Employment effect of FDI

This article was downloaded by: [KU Leuven University Library]On: 02 August 2015, At: 09:17Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: 5 Howick Place, London, SW1P 1WG

The Journal of Development StudiesPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/fjds20

Foreign Ownership and EmploymentGrowth in a Developing CountryRobert E. Lipsey c a , Fredrik Sjholm b & Jing Sun ca NBER and City University of NYb Lund University and Research Institute of Industrial Economicsc University of IdahoPublished online: 04 Jun 2013.

To cite this article: Robert E. Lipsey , Fredrik Sjholm & Jing Sun (2013) Foreign Ownershipand Employment Growth in a Developing Country, The Journal of Development Studies, 49:8,1133-1147, DOI: 10.1080/00220388.2013.794264

To link to this article: http://dx.doi.org/10.1080/00220388.2013.794264

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (theContent) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Foreign Ownership and Employment Growth in aDeveloping Country

ROBERT E. LIPSEY,*, FREDRIK SJHOLM** & JING SUN

*NBER and City University of NY, **Lund University and Research Institute of Industrial Economics,

University of Idaho

Final version received February 2013

ABSTRACT Many developing countries would like to increase employment in the formal sectors. One way toaccomplish this goal may be to encourage the entrance of foreign firms. We examine employment growth inIndonesian plants taken over by foreign owners from domestic ones. We also examine the effect of FDI duringdifferent trade regimes and the timing of employment effects following an acquisition. For plants that change thenationality of ownership, we find a strong effect of shifts from domestic to foreign ownership in raising the growthrate of employment, but no significant effects of shifts from foreign to domestic ownership.

1. Introduction

One of the possible consequences of inward foreign direct investments (FDI) for developing countries,and one that is of particular interest to their governments, is the extent to which the investment createsnew jobs in the industrial, or modern sector, to help in the transformation of the economies. TheLewis (1954) notion of a need to move people out of agriculture and into the modern sector is still agoal for many developing countries (Asian Development Bank, 2005). There are several ways inwhich inward FDI might play this role.For instance, there is considerable evidence that foreign-owned firms are relatively efficient, and

may for that reason have access to foreign markets that would not be within the reach of domesticallyowned firms. They may also have wider contacts and knowledge of world markets and better access tofinancing, all advantages that should provide a positive effect on their employment.1 On the other side,the foreign-owned firms may compete with domestically owned firms for some markets, so that thelosses of employment by domestically owned firms may offset, to some extent, the gains in theforeign-owned firms. In addition, the foreign-owned firms may tend to be more capital-intensive thandomestically owned firms, and more intensive in the use of imported intermediate products, so that anincrease in their sales adds less to employment than would a corresponding increase by domesticallyowned firms.Considering the potential importance of FDI for formal sector employment growth in developing

countries, there is an unfortunate lack of existing studies. One exception is Karlsson, Lundin, Sjholm,& He (2009) which finds higher growth in employment in foreign than in domestically owned Chinesefirms. However, from such comparisons it is not possible to disentangle a selection effect that

Correspondence Address: Dr Jing Sun, College of Business and Economics, University of Idaho, 875 Campus Drive, PO Box443161, Moscow 83844, USA. Email: [email protected]*Robert E. Lipsey was a Research Fellow, NBER, and Professor of Economics at City University of New York when he sadlypassed away on 11 August 2011.An Online Appendix is available for this article which can be accessed via the online version of this journal available at http://dx.doi.org/10.1080/00220388.2013.794264.

The Journal of Development Studies, 2013Vol. 49, No. 8, 11331147, http://dx.doi.org/10.1080/00220388.2013.794264

2013 Taylor & Francis

Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

foreign owners acquire relatively good firms from a pure growth effect of foreign ownership (Arnold& Javorcik, 2009).We contribute to the existing literature in several aspects. First, we examine employment growth

after foreign acquisitions of domestically owned establishments and domestic acquisitions of foreign-owned establishments. If foreign ownership provides superior technology or better access to worldmarkets, establishments should tend to raise their employment after foreign takeovers. If theseadvantages require continued foreign ownership, there may be employment losses when a foreign-owned establishment is acquired by a domestic firm. On the other hand, if the technological or othergains from foreign ownership are retained in the establishment, its level and growth of employmentmay continue after a domestic acquisition.Second, given a possible relationship between trade regimes and development in a developing

economy like Indonesia, we test whether a countrys trade regime might affect the employment effectof FDI. FDI flows attracted to a country by cheap production costs will presumably respond to anexport-oriented policy by expansion (Balasubramanyam, Salisu, & Sapsford, 1996). A nice feature ofour long time series is that it enables us to examine the employment effect of FDI during differentIndonesian trade regimes.Last, our panel data allow us to track the pattern of employment changes following a foreign

acquisition. Our results show that the employment growth is concentrated in the year of acquisition.This suggests that acquisitions can be associated with very substantial additions to resources, contest-ing the conventional wisdom that green-field investment adds resources to the recipient country, butacquisitions only change ownership.

