514-1060-1-PB

Economics and Finance of IndonesiaVol. 61 No. 3, December 2015 : 196-213p-ISSN 0126-155X; e-ISSN 2442-9260196

Oil Exploration Economics: Empirical Evidence from IndonesianGeological Basins I

Harry Patriaa,�, Vid Adrisonb

aEP EnergybFaculty of Economics and Business Universitas Indonesia

Abstract

Oil exploration has been subject to economic research for decades. Earlier studies of exploration modelsare mostly discussed the behavior of exploration at the macro-level analysis such as field, firm, region, andcontinental. This paper then focuses on the geological and economic factors that determine the well-drillingdecision at the micro-level using disaggregated panel data of 32 geological basins in Indonesia over theperiod of 2004–2013. This study shows that the number of drilled wells is determined significantly by the lagof success rate, lag of discovery size, lag of global oil price, and regional location of geological basin.Keywords: Drilling; Geological Variables; Economic Variables; Exploration

AbstrakEksplorasi migas telah menjadi subyek ekonomi dalam beberapa dekade. Studi-studi sebelumnya denganmodel eksplorasi, kebanyakan mengembangkan model Fisher (1964), secara umum dikelompokkanoleh persamaan yang menjelaskan respons eksplorasi pada tingkat makro menggunakan lapangan,perusahaan, wilayah, dan kontinental. Paper ini fokus pada analisis faktor-faktor geologi dan ekonomiyang menentukan tingkat sumur pemboran pada tingkat mikro menggunakan data panel dari 32 basindi Indonesia dalam periode 2004–2013. Hasil empiris menunjukkan bahwa tingkat sumur pemboranditentukan secara signifikan berdasarkan tingkat keberhasilan pemboran, ukuran temuan dan harga minyakpada tahun sebelumnya serta lokasi basin geologis.Kata kunci: Pengeboran; Variabel Geologi; Variabel Ekonomi; Eksplorasi

JEL classifications: L71; Q35

1. Introduction

Petroleum exploration has been subject to eco-nomic research for decades, but many studieshave been concentrated in the US with some stud-ies conducted for the United Kingdom and Norway.Earlier studies of exploration models, which aremostly based on Fisher (1964), have been gener-ally characterized by equations describing the be-havior of exploration at the macro-level analysis

IThis present paper is developed based on an academic the-sis at the Faculty of Economics and Business, Economics Pro-gram, Universitas Indonesia , which was nominated as the bestthesis in 2015.�Correspondence address: Gd. Pascasarjana Fakultas

Ekonomi Universitas Indonesia Lt.2 Kampus UI Depok, 16424.E-mail: [email protected].

using various levels of empirical analysis includ-ing field (Attanasi & Drew 1986), firm (Ghouri 1991;Forbes & Zampelli 2000), region (Fisher 1964; Er-ickson & Spann 1971; Pindyck 1978; Kolb, 1979),and continent (Mohn & Osmundsen, 2008).

On the other hand, some studies have been dis-satisfied with macro-level analysis in explainingthe exploration behavior, as Mohn & Osmundsen(2008) reported that many exploration articles ex-cept Ghouri (1991) and Iledare (1995) fell into thiscategory. An alternative approach is to view ex-ploration behavior from the company’s perspectiveand the base the econometric models on microdata. In fact there is a specific impact of geologicalbasins on the drilled wells. However, highly aggre-gated data relating to exploration behavior at thegeological basin have been limited due to availabil-

Economics and Finance of Indonesia Vol. 61 No. 3, December 2015

[email protected]

Patria, H. & Adrison, V./Oil Exploration Economics: Empirical Evidence... 197

ity of geological data. Thus, little attention has beenfocused on addressing this specific issue into a mi-cro analysis at the geological basin level.

There has been less number of empirical analysesthat use the geological basin level, which is consid-ered more representative to explain the pattern ofexploratory (Managi et al. 2005; Mohn & Osmund-sen 2008). Moreover, empirical studies on Indone-sia’s exploration have been relatively few in num-ber, which is probably caused by unavalability ofdata. Nasir (2011) regressed the drilling decisionusing time series data consisting of seismic sur-vey, discovery rate, exploration cost, oil price, andlevel of consumption. However, those parametershave insignificantly explained the drilling decisionand recommended further researches using disag-gregate analysis which include region and offshorearea. In fact, there is a specific impact of geologicalbasin characteristics on well-drilling decision.

The purpose of this paper is to examine the im-pact of natural geology on exploration decision,using Indonesian basin as a case study. The In-donesian exploration is an interesting case sinceIndonesia has been facing a new paradigm due tothe shift of exploration drilling from mature fieldsin the west to emerging fields in the east pronein terms of increasing national oil reserves. Afterthe oil period ended in 2001, many discoveriesof natural gas have been developed. These arelargely dominated by offshore fields at the east-ern region including South Makassar, Tomori, Ara-fura, Salawati, and Papuan Basin, with several ofthem consisting of offshore and deep water devel-opment. Moreover, USGS (2007) reported that thenumber of wildcat wells drilled each year in Indone-sia had been the largest among Asian countries,with 3,794 of 13,036 wells drilled over the periodof 1961–2001, 36 mil2/well of current growth in de-lineated prospective area per wildcat, and 0.149 ofrichness of total oil discoveries of total delineatedprospective area (USGS 2007).

Here we investigated the second paradigm ofshifting onshore to offshore drilling, in which theexploring companies are assumed to choose alevel of investment that maximizes the firm’s valueafter balancing expected revenues against therisks involved in exploration and the correspondingcosts. Combining the investment characterization,the study continued to an expression of the totalamount of wells drilled in each geological basin in

terms of estimates of anticipated returns and an-ticipated risk. Finally, we completed our analysiswith geological-economic variables into an empir-ical model.

This study makes two contributions towards under-standing the economics of oil and gas exploration.First, this study shows how the geological naturesignificantly affects drilling decision more than eco-nomic variables. This result has a similar conclu-sion to IPA’s (2012), which stated that the numberof exploration wells drilled had been relatively sta-ble during the fluctuation of oil prices which onlychanged the number of drilling projects. Second,our empirical analysis explicitly takes account ofoffshore percentage and basin’s region concerningthe success rate and discovery size to explain thenew exploration paradigm of "West to East Prone"and "Onshore to Offshore Drilling".

The paper is systematically organized as follows.First, the background and objective are briefly de-scribed. Section 2 provides an overview of earlierliteratures. An explanation of theoretical frameworkis offered in Section 3, before an empirical methodof well-drilling decision is derived in Section 4. Fur-thermore, Section 5 consists of the research data.Econometric results are presented and discussedin Section 6 before several concluding remarks areoffered in Section 7.

