Volume 26, Number 2, December 2001 Determnants of Prvatsaton n Selected Sub-Saharan Afrcan Countres: Is Prvatsaton Poltcally Induced? Osman Sulman and Ghrmay Ghebreysus Whle Afrcan governments and nternatonal donors generally support prvatsaton for stablsaton as an mperatve of publc fnance problems, academcans are more nclned to dscuss the effcency gans of prvatsaton. Antthetcally, some poltcal economsts argue that whereas Afrcan governments actually de-emphasse prvatsaton, donors manly nsst on prvatsaton to promote neoclasscal vews wthout offerng an alternatve to state reform. Ths paper realses that the man purposes of prvatsaton n Sub-Saharan Afrca have, so far, been multdmensonal. It envsages, emprcally, the determnants of prvatsaton n Sub-Saharan Afrca usng a probt model over the perod 1970-1994. The results are supportve of the hypothess that prvatsaton n the sample countres s nduced by macro -nstablty and poltcal bas. I. Introducton The lterature on prvatsaton n Sub-Saharan Afrca and developng countres n general s very complex and multdmensonal. There s no clear consensus among governments, nternatonal organsatons, and academcans as to what exactly s the purpose of prvatsaton. Whle Afrcan governments, the IMF, and the World Bank (donors) seem to target manly stablsaton as an mperatve of publc fnance problems wth effcency as a secondary axomatc belef, much of the academc lterature emphasses the effcency and reversal of prce dstortons as gans of prvatsaton. On the other hand, as envsaged n Secton III, some poltcal economsts have serous doubts about the ntentons of both Afrcan governments and donors. Accordng to these economs ts, whle Afrcan authortaran governments de-emphasse prvatsaton because of power loss, donors have manly nssted on prvatsaton to promote neoclasscal vews of state mnmsaton wthout offerng an alternatve approach to state reform. Prvatsaton programs have been ntroduced by a number of Afrcan countres n the last two decades wth the ostensble vew to mprovng economc performance of publc enterprses and creatng a market economy. Yet the programs produced mxed results. In most of these countres poorly performng publc enterprses were sold to the prvate sector Economcs Department, Mllersvlle Unversty, P.O.Box 1002, Mllersvlle, PA 17551-0302, U.S.A. and Gramblng State Unversty, respectvely. The authors would lke to thank an anonymous referee for helpful suggestons. 33
to releve the government of heavy fnancal burdens and obvate macroeconomc problems. In fact, as shown below, t has been alleged that most Sub-Saharan Afrcan countres essentally prvatse because of condtonal assstance provded by the World Bank and the Internatonal Monetary Fund (IMF) subject to prvatsaton and other macroeconomc adjustments. 1 To mantan ther power whle submttng to such pressure, many authortaran governments n Sub-Saharan Afrca apply prvatsaton by favorng ther alles. The expectaton, therefore, s that X-neffcency mght result n both the prvate and publc sectors, or state-owned enterprses (SOEs). In addton, as Due (1993, pp. 1981-1988) ponts out by not exposng parastatals to open competton, a worsenng of ncome dstrbuton usually ensues. It has been argued by Bennell (1997, pp. 1785-1803) that publc floataton of SOE shares has the added poltcal and economc advantages of enablng relatvely large numbers of ctzens to acqure them. Otherwse, deferred publc offerngs, where the government stpulates that a certan percentage of shares wll be sold to the publc as tme passes, can be consdered. In the absence of stock exchanges, trust funds can be establshed that allow the government to keep SOE shares untl they can be sold to the publc. In fact, n some of these countres, prvatsaton polces were enacted for deologcal and opportunstc goals. As Rddell (1999, pp. 309-335) contends, t s naïve to beleve that government wll execute ncome redstrbuton reforms that wll challenge ther poltcal nterests. In a general sense, the prvatsaton programs dd not produce the desred effcency gans snce the ncentve was to reduce government defct and to advance poltcal agendas rather than offerng an opportunty to buld a strong prvate sector. Ths study s, therefore, concerned wth assessng the extent to whch prvatsaton decsons are subject to macroeconomc changes and poltcal bas n selected Sub-Saharan Afrcan countres. The proclamed effcency gans of prvatsaton have been dscussed extensvely by numerous artcles, ncludng Brock and Magee (1978, pp. 246-50), Becker (1983, pp. 371-400), Ake and Mohammed (1986, pp. 161-80), Clague (1991, pp. 507-30), Fernandez and Rodrk (1991, pp. 1146-55), Bruton (1992), Dollar (1992, pp. 523-44), Bates and Krueger (1993), Bradburd (1993), Drazen and Grll (1993, pp. 598-607), Campos and Esfahan (1996, pp. 451-85), and Rose-Ackerman (1998, pp. 35-57), among others. To our knowledge, none of these studes focused on the determnants of prvatsaton n Sub- Saharan Afrca. We conjecture that economc polces follow a set of contracts between the government (prncpal) and varous nterest groups (agents) that desre the promoton of a free market system or effcency gans (proft maxmsaton). Alternatvely, prncpal-agent contracts work under both complete certanty and uncertanty wth and wthout perfect nformaton and can form the bass of analyss. The theoretcal prncpal-agent (P-A) model s ntended to exhbt the expected proft maxmsaton (effcency) condtons. At the outset, t becomes evdent that proft-sharng s the underlyng latent condton for the decson to prvatse subject to (explaned by) poltcal motves and the nevtable macroeconomc problems. Specfcally, whle ths study agrees and attempts to llustrate theoretcally the effcency benefts of prvatsaton, ts man purpose s to attract attenton, emprcally, to the 1. For an explanaton of polcy-based lendng n Sub-Saharan Afrca, see Cassen (1994). 34
