Privatization in Emerging Markets

Transcription

1 Journal of Economic Integration 15(1), March 2000; Privatization in Emerging Markets Joshua Aizenman Dartmouth College an the NBER Abstract This paper shows two examples where privatization may lea to large effi - ciency gains by changing the menu of taxes. First, social security privatization increases the equity position of the mile class, inucing the meian voter to internalize a higher fraction of the costs of high taxes on capital, thereby reuc - ing the capital tax rate. Secon, reucing the public sector involvement in import competing activities is shown to lower the public sector s benefits from protection, reucing thereby the equilibrium tariff rate. These inirect effects of privatization escribe in the paper are external to the privatize activity. (JEL Classifications: F13, H21) <Key Wo rs : I m p e rfect capital mobility, Social security privatization, Import competition, Public enterprises> I. Introuction One of the remarkable evelopments of recent years has been the privatization eff o rt in countries characterize by the eep involvement of the State. 1 Privatization has been frequently part of a broaer attempt to liberalize economies characterize by stagnation. 2 The purpose of this paper is to * Corresponence Aress: 6106 Rockefeller Hall, Economics Department, Dartmouth College, Hanover, NH 03755, U.S.A. (Tel) , ( ) J.Aizenman@Dartmouth. e u Center for International Economics, Sejong Institution. All rights reserve. 1. See Worl Bank [1995] for an overview of privatization.

2 Joshua Aizenman 146 investigate the implications of privatization on the esign of taxes. As is well recognize by the literature, privatization leas to irect economic eff e c t s ue to the ownership change of prouctive assets. What is less appreciate, however, is that privatization may lea to other subtle effects by changing the menu of policies supporte by the meian voter. For example, in various countries the State was (an in some still is) irectly involve in the prouction of import competing consumer goos an services, as well as in the p rouction of key inputs. The alliance between the State an the import competing sector tilts the balance in favor of protective policies, as the branch of the state running the inustry frequently enjoys irect access to the political process. Privatization of such inustries woul reuce their bargaining power in the political process, ultimately reucing protection, an enhancing efficiency. Ignoring these inirect gains may lea one to unerestimate the importance of privatization. 3 A less obvious example along this line is social security privatization, where the state has been heavily involve in esigning an running the pay-as-you-go system. The purpose of this paper is to show that in an economy where policies are etermine by the meian voter, privatization woul lea to efficiency gains by changing the menu of taxes, in aition to all the other effects recognize by earlier literature. To illustrate this point we iscuss in section II the political economy implications of social security privatization in an emerging market economy. Section III stuies privatization of import competing activities, an section IV conclues. 2. For example, the Economist commente on December 6, 1997 The thir big change in Latin America s business environment, along with the taming of inflation an the liberalization of trae, has been the ismantling of the proprietorial state. Latin America has privatize with remarkable spee an on a vast scale. By one count, over the past ten years 279 companies have been sol for a total of $90 billion in the seven largest Latin American countries. Between 1990 an 1995, Latin America accounte for more than half (by value) of all privatizations in the eveloping worl, incluing Central an Eastern Europe. 3. For example, Lance Taylor recently wrote Privatization brings no obvious prouctivity gains, an if one in a slapash fashion it can aversely perturb savings, investment, an financial flows. See Blecker, e. [1996].

