Growth and Foreign Capital Enisse Kharroubi
Abstract Recent empirical work has shown that over the long run, current account de…cits are associated with lower growth, especially in developing countries. This paper shows that this can hold in an economy where (i) entrepreneurs of di¤erent productivities can raise capital from domestic and foreign …nanciers and (ii) domestic …nanciers have a comparative advantage in …nancing low productivity entrepreneurs. In this framework, low productivity entrepreneurs can outbid higher productivity entrepreneurs on the domestic capital market. When this happens, the economy attracts more capital from abroad but su¤ers capital misallocation and low total factor productivity as a large part of domestic investment goes to low productivity projects. Introducing workers into the model and allowing for endogenous savings, we show that if the labor share in value added is su¢ ciently large, aggregate savings and investment are typically lower with larger foreign capital in‡ows. Finally the paper contrasts …nancial openness and …nancial autarky highlighting that the trade-o¤ lies between an enhanced borrowing ability on the one hand and potential capital misallocation as well as reduced savings on the other hand. Keywords: …nancial integration, borrowing constraint, growth, domestic savings. JEL Classi…cation: D82, E44, F36, G15, G21, O16.
Bank for International Settlements. Address: Centralbahnplatz 2, CH-4051 Basel. e-mail: …rst name.surname(at)bis.org. The views expressed herein are those of the author and do not necessarily re‡ect the views of the BIS. Usual disclaimers apply.
1
1
Introduction.
One of the main di¤erences between open and closed economies is the ability to borrow resources from the rest of the world. Open economies can for instance …nance productive investments that could not have been undertaken on the basis of domestic savings alone. Following this approach, an open economy typically borrows from the rest of the world, i.e. runs a current account de…cit when the return to domestic investment exceeds the return foreign …nanciers ask for. Current account de…cits should therefore re‡ect high returns on domestic investment and everything else equal, high investment and high growth.1 However, both anecdotal evidence and a systematic empirical examination suggest a very di¤erent picture. For example, Gourinchas and Jeanne (2007) compare Madagascar to South Korea. Over 1980-2000, Madagascar invested on average 2.8% of its GDP while its current account balance to GDP was about -6% . To the opposite, South Korea invested on average 32% of GDP while its current account was approximately balanced over the same period. More systematic evidence provided in next section goes in the same direction: In developing countries, investment and growth have been positively not negatively associated with current account balance. This paper aims at providing a simple framework which can account for the positive relationship between current account balance on one side and investment and growth on the other.2 We consider an economy where entrepreneurs with di¤erent productivity can borrow from domestic and foreign …nanciers, subject to some borrowing constraints. Moreover, we make the core assumption that domestic …nanciers have a comparative advantage in …nancing low productivity entrepreneurs. This means that they can typically monitor low productivity entrepreneurs more e¢ ciently, or they can recover their assets following a default at a lower cost, etc.3 Under this assumption, domestic …nanciers play a certi…cation role for low productivity entrepreneurs vis-a-vis foreign …nanciers: borrowing from domestic …nanciers allows low productivity entrepreneurs 1 In a world of perfect capital mobility, the return to capital should be equalized across all countries. However, investment should still be larger in countries with larger current account de…cit. 2 Note that the positive relationship between current account balance and investment implies that current account balance and domestic savings should also be positively related. 3 This is the key assumption of the paper. Its justi…cation lies in the fact that domestic lenders are more accustomed than foreign lenders to deal with domestic entrepreneurs. Domestic lenders can therefore recover their debts more easily, i.e. at a lower cost, when an entrepreneur tries to escape its debt repayments. See Mian (2006) for an empirical investigation of this proposition. See also Guiso, Sapienza and Zingales (2004) which shows that investment is highly correlated to savings at the county level in Italy.