SULIMAN AND GHEBREYSUS: DETERMINANTS OF PRIVATISATION macroeconomc determnants of prvatsaton and the poltcal manner n whch prvatsaton s appled. Demonstraton of both the agreed upon (theoretcal) effcency gans and the clamed emprcal questons are crucal as we allude to the resultng perncous ncome dstrbuton effects whch may offset any effcency gans. In addton, Adam (1992) demonstrates that the effcency gans from the reducton of agency problems s mu ch more complex. Adam succnctly dstngushes between productve and allocatve effcency. He argues that prvatsaton can ndeed lead to lower costs due to the ownershp shfts and change n ncentve structures whch are related to agency problems. Besdes, Adam argues that the ssue of allocatve effcency, whch algns consumer prces wth the margnal cost of producton, s closely lnked wth the ssue of market lberalsaton and competton whereas state enterprses are ether n natural monopoly sectors or present serous barrers to entry due to ther sze and advantageous access to technology and other nputs. Thus, for prvatsaton to be successful, t has to be lnked to competton and regulaton polcy whch has been gnored by both donors and the Afrcan governments. That s, unless delberate prvate-publc corrupton of some Afrcan governments amng at avodng competton to reap monopoly rents s elmnated, prvatsaton s not lkely to succeed. Therefore, the effcency ssue may have largely been a secondary axomatc belef n the context of structural adjustment. Thus, to a large extent, the man focus has been on stablsaton polces amed at the publc fnance mperatve of adjustment where SOEs have been a net dran on the fnancal resources of the government. Theoretcally, as Adam (1994) enuncates, under prvate ownershp, frms wll only reman n exstence as long as they are vable; otherwse the market mechansm wll reallocate ther resources to more effcent uses. Practcally, poltcal bas and the alleged stablsaton of the macroeconomy are, n essence, externaltes that stand n the way of the market mechansm. Hence, the present paper envsages emprcally two questons: (1) s prvatsaton nduced by macroeconomc problems (varables)? and (2) s prvatsaton poltcally nduced? The queston of ncome redstrbuton effects resultng from poltcal bas, and whether or not prvatsaton leads to some reversal of prvate dstortons, can be tackled elsewhere. Adam et al. (1992), examne prvatsaton from an explctly economc standpont, focusng more on the specfc effects of ownershp transfer on economc performance as appled to the market structures and characterstcs of many developng countres. Evdence derved from the apparent oppressve condtons n most of those countres and the ncreasng patterns of mgraton out of them suggest a worsenng of ncome dstrbuton. For example, n Sudan the publc enttes have essentally been sold to supporters of the regme, whle successful tradtonal busnesses have been overtaken and forced to leave ther enterprse to party afflates. The UNDP report reveals that n 1997 alone, 374.4 thousand Sudanese sought asylum n other countres, and n 1994, 282.6 (per 100,000 people) were jaled. The next secton wll present a bref descrpton of the underlyng prncpal-agent model. Secton III dscusses the motves behnd prvatsaton and the poltcal economy of the related ssues. Secton IV presents the emprcal results. Secton V offers some concludng remarks. 35
II. The Prncpal Agent Model: An Overvew Although t s more commonly argued that the publc sector s more X-neffcent, our study purports that n a poltcal allance scenaro, both the publc and prvate sector are subject to X-neffcency. The neoclasscal explanaton focuses on agency theory, whch, as Haskel and Sanchs (1995, pp. 301-320) explan, examnes the relatonshp between managers/owners and model manageral effort and determnes effort n a contractual arrangement under asymmetrc nformaton. As dscussed n Bös (1991) and Vckers and Yarrow (1993, pp. 111-32), these models predct that X-neffcency s greater n the publc sector f montorng arrangements are nferor to the prvate sector. That s, f asymmetrc nformaton exsts. However, the present paper takes the extreme, but realstc, assumpton of symmetrc nformaton on the premse that when the prvate sector and publc sector are poltcally alled for the same cause, they are nseparable. Ths accords wth Johnston s (1998, pp. 69-90) argument that n such entrenched corrupton settngs many ctzens and nvestors see corrupton as nevtable and reform as futle. As such, entrenched corrupton dverts development resources and saps the poltcal and socal vtalty that socetes need. It also ressts many of the nsttutonal and cvl servce reforms commonly deployed n advanced natons. That s, as Johnson adds, entrenched corrupton s pervasve, organsed, and monopolstc wth no meanngful poltcal opposton or economc competton. After elmnatng ther compettors, the alled prvate and publc sector offcals can extract monopoly rents as enuncated n the Afrcan Development Report (1997, pp. 79-164). As Kltgaard (2000, pp. 2-5) explans, systematc corrupton dstorts ncentves, undermnes nsttutons, and redstrbutes wealth and power to the undeservng. Kltgaard argues that when corrupton undermnes property rghts, the rule of law, and ncentves to nvest, economc and poltcal development are crppled. Rent-seekng s further compounded n these countres by dscretonary allocaton of foregn currency and foregn currency black markets. In the model the government acts as the prncpal (acton-taker) and the collaboratve