1.1. Stylized Facts

To focus the discussion, this paper presents themain stylized facts for a model of oil and gas ex-ploration. Figure 2 shows the geological and eco-nomic data of east prone on the left side and westprone on the right side. Geological data consistof success rate (S_c), oil discovery size (D_O),gas discovery size (D_G), and offshore percent-age (off). Economic data consist of oil price (p),spending per well (S_p), and value per well (V _w).

US Geology Survey (2007) reported that the firstbasin was discovered in North Sumatra in 1885.Since the first discovery, exploration drilling hassignificantly increased, most of which takes placein western Indonesia such as East Java Basin,South Sumatera Basin, Central Sumatra Basin,Northwest Java Basin, Sarawak Basin, and MalayBasin; while drilling in eastern Indonesia has beenconducted in Kutei Basin, Pamusiman Tarakan


Patria, H. & Adrison, V./Oil Exploration Economics: Empirical Evidence...198

Figure 1: Ultimate Recovery Factor of Oil and Gas and Prospective Locations in IndonesiaSource: SKK Migas (2015) and Indonesian Petroleum Association (2014)



Figure 2: Geological and Economic Data of Indonesian Geological BasinsSource: Author, using Stata 13.0 for the unpublished data of Indonesian exploration, accessed through the author’ account in SKK

Migas in February 2015

Figure 3: Indonesian Geological Basins in Western and Eastern of IndonesiaNote: Bengkulu to West Natuna are located in the western part of Indonesia (west prone), while the rest of the basins

are located in the eastern part of Indonesia (east prone).Source: US Geology Survey (2007)



Basin, Bintuni Basin, and Sulawesi Basin. The In-donesia Geological Basins consist of 2 (two) mainparts of geological region. Western Indonesia isrelatively underlain by mature fields and mainlyconsists of oil discoveries. On the other hand,the eastern region has been largely dominated byemerging fields and mainly consists of natural gasdiscoveries.

The diversity of geological regions qualitatively in-dicates the impact of natural geology to well-drillingdecision. Mature basins such as Central Suma-tra, South Sumatra, North West Java, and EastJava are characterized by well-established explo-ration models, minor technical challenges, andmaximized existing infrastructures. Therefore, ex-pected discovery rates are often relatively high forsuch areas. However, the average discovery size isusually limited due to natural declining effect. Onthe other hand, emerging basins such as SouthMakassar, Tomori, Salawati, and Papuan Basinsare distinguished by poorer comprehension of thegeology, technological challenges, and limitation ofinstalled infrastructures. Thus, the exploration riskis higher as indicated by relatively lower successrates, but the expected rewards are often relativelyhigh in discovery size.

In the model, oil prices influence exploration drillingdecision, which leads to discovery. On the otherhand, gas price is represented by oil price forsome reasons. National gas prices are generallycomprised of pipe gas and liquefied natural gas(LNG) prices. LNG prices, as a trading commodity,are generally regulated based on world oil price.Since LNG has been mostly exported to Japan,contracted prices are determined by Japan CrudeCocktail (JCC) in the equation of LNGPrice �α�JCC�β, wherein Îs represents a slope of priceand Îš cost of transportation. On the other hand,prices of pipe gas in certain fields may be differ-ent, which are subject to government regulation.Therefore, world oil prices are more appropriate inexplaining the profit return in short- and long-terminvestment.

This study uses an assumption of fixed traditionalinputs which consist of labor, capital, and inter-est rate. Existing literatures show that explorationmodels have used this assumption, excluding theimpact of labor, capital, and interest rate (Dahl &Duggan 1998; Iledare et al. 2000; Mohn & Os-mundson 2008). Exploration companies have em-

ployed exploration staff for the long term (Mohn &Osmundson 2008). A standby capacity has beenmaintained over the business cycle, almost inde-pendent of fluctuations in the oil price and explo-ration activity from year to year. From the explo-ration companies’ side, exploration activities aremoderately capital-intensive, as all capital equip-ment is hired for specific acitivities. I therefore hy-pothesized a subordinate role for traditional inputssuch as capital and labor in decisions concern-ing exploration activity. This means that fluctua-tions in exploration activity from one year to an-other are driven mainly by the oil price, as wellas the availability and quality of exploration op-portunities (Mohn & Osmundson 2008). Moreover,the role of traditional inputs has not been well ex-plained in the empirical studies of exploration be-havior (Dahl & Duggan 1998). Attempts have beenmade to include interest rates and user cost of cap-ital (e.q. Pindyck 1978; Peseran 1990), but theirrole is generally not justified by econometric ev-idence. Interest rate variables are therefore nor-mally not included in modern empirical explorationstudies (e.g., Iledare et al. 1999; Farzin 2001). Thestrongest argument for this simplification is that dy-namic optimisation is not well supported by previ-ous attempts to explain the economics of oil andgas exploration.

2. Literature Review

The combination of geological and economic vari-ables in empirical models of exploration datesback 40 years, with Fisher (1964) claimed to haveopened this field of research with his seminaleconometric studies of US oil and gas exploration.Fisher (1964) used a three-stage model with es-timating equations for total wildcat wells, successratio, and the average size of discovery for dif-ferent US Petroleum Administration Defense Dis-tricts (PADD) over the period of 1946–1955. Theexplanatory variables include oil prices, seismiccrews, and proxy variables for drilling costs. Hiscontribution is important to explain how economicvariables affect well-drilling decision. When eco-nomic incentives increase, not only the total num-ber of wells drilled goes up, but the average char-acteristics of the drilling prospects change becauseit now becomes worthwhile to drill wells with poorerprospect.



Since the pioneering work of Fisher (1964), a sub-stantial literature on oil and gas exploration hasdeveloped. In the 1970s, empirical analyses weremostly based on the aggregate level. Since themid-1980s, a growing number of studies have usedstate, regional, or firm-level data. There are clearadvantages to using micro-level data, since aggre-gation of data across distinctive geologic provincesmay obscure the effects of economic and policyvariables on the pattern of exploratory activities(e.g. Pindyck 1978). Concentrating on gas explo-ration, Pindyck (1978) estimated a similar modelon a broader data set, with quite different resultsfrom those of Fisher (1964) and Erickson & Spann(1971). Kolb (1979) focused on oil-prone districtsin a slightly more disaggregated approach. Theseearly Fisher models had a simple structure thatlargely could be justified based on economic fun-damental principles.

A growing number of studies have used state,regional, or firm-level data since the mid-1980s(Managi et.al. 2005). However, the most commonperspective has been based on aggregate data forregions, countries, or groups of countries. More-over, there are clear advantages to using microfirm-level data, since aggregation of data acrossdistinctive geologic provinces may obscure the ef-fects of economic and policy variables on the pat-tern of exploratory activities (e.g., Pindyck 1978).Although the lack of data at the field level has beenviewed as a major obstacle to carrying out dis-aggregated analysis, field-level behavior has beenconsidered too erratic to model successfully in em-pirical model (Attanasi 1979).

