Research n Busness and Economcs Journal Volume 12 Is there a relatonshp between corrupton and job creaton n resource rch ol and gas developng countres? ABSTRACT Olumde Ijose Governors State Unversty Jesus Valenca Slppery Rock Unversty of Pennsylvana Though the relatonshp between corrupton and development has attracted ncreased research attenton n the last 5 years, past studes have not segmented developng countres n a unque manner n examnng ths relatonshp. Usng data from the Economc Intellgence Unt, Transparency Internatonal and the Hertage Foundaton, ths paper examnes the relatonshp between corrupton and employment growth n ol rch developng countres between the 1999 and 2006-tme perod. The perod was selected as t represents an era of hgh ol and gas prces and n theory, an enhanced ablty to create jobs. The paper wll deepen the understandng of the mpact of corrupton on employment n emergng/developng resource rch countres. Keywords: Corrupton, job creaton, ol and gas, developng countres Copyrght statement: Authors retan the copyrght to the manuscrpts publshed n AABRI journals. Please see the AABRI Copyrght Polcy at http://www.aabr.com/copyrght.html Is there a relatonshp, Page 1
Research n Busness and Economcs Journal Volume 12 INTRODUCTION Economes are hard pressed to create good jobs at a reasonable clp, f ther resources are suboptmally utlzed. Snce corrupton creates neffcences, there should be a drect relatonshp between job creaton and the ncdence of corrupton. Corrupton s however but one force affectng job creaton. The openness of an economy also nfluences ts ablty to create jobs. If t s true, that the more nsttutonalzed corrupton s n an economy, the less open t becomes, t means that ceters parbus, the more corrupt a country s, the less compettve t becomes, and the more constraned ts ablty to create jobs becomes. The relatonshp between natural resources and slow economc growth n emergng economes has been well-demonstrated (Sachs and Warner, 1995, Gylfason, 2001). Lkewse, the negatve correlaton between slow economc growth and job creaton has been well documented. Ths fndng s counter to classc economc theory whch holds that ceters parbus, the more abundant the supply of natural resources n a country, the better for the country s economy. It s a fact that today, asde from the Unted States, the rch countres and the major emergng economes do not possess an abundant stock of natural resources. Most economsts agree that a plausble reason for the seemng contradcton s that, other mportant job creatng factors tend to be crowded out n countres wth abundant natural resources. Gylfason (2001) for nstance, ponts to a dearth of nvestment n human captal, Sachs and Warner (2001) argue that natural resources tend to crowd out manufacturng, whle Kronenberg (2004) contends that natural resource rch countres are plagued by rent seekng behavor and wdespread corrupton that negatvely affects ther economc performance. There has been no prevous emprcal work on the relatonshp between corrupton, economc openness and job creaton n ol and gas domnated economes. Ths paper analyses ths relatonshp, usng measures of economc openness and corrupton estmates from Hertage Foundaton and employment data from Economc Intellgence Unt. The study covers employment growth n the perod between 1999 and 2006 as ths represents a perod of sgnfcantly ncreased earnngs for ol and gas endowed countres. The countres studed were selected, because sales from crude ol represent a majorty of ther earnngs and because of the exstence of relable data. Ths meant that a lack of consstent data led to the excluson of nterestng countres such as Ngera, Iran, Iraq, Kuwat and Angola from the study. To control for the level of development n the selected countres, Norway, a developed country, endowed wth huge ol and gas deposts, was ncluded n the regresson equaton. The expectaton s that everythng beng equal, gven the sgnfcant ncrease durng the perod under revew, the countres should demonstrate sgnfcant gans n employment. Secton 2 of ths paper provdes a bref overvew of the lterature on corrupton and economc openness and how they may affect job creaton. The next secton descrbes the data methodology used n the regresson analyss. In secton 4, the regresson results are dscussed whle concludng remarks are presented n secton 5. WHAT IS CORRUPTION? The conceptualzaton of corrupton n ths paper ncludes all of t economc forms, publc and prvate. Publc sector corrupton, ncludes actvtes such as the blatant use of publc money for prvate gan such as trade msprcng where publc contracts are over-nvoced for prvate gan and the payment of brbes to a publc offcal to gan undue advantage such as evadng taxes, wnnng a compettve Is there a relatonshp, Page 2