3 147 Privatization in Emerging Markets II. Social Security Privatization in Emerging Markets - - -Labor an Capital Taxes The aging of the baby-boomers an the backlash against the welfare state has focuse attention on the future of social security. One of the innovative schemes ealing with it is the privatization of social security funs. This reform has been frequently avocate by economists, an a version of it was aopte in 1981 by Chile, followe recently by other emerging markets (incluing Mexico an Bolivia in 1997). 4 Most of the analysis assessing these evelopments focuse on the impact of the re f o rm on saving. This paper argues that in an economy where the menu of taxes is the outcome of a political economy perspective, social security privatization woul lea to efficiency gains by changing the menu of taxes. To illustrate this issue, consier an economy where efficiency consierations call for a relatively low tax rate on capital income. 5 If the share of capital owne by the mile class is small, the meian voter woul impose a tax on capital income that excees the efficient tax by a large margin, reflecting the beggar my (capitalist) neighbor attitue. Granting a greater equity position to the mile class woul inuce it to internalize a higher fraction of the costs of high taxes on capital, reucing thereby the capital tax rate supporte by the meian voter. The purpose of this paper is to illustrate rigorously this argument for an emerging market, where agents can engage in capital flight an black market investment to shelter their investment income. We consier an economy compose of three classes, where the meian voter belongs to the mile class. The existing system finances a given transfer scheme with a mixture of capital an income tax. Agents have access to a gray capital market, sheltering their investment in tax free opportunities. The re t u rn to gray capital iminishes with the volume investe, as is the case when evaing 4. For an overview of social security reforms see Diamon [1996], an the papers in Felstein [1996]. For a review of the Chilean reform, see Diamon [1993] an Ewars [1996]. 5. This woul be the case if the supply of labor were relatively inelastic, whereas the supply of taxable capital were relatively elastic. The concern regaring saving s ouble taxation may provie another rationale for a low tax on capital income.

4 Joshua Aizenman taxes is harer for larger transactions. The gray capital market has several i n t e r p retations, like capital flight to overseas markets in eveloping countries, as well as omestic black market activities. Social security privatization is viewe as a policy that transfers equity from the high class to the other classes. i.e., prior to privatization, some of the equity of the high class p rovie the tax income reistribute v i a social security to other classes. After social security privatization, some of this equity is effectively transferre from the high class to the other classes. This amounts to a rop both in equity position an in the tax collecte from the high class. A. The Moel We consier a simple ex-ante/ex-post moel. The omestic output is prouce with labor an capital. The economy is compose of L agents, each supplying inelastically 1 unit of labor. Agents iffer only in their enowment of capital. There are three classes, each compose of a thir of the population. The ownership of the aggregate stock of capital, K, is istribute among the low, mile an high classes (enote by l, m, h) with shares s l, s m, s h, respectively. We assume that 0 s l s m <<s h. Capital investe at home is subject to a capital tax of. Capital investe abroa via capital flight evaes taxes, but is subject to raising transaction costs of tax evasion. We capture this possibility by assuming that the yiel on capital investe overseas epens negatively on the magnitue of capital flight. Let enote the share of capital investe omestically by the re p resentative agent [hence (1 ) is investe overseas], an let r * be the net yiel on capital investe overseas. The gray market investment opportunities are summarize by a re u c e form equation, r * = r * ( ); r* > 0. (1) Greater effective integration of the omestic capital market with the global market implies lower r *.

5 1 4 9 Privatization in Emerging Markets Domestic output, y, is y = [ L ] [ K ] 1 (2) where K is the aggregate capital owne by omestic agents. The pre-tax real wages an rental rates on capital, (w, r k ), are, re s p e c t i v e l y w = MP L = K L The government buget constraint is 1 ;r k = MP K = (1 ) L K (3) G 0 = wl + r k K (4) where is the labor tax rate. The initial equilibrium is characterize by an exogenously given fiscal revenue target, G 0. This revenue is re i s t r i b u t e among the three classes. Taking the value of the fiscal target an the patt e rns of reistribution given, the mile class etermines the tax rate on capital ( ) an on labor ( ), at rates that maximize its utility. A re p resentative capitalist is iniff e rent to investing at home or overseas (or in the omestic gray market activities) if the net re t u rns are equalize-- r * = (1 )(1 ) from which we infer that L K (5) = ( ); where r*, = = (1 )[ r*, + ] < 0 log r* log (5') B. The Political Equilibrium We turn now to escribe the set of taxes supporte by the mile class. The net income of class i (i = l, m, h) is