Exploration models have been mostly developedon oil price rather than natural gas price for somereasons (Kolb 1979; Ghouri 1991; Dahl & Dug-gan 1998; Forbes & Zampelli 2000; 2002; Mohn& Osmundson 2008). Most studies found that nat-ural gas price was insignificant to explain the well-drilling decision (Dahl & Duggan 1998; Kolb 1979;Mohn & Osmundson 2008; Boyce & Nostbakken2011). Moreover, natural gas prices in Indonesiahave been regulated by the government insteadof representing its real production costs. For somecases, we would even find different prices of nat-ural gas sourced from the same region. Differenti-ation of natural gas price in the same region maycreate error in determining which natural gas priceto be selected to represent the basin. Thus, thispaper prefers to select oil price over natural gas

price.

2.1. Theoretical framework

Petroleum exploration aims to find new reservesthrough comprehensive process that consists ofGeology & Geophysics Study (G&G) and wildcatdrilling in potential areas1. The present frameworkof petroleum exploration is developed based ontheoretical economics. My theoretical point of de-parture is a simple model of geological and eco-nomic variables associated with petroleum explo-ration. Based on the pioneering work of Fisher(1964) and recent literatures, I developed a theo-retical framework to explain the impact of depen-dent variables that generally consist of geologicalvariables: success rate, discovery size, region, andlocation. On the other hand, the economic vari-ables were defined as cost of production (spendingper well) and oil price. Those dependent variableswere used to explain the impact on the number ofwells drilled in certain basins. Adopting previousliteratures to define the geological and economicvariables, the framework is simply illustrated in Fig-ure 4.

This paper modified a block diagram of the econo-metric model of Challa (1974) by emphasizingon petroleum exploration and maximizing the ex-pected return considering the estimated risk of ex-pected return. Adopting the work of Mohn & Os-mundsen (2008), it formulated a production func-tion Y � F pZ,W q � WαZβ , wherein W is definedas number of wells drilled, Z represents geologi-cal variables with an assumption of decreasing re-turn to scale, α�β 1. Eventhough the theoreticalmodel present an analogy of Cobb-Dauglass, how-ever α and β cannot be indirectly interpreted by aconcept of degree of substainability. In this specificformula, Îs represents a level of drilling intensity ofan exploration company while β represents a levelof intensity of geological variables which are purely

1Petroleum drilling consists of 3 (three) types classifiedbased on the objectives and stages of drilling: wildcat drilling,delineation drilling, and development drilling. Wildcat drillingaims to ensure petroleum reserves in geological basins. Wild-cat drilling becomes an essential stage due to its higher invest-ment and risk compared to the subsequent stages. Delineationdrilling tends to determine the reserves based on reservoir char-acteristics. Development drilling aims to increase the productionrate and enhance the recovery factor.



Figure 4: Conceptual Framework for the Indonesian Exploration ModelSource: Author

exogenous as given as per geological basins. Asoil prices are determined by global supply and de-mand, its characteristics become more exogenous.

The net benefits of finding reserves is the expected(E) discounted present revenues of oil and gasproduction minus total cost of exploration. Subse-quently, total revenue is formed by both oil priceand production funtion. Furthermore total cost ofexploration is explained by the function of C �φwW � φwZ, wherein φw is the specific cost ofwell exploration. The choice of number of explo-ration wells to drill depends upon the future pricesexpected price at time, given price pptq is pepsq �pptqeps�rqt.

The objective is then to maximize the correspond-ing profit function:

π �n

i�1

rpF pZi,Wiqeps�rqt � φwWie

�rt �

φwZie�rts (1)

The objective subsequently to maximized by differ-entiation of equation (1) as follows:

dπ

dW�

n

i�1

�p.eps�rqt

dY

dW� φwe

�rt

�(2)

In terms of maximizing profit, the first order condi-tion is given by:

n

i�1

�p.eps�rqt

dY

dW� φwe

�rt

�� 0 (3)

p.eps�rqtdY

dW� φwe

�rt (4)

Substituting dYdW � α.Y

W into equation (4):

p.eps�rqt�α.Y

W

�� φwe

�rt (5)

p.estα.Y � φwW (6)

Rearranging the equation, the final equation isgiven by:

W �pest.α.Y

φw(7)

Substituting Y � fpZ,W q � WαZβ into the equa-tion:

W �pest.α.WαZβ

φw(8)

Therefore, the corresponding a number of wellsdrilled is explicitly given by:

W 1�α �pest.α.Zβ

φw(9)

Then, the final equation of wells drilled given by:

W �

�pest.α.Zβ

φw

11�α

(10)

With regard to the final equation, dWdP ¡ 0 indicatesthat the expected price of oil has a positive impacton the number of wells drilled, whereas dW

dφw 0

indicates that the cost of exploration has a negativeimpact on the number of wells drilled.



3. Method

3.1. Empirical method

This section elaborates the empirical method toanalyze the well-drilling function. Adopting Fisher(1964), method used in this study provides geolog-ical variables that consist of success rate, averagediscovery size of oil, and average discovery size ofgas. To specifically address the impact of geologi-cal basin maturity in Indonesia, I constructed sev-eral proxies as per region and offshore, while theeconomic variables consist of oil price and explo-ration cost as denoted by spending per well. By in-tegrating those geological and economic variables,the corresponding well-drilling model is specificallynotated by:

Wit �αi � β2iDOi,t�1� β3iD

2O,t�1 � β4iDGi,t�1looooooooooooooooooooooomooooooooooooooooooooooon

Geological Variables

�β5iD2G,t�1 � β6iSCi,t�1looooooooooooooomooooooooooooooon


�β7iOfi,t�1 � β8iRegilooooooooooooomooooooooooooonGeological Variables

� β9i lnPt�1 � β10i lnSpit�1 � eitloooooooooooooooooooomoooooooooooooooooooonEconomic Variables

(11)

Wit represents the number of wells drilled andDOi,t�1 and DGi,t�1 represent the lag of discov-ery size of oil and gas, respectively. The geologicalvariable of SCi,t�1

indicates the lag of success rate.Moreover, the proxies of lag of offshore explorationand region are represented by Ofi,t�1 and Regi,respectively. The economic variables consist of lagof oil price Pt�1 and spending per well SCi,t�1 .Therefore, β2i to β8i represent the coeffcient of ge-ological variables for geological basin i, whereasβ9i and β10i represent the coeffcient of economicvariables, αi and eit indicate respectively the inter-cept and error of empirical model.