Research n Busness and Economcs Journal Volume 12 contract, wnnng a tender offer, wnnng a prvatzaton bd, or wnnng access to a lcense to explot natural resources. Prvate sector corrupton ncludes, over-nvocng of mport transactons as a tool for facltatng transfer payments, evadng taxes or to conceal llegal payments, under-nvocng of mport prces to reduce mport dutes, to avod payng taxes or to foster captal flght (de Boyre, Pak and Zdanowcz, 2004). Though corrupton s an age-old problem (Bardham, 1997), there are dvergent vews on how t has mpacted natonal economes. Some researchers (Leff, 1964, Huntngton, 1968, Lu, 1985, Len, 1986) have argued that corrupton has an effcency effect n developng countres where bureaucraces tend to dstort effcent economc actvtes and to hnder entrepreneural behavor as practced n western market economes. In essence, these researchers argue that corrupton should have a postve effect on economc growth and employment. A second group (Rose-Ackerman, 1978, Shlefer and Vshny, 1993) contends that the mpact of corrupton depends on the nature of government. That corrupton hnders growth n countres where power s decentralzed but not n countres where power s centralzed. Shlefer and Vshny (1993) use ths argument n contendng that the ncrease n corrupton n Russa post the communst Sovet Unon was due to the ensung decentralzaton of corrupton. A thrd group argues that corrupton mpacts negatvely on statc effcency, nvestment and growth (Romer, 1994, Mauro, 1995, Tanz, 1998, Lete and Wedman, 1999). Emprcal studes that have found a negatve relatonshp between corrupton and ndces of economc growth nclude the works of Mauro, 1995 and We, 1997 that found a negatve relatonshp between the ncdence of corrupton and nvestment and growth and those of Kruger, 1974 and Torrez, 2002 that found a smlar effect on nternatonal trade. Ths paper s based on the hypothess that corrupton has a negatve effect on job creaton and thus, belongs to the school of thought that dscredts the argument that corrupton has an effcency effect. METHODOLOGY Data Sources The unt of analyss for the study was an ol and gas producng country, wth a country defned as a dstnct autonomous entty, able to take ndependent decsons on the structure of ts nternal nsttutonal envronment and ts domestc economc, socal and poltcal polces. Longtudnal data was collected for the perod between 1999 and 2006 from the webstes of Economst Intellgence Unt and Hertage Foundaton. Data was collected for the followng countres: Algera, Azerbajan, Kazakhstan, Mexco, Norway, Russa and Venezuela. Dependent Varable The dependent varable for the study was employment growth. Data was collected from the Economst Intellgence Unt for the perod between 1999 and 2006 and ndcates yearly changes n job creaton. Analyss was done by poolng the ndvdual country data. Countres are presumed to consttute relatvely homogenous economc/poltcal enttes, thus they are the logcal unts wthn whch to dentfy varatons n the ndependent varable, and to assess the mpact of these varatons on job creaton. Is there a relatonshp, Page 3
Research n Busness and Economcs Journal Volume 12 Independent Varables Nne ndcators of a country s ablty to create jobs were dentfed from the Hertage Foundaton s Index of Openness, Transparency Internatonal s Corrupton Perceptons Index (CPI) and the Economst Intellgence Unt Country Reports database. These ndcators nclude varables measurng economc freedom, corrupton and economc growth. As the extent of a country s economc openness, level of corrupton and rate of economc growth s expected to affect job creaton, each measure s expected to correlate wth employment growth. Retroactvely analyzng the effect of economc openness, corrupton and economc growth on employment growth n selected natural resource rch countres, wll help n understandng how these countres n general can leverage ther natural resource wealth to create