entrepreneurs as agents. As such, prvatsaton actons are taken by the government. The extent of effcency (neffcency) wll depend on the expedency wth whch the acton s taken. To the extent that government poltcses prvatsaton to gan more control by favorng nterest groups that are loyal to the ncumbent authortes, prvatsaton may yeld perverse results. By tryng to empower only loyal nterest groups, the government s, n essence, movng to secure ts poltcal power by ganng full control of the economc actvty of relevant consttuents. That s, the authortes are delberately enterng nto a contract to undertake economc actvty. Wth the combnaton of rsk sharng and access to nformaton, the prncpal-agent problem offers a frst-best soluton f the nterests of the two sdes converge whle nformaton s shared costlessly. Under nformaton asymmetry, however, a frst-best soluton s possble f and only f the prncpal mo ntors the agent s actons and gans free access to nformaton; otherwse second-best solutons are possble wth the so-called agency loss arsng. In the Afrcan stuaton, however, there s a vrtual symmetry of nformaton and shared X-neffcency snce the agent and the prncpal are nseparable alles through ethnc or relgous bases. 36
SULIMAN AND GHEBREYSUS: DETERMINANTS OF PRIVATISATION It s generally presumed that the outcome s sgnfcantly nfluenced by the agents, and that agents actons are not readly observable by the prncpal. The prncpal can control some observable behavor of the agent through the nvestment (output) acqured by the prncpal and a known channel of predetermned contract whereby the prncpal observes the actons of the agents, through poltcal support and so forth. When agents actons are based on unshared observaton wth the prncpal, the prncpal cannot detect whether the agent used hs nformaton n a way that best serves the prncpal s nterests. Thus, we run nto a moral hazard problem. Plausbly, the prncpal has to determne a contractng rule that wll nduce the agent to behave n an optmal convergng soluton, know as ncentve compatblty. 2 III. Emprcal Issues: Why Prvatse? Prvatsaton programs have been ntroduced by a number of Afrcan countres n the 1980s and the 1990s. The motves behnd prvatsaton are complex and controversal. Importantly, the bulk of researchers are more nclned to beleve that most Afrcan prvatsatons take place as part of a broader structural adjustment program under the auspces of nternatonal development agences or the so-called donors, ncludng the World Bank and the Internatonal Monetary Fund (IMF). 3 Both the World Bank and the IMF are concerned wth the necessary macroeconomc stablty as they see t, whch usually entals fscal dscplne and prvatsaton. Among researchers are those who queston these donor polces and dscount the exstence of entrenched corrupton settngs. Ths alters the adduced causal nferences n ths paper and prncpal-agent relatonshps. Such a vew s enuncated by Afrcansts such as Callaghy (1989, pp. 15-38), allegng that to cope wth the Thrd World debt crss, Western states and actors have brought to promnence the domnant neoclasscal vew of development to countres dependent on the World Bank and the IMF. 2. Pratt and Zechauser (1985) and Ul Haque and Mrakhor (1987) pont out that the followng prncples came as a result of the prncpal-agent analytcal framework: (a) when nformaton access s expensve and the agency loss s hghest, the nterests of the prncpal and agent do not converge; (b) an optmal level of montorng s mantaned by the prncpal subject to the cost of montorng; (c) under relatvely expensve montorng, montorng wll be reduced n quantty and/or qualty; (d) wthn an array of real-lfe cases, successful montorng of such ndcators as nvestment (proft) s at least partly effectve n the reducton of agency loss; (e) good reputaton whch could be lost through napproprate or ltgous behavor; (f) repettve actons wll, over tme, reduce uncertanty and mtgate the ncdence of moral hazard; and (g) the benefts that accrue, as aresult of agency-loss reducton, wll be shared by both the prncpal and the agent. Hence, both the prncpal and the agent are nterested n fndng an ncentve and montorng plan that gves outcomes smlar to a free-nformaton stuaton. The role of moral behavor n reducng agency loss has been succnctly addressed by Arrow (1970, 1971). 3. Whle many economsts perceve structural adjustment programs adopted by the World Bank and IMF as marketorented approaches of reform (ncludng prvatsaton), these have been crtcsed by a number of Afrcan poltcal economsts as a dscourse and practce of post-colonal nternatonal economy and not merely an economc program. See, for example, Blackng (1999, pp. 207-227). 37
Central to ths vew are export-led growth and mnmalsts state, wth the market beng the major nstrument of reform. Yet, Kahler (1990, pp. 33-62) ponts out, that except for tryng to remove the state from the economy, neoclasscal theory does not offer an alternatve approach to state reform. Ths has posed a major mpedment to effectve economc adjustment n the Thrd World. It has, furthermore, been contended that the IMF and World Bank actons are antthetcal to what they recommend. As Callaghy (1989, pp. 115-38) argued, much of ther actvty renforces the role of the state to mprove montorng, data-gatherng, parastatal reform, and prvatsaton of SOEs, among other thngs. It should be mentoned here that ths paper argues that because Afrcan countres need these donors fnancal support as a result of perverse macroeconomc condtons, they submt to donors pressure to reduce government and prvatse. To make sure that prvatsaton does not dmnsh ther power, they apply t n a poltcal way. Both Callaghy (1989, pp. 