This paper developed lag variables on explorationmodel as well as expected to handle the poten-tial endogeneity. Those variables consist of depen-dent variables, wells drilled, where geological andeconomic variables are actually exogenous. How-ever, providing lag variables at different year hasnot yet represented the field drilling decision. For

some cases, firms can response on monthly ba-sis to change their drilling plan. When the existingdrilling has found only dry holes with lower successrate, at certain level they could reconsider their de-cision to prevent higher cost and risk.

This paper provides panel data that combine bothcross sections and time series. The cross sectionsare represented by 32 geological basins in Indone-sia over the last ten years as time series data. An-nual data for the model variables of 2003 to 2013were obtained directly or derived from the officialdatasets of (i) Woodmac, and (ii) the Task Forcefor Upstream Oil and Gas of Republic of Indonesia(SKK Migas). For qualitative analysis, descriptivestatistics of those data are presented in Table 1.

The exploration model can be estimated by econo-metric methods, taking proper account of charac-teristics of the data-generating process. Each welldrilled is an integer, while discovery the explana-tory variables are non-integer data, continue andcategorical. Theoretically, one of regression mod-els in this specific case is Poisson regression. Theregression model analyzes the distribution with in-tensity paramete µ that is determined by explana-tory variables. One of assumptions that should besufficient is equidispersion, wherein the variancevalue of response variable Y at X � x shouldbe equivalent to the mean value, formulated asV arpY |xq � EpY |xq � µ.

With regard to the descriptive data, wells drilledhave a variance of 8.2737 and mean of 1.5770.The same value of discovery size, oil price andspends per well. Consequently, the characteristicsof data will potentially create an overdispersion.The overdispersion can cause standard error foreach parameters tend to be lower. For example, ifwe keep using the Poisson regression, the insignifi-cance variable could be interpreted as significanceone. As the implication, result summary could beinappropriate. In this overdispersion case, nega-tive binomial is the appropriate method of Poissonmodel.

Poisson and negative binomial models are de-signed to analyze count data. The "rare events"of the nature of wells drilled are controlled in theformulas of negative binomial regression. Further-more, Poisson and negative binomial regressionmodels differ in terms of their assumptions of theconditional mean and variance of the dependentvariable. Poisson models assume that the condi-



Table 1: Geological and Economics Data of Indonesia during 2004–2013

Variable Mean Standar Deviation Minimum Maximum

Wells drilled W 1.5771 2.876407 0 17Discovery size of oil DO 1.4003 5.531158 0 60Discovery size of gas DG 2.3485 16.17237 0 248.89Success rate Sc 0.1500 0.250464 0 1Offshore variable Off 0.1833 0.360601 0 1Oil Price of ICP P 91.4696 19.69355 58.015 118.39Spends per well Sp 70.50353 76.5068 0.156694 423,769

Source: Author, using STATA 13

Table 2: Statistic Descriptive of Geological and Economic Variables of Petroleum Exploration in Indonesiaduring 2004–2013

Variable Mean Variance Skewness Kurtosis

Wells drilled W 1.5771 8.273717 2.713073 11.53123Discovery size of oil DO 1.4003 16.48744 5.257498 44.76493Discovery size of gas DG 2.3485 262.7516 8.789477 88.25679Success rate Sc 0.1500 0.0627326 1.730904 5.366517Offshore variable Off 0.1833 0.1300334 1.612934 3.806472Oil Price of ICP P 91.4696 387.836 -0.3371503 1.775671Spends per well Sp 70.50353 5853.29 2.530098 10.63421

Source: Author, using STATA 13

Figure 5: Wells Drilled Data in Indonesian BasinsSource: Author, using STATA 13



tional mean and variance of the distribution areequal. Negative binomial regression models do notassume an equal mean and variance and particu-larly correct for overdispersion in the data, whichis when the variance is greater than the condi-tional mean. Due to the nature of the distribution ofdependent variable, I chose the negative binomialmodels. For the result, generally negative binomialmodels are interpreted by an incident rate ration. Itis a ratio based on the rate or incident of counts.

3.2. Empirical model

This section elaborates the result and discussionin determining the explanatory variables of explo-ration decision. The model was estimated by nega-tive binomial regression both fixed effects and ran-dom effects in order to check the model stability. Allresults were interpreted through coefficient valueand significance of explanatory variables. This sec-tion explores the comparison between result es-timation and hypothesis in theoretical model. Tosupport the result, this section was completed bysupporting data and information to compare the re-sult estimation and its real condition.

The empirical model was developed to identify thedepletion effect on oil exploration to represent thematurity of geological basins in Indonesia. The de-pletion effect was identified by a square of dis-covery size of oil. In addition, I also completed asquare of discovery size of gas in order to com-pare the results. Furthermore, I constructed an in-teraction variables consisting of region proxies andspending per well. For oil prices, I completed themodel with local and global oil price as expectedprice released by OPEC to identify which one thatshould be selected as the main reference in drillingdecision. Integrating all modification, the empiricalmodel is mathematically given by:

Wit �αi � β2iDOi,t�1 � β3iD2O,t�1 � β4iDGi,t�1looooooooooooooooooooooomooooooooooooooooooooooon


�β5iD2G,t�1 � β6iSCi,t�1looooooooooooooomooooooooooooooon


�β7iOfi,t�1 � β8iRegilooooooooooooomooooooooooooonGeological Variables

� β9i lnPt�1 � β10i lnSpit�1 � eitloooooooooooooooooooomoooooooooooooooooooonEconomic Variables

(12)

Wit represents the number of wells drilled andDOi,t�1

and, DGi,t�1represent the lag of discov-

ery size of oil and gas respectively. Meanwhile,other geological variable of SCi,t�1 indicates thelag of success rate. Depletion effects for oil andgas are respectively represented by D2

O,t�1 andD2G,t�1, while the proxies of lag of offshore explo-

ration and region are represented by Ofi,t�1 andRegi respectively. Economic variables consist oflag of oil price Pt�1 and spending per well SCi,t�1

.Therefore, β2i to β8i represent the coeffcient of ge-ological variables for geological basin i, whereasβ9i and β10i represent the coeffcient of economicvariables. αi and eit indicate respectively the inter-cept and error of empirical model.

4. Results and Analysis

The estimation results for discovery size of oil andgas showed positive coefficients and were statis-tically significant with respect to number of wellsdrilled. The coefficients for lag of discovery size ofoil and gas were statistically significant for all re-gression models, both fixed and random effects.Thus, those results were quite stable for all re-gression models. Those coefficients had a positivesign as I would expect in the hypothesis and the-oretical model. Furthermore, I obtained similar re-sults that supported the existing literatures on ex-ploration model (Fisher 1964; Kolb 1979; Mohn &Osmundson 2008)2. The significance contributesto explain why Nasir’s study (2011) was not signif-icant with respect to the number of wells drilled inIndonesia. Thus, the disaggregate analysis at geo-logical basin level can explain the effect of discov-ery size in determining the number of wells drilled.