6 Joshua Aizenman R i = (1 ) wl 3 + (1 ) s i Kr k + (1 ) s i Kr * + q i G 0 (6) where q i is the transfer share of class i, an is assume to be exogenously given. Taxes are set by the meian voter, accoring to MAX[R m ], (7) subject to the government buget constraint, (4), an the competitive equilibrium conitions. Applying (3) - (4) to (6) we substitute for the wage tax, obtaining the mile class income as a function of the capital tax R m ( )= + (1 )[(1 )s 3 m + ] [L] [ K ]1 3 + (1 )s m K r * + (q m 1 3 )G 0 (8) From which we infer that the first orer conition efining the optimal tax on capital is R m = 1 + (1 ) y[1+ ] s m = 0 3 r * where = (1 )y [(1 )K + (1 ) (1 )2 y Kr * ]. (9) Figure 1 summarizes a simulation of the association of the mile class capital ownership share (s m ) with the tax rate on capital. The ownwar sloping curve traces the political equilibrium values of (s m ; ). In our framework, labor supply is inelastic, whereas the supply of capital for use in the omestic economy is elastic. This in turn implies that the efficient tax on capital is zero. The contours in Figure 1 report the welfare cost of the tax on capital, relative to the efficient benchmark case where the tax on capital is z e ro. Note that in our example, if the mile class equity share is 10%, it supports a relatively high tax rate on capital (46%), espite its relatively high welfare cost (8%). Increasing the ownership share to 0.2 woul reuce the capital tax to about 13%, an reuce the welfare cost to about 1.2%. We may unerstan better the above results by noting that the first orer

7 151 Privatization in Emerging Markets conition etermining the capital tax, (9), can be rewritten as a conition equating the marginal benefit of raising the capital tax to the marginal cost (both from the perspective of the mile class). The marginal benefit is efine as the income gain for the mile class were its equity share zero. The marginal cost is efine by the rop the mile class equity income cause by a higher capital tax rate. F i g u re 1 The Mile Class Capital Ownership Share an Tax Rate on Capital S m Notes: 1. The ownwar sloping curve traces the political equilibrium values of (s m, ). 2. The contours report the welfare cost of capital tax 3. The simulation assumes K = 1;L = 1;r * = 2 ;G 0 = 0.1; = 2/ MB = MC where where MB = 1 3 y[1+ + (1 ) ];MC = s m (9') Applying (5') to (9'), collecting terms, it follows that

8 Joshua Aizenman MB = 1 3(1 )( + r*, ) y [ r *, (1+ r*, )]; MC = s m (1 )y(1 )(1+ r*, ) + K r * 1 [ {(1 ) r*, }] (1 )( + r*, ) (10) The marginal benefit curve (MB) in Figure 2, panel I, measures the marginal benefits to the mile class resulting from shifting the buren of taxes from labor to capital. The curve correspons to the example use in Figure 1, where = 2. r Starting with zero tax on capital, a higher capital tax rate *, increases revenue from capitalists, allowing a rop in the wage tax, benefiting the mile class. This effect iminishes as we move upwars on the capital tax Laffer curve - - both the tax revenue an the wage gains attribute to the higher tax revenue collecte from capital rop ue to more intense capital flight. The marginal benefit from the tax will isappear at point L w h e re r MB = 0, or = *,. Point L also etermines the tax on capital if the 1+ r *, mile class oes not own any equity. The marginal cost stems from the fact that a higher capital tax inuces a negative income effect, pro p o rtional to the equity position. The income attribute to capital ownership rops both ue to the tax on capital, as well as to the inuce rop in the gray market return. The sum of both effects efines the marginal cost of the capital tax, an is plotte by the ashe MC curve for s m =0.066(see Figure 2, panel I). The otte curve correspons to the marginal cost if the equity position triples. Note that a higher equity position increases the marginal cost of capital tax for the mile class, shifting MC upwars, inucing a rop in the tax rate. A greater integration with the global capital market implies that a given tax rate inuces greater capital flight, exacerbating the welfare cost of the capital tax, an leaing to a lower capital tax rate. This effect is capture by comparing the two panels of Figure 2. Greater integration implies a smaller r *,. Inee, reucing it from 2 to 1 implies that the capital tax rate supporte by the mile class rops from 2/3 to 1/2 if the mile class equity share is zero [an from 10% to 0% if the mile class equity share is 20%].

