Does Competition for Capital Discipline Governments

0 downloads 0 Views 276KB Size Report
Does competition to attract mobile capital discipline governments, reducing ... Even if capital market integration does discipline governments and improve policy, ...
Does Competition for Capital Discipline Governments? Decentralization, Globalization and Corruption Hongbin Caia

Daniel Treismanb

Many political economists believe that competition among countries—or regions within them—to attract mobile capital should discipline their governments, rendering them less corrupt and more friendly toward business. This argument surfaces repeatedly in debates over both political decentralization and globalization. We argue that it is based on an assumption— countries or regions start out identical—that is quite unrealistic. We reexamine the standard model that predicts a disciplining effect of capital mobility, and show that if units are sufficiently heterogeneous exactly the opposite prediction often follows. If some units are exogenously much more attractive to investors than others (and competition for capital is intense), the only equilibrium under capital mobility will involve polarization. Initially disadvantaged units will actually be more corrupt, more starved of capital, and slower to grow if capital is mobile than if it is not. By contrast, exogenously attractive units will do more to woo investors, suck capital out of their lower productivity counterparts, and grow faster. We suggest this may help explain the disappointing results of liberalizing capital flows within the Russian federation and in sub-Saharan Africa.

July 2002 a

b

Department of Economics, University of California, Los Angeles Department of Political Science, University of California, Los Angeles, 4289 Bunche Hall, Los Angeles CA 90095-1472

1 Introduction Does competition to attract mobile capital discipline governments, reducing corruption and increasing growth-promoting public investments? The answer to this question has important implications for two key topics in political economy—the consequences of political decentralization and the desirability of liberalizing international capital flows.1 If competition over capital renders governments more honest and effective, decentralizing authority to regional governments within large states should improve governance.2 Freeing international capital flows should also create pressures toward less corrupt and more business-friendly policies in countries around the world.3 A consensus seems to be forming around the view that competition for capital does lead to better governance. Scholars of federalism have argued that interregional competition punishes wasteful or dishonest governments with capital flight (Qian and Roland 1998). Montinola, Qian, and Weingast (1995, p.58) contend that the contest among subnational units to attract capital and labor induces them “to provide a hospitable environment for factors, typically through the provision of local, public goods,” and to guarantee secure property rights and infrastructure. They argue that in China, competition among provinces, cities, and townships for foreign investment has caused many to adopt pro-business laws, regulations, and tax systems (Ibid, p.77). This view of capital mobility as a constraint on corrupt or foolish policies is also common in the globalization debate. In Obstfeld’s words, a “main potential positive role of international capital markets 1

In this paper, we focus on questions of capital mobility and do not consider the many important debates over the effects of increasing international trade and trade openness on corruption and growth. Even if capital market liberalization does not have unequivocally positive effects, trade liberalization might still be a good thing.

2 3

We assume that capital is generally more mobile between regions within a state than between states.

Even if this argument is right, there may be other reasons to be cautious about capital market integration. Competition over capital might at times be too intense, leading governments to invest too much in services to business and too little in those favored by labor—they might build too many “business centers and airports but not enough parks or libraries” in the phrase of Keen and Marchand (1997). Or competition to attract capital might cause governments to set taxes on capital too low and those on less mobile factors (land, labor) too high (Rodrik 1998). Even if capital market integration does discipline governments and improve policy, the greater volatility associated with openness may have damaging effects (see, e.g., Demirgüç-Kunt and Detragiache 1998). Nevertheless, the claim

is to discipline policymakers who might be tempted to exploit a captive domestic capital market. Unsound policies—for example, excessive government borrowing or inadequate bank regulation—would spark speculative capital outflows and higher domestic interest rates” (Obstfeld 1998, p.10). Even one wellknown skeptic on the benefits of globalization declares himself sympathetic to the argument that “opening the capital account imposes ‘discipline.’ Countries are ‘forced’ to have good economic policies, lest capital flow out of the unit” (Stiglitz 2000, p.1080). The Economist magazine goes still further, contending that: “Integration makes it harder to be a tyrant… Oppression is more difficult with open borders: people can leave and take their savings with them” (The Economist 2001).4 In this paper, we contend that—while such effects are certainly possible—they rely on strong assumptions that are unlikely to hold for many real world cases. Critically, the standard model motivating such predictions assumes that regions or countries (henceforth, “units”) are identical. Analysts then focus only on symmetric equilibria, in which by definition units converge on the same policies or tax rates. We show that given alternative, empirically plausible assumptions, exactly the opposite logic may apply. If some units start out with higher capital productivity than others, symmetric equilibria will not exist. Given substantial initial differences—and sufficiently intense competition for capital—the initially disadvantaged units will actually be more corrupt, more starved of capital, and slower to grow in equilibrium under capital mobility than if they had effective capital controls. By contrast, under capital mobility, exogenously attractive units will invest relatively more in business services, suck capital out of their lower productivity counterparts, and grow faster. If units have sufficiently heterogeneous endowments, capital mobility benefits the advantaged at the expense of the disadvantaged. This follows not from some new and far-fetched model, but from reexamination of the standard model from which the more optimistic predictions about capital mobility were derived. To put it concretely, even if Chad’s government were to invest massively in business that capital mobility enforces growth-promoting policies and honest governance is one of the most powerful in the globalist’s repertoire. 4

See also Wilson (1999, p.298).

2

infrastructure, it is unlikely it would be able to attract money out of the capital markets of New York or to compete in productivity with the industrial zones of East Asia. Even if Buryatia were to lay down highspeed fibre-optic cables, it would not divert much business investment from Moscow and St Petersburg. Under capital immobility, governments have some incentive to stimulate the investment of domestically generated savings—they will be able to tax the profits. Under capital mobility, domestic savings will flee the unit’s undeveloped infrastructure and political risk in search of more promising and secure returns. Knowing that they cannot compete, governments in severely underdeveloped units facing capital mobility will give up on pro-business policies and focus instead on predation—they will face less, not more, effective discipline. To demonstrate our point, we study the standard model generally used to argue that capital mobility disciplines governments. In this model, governments of two units allocate fiscal revenue between spending to improve the business environment (“infrastructure”) and self-consumption (“corruption”).5 Infrastructure increases capital productivity. Our main innovation is to allow (in the production function) for different initial conditions in the units. We first reproduce the standard result that when units start out identical, a symmetric equilibrium exists in which the governments invest more in infrastructure and less on corrupt consumption under capital mobility than under capital immobility. However, a symmetric equilibrium does not exist if the units do not start out identical. In this case, two types of equilibrium can exist—participation equilibria and polarization equilibria. In participation equilibria, both governments invest in infrastructure and both units get positive amounts of capital. We show that the greater is the initial asymmetry in capital productivity between units, the greater will be the disparity in the equilibrium capital allocation and infrastructure investment under capital mobility. If initial asymmetry is high enough, the government of the initially disadvantaged unit will invest less in infrastructure, steal more of the budget, and receive less capital under mobility than under immobility. In the extreme, there exists only a polarization equilibrium, in which the government of

5

It is straightforward to extend the model to include public good provision in government expenditure.

3

the disadvantaged unit makes no infrastructure investment and the unit gets no capital. In general, competition for capital exacerbates initial inequality. Not only the disadvantaged unit is worse off; under certain conditions, total social welfare in the unit (or the whole system)—measured by aggregate output— also falls as capital becomes mobile. Recognizing this helps explain some otherwise surprising empirical cases. Internal capital flows have been liberalized recently in both China and Russia. While there is some impressive evidence of competition among the more developed coastal provinces of China and among cities within them, there is little evidence of any salutary effect of competition on the inland provinces.6 In Russia, capital appears to have flowed out of capital-poor into a few capital- and infrastructure-rich regions, exacerbating interregional inequality. Regions that had low initial levels of infrastructure and human capital tended to spend a smaller share of their budget on market infrastructure, transport and communications; to rank lower in investment ratings of institutional quality; to attract smaller—or even negative—net investment inflows; and to experience slower growth of small businesses. Many developing countries liberalized their capital accounts in the 1980s and 1990s. Some— usually the upper middle-income ones—experienced large inflows of capital, which helped stimulate growth. However, others—in particular, some Sub-Saharan African countries—suffered net capital outflows. During these decades, there was no noticeable, general improvement in the quality of African governance, and the continent continued to fall further behind the rest of the world in output.7 We present evidence that is consistent with our model but not with the more optimistic arguments about capital mobility in a later section. Our argument is related to several others. Students of economic growth noticed some time ago

6

For instance, Jian, Sachs and Warner (1996) found that the more developed coastal provinces began diverging in output from the less developed inland provinces in the 1990s after international trade and investment flows were liberalized.

7

GDP growth in Sub-Saharan Africa averaged 1.7 percent a year in 1980-90, compared to 3.1 percent in the world as a whole; in 1990-97, the region’s growth averaged 2.1 percent a year, compared to 2.3 percent worldwide (World Bank 1999, p.211).

4

that countries’ incomes were not converging in the way that simple neoclassical models predicted (see, for instance, Romer 1994, Barro and Sala-i-Martin 1995). Common explanations posit that capital is more productive when combined with high levels of human capital, infrastructure, or property rights protection (Lucas 1990, Mankiw, Romer and Weil 1992, Sachs, Borensztein, De Gregorio, and Lee 1998). Most previous treatments have not noted, however, that such initial productivity differences also undermine the argument that capital mobility disciplines governments. The novelty of our argument is to show that competition for capital will not necessarily cause governments to converge on capital-friendly policies. In fact, poor countries will often choose to invest less in growth-promoting policies, even knowing they will lose capital as a result. Rogowski (2001) makes an argument very similar to our own. He uses a spatial model of policy preferences to explore the extent to which the median voter (worker) will vote for environmental or labor policies that accommodate—and thus attract—mobile capital. He finds that under capital mobility the median voter will accommodate capital more in countries that have a higher capital/labor ratio and greater “natural” advantages as a production site. This should lead to greater divergence among countries’ policies under capital mobility than in a world of strict capital controls. There are two main differences with our approach. First, Rogowski’s model does not include taxation or public services that are costly to provide. The cost to voters of policies that alienate capital shows up in lower labor productivity and wages as capital flows out. But another possible cost of capital outflows are that they shrink the tax base, reducing government revenues that could be spent on public goods. We are able to show that even when the cost of a shrinking tax base is modeled, fear of capital flight will not cause very capital-poor countries or regions to compete by implementing more capital-friendly policies. Second, Rogowski assumes that governments are constrained to implement the policies favored by the median voter. We allow for a range of government objectives—from pure predation to pure benevolence. This renders the model readily applicable to non-democracies. It also permits us to investigate the level of corruption and waste—the central concerns of this paper—and to show that in very capital-poor countries, corruption will actually increase under capital mobility.