2. Previous Literature

There have been relatively few studies of the effects of takeovers of domestic firms or plants byforeign firms on levels of employment in the target plants. Moreover, almost all of the existing studiesexamine the issue in developed economies. Bandick & Karpaty (2007) found some evidence of apositive effect on numbers of employees of acquisitions by foreign firms of Swedish firms: theincrease in the number of skilled employees was larger than the decline in the number of unskilledworkers. In another study on Swedish data, Bandick & Grg (2010) found that the positive effect onemployment was restricted to acquisitions of exporting firms. In a paper on mergers and acquisitionsin Finland comparing cross-border with other types, going beyond manufacturing to include serviceindustries and using matching methods to assure the comparability of acquired firms to those notchanging ownership, Lehto & Bckerman (2008) found that cross-border acquisitions in Finland led toreductions in employment in manufacturing, but effects in non-manufacturing were much weaker. Anexamination of foreign acquisitions in UK manufacturing between 1989 and 1996 by Girma (2005, p.165), using propensity score matching, concluded that foreign acquisition did not result in any clearemployment effects, but there were some signs of a negative effect in the case of larger take-overtargets, and beneficial impacts amongst smaller ones.Among developing host countries, there have been some studies for China. A paper on China by

Gong, Grg, & Maioli (2007) compared the employment effects of domestic privatisation with thoseof foreign acquisition of state-owned enterprises, using propensity score matching and differences indifferences. However, their study controlled for firm output growth, and therefore was an examinationof productivity effects rather than employment effects. They found that given firm output growth,domestic privatisation led to reductions in employment growth, and therefore increases in productivity,compared with firms that did not change ownership, while foreign acquisitions were associated withhigher employment growth. The authors suspected that foreign acquirers, in particular, were con-strained by promises made with respect to employment in securing permission for acquisitions.Other studies that focused on productivity rather than employment levels, or on effects of ownership

rather than acquisition, included, for China, Karlsson, Lundin, Sjholm, & He (2009, p. 198) findinghigher employment growth in foreign-owned firms, Takii (2005), which found higher productivity

1134 R.E. Lipsey et al.

Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

levels in foreign-owned plants in Indonesian manufacturing, and Arnold & Javorcik (2009), whichfound positive productivity effects from foreign acquisition in Indonesian manufacturing between1983 and 1996.There is also a literature that examines how FDI affects productivity in local firms through

spillovers. The extent of spillovers seems to differ across countries (Grg & Greenaway, 2004;Harrison & Rodrguez-Clare, 2010; Lipsey & Sjholm, 2005), but a number of studies find positivespillovers in Indonesia (Blalock & Gertler, 2008, 2009; Javorcik, 2004).

3. Foreign Plants in Indonesian Manufacturing

We analyse Indonesian manufacturing data supplied by the Indonesian Statistical Office for the period1975 to 2005 for all manufacturing plants with more than 20 employees. A plant identification codeenables us to construct a panel and follow individual plants over time.Employment in manufacturing plants with more than 20 employees increased from fewer than

700,000 in 1975 to about 4 million in 1997 and later years (Figure 1). That growth was driven mainlyby a strong increase in employment in domestically owned private plants, which remained close tothree-quarters of all plants during the entire period. Foreign establishments have played an increasingrole in Indonesian manufacturing employment. Plants with some foreign ownership, accounting forless than 10 per cent of manufacturing employment in 1975, employed around 20 per cent in 1997, atthe time of the Asian crisis. After that the share declined slightly, but then recovered to 20 per centagain in 2005. The share of government-owned plants, much larger than that of foreign plants in 1975,shrank steadily after the late 1980s, and was only 5 per cent of manufacturing employment in 2005.The industry sectors and the ownership groups differed in some important aspects. One extreme

difference was in size: government-owned plants were far larger than domestically owned private plants,five times as large, on average, in 1975 and still over three times as large in 2005. They were muchlarger also within the industry groups, with a few exceptions (Table 1). Foreign-owned plants were alsomuch larger than domestically owned private plants, about three times as large in both beginning and endyears. In 2005, the foreign-owned plants were larger than domestically owned private plants in everygroup. The size disparity may be an element in the frequency and success of takeovers.

Figure 1. Employment in Indonesian manufacturing, by ownership.

Foreign Ownership and Employment Growth in a Developing Country 1135

Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

To the extent that we can associate the share of blue-collar workers in total employment with theaverage skill level in an establishment, it appears that foreign firms tended to use a slightly higher-skilllabour force than private domestic firms in the same industry. Government-owned plants operated withthe lowest proportions of blue-collar workers consistently across almost all industries. Only govern-ment-owned plants employed work forces with 30 per cent or more of white-collar workers, almost 40per cent in a few cases, while private domestic plants employed more than 20 per cent white-collarworkers in only one industry group in one year.The changes in the share of Indonesian manufacturing employment in foreign-owned plants,

discussed above, came about in several different ways. One was takeovers of domestically ownedplants by foreign firms, offset by takeovers of foreign-owned plants by Indonesian owners. Anotherwas the establishment of new plants by foreign owners and the demise of existing plants. A thirdsource of change was any differences in average rates of growth in employment between locallyowned and foreign-owned plants.These different sources of growth of employment in foreign-owned plants are shown in Table 2. Up

through 1989, foreign takeovers accounted for a large part of total growth in employment in foreign-owned manufacturing establishments, but they were offset by declines in such employment from localtakeovers of foreign-owned plants. After 1989, the foreign takeovers added more to the foreign-ownedshare than the domestic takeovers took away.The path of takeover activity between foreign and domestic owners, in terms of numbers of

takeovers, is described in Figure 2. The numbers of takeovers had been fairly similar in the twodirections until the 1990s, but since then foreign takeovers have been more numerous, except in 1997during the Asian crisis. However, the net effect of foreign and domestic takeovers was less importantas a source of employment growth in foreign-owned establishments than the combination of theestablishment of new foreign-owned plants and their more rapid growth.