115-38) and Kahler (1990, pp. 33-62) contend that prvatsaton was not popular among poltcans n Sub-Saharan Afrca before the 1990s. Poltcans and managers of SOEs were afrad of losng power and jobs. Moreover, labor unons, whch are mostly concentrated n SOEs were generally opposed to prvatsaton. Callaghy (1990, pp. 257-320) argued that as ndependence came to Afrca n the late 1950s, poltcal rather than economc logc prevaled. He contends that a new poltcal class began to emerge that used the state as ts nstrument of acton and source of power, status, rents, and other forms of wealth. The state was used to begn buldng an economc base for tself. Thus, n contrast to the onerous prvate captalst actvty, all groups concentrated on the weak and vulnerable, newly autonomous state to reach power and wealth. The result was a formaton of networks to buld support through rent dstrbuton, creaton of a large parastatal sector, and purchase of urban support va state welfare servces and subsdes. In ths regard, although Callaghy s correct n questonng the prevalence of poltcal motves of the rulng class, t can, nonetheless, be argued that corrupt poltcans manly contnued along the same way of power and wealth formaton nherted from ther predecessors, the ex-colonsers. Notwthstandng, Callaghy s (1990, pp. 257-320) pont that new poltcal groups attempted to use rents from the state to reward alles and, at the same tme, crosscut and control complcated ethnc, regonal, relgous, clentelstc, and emergng class tes, s well taken. As a matter of fact, one major task of ths paper s to nvestgate whether or not prvatsaton s poltcally-nduced, and whether or not t s determned by donors' macroeconomc stablsaton polces. In the 1990s, however, many governments went a long way toward prvatsaton. Gven poltcal lberalsaton and mult-party electons n some countres, such as South Afrca, Tanzana, and Zamba, prvatsaton allowed newly elected governments to change prevous statst polces. Mostly, the prolferaton of prvatsaton n the nnetes was a result of the relentless pressure of the World Bank and the IMF. As Callaghy (1989, pp. 115-38; 1990, pp. 257-320; 1993, pp. 463-519) and Toye (1991, pp. 151-200) pont out, rrespectve of many prvatsaton falures n Afrca, the World Bank and the IMF rushed to clam the success of reform endeavors. Several factors are mentoned by the World Bank Report (1995) as testmony for ths success. Two mportant contrbutors to success are dvesture (degree of prvatsaton) and mposton of hard budget constrants (spendng cuts or government shrnkage) on SOEs n leu of the gradual declne of drect transfers, elmnaton 38
SULIMAN AND GHEBREYSUS: DETERMINANTS OF PRIVATISATION of drect and ndrect subsdes, mprovng the prce mechansm, and stpulaton (control) of credtworthness (loans) by lenders. Another mportant factor s admnstratve reform, through hgher manageral expectatons and more autonomy gven to SOE managers. Yet as Bennell (1997, pp. 1758-1803) argued, reform of SOEs becomes all the more mportant gven that attempts of restructurng them, usng performance and management contracts wth exstng managements, have not been successful. The World Bank report (bd.) ndcated that fnancal sector reform was crucally mportant n prvatsng SOEs. Clark (1998) reports that n 1990 Afrcan stock exchanges (excludng the domnant market of South Afrcan) had a total market captalsaton of $8.7 bllon, or 5.9 percent of the contnent s total captalsaton. By 1995 ths had ncreased to $26.9 bllon, or 8.8 percent of Afrca s market captalsaton. The non-south Afrcan markets have rsen a cumulatve 211 percent over the fve-year perod. However, as Callaghy (1990, pp. 257-320) and Lpumba (1994) ndcate, most of the ownershp (shares) transfer has been acqured by the government s close supporters. They are essentally the ones who have money to partcpate n the market. It s noteworthy that n hs crtcal analyss of the World Bank s (1994) report, Lpumba (1994) ponts out that the domnant opnon among ntellectuals n Afrca s that structural adjustment programs are part of the problem rather than of the soluton to the Afrcan economc crss. The World Bank report (1994) argues that the growth rate of output (percentage change n per capta real GDP) remans low even for countres that have mproved ther macroeconomc performance because adjustment efforts have not gone far enough n areas of publc enterprse reform and prvatsaton. The problem wth ths argument, as Lpumba (bd) contends s that t s unrealstc to expect success of prvatsaton durng the md-1980s, the same perod when macroeconomc adjustments were mplemented. In the md-1980s, the focus of reforms was on macroeconomc stablsaton and exchange rate adjustment. Part of the stablsaton package s prvatsaton (hgher unemployment of workers prevously workng for SOEs) and devaluaton of currency whch means lower real wages (hgher nflaton). Ths (together wth exchange controls) leads to a deteroraton of the real exchange rate and foregn currency black markets prolferaton. IV. Emprcal Analyss Evdence from the above dscusson of several Sub-Saharan Afrcan economes suggests that sluggsh economc performance that usually precedes polcy-shftng of transferrng publc enttes nto prvate enterprses, forces those countres to reluctantly follow World Bank and IMF prvatsaton polces. It, furthermore, ndcates that governments, n order to mantan ther power, usually sell publc enttes to ther alles based on ethnc (trbal) and/or relgous afflatons. To be sure, many of the Afrcan countres have authortaran regmes that mpose a certan governng group wth the ntenton of ndoctrnatng certan relgous or trbal belefs. To ensure complete power, prvatsaton has often been used to control the country economcally and poltcally. Examples abound. Table 1 presents a lst of twenty-one such countres. Almost all of the Afrcan countres wth authortaran governments have adopted prvatsaton on pure poltcal grounds. Therefore, although prvatsaton s ostensbly performed for effcency purposes, the underlyng motvatons are, often-tmes, poltcal. 39