9 153 Privatization in Emerging Markets F i g u re 2 The Marginal Cost an Marginal Benefit of Capital Taxes [the Mile Class Perspectives] MC[s=0.2] 0.05 MC[s=0.066] 0 L 0.05 MB I - rawn for r *, = 2 MC[s=0.2] 0.05 MC[s=0.066] 0 L' 0.05 MB I - rawn for r *, = 1

10 Joshua Aizenman 154 C. Social Security Privatization We turn now to evaluate the impact of social security privatization. We assume that the privatization scheme is revenue neutral at the initial tax rates. A transfer of equity ownership from the high to the mile class leas to a rise in the net income of the mile class. It is assume that there is a matching rop in the irect transfer from the government to the mile class, so that the net income effect is zero. This compensation is summarize by (11) - - q m = R / s m m = < 0 s m R m / q m (1 )(1 ) +(1 )Kr * G 0 (11) A similar ajustment (but in the opposite irection) takes place for the high class. The net effect of the equity infusion to the mile class can be trace with the help of our analysis from the previous section, summarize by (10). Note that the results in the previous section were inepenent from the transfer share of the mile class, q m. Hence (inepenently of the income effects), social security privatization tens to reuce the capital tax supporte by the mile class. D. Discussion The moel escribe in Section II is a static version, esigne to eliver the argument in the simplest manner. The logic of our iscussion can be applie to a ynamic, overlapping generation moel. Privatization is not a panacea. Partial privatization, with excessive regulatory intervention by the government, may inuce significant welfare costs. Examples -- Attempts to force funs free by social security privatization to be investe mostly in omestic government bons [as has been the case in Mexico an Bolivia] amounts to an implicit tax on the new pool of privatize savings. It is equivalent to force saving, use to fun public ebt at below the market interest rate. Our analysis can be extene to aress the above issues in a moel that allows for

11 155 Privatization in Emerging Markets restrictions on capital mobility, along the line of Aizenman an Guiotti [1994]-- consiering an economy where there are no lump sum taxes, an all tax collection is costly. In these circumstances the t re a s u ry woul rely on restrictions on capital mobility as a secon best policy aiming at increasing the omestic tax base. Attempts to restrict the privatize social security funs to be investe mostly in omestic assets woul limit the private sector s gains from global iversification. These gains may be especially large in emerging markets characterize by limite GDP iversity, an re l a t i v e l y high reliance on commoity trae. Non competitive management of the privatize social security funs may encourage waste an rent seeking, as apparently has been the case in Chile (see Diamon [1993]). With uncertainty, social security privatization has other effects that are ignore by the present paper (e.g., it increases the exposure of the mile class to equity risk). III. Import Competing Goos, Protection, an Privatization Various eveloping countries are characterize by inflate public sector that operates uner the shiel of a high level of protection. In this section we illustrate that the logic of our iscussion is applicable to these circumstances. We consier the simplest example -- a two goos, small economy, w h e re the government owns a fraction s g of the import competing sector. E x p o r tables, X, an importables, Y, are prouce using the services of labor (L), an a sector specific immobile capital. The international price of both goos is normalize to 1. Labor is fully mobile between sectors, an all markets are competitive. Government utility, U g, is a weighte average of consumer s welfare (enote by W, evaluate in terms of goo X) an government s income (enote by R) 6 -- U g = W + (1 )R (12) 6. For a erivation of such a reuce form in a political economy equilibrium see Grossman an Helpman (1996).

12 Joshua Aizenman 156 Government s revenue is the sum of its revenue from a tariff (the rate of which is ), plus any surplus stemming from the ownership of pro u c t i v e assets ( ) -- R = IM y + = IM y + s g [(1+ )Y s L y ] (13) where IM y enotes the imports of goo y, Y s is the omestic prouction of importables, the wage, an L y the labor employe in sector y. A. Optimal Tariff an Public Sector s Ownership The government etermines the tariff at a rate that maximizes (12), leaing to a first orer conition given by W R 0 = + (1 ) (14) Alternatively, as shown in the Appenix 0 = IM y s g (Y s L y ) + (1 )[IM (1 y IM, ) + s g (Y s L y )], (15) w h e re IM, = log IM y is the import s elasticity with respect to the tariff log rate. The first term of the RHS of (15) weighs the impact of a marginal tariff on consumer s welfare. Welfare rops by the averse income effect of the higher cost of imports [proportional to the level of imports] an by a rop in the private s sector income [which in the Appenix is shown to equal the increase in public enterprise surplus]. The secon terms weigh the impact of the tariff on government income. It is the sum of the net change in tariff revenue plus the resultant increase in government surplus. Alternatively, (1 2 )[IM y + s g (Y s L y )] = (1 ) IM y IM, The term 1 2 (15') is the government s utility gain from transferring a unit of p u rchasing power from public to government use [= + (1 )]. This is a m e a s u re of the egree to which the govern m e n t s agena iverges fro m that of the private sector s agena. It must be positive if the govern m e n t