5

Several previous papers analyzed asymmetric tax competition. Bucovetsky (1991) presented a model in which smaller countries have lower tax rates at equilibrium because the benefit from capital has a larger per capita impact than in larger countries. Kanbur and Keen (1993), in a model with commodity taxes and transportation costs, found that governments of geographically small countries should set the tax rate lower, because the shorter distance for arbitrageurs to travel reduces the rents the government can extract. We do not examine the effects of country size.

2 The model We study a standard model often used in the existing literature to argue that capital mobility disciplines governments.We keep the model as simple as possible to make our points clear. Two regions or countries (“units”), indexed i = 1, 2 , are governed by two governments, G1 and G2. Each government can invest in infrastructure to improve the business environment in its territory; let I i be Gi's infrastructure investment. There is a fixed amount of capital, K; let ki be the amount of capital invested in unit i . Given I i and ki , we assume that the aggregate production function of unit i, Fi , takes the following form:

F1 = (αI1 + βI2 + a1 )k1 − 0.5γk12 − 0.5δI12 F2 = (αI2 + βI1 + a2 )k2 − 0.5γk22 − 0.5δI22

(1)

Without infrastructure investments, the production functions are standard quadratic functions. Infrastructure investments increase the returns to capital. The parameter α > 0 measures the increase in the return to capital in a country or region caused by one unit of infrastructure investment there. “Infrastructure” should be interpreted broadly here: it represents anything that increases the productivity of capital in the unit and that is costly for the government to provide. Thus, it includes physical infrastructure (transportation, telecommunications, etc.), education, public health, and a system of wellenforced property rights and legal protections. Infrastructure investment in one unit may also create externalities for the other unit. For example, better infrastructure in unit 2 reduces local transportation costs, benefiting firms in unit 1 that use inputs from or sell products to unit 2. To allow for investment

6

externalities, β ≥ 0 measures the increase in the return to capital in one country or region caused by one unit of infrastructure investment in the other. In general, infrastructure investment in one unit affects it more than its neighbor, so we assume that α > β . Finally, γ ≥ 0 measures how fast the marginal return to capital diminishes as capital increases, and δ > 0 measures how fast the marginal return to infrastructure investment diminishes as investment in it increases. Our innovation is to introduce asymmetry into the otherwise standard model. Due to differences in the two units’ initial conditions (e.g., natural endowments, human capital, initial stock of infrastructure), the exogenous attractiveness of business environments in the two units will differ. To capture this, we assume that a1 ≥ a2 > 0 . Note that in the production function Fi , ai > 0 represents the marginal return to capital in unit i when ki = Ii = I j = 0 . Hence ai really measures the quality of unit

i ’s initial business environment before the government decides how much it wants to spend improving it and entrepreneurs decide where to invest. Countries—and regions within large countries—often differ greatly in the quality of their business environments at the moment when they contemplate opening their capital markets. To keep things simple, except for this difference in initial business environments, we assume the two units are identical. Of course, in the special case when a1 = a2 , the two countries are completely identical. From the production functions in equation (1), the marginal returns to capital and to infrastructure are, for i = 1, 2, j = 3 − i ,

∂Fi = αI i + βI j + ai − γki ∂ki ∂Fi = αki − δI i ∂I i

(2)

We assume that a1 ≥ a2 ≥ γK so that returns to capital are always non-negative. In other words, capital is

7

scarce and additional investments never reduce total output.8 An important modeling issue concerns the objective functions of the governments. In some existing works, governments are assumed to be benevolent and to maximize social welfare under their jurisdictions; in others, they are assumed to be Leviathans that maximize their own consumption or tax revenues. In this paper, we consider a general situation in which governments (G1 and G2) are partially self-interested, in the sense that they care about both social welfare and their own consumption. Specifically, a government’s objective function includes: (1) total output within the unit net of taxes, (2) public goods provided for residents, and (3) the government’s own consumption from money corruptly stolen from fiscal revenues. Specifically, the payoff functions of G1 and G2 are assumed to be

Ui = (1 − ti ) Fi + mi + λci

(3)

where ti is the (sales) tax rate, mi is the amount of public good provision, ci is the government’s own consumption from money corruptly stolen and λ is the corruption propensity of governments (assumed to be the same in both units). For simplicity, we assume that utility from public good provision is linear. (The analysis of this paper can be straightforwardly extended to situations in which the utility from public good provision is strictly concave.) Under this assumption, when λ < 1 , the governments are completely honest and maximize social welfare because they will never choose positive corruption. To focus on interesting cases, we assume that λ ≥ 1 . This implies that mi = 0 , so mi will be ignored except in welfare analysis.9 If λ approaches infinity, then the government is totally corrupt and cares only about its own

8

With free disposal, capital must have non-negative returns. Our assumption that returns to capital are non-negative even without infrastructure investments is made for convenience. Without this assumption, we would need to impose technical conditions later to ensure non-negative returns to capital in various cases, but not much additional insight could be gained.

9

This does not mean that literally no public goods are provided. Certain vital public goods can be provided with earmarked funds, while the mi ’s represent discretionary government expenditure on public goods.

8

corrupt consumption. Each government is endowed with initial fiscal revenue S ≥ 0 . The budget constraint of government Gi is Ii + ci = S + ti Fi . We study the game in which G1 and G2 simultaneously choose how much infrastructure investment to make. Our focus will be on the effect of capital mobility on the governments’ investment incentives. In particular, we will compare two polar cases: (1) capital is completely immobile in the sense that capital allocation across the two units is fixed at some historically determined level; and (2) capital is perfectly mobile in the sense that there is no cost for capital to cross borders. Perfect mobility of capital implies that the after-tax marginal productivity of capital should be equalized across units. Of course, the real world lies somewhere in between these two polar cases. Nevertheless, by studying and comparing them, we hope to shed light on what happens as capital mobility increases. When capital is mobile, governments compete to attract capital to their territory by investing in infrastructure. The intensity of the competition depends on how fast the marginal return to capital diminishes as the stock of capital increases, which is captured by the parameter γ . If the marginal return to capital diminishes fast ( γ is large), then competition for capital is not intense, especially if the total amount of capital available is high so that one unit becomes satiated in capital. In thinking about countries or large economic units, it is likely that γ is relatively small—the marginal return to capital diminishes slowly.10 Even the richest countries—or regions—are certainly not satiated in capital. To present the main idea in the simplest way, we will first assume that the tax rates in the two units are exogenously fixed at the same level: t1 = t 2 = t ≥ 0 . Then, in Section 4, we will briefly discuss changes to the model caused by relaxation of this assumption.

10

In fact, in the development economics literature, the assumption of constant or even increasing returns is quite common and is a key element in the theory of a “development trap”. Similarly, in economic geography, the assumption of constant or increasing returns is often invoked to explain geographic concentration of economic activities (e.g., cities, software industry clustering in Silicon Valley).

9

2.1 Equilibrium under capital immobility Suppose unit i is endowed with capital of ki > 0 , where k1 + k 2 = K . Government Gi chooses (ci , Ii ) to maximize Ui = (1 − t ) Fi + λci subject to its budget constraint ci = S + tFi − Ii . Substituting the budget constraint into the objective function, we get the following first order condition

(1 − t ) ∂Fi ∂I i + λ (t ∂Fi ∂I i − 1) = [1 + (λ − 1)t ] ∂Fi ∂I i − λ = 0

(4)

From Equation (2), we have

Ii = [αki −

λ ]/δ 1 + (λ − 1)t

(4′)

The second order condition is clearly satisfied, so Equation (4′) gives Gi ’s optimal infrastructure investment as long as αki ≥ λ [1 + ( λ − 1)t ]. Let us denote the solution I1 ( k1 ) . The corrupt consumption of Gi is then ci = S + tFi − Ii . We assume that S is sufficiently large that ci ≥ 0 is satisfied for both i. Adding I1 and I 2 gives the aggregate infrastructure investment in the whole economy

I = I1 + I 2 = αK δ − 2 λ [δ + δ ( λ − 1)t ] . Note that the difference in infrastructure investment, I1 − I 2 =

α (k1 − k2 ) , depends only on the difference in capital allocation and is independent of the δ

exogenous asymmetry in initial conditions, a1 − a2 . In order to focus on the most interesting cases, we assume that αK ≥ 4τ , where

τ = λ [1 + ( λ − 1)t ] represents the opportunity cost of infrastructure investment for the governments. This says that the effect of infrastructure on capital is sufficiently large that it is worthwhile for a government to invest in infrastructure as long as it has 25 percent of the total capital. The choice of 4

10

times τ as the threshold is for convenience.11 For future reference, note that τ ≥ 1 because λ ≥ 1 . Moreover τ increases in λ : the greater the propensity for corruption, the larger the opportunity cost of infrastructure investment. Furthermore, τ decreases in t : the higher the tax rate, the greater the fiscal revenue, and hence the smaller the opportunity cost of infrastructure. When t = 0 , τ = λ ; when t = 1 ,

τ = 1. The proposition below presents the equilibrium outcome for the case with capital immobility.