Table 1 Sze and Suspected Motvaton of Prvatsaton, 1995 Country Sze of Publc Sector (% of GDP) Motvaton Angola 64.1 ethnc Burund 13.0 ethnc Cameroon 21.6 ethnc Central Afrcan Republc 26.6 ethnc Chad 29.4 ethnc Congo, People s Republc of the 38.8 n/a Côte d Ivore 24.8 ethnc Djbout 42.0 n/a Ethopa 27.6 ethnc Ghana 19.6 ethnc Lbera 50.5 ethnc Madagascar 13.3 ethnc Mozambque 43.4 ethnc Ngera 15.8 relgous Rwanda 112.4 ethnc Somala 51.3 ethnc Sudan 25.0 relgous Uganda 15.1 ethnc Zare 32.6 ethnc Zamba 32.6 ethnc Zmbabwe 45.0 ethnc Source: Afrcan Development Bank, Afrcan Development Report 1997 and nformaton from varous sources. Do the features of poltcal motves explan why these poltcal motvatons resulted n prvatsaton of publc enterprses? The answer can be clarfed by conductng a probt analyss of prvatsaton endeavors condtonal on poltcal motves and macroeconomc characterstcs. The data ncludes the twenty-one countres ncluded n Table 1. Annual data over the perod 1970-1994 has been used. Prvatsaton efforts and publc reform n most of these countres started n the 1970s and ntensfed n the 1980s and 1990s. Therefore, for comparatve purposes the sub-sample 1980-1994 has been examned. The dependent varable ( Y ) s the decson (probablty) of whether to prvatse or not, gven the country s characterstcs or the values of the explanatory varables (poltcal motves and the specfed macroeconomc varables). The dependent varable s a dummy varable where Y = 1 f the decson to prvatse s sgnfcantly dependent on the descrbed characterstcs (or the sx explanatory varables); Y = 0, otherwse. Y, or the probablty of prvatsaton decson s related to sx varables lsted n Table 3. These varables are the sze of the prvate (or alternatvely publc) sector (SP), per capta real GDP growth ( GDP), spendng on defense (SD) - a measure of authortaran power and budget defct problems, changes n the real exchange rate (RE) - a measure of nflaton and ndcator of the sze of the nformal foregn currency market, control of bankng (CB) - 40
SULIMAN AND GHEBREYSUS: DETERMINANTS OF PRIVATISATION access to lendng (an ndcator of banks (whch are manly government controlled) support to government alles, and an ndrect ndcator of nflatonary pressures), and afflaton to a party or ethnc group (AP). AP represents a dummy ndependent varable X, consdered as follows: X = 1, f poltcal motves (relgous, ethnc, etc.) resulted n prvatsaton; X = 0, otherwse. To do ths, numbers have been assgned to the poltcally motvated cases defned n Table 1. Table 2 Summary of Prvatsaton of Publc Enterprses (1995-1999) Transactons Completed 95 96 97 98 99 Agrculture 41 Fnancal Transactons by Sector Manufac - turng Servces Trade Angola.. 56...... 9.. 29 9 9 275 Burund.. 13 3.... 21 4 6 9 2.. Cameroon 3 23 5 4.. 24 4 7 12 1.. Central Afrcan Republc.......... 8 4 9 14.... Chad 13 6...... 9 5 6 8 3.. Congo, Dem. 44 15...... 20 5 14 22.... Congo, Rep. 2 3...... 3 1 6 10 1.. Other Côte d Ivore 27 21 15 11 6 40 5 33 25 2.. Djboute...................... Ethopa 9.... 1...... 6 3 1 115 Ghana 23 20 27 4 8 44 8 101 50 30.. Lbera...................... Madagascar 24...... 1 31 4 33 14 4.. Mozambque 112 38 31.... 75 2 256 120 126.. Ngera.......... 23 24 23 10 1.. Rwanda.......... 1.. 1...... Somala...................... Sudan.......... 11 1 8 11 1.. Uganda 23 18 2.. 3 19 5 29 26 12.. Zamba 60 91 55 16 10 71 6 69 102 20.. Zmbabwe.... 3.. 1 3 1 1 1.... Source: The World Bank, Afrcan Development Indcators 2001, p. 259. Note: The rest of the sample countres data s not avalable. The sze of the publc sector varable (SP) (or government defct as a rato of GDP) s ntended to ndcate the extent to whch the exstng degree of publc defcts mpnges on the decson to prvatse. It has been argued that prvatsaton has been drected toward the publc fnance mperatve of macro-stablty where SOEs have been a net dran on the coffers of treasures. The spendng on defense varable (SD) s meant to show the extent to whch governments n these developng Afrcan economes are nsecure and would lke to stay n
power at all expense, thus partly explanng ther poltcal motves to prvatsaton. Importantly, large spendng on defense n most of these Afrcan countres has compounded the publc debt problems. The real exchange rate (RE) s taken as the rato of the foregn prce (tmes the nomnal exchange rate) dvded by the domestc prce. The consumer prce ndex has been used as a measure of the domestc prce. The foregn prce s the prce of the country aganst whch the Afrcan country s currency s pegged. Usually ths foregn country s also the largest tradng partner. The real exchange rate (RE) serves a dual purpose. Partly, as Campos and Esfahan (1996, pp. 451-85) contend, exchange rate deteroraton and the concurrent hgh epsodes of nflaton result from neffcent poltcally-orented polces whch cause severe recessons and multple downturns n the economy. Ths justfes the ncluson of the economc growth varable ( GDP). Usng real per capta GDP wll help allevate the problem of heteroskedastcty (whch s common n cross-sectonal data) because t adjusts the dfferent countres ncome by populaton sze. To avod the possble smultanety (bdrectonal causalty) between ths economc growth varable (GDP) and the sze of the publc sector varable (SP), we subtracted total government spendng from GDP. That