13 157 Privatization in Emerging Markets prefers not to rebate the tariff income to the public, as we assume it to be the case. The LHS of (15') measures the marginal benefit of a higher tariff [from the government perspective], equal to the prouct of 1 2 times the extra revenue from tariff (at the given level of imports) plus the increase in rents associate with importables. The RHS is the marginal cost, inuce by a rop in tariff income ue to the rop in imports triggere by the tariff hike. Equation (15') reveals that a higher ownership share of the public sector s g increases the marginal benefit of the tariff, without impacting on the marginal cost. Hence, it will lea to a higher optimal tariff. Formally, by applying the implicit function theorem to (15) we infer that increasing the governm e n t s ownership share of importables woul increase the optimal tariff [see the Appenix for erivation]. s g = (1 2 )(Y s L y ) 2 U g > 0 2 (16) In signing the RHS of (16) we note that the secon orer conition for maximizing government utility implies g 2 U. Note also that Y s L 2 > 0 y m e a s u res the impact of a marginal tariff change on rents in sector Y. In the appenix we show that Y s L y > 0. Hence, a higher share of import ables prouce by the government woul increase the optimal tariff chosen by the government, as it increases the marginal benefit of a tariff hike. Alternat i v e l y, a lower public ownership of the impor t competing sector woul reuce the benefits from the tariff [from the govern m e n t s perspective], resulting in lower protection. This suggests that successful privatization woul lea overtime to a rop in protection. IV. Discussion UOur analysis of protection an public sector ownership consiere the case where the public enterprises are generating rents. Frequently, howev-

14 Joshua Aizenman 158 er, public enterprises in eveloping countries are not profitable. This is consistent with the logic of our paper once we recognize that the rent of public enterprises may be use [an abuse] towars other political goals [e.g., a politically motivate expansion of employment, financing political activities, etc.]. Uner these circumstances the rent relevant for our iscussion is the gross rent, before subtracting all the costs of politically motivate activities. 7 ULobbying for protection [as in Grossman an Helpman [1994]] may mitigate the impact of privatization, as the new owners will use part of their rent to lobby. Yet, lobbying woul not reverse the above results -- in the initial equilibrium the government got irectly all the surplus in the activity uner its ownership, whereas in the lobbying game it will get in most cases only a fraction of that surplus. Note also that with uncertainty, privatization may lea to other large inirect ef fects, like reucing gover n m e n t s unfoune contingent liabilities. UOur analysis points out that privatization in eveloping countries may entail positive externalities by changing the menu of taxes. The two examples escribe above illustrate that in economies where taxes are etermine by the meian voter, privatization may lea to efficiency gains by reucing the tax on capital [with social security privatization] an tariff s [with public enterprise liberalization]. Earlier literature ealing with privatization of public enterprise pointe out that, unless privatization is accompanie by liberalization measures, it is unlikely to result in significant gains in economic efficiency [see van e Walle [1989]]. Hence, a public enterprise privatization effort is more likely to succee if it is part of a comprehensive rationalization an trae liberalization efforts. Similar conclusions apply for social security privatization - - its gains are greater an more transparent if it is accomplishe with liberalization of financial markets, allowing gre a t e r international iversification an freer management of private funs. UOur analysis suggests that privatization an liberalization are complem e n t a ry steps, re i n f o rcing each other through the political process. This result shoul be qualifie as it ignores other obstacles to privatization. Fre- 7. Note that privatization in such an economy may curb the political rent seeking, increasing the resultant efficiency gains.