Proposition 1: Under capital immobility, the governments’ infrastructure investments are I i = (αki − τ ) δ , and corrupt consumption levels are ci = S + tFi − I i . The aggregate infrastructure investment level is independent of how capital is allocated across units, and is given by I = (αK − 2τ ) δ .

Proposition 1 shows that infrastructure investment in one unit depends only on its capital endowment. As long as both units have more than τ α units of capital, both will invest positively in infrastructure. When capital is evenly distributed between the two units ( k1 = k 2 = 0.5K ), infrastructure investments will be the same. In this case, total output is greater in unit 1 than in unit 2 ( F1 ≥ F2 ) since

a1 ≥ a2 . Hence c1 ≥ c2 : corrupt government consumption in unit 1 is greater than that in unit 2.

2.2 Participation equilibrium under capital mobility Now suppose capital is perfectly mobile across units. Under capital mobility, capital will flow from the unit with the lower after-tax marginal rate of return to capital to the unit with the higher rate. At an interior equilibrium, the rates in the two units must be equalized:

11

If the effect of infrastructure on capital is not large enough, then technologically it becomes difficult or impossible for both regions to invest in infrastructure. When αK ≤ τ , it is not worthwhile for any region to invest in infrastructure even if it has all the capital.

11

(1 − t ) ∂F1 ∂k1 = (1 − t ) ∂F2 ∂k 2

(5)

Solving for k1 , k 2 , (recalling that k1 + k 2 = K ) we have

(α − β )( I1 − I 2 ) a1 − a2 + 2γ 2γ (α − β )( I1 − I 2 ) a1 − a2 k 2 = 0.5K − − 2γ 2γ k1 = 0.5K +

(6)

From (6), capital allocation depends on two factors: (1) an endogenous factor, I1 − I2 , the difference between the two governments’ infrastructure investments; and (2) an exogenous factor, a1 − a2 , the difference between the two units’ initial conditions. While the existing literature has focused on the relationship between I1 − I2 and k1 − k2 , we will show that the exogenous factor, a1 − a2 , can have a profound impact on both I1 − I2 and k1 − k2 in equilibrium. G1’s problem is to choose I1 to maximize U 1 = (1 − t ) F1 ( k1 , I1 , I 2 ) + λc1 subject to its budget constraint and subject to the capital mobility constraint (6), given G2’s choice of I 2 . Solving the problem gives G1’s best response function I1 ( I 2 ) . G2’s best response can be found similarly. A Nash equilibrium of the investment game is a pair of strategies ( I1 , I2 ) such that both G1 and G2 play a best response against the other’s strategy. To solve for G1’s best response, substituting its budget constraint into its objective function, we derive the following first order condition

[1 + ( λ − 1)t ][∂F1 ∂I 1 + ∂F1 ∂k 1 ∂k1 ∂I 1 ] − λ = 0

12

(7)

Comparing this with (4), the first order condition in the case with capital immobility, we see that the difference is the term ∂F1 ∂k 1 ∂k1 ∂I 1 , which measures the indirect effect of infrastructure investment

I1 on unit 1’s output due to the additional capital it attracts to the unit. Previous papers have often pointed to this infrastructure- and output-increasing effect of capital competition to argue that fiscal decentralization and the liberalization of capital controls increase welfare (see, e.g., Qian and Roland 1998). Two assumptions, sometimes made explicitly and sometimes implicitly, are crucial for the argument: that the units are completely identical and that capital has rapidly diminishing returns. As we show below, when these assumptions are relaxed, no symmetric equilibrium exists. In the proposition below, we provide necessary conditions for the existence of a participation equilibrium. A more specific version with necessary and sufficient conditions is contained in the Appendix, along with all technical proofs.

Proposition 2: There exists a participation equilibrium, in which both units invest positively in infrastructure only if (i) (3α + β )(α − β ) < 4γδ and (ii) K ≥ δ ( a1 − a2 ) | γδ − α (α − β )| .

Proposition 2 shows that for the existence of a participation equilibrium, capital and infrastructure must have rapidly diminishing returns (large γ and δ ), capital cannot be too scarce (relatively large K ), and initial inequality between the two units cannot be too large (small a1 − a2 ).

Corollary 1: Suppose the two units are identical: a1 = a2 = a . When (3α + β )(α − β ) < 4γδ , there exists a symmetric equilibrium in which I1 = I 2 =

γ [(α + β ) K − 4τ ] + 2(α − β )a and each gets half of the 2[2γδ − (α + β )(α − β )]

total capital. The total infrastructure investments are greater than those when capital is immobile.

13

Corollary 1 reproduces the result in the existing literature that competition for capital forces governments to invest in welfare-enhancing infrastructure (e.g., Qian and Roland 1998). As is clear from Corollary 1, this result depends on the assumptions that the units are identical and that capital and infrastructure have rapidly diminishing returns. When some of these assumptions fail to hold, a symmetric equilibrium simply does not exist. More generally, when the conditions of Proposition 2 do not hold, no participation equilibrium exists. In such cases, we need to consider polarization equilibria.

2.3 Polarization equilibrium under capital mobility In this section, we explore conditions under which there exists a polarization equilibrium, in which G1 invests in infrastructure and gets all the capital, while G2 makes no investment and gets no capital.

Proposition 3: There exists a polarization equilibrium in which G2 makes no infrastructure investment and G1 invests I1 = (αK − τ ) δ , so that unit 2 gets no capital while unit 1 gets all the capital, when (i)

4γδ − (3α + β )(α − β ) ≤ 0 ; and (ii) K ≥ [τ (α − β ) − δ (a1 − a2 )] [α (α − β ) − γδ ] ; and (iii) a technical condition holds (see the Appendix). In this equilibrium, corrupt consumption levels of the governments are c1 = S + tF1 ( I1 , K ) − I1 and c2 = S .

Proposition 3 gives a set of sufficient conditions for a polarization equilibrium to exist. In this equilibrium, the government of the disadvantaged unit (G2) has no chance of attracting capital via infrastructure investment, and so invests nothing. Expecting to “beat” G2, G1 makes a large amount of infrastructure investment. Consequently, all the capital flows to unit 1. From the conditions of Proposition 3, it is clear that polarization is more likely to occur when initial inequality a1 − a2 is relatively large. This is very intuitive. When a1 − a2 is large, G2 sees little hope in competing with G1, and G1 finds it easy to “drive out” G2. In addition, for polarization to occur,

14

γ must be small—that is, returns to capital must not diminish too fast. When γ is small, small changes in infrastructure investment in one unit, holding infrastructure investment by the other unit fixed, can have a large impact on capital allocation across units. In other words, the smaller γ is, the more intense is competition for capital.12 Comparing the polarization equilibrium under capital mobility (Proposition 3) with the outcome under capital immobility (Proposition 1), we have

Proposition 4: Suppose the conditions of Proposition 3 are satisfied. Under capital mobility, G2 spends nothing on infrastructure and gets less capital and a smaller payoff in equilibrium than under immobility, while G1 spends more on infrastructure and gets more capital and a higher payoff. The infrastructure investment by G1 in the polarization equilibrium under capital mobility is greater than the total infrastructure investments by G1 and G2 when capital is immobile.

Proposition 4 shows that under certain conditions, competition for capital exacerbates initial inequalities and hinders economic development in the disadvantaged unit. Such competition does not discipline the government in the disadvantaged unit by forcing it to reduce corruption and improve its business environment through infrastructure investment. Instead, the government becomes more corrupt and spends all its resources on its own consumption. The government in the advantaged unit invests more

On the other hand, when γ is too small (e.g., very close to zero), then condition (iii) in Proposition 3 may not hold, so the polarization equilibrium fails to exist. In this case, since a small infrastructure investment advantage can cause a large amount of capital to cross the border, G2 may try to overtake G1 and get all the capital. Consequently, there will be no pure strategy equilibrium in this case. To see this, consider the special case in which γ = 0 : capital

12

k1 = K , k 2 = 0 whenever I1 > I 2 ; and k1 = 0, k 2 = K whenever I1 < I 2 . Consider I1 > 0, I 2 = 0 . Given I 2 = 0 , G1’s best response would be I1 = (αK − τ ) δ . But then if G2 invested (αK − τ ) δ + ε , all the capital would flow to unit 2. Hence I1 > 0, I 2 = 0 cannot be an equilibrium. We will not try to characterize mixed strategy equilibrium of the model,

exhibits constant returns. In this case, capital allocation is discontinuous:

because not much new insight can be gained. We work out a numerical example (available from the authors upon request) with both a participation and a polarization equilibrium to illustrate that given reasonable parameters, the main propositions in the paper will hold.

15

in infrastructure in absolute terms. Its corruption consumption may also increase in absolute terms since it has a larger budget due to the capital inflow and better business environment; but it may well decrease in relative terms since the total budget is much larger.

3 Welfare analysis We define social welfare as total after-tax output plus public goods provided in both units. As a benchmark, we solve for the first best solution. The social planner chooses infrastructure investment

( I1* , I 2* ) and capital allocation ( k1* , k 2* ) to maximize Π = (1 − t ) F1 + (1 − t ) F2 + m subject to the aggregate budget constraint I1 + I2 + m = 2 S + t ( F1 + F2 ) . Substituting the budget constraint into the objective function, we get the following first order conditions:

αk1 + β k 2 − δI1 = 1 = αk 2 + β k1 − δI 2 αI1 + βI 2 + a1 − γk1 = αI 2 + βI1 + a2 − γk 2

(8)

Solving these equations yields Ii =

0.5(α + β ) K − 1 0.5(α − β )( ai − a j ) + δ γδ − (α − β )2

ki = 0.5K +

0.5δ ( ai − a j )

(9)

γδ − (α − β )2

For the second order condition, note that k1 + k 2 = K is always binding, so the social welfare function

Π = F1 + F2 − I1 − I 2 + 2 S can be expressed as Π( I1 , I 2 , k1 ) . It can be checked that Π( I1 , I 2 , k1 ) is concave in ( I1 , I 2 , k1 ) when γδ − (α − β ) 2 > 0 . Therefore, Equation (9) gives the first best allocation when γδ − (α − β ) 2 > 0 and I2 ≥ 0 and k 2 ≥ 0 . When γδ − (α − β ) 2 ≤ 0 , the social welfare function is convex. In this case the solution must be at the corner, hence k1 = K and k 2 = 0 . By Equation (8),

I1 = (αK − 1) δ and I 2 = max {0, ( βK − 1) δ } .