2
to increase their ability to raise capital from foreign …nanciers.4 As a result of this complementarity, low productivity entrepreneurs are ready to pay a premium -relative to their productivity- on capital borrowed from domestic …nanciers because this raises their ability to borrow from foreign …nanciers, and hence their pro…ts. Turning to high productivity entrepreneurs, neither domestic nor foreign …nanciers have a comparative advantage in …nancing them. Borrowing from domestic …nanciers therefore reduces the ability of a high productivity entrepreneur to raise capital from foreign …nanciers. High productivity entrepreneurs then require a discount -relative to their productivity- to borrow from domestic …nanciers since whenever they borrow more from domestic …nanciers, they need to reduce the amount of capital borrowed from foreign …nanciers and hence to forego a source of pro…ts. Two di¤erent cases can then happen. When the di¤erence in entrepreneurs’productivity is su¢ ciently large, high productivity entrepreneurs can still o¤er to domestic …nanciers a higher return than low productivity entrepreneurs despite the premium/discount stemming from borrowing constraints. They then borrow from domestic …nanciers and since their borrowing constraint features substituability, high productivity entrepreneurs only borrow a limited amount of capital from foreign …nanciers. Turning to low productivity entrepreneurs, they are unable to borrow domestically, it is too costly for them. Then given that their borrowing constraint features complementarity, low productivity entrepreneurs also raise a rather limited amount of capital from foreign …nanciers. Aggregate foreign borrowing is therefore relatively low while total factor productivity (TFP, henceforth) is rather high given that domestic …nanciers’ capital is allocated to high productivity projects. On the contrary, when the premium that low productivity entrepreneurs pay to domestic …nanciers is su¢ ciently large -or equivalently, when the discount that high productivity entrepreneurs require from domestic …nanciers is su¢ ciently large-, low productivity entrepreneurs can outbid high productivity entrepreneurs on the domestic capital market, they can o¤er a higher return to domestic …nanciers. Low productivity entrepreneurs hence borrow domestically and given that their borrowing constraint features complementarity, they can also raise a relatively large amount of capital from foreign …nanciers. Turning to high productivity entrepreneurs, they are excluded from the domestic capital market 4 See Holmström and Tirole (1997) for a micro foundation of the idea that …nancial intermediaries can play a certi…cation role of borrowers.
3
and since their borrowing constraint features substituability, they borrow a relatively large amount of capital from foreign …nanciers. Aggregate foreign borrowing is therefore relatively large while TFP is rather low given that domestic …nanciers’capital is allocated to low productivity projects. Comparing these two cases shows that a larger amount of capital borrowed from abroad is associated with a lower not higher TFP in the economy. Here foreign borrowing and TFP are negatively correlated because, at the margin, foreign borrowing is systematically allocated to low productivity projects. This framework hence provides a simple and intuitive explanation for the empirical results that have been developed in the recent literature on the role of foreign capital in the growth process (among others Aizenman et al. (2006), Gourinchas and Jeanne (2007) and Prasad, Rajan, and Subramanian (2007) point out a positive relationship between growth and current account balance). Two points are further worth noting. First, this framework does not make any point about causality: foreign borrowing and TFP are negatively associated, but ultimately this relates to entrepreneurs’borrowing constraints and the distribution of productivity among entrepreneurs. One interpretation to this negative correlation consists in linking domestic …nanciers’certi…cation role to low productivity entrepreneurs’opacity: when low productivity entrepreneurs display high opacity, domestic …nanciers’certi…cation role is larger and so is the premium low productivity entrepreneurs can pay on the domestic capital market. The two cases described above can hence be interpreted as low and high opacity cases: foreign borrowing is low -and TFP is high- when low productivity entrepreneurs display low opacity, but foreign borrowing is large -and TFP is lowwhen low productivity entrepreneurs display high opacity. Second, the key assumption in this framework can be summarized following the "opacity" interpretation, saying that productivity correlates negatively among entrepreneurs with opacity: high productivity entrepreneurs are assumed relatively transparent, hence domestic and foreign borrowing are substitutes. But low productivity entrepreneurs are assumed relatively opaque so that domestic and foreign borrowing are rather complements. As is clear, this assumption has no implications for the overall borrowing capacity. In particular, high productivity entrepreneurs can still enjoy a larger overall ability to raise capital than low productivity entrepreneurs. In other words, the key assumption in this framework is about the slope of the borrowing constraints not their level (see for instance