jobs. From ths understandng comes dfferent theoretcal, emprcal and polcy makng attrbutes, whch wll nform future theoretcal, emprcal and polcy makng actvtes. The measurement of each of the nne varables s brefly descrbed below. Busness Freedom: The ablty to create, operate and close an enterprse quckly and easly. Trade Freedom: A composte measure of the absence of tarff and non-tarff barrers that affect mports and exports of goods and servces. Monetary Freedom: Combnes a measure of prce stablty wth an assessment of prce controls. Freedom from Government: Includes all government expendtures, ncludng consumpton and transfers. Fscal Freedom: A measure of the burden of government from the revenue sde. The measure ncludes both tax burden and the overall amount of tax revenues as a proporton of GDP. Property Rghts: An assessment of the ablty of ndvduals to accumulate prvate property secured by clear laws that are enforced by the state. Investment Freedom: An assessment of the free flow of captal especally foregn nvestment. Fnancal Freedom: A measure of bankng securty as well as ndependence from government control. Gross Domestc Product Per Capta: GDP n US$ per capta at current prces and exchange rates Source: Hertage Foundaton 2007 STATISTICAL ANALYSIS The data conssted of 10 varables collected on seven countres durng 9 years for a total of 630 observatons. The data s referred to n the econometrc lterature as panel data. The research queston s the mpact that corrupton has on the ablty of countres to generate employment. Ths meant that employment growth (empgrwt) wll be the dependent varable. The ndependent varables comprsng the data set were named as follows: Is there a relatonshp, Page 4
Research n Busness and Economcs Journal Volume 12 Gdpch = percentage change n GDP Corrupt = corrupton ndex Regul = freedom from regulaton ndex (busness freedom) Trade = trade freedom Fscal = a measure of the burden of government from the tax revenue sde. Govnterv = a measure of the degree of freedom from government. Monetstabl = a measure of prce stablty and absence of prce controls. Invest = a measure of nvestment freedom how freely captal can flow. Fnance = a measure of fnancal freedom: bankng securty and freedom from bankng regulaton Accordng to economc theory, the creaton of jobs s of course postvely related to the growth of gross domestc product (GDP) ncreases n real GDP should accompany ncreases n labor employment. The paper thus examnes corrupton as a force that reduces the effectveness of GDP growth on the premse that t causes an neffcent allocaton of scarce resources n pursut of economc actvty, that would otherwse be allocated more effcently. The ntal estmates estmated employment growth as a functon of GDP growth and the 8 varables measurng the varous economc freedoms as defned above. Empgrwt = α + β ( gdpch) + β ( corrupt) + β ( regul) + β ( trade) +... + β ( fnance) + υ + ε,t 1, t 2, t 3, t 4, t 9, t, t '' refers to country and 't' refers to year t. [1] Equaton [1] was estmated usng a fxed effects estmator. The term υ corresponds to the country specfc term that would defne a dfferent ntercept equaton for country. The term ε represents an assumed whte nose error term wth expected zero mean. The results from usng a fxed effects estmator for equaton [1] were dsappontng to say the least. Corrupton had the wrong sgn and most coeffcents were not statstcally sgnfcant. Corrupton as an Endogenous Varable Corrupton s perhaps better vewed as a varable, whch s endogenously generated alongsde forces that also determne job creaton. It s ndeed possble that corrupton affects job creaton, whch affects subsequent levels of corrupton for better, or worse. If that s the case, the ncluson of corrupton as an explanatory varable n an equaton attemptng to explan employment growth leads to smultaneous causalty bas. In that case, OLS estmates are known to be nconsstent because the explanatory varable corrupton s correlated wth the error term ε. Is there a relatonshp, Page 5