15 159 Privatization in Emerging Markets quently, the parties that woul lose from privatization an liberalization are easier to ientify, an are better organize than the parties that woul gain from it. This implies that the ajustment of taxes to the new economic environment woul be protracte an time consuming. In these circumstances a moifie version of our moel continues to hol, once proper iscounting is applie to the costs an the benefits of privatization. UThe inirect effects of privatization escribe in the paper are external to the privatize activity. Hence, these benefits are not accounte for in a conventional cost benefit assessment of the privatize projects. Our examples illustrate that ignoring these effects may lea one to unerestimate the potential gains of privatization. References Aizenman, J. an P. Guiotti [1994], Capital Controls, Collection Costs, an Domestic Public Debt, J o u rnal of International Money an Finance, February; pp Blecker, R. A., e. [1996], U.S. Trae Policy an Global Growth: New Direc - tions in In t e rnational Economy, Economic Policy Institute Series, Armonk, N.Y. an Lonon: Sharpe. Diamon, P. A. [1996], Proposals to Restructure Social Security, Journal of Economic Perspectives 10 (3); pp Diamon, P. A. [1993], Privatization of Social Security: Lessons from Chile, NBER Working Paper No Ewars, S. [1996], The Chilean Pension Reform: A Pioneering Program, NBER Working Paper No Felstein, M., e. [1996], Privatizing Social Security, Forthcoming, University of Chicago Press. Grossman, G. M. an E. Helpman [1996], Electoral Competition an Special Interest Politics, Review of Economic Stuies 63(2); pp G rossman, G. M. an E. Helpman [1994], Protection for Sale, A m e r i c a n Economic Review; 84(4); pp van e Walle, N. [1989], Privatization in Developing Countries: A Review of the Issues, Worl Development 17(5); pp Worl Bank [1995], Bureaucrats in Business: The Economics an Politics

16 Joshua Aizenman of Government Ownership Policy Research Report Series, Oxfor an New York: Oxfor University Press for the Worl Bank. Appenix This Appenix summarizes the steps leaing to (16). First, note that the optimal consumption of the private sector is obtaine by MAX[ u(x c,y c ) {X c + (1+ )Y c I} ] X c,y c (A1) w h e re u is a neoclassical utility function, X c, Y c a re the consumption of e x p o rtables an importables, re s p e c t i v e l y, an I is the private sector income -- I = X s + (1+ )Y s s g [(1+ )Y s L y ] From the first orer conitions corresponing to (A1) we infer that (A2) W = u u X = X c +(1+ )Y c. (A3) w h e r e u X is the marginal utility of X. From the buget constraint [I = X c + (1+ )Y c ] an (A2) it follows X c + (1+ )Y c = I Y c = { X s + (1+ )Y s } + Y s s g (1+ )Y s L y [ ] s Y s g[ L y ] Y c. (A4) Recall that labor is mobile between the two sectors, hence X s L x = ; (1+ ) Y s L y = ; an L x + L y = 0, (A5) from which we infer that X s +(1+ )Y s = [L x + L y ] = 0 (1+ )Y s L y = 0 Applying (A6) to (A3) an (A4) we infer that (A6) W = Y s s g [ Y s L y ] Y c = IM y s g [ Y s L y ]. (A7)

17 161 Privatization in Emerging Markets Equation (15) is obtaine by substituting (A7) for W in (14), after collecting the various terms. We conclue the Appenix by showing that Y s L y a conition > 0, that is applie in signing (16). Note that equilibrium in the labor market implies that where L x + L y = L L is the supply of labor. Thus, L x + ˆ L y L y = 0, L x + Ly =1, L x ˆ (A8) (A9) where L X ; L y correspon to the labor share in the relevant sectors. Applying (A5) we infer that ˆ L x X ', L x = ˆ ; ˆ L x Y ', L y + ˆ = ˆ (A10) where X ',L x = log2 X s log L x 2 ; Y ', L y = log2 Y s log L y 2. Applying (A9) to (A10) we infer that = 1 + where = L y / Y ',L y L y / Y ', L y + L x / Y ', L x < 1. (A11) Thus, Y s L y = Y s L y 1 + = 1 1+ [(1+ )Y s L y ] > [(1+ )Y s L y ], (A12) Note that (1+ )Y s L y a re the rents in sector Y, hence (1+ )Y s L y >0, an Y s L y > 0.