Table 1. Average number of employees per establishment and the share of blue-collar workers, 1975 and 2005

Private-domestic Govt-domestic Foreign

Sector ISIC

Averageemployees per

plantShare of blue-collar workers





1975Total 75 0.88 365 0.75 219 0.77

31 91 0.88 537 0.75 179 0.8132 72 0.93 507 0.81 431 0.9033 58 0.82 90 0.86 146 0.8134 52 0.84 228 0.71 157 0.7835 74 0.83 243 0.68 167 0.6436 41 0.88 385 0.71 325 0.8537 174 0.82 72* 0.65* 96* 0.75*38 87 0.86 210 0.72 223 0.7339 47 0.90 191* 0.82* 167* 0.92*

2005Total 157 0.85 481 0.74 563 0.79

31 135 0.85 507 0.74 517 0.7532 206 0.89 204 0.85 1060 0.8933 168 0.87 116 0.83 280 0.8334 145 0.78 519 0.75 647 0.7835 178 0.79 530 0.68 389 0.7036 89 0.87 725 0.67 398 0.8037 205 0.78 1822* 0.75* 215 0.7638 142 0.82 619 0.66 536 0.8039 120 0.87 287* 0.90* 664 0.87

Note: *Fewer than five observations.


Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

4. Econometric Approach

We begin the econometric analysis by treating growth in employment as a function of various plantcharacteristics:

ln Lit ln Lit ln Lit1 Plantit1 X

wOwnershipit X

tYear dummy XindInd dummyj

XRReg dummy "it

(1)

where i indexes firms, and t indexes year.The variables included in the model are: L: employment; Plant: a vector of lagged plant

characteristics, that is plant size measured by employment, energy intensity (quantity of energyper employee), which is a proxy for physical capital intensity, and inputs of intermediate goods,defined as raw materials, fuel, and lubricants, per employee; Ownershipi: ownership dummyvariables indicating three ownership categories, private domestically owned, private foreign-owned, and government-owned; and dummy variables for year, industry (two-digit ISIC), and region(provinces aggregated into five regions).The plant control variables might be endogenously determined and we try to control for this possibility

by lagging them one period. Hence, we assume that growth in employment between period t and t + 1 is

Figure 2. The number of takeovers in Indonesian manufacturing, 19762005.

Table 2. Employment growth in foreign-owned manufacturing establishments in Indonesia, by source of growth,19752005

Year Foreign Foreign takeover Domestic takeover Othera

1975~1979 49,379 21,190 10,765 38,9541980~1984 9,197 18,463 27,435 18,1691985~1989 30,615 47,488 47,997 31,1241990~1994 384,856 182,561 87,909 290,2041995~1999 135,759 216,927 181,210 100,0422000~2005 108,500 300,782 110,081 82,201

Note: a New establishments minus disappearances, firm growth after takeover, and miscellaneous changes.


Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

caused by, for instance, the size in period t. Labour productivity, as measured by value-added peremployee, was included in some specifications, but it added nothing to the equation and was dropped.Ownership is in the data divided into shares of foreign, government-domestic and private-domestic.

Foreign establishments are defined as plants with any foreign ownership, and the dummy variable hasthe value one for all periods with foreign ownership.2 Government-owned establishments are defined asplants without foreign ownership but with any government (central or local) ownership. The remainingplants are defined as private-domestically owned. In some later estimations ownership is instead adummy on foreign acquisitions of domestically owned plants and a dummy on domestic acquisitions offoreign-owned plants. The foreign acquisition dummy variable is one for all the years after foreigntakeover, and the domestic acquisition dummy variable is one for all the years after domestic takeover.The universe we examine in the estimations on takeovers includes all firms except those that experiencemultiple ownership changes. Firm-specific effects and time dummies are included in the regressions.One problem with estimating the causal effect of an acquisition on employment is the possible

endogeneity of firms being acquired. Acquisitions may not be random with respect to factors thatdetermine future growth. This means that estimates on employment growth may be biased if non-randomness is not taken into account. More specifically, firms that become foreign owned might besuch that they would in any case develop differently than firms that remain domestically owned. Thismeans that estimates on outcome variables (such as employment) become biased.One often-used approach to control for selection bias is to use an instrumental variable (IV)

approach. However, instrumental variables are arguably difficult in the context of FDI and employ-ment, since most variables that affect foreign acquisitions also could have an effect on most outcomevariables including employment.We therefore use propensity score matching (PSM) combined with the more general difference-in-

differences technique, as suggested by, for example, Blundell & Costa Dias (2000), Heyman, Sjholm,& Gustavsson Tingvall (2007), and Arnold & Javorcik (2009). The matching procedure aims to find agroup of non-acquired plants that display the same characteristics as the group of acquired plants priorto the acquisition. The matched non-acquired plants are the counterfactuals to approximate theemployment growth that the acquired plants would have had if they were not acquired. Specifically,the identifying assumption for the matching technique is that, conditional on all relevant covariates,the employment growth rates are independent of the acquisition status (conditional independence), soacquisition status is random.3 The conditional independence assumption is not directly testable(Wooldridge, 2010, p. 910), but our dataset provides us with rich control variables to help thisassumption to hold. Moreover, the combination of propensity score matching and difference-in-differences estimator removes the time-invariant bias. In other words, this estimation strategy allowsus to control for observable and unobservable but constant differences between acquired and non-acquired plants. However, it should be noted that the methodology does not solve possible endogene-ity problems caused by time-varying unobservable variables.To obtain the matched control group, we estimate the propensity score by fitting a probit model. We

then apply the difference-in-differences estimator on the matched samples to estimate the impact ofacquisitions on employment growth. The difference in employment growth trends after acquisition canbe attributed to the effects of acquisition. Below is our DD estimator:

DD E Ltreatedpost Xj

E Lcontrolpost Xj

E Ltreatedpre Xj

E Lcontrolpre Xj

(2)

L is employment growth rates (difference in log employment) or, in some estimations, employment itself.Post refers to the post-acquisition period, which could be in the year of acquisition, or one year after, orthe average of the whole post-acquisition periods. Pre refers to the period before acquisition. Similarly itcould be one year before the acquisition, or the average of the all the years before acquisition. Thedifference in the second parenthesis corrects the selection bias in the pre-acquisition period. In ourmatched samples, the control group for foreign takeovers is taken from the plants that are alwaysdomestic, while the control group for domestic takeovers is taken from the plants that are always foreign.


Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

The model specification for the propensity score includes age, employment, the share of white-collarworkers, inputs, energy use, and productivity. All variables (except age) are lagged one year and thesquare of all variables are also included. Moreover, industry and year specific effects are included.Table 3 shows that large domestically owned plants with low productivity and a high share of white-collar workers are relatively likely to be acquired by foreign owners. By contrast, foreign-owned plantsthat are small, with low productivity and energy intensity, are relatively likely to be taken over bydomestic owners. Hence, foreigners and domestic actors tend to acquire different types of plants. Byconstructing a matched sample based on the probability of takeover, the selection problem is reduced.We employ a nearest neighbour matching technique with replacement to construct our matched

sample of plants. In the case of foreign takeover, each domestic plant that would be acquired later byforeign owners is matched to an always domestic plant that has the closest propensity score. The sameapproach is used for domestic takeovers. Moreover, the matched treated and control units are from thesame year and same industry.Of the 1,036 foreign takeovers, 377 are in the treatment group. The loss in the number of foreign

takeovers from the treatment group is mainly because there are 442 foreign takeovers reported to havetaken place in the second year after the plant starts operation, and another 108 foreign takeovers arematched to domestic plants that do not report employment growth before the matched year, and thusthere is no employment growth in the pre-acquisition period with which to compare. Another 109takeovers are dropped because there are some missing values in the observed characteristics used toestimate propensity scores. Of 652 domestic takeovers, 291 are included in the treatment group; 233domestic takeovers are dropped because they are reported to have taken place in the second year ofoperation, and another 128 domestic takeovers are dropped because of missing values. It is a cause for

Table 3. Probit results for foreign and domestic takeovers

Foreign takeover Domestic takeover

Age 0.145 0.062***[0.007] [0.013]

Age squared 0.003*** 0.002***[0.000] [0.000]

Employment 0.549*** 0.450***[0.076] [0.154]

Employment squared 0.029*** 0.021[0.007] [0.014]

Ratio of white-collar workers 0.678*** 0.872*[0.232] [0.447]

Ratio of white-collar workers squared 0.288 0.935*[0.329] [0.566]

Inputs 0.008 0.004[0.037] [0.076]

Inputs squared 0.001 0.003[0.002] [0.004]

Energy 0.013 0.094*[0.037] [0.054]

Energy squared 0.000 0.006[0.003] [0.004]

Productivity 0.217** 0.097[0.086] [0.085]

Productivity squared 0.015*** 0.010**[0.004] [0.005]

Year fixed effects Yes YesIndustry fixed effects Yes YesNo. of observations 221,068 9,417Chi-squared 1,414 420Pseudo R2 0.1897 0.1345

Notes: *Significant at the 10% level; **Significant at the 5% level; ***Significant at the 1% level.


Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

concern that so many takeovers are dropped because they are reported to take place in the second year ofexistence. However, the regression analyses were carried out on samples with and without takeovers inthe second year, and the results were robust.To identify the treatment effect, the so called balancing property of the propensity score must be

fulfilled. This means that observations with the same propensity score must have the same distributionof characteristics, independently of treatment status. Observations with the same propensity score areon average observationally identical, implying that exposure to treatment is random. In other words,the balancing property ensures that sufficiently good matches are found for all treated observations.Tests are conducted to make sure that our matched sample is balanced in the sense that the treated

and control units have similar pre-takeover values on the control variables (Online Appendix TablesA1 and A2). There are large differences in the unmatched sample between characteristics between thetreated and control groups. The differences in means of the control variables are in general notsignificant between treated and control units in the matched sample. The one exception is that treatedplants tend to have a larger share of white-collar workers even in the matched sample.Moreover, the difference-in-differences approach assumes that the acquired and non-acquired plants

should follow a similar employment growth trend without the acquisition, given that they have similarplant characteristics before acquisition. Figures A1 and A2 in the Online Appendix show growth inemployment before and after foreign acquisitions of domestic plants and before and after domesticacquisitions of foreign plants. There are no obvious differences in the employment trend in treated andcontrol groups before acquisitions. Hence, it seems that the assumption of a common growth trendbefore treatment is reasonably fulfilled. However, the development after acquisitions is strikinglydifferent between foreign and domestic acquisitions, with a sharp increase in employment growth theyear of acquisition in the former group and with no clear effect in the latter group.

5. Econometric Results

5.1 Determinants of the Rate of Plant Employment Growth

We start in Table 4 with simple OLS analyses on the whole universe of manufacturing plants. Theequations include the ownership variables, Foreign and Government, and the reference group istherefore domestic-private firms. The coefficient for Foreign is positive and statistically significant,

Table 4. Ownership and growth in employment, OLS estimations

Total employment Blue-collar workers White-collar workers

Foreign (t) 0.060 0.060 0.036(20.74)*** (19.93)*** (10.27)***

Government (t) 0.023 0.021 0.011(6.28)*** (5.31)*** (2.01)**

Size (t1) 0.038 0.036 0.026(47.47)*** (46.27)*** (27.38)***

Energy (t1) 0.011 0.011 0.005(31.86)*** (29.51)*** (9.45)***

Inputs (t1) 0.012 0.012 0.009(27.53)*** (25.17)*** (13.76)***

Time dummy Estimated Estimated EstimatedInd. dummy Estimated Estimated EstimatedRegion dummy Estimated Estimated Estimated

R-square 0.028 0.021 0.006No. of obserations 324,387 324,268 277,653

Notes: A constant is included in all estimations. Energy, and Inputs are in log form. T- values based on robust(cluster at plant level), standard deviations are in parentheses. *significant at the 10 per cent level; **significant atthe 5 per cent level; ***significant at the 1 per cent level.


Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

indicating a rate of growth in employment 6 percentage points higher in foreign-owned than indomestic-private plants. The coefficient for government is also positive and statistically significant.The equation includes plant characteristics that might affect employment growth. Large firms have

comparatively low growth rates, as has been found in previous studies (for example, Karlsson et al.,2009). Plants that are more energy intensive and use more raw materials are associated with higheremployment growth rates in general. The last two columns examine growth of the numbers of blue-and white-collar workers. The positive effect on the employment growth of blue-collar workers issubstantially larger than the effect on white-collar workers: 6 percentage points compared to 3.7percentage points. The effect of government ownership is also higher for blue- than for white-collarworkers, but both effects are small compared to the effect of foreign ownership. Finally, the negativeeffect of size and the positive effect of input per employee primarily affect blue-collar workers, as isalso the case for the positive effect from energy intensity.The evidence of Table 4 is that foreign-owned plants tend to increase their employment 6 percentage

points faster than private domestically owned plants over the years of their existence, given the othercharacteristics of the plants.

5.2 Foreign Takeovers and Employment Growth

We separate the effects of foreign takeovers from those of foreign ownership in general in Table 5. TheOLS estimate of the effect of foreign ownership aside from foreign acquisition effects is about 5.5percentage points per year faster growth in employment. The effect of foreign acquisition is subse-quent growth in employment at a rate 9 percentage points faster.

Table 5. Acquisitions and growth in employment

Total employees Total employees Blue-collar workers White-collar workers

OLS Fixed effect Fixed effect Fixed effect

Always foreign 0.055 (16.50)***

Before foreign acquisition 0.078(10.94)***

Foreign acquisition 0.088 0.108 0.122 0.063(14.68)*** (6.11)*** (6.61)*** (2.98)***

Before domestic acquisition 0.031(5.83)***

Domestic acquisition 0.004 0.030 0.027 0.044(0.69) (1.58) (1.38) (1.79)*

Government 0.023 (6.37)***

Size (t1) 0.038 0.426 0.425 0.353(48.01)*** (27.16)*** (27.23)*** (16.47)***

Energy (t1) 0.011 0.002 0.002 0.004(32.02)*** (0.61) (0.58) (0.84)

Inputs (t1) 0.012 0.007 0.006 0.002(27.50)*** (1.35) (1.18) (0.18)

Time dummy Estimated Estimated Estimated EstimatedInd. dummy Estimated Region dummy Estimated R-square 0.028 0.264 0.214 0.068No. of observations 327,843 15,347 15,333 14,580

Notes: A constant is included in all estimations. Energy, and Inputs are in log form. T- values based on robust(cluster at plant level), standard deviations are in parentheses.significant at the 10 per cent level; **significant atthe 5 per cent level; ***significant at the 1 per cent level.


Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

We have also included a dummy variable indicating the period before a domestic plant was takenover by foreign owners, and a similar dummy variable for foreign plants that were taken over bydomestic owners. These dummy variables capture pre-acquisition growth rates in employment and areboth positive and statistically significant, indicating that all takeovers are of high-growth plants.Including these dummy variables has no impact on the result for foreign acquisitions.The fixed-effect approach looks at growth in employment within a firm before and after the

acquisition and removes the time-constant unobserved plant characteristics that could confound theexplanation of acquisition effects. Only firms that change ownership are included. Fixed-effectestimates raise the foreign acquisition effect to 11 percentage points. The effect on blue-collar workersis about twice as large as the effect on white-collar workers. Moreover, the results indicate thatdomestic acquisition reduces the subsequent rate of employment growth, although only the effect onwhite-collar workers is statistically significant.The effect of FDI on employment might differ between trade regimes (Balasubramanyam et al.,

1996). FDI flows drawn to a developing country to take advantage of cheaper labour costs wouldrespond to an export-oriented policy by expansion. By contrast, FDI induced by import substitutionpolicy is limited by the size and income level of the host country market. Moreover, Melitz (2003)suggests that trade liberalisations will have substantial reallocation effects with expanding employmentin high productivity firms (foreign) and declining employment in low productivity firms (domestic).To test for the possible impact of differences in trade regimes, suggested above, we divide

Indonesias history since 1975 into three periods, which we think of as an import substitution period,19751985; a trade liberalisation period, 19861996; and the crisis and post-crisis period, 19972005.4 The results, shown in Table 6, support the idea that the effects of FDI on host countries areaffected by trade regimes. During the trade liberalisation period, 19861996, the employment growthrate effect of foreign acquisition was as high as 17 percentage points. In contrast, foreign takeovers hadno significant effects on employment growth rates during the earlier, import substitution, period 19751985. It is interesting to note that the coefficient for foreign acquisitions during the crisis period ispositive. Manufacturing employment has not increased in Indonesia in recent years, which is a sourceof concern among policy-makers (Chatani, 2012). Our result suggests that this development inmanufacturing employment is at least not caused by foreign acquisitions, but that such acquisitionsinstead seem to increase employment also in times of overall stagnation. One possible explanation isthat foreign-owned firms could have greater access to overseas financing and would expand followingthe East Asian financial crisis, while lack of liquidity during the crisis greatly constrained domestic-owned firms ability to take advantage of improved terms of trade (Blalock, Gertler, & Levine, 2008).