Research n Busness and Economcs Journal Volume 12 Instrumental Varables Regresson Ths estmaton technque was desgned specfcally to deal wth stuatons where for some reason an explanatory varable s correlated wth the error term n a regresson. The smultaneous causalty case nvoked here s but one of several theoretcal cases where correlaton between a regressor and the error term s the result. Possble Instruments for the varable Corrupton The dffculty of usng the nstrumental varable technque usually has to do wth fndng a sutable nstrument for the endogenous regressor n the estmatng equaton as the nstruments have to be correlated wth the endogenous varable corrupton but also have to have zero correlaton wth the error term ε n the equaton. The nstruments selected also have to be exogenous to the system of varables that determne employment growth and corrupton. Three Sutable Instruments for Corrupton The varables trade, nvest, and monetstabl appear to be sutable nstruments n that they are correlated to corrupton but can be thought of as exogenously determned. The varable trade measures a country s freedom to trade. It s a composte measure of the absence of tarff and non-tarff barrers that affect a country s level of mports and exports. Trade freedom s an economc polcy varable determned through a country s poltcal process and thus determned exogenously to the process beng modeled n ths study. The same can be sad about the varable nvest, whch measures the extent to whch fnancal captal can flow freely, especally foregn captal. The thrd nstrumental varable monetstabl measures monetary freedom n the sense that t combnes a measure of prce stablty wth an assessment of prce controls. In essence, ths posts that prce stablty and the degree of prce controls n the economy are also economc polcy varables that are determned outsde the system beng examned n ths study. An Instrumental Varable Approach: Statstcal Results In essence, the method of nstrumental varables s a two stage least squares estmatng technque. Table 1 shows the results of the nstrumental varable approach. The varable country represented the cross sectonal dmenson and the varable year represented the tme-seres dmenson n the data. The results shown n table 1 are the result of usng a random effects estmator of the coeffcents. The selecton of the random effects estmator can be defended because the sample of seven countres s but a subset of the ol exportng countres for whch nference are beng made regardng the relatonshp between corrupton and ts mpact on job creaton. Is there a relatonshp, Page 6
Research n Busness and Economcs Journal Volume 12 TABLE 1 Frst-stage G2SLS regresson Number of obs. = 61 Wald ch (4) = 28 Prob > ch2 = 0.0000 corrupt Coef. Std. Err. z P> z [95% Conf. Interval] gdpch -.506.359-1.41 0.159-1.211.198 monetstabl.437.141 3.10 0.002.160.713 trade.446.250 1.78 0.075 -.044.937 nvest.296.158 1.88 0.061 -.013.606 _cons -29.630 16.190-1.83 0.067-61.362 2.102 G2SLS random-effects IV regresson Number of obs = 61 Group varable: country Number of groups = 7 Obs per group: mn = 8 avg = 8.7 max = 9 F (2,59) = 4.77 Corr (u_, X) = 0 Prob > F = 0.012 empgrwt Coef. Std. Err. t P> t [95% Conf. Interval] corrupt.145.054 2.67 0.010.036.255 gdpch.194.087 2.21 0.031.018.369 _cons -4.695 2.376-1.98 0.053-9.449.059 sgma_u 1.429 sgma_e 3.458 rho.1454 (fracton of varance due to u_) Analyss of the results The results of the frst stage n ths two-stage least-squares estmatng technque shows that corrupton s well explaned by the nstruments monetstabl, trade and nvest. All coeffcents for the 3 nstrumental varables are statstcally sgnfcant. Accordng to the estmates: A. Growth of real GDP reduces the ndex value for corrupton, whch means that corrupton ncreases. B. Increased prce stablty and ncreased freedom from prce controls ncreases the ndex value of corrupton, whch means that corrupton falls. Is there a relatonshp, Page 7
Research n Busness and Economcs Journal Volume 12 C. Increased trade freedom ncreases the ndex value of corrupton, whch means that corrupton falls. D. Increased nvestment freedom ncreases the ndex value of corrupton, whch means that corrupton falls. It s reassurng that greater nvestment and trade freedom reduces corrupton and that prce stablty and freedom from prce controls do the same. The second stage n ths nstrumental varable estmatng technque uses the results from the frst stage n order to estmate the relatonshp of prmary nterest n ths study: how corrupton affects employment growth. The results n table 1 show that all coeffcents are statstcally sgnfcant. The coeffcent on corrupton s sgnfcant at the 1% level of