4
Song et al. (2011) which provides a growth model replicating the Chinese experience based on the assumption that higher productivity …rms face tighter …nancial constraints). As a …nal step the paper introduces an endogenous decision for savings to show that in the presence of complementarity in the borrowing constraint of low productivity entrepreneurs, economies with large foreign capital in‡ows to tend save less, invest less and have a lower total factor productivity than economies with low foreign capital in‡ows. We introduce workers and show that when low productivity projects command a su¢ ciently large labor share compared with high productivity project, aggregate savings and investment are everything else lower when the economy attracts more capital from abroad. The mechanism is relatively simple here: When capital borrowed from abroad is large, low productivity entrepreneurs enjoy a large output and so do workers for labor income. As a result, workers reduce their current savings to smooth consumption since they expect a relatively large future labor income. On the contrary, when the amount of capital borrowed from abroad is low, low productivity entrepreneurs’ output is also relatively low because capital goes essentially to high productivity projects. Workers’future labor income is therefore also relatively low and as a result, workers smooth consumption by raising current savings since they expect a relatively low future labor income. To put it in a nutshell, when the labor share of low productivity projects is su¢ ciently large, then aggregate savings and investment are larger when the amount of capital borrowed from abroad is lower. Yet, these results do not imply that the economy is better-o¤ under …nancial autarky. Under …nancial autarky, it is true that domestic savings go to high productivity projects because entrepreneurs o¤er a return to domestic …nanciers which meets their marginal productivity. High productivity entrepreneurs can hence always o¤er a higher return to domestic …nanciers than low productivity entrepreneurs. Under …nancial openness, domestic savings can be misallocated because low productivity entrepreneurs can possibly outbid high productivity ones. However, …nancial openness also provides entrepreneurs with a new source of …nance. Allocation problems must therefore be trade-o¤ against the ability to raise capital from abroad to assess the bene…ts and drawbacks of …nancial openness and …nancial autarky. This paper relates to two strands of literature. The …rst deals with the e¤ect of …nancial openness and
5
capital ‡ows on domestic savings and investment. In their seminal paper, Feldstein and Horioka (1980) show that among OECD countries, the correlation between investment and domestic savings is large and hence di¢ cult to reconcile with a view of capital being highly mobile. Rodrik (1998) argues that foreign savings cannot account for a large share of investment even in widely open countries. Aghion, Comin and Howitt (2006) point out that domestic savings can raise a country attractiveness for FDI. A number of papers have tried to determine the e¤ect of …nancial integration on domestic savings and investment (Obstfeld (1998), Bosworth and Collins (1999) or Razin Sadka and Yuen (1999)). Similarly, Caballerro and Krishnamurthy (2001) focuses on the e¤ects of exogenously given domestic and international borrowing constraints on real and …nancial variables. Finally Prasad, Rajan and Subramanian (2007) study the growth e¤ects of the access to foreign capital using macro and industry level data. The contribution of this paper is here to provide a mechanism to account for a possibly positive relationship between current account and growth. Moreover the paper highlights the dampening role of …nancial development in this relationship. Secondly this paper relates to the literature on the cost of capital e¤ects of …nancial liberalization. Bekaert, Harvey and Lundblad (2001), Bekaert, Harvey and Lumsdaine (2002) or Blair Henry (2003) all show that …nancial liberalization reduces signi…cantly the cost of capital for …rms, which constitutes a powerful channel through which liberalization a¤ects investment and growth. Kose, Prasad and Terrones (2003) show that …nancial integration has positive growth e¤ects but mostly in developed countries. The paper is organized as follows. A review of the recent empirical literature on the relationship between growth and current account balance is provided in section 2. Section 3 lays down the basic model and its main assumption. Section 4 derives the equilibrium of the economy and the main result of the paper. Section 5 extends the basic model to allow for endogenous savings while section 6 constrats …nancial openness and …nancial autarky. Conclusions are eventually drawn in section 7.