Table 6. Acquisitions and growth in employment in different time periods, fixed effects (only acquired plants)

19751985 19861996 19972005

Foreign Acquisition 0.002 0.174*** 0.125***(0.118) (0.048) (0.029)

Domestic Acquisition 0.037 0.008 0.001(0.040) (0.041) (0.035)

Size (t1) 0.505*** 0.551*** 0.654***(0.061) (0.030) (0.024)

Energy (t1) 0.002 0.007 0.013***(0.007) (0.006) (0.005)

Inputs (t1) 0.032* 0.004 0.000(0.018) (0.009) (0.006)

Time dummy Estimated Estimated EstimatedR-square 0.28 0.35 0.36No. of observations 1,644 5,459 7,483

Notes: Only plants with one takeover either foreign or domestic, are used. A constant is included in allestimations. Size, Energy and Inputs are in log form. Standard errors clustered at plant level. *significant at the10 per cent level; **significant at the 5 per cent level; ***significant at the 1 per cent level.


Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

5.3 Matched Comparisons of Domestic and Foreign Takeovers

The results above might be biased if acquisitions are not random and if acquired plants differ fromnon-acquired plants. We therefore continue with our analysis on the matched sample of acquired andnon-acquired plants which will reduce any selection bias. The results are shown in Table 7. Foreigntakeovers raise the growth rate of employment by 11 percentage points on average during theacquisition and post-acquisition period, after correcting for the pre-acquisition differences betweenacquired and non-acquired plants. This is similar to the fixed-effect estimate. Domestic takeovers,according to the matched comparison, do not affect employment growth rates in the acquisition yearand the following two years. However, under the current model specification, the domestic takeoversraise the employment growth rates by 7 percentage points on average during the post-acquisitionperiod. As a robustness check, we experimented with different model specifications in the probitmodel. The positive effects of foreign takeovers on employment growth are robust to various modelspecifications, while the effects of domestic takeovers on employment growth are mostly negative orinsignificant.While the employment growth rates in foreign takeovers do not differ significantly from those of

plants remaining domestically owned in the first and second years after the takeover, the impact of theforeign takeovers continues, because the acquired plants grow so much in the year of takeover that thesame growth rate after takeover implies a considerably larger absolute growth in employment in thefollowing years in the acquired plants, relative to domestic plants, as is shown in Table 8, which showsthe increase in levels of employment. There are no similar effects in absolute terms from domesticacquisitions of foreign plants. The concentration of employment growth rate changes in the year ofacquisition with the consequent carryover of absolute employment growth into the following years.

Table 7. Estimated effects of takeovers on employment growth rates after takeover, propensity score matching

Foreign takeover (control: alwaysdomestic)

Domestic takeover (control: alwaysforeign)

DD Standard error DD Standard error

Acquisition year 0.314*** (0.0564) 0.037 (0.0466)One year after acquisition 0.0305 (0.0469) 0.014 (0.0381)Two year after acquisition 0.009 (0.0487) 0.054 (0.0483)Average of post-acquisition 0.109*** (0.0368) 0.073* (0.0377)


Table 8. Estimated effects of takeovers on employment after takeover, propensity score matching

Foreign takeover (control: alwaysdomestic)

Domestic takeover (control: alwaysforeign)

DD Standard error DD Standard error

Acquisition year 134** (54.9) 23 (28.3)One year after acquisition 118** (55.9) 12 (36.8)Two year after acquisition 186*** (54.4) 1 (31.8)Average of post-acquisition 158*** (50.2) 2 (33.4)

Notes: For foreign takeovers, the average number of years after acquisition for both treated and control group isapproximately six years. For domestic takeovers, both treated and control groups have on average six years afteracquisition. The pre-acquisition for this calculation uses information at one year before acquisition. It would notchange the story if the average from all the years before acquisition is used instead.Standard errors arebootstrapped. *Significant at the 10% level; **Significant at the 5% level; ***Significant at the 1% level.


Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

One implication of this concentration in the year of acquisition is that the usual assumption thatGreenfield investment adds resources to the recipient country but acquisitions only change ownershipneeds to be put in to perspective (Nocke & Yeaple, 2007, p. 357). Acquisitions can be associated withvery substantial additions to resources, quite apart from any gains that might arise from transfers fromless-skilled to more-skilled management.One related issue is whether the employment growth in foreign-owned plants is at the expense of the

domestically owned plants in the same industry because of a business stealing effect (Harrison &Rodrguez-Clare, 2010). In other words, when foreign ownership boosts employment growth inacquired plants, how has the industry-level employment growth been affected? We examine thisissue by correlating the growth in foreign takeovers with the growth in domestic (private andgovernment) plant employment at an industry level after controlling for industry and year fixed effects(Table 9). The results suggest that a rise in employment growth in foreign takeovers is correlated withemployment growth in plants that remain domestic. One reason could be an increased market forsuppliers of intermediate goods.5 We made some simple calculations to examine this issue with the useof inputs as a dependent variable in fixed effect estimations on foreign acquisitions (not shown).Foreign acquisitions increased the use of inputs by 42 per cent, which supports the hypothesis ofincreased markets for input goods. Another reason could be that domestic firms learn to export due tothe presence of foreign plants in the same industry and thereby increase output and employment.6

6. Conclusions

Modern sector employment growth is in many developing countries as important as it is difficult toachieve. This article examined whether FDI can play a role in manufacturing employment growth. Theanalysis contributes to a relatively scarce existing literature in several respects: by examining thecausality of FDI on employment; by looking at the effect of trade regimes; and by examining thetiming of employment effects.Foreign-owned Indonesian manufacturing plants grew more rapidly in employment than plants that

remained in Indonesian ownership during 19752005, given the other characteristics of the plants. Themore rapid growth is confirmed by several tests of the data, including a close examination of takeoversof locally owned plants by foreigners and of foreign-owned plants by local owners. Employment inplants that were foreign-owned throughout our period grew, on average, about 5.5 percentage pointsfaster than always domestically owned plants. Plants that were acquired by foreigners grew about 11percentage points faster than their pre-acquisition level, according to fixed-effect estimates.Considering that foreign plants are on average considerably larger than domestic plants, the differencein the number of jobs created was large.The propensity score matching consistently confirmed the advantages of foreign ownership for

employment growth. There is also some indication that the employment growth effects of foreignownership are sensitive to host country trade policy, with liberalisation encouraging the expansion of