The result is consistent to represent the real condi-tion where exploration companies review the pre-vious geological data such as discovery size andsuccess rate for future investment decision. Interms of analyzing depletion effect, the coefficientof square discovery size of oil indicates the sig-nificance with respect to number of wells drilled.

2Detail summary of exploration models can be seen in Table5.



The results support the real condition where oil ex-ploration has entered a mature stage since it hasbeen exploited for decades. Thus, it has an inverserelation with respect to wells drilled. On the otherhand, the discovery size of gas tends to be higherthan the discovery size of oil. Consequently, it haspositive relation with respect to wells drilled. Theresult is supported by the fact which cumulative re-coverable gas discoveries in Figure 1. Cumulativerecoverable for gas is higher than oil in terms ofincreasing gas discoveries in Indonesia.

The estimation result for success rate indicatedpositive coefficient and was statistically significantwith respect to number of wells drilled. The coeffi-cients for lag of success rate were statistically sig-nificant for all regression models, both fixed andrandom effects. Thus, those results were quite sta-ble for all regression models. Those coefficientshad a positive sign as I would expect in hypothe-sis and theoretical model. I obtained similar resultsthat supported the existing literatures on explo-ration model (Boyce & Nostbakken 2011). The sig-nificance contributes to explain why Nasir’s study(2011) was not statistically significant with respectto the number of wells drilled in Indonesia. Thus,the disaggregate analysis at geological basin levelcan explain the effect of success rate in determin-ing the number of wells drilled.

In addition to being consistent with respect to hy-pothesis and theoretical model, the result repre-sents the real fact in petroleum exploration. Ex-ploration companies tend to invest in proven fieldswhich have higher success rate and discovery size.Geological data show that exploration drillingswhich consist of 2 wells drilled in Nam Con Son in2011, 1 well drilled in North New Guinea in 2007,2 wells drilled in Pasir-Asem Asem in 2006, and 2wells drilled in West Natuna in 2005 had found nosignificant discoveries, or generally known as dryholes.

The estimation result for offshore variable indi-cated positive coefficient but was statistically in-significant with respect to the number of wellsdrilled. The coefficients for lag of success rate werestatistically insignificant for all regression models,both fixed and random effects. Thus, those resultswere quite stable for all regression models. Thosecoefficients had a positive sign as I would expectin the hypothesis and theoretical model. The in-significance is caused by distribution of data in off-

shore wells being lower than in onshore wells. Theresult is in line with the fact that offshore wellshave been dominated by emerging fields in east-ern part of Indonesia, whereas onshore wells havebeen dominated by mature fields in western partof Indonesia. SKK Migas (2015) reported that off-shore drilling had been developed intensively ineastern areas of Indonesia such as Arafura Basin,Kutei Bain, Bintuni Basin, South Makassar Basin,Tarakan Basins, and Pasir-Asem Asem Basin. In-donesia has developed deep water development inEast Kalimantan which has relatively high gas re-serves. In western areas, Indonesia had developedoffshore drillings in areas such as Northwest Java,West Natuna-Penyu Basin, and East Java Basin.Therefore, I concluded that even though the lag ofoffshore had a positive coefficient with respect tothe number of wells drilled, the coefficient was sta-tistically not significant for all specification models.

The estimation result for region proxies indicatedpositive coefficient and was statistically significantwith respect to the number of wells drilled. The co-efficients for lag of success rate were statisticallysignificant for all regression models, both fixed andrandom effects. Thus, those results were quite sta-ble for all regression models. Those coefficientshad a positive sign as I would expect in the hy-pothesis and theoretical model. The result is illus-trated with distribution of data in Figure 6. Theresult is in line with the fact that those emergingand mature fields are mostly located respectivelyin eastern and western part of Indonesia. Explo-ration companies tend to drill intensively in easternIndonesia, which has relatively had higher successrate and discovery size.

Meanwhile, region proxies show that eastern areashave become more interesting than western areasfor exploration companies. Eastern areas consistof emerging basins which have higher discoverysize and success rate than mature basins mostlylocated in western areas. However, offshore vari-ables are not statistically significant to well-drillingdecision due to onshore development having beenestablished longer than offshore development. Fur-thermore, offshore exploration tends to have higherrisk in terms of exploration cost. The results showthe impact of spending per well and region on in-teraction variable for emerging basins which aremostly located in the eastern part of Indonesia.This indicates that interaction variable has an in-verse relationship with respect to wells drilled.



Table 3: Estimation Result of Wells Drilled Model

Variables Wells drilledFixed Effects Random Effects

(1) (2) (3)a (4)b

Lag of discovery size of oil mmboe 0.0226* 0.106*** 0.0243** 0.110***(0.0125) (0.0295) (0.0117) (0.0282)

Lag of discovery size of oil2 mmboe -0.00164** -0.00170***(0.000646) (0.000623)

Lag of discovery size of gas mmboe 0.00842** 0.0262** 0.00907** 0.0269**(0.00391) (0.0125) (0.00386) (0.0123)

Lag of discovery size of gas2 mmboe -0.000115 -0.000118(7.79e-05) (7.74e-05)

Lag of success rate % 1.191*** 0.746** 1.186*** 0.744**(0.294) (0.330) (0.288) (0.320)

Lag of offshore variable % 0.174 0.149 0.144 0.141(0.245) (0.244) (0.242) (0.236)

Region proxies east = 1 0.537*** 1.021*** 0.556*** 1.018***(0.202) (0.280) (0.201) (0.275)

Lag of spending per well M USD/well -0.00200 -9.46e-05 -0.00226 -0.000213(0.00171) (0.00159) (0.00171) (0.00158)

Region* Lag of spending per well M US$/well -0.0121** -0.0121**(0.00595) (0.00588)

Lag of Oil Price OPEC USD/barrel 0.00711** 0.00703** 0.00690** 0.00681**(0.00307) (0.00300) (0.00303) (0.00294)

Lag of ICP USD/barrel 0.00390 -0.00128(0.0129) (0.00679)

Constant -2.439*** -2.556** -2.171*** -2.101***(0.587) -1.003 (0.424) (0.585)

ln _r 18.55417 17.01599ln _s 19.8389 18.11373r 1.14e+08 2.45e+07s 4.13e+08 7.36e+07Observations 279 279 279 279Number of t 9 9 9 9Log-likelihood -377.96878 -366.18769 -412.48106 -399.7523

Note: Interpretation of result using incident rate ratio can be seen in attachment.Note: Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1Note: a Likelihood-ratio test vs. pooled: chibar2(01) = 2.5e-05 Prob>=chibar2 = 0.498Note: b Likelihood-ratio test vs. pooled: chibar2(01) = 2.9e-08 Prob>=chibar2 = 0.500Note: Akaike’s information criterion (AIC) = 841.5047Note: Bayesian information criterion (BIC) = 917.7601Source: Author, using STATA 13