6
2
Stylized facts on current account balance and growth
The traditional way of thinking about the relationship between current account and growth focuses on di¤erences between the domestic and the international returns to capital: a country runs a current account de…cit when the domestic return to capital is larger than the international cost of capital. Such current account de…cits raise investment and thereby growth. Hence current account de…cits are theoretically associated with larger economic growth in as much as they re‡ect arbitrage opportunities. The di¢ culty in testing empirically this prediction consists in obtaining a proper empirical assessment of anticipated changes in the net return to capital. This problem can be bypassed assuming that expectation errors on anticipated changes in the net return to capital are uncorrelated across countries or across time. If expectation errors are uncorrelated across countries, then countries running a current account de…cit should on average bene…t from a positive arbitrage opportunity. On the contrary countries running a current account surplus should on average su¤er a negative arbitrage opportunity. Hence investment should be larger on average across countries running current account de…cits than across countries running current account surplus. Weighted Average Investment (% GDP) in capital importing countries Weighted Average Investment (% GDP) in capital exporting countries 29
27
25
23
21
19
17
15 1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
Source: WDI Indicators and author's calculations.
Fig. 1: Average investment rate in capital importing and capital exporting countries. Empirical evidence goes however in the opposite direction. Average investment has always been larger in
7
countries with a current account surplus than in countries running a current account de…cit.5 Over a twenty years period (1984-2003), the average di¤erence in the investment to GDP ratio has been around 4 pp in favor of current account surplus countries (3 pp over the very last years). Therefore if investment -as a share of GDP- increases with the return to capital, it turns out that economies have been running a current account de…cit when the return on capital was low. This conclusion is incompatible with the view that associates current account de…cits with a high return on domestic capital. Now if we consider instead that expectation errors on the net return to capital are uncorrelated across time, then the correlation between long term average current account balance and long term investment (both as a share of GDP) should be negative across countries. Considering a su¢ ciently long time period, expectation errors on the net return to capital should cancel out and countries with a high return should be net borrowers with high investment while countries with a low return should be net …nanciers with low investment. 40
1980-2004 Average Investment to GDP (% GDP)
Source: WDI Indicators and author's calculations.
35
30
25
20
15
10
5 -8
-6
-4
-2
0
2
4
6
1980-2004 Average Current Account Balance (%GDP)
Fig. 2: Average investment to GDP and average current account balance to GDP. However as previously the data does not con…rm this prediction. Indeed, the cross country correlation between average current account balance and investment is, if anything, positive, thereby validating the positive association between current account balance and investment. Moreover Gourinchas and Jeanne 5 See
appendix for the list of countries in the sample of which computations are carried out. Average investment to GDP is weighted by the relative contribution of each country to the category it falls in.
8
(2007) con…rm this result and extend it to the relationship between long run growth and long run current account balance showing that they are positively and not negatively associated across a large pool of developing countries. The relationship between current account balance and growth is indeed analyzed in more details in Prasad, Rajan and Subramanian (2007). In particular they provide evidence on a large sample of developing countries that growth both at the macro level and at the industry level has been slower in countries which have been running larger current account de…cit. Another way to look at this question consists in comparing the average current account balance of countries with relatively high growth to that of countries with relatively low growth. To do so, based on the same sample as previously, countries are divided between those with above median GDP per capita growth and those with below median GDP per capita growth, the median being computed for the period under consideration. Average Current Account Deficit (%GDP) Below median growth countries
Above median growth countries
5
4
3
2
1
0 1980-1984
1985-1989
1990-1994
1995-1999
1980-2003
Source: WDI Indicators and author's calculations.
