire points for each sample and the mean of Fire Radiative Power andBrightness of the fires detected in that sample Panel B reports estimates for Column (4) of Table and Panel C reports estimates of Column(3) of Table All variables are as described in Section 3 and 4lowast plt01 lowastlowast plt005 lowast lowast lowast plt001

40

Table A5 Palm Oil Price Increases and District Fires (Monthly Lags)



(0364) (0357) (0357) (0358) (0361) (0360)Suitability x Price[t-2] 2648 2487 2438 2531 -0943 -0960





(0491) (0488) (0489) (0494) (0350) (0350)Sustainability Index 1103 1226 1214 1231 00612 0122

(0359) (0358) (0357) (0360) (00944) (00948)Log Real Palm Oil[t-1] 0449 1355 2844 2886 1226

(0244) (0251) (0330) (0334) (0182)Log Real Palm Oil[t-2] 0901 -1692 -2865 -2909 -1104


(0358) (0371) (0394) (0395) (0177)Log Real Palm Oil[t-4] 0340 -1315 -0947 -0952 1230

(0523) (0556) (0553) (0556) (0278)Log Real Palm Oil[t-5] 1101 00580 -0972 -1003 -1880

(0506) (0515) (0518) (0520) (0322)Log Real Palm Oil[t-6] -0159 1933 3553 3595 1779

(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591


Robust standard errors clustered at district level in parentheses Dependent variable is the log of district level fire activityColumn 4 controls for a linear time trend by interacting log forest cover at district level with a linear time trend Price ininteraction term is specified above the column All models run using random effectslowast plt01 lowastlowast plt005 lowast lowast lowast plt001

41

Table A6 Magnitude of Impact for Increased Rubber Prices on Fires (Quarterly Lags)

Increase in Prices 05 Std Dev 1 Std Dev 10 PercentMonth Lag1 -0347 -0693 -00122 -0041 -0082 -00013 -0273 -0546 -00094 0041 0082 0001Net Effect -0620 -1240 -0021Average Effect -0155 -0310 -0005

Column 2 shows the marginal effect of a 05 standard deviation increasein real rubber prices column 3 shows a 1 standard deviation in real rub-ber prices and column 4 shows a 10 percent increase in real rubber pricesCalculations of marginal impacts made from column 6 in Table 7

Table A7 Magnitude of Impact for Increased Rubber Prices on Fires (Monthly Lags)

Increase in Prices 05 Std Dev 1 Std Dev 10 PercentMonth Lag1 -0405 -0810 -00142 0407 0813 00143 -0593 -1185 -00204 0238 0475 00085 -0226 -0451 -00086 -0002 -0003 0000Net Effect -0580 -1161 -0019Average Effect -0097 -0193 -0003

Column 2 shows the marginal effect of a 05 standard deviation increasein real rubber prices column 3 shows a 1 standard deviation in real rub-ber prices and column 4 shows a 10 percent increase in real rubber pricesCalculations of marginal impacts made from column 6 in Table

42

Table A8 Number of Districts Per Province


(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951

(0416) (0365) (0330) (0342)Province 1990 FE Yes Yes Yes Yes Yes YesYear FE Yes Yes No Yes Yes NoIsland by Year No No Yes No No YesLinear Forest Trend No No Yes No No YesR2 0144 00451 0145 0131 00422 0209Number of Clusters 23 23 23 33 33 33Number of Observations 322 322 322 462 458 458

All columns are Fixed Effects regressions Dependent variable is the log of the number of fires per province per yearStandard errors clustered by 1990 province boundaries in parentheses Columns 1-3 exclude Java the Maluku Islands andthe Lesser Sunda Islandslowast plt01 lowastlowast plt005 lowast lowast lowast plt001

43

Table A9 Number of Splits per Province per Year


(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984

(0408) (0411) (0379) (0333) (0323) (0339)Palm Oil Suitability 00194 000205 00538 00239