Figure 6: Distribution of Wells drilled in Eastern and Western Part of IndonesiaSource: Author, using STATA 13

Estimation result for oil price indicated positive co-efficient and was statistically significant with re-spect to the number of wells drilled. The coeffi-cients for lag of global price were statistically sig-nificant for all regression models, both fixed andrandom effects. However, the coefficients for lagof local price were statistically insignificant for allregression models, both fixed and random effects.Thus, those results were quite stable for all regres-sion models. Those coefficients of global oil pricehad a positive sign as I would expect in the hy-pothesis and theoretical model. However local oilprice failed to explain the wells drilled in Indone-sia. In reality, exploration companies exercise theprofitability of investment using the global price. Onthe other hand, local price does not purely repre-sent market price because it has been regulatedby the government. Government tends to regulateoil price by considering various factors such assubsidy and economic condition. Therefore, globalprice is the main reference for many explorationcompanies in considering the number of wells tobe drilled.

Estimation result for exploration cost, spending perwell, indicated negative coefficient and was statis-tically insignificant with respect to the number ofwells drilled. The coefficients for lag of spendingper well had an inverse relation with respect towells drilled in all regression models, both fixed andrandom effects. Thus, those results were consis-tent for all regression models. Those coefficients

had a negative sign as I would expect in the hy-pothesis and theoretical model, representing thefact that exploration companies actually considerexploration cost in terms of the profit return. The re-sults show the impact of spending per well and re-gion in interaction cost for emerging basins whichare mostly located in the eastern part of Indone-sia. This indicates that interaction variable has aninverse relationship with drilling cost, although therelationship was not statistically significant in theirestimations.

Estimation result for interaction cost of spendingper well and region had an inverse relation and wasstatistically significant with respect to the numberof wells drilled. Those coefficients were statisticallysignificant for all regression models, both fixed andrandom effects. Those coefficients had an inverserelation as they consist of region and spending perwell which had an inverse relation with respect tothe number of wells drilled. In reality, the result isin line with the trade-off that exploration companieshave in terms of drilling in the frontier or emergingfields with higher exploration cost or drilling in ma-ture fields with lower exploration cost.

5. Conclusion

This paper developed an empirical model of explo-ration economics for oil and gas at geological basin



level in Indonesia during 2003–2013. The resultsshowed that geological and economic variableshad significant impact with respect to the numberof wells drilled. Wells drilled were explained sig-nificantly with positive sign by lag of success rate,lag of discovery size of oil, lag of discovery size ofgas, and region of geological basin. On the otherhand, offshore proxies were statistically insignifi-cant to determine the number of wells drilled due tothe fact that onshore drillings have been intensivelydrilled longer than offshore drilling. As a main con-tribution, I find those significant variables to deter-mine the number of wells drilled in Indonesia andexplain why Nasir’s study (2011), which as far asI know is the only exploration study in Indonesiawhich uses time series data, has failed to find ge-ological variables that significantly determine thenumber of wells drilled. This occurred because ofthe different level between the aggregate and dis-aggregate data.

For the economic variables, wells drilled were ex-plained significantly with positive sign by the globaloil price. The spending per well even had an in-verse relation as the expected hypothesis and the-oretical model, although it failed to explain the wellsdrilled. The interaction cost of spending per welland region had an inverse relation and was statisti-cally significant with respect to the number of wellsdrilled. The results support the fact that explorationcompanies have a trade-off in terms of drilling infrontier or emerging fields with higher explorationcost or mature fields with lower spending per well.

References

[1] Attanasi, E 1979, ’The nature of firm expectations inpetroleum exploration’, Land Economics, vol. 55, no. 3, pp.299–312.

[2] Attanasi, ED & Drew, LJ 1986, ’Field Expectation andthe Determinants of Wildcat Drilling’, Southern EconomicJournal (pre-1986), vol. 44, no. 1, pp. 53–67.

[3] Boyce, JR & Nostbakken, L 2011, ’Exploration and Devel-opment of U.S. Oil and Gas Fields, 1955–2002’, Journal ofEconomic Dynamics & Control, vol. 35, pp. 891–908 .

[4] Challa, K 1974, ’Investment and Return in Exploration andthe Impact on Oil and Natural Gas Supply’, Energy Labo-ratory Report. No. MIT-EL 74-016, MIT Energy Lab.

[5] Dahl, C & Duggan, TE 1998, ’Survey of Price Elastici-ties from Economic Exploration Models of US Oil and GasSupply’, Journal of Energy Finance & Development, vol 2,pp. 129–169.

[6] Erickson, EW & Spann, RM 1971, ’Suply in a RegulatedIndustry: the Case of Natural Gas’, Bell Journal of Eco-

nomics and Management Science, vol 2, no. 1, pp. 94–121.

[7] Farzin, Y 2001, ’The Impact of oil price on additions to USproven reserves’, Resources and Energy Economics, vol.23, no. 3, pp. 271-292

[8] Fisher, FM 1964, Supply and Costs in the U.S. PetroleumIndustry: Two Econometric Studies, Johns Hopkins Press,Baltimore, MD

[9] Forbes, KF & Zampelli, EM 2000, ’Technology and the Ex-ploratory Success Rate in the U.S. Offshore’, The EnergyJournal, vol. 21, no. 1, pp. 109–120.

[10] Forbes, KF & Zampelli, EM 2002, ’Technology and the ex-ploratory success rate in the U.S. onshore’, The QuarterlyReview of Economics and Finance, vol. 42, no. 2, pp. 319–334.

[11] Ghouri, SS 1991, ’World Oil and Gas Exploration Treands:A Comparative Study of National and US Private Com-panies’, Ph.D. Thesis, Colorado School of Mines, Golden,CO.

[12] Iledare, OO 1995, ’Simulating the effect of economic andpolicy incentives on natural gas drilling and gross reserveadditions’, Resources and Energy Economics, vol. 17, no.3, pp. 261–279.

[13] Iledare, OO, Pulsipher, A & Bauman, R 1999, ’Effects ofan increasing role for independents on petroleum resourcedevelopment in the gulf of Mexico OCS region’, The En-ergy Journal, vol. 16, no. 2, pp. 59–76.

[14] Iledare, OO, Pulsipher, AG & Olatubi, WO 2000, ’ModelingPetroleum Productivity in the U.S. Gulf of Mexico OuterContinental Shelf (OCS) Region’, Energy Studies Review,vol. 9, no. 2, pp. 86–95.

[15] [IPA] Indonesian Petroleum Association 2012, ’The Roleof Exploration in Meeting Indonesia’s Future Oil and GasNeeds’, presentation slide on ESDM-IPA Exploration Fo-rum July 2012.