s, we used prvate spendng plus net exports as an nstrumental varable for GDP. Note that whle the Keynesan model assumes that government spendng s exogenous (a polcy varable) to GDP, Wagner s Law asserts that government spendng depends on (s nduced by) GDP (Brrd (1971, pp. 1-26), Afexentou et al. (1991, pp. 316-333)). Hence, allevatng the possble smultanety bas problem s mportant. Worsenng of macroeconomc condtons expedtes the process of publc enterprse reform, most of whch may have become a burden on the publc sector. In addton, the expanson of prvatsed schemes wdens the foregn exchange gap as the need for mo re foregn captal arses. Foregn exchange s an absolute constrant on the success of prvatsed projects n that even f domestc savng s suffcent, domestc nvestment may stll be low wthout the avalablty of foregn captal. Put dfferently, a salent explanaton of the ncluson of the real exchange rate as an explanatory varable s the openness of trade pror to prvatsaton. Another varable s the control of the bankng system to have ready access to selectve lendng (CB) or soft credt. To help ther supporters n purchasng and mprovng the prvatsed frms, governments control the bankng system and provde loans to supporters. Thus fraudulent schemes of credtworthness determnaton are used. Ths usually happens through selectve appontments to key postons n the bankng system, ownershp of certan banks, or natonalsaton. As a proxy for ths varable, we used the percentage change n total loans whch wll also act as an ndcator of nflatonary pressures. We could not fnd data on loans to prvatsed frms. Lastly, an mportant (dummy) varable s the degree of party (or ethnc group) afflaton (AP), based on the motvatons shown n Table 1. Governments whch undertake selectve (poltcal) prvatsaton usually belong to a certan party or ethnc group. Ther polces are, thus, more lkely to be partsan than natonalstc. Meanwhle, any expected effcency benefts may have been thwarted through harmful ncome dstrbuton effects. Based on the above, the sgns of the coeffcents whch dsplay the lkelhood that prvatsaton s poltcally motvated are all apror expected to be postve. It goes wthout sayng, that the set of all explanatory varables collectvely acts as proxy for the factors underlyng the proft-sharng relatonshp (represented emprcally by the decson to 42
SULIMAN AND GHEBREYSUS: DETERMINANTS OF PRIVATISATION prvatse) between the prncpal (government) and agent (prvate frm) or prvatsaton benefcares. That s, the underlyng latent proft-sharng (prvatsaton) decson between the government and ts alles (agent) s a lnear functon (wth parameter vector β) of the sx prescrbed explanatory varables. The dependent dummy varable Y = 1, f the estmated 2 prob ( Y = 1) s hgh (exceeds one-half) as can be approxmated by the pseudo R. Hence, we hypothesse that the probablty a regme wll pursue poltcal prvatsaton ncreases wth the proft-sharng motves emboded n a state of nature explaned by the set of the envsaged explanatory varables. The selecton of these varables has been based on prevous emprcal works by Dollar (1992, pp. 523-44), Campos and Esfahan (1996, pp 451-85), and Plane (1997, pp. 161-78), among others. All of these studes, however, focused manly on the effects of a specfc varable on prvatsaton. To our knowledge, none of them envsaged the possble set of macroeconomc determnants of prvatsaton. There s also an abundance of evdence from these studes showng that ths set of explanatory varables comprses a number of varables known to be addng to the underdevelopment problem of Sub-Saharan Afrcan countres. In ths respect, there mght be other varables of smaller effects that could have been ncluded n the set, such as some measure of development (see Plane (1997, pp. 161-78)). However, gven that these varables mght be correlated wth the specfed explanatory varables, we preclude such varables to mantan suffcent degrees of freedom for the probt model to be utlsed n the analyss. Ths s all the more vald consderng that hgher development may yeld ambguous results as they may encourage or dscourage poltcally motvated prvatsaton. The model used for testng can be explaned as: prob ( Y =1) = prob ( X β +ε > 0) = prob ( ε > Xβ) ; and prob ( Y = 0) f ( X β +ε 0). Therefore, Y = β + βsp + β SD + β RE + βcb + β AP + β GDP + ε. 0 1 2 3 4 5 6 ε = random varable wth zero mean and constant varance and covarance. Thus 2 ε ~ N (0, σ ). It thus becomes clear that Y s a measure of the decson (probablty) of whether to prvatse or not. Observe that the probablty of ( Y = 1) equals the probablty of ( X β +ε > 0) equals the probablty of ( ε > Xβ) or that the hgher the value of X, the more the macroeconomc varables and poltcal bas of country are resultng n a hgher lkelhood that country wll prvatse whenever t becomes poltcally feasble. The data sources are lsted n Table 3. Equatons 1 and 2 n Table 3 compare the dstrbuton of predcted and actual outcomes (perods when prvatsaton occurred) for the full perod 1970-94 and the sub-perod 1980-90 (Equatons 3 and 4). Because prvatsaton endeavors ntensfed n the 1980s and 1990s, we selected the subperod 1980-1994, together wth the entre sample perod of 1970-1994. The statstcal results essentally stayed the same ndcatng that the possblty of a structural shft after 1980 s not worrsome. As shown n Table 3, the model has a relable goodness of ft level as measured by the 2 pseudo R, rangng between.69 to.71 for the total sample perod, 1970-1994, whereas the range was somewhat hgher for the subperod 1980-1994, rangng from.74 to.82. The hgher range of probabltes for 1980-1994 s consstent wth the ntensfed prvatsaton efforts durng the 1980s and the 1990s. The model also makes correct predctons 89 percent 43