(00297) (00521) (00316) (00528)Districts Yes Yes Yes Yes Yes Yes

Yes Yes No Yes Yes NoProvince 1990 FE Yes Yes Yes Yes Yes YesYear FE Yes Yes No Yes Yes NoIsland by Year No No Yes No No YesLinear Forest Trend No No Yes No No YesR2 00434 00605 0148 00255 00575 0225Number of Clusters 23 23 23 33 33 33Number of Observations 322 322 322 458 458 458

All columns are Fixed Effects regressions Dependent variable is the log of the number of fires per province per year Standarderrors clustered by 1990 province boundaries in parentheses Columns 1-3 exclude Java the Maluku Islands and the LesserSunda Islandslowast plt01 lowastlowast plt005 lowast lowast lowast plt001

44

Introduction

Why Do People Burn the Forest

Administrative changes in Post-Suharto Indonesia

Indonesias palm oil industry

The Characteristics of Palm Oil

Data

Active Fire Data

Palm Oil Suitability Index

Price Data

Additional Data

Methodology

Weak Governance and Fires the Effect of District Splits

Exogeneity of District Splits

Samples

Global Palm Oil Prices and Fires

Results

Weak Governance and Fires The Effect of District Splits

Main Results

Robustness of the District Split Results

What Drives the District Split Result

Fires or Other Forms of Deforestation


Conclusion

Additional Tables and Figures

1 Introduction

Each year between July and November equatorial Asia experiences a dense smokey haze as Indonesiarsquosforests are illegally burnt to clear land for agriculture The fires cause numerous complications forIndonesia and neighboring Malaysia and Singapore and have become a contentious internationalrelations issue in the region The haze causes significant respiratory issues in affected populations(Frankenberg Mckee and Thomas 2005) Studies have estimated a 12 percent increase in infantmortality in regions exposed to the haze in 1997 and over 100000 premature deaths in IndonesiaMalaysia and Singapore during 2015 a year of particularly intense fire activity (Jaychandran 2009Koplitz et al 2016) The fires also have substantial environmental consequences especially when theburning occurs on peatland a dense forest that emits substantial amounts of carbon when burnt (Pageet al 2002 Alisjahbana and Busch 2017) During 2015 Indonesia emitted more carbon than the entireEuropean Union and ranked third for total carbon pollution globally (Huijen et al 2016 Edwards andHeiduk 2015) Forest fires contribute an overwhelming amount of this pollution accounting for 69percent of Indonesiarsquos carbon emissions over the period 1989-2008 (Carson et al 2012) The economicconsequences are also severe with estimates suggesting that the fires in 2015 cost the Indonesianeconomy $161 billion USD equivalent to 19 percent of GDP a figure which fails to consider additionaladverse impacts on Singapore and Malaysia (World Bank 2016) This paper provides systematicquantitative evidence on some of the political and economic forces behind these annual fire events

The forest fires are considered to primarily be the result of slash and burn land conversionpractices generally to convert land to palm oil (Dauvergne 1998 Dennis et al 2005 Purnomo etal 2017) Palm oil is Indonesiarsquos biggest agricultural export and has experienced rapid growth sincethe turn of the century with a deliberate government goal to increase production of palm oil as apoverty reduction strategy (Gatto et al 2017 Edwards 2017 Edwards and Heiduk 2015) Palmoil concessions covered under 4 million hectares before 2000 and over 12 million hectares by 2014(Edwards 2017) Most of the expansion of palm oil has come from converting forest-covered land topalm plantations (Furumo and Aide 2017)

This paper provides causal estimates motivated by two of the main hypothesized channels behindthe fires (1) the role of weak governance and the (lack of) incentives for local government officials toenforce laws outlawing the fires and (2) the returns from land conversion to palm oil driven by globalpalm oil demand To do so we combine a number of remotely-sensed and administrative data sourcescovering the early 2000s to the mid-2010s on fires rainfall forest cover administrative boundariesagricultural conditions (the suitability of local regions for palm oil conversion) and agriculturalprices of the two fastest-growing crops in terms of land area palm oil and rubber In supplementalexploratory analysis to assess the role of the fires in overall deforestation in Indonesia we also utilizedata on deforestation Because of the relatively unavailability and distinct unreliability of surveyor administrative data on fires and deforestation we instead rely on remotely-sensed data on theseoutcomes which has come available in recent years