16

Comparing Equation (9) with Proposition 1, we see that there are three factors that make the outcome of decentralization with capital immobility deviate from the first best solution. First, under decentralization, governments care about their own corrupt consumption as well as output, introducing distortions in resource allocation. This is reflected in that the opportunity cost of infrastructure investment under decentralization, τ , is greater than one. Second, governments do not take into account the positive externalities of their infrastructure investment on the other unit, so their equilibrium infrastructure investments are independent of the externality parameter β . These two effects reduce the incentives of governments to invest in infrastructure; consequently, total infrastructure investment under capital immobility, I = (αK − 2τ ) δ , is smaller than that in the first best solution, I * = [(α + β ) K − 2)] δ . A third effect is that the capital distribution under capital immobility may not be optimal, leading to further distortion in governments’ infrastructure investments. When the two units are identical and certain conditions hold, Corollary 1 shows that competition for capital generates additional incentives for infrastructure investment that are absent under capital immobility. With symmetric units, competition for capital will in general increase social welfare by offsetting governments’ tendency to under-invest due to corruption and externalities. However, there is a caveat. Comparing Corollary 1 and (9) reveals that, when the two units are identical ( a1 = a2 = a ), it is possible for infrastructure investments in the symmetric equilibrium under capital mobility to exceed the first best level. Therefore, when all the conditions hold so that there exists a participation equilibrium and the initial conditions of the two units are sufficiently close, competition for capital increases social welfare so long as governments do not invest “too much”. On the other hand, when γδ − (α − β ) 2 ≤ 0 , a participation equilibrium does not exist under capital mobility. When the conditions of Proposition 3 hold, the polarization equilibrium has the same capital allocation as the first best solution ( k1 = K , k 2 = 0 ). By Proposition 4, the total infrastructure investment under capital mobility is greater than that under capital immobility. Therefore, the social welfare is again greater under capital mobility than under capital immobility.

17

The welfare comparison between capital mobility and immobility can be different when

(α − β ) 2 ≤ γδ ≤ α (α − β ) . In this case, the first best solution calls for both units to make infrastructure investments, while in the polarization equilibrium of Proposition 3 only G1 invests in infrastructure. The next proposition shows that in this case, it is indeed possible that capital mobility reduces social welfare relative to capital immobility. For concreteness, we assume that under capital immobility the capital allocation across units is even.

Proposition 5: Suppose the conditions of Proposition 3 hold. When (α − β ) 2 ≤ γδ ≤ α (α − β ) and a

technical condition holds (see the Appendix), social welfare in both units in the polarization equilibrium under capital mobility is lower than that under capital immobility.

Proposition 5 says that the greater inter-unit inequality caused by polarization may hurt economic performance—and social welfare—systemwide. The idea is simple. In our model, infrastructure investments create positive externalities, improving the productivity of capital in the other unit. When such externalities are sufficiently strong, the first best solution calls for the disadvantaged unit to invest in infrastructure and to be allocated some capital. Polarization under capital mobility does not take advantage of such externalities, and hence may reduce output and social welfare.13

4

Robustness

4.1

Intermediate degree of asymmetry

To make our main point in the simplest possible way, we have chosen to compare two polar cases. We showed that: (1) when the two units are identical or sufficiently similar, a participation equilibrium exists

13

Even if infrastructure externalities are small, so that output under capital mobility is greater than under capital immobility, there may be other benefits to internationally (or inter-regionally) balanced development that our narrowly defined social welfare function does not capture.

18

under certain conditions; and (2) when the two units are sufficiently different, a polarization—but no participation—equilibrium exists under some conditions. The arguments for capital mobility that focus on the symmetric case rely on strong assumptions (identical—or very similar—units, rapidly declining marginal returns) that may not be realistic in many applications. We have shown that if these assumptions are not met (e.g., if units have substantially different exogenous productivity), the conclusion about competition for capital is reversed. This point does not depend on the peculiarities of these two polar cases, and is also valid for intermediate degrees of difference between the two units. Suppose that the asymmetry is not extreme and that a participation equilibrium exists. Note that in the participation equilibrium, the differences in infrastructure investments and in capital allocation between the two units are (see the proof of Proposition 2) I1 − I 2 =

α (a1 − a2 ) δ (a1 − a2 ) and k1 − k 2 = γδ − α (α − β ) γδ − α (α − β )

Clearly, both are strictly increasing in a1 − a2 , the asymmetry in the initial conditions of the two units. As asymmetry increases starting from symmetry ( a1 − a2 = 0 ), equilibrium inequality between the two units in both infrastructure and capital rises. When γδ − α (α − β ) is relatively small, the endogenous inequality between the two units increases rapidly in the asymmetry of initial conditions. When the exogenous asymmetry becomes sufficiently large, the endogenous inequality will hit the extreme upper bound— polarization. Thus, although we compare the polarization and participation equilibria in previous sections, the main results hold more generally comparing different participation equilibria, so long as the exogenous difference in initial conditions is sufficiently large. In short, when two units start off with sufficiently different levels of capital productivity, competition for capital does not motivate the government of the disadvantaged unit to invest more in business-friendly policies and less in corrupt consumption. On the contrary, such governments will invest less in infrastructure and become more corrupt. The greater the initial asymmetry ( a1 − a2 ), the less the

19

disadvantaged unit will invest in infrastructure, and the less capital it will receive. Beyond a certain degree of initial asymmetry, the disadvantaged unit becomes more corrupt and receives less capital than under capital immobility. Not only is welfare unevenly distributed, total social welfare of both units may even be lower. While in the polarization equilbrium G2 makes zero infrastructure investment and unit 2 receives no capital at all, this is not at all essential to the argument.

4.2 Endogenous tax competition So far we have assumed that tax rates are fixed. In this section, we will argue that the main insights of the model are still valid when endogenous tax competition is feasible. First consider the case of capital immobility. Substituting in the budget constraint, G1’s objective function can be written as U 1 = (1 + (λ − 1)t1 ) F1 + λS − λI1 . Clearly, when λ = 1 , it does not matter what tax rate G1 chooses. When λ > 1 , G1 will choose the highest possible tax rate. Let t ≤ 1 be the upper bound of tax rate for both G1 and G2. So when λ > 1 , t1 = t 2 = t . It is easy to see that Proposition 1 still holds except that the tax rate now is t . It is also easy to see that the first best infrastructure investments and capital allocation given by Equation (9) are still correct, because the first best solution corresponds to λ = 1 , in which case the exact level of the tax rate does not matter. Now we turn to the case of capital mobility. Consider first the participation equilibrium when the two units are relatively equal and γ is relatively large. Capital mobility equation (5) now becomes

(1 − t1 )(αI1 + βI 2 + a1 − γk1 ) = (1 − t 2 )(αI 2 + βI1 + a2 − γk 2 ) . Together with k1 + k 2 = K , we have, for

i = 1, 2, j = 3 − i , ki =

(1 − ti )(αIi + βI j + ai ) − (1 − t j )(αI j + βI i + a j ) + (1 − t j )γK

γ (2 − t1 − t 2 )

It follows that

20

∂ki (1 − ti )α − (1 − t j ) β = ∂I i γ ( 2 − t1 − t 2 ) (1 − t j )[(α + β )( I1 + I 2 ) + a1 + a2 − γK ] ∂ki =− γ (2 − t1 − t 2 ) 2 ∂ti Obviously ∂ki ∂ti < 0 . Also, when (1 − ti )α > (1 − t j ) β , ∂ki ∂I i > 0 . This means that only when a unit’s tax rate is not too high relative to its rival’s tax rate can infrastructure investment increase capital inflow to the unit. Aside from the fact that governments have two policy instruments (infrastructure and tax rate) to compete for capital, a participation equilibrium, if it exists, can be derived in the same way as before. The equilibrium tax rate will be lower than under capital immobility since governments choose the maximum tax rate under capital immobility. Moreover, equilibrium infrastructure investment is lower than if the tax rate was exogenously fixed. In other words, because governments can lower their tax rates to attract capital, their incentive to invest in infrastructure to compete for capital is reduced.14 Although the derivation becomes very cumbersome, the analysis of the participation equilibrium parallels that under a fixed tax rate. The conditions for the existence of participation equilibria are likely to be more stringent. Finally, we argue that polarization can still occur in equilibria under endogenous tax competition. In a polarization equilibrium, G2 makes no infrastructure investment ( I 2 = 0 ) and also sets a tax rate of zero ( t 2 = 0 ). (Since G2 expects to get no capital, there is nothing to lose in lowering the tax rate all the way to zero.) Expecting to get all the capital, G1 will choose a level of infrastructure investment at

I1 = (αK − τ 1 ) δ , where t1 (and hence τ 1 ) is chosen so that the after-tax return to capital in unit 1 is sufficiently great that G2 cannot profitably overtake G1. If G1 has to set t1 = 0 = t 2 , then Proposition 3 will hold. (In this case, G1 spends S on infrastructure.) If G1 can set a positive tax rate, then Proposition 3 only needs some minor modifications. Because of the substitution between the two policy instruments,

14

This echoes the theme of Cai and Treisman (2001), which shows that governments may substitute one policy instrument (here the tax rate) for another (e.g., infrastructure investment) in competing for capital, rendering the welfare consequences of fiscal federalism ambiguous.