Table 9. Estimated effects of foreign acquisitions on growth of domestically owned plants (dependent variable:industry-level employment growth rate of domestic firms in decimals)

Regression1 Regression2 Regression3 Regression4

Employment growth rate of foreign takeovers(in decimals)

0.094***(0.001)

0.094***(0.001)

0.032***(0.007)

0.018***(0.006)

Industry fixed effects N N Y YYear fixed effects N Y N YNo of observations 275 275 269 269Adjusted R-squared 0.9916 0.9928 0.0598 0.3605



Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

employment through foreign takeover. There were indications in several tests that there was a declineof employment growth in shifts from foreign to domestic ownership, although that result is notstatistically significant.Most of the employment effects of foreign takeovers took place in the year of takeover. There was

relatively little effect on growth rates in the following years, but the absolute additions to employmentin the years after takeover were larger than they would have been under continued local ownershipbecause the base was much larger. The negative or insignificant effect of domestic acquisition onforeign-owned plants, as in the fixed effects estimate and the difference-in-differences estimate from amatched sample, shows that the advantages of foreign-owned plants that accounted for more rapidgrowth required continued foreign ownership. They are apparently lost if the plant returns to domesticownership.One possible implication of the concentration of growth in the year of acquisition is that the

distinction between greenfield investment and acquisitions is not as sharp as is often assumed.Many of the acquisitions apparently involve major changes in the size and possibly other dimensionsof the target firms.The results also contribute to the recent discussion on job creation in large versus small businesses.

Much of the existing evidence shows an inverse relationship between firm size and employmentgrowth rates (Neumark, Wall, & Zhang, 2011), which has in some countries led to various policyinterventions in favour of small businesses. However, our results suggest that bigger firms have higherchances of being acquired by foreign firms, which boosts their employment growth even aftercontrolling for the firm size. This has important implications for developing countries like Indonesiatrying to catch up with the developed economies. Foreign-owned firms tend to be bigger and moreproductive. Thus, their expansion will help to address the issue of resource misallocation in developingeconomies (Hsieh & Klenow, 2009) and improve productivity and growth.7

Acknowledgements

We are grateful for comments and suggestions by a large number of conference and seminarparticipants and from two anonymous referees and the editor of the journal. Fredrik Sjholm gratefullyacknowledges financial support from the Ragnar Sderberg Foundation. The data have been providedby the Indonesian Bureau of Statistics. Programming codes are available upon request.

Notes

1. See Lipsey (2004) for a survey on host country effects of FDI. For studies on FDI in Indonesia, see, for example, Blomstrm& Sjholm (1999), Lipsey & Sjholm (2004), Takii (2005), Blalock & Gertler (2008), and Arnold & Javorcik (2009). Forstudies on multinational firms and labour markets in high income countries, see Bandick and Karpaty (2007), Lehto andBckerman (2008), Hakkala, Heyman, & Sjholm (2008), and Bandick & Grg (2010).

2. Using alternative definitions of foreign ownership with different required ownership thresholds has no major impact on theresults.

3. Wooldridge (2010, p. 910) discusses another identifying assumption called the overlap assumption. Following Arnold &Javorcik (2009), we exclude observations outside the common support in our matching estimation. The common support isbound by the lowest propensity score of a treatment observation and the highest propensity score of a control observation.

4. See Aswicahyono, Bird, & Hill (1996), Aswicahyono, Dionisius, & Hill (2008) and Aswicahyono & Hill (2002) fordiscussions on Indonesias policy regimes, and for similar distinctions in different periods.

5. We need to use an industry classification at a two-digit level of ISIC (nine industries) because of changes in classificationover time at lower levels of aggregation. That means that suppliers of intermediate goods often are in the same industry asproducers of finished goods.

6. Swenson (2008) finds that a growing presence of multinational firms in China enhances the export capabilities of localdomestic firms. We tried to conduct a similar analysis but were unsuccessful because of poor export data.

7. A recent study by Braguinsky, Branstetter, & Regateiro (2011) is interesting in this respect. It explains Portugals anemicgrowth and low productivity in recent decades by a leftward shift in the firm size distribution.


Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

References

Arnold, Jens Matthias, & Javorcik, Beata s. (2009). gifted kids or pushy parents? foreign acquisitions and Plant Productivity inIndonesia. Journal of International Economics, 79, 4253.

Asian Development Bank. (2005). Labor markets in Asia: Promoting full, productive, and decent employment. Manila: AsianDevelopment Bank.

Aswicahyono, H. H., Bird, Kelly, & Hill, Hal. (1996). What happens to industrial structure when countries liberalize? Indonesiasince the mid 1980s. Journal of Development Studies, 32, 340363.

Aswicahyono, Haryo, Dionisius, Narjoko, & Hill, Hal. (2008, November). Industrialization after a deep economic crisis:Indonesia. Paper presented at the 11th International Convention of the East Asian Economic Association, Manila.

Aswicahyono, Haryo, & Hill, Hal. (2002). Perspiration versus inspiration in Asian industrialisation: Indonesia before the crisis.Journal of Development Studies, 38, 138163.

Balasubramanyam, V. N., Salisu, M., & Sapsford, David. (1996). Foreign direct investment and growth in EP and IS countries.The Economic Journal, 106, 92105.

Bandick, Roger, & Grg, Holger. (2010). Foreign acquisition, plant survival, and employment growth. Canadian Journal ofEconomics, 43, 547573.