Despite successive central government attempts to reduce the prevalence of fires including President

1







2





3




4







5







6






7









8





9

3 Data


31 Active Fire Data





10






33 Price Data





11



34 Additional Data



12


4 Methodology









13












14








15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion








2





3




4







5







6






7









8





9

3 Data


31 Active Fire Data





10






33 Price Data





11



34 Additional Data



12


4 Methodology









13












14








15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion






3




4







5







6






7









8





9

3 Data


31 Active Fire Data





10






33 Price Data





11



34 Additional Data



12


4 Methodology









13












14








15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion





4







5







6






7









8





9

3 Data


31 Active Fire Data





10






33 Price Data





11



34 Additional Data



12


4 Methodology









13












14








15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion








5







6






7









8





9

3 Data


31 Active Fire Data





10






33 Price Data





11



34 Additional Data



12


4 Methodology









13












14








15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion








6






7









8





9

3 Data


31 Active Fire Data





10






33 Price Data





11



34 Additional Data



12


4 Methodology









13












14








15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion







7









8





9

3 Data


31 Active Fire Data





10






33 Price Data





11



34 Additional Data



12


4 Methodology









13












14








15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion










8





9

3 Data


31 Active Fire Data





10






33 Price Data





11



34 Additional Data



12


4 Methodology









13












14








15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion






9

3 Data


31 Active Fire Data





10






33 Price Data





11



34 Additional Data



12


4 Methodology









13












14








15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion


3 Data


31 Active Fire Data





10






33 Price Data





11



34 Additional Data



12


4 Methodology









13












14








15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion







33 Price Data





11



34 Additional Data



12


4 Methodology









13












14








15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion




34 Additional Data



12


4 Methodology









13












14








15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion



4 Methodology









13












14








15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion













14








15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion









15



412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion




412 Samples





16








17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion









17

5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion


5 Results



511 Main Results




(1) (2) (3) (4) (5) (6)Districts 0108 00302 0117 00561 00682 0106

(000890) (00108) (00092) (00068) (00074) (000844)Log Rain -2797 -0625 -2181 -0532






18






19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion







19



(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion




(1) (2) (3) (4) (5) (6)Splits [t-2] 00187 000527 0172 -000514 00147 000482

(00604) (00473) (00582) (00548) (00460) (00575)Splits [t-1] 00868 01000 0176 0109 0133 00723

(00393) (00329) (00415) (00329) (00295) (00409)Splits [t] 00336 -00257 00506 -00002 -00207 00645

(00419) (00363) (00253) (00407) (00361) (00333)Splits [t+1] 00626 -000139 00554 00401 00147 -000506

(00416) (00348) (00317) (00391) (00337) (00391)Splits [t+2] 00298 -00329 -00173 -00028 -00323 -00565

(00403) (00354) (00255) (00396) (00361) (00300)Log Rain -1862 -2184 -0687 -1880 -2259 0189


(00001) (00001) (00001) (00001) (00001) (00002)Districts 00873 0125 00717 00819







20



(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion




(1) (2) (3) (4) (5) (6)Splits[t-1] -00213 00269 00473 -00652 -000444 -00506

(00550) (00683) (00777) (00726) (00840) (00974)Splits[t] 00396 -00154 -000796 00334 00579 0133

(00940) (0103) (0118) (00912) (00985) (0102)Splits[t+1] 0152 0230 0210 0000397 00486 -00262

(00870) (00963) (0110) (00943) (0110) (0131)Splits[t+2] 00103 00759 00688 -0182 -0173 -0194

(00737) (00902) (0115) (00714) (00873) (00990)Splits[t+3] -00564 -0151 -0111 -0158 -0139 -0125

(00609) (00680) (00789) (00641) (00764) (00843)Splits[t+4] 00398 0109 0192 00159 00665 00824