21

the infrastructure investment by G1 in the polarization equilibrium may or may not exceed the total infrastructure investments by G1 and G2 under capital immobility (modification of Proposition 4). Consequently, with endogenous tax competition, it is more likely that social welfare under capital mobility is smaller than under capital immobility; i.e., Proposition 5 is more likely to hold. In summary, two general points can be made about endogenous tax competition in our model. First, because governments can substitute tax competition for infrastructure investments, the incentives to make infrastructure investments are reduced in both the participation and polarization equilibria. Second, polarization can still occur with endogenous tax competition, and the outcome is even worse when it does.

5 Applications 5.1 Interregional capital flows in post-communist Russia One setting our model might illuminate is the competition over private investment among regions within a large country that has recently undergone capital market liberalization. (Although it includes just two units, a similar logic is likely to apply when there are more.) Based on the model, one might expect to see the regions richest in infrastructure, resources, and human capital at the start of liberalization adopting pro-business policies and reducing corruption. Exogenously less attractive regions should be becoming more corrupt. These regions should receive less investment than before—and maybe even net outflows of capital to the more exogenously attractive regions. In Russia, the transition from communism since 1991 has liberated private capital to flow relatively unimpaired among the federation’s 89 regions. The resulting financial system is imperfect in numerous ways. Still, by the mid-1990s there were few remaining restrictions on the flow of funds from one region to another. Does the pattern of change fit our model’s predictions? Measuring change over time in the degree of corruption of Russian regional governments is difficult for obvious reasons. However, we found three alternative indicators of the degree to which governments’ policies were favorable to business. First, the Finance Ministry’s reports of regional budget

22

execution provide some evidence of the way regional governments allocate funds. Since 1996, these accounts have included “spending on the development of markets”. We constructed an indicator of market infrastructure spending equal to the share of the region’s budget that was spent on development of markets, transport, roads, communications, and information technology. We averaged this for 1996-98. Second, many authors have noted the particularly pernicious effects of local and regional government corruption on the startup and survival of new small businesses (e.g., Frye and Shleifer 1997). We treated such small enterprises as canaries in the mine-shaft, and estimated change in the degree to which a region’s political environment is “business-friendly” by the percentage change in the number of small enterprises registered in a region, from 1995, the first year for which we had figures, to 1999.15 Third, we used an index of the “institutional potential” of the region defined as “the degree of development of the leading institutions of a market economy”, constructed by the business magazine Ekspert. We used the rating for 2001, the latest available, as a cumulative measure of the extent of enactment of pro-market policies in the transition period. To measure the initial attractiveness to investors of the regions, we constructed an index of infrastructure and human capital as of the early 1990s. This index was the sum of five separate indicators—the percentage of urban families with home telephones as of 1990, the number of public buses per 1,000 inhabitants as of 1992, the percentage of roads that were paved as of 1990, the number of research and development organizations as of 1992, and the percentage of the employed population with higher education as of 1995 (the first year for which this was available)—each of which was expressed as a percentage of the mean.16 Our index was the log of this sum, since Moscow and St Petersburg came out so far above the other regions. Figure 1 shows that in regions that started the reform period with relatively more developed infrastructure and human capital, the regional government tended to devote a larger share of total 15

There was a change in Goskomstat’s definition of small enterprises in 1996, but since this was the same for all regions and we are interested in the cross-regional comparison, we have ignored it.

16

For lack of a theoretical reason to do otherwise, we attribute equal weight to each indicator.

23

expenditures to market-supporting infrastructure in the late 1990s. The correlation was .59 (significant at p < .001). In Moscow and St Petersburg, the two regions with by far the most developed initial infrastructure and human capital, governments allocated more than seven percent of total spending to this. By contrast, in infrastructure-poor Altai and Tyva republics, governments devoted less than two percent of expenditures to market-supporting infrastructure.17 Figure 2 shows that the number of small enterprises tended to grow much faster in the late 1990s in regions that started the decade with better infrastructure and human capital. This may reflect just the productivity-enhancing effects of infrastructure and human capital. But it may also reflect in part more business-friendly policies in infrastructure-rich regions. The latter interpretation is supported by statistical analysis. We calculated the partial correlation between the percentage change in the number of small enterprises in 1995-9 and the average share of regional government spending devoted to market infrastructure in 1996-8, controlling for our index of initial infrastructure and human capital. The correlation was .32 (significant at p < .01), suggesting that regardless of their initial infrastructure and human capital, regions that adopted pro-market policies saw a larger increase in the number of small enterprises. Controlling for regional policy (the budget share spent on market infrastructure in 1996-8), there was no correlation between the initial conditions index and the growth of small firms. Figure 3 plots Ekspert magazine’s rating of the degree of development of market institutions in regions as of 2001 against the initial infrastructure and human capital index. By 2001, capital mobility should have had time to influence government policy on this. There is a highly significant positive relationship between initial infrastructure and the development of market institutions.18 But do the more capital-friendly policies in regions that started with more developed infrastructure and human capital translate into higher inflows of investment capital? Is capital flowing out

17

The correlation excluding Moscow and St Petersburg is .40, still significant at p < .01.

18

Since the rating is just an ordinal ranking (and Moscow and St Petersburg cannot rise above the top and second slots despite extremely developed infrastructure), the effect appears to taper off at high levels. But this is probably just an artifact of the ordinal rankings.

24

of regions with poorer infrastructure and policies into those that are more business-friendly? The data available to judge this are not ideal. We calculated two alternative estimates of net regional capital inflows. First, we computed the difference between total investment in non-financial assets and total savings of the population in each region in 1998, the latest year for which we had data. Second, we calculated the difference between total bank credits issued and total savings in each region, also in 1998.19 These are both imperfect estimates of the net inflow of private capital. For instance, both credit issues and non-financial investment include government loans. Still, they serve as rough approximations.20 Figure 4 shows that regions that started with better infrastructure and human capital tended to receive a net inflow of capital, while those with poorer infrastructure and human capital were exporting locally generated savings. Is it the infrastructure and human capital themselves that attract the investment flows, or is it associated business-friendly policies? Statistical analysis suggests that it is both. The partial correlation between the investment-minus-savings variable and the economic policy variable (the government’s spending on market infrastructure), controlling for initial attractiveness (the infrastructure and human capital index), is positive and significant (.28, p < .02). So is the partial correlation between the investment-minus-savings variable and initial attractiveness, controlling for economic policy (.24, p < .04). Figure 5 plots initial infrastructure and human capital against regional credit issued minus savings. In regions with better infrastructure and human capital—most notably, Moscow and St Petersburg— larger amounts were issued in credit by the banking system than the population saved. But those regions with the worst initial conditions received far less in banking system credits than their populations saved.21

19

In both cases, the savings variable is measured as the difference between total income of the population and expenditures on goods and services plus taxes and other obligatory payments; it includes growth in bank deposits, purchases of securities and foreign currency, and the surplus of incomes of the population over their recorded outlays.

20

If government loans and investments go disproportionately to aid less developed regions, this should reduce the positive relationship between investment climate and credits or investment that we found.

21

Statistical analysis suggests it may have been the infrastructure and human capital that explains this rather than government policies: the partial correlation between the credit-minus-saving variable and the economic policy variable, controlling for initial attractiveness, is insignificant; that between the credit-minus-saving variable and the initial attractiveness variable, controlling for economic policy, is highly significant (.68, p < .01). However, this just reflects the high initial attractiveness and credit receipts of Moscow and St Petersburg. When these are excluded, the

25

A final indicator of capital flows is the level of foreign investment in the regions. Figure 6 shows that foreign investment tended to be higher in regions with higher initial infrastructure and human capital. Regional government spending on market infrastructure was not correlated with the 1998 foreign investment variable, but the Ekspert rating of “development of market institutions” for 1996 was, suggesting that policy may have helped to attract foreign investment. Thus, in the period since economic reform liberalized capital movements, the regional pattern of government policy and capital flows fit the predictions of our model quite closely. In regions that had initial advantages in infrastructure and human capital, governments have tended to spend relatively more on market development and infrastructure, and the number of small enterprises has increased more rapidly (apparently stimulated by the relatively higher government infrastructure spending rather than inherited advantages). By 2001, such initially-advantaged regions had developed market-supporting institutions that were rated more highly than those of other regions. Better initial infrastructure and human capital also correlated with larger net inflows of capital, apparently induced at least in part by the more business-friendly patterns of government spending and the better market institutions in such regions. [Figures 1-6 about here]

5.2 Capital account liberalization in the developing world The 1980s and 1990s witnessed a worldwide trend toward capital account liberalization, during which many developing countries opened their markets to external capital. Did the newly open developing countries invest in infrastructure, reform their bureaucracies, and improve their business environments sufficiently to attract in large flows of mobile capital? Did capital mobility provide the discipline that the conventional analysis predicts?

partial correlation between credit and economic policy, controlling for initial attractiveness, is greater than that between credit and initial attractiveness, controlling for economic policy.

26

Several points need to be considered when applying our model to international competition. First, like the previous literature we critique, we treat the extent of capital mobility as exogenous. In fact, however, national governments will be less likely to open their capital markets if they expect net capital outflows. Thus, actual net outflows of capital from low-productivity countries should be rare. Rather, we should expect such countries to be reluctant to remove capital controls. This conforms with the observation that it usually takes considerable IMF pressure to get Third World countries to liberalize. Second, when Third World countries do open their capital markets, such reforms are often part of a package enacted under pressure from international organizations. Donors often insist on anti-corruption reforms as a condition for receiving international funds, and international organizations such as the World Bank also often fund infrastructure projects directly. Thus, to assess the effects of capital mobility on government quality, one needs to control for the interventions of international donors.22 Given this, our model would imply that countries where initial capital productivity is low: (a) will be reluctant to open their capital markets, (b) will receive at most small net inflows of capital when they do, and (c) will not generally reduce their levels of corruption after capital market liberalization, unless bribed or pressured to do so by international donors.23 These predictions appear highly consistent with actual experience in the 1990s. The developing world’s share in global capital flows was tiny, representing less than 8 percent of the total in 2000 (see Table 1). At the same time, this share was dropping—from 11.8 percent in 1991 to 7.6 percent in 2000, despite growth in the developing countries’ share in global output from 19.8 to 22.5 percent. Net private inflows of capital to developing countries were very low in most years—they peaked in 1996 at $129 billion, compared to total global capital flows that year of $2.4 trillion. After the Asian

22

For these reasons, applications to capital competition within countries—like that presented in the previous section—are less complicated. Within a country, liberalization of capital flows is often simply imposed on regional governments by the center and central pressures for regional government reform are likely to be relatively uniform.