Bandick, Roger, & Karpaty, Patrik. (2007). Foreign acquisition and employment effect in Swedish manufacturing (WorkingPaper No. 10). rebro University.

Blalock, Garrick, & Gertler, Paul J. (2008). Welfare gains from foreign direct investment trough technology transfer to localsuppliers. Journal of International Economics, 74, 402421.

Blalock, Garrick, & Gertler, Paul J. (2009). How firm capabilities affect who benefits from foreign technologies. Journal ofDevelopment Economics, 90, 192199.

Blalock, Garrick, Gertler, Paul J., & Levine, David I. (2008). Financial constraints on investment in an emerging market crisis.Journal of Monetary Economics, 55, 568591.

Blomstrm, Magnus, & Sjholm, Fredrik. (1999). Technology transfer and spillovers: Does local participation with multi-nationals matter? European Economic Review, 43, 915923.

Blundell, Rrichard, & Costa Dias, Monica. (2000). Evaluation methods for non-experimental data. Fiscal Studies, 21, 427468.Braguinsky, Serguey, Branstetter, Lee G., & Regateiro, Andre. (2011). The incredible shrinking Portuguese firm (NBERWorking

Paper 17265, July). Cambridge MA: NBER.Chatani, Kazutoshi. (2012). Economic growth, employment creation, and poverty alleviation. In Hall Hill, Muhammad Ehsan

Khan, & Juzhong Zhuang (Eds.), Diagnosing the Indonesian economy: Toward inclusive and green growth (pp. 301340).London and Manila: Anthem Press and Asian Development Bank.

Girma, Sourafel. (2005). Safeguarding jobs? Acquisition FDI and employment dynamics in UK manufacturing. Review of WorldEconomics, 141, 165178.

Gong, Yundan, Grg, Holger, & Maioli, Sara. (2007). Employment effects of privatisation and foreign acquisition of Chinesestate-owned enterprises. International Journal of the Economics of Business, 14, 197214.

Grg, Holger, & Greenaway, David. (2004). Much ado about nothing. The World Bank Research Observer, 19, 171197.Hakkala, Katariina, Heyman, Fredrik, & Sjholm, Fredrik. (2008). Multinational firms and job tasks (Working Paper No. 781).

Stockholm: The Research Institute of Industrial Economics.Harrison, Ann E., & Rodriguez-Clare, Andres. (2010). Trade, foreign investment, and industrial policy for developing countries.

In Dani Rodrik, and Mark Rosenzweig (Eds.), Handbook of Development Economics. Vol. 5 ( pp. 40394214) Amsterdam:North-Holland.

Heyman, Fredrik, Sjholm, Fredrik, & Gustavsson Tingvall, Patrik. (2007). Is there really a foreign ownership wage premium?Evidence from matched employeremployee data. Journal of International Economics, 73, 355376.

Hsieh, Chang-tai, & Klenow, Peter J. (2009). Misallocation and manufacturing TFP in China and India. Quarterly Journal ofEconomics, 124, 14031448.

Javorcik, Beata S. (2004). Does foreign direct investment increase the productivity of domestic firms? In search of spilloversthrough backward linkages. American Economic Review, 94, 605627.

Karlsson, Sune, Lundin, Nannan, Sjholm, Fredrik, & He, Ping. (2009). Foreign firms and Chinese employment. The WorldEconomy, 32, 178201.

Lehto, Eero, & Bckerman, Petri. (2008). Analysing the employment effects of mergers and acquisitions. Journal of EconomicBehavior & Organization, 68, 112124.

Lewis, W. Arthur. (1954). Economic development with unlimited supplies of labour. Manchester School of Economic and SocialStudies, 22, 139191.

Lipsey, Robert E. (2004). Home- and host-country effects of foreign direct investment. In Richard E. Baldwin, and Alan Winters(Eds.), Challenges to globalization ( pp. 333382). Chicago, IL: University of Chicago Press.

Lipsey, Robert E., & Sjholm, Fredrik. (2004). Foreign direct investment, education, and wages in Indonesian manufacturing.Journal of Development Economics, 73, 415422.

Lipsey, Robert E., & Sjholm, Fredrik. (2005). Host country impacts of inward FDI: Why such different answers? In TheodoreH. Moran, Edward M. Graham, & Magnus Blomstrm. (Eds.), The impact of foreign direct investment on development: NewMeasurements, new outcomes, new policy approaches (pp. 2344). Wasshington, DC: Institute for International Economics.


Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

Melitz, Marc. (2003). The impact of trade on intra-industry reallocations and aggregate industry productivity. Econometrica, 71,16951725.

Neumark, David, Wall, Brandon, & Zhang, Junfu. (2011). Do small businesses create more jobs? New evidence for the UnitedStates from the national establishment time series. Review of Economics and Statistics, 93, 1629.

Nocke, Volker, & Yeaple, Stephen. (2007). Cross-border mergers and acquisitions vs. greenfield foreign direct investment: Therole of firm heterogeneity. Journal of International Economics, 72, 336365.

Swenson, Deborah L. (2008). Multinationals and the creation of Chinese trade linkages. Canadian Journal of Economics, 41,596618.

Takii, Sadayuki. (2005). Productivity spillovers and characteristics of foreign multinational plants in Indonesian manufacturing19901995. Journal of Development Economics, 76, 521542.

Wooldridge, Jeffrey. 2010. Econometric analysis of cross section and panel data. 2nd ed. Cambridhe MA: MIT Press.


Dow

nloa

ded

by [K

U Le

uven

Univ

ersity

Libr

ary] a

t 09:1

7 02 A

ugus

t 201

5

Employment effect of FDI

Documents

Transcript of Employment effect of FDI