(00569) (00709) (00805) (00566) (00663) (00750)Log(Rain) -1653 -2273 -2413 -1656 -2265 -2410

(0148) (0214) (0220) (0147) (0215) (0222)Unemployment 00521 00602 00609 00526 00592 00584

(000872) (000968) (00112) (000872) (000969) (00110)




21









22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion










22










23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion











23




24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion





24

Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion


Tabl

e6

Def

ores

tatio

nan

dFi

res

(200

2-20

08)

Fore

sted

Isla

nds

Sum

ater

aK

alim

anta

nSu

law

esi

Papu

aFo

rest

edIs

land

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)Fi

res

-44

30

-2

949

-43

91

-1

533

-5

339

-15

65

-2

11

-4

86

(0

721

)(0

661

)(1

076

)(0

284

)(1

067

)(5

95e

-08)

(497

)(9

83)

Con

stan

t-2

930

4

-274

24

-142

37

-891

41

-42

59-1

4033

3-2

439

90-3

089

48(1

018

8)(1

905

6)(8

070

)(6

050

5)(5

209

)(1

8409

4)

1872

919

330

Yea

rFE

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Prov

FEN

oY

esY

esY

esY

esY

esY

esY

esR

20

600

080

70

930

079

60

792

081

30

810

080

7N

147

147

6328

4214

147

145

Fire

sC

onfid

ence

Lev

el80

8080

8080

8070

90M

ean

Fire

s10

797

1079

710

615

2782

822

11

331

417

399

549

1S

DF

ires

2064

020

640

1749

633

720

227

648

11

3022

211

841

Mea

nD

efor

este

d-6

673

9-6

673

9-8

393

8-1

5785

9-1

341

3-7

625

0-6

673

9-6

673

9S

DD

efor

este

d10

894

810

894

812

308

314

172

616

351

1196

93

1089

48

1089

48

Clu

ster

-rob

usts

tand

ard

erro

rsin

pare

nthe

ses

Dep

ende

ntva

riab

leD

efor

este

dis

aco

unto

fhow

man

ypi

xels

offo

rest

cove

rwer

elo

stin

agi

ven

year

fore

ach

prov

ince

Eac

hpi

xeli

s25

0x25

0msq

uare

Fire

sis

aco

untv

aria

ble

ofth

enu

mbe

roffi

res

perp

rovi

nce

Col

umns

1-2

incl

ude

allf

ores

ted

isla

nds

colu

mns

3-6

cons

ider

only

indi

vidu

alfo

rest

edis

land

san

dco

lum

ns7-

8va

ryth

efir

eco

nfide

nce

leve

l

25








26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion









26












(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion













(0161) (0163) (0183) (0181) (00939)Log Real Rubber[t-3] -1515 -1537 -1743 -2162

(0167) (0167) (0232) (0242)Log Real Rubber[t-6] -0176 -0188 -1226 -1416

(0144) (0145) (0236) (0260)Log Real Rubber[t-9] -1197 -1186 2814 2272

(0145) (0144) (0356) (0345)Log Real Rubber[t-12] 0178 0178 -1736 -1703




27



3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion




3 0236 0472 02086 0151 0302 01349 0045 0090 0040

12 0086 0172 0134Net Effect 0518 1036 0516






28



1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion




1 0240 0481 02122 -0220 -0440 -01943 0223 0447 01974 0254 0508 02245 -0380 -0759 -03356 0359 0719 0317





6 Conclusion



29





30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion






30

















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion


















31

















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion


















32

















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion


















33







34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion








34




35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion





35



36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion




36

























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion


























Village RoadData





Rupiah PalmPrices



USD RubberPrices






37





38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion






38

Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion


Tabl

eA

3C

orre

latio

nbe

twee

nnu

mbe

rofd

istr

icts

plits

and

pote

ntia

lcor

rela

tes

tofir

eac

tivity

Prov

ince

Nat

iona

l

Rai

nfal

lU

nem

ploy

men

tPa

lmSu

itabi

lity

Fore

st(2

000)