of the tme. That s, most of the hgh values of the explanatory varables correspond to a dependent dummy varable value of unty (mplyng the decson to prvatse), whereas most of the low values of the explanatory varables correspond to a dependent dummy varable of zero (mplyng the decson not to prvatse). In addton, the coeffcent estmates are consstent wth apror expectatons ndcatng the correct sgns and they are all statstcally sgnfcant. Further, there s no cause for undue concern about heteroscedastcty. The absence of heteroscedastcty and the log lkelhood transformaton test stated by Greene (1997), suggests that the estmators are consstent. Table 3 Probt Estmaton Results for the Perod 1970-1994 and 1980-1994 Equaton Constant * SP SD RE CB AP GDP Log. Pseudo Hetero lkelhood R 2 scedastcty 1970-1994 1 14.61 32.89 49.62 17.67 5.19 21.33 32.64 ( 3.22) (2.56) (2.17) (2.45) (1.01) (2.68) (3.80) 2 13.92 34.23 46.15 18.26 5.08 18.92 27.51 ( 3.01) (2.64) (2.43) (2.81) (1.53) (2.44) (4.01) 1980-1994 3 12.84 ( 3.60) 4 15.47 ( 2,26) 27.93 (2.52) 24.84 (2.34) 47.11 (3.22) 41.29 (2.98) 16.29 (2.52) 17.43 (2.54) 6.32 (1.09) 3.88 (1.13) 19.26 (2.93) 20.01 (2.77) 30.33 (3.90) 34.28 (2.89) Source: World Bank, World Development Report 1997, Oxford: Oxford Unversty Press, 1997. Notes: World Bank, World Tables, 1994. Internatonal Monetary Fund, Internatonal Fnancal Statstcs, varous ssues. Afrcan Development Bank, Afrcan Development Report 1996. * In the probablty model E Y ): β 0 + βx and 19.75.69 17.22 20.34.71 11.31 21.10.82 14.24 16.58.74 6.59 (, the normal equatons are = ( X X)( X X ) β = Y ( X X) βˆ = βˆ X (becausey = 0 ). Thus, the constant term s negatve. To force the mean of the error term 0 dstrbuton to be zero, any non-zero fxed porton of the dependent varable s added to the constant term. That s the constant term changes by the dfference between the sample mean of the error term and zero. 1. The dependent varable Y ) s a dummy varable. ( 2. Numbers n parentheses are the t-ratos. 3. The lkelhood rato (LR) = 2[log L restrcted log L unrestrcted] 4. Heteroscedastcty Test = L RH = 2[log L rv log r] where L RH = the log lkelhood functon; rv = error varance n the sub-sample; r = error varance n the entre sample. 5. Pseudo R 2 = {1 (1 log L restrcted/log L unrestrcted)} 6. The sample ncludes 21 countres lsted n Table 1. The only varable that seems to be nsgnfcant s the control of the bankng system (CB). Ths mght be a result of data contamnaton, snce we were not able to fnd data other than the change n total loans from the bankng system over the sample perod. Ideally, t would have been better to nclude only loans to prvatsed frms. 44
SULIMAN AND GHEBREYSUS: DETERMINANTS OF PRIVATISATION V. Concludng Remarks Ths paper examned prvatsaton epsodes n a sample of Sub-Saharan Afrcan countres that were under authortaran regmes as of 1970. After surveyng the multdmensonal motves of prvatsaton n Sub-Saharan Afrca, the paper focused ts attenton on the macroeconomc nstablty and poltcal motvatons as essental determnants of prvatsaton n those countres. A bref theoretcal dscusson employed a prncpal-agent model to nvestgate the mutually benefcal proft-sharng collaboraton between the prncpal (government) and the agent (prvatsed frm). That s, the theoretcal dscusson s ntended to show effcency gans (or statng of proft maxmsaton condtons) expected from prvatsaton. But the emprcal queston s whether prvatsaton s poltcally nduced, besdes beng a result of macroeconomc problems. The emprcal model dentfed a set of varables desgned to act as a proxy that creates the rght crcumstances for an expedent agent to collaborate wth the government (prncpal) based on relgous or ethnc bases and control the country economcally and poltcally. A rudmentary probt model has been used to do the emprcal work. The results n general are supportve of the basc hypothess that prvatsaton n the sample countres s nduced by macro-nstablty and poltcally motvated. Thus, one of the reasons why prvatsaton s not very effectve n Sub-Saharan Afrca s that t needs to be lnked to competton and regulaton polcy. Because of the smplcty of the probt model developed n ths study, t s mportant to take the emprcal results wth cauton. Surely, the model can be refned by ncorporatng more explanatory varables, and more mportantly, by consderng dfferent subperods and better specfcatons of the dependent, dummy, varable. Notwthstandng ts lmtatons, however, the model explans that prvatsaton responds to macroeconomc pressure and that t can be poltcally drven. References Adam, C. (1994), Prvatzaton and Structural Adjustment n Afrca, n Negotatng Structural Adjustment n Afrca, ed. by W.V.D. Geest, New York: Unted Natons Development Programme. Adam, C., W. Cavendsh, and P. Msty (1992), Adjustng Prvatzaton: Case Studes from Developng Countres, London: James Curry Ltd. Afexentou, P., and A. Serlets (1991), A Tme Seres Analyss of the Relatonshp Between Government Expendture and GDP n Canada, Publc Fnance Quarterly, 19(2), 316-333. Afrcan Development Report (1997), Fosterng Prvate Sector Development n Afrca, Oxford: Oxford Unversty Press, 79-164. Ake, B., and S. Mohammed (1986), Controls, Corrupton, and Compettve Rent-Seekng n LDCs, Journal of Development Economcs, 21(1), 161-80. Arrow, K.J. (1970), Essays n the Theory of Rsk-Bearng, Amsterdam: North Holland Publshng Co. (1971), Insurance Rsk and Resource Allocaton, n Essays n the Theory of 45