Fig. 3: Current account de…cits and GDP per capita growth The current account balance has always been systematically larger in countries with high growth compared to countries with low growth. Interestingly, the di¤erence in average current account de…cit between countries with above median GDP per capita growth seems to have increased over time. Over 1980-1984, it was about 0.4pp of GDP while over 1995-1999, it was more than 1pp of GDP, thereby indicating an increased
9
polarization of average current account positions across “high growth”countries and “low growth countries”. If the traditional approach to current account balance and growth cannot account for the stylized facts raised above, that begs the question of how to account for these empirical regularities. The remainder of the paper is dedicated to provide a simple framework in which capital in‡ows and growth are not only driven by di¤erences in the domestic and the international return to capital. In particular heterogeneity of domestic agents in their productivity and their credit constraints will be shown to be key to understand how capital in‡ows can be negatively related to growth.
3
The model
3.1
Main assumptions
Let us begin with the simplest version. We consider an economy with a single good and a unit mass of agents, each living for one period. Agents are either entrepreneurs or …nanciers. There are two types of …nanciers, domestic and foreign …nanciers, and two types of entrepreneurs, high and low productivity entrepreneurs. There is an equal number of …nanciers and entrepreneurs as well as an equal number of each type of …nanciers and entrepreneurs. Domestic …nanciers are endowed with some initial capital w at the beginning of each period t which they can lend at the domestic opportunity cost of capital denoted rd . Foreign …nanciers have deep pockets and can supply any amount capital at the beginning of each period t at the international opportunity cost of capital rf . Entrepreneurs are also endowed with some initial capital w. They also have an investment opportunity at the beginning of each period t. A high (low) productivity entrepreneur reaps Rh (Rl ) in period t + 1 after investing one in period t, with Rh > Rl > rf . Entrepreneurs can raise capital from domestic and foreign …nanciers but face borrowing limits. We assume that the amount of capital Lhf a high productivity entrepreneur can borrow from foreign …nanciers decreases with the amount of capital Ld it borrows from domestic …nanciers. On the contrary, the amount of capital Llf a low productivity entrepreneur can borrow
10
from foreign …nanciers increases with the amount of capital Ld it borrows from domestic …nanciers. To simplify, we will assume that borrowing limits are linear, the parameters ( l ;
l)
and (
h;
h)
being all
positive.6 low productivity entrepreneurs: Llf
lw
Lhf
hw
high productivity entrepreneurs:
+
l Ld
(1)
h Ld
Two reasons motivate the complementarity embedded in the borrowing constraint for low productivity entrepreneurs. First domestic …nanciers can monitor relatively easily low productivity projects as they are presumably rather unsophisticated and rely on simple technologies. Of course, this also holds for foreign …nanciers. However, in developing countries, low productivity projects are usually domestic projects while high productivity projects being usually associated with foreign direct investment-. This means that domestic …nanciers have an advantage over foreign …nanciers because they should be used to contract with low productivity entrepreneurs. Domestic …nanciers can therefore play a certi…cation role of low productivity entrepreneurs vis-a-vis foreign …nanciers because of the unsophisticated and easy-to-monitor technology as well as the habit to contract with such entrepreneurs.
Fig. 4: Entrepreneurs’borrowing constraints. 6A
micro foundation for these borrowing limits is detailed in appendix 1.
11
By contrast, foreign …nanciers are probably less used to contract with low productivity entrepreneurs which explains why they may have more di¢ culties to monitor them, in spite of their relatively simple and unsophisticated technology. Turning to high productivity entrepreneurs, their borrowing constraint features substituability. Here the argument is that high productivity projects imply rather sophisticated technologies which are complicated to monitor for all …nanciers, including foreign …nanciers. Sophistication therefore make domestic and foreign …nanciers symmetric, so that no investor has a comparative advantage in dealing with high productivity entrepreneurs. Finally we consider a last assumption in this framework which contributes to make the model more realistic in talking about developing economies: the economy is assumed to be short of capital in the absence of foreign …nanciers. Put di¤erently entrepreneurs’ borrowing limits w.r.t. foreign …nanciers do not bind when domestic …nanciers’lending capacity is exhausted. This is trivial for low productivity entrepreneurs who do not face any borrowing limit w.r.t. domestic …nanciers and can therefore borrow any amount of capital on the domestic capital market. For high productivity entrepreneurs, this assumption means that the borrowing constraint Lhf
3.2
hw
h Ld
does not bind for Lh = w, i.e.
h
>
h.