Inte

rest

Rat

ePa

lmO

ilPr

ice

Palm

Oil

Pric

eL

ag[t

+1]

Num

bero

fSpl

its0

001

-00

010

000

000

01

555

-16

31-1

633

(00

02)

(00

13)

(00

43)

(00

00)

(14

22)

(17

70)

(16

00)

[07

43]

[09

19]

[09

98]

[00

00]

[02

95]

[03

75]

[03

27]

All

colu

mns

are

OL

Sre

gres

sion

sof

the

num

bero

fspl

itsin

agi

ven

year

eith

erat

prov

ince

leve

l(C

olum

ns2-

5)or

natio

nally

(col

umns

6-8)

onth

eco

lum

nva

riab

les

and

aco

nsta

ntte

rmS

tand

ard

erro

rsar

ein

pare

nthe

ses

and

p-va

lues

are

insq

uare

brac

kets

39



Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion




Percentile 80 60 70 90 100 75 95


(00175) (00155) (00166) (00244) (00273) (00211) (00220)Log(Rain) -1965 -1799 -1858 -1804 -1758 -2147 -1772

(0390) (0342) (0371) (0517) (0548) (0482) (0583)Log(ForestCover) 0940 0873 0895 0993 1005 0997 1117



(00404) (00370) (00382) (00519) (00520) (00476) (00452)Splits[t] 00376 00390 00418 00308 00294 00338 00116

(00410) (00331) (00362) (00532) (00567) (00417) (00494)Splits[t+1] 00614 00465 00553 00644 00892 00639 00711

(00268) (00226) (00225) (00331) (00348) (00318) (00313)Splits[t+2] -00150 -00184 -00159 -00274 -00457 -00178 -00322

(00289) (00246) (00256) (00423) (00435) (00354) (00341)Log(Rain) -2106 -1890 -1990 -2302 -2082 -2499 -1662

(0425) (0346) (0380) (0596) (0628) (0516) (0596)Log(ForestCover) 1053 0929 0968 1155 1170 1113 1219




40
















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion

















(0343) (0354) (0389) (0391) (0234)Log Real Rubber[t-1] -1771 -1749 0989 -0319

(0217) (0218) (0245) (0254)Log Real Rubber[t-2] 1775 1767 0966 2485

(0272) (0275) (0612) (0676)Log Real Rubber[t-3] -2604 -2636 -5562 -6788

(0240) (0242) (0605) (0672)Log Real Rubber[t-4] 1042 1056 1101 1251

(0161) (0161) (0501) (0502)Log Real Rubber[t-5] -0992 -0996 1541 1550

(0160) (0160) (0449) (0461)Log Real Rubber[t-6] 000129 -000943 -0352 -0591



41







42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion








42



(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion




(1) (2) (3) (4) (5) (6)Districts 000408 000346 0000840 00313 00330 00154

(00195) (00217) (00379) (00209) (00243) (00398)Log Rain -2511 -2095 -2539 -1951



43



(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion




(1) (2) (3) (4) (5) (6)Splits [t-2] -000594 000954 00379 -00539 -000961 00476

(00821) (00835) (00674) (00674) (00744) (00633)Splits [t-1] 00416 00615 -000169 00209 00772 00216

(00297) (00330) (00567) (00300) (00314) (00535)Splits [t] 00301 00299 000904 00128 00136 000407

(00464) (00448) (00512) (00359) (00325) (00484)Splits [t+1] 00560 00558 00388 00549 00557 00350

(00360) (00342) (00428) (00315) (00281) (00406)Splits [t+2] -00189 -00196 -000740 -00226 -00230 000599

(00392) (00369) (00235) (00388) (00340) (00271)Log Rain -2494 -2497 -2120 -2502 -2530 -1984





44

Introduction





Data

Active Fire Data


Price Data

Additional Data

Methodology



Samples


Results


Main Results





Conclusion


Where There is Fire There is Haze: The Economic...

Documents

Transcript of Where There is Fire There is Haze: The Economic...