[16] [IPA] Indonesian Petroleum Association 2014, ’ImprovingOil and Gas Investment Climate in Indonesia’, presenta-tion slide on Press Conference 2014 IPA Annual GeneralMeeting.

[17] Kolb, JA 1979, An econometric study of discoveries of nat-ural gas and oil reserves in the United States, 1948 to1970, Arno Press, New York.

[18] Managi, S, Opaluch, JJ, Jin, D & Grigalunas, TA 2005,’Technological Change and Petroleum Exploration in theGulf of Mexico’, Energy Policy, vol. 33, pp. 619–632.

[19] Mohn, K & Osmundsen, P 2008, ’Exploration Economicsin a Regulated Petroleum Province: The Case of the Nor-wegian Continental Shelf’, Energy Economics, vol. 30, pp.303–320.

[20] Nasir, A 2011, ’Proyeksi Supply melalui Analisa KegiatanEksplorasi dan Penemuan Minyak dan Gas Bumi di In-donesia’, Thesis, Faculty of Economics and Business, Uni-versitas Indonesia.

[21] Peseran, MH 1990, ’An Economic Analysis of Explorationand Extraction of Oil on the UK Continental Shelf’, OxfordInstitute for Energy Studies. EE7.

[22] Pindyck, RS 1978, ’The Regulatory Implications of ThreeAlternative Economic Supply Models of Natural Gas’, TheBell Journal of Economics and Management Science, vol.5, pp. 633–645.

[23] SKK Migas 2015, Laporan Tahunan SKK Migas 2014.Available from: <www.skkmigas.go.id>. [14 February2015].

[24] [USGS] U.S. Geological Survey 2007, Statisctics ofPetroleum Exploration in the World Outside the United


www.skkmigas.go.id


Figure 7: Empirical Results Incident Rate RatioNote: Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1.

Source: Author, estimated using Stata 13 and Woodmac Data (2015)


Patria, H. & Adrison, V./Oil Exploration Economics: Empirical Evidence... 211Ta

ble

4:S

um

mar

yo

fE

xplo

rati

on

Mo

del

-(P

art

1)

No

Ref

eren

siD

ep.V

aria

bel

Inde

pend

entV

aria

bles

R2

Geo

logi

cal

Eco

nom

icA

dditi

onal

/Pro

xies

1F

ishe

r(1

964)

1W

wS

EIS

MIC

+*,

O/W

ws(

-1)+

*,W

ws/

Ww

(-1)

-*,G

/Ww

s(-1

)-*

Poi

ldefl

ated

WP

I+*

0,84

Sam

ple:

46–5

5P

AD

D1-

5,C

TW

ws/

Ww

Ww

s/W

w(-

1)+

*,O

/Ww

s(-1

)-*,

G/W

ws(

-1),

Dep

t(-1

)-*,

Sei

smic

+*

Poi

ldefl

ated

WP

I0,

71

O/W

ws

O/W

ws(

-1),

Ww

s/W

w(-

1)+

*,G

/Ww

s(-

1)+

*P

oild

eflat

edW

PI-

*0,

85

G/W

ws

O/W

ws(

-1),

Ww

s/W

w(-

1)+

*,G

/Ww

s(-

1)P

oild

eflat

edW

PI-

*0,

32

Ww

Sei

smic

+*,

O/W

ws(

-1)+

*,W

ws/

Ww

(-1)

-*,G

/Ww

s(-1

)-*,

Dep

th(-

1)-*

Poi

ldefl

ated

bydr

illco

st+

*D

*TX

shut

dow

n,D

*non

Txs

hutd

own

+*

0,91

Ww

s/W

wW

ws/

Ww

(-1)

+*,

O/W

ws(

-1)+

*,G

/Ww

s(-1

),D

epth

(-1)

,Sei

smic

+*

Poi

ldefl

ated

bydr

illco

sts

0,72

O/W

ws

O/W

ws(

-1)+

*,W

ws/

Ww

(-1)

+*,

G/W

ws(

-1)-

*P

oild

eflat

edby

drill

cost

-*0,

86

G/W

ws

O/W

ws(

-1),

Ww

s/W

w(-

1)+

*,G

/Ww

s(-

1)P

oild

eflat

edby

drill

cost

-*0,

23

Sei

smic

Sei

smic

(-1)

+*

4D*P

AD

Dso

me*

,D

*Txs

hut

dow

non

Tx,

D*T

Xsh

utdo

wn+

*0,

97

2E

ricks

on&

Spa

nn(1

971)

1W

wp

Dep

th(-

1)+

*,W

wps

/Ww

p(-1

)-*,

Ww

larg

e/W

wU

S+

*,D

eflat

edP

oil+

*,D

eflat

edP

gas

D*P

AD

Dso

me*

,D

*PA

DD

som

e*,

D*T

Xsh

utdo

wn,

D*N

onT

Xsh

utdo

wn

0,97

Sam

ple:

46–5

5P

AD

D1-

5,C

TW

wps

/Ww

pD

eflat

edP

oil,

Defl

ated

Pga

sD

*PA

DD

,D

*TX

shut

dow

n,D

*Non

TX

shut

dow

n0,

8

O/W

wps

Ww

ps/W

wp(

-1)

Defl

ated

Poi

l,D

eflat

edP

gas,

D*P

AD

Dso

me*

,D

*TX

shut

dow

n-*,

D*N

onT

Xsh

utdo

wn

0,89

G/W

wps

Ww

ps/W

wp(

-1)

Defl

ated

Poi

l,D

eflat

edP

gas,

D*P

AD

Dso

me*

,D

*TX

shut

dow

n-*,

D*N

onT

Xsh

utdo

wn

0,6

G/W

xsW

xs/W

x(-1

)D

eflat

edP

gas+

*,D

eflat

edP

oil

3D

*Dis

tsom

e*

0,27

3A

ttana

si(1

979)

1W

w°

Ww

Exp

ecet

edgr

oss

profi

tno

tdi

s-co

unte

d+*,

$val

ueof

disc

over

ies(

0),

and

(01)

to(-

10)+

*

0,92

Sam

ple:

Den

ver

Bas

in49

–73

4K

olb

(197

9)1

OP

gas(

nr),

Poi

l(nr)

,WP

I(nr

)D

*reg

iona

llan

din

put+

*0,

70S

ampl

e:46

–69

Gna

Pga

s(nr

),P

oil(n

r),W

PI(

nr)

D*r

egio

nall

and

inpu

t+*

0,78

8U

SG

ulfR

egio

nsW

wW

ws/

Ww

(-1)