23

Another simplifying assumption of the model is that total capital worldwid, K, is fixed. In fact, the world capital stock grows over time, so one should expect to see small annual inflows even into low-productivity countries.

27

financial crisis of 1997-9, net flows even turned negative.24 At the end of a decade of capital market integration, private capital appeared on balance to be flowing out of, rather than into, the developing world.25 Even this paints too rosy a picture of the capital accounts of the least competitive economies. Capital inflows to the developing world were highly concentrated on a dozen or so success stories in Latin America and Asia—countries such as China, Mexico, and Brazil. Despite significant capital market liberalization in many countries and low capital saturation, Africa saw almost none of the increase. Private capital inflows to Sub-Saharan Africa fell from 3.9 percent of the region’s GNP in 1975-82 to 1.8 percent in 1990-98 (UNCTAD 2000). Inflows to North Africa fell even more drastically, from 7.2 to 0.8 percent of GNP in the same period. In the 1990s, in both regions outbound profit remittances and interest payments were larger than private capital inflows. While developing countries’ share of worldwide foreign direct investment (FDI) inflows increased from 17.1 percent in 1988-90 to 21.4 percent in 19982000, Africa’s share fell from 1.8 to 0.8 percent during the same decade (UNCTAD 2001, p.256), and most of that 0.8 percent was concentrated on countries with oil and mining sectors (and therefore a natural productivity advantage) (Mishra, Mody, and Murshid 2001). During this period, annual FDI to a number of countries—Algeria, Libya, Morocco, Benin, Liberia, Mauritania, Niger, Nigeria, Rwanda, Sierra Leone, and Swaziland—fell in absolute terms (Ibid, p.292). Capital flight often increased after capital market liberalization. In various countries that opened their capital markets—Egypt, Mauritius, Uganda—capital outflows by residents rose substantially in relative terms in the 1990s (UNCTAD 2000, p.37). Even before capital accounts were liberalized, huge sums left the continent as unofficial capital flight. Starting from the “errors and omissions” data in

24

The picture would not be changed by including short term capital flows. We do not have data on such flows broken down into private and public components, but total short term inflows came to just –$18.3 bn in 1999 and +$3.5 bn in 2000.

25

Another way of gauging total capital flows is to look at the current account, which measures the difference between domestic savings and investment. A current account surplus indicates net outflows. Developing countries’ aggregate current account was in surplus of $60.3 billion in 2000 (World Bank 2001, Ch.2).

28

balance of payments statistics, and correcting this for underreported external borrowing and trade misinvoicing, Boyce and Ndikumana (2001) estimate that capital flight for 25 Sub-Saharan African countries between 1970 and 1996 exceeded their total accumulated external debt. By the end of the 1990s, African residents held a larger proportion of their wealth overseas than residents of any other continent.

Table 1: Private capital flows to developing countries in the 1990s ($ bn) 1991 1992 1996 1997 1998 1999 62.1 99.3 279.3 299.8 280.3 219.2 1. Longterm private capital inflows to developing countries 16.9 93.1 150.3 228.3 188.5 246.9 2. Capital outflows plus “errors & omissions” 45.2 6.2 129.0 71.5 91.8 -27.7 3. Net private inflow (1) – (2) 12.4 13.2 14.4 9.9 7.6 4. Developing countries’ 11.8 share in total global private capital flows 19.2 22.1 23.2 21.6 21.7 5. Developing countries’ 19.8 share in total global output

2000 257.2

306.6 -49.4 7.6 22.5

Source: World Bank (2001, Tables 2.1, 2.2, 2.3).

Why have holders of capital—both international and domestic—generally avoided the African countries? Our model suggests two reasons. First, as emphasized by the new growth literature, capital will be more productive when combined with infrastructure, human capital, or other factors (which might include an effective law enforcement bureaucracy). If countries start with too low an initial level of these, they will not be able to lure capital away from their rivals. Second, given the hopelessness of attracting capital through investment in such productivity-enhancing factors, they will reduce their growthpromoting investments and embezzle all or most tax revenues. A variety of evidence suggests that investors do expect returns on investments in Africa to be lower because of poor infrastructure and human capital or greater expropriation risk. The increasing infrastructure gap, along with poor macroeconomic policies and political risk, are among the main reasons

29

business respondents give to explain why they invest so little in Sub-Saharan countries (Bhattacharya, Montiel, and Sharma 1997).26 African countries have only 55 kms of rural highways per thousand square kilometers, compared to more than 800 kms in India, and the average number of telephones per capita in Asia is 10 times higher than in Africa (Collier and Gunning 1999, pp.71-2). We do not have a way of testing the prediction that corruption will increase as countries give up on attracting capital. But since liberalizing their capital markets these countries do not seem to have devoted a noticeable increase in funds to building infrastructure that would attract international business. In short, capital account liberalization does not appear to have resulted in a significant net inflow of capital into the most underdeveloped countries. And there is little evidence that governments in these countries have been pressured by the competition for capital to institute more business-friendly policies.

6 Conclusions A common view among political economists is that competition between governments to attract capital should render them less corrupt and more friendly toward business. This view—often adduced in debates over both political decentralization and globalization—is usually justified by reference to a simple, standard model, in which all governments are forced to adopt good policies by the fear of losing mobile capital. Reexamining this model, we showed that the disciplining effect of capital competition only exists under certain conditions, and given alternative, empirically plausible assumptions exactly the opposite logic may apply. Only if units are initially quite similar (have similar resources, human capital, infrastructure) will capital mobility improve government quality. If the units start out sufficiently heterogeneous (and 26

The point that investment is deterred by political risk is quite consistent with our argument. Country risk includes both an exogenous and an endogenous component. Risk is higher in countries subject to earthquakes (exogenous) and in those where governments expropriate investors (endogenous). The exogenous risk factor is part of what we include in the initial productivity parameter. According to our model, investment should be higher where exogenous risk is lower. Endogenous risk reflects government policy, which we model as an “infrastructure” investment decision. We predict that governments will invest less in infrastructure—including, say, protections against

30

competition for capital is sufficiently intense), the only equilibrium under capital mobility will involve polarization. The government of the less initially productive unit will do less to attract capital and embezzle more than it would under a system of effective capital controls. Capital will flow out of the exogenously disadvantaged unit into the more advantaged one. In such a world, capital market liberalization threatens to reduce growth and increase corruption in the poorest countries, while benefiting the richest. This might explain why results of capital market liberalization in some developing countries have been disappointing. In much of Sub-Saharan Africa, the lowering of capital controls seems mostly to have facilitated capital flight, making it easier for corrupt officials to transfer their gains into safe Swiss bank accounts. Within countries that have liberalized internal capital flows—Russia, for instance—the more developed regions seem to have siphoned savings from their poorer counterparts, reducing investment in the latter and increasing interregional inequality. Competition for capital does not appear to have reduced corruption in the country’s less well-endowed regions, many of which are ruled by semi-authoritarian regimes that have become increasingly predatory. Our argument should not be taken as a blanket endorsement of capital controls. There are wellknown efficiency reasons for favoring the free flow of capital. Instead, our paper suggests that for disadvantaged countries or regions to reap the benefits of free capital flows, certain conditions need to be met. External aid—if used to fund market infrastructure, improve human capital, or insure against exogenous risks—may reduce the initial productivity gap between countries to the point where capital competition will have the desired, disciplining effect. In a decentralized state, centrally funded public investment in infrastructure and human capital in underdeveloped regions may enable such regions to compete, giving their governments an incentive to reform themselves. Such external interventions—if sufficiently large and well-directed—can turn polarizing competition into a beneficial force. The challenge, of course, is how such central investments or international aid can be turned into infrastructure expropriation by their own bureaucrats—if their exogenous productivity parameter is lower. Thus, endogenous country risk will be higher—and private investment lower—in initially-low-productivity countries.

31

and human capital without being diverted by existing corrupt governments. To this, we do not have any simple answers.

32

Appendix: Proofs Proposition 2’: There exists a participation equilibrium, in which both regions invest positively in

infrastructure, if one of the following is true: 1. (i) γδ − α (α − β ) > 0 ; and (ii) K ≥ δ ( a1 − a2 ) [γδ − α (α − β )] ; and (iii)

γ [(α + β ) K − 4τ ][γδ − α (α − β )] > [(α + β )γδ − 2αβ (α − β )]a1 − [(3α − β )γδ − 2α 2 (α − β )]a2 2. (i) 0.25( 3α + β )(α − β ) < γδ < α (α − β ) ; and (ii) K ≥ δ ( a2 − a1 ) [γδ − α (α − β )] ; and

γ [(α + β ) K − 4τ ][γδ − α (α − β )] < [(α + β )γδ − 2αβ (α − β )]a2 − [(3α − β )γδ − 2α 2 (α − β )]a1 Proof : From (6), ∂k1 ∂I1 = ∂k 2 ∂I 2 = 0.5(α − β ) γ . Using (2) and (6), we can rewrite (7) as

[4γδ − (3α + β )(α − β )]I1 + (α − β ) 2 I 2 = γ [(α + β ) K − 4τ ] + (3α − β )a1 − (α + β )a2

(*)

where τ = λ [1 + ( λ − 1)t ] . For this to be G1’s best response, the second order condition requires that

− ∂ 2U 1 ∂I12 = 4γδ − (3α + β )(α − β ) > 0

(∆ )

Under the same second order condition, G2’s best response similar to (*) can be derived. Solving these best responses gives

γ [(α + β ) K − 4τ ][γδ − α (α − β )] + [(3α − β )γδ − 2α 2 (α − β )]ai − [(α + β )γδ − 2αβ (α − β )]a j Ii = 2[2γδ − (α + β )(α − β )][γδ − α (α − β )] Substituting these back into (6) gives ki = 0.5K +

0.5δ (ai − a j )

γδ − α (α − β )

.