Rsk-Bearng, Chcago: Markham. Bardhan, P. (1984), The Poltcal Economy of Development n Inda, Oxford, U.K.: Basl Blackwell. Bates, R. (1981), Markets and States n Tropcal Afrca, Los Angeles: Unversty of Calforna Press. Bates, R., and A. Krueger (1993), Poltcal and Economc Interacton n Economc Polcy Reform, Oxford, U.K.: Basl Blackwell. Becker, G. (1983), A Theory of Competton among Pressure Groups for Poltcal Influence, Quarterly Journal of Economcs, 98(3), 371-400. Bennell, P. (1997), Prvatzaton n Sub-Saharan Afrca: Progress and Prospects Durng the 1990s, World Development, 25(11), 1785-1803. Brrd, R. (1971), Wagner s Law of Expandng State Actvty, Publc Fnance, 26, 1-26. Blackng, S. (1999), Structural Adjustment: Why It Wasn t Necessary and Why It Dd Not Work, Revew of Afrcan Poltcal Economy, 26(80), 207-227. Borch, K. (1962), Equlbrum n Rensurance Market, Econometrca, 30(2), 424-44. Bös, D. (1991), Prvatzaton: A Theoretcal Treatment, Oxford: Oxford Unversty Press. Bradburd, R. (1993), Why Thngs Have to Get a Lot Worse before They Get Better: Insttutons, Investments, Downturn, and Reform, World Bank, Polcy Research Department, Washngton, D.C. Brock, W., and S. Magee (1978), The Economcs of Specal Interest Poltcs: The Case of the Tarff, Amercan Economc Revew, 68(2), 246-50. Bruton, H. (1992), The Poltcal Economy of Poverty, Equty, and Growth: Sr Lanka and Malaysa, New York: Oxford Unversty Press. Callaghy, T. (1989), Toward State Capablty and Embedded Lberalsm n the Thrd World: Lessons for Adjustment, n Fragle Coaltons: The Poltcs of Economc Adjustment, ed. by J.M. Nelson, Washngton, D.C.: Overseas Development Councl, 115-138. (1990), Lost Between State and Market: The Poltcs of Economc Adjustment n Ghana, Zamba, and Ngera, n Economc Crss and Polcy Choce: The Poltcs of Adjustment n the Thrd World, ed. by J.M. Nelson, Prnceton, N.J.: Prnceton Unversty Press, 257-320. (1993) (ed.), Hemmed In: Responses to Afrca s Economc Declne, New York: Columba Unversty Press, 463-519. Campos, J., and H. Esfahan (1996), Why and When Do Governments Intate Publc Enterprse Reform? The World Bank Economc Revew, 10(3), 451-85. Cassen, R. (1994), Structural Adjustment n Sub-Saharan Afrca, n Negotatng Structural Adjustment n Afrca, ed. by W.V.D. Geest, New York: Unted Natons Development Programme. Clague, C. (1991), Relatve Effcency, Self-Contanment, and Comparatve Costs of Developed Countres, Economc Development and Cultural Change, 39(3), 507-30. Clark, R. (1998), Afrca s Emergng Securtes Markets, Westport, CT: Quorum Books. Dollar, D. (1992), Outward-Orented Developng Economes Really Do Grow More Rapdly: Evdence from 95 LDCs, 1976-1985, Economc Development and Cultural Change, 40, (3, Aprl), 523-44. 46
SULIMAN AND GHEBREYSUS: DETERMINANTS OF PRIVATISATION Domberger, S., and J. Pggot (1986), Prvatzaton Polces and Publc Enterprses: A Survey, Economc Record, 62(177), 145-62. Drazen, A., and V. Grll (1993), The Beneft of Crses for Economc Reform, Amercan Economc Revew, 83(3), 598-607. Due, J. (1993), Prvatzaton n Tanzana and Zamba, World Development, 21(12), 1981-1988. Edwa rds, S. (1993), The Poltcal Economy of Inflaton and Stablzaton n Developng Countres, NBER Workng Paper 4319, Natonal Bureau of Economc Research, Cambrdge, Massachusetts. Fernandez, R., and D. Rodrck (1991), Resstance to Reform: Status Quo Bas n the Presence of Indvdual-Specfc Uncertanty, Amercan Economc Revew, 81(5), 1146-55. Greene, W. (1997), Econometrc Analyss, 3 rd ed. New York: Macmllan. Haskel, J., and A. Sanchs (1995), Prvatzaton and X-Ineffcency: A Barganng Approach, Journal of Industral Economcs, XLIII (3), 301-320. Johnston, M. (1998), What Can Be Done about Entrenched Corrupton? n Annual World Bank Conference on Development Economcs 1997, eds. by B. Pleskovac and J.E. Stgltz, Washngton, D.C.: The World Bank, 69-90. Kahler, M. (1990), Orthodoxy and Its Alternatves: Explanng Approaches to Stablzaton and Adjustment, n Economc Crss and Polcy Choce: The Poltcs of Adjustment n the Thrd World, ed. by J.M. Nelson, Prnceton, N.J.: Prnceton Unversty Press, 33-62. Kltgaard, R. (2000), Subvertng Corrupton, Fnance and Development, June, 2-5. Laffont, J., and J. Trole (1993), A Theory of Incentves n Procurement and Regulaton, Cambrdge, M.A.: MIT Press. Lpumba, N. (1994), Afrca Beyond Adjustment, Washngton, D.C.: Overseas Development Councl. Plane, P. (1997), Prvatzaton and Economc Growth: An Emprcal Investgaton from a Sample of Developng Market Economes, Appled Economcs, 29, 161-178. Pratt, J., and R. Zechhauser (1985), Prncpals and Agents: The Structure of Busness, Boston, Mass.: Harvard Busness School Press. Rddell, R. (1999), The End of Foregn Ad to Afrca? Concerns about Donor Polces, Afrcan Affars, 98(392), 309-335. Rodrk, D. (1994), Poltcal Economc Markets: PERTS and PESTS n Food and Agrculture, Amercan Journal of Agrcultural Economcs, 64(5), 821-33. Rose-Ackerman S. (1998), Corrupton and Development, n Annual World Bank Conference on Development Economcs 1997, eds. by B. Pleskovac and J.E. Stgltz, Washngton, D.C.: The World Bank, 35-57. Toye, J. (1991), Ghana, n Ad n Power: The World Bank and Polcy-Based Lendng, Vol. 2, Case Studes, ed. by P. Mosley et al., London: Routledge, 151-200. Ul Haque, N., and A. Mrakhor (1987), Optmal Proft-Sharng Contracts and Investment n an Interest-Free Economy, n Theoretcal Studes n Islamc Bankng and Fnance, eds. by M.S. Khan and A. Mrakhor, Houston, TX: Insttute for Research and Islamc Studes. 47
Vckers, J., and G. Yarrow (1991), Economc Perspectves on Prvatzaton, Journal of Economc Perspectves, 5(2), 111-32. World Bank (1994), Adjustment n Afrca: A World Bank Polcy Research Report, New York: Oxford Unversty Press. (1995), Bureaucrats n Busness, Washngton, D.C.. (2001), Afrcan Development Indcators 2001, Washngton, D.C.. 48