,O

/Ww

s(-1

),G

/Ww

s(-1

)-*,

Dep

th-*

,Sei

smic

+*,

Poi

ldefl

ated

WP

I+*

0,75

46–5

87

Dis

t,C

TW

ws/

Ww

O/W

ws(

-1),

Ww

s/W

w(-

1)+

*,G

/Ww

s(-

1)+

*,D

epth

,Sei

smic

+*,

0,65

48–7

07

Dis

t,C

TO

/Ww

sO

/Ww

s(-1

)+*,

Ww

s/W

w(-

1),

G/W

ws(

-1)

Poi

ldefl

ated

WP

I-*

Poi

ldefl

ated

WP

I+*

0,5

G/W

ws

O/W

ws(

-1),

Ww

s/W

w(-

1)+

*,G

/Ww

s(-

1)+

*P

oil

defla

ted

WP

I,P

gas

defla

ted

byW

PI

0,30

Ww

Poi

lde

flate

dW

PI+

*,P

gas

defla

ted

WP

I+*,

capa

city

utili

zatio

n,dr

illco

st-*

,4m

aW

WS

/WW

-*,

Ro

plus

Rg.

7D*d

ist*

0,88

cont

inue

d...



Tab

le5:

Su

mm

ary

of

Exp

lora

tio

nM

od

el-

(Par

t2)

No

Ref

eren

siD

ep.V

aria

bel

Inde

pend

entV

aria

bles

R2

Geo

logi

cal

Eco

nom

icA

dditi

onal

/Pro

xies

Ww

s/W

wP

oil

defla

ted

WP

I,P

gas

defla

ted

WP

I+*,

capa

city

utili

zatio

n,dr

illco

st+

*R

opl

usR

g.7D

*dis

t*0,

67

O/W

wos

Poi

ldefl

ated

WP

I,P

gas

defla

ted

WP

I,ca

paci

tyut

iliza

tion,

drill

cost

4ma

WW

S/W

W-*

,Ro,

7D*d

ist*

0,57

5G

hour

i(19

91)

1W

xD

eflat

edP

oil+

*,P

oil(-

1),P

oil(-

2),P

oil(-

3),P

oil(-

4),E

xplo

ratio

nco

sts-

*0,

90

Sam

ple:

AR

CO

US

74-8

9T

SW

xD

eflat

edP

oil+

*,P

oil(-

1)+

*,P

oil(-

2),

Poi

l(-3)

+*,

Poi

l(-4)

+*,

Exp

lora

tion

cost

s-*

0,94

Mob

il70

-88

US

TS

Wx

Defl

ated

Poi

l+*,

Poi

l(-1)

+*,

Poi

l(-2)

,P

oil(-

3)+

*,P

oil(-

4)+

*,E

xplo

ratio

nco

sts-

*

0,96

Phi

lips

70-8

8U

ST

SW

xD

eflat

edP

oil+

*,P

oil(-

1),

Poi

l(-2)

+*,

Poi

l(-3)

+*,

Poi

l(-4)

+*,

Exp

lora

tion

cost

s-*

0,94

She

ll70

–88

US

TS

6F

orbe

s&

Zam

pelli

(200

0)2

Ww

s/W

wD

epth

+*,

Ww

gs/W

w+

*P

oil+

*ln

(une

xplo

red

acre

age)

+*,

Tech

nolo

gi-

calp

roxi

es+

*0,

85

Sam

ple:

US

EIA

FR

S10

firm

s19

78-

1998

CT

7M

ohn

&O

smun

dsen

(200

8)2

WO

+G

(-1)

-***

,°O

+G

(-2)

+*,

Wd(

-1)-

***

P+

**,P

(-1)

+**

acre

age(

-1)+

*0,

51N

orw

egia

nC

ontin

enta

lS

helf

1965

–20

04C

TW

w°

O+

G(-

1)-*

**,W

d(-1

)-**

*P

+**

,P(-

1)+

**ac

reag

e(-1

)+*

0,5

Wa

°O

+G

(-1)

-***

,°

O+

G(-

2)-*

**,

Wd(

-1)-

**P

+**

,P(-

1)+

*ac

reag

e(-1

)+*

0,61

8B

oyce

&N

ostb

akke

n(2

011)

2W

Wxs

/W**

*,O

ffsh

ore+

**,

O,

G,

dept

h,E

OR

+*,

new

CB

MP

oil*

**,

Pga

s,D

rillc

ost,

real

inte

rest

rate

+**

Yea

r-**

*,La

nd-*

*-

US

Sta

te19

55–2

002

Ww

Wxs

/W**

*,O

ffsh

ore+

**,

O,

G,

dept

h,E

OR

+*,

new

CB

MP

oil*

**,

Pga

s,D

rillc

ost,

real

inte

rest

rate

Yea

r-**

*,La

nd-*

*

Wde

vW

xs/W

***,

Off

shor

e+**

*,O

-*,

G+

***,

dept

h,E

OR

+*,

new

CB

MP

oil-*

**,

Pga

s-**

*,D

rillc

ost-

**,

real

in-

tere

stra

teY

ear-

***,

Land

-**

-

9N

asir

(201

1)2

WS

eism

ic+

*,°

O+

G(-

1)E

xplo

ratio

n&D

evel

opm

ent

cost

-,w

orld

Poi

l+*,

OilC

onsu

mpt

ion+

*0,

78

Tim

eS

erie

s19

80–2

009

Indo

nesi

a

Not

e:1

estim

atio

nre

sults

wer

ere

port

edby

Dah

l&D

ugga

n(1

998)

Not

e:2

estim

atio

nre

sults

wer

eta

ken

from

rele

vant

relit

erat

ures

Sou

rce:

Aut

hors

’cal

cula

tion



Table 6: Description of Variables

(-1) = Lag of year Pgas = Gas priceCT = Cross section time series data Poil = Oil priceD* = Dummy variable Seismic = Seismic crewsD*dist = District dummy variable TS = Time series datadepth = Kedalaman sumur TX shut down days = The number of days a Texas was shut downG = Gas reserves discovered W = Totals wells drilledO = Oil reserves discovered Wds = Successful developoment wellsma = A moving average Gas = Associated gasGa = New associated gas reserves found Ww = New field wildcat wells drilledGna = New nonassociated gas reserves found Wwgs = Successful gas wildcat wells drilledPADD = Petroleum administrative defense district Wws = Successful wild cat wells drilledWx0s = Successful oil exploratory wells drilled Wg = Total gas wells drilledWo = Total oil wells drilled Wxs = Successful exploratory wells drilledWPI = Wholesale price index Wx = Exploratory wells drilledWs = Total successful wells drilled Wxgs = Successful gas exploratory wells drilled

Source: Dahl & Duggan (1998)

States and Canada Through 2001, U.S Department of theInterior. Circular 1288.


514-1060-1-PB

Documents

Transcript of 514-1060-1-PB