confdence, and those on gdpch and the constant term are sgnfcant at a 5% level of sgnfcance or better. Accordng to the results: A. An ncrease n the ndex value of corrupton (a reducton n corrupton) ncreases employment growth. The margnal effect beng an ncrease of 0.14 ncrease n employment growth for a onepont ncrease n the ndex value of corrupton (a reducton n the level of corrupton). B. The elastcty of employment growth wth respect to changes n corrupton was estmated at 2.876. Ths elastcty was measured at the mean level of corrupton for all countres n the sample (37.67). Ths estmate shows that there are sgnfcant returns to fghtng corrupton. C. The elastcty of employment growth wth respect to changes n real GDP growth was estmated as 0.581. Ths elastcty was measured at the mean level of DPG growth for all countres n the sample (5.72%). Country-Specfc Inferences The random effects estmatng technque facltated the estmaton of the followng equaton for each country n the sample: Empgwth = -4.695 + 0.1942( Gdpch) + 0.1458( Corrupt) + ν + ε [2],t, t, t, t where the subscrpt denotes country and t denotes year t For purposes of llustraton, let us refer to equaton [2] as y, t = xβ + ν + ε, t. Under a random effects estmator the coeffcent ν s a country specfc random error derved from the characterstcs partcular to country. Table 2 shows the means for each of the elements of equaton [2] by country as well as the means for corrupton and gdp change. The table also shows the employment growth rates estmated by the model: xβ + ν. Is there a relatonshp, Page 8
Research n Busness and Economcs Journal Volume 12 TABLE 2 Country Corrupt gdpch Constant xβ ν xβ + ν Algera 44.777 4.188-4.695 2.649 0.861 3.511 Azerbajan 21.333 13.711-4.695 1.079-0.234 0.844 Kazakhstan 21.444 8.122-4.695 0.010 0.434 0.444 Mexco 42 3.222-4.695 2.056 2.579 4.636 Norway 88.777 2.588-4.695 8.756-7.934 0.822 Russa 27.333 5.433-4.695 0.346 0.186 0.533 Venezuela 18.555 2.544-4.695-1.494 4.216 2.722 The levels of corrupton shown on table 2 are averages for each country durng the years sampled: 1998 through 2006. In terms of corrupton the most corrupted n the sample s Venezuela (18.55) followed by Azerbajan (21.3), Kazakhstan (21.4), Russa 27.3), Mexco (42), Algera (44.7) and far away from the rest, Norway (88.7). Table 3 shows the employment growth rates that would have taken place had the ndex of corrupton for every country been 10 ponts greater than they actually were durng the years sampled, all else beng equal. That means f levels of corrupton had been less than they actually were. TABLE 3 %Change %Change Corrupt gdpch constant xβ ν xβ + ν emplymt corrupt Elastcty Algera 54.777 4.188-4.695 4.108 0.861 4.969 41.541 22.332 1.860 Azerbajan 31.333 13.711-4.695 2.537-0.234 2.303 172.719 46.875 3.684 Kazakhstan 31.444 8.122-4.695 1.468 0.434 1.903 328.184 46.632 7.037 Mexco 52 3.222-4.695 3.515 2.579 6.095 31.458 23.809 1.321 Norway 98.777 2.588-4.695 10.215-7.934 2.280 177.394 11.264 15.748 Russa 37.333 5.433-4.695 1.805 0.186 1.991 273.485 36.585 7.475 Venezuela 28.555 2.544-4.695-0.035 4.216 4.180 53.580 53.892 0.994 As table 3 shows, there appears to be great returns to fghtng corrupton. The emprcal analyss ndcated an elastcty of 2.87 on average for a 1% reducton n corrupton by all countres. That elastcty was however estmated at 37.67, the average level of corrupton for all countres n the sample. The results ndcate that the returns on fghtng corrupton are vared gven the country-specfc estmates obtaned n the model. The percentage change n employment s of course greatly affected by the employment performance each country experenced durng the perod sampled. Thus, those wth hgh employment growth rates between 1998 and 2006 stand to gan smaller returns fghtng corrupton. Venezuela, the most corrupt n the sample, gans the least smply because a 10-pont ncrease n the corrupton ndex represents a large reducton n corrupton and ts employment growth performance durng the perod sampled was relatvely good. The hgh return on corrupton reducton that Norway experences, the least corrupt country n the sample, s due to the small reducton n corrupton Is there a relatonshp, Page 9