Entrepreneurs
Consider a high productivity entrepreneur who maximizes pro…ts. It needs to choose the amounts of capital (Ld ; Lf ) to raise respectively from domestic and foreign …nanciers, which solve
max (w + Ld + Lf ) Rh
Ld ;Lf
rd Ld
rf Lf (2)
s.t. Lf
hw
h Ld
High productivity entrepreneurs face a simple cost vs. quantity trade-o¤. For example, it may be less costly to borrow from foreign …nanciers, but borrowing from domestic …nanciers may allow to raise more capital. The solution is therefore as follows: Denoting rh = (1 from foreign …nanciers Lhf = rd
hw
if rd
h ) Rh +
h rf ,
a high productivity entrepreneur borrows
rh , while it borrows from domestic …nanciers Lhd = (
h= h) w
if
rh . High productivity entrepreneurs therefore prefer to borrow from domestic (foreign) …nanciers when
12
the cost rd to do so is su¢ ciently low (large). An important property for high productivity entrepreneurs’ optimal borrowing is that the threshold rh is lower than the return to capital Rh since the parameter
h
is positive. The presence of foreign …nanciers
therefore reduces below the return to capital Rh the maximum cost for domestic capital rh high productivity entrepreneurs can sustain. This is due to the substituability between domestic and foreign borrowing in high productivity entrepreneurs’ credit constraint. As a result, domestic and foreign borrowing should at the margin be equally pro…table. Given that foreign borrowing is always pro…table, the maximum cost for domestic capital rh that entrepreneurs can accept, should necessarily be strictly lower than the return to capital Rh so that domestic borrowing is also pro…table. Turning now to low productivity entrepreneurs, their problem consists in choosing the amounts of capital (Ld ; Lf ) to raise respectively from domestic and foreign …nanciers which solve
max (w + Ld + Lf ) Rl
Ld ;Lf
rd Ld
rf Lf (3)
s.t. Lf
lw
+
l Ld
Low productivity entrepreneurs face a di¤erent trade-o¤ than high productivity entrepreneurs. When a low productivity entrepreneur borrows from domestic …nanciers, this actually raises its ability to borrow from foreign …nanciers. Hence, given that borrowing from foreign …nanciers is always pro…table, -Rl > rf -, low productivity entrepreneurs choose to borrow from foreign …nanciers the maximum amount compatible with the borrowing limit: Llf =
lw
+
l l Ld .
But they borrow from domestic …nanciers only if the cost rd to do
so is not too large. More precisely, they e¤ectively borrow from domestic …nanciers if and only if rd with rl = (1 +
l ) Rl
l rf .
rl
Since they do not face any constraint on domestic borrowing, their demand for
domestic capital is simply as large as possible if rd
rl .
An important property of low productivity entrepreneurs’optimal borrowing is that the threshold rl is larger than the return to capital Rl since the parameter
l
has been assumed to be positive. Low productivity
entrepreneurs therefore borrow from domestic …nanciers even if the opportunity cost of domestic capital rd is above their return to capital Rl . The presence of foreign …nanciers actually leads low productivity
13
entrepreneurs to raise the threshold opportunity cost of capital rl beyond their productivity Rl . This relates to the complementarity between domestic and foreign borrowing in low productivity entrepreneurs’ credit constraints. When low productivity entrepreneurs borrow more from domestic …nanciers, they can raise more capital from foreign …nanciers which raises pro…ts since borrowing from foreign …nanciers is pro…table. Low productivity entrepreneurs therefore accept to make losses on domestic borrowing in as much as this is compensated by larger foreign borrowing and hence larger pro…ts.