For these to define a participation equilibrium, it must be that Ii > 0, and ki > 0 for i = 1, 2 and the second order condition ( ∆ ) holds. Note that γδ − α (α − β ) > 0 implies ( ∆ ), which in turn implies that

2γδ − (α + β )(α − β ) > 0 , (3α − β )γδ − 2α 2 (α − β ) > 0 and (α + β )γδ − 2αβ (α − β ) > 0 . Since in equilibrium k1 − k 2 =

δ (a1 − a2 ) α (a1 − a2 ) and I1 − I 2 = , we have k1 > k 2 and I1 > I 2 if γδ − α (α − β ) γδ − α (α − β )

γδ − α (α − β ) > 0 . Therefore, when γδ − α (α − β ) > 0 , the existence of the equilibrium with both

33

units making positive investments requires that k2 > 0 and I 2 > 0 , which is case 1. When

0.25(3α + β )(α − β ) < γδ < α (α − β ) , the requirements are k1 > 0 and I1 > 0 , which is case 2. Q.E.D Proof of Corollary 1: When a1 = a2 = a , in both cases of Proposition 2’, condition (ii) holds trivially,

and condition (iii) becomes γ [(α + β ) K − 4τ ] + 2(α − β )a ≥ 0 , which is clearly satisfied, given our assumption that αK ≥ 4τ . Thus Proposition 2’ boils down to one very simple condition

(3α + β )(α − β ) < 4γδ . When this condition holds, there exists a symmetric equilibrium in which k1 = k 2 = 0.5K and I1 = I 2 = γ [(α + β ) K − 4τ ] + 2(α − β )a . In this symmetric equilibrium, we also 2[2γδ − (α + β )(α − β )] should have the returns to capital to be positive: ∂F1 ∂k1 = ∂F2 ∂k 2 = (α + β ) I1 + a − 0.5γK ≥ 0 , which requires that [α (α + β ) − γδ ] K + 2aδ − 2(α + β )τ ≥ 0 . Clearly, when a is sufficiently large, this condition is satisfied. Specifically, under our assumptions that a1 = a2 ≥ γK and αK ≥ 4τ , the condition is easily satisfied. To see this, note that since a1 = a2 = a ≥ γK , [α (α + β ) − γδ ]K + 2 aδ − 2(α + β )τ is greater than [α (α + β ) + γδ ]K − 2(α + β )τ = (α + β )(αK − 2τ ) + γδK ≥ 0 . By Proposition 1, the total infrastructure investment when capital is immobile is (αK − 2τ ) δ . Some algebra reveals that whenever [α (α + β ) − γδ ] K + 2aδ − 2(α + β )τ ≥ 0 , we have

2

γ [(α + β ) K − 4τ ] + 2(α − β )a αK − 2τ ≥ δ 2[2γδ − (α + β )(α − β )]

Q.E.D.

Proof of Proposition 3: Suppose the first condition of Proposition 2, 4γδ − ( 3α + β )(α − β ) ≤ 0 , is

violated. If both units stay in the contest for capital, each government’s objective function is convex in its infrastructure investment. This implies that the best responses of both G1 and G2 must be at their corresponding corner solutions. That is, given I 2 , G1 should invest either I1 = 0 , or at least

I 2 + [γK − (a1 − a2 )] (α − β ) so that k1 = K and k 2 = 0 ; given I1 , G2 should invest either I 2 = 0 , or at least I1 + [γK + (a1 − a2 )] (α − β ) so that: k 2 = K and k1 = 0 (Equation 6). Therefore, if an equilibrium exists in this case, it must be polar. Consider a hypothetical polarization equilibrium in which all the capital flows to unit 1: k1 = K and k 2 = 0 . If G2 expects to have no capital flowing into unit 2, it will not invest in infrastructure, since infrastructure does not increase output without capital. Without competition for capital from G2, G1’s optimal investment policy is given by Equation (4′):

I1 = (αK − τ ) δ . For this to be G1’s best response to I 2 = 0 , it requires that

34

(αK − τ ) δ ≥ [γK − (a1 − a2 )] (α − β ) since G1 must invest at least [γK − (a1 − a2 )] (α − β ) to attract all the capital. This condition means that K ≥ [τ (α − β ) − δ ( a1 − a2 )] [α (α − β ) − γδ ] . For G2 to be willing to give up and invest nothing in infrastructure, it must be that no other strategy yields higher payoff than zero investment. Because of its convex payoff function, this requires that G2’s investment choice I1 + [γK + (a1 − a2 )] (α − β ) that would attract all the capital to unit 2 given G1’s choice of I1 makes G2 worse off than I2 = 0 . That is, (1 − t ) F2 + λ ( S + tF2 − I2 ) ≤ λS , or equivalently, F2 I 2 ≤ τ . Some algebra can show that this will hold if the following condition is satisfied

[α (α − β ) − γδ −

0.5(α − β ) 2 (γδ − 2αβ ) (α − β ) 2 (βτ − a2δ ) ]K ≤ [τ (α − β ) + δ (a1 − a2 )] + γδ γδ

Under this and the conditions specified in the Proposition, there exists a polarization equilibrium. Q.E.D. Proof of Proposition 4: Clearly G2 is worse off in the polarization equilibrium under capital mobility,

since it could choose zero infrastructure investment under capital immobility but does not. G1’s payoff in the polarization equilibrium under capital mobility is U 1 = [1 + ( λ − 1)t ] F1 ( I1 , K ) − λI1 + λS , and its payoff under capital immobility is U 1 = [1 + ( λ − 1)t ]F1 ( I1 , I 2 , k1 ) − λI1 + λS . Thus, U 1 ≥ U 1 if and only if F1 ( I1 , K ) − τI1 ≥ F1 ( I 1 , I 2 , k1 ) − τI1 . Let V ( K ) = F1 ( I1 , K ) − τI1 , V ( k1 ) = F1 ( I1 , I 2 , k1 ) − τI1 . We know that V ( K ) = V ( K ) . By the Envelope Theorem, ∂V ( k1 ) ∂k1 = ∂F1 ( I1 , I 2 , k1 ) ∂k1 > 0 , therefore, V ( k1 ) ≤ V ( K ) . Finally, by Proposition 1, the total infrastructure investment when capital is immobile is (αK − 2τ ) δ , which is less that I1 = (αK − τ ) δ .

Q.E.D.

Proof of Proposition 5: In the polarization equilibrium under capital mobility, since I2 = 0, k2 = 0 , the

total output is simply F1 = (αI1 + a1 ) K − 0.5γK 2 − 0.5δ I12 . From Proposition 1, under capital immobility, the total output is

F1 + F2 = 0.5[(α + β )( I1 + I 2 ) + a1 + a2 ] K − 0.25γK 2 − 0.5δ I12 − 0.5δ I 22 Since I1 = (αK − τ ) δ , I1 = I 2 = ( 0.5αK − τ ) δ , some algebraic calculation can show that F1 ≤ F1 + F2 if and only if

[γδ − α (α − 2 β )] K 2 − 2[2 βτ + δ (a1 − a2 )]K ≥ 2τ 2 .

35

Q.E.D.

Figure 1: Regional Spending on Markets and Infrastructure as Percentage of

Regional spending on markets and infrastructure, 1996-98 average (%)

Total Regional Government Expenditures, Russia 1996-98 14 S-P

12

10 Kal

8

Mos C

Chel Yar Kem Tula Orel OrenVor Nizh Novs Lip Chuv Sam Kirov Penza Sar Mos O Novg Volg Belg Perm Brya Omsk Khab Sverd Vlad Kurg Udm Ryaz Kursk Smol Tver KalKrasdRost ArchUly Len TyumPrim Alt KIrk Krasy Chit Mari Tat NMurm Oss Kost BurIvano Bash Toms Ingu Yevr Khak Tamb Sakhl K-Ch Kalm Mord Astr AdygKar PskovStav VoloAmur Maga Sakha Dag Tyva Komi

6

4

2

Alt R

0 .2

.4

.6

.8

1.0

1.2

1.4

1.6

Initial infrastructure and human capital, early 1990s

Correlation: .59 (p < .01); without Moscow and St Petersburg, .40 (p < .01).

Source: Goskomstat, Ministry of Finance. Spending on “development of market infrastructure” + “transport, roads, communications, and information technology” as percentage of total regional budget expenditures. “Initial infrastructure and human capital” is log of sum of (1) percentage of urban families with home telephones as of 1990, (2) number of public buses per 1,000 inhabitants as of 1992, (3) percentage of roads that were paved as of 1990, (4) number of research and development organizations as of 1992, and (5) percentage of the employed population with higher education as of 1995, each expressed as proportion of mean.

36

Figure 2: Regional growth of small enterprises, Russia 1995-99

Change in number of small enterprises, 1995-99 (%)

200

Kal

100

Stav

S-P

Belg Chel Rost Mos O Omsk Sam Nizh Novs Len Mari Sakha Pskov Ryaz Lip Yar Vor Sar Chit Tamb MordUdm Alt K Khab Sakhl Adyg Uly KurgKost OrelKalm Bur Sverd Kirov Tula Volo Maga K-Ch Toms Kursk Bash Ivano Smol Irk Komi Tat Kar Chuv Vlad Arch Perm Krasd Brya Volg Novg Astr Kem Krasy Prim Penza Tver AmurMurmKal Alt R Oren Tyum Ingu Yevr Khak Tyva

0

Mos C

Dag

-100 .4

.6

.8

1.0

Initial infrastructure and human capital, early 1990s

Correlation: .30 (p < .01); without Moscow and St Petersburg, .28 (p < .02). Source: Goskomstat.