Research n Busness and Economcs Journal Volume 12 (percentage-wse) that a 10-pont ncrease would brng coupled wth a relatvely poor employment growth record between 1998 and 2006. CONCLUSION Ths study examned the relatonshp between corrupton and economc openness on job creaton n a subset of countres. The result suggests that there should be lttle doubt that economc openness and reductons n the ncdence of corrupton wll have a postve mpact on job creaton, as consstent wth predctons from the openness, growth, rent seekng and other lterature (Torrez, 2002; Kaufman, 1997; Tanz and Davood, 1997), the emprcal results demonstrated the expected negatve relatonshp between corrupton and employment creaton. Interestngly, the analyss shows that a reducton n the ncdence of corrupton leads to an ncrease n the job creaton capablty of a country. Granted that corrupton cannot be reduced solely by wll power or by an understandng of ts systemc negatve consequences, t can be argued that polcy makers need to put a genune market based system n place, f they are serous about fghtng corrupton. Although, the data covered only a subset of all ol and gas domnated economes, there s strong reason to beleve that the results would hold true f data had been avalable on all of them. In fact, the results suggest that t would only have been strengthened wth the avalablty of relable data on countres such as Ngera, Angola, Equatoral Gunea, Chad and Sudan. Research on the underlyng poltcal, soco-cultural, nsttutonal and hstorcal factors undergrdng corrupton n ol and gas rch economes would also be useful. Such research wll help us understand the extent to whch fnancal nflows and external pressures can n realty provde a sgnal to these countres to put n place, real market based economc systems and workable nsttutonal checks and balances. REFERENCES Bardham, P. (1997). Corrupton and Development: A Revew of the Issues. Journal of Economc Lterature: XXXV, 1320-1346. de Boyre, M., Pak, S.J., and Zdanowcz, J. (2004). Money Launderng and Income Tax Evason: The Determnaton of Optmal Audts and Inspectons to Detect Abnormal Prces n Internatonal Trade. Journal of Fnancal Crme: 12:2, 123-130. Gylfason, T. (2001). Natural Resources, Educaton and Economc Development. European Economc Revew: 45(4-6), 847-859. Huntngton, S. (1968). Poltcal Order n Changng Socetes. New Haven: Yale Unversty Press. Kaufmann, D. (1997). Corrupton: the Facts. Foregn Polcy: 107, 114-31. Krueger, A. 1974. The Poltcal Economy of the Rent-Seekng Socety. Amercan Economc Revew: 64, 291-303. Leff, N. (1964). Economc Development Through Bureaucratc Corrupton. The Amercan Behavoral Scentst: 8(2), 8-14. Is there a relatonshp, Page 10
Research n Busness and Economcs Journal Volume 12 Len, D. (1986). A Note on Comparatve Brbery Games. Economc Letters: 22, 337-41. Lete, C. and Wedman, J. (1999). Does Mother Nature Corrupt? Natural Resources, Corrupton and Economc Growth. IMF Workng Paper WP/99/85. Lu, F. (1985). An Equlbrum Queung Model of Brbery. Journal of Poltcal Economy: 93(4), 760-81. Mauro, P. (1995). Corrupton and Growth. Quarterly Journal of Economcs: 11093, 681-712. Romer, P. (1994). New Goods Old Theory and the Welfare Costs of Trade Restrctons. Journal of Development Economcs: 43(1), 5-38. Rose-Ackerman, S. (1978). Corrupton: A Study n Poltcal Economy. New York: Academc Press. Sachs, J.D. and Warner, A.M. (2001). The Curse of Natural Resources. European Economc Revew: 45(4-6), 827-838. Shlefer, A and Vshny, R. (1993). Corrupton. Quarterly Journal of Economc, 108(3), 599-617. Tanz, V. (1998). Corrupton around the World: Causes, Consequences, Scope and Cures. IMF Staff Papers: 45(4), 559-94. Tanz, V. and Davood, H. (1997. Corrupton, Publc Investment, and Growth. IMF Workng Paper. WP/97/139. Torrez, J. (2002). The Effect of Openness on Corrupton. The Journal of Internatonal Trade & Economc Development. 11(4) 387-403. We, S.J. 1997. How Taxng s Corrupton on Internatonal Investors? Natonal Bureau of Economc Research Workng Paper, 5755: May Authors Note Data for ths paper was collected from the 2007 Hertage Foundaton Index of Economc Freedom and the Economst Intellgence Unt. As much as the authors would have lked to nclude every ol and gas domnated economy n the emprcal analyss, the countres n the dataset were those wth consstent data across the years studed. As such, nterestng countres lke Ngera and Saud Araba were not ncluded n the emprcal analyss. Is there a relatonshp, Page 11