4
The equilibrium
In a closed economy where there would be no foreign …nanciers, high productivity entrepreneurs would pay a gross interest rate Rh while low productivity entrepreneurs would pay a gross interest rate Rl to borrow from domestic lenders. Given that Rh > Rl , high productivity entrepreneurs would always outbid low productivity entrepreneurs and domestic …nanciers would always …nance high productivity projects. This logic however disappears in the current framework with foreign …nanciers. Since high productivity entrepreneurs can substitute borrowing from domestic …nanciers with borrowing from foreign …nanciers, the gross interest rate they are willing to pay for domestic borrowing is reduced compared with the case of …nancial autarky. By contrast, for low productivity entrepreneurs, borrowing from domestic …nanciers complements borrowing from foreign …nanciers. As a result, the gross interest rate low productivity entrepreneurs are willing to pay for domestic borrowing goes up compared to …nancial autarky because doing so, they can raise more capital from domestic …nanciers and thereby raise more capital from foreign …nanciers. Hence whether high productivity entrepreneurs can still outbid low productivity entrepreneurs on the domestic capital market depends on how large are di¤erences in productivity relative to di¤erences in borrowing constraints. When productivity di¤erences dominate, then high productivity entrepreneurs can still outbid low productivity entrepreneurs but if di¤erences in borrowing constraints dominate -e.g. complementarity between domestic and foreign borrowing for low productivity entrepreneurs is su¢ ciently strongthen low productivity entrepreneurs will be able to outbid high productivity entrepreneurs. The next section draws the implications of these two di¤erent situations in terms of aggregate growth and aggregate foreign 14
borrowing.
4.1
The case where productivity di¤erences dominate
Consider …rst the case where high productivity entrepreneurs can outbid their low productivity counterparts. This means that the threshold above which high productivity entrepreneurs stop borrowing from domestic …nanciers is larger than the threshold for low productivity entrepreneurs, i.e. rh > rl .7 The opportunity cost of domestic capital rd is hence equal to rh and only high productivity entrepreneurs borrow from domestic …nanciers. Given that there is an equal number of high productivity entrepreneurs and domestic …nanciers, each high productivity entrepreneur can borrow w from domestic …nanciers and given that high productivity entrepreneurs can also borrow Fh =
hw
hw
hw
>
h w,
from foreign …nanciers. Low productivity
entrepreneurs do not borrow from domestic lenders because this is too costly for them. Yet, they can still borrow from foreign …nanciers Fl =
l w.
We can hence compute the economy’s total borrowing from foreign
…nanciers Fp which also represents its current account de…cit, as the sum of high and low productivity entrepreneurs borrowing from foreign …nanciers:
Fp = Fh + Fl =
hw
hw
+
lw
(4)
Turning now to investment, high productivity entrepreneurs invest a total amount Ih = w + w + Fh which is the sum of high productivity entrepreneurs’ capital w, high productivity entrepreneurs’ borrowing from domestic …nanciers’s w and high productivity entrepreneurs’ borrowing from foreign …nanciers Fh . Low productivity entrepreneurs total investment which writes as Il = w + Fl is simply the sum of low productivity entrepreneurs’capital w and low productivity entrepreneurs’borrowing from foreign …nanciers Fl .The economy’s average total factor productivity is R =
Rp =
7 Note
this case requires
h
Ih Rh +Il Rl , Ih +Il
which simpli…es as
w + w + Fh R h + w + Fl R l w + w + Fh + w + Fl
< 1 as a necessary condition to hold given that we have assumed that Rl > rf .
15
(5)
4.2
The case where borrowing constraint di¤erences dominate
Consider now the case where low productivity entrepreneurs can outbid high productivity entrepreneurs on the domestic capital market. This means that the cost of domestic capital, low productivity entrepreneurs can sustain is larger than the cost of domestic capital that high productivity entrepreneurs can sustain, i.e. rl > rh . The opportunity cost of domestic capital is hence equal to rl and only low productivity entrepreneurs borrow from domestic …nanciers. Low productivity entrepreneurs and domestic …nanciers being in equal number, each low productivity entrepreneur can borrow w from domestic …nanciers. Moreover, low productivity entrepreneurs can borrow Fl =
lw
+
lw
from foreign …nanciers. High productivity
entrepreneurs do not borrow from domestic lenders here because this is too costly for them. Yet, they can still borrow from foreign …nanciers an amount Fh =
h w.