37

1.2

1.4

Figure 3: Degree of development of leading insitutions of a market economy, 2001; rating compiled by staff of Ekspert magazine. 100

Degree of development of market institutions 2001

80

60

40

20

Mos C

S-P

Mos O Sverd Sam Rost Chel Krasd Tat Novs Prim Krasy Nizh Sar Irk Kem Stav Bash Len Perm Volg Omsk Tyum Alt KYar Volo DagOren Vor KalUdm Kar Vlad Sakha Tula Khab Kirov Toms Komi Arch Ryaz Penza Sakhl Kal Belg Chuv Tver Brya Uly Bur Astr Tamb Novg Ivano Lip Smol AmurMurm Kursk Chit Mari Kost Maga N Oss Mord Orel Kurg Pskov Adyg Khak K-Ch Alt R Kalm Tyva Yevr Ingu

0 .4

.6

.8

1.0

1.2

Initial infrastructure and human capital, early 1990s

Correlation: .59 (p < .01) ); without Moscow and St Petersburg, .67 (p < .01). Source: Goskomstat, Ekspert (www.ekspert.ru). Index of development of market institutions adjusted so that “most developed” is 89, “least developed” is 1.

38

1.4

Figure 4: Investment minus savings in Russia’s regions, 1998 30

20

Sakhl

Sam

Novs S-P

Chuv Len

Kal Toms Uly Stav Nizh Kirov Omsk SarSverdMos O Khak Tyum Udm Tver Penza Krasd Bur Mari Kursk Mord Novg Lip Prim Volg Astr Vor Chel Rost Oren Ivano Perm Alt KYar Volo Amur Kost Kal Chit TambVlad Orel K-Ch Komi Irk Kem Krasy Khab Ryaz Kurg ArchPskov Maga Smol Tula Sakha Murm Yevr Brya

10

Investment minus savings 1998 (% GRP)

Mos C

Belg Bash Tat

0

-10

Adyg Kar

-20

N Oss Kalm

Alt R Dag

-30 Tyva

-40 .4

.6

.8

1.0

1.2

Initial infrastructure and human capital, early 1990s

Correlation: .45 (p < .01); without Moscow and St Petersburg, .45 (p < .01). Source: Goskomstat.

39

1.4

1.6

Figure 5: Credit issued minus savings in Russia’s regions, 1998 120 Mos C

100

80

Credit issued minus savings 1998, (% GRP)

60

40 S-P

20 Sam Novs Kal Bash Sverd Omsk Uly Tat Prim Nizh KirovChuv Stav Udm Tver Yar Krasd Mari Volg Khak Lip Mord Kursk Belg Rost Mos O Amur Len Volo Penza Perm Alt KChel Bur Vor Sar Irk Toms Khab Chit Tamb Krasy Ivano Arch Tyum Novg OrenKal Ryaz Smol KurgKost Vlad Orel Murm Astr Komi Kem K-Ch Pskov SakhaSakhlBrya Tula Yevr Adyg Kar Maga N Oss Alt R

0

-20

-40

Kalm Dag

Tyva

-60 .4

.6

.8

1.0

1.2

Initial infrastructure and human capital, early 1990s

Correlation: .78 (p < .01); without Moscow and St Petersburg: .43 (p < .01). Source: Goskomstat.

40

1.4

1.6

Figure 6: Foreign investment in Russia’s regions, 1998 5

4

Mos C

3

Tat Mos O

Log of foreign investment, 1998 (mn $ US)

Omsk

S-P

Krasd Vlad Sakha Komi Sam Novs Len Tyum BelgNizh Sakhl Oren Irk Sverd Toms VolgPrim Stav KalBash Chel Maga Novg Perm Khab Sar Kal Orel Smol Tula Arch Yar Rost Bur LipKursk Chit Mord Murm Kem VoloAstr Krasy Udm Alt K TverPenza Ryaz Kar Pskov Vor K-Ch Tyva Kost Chuv Adyg Brya Kurg

2

1

Yevr Amur Tamb Kirov Mari Uly Ivano Dag N Kalm Oss Alt R Khak

0

-1 .4

.6

.8

1.0

1.2

Initial infrastructure and human capital, early 1990s

Correlation: .60 (p < .01); without Moscow and St Petersburg, .55 (p < .01). Source: Goskomstat, Rossiiskiy Statisticheskiy Yezhegodnik 2000, p.539-40.

41

1.4

1.6

References Arteta, Carlos, Barry Eichengreen, and Charles Wyplosz. 2001. “When Does Capital Account Liberalization Help More Than It Hurts?” NBER Working Paper No. 8414, August. Barro, Robert and Xavier Sala i Martin. 1995. Economic Growth, New York: McGraw Hill. Bhattacharya, Amar, Peter J. Montiel, and Sunil Sharma, “Can Sub-Saharan Africa Attract Private Capital Inflows?” Finance & Development, June 1997, pp.3-6. Borensztein, Eduardo, José de Gregorio and Jong-Wha Lee, 1998, “How Does Foreign Direct Investment Affect Economic Growth?” Journal of International Economics, 45, pp.115-35. Boyce, James K. and Léonce Ndikumana. 2001. “Is Africa a Net Creditor? New Estimates of Capital Flight From Severely Indebted Sub-Saharan African Countries, 1970-96,” University of Massachusetts, Amherst: manuscript. Boyer, R. 1996. "The Convergence Hypothesis Revisited: Globalization but Still the Century of Nations?" in S. Berger and R. Dore, eds., National Diversity and Global Capitalism, Ithaca and London: Cornell University Press, 1996. Bucovetsky, Sam. 1991. “Asymmetric Tax Competition”, Journal of Urban Economics, 30, 167-181. Cai, Hongbin and Daniel Treisman. 2001. “State Corroding Federalism: Interjurisdictional Competition and the Weakening of Central Authority”, mimeo, UCLA. Calvo, Guillermo A., Leonardo Leiderman, and Carmen M. Reinhart. 1996. “Inflows of Capital to Developing Countries in the 1990s,” Journal of Economic Perspectives, 10, 2, Spring, pp.123-39. Collier, P. and J.W. Gunning. 1999. “Explaining African Economic Performance,” Journal of Economic Literature, 37, March, pp.64-111. Demirgüç-Kunt, A., and E. Detragiache. 1998. “Financial Liberalization and Financial Fragility,” Annual Bank Conference on Development Economics, Washington, DC: World Bank. The Economist, “A Survey of Globalisation,” September 29, 2001, p.15. Eichengreen, Barry and Michael Mussa. 1998. “Capital Account Liberalization and the IMF,” Finance and Development, 35, 4. Fernandez-Arias, E. 1996. “The New Wave of Private Capital Inflows—Push or Pull?” Journal of Development Economics, 48, 2, March, pp.389-418. Frye, Timothy and Andrei Shleifer. 1997. “The Invisible Hand and the Grabbing Hand,” American Economic Review, 87, 2, pp.354-8. Jian, Tianlun, Jeffrey D. Sachs, and Andrew M. Warner. 1996. “Trends in Regional Inequality in China,” China Economic Review, 7, 1.

42

Kanbur, Ravi and Michael Keen. 1993. “Jeux Sans Frontieres: Tax Competition and Tax Coordination When Countries Differ in Size”, American Economic Review, September, 83,4, pp.877-92. Keen, Michael and Maurice Marchand. 1997. “Fiscal Competition and the Pattern of Public Spending,” Journal of Public Economics, 66, 1, 33-53. Klein, Michael W., and Giovanni Olivei. 1999. “Capital Account Liberalization, Financial Depth and Economic Growth,” NBER: WP W7384, October. Lucas, Robert E. Jr. 1990. “Why Doesn’t Capital Flow from Rich to Poor Countries?” American Economic Review Papers and Proceedings, 80, May, pp.92-6. Mankiw, Gregory N., David Romer, and David N. Weil. 1992. “A Contribution to the Empirics of Economic Growth,” Quarterly Journal of Economics, 107, 2, pp.407-37. Mishra, Deepak, Ashoka Mody, and Antu Panini Murshid. 2001. “Private Capital Flows and Growth,” Finance and Development, June, 38, 2. Montinola, Gabriella, Yingyi Qian, and Barry R. Weingast. 1996. “Federalism, Chinese Style: The Political Basis for Economic Success,” World Politics, 48, 1, pp.50-81. Obstfeld, Maurice. 1998. “The Global Capital Market: Benefactor or Menace?” Journal of Economic Perspectives, 12, 4, Fall, pp.9-30. Qian, Yingyi and Gérard Roland. 1998. “Federalism and the Soft Budget Constraint,” American Economic Review, December, 88, 5, pp.1143-62. Rogowski, Ronald. 2001. “Does Globalization Imply Convergence? Reconsidering the Theory,” UCLA: manuscript. Romer, Paul M. 1994. “The Origins of Endogenous Growth,” Journal of Economic Perspectives, 8, 1, pp.3-22. Stiglitz, Joseph E. 2000. “Capital Market Liberalization, Economic Growth, and Instability,” World Development, 28, 6, pp.1075-1086. Tiebout, Charles. 1956. “A Pure Theory of Local Expenditures,” Journal of Political Economy, 64, October, 416-24. UNCTAD. 2000. Capital Flows and Growth in Africa, New York: United Nations Conference on Trade and Development. --------------. 2001. World Investment Report 2001, New York: United Nations Conference on Trade and Development. Wilson, John D. 1999. “Theories of Tax Competition,” National Tax Journal, 52, 2, pp.269-304. World Bank. 1999. World Development Report 1998-9: Knowledge for Development, Washington DC: World Bank. World Bank. 2001. Global Development Finance, Washington DC: World Bank.

43