The economy’s total borrowing from foreign
…nanciers which represents its current account de…cit is equal in this case to
Fb = Fh + Fl =
hw
+
lw
+
lw
(6)
Investment from low productivity entrepreneurs is therefore Il = w + w + Fl . This is the sum of low productivity entrepreneurs’capital w, low productivity entrepreneurs’borrowing from domestic …nanciers’s w and low productivity entrepreneurs’borrowing from foreign …nanciers Fl . High productivity entrepreneurs’ total investment writes as Ih = w + Fh . It is the sum of high productivity entrepreneurs’ capital w, and high productivity entrepreneurs’borrowing from foreign …nanciers Fh . The economy’s average total factor productivity is R =
Ih Rh +Il Rl , Ih +Il
which simpli…es as
Rb =
w + Fh R h + w + w + Fl R l w + Fh + w + w + Fl
16
(7)
4.3
Total factor productivity and foreign capital.
Comparing these two cases yields a simple conclusion: Average TFP is larger when productivity di¤erence dominate, but capital in‡ows are larger when borrowing constraint di¤erences dominate
Rp > Rb and Fp < Fb
The …rst case where high productivity entrepreneurs can outbid their low productivity counterparts tends to exhibit both high TFP and low foreign borrowing while the case where low productivity entrepreneurs can outbid their high productivity counterparts exhibits low TFP but high foreign borrowing. A negative correlation therefore emerges between average TFP and foreign borrowing. The reason is that the economy as a whole borrows more capital from abroad when low productivity entrepreneurs are able to outbid high productivity entrepreneurs from the domestic capital market. The complementarity between domestic and foreign borrowing then allows low productivity entrepreneurs to borrow more from foreign …nanciers and to increase investment at the cost of a lower TFP given that a larger share of total investment goes to low productivity projects. In a nutshell, the negative correlation between foreign borrowing and average total factor productivity growth emerges as a direct implication of the assumption that domestic and foreign capital are more likely to be complements for lower productivity projects and more likely to be substitutes for higher productivity projects.
5
Endogenous savings and investment
In the previous section, we have shown that higher foreign borrowing is associated with lower average total factor productivity. The empirical evidence presented above however shows that savings and investment are negatively associated with foreign borrowing while in the previous framework, savings are exogenous and investment correlates positively with foreign borrowing.8 We therefore extend the model developed above, 8 Investment is simply the sum of savings and foreign borrowing. When savings are exogenous, investment and foreign borrowing are similar up to a constant.
17
to allow for endogenous savings and investment to determine how they correlate with foreign borrowing. To do so, we enrich the current framework as follows.
5.1
Timing
At the beginning of period t, agents start with some capital endowment. They learn whether they will be high productivity entrepreneurs, low productivity entrepreneurs or domestic …nanciers. They then take a consumption/saving decision. Entrepreneurs then borrow from domestic and/or foreign …nanciers and invest. In period t + 1, entrepreneurs reap their output, pay back domestic and foreign …nanciers they have borrowed from in period t, make a bequest to the next generation of agents and disappear. Agents in the next generation then start period t + 1 with the last generation’s bequest. They learn whether they will be high productivity entrepreneurs, low productivity entrepreneurs or domestic …nanciers. Agents therefore begin each period with the same endowment.
5.2
Preferences
Agents have separable log preferences over period t consumption and period t + 1 bequest. Agents begin with an endowment et in period t, and take a consumption decision ct and a savings decision st . Agents then reap an income
t+1
in period t + 1 which depends on the amount of savings st in period t and they make a
bequest bt+1 to the next generation out of this income it+1 . The period t + 1 generation’s initial endowment is therefore the period t generation bequest et+1 = bt+1 . The consumption-savings decision (ct ; st ) typically di¤ers across agents since agents know their type and the period t + 1 income it+1 when they take this decision. Denoting
the discount factor, an agent’s problem in period t therefore writes as
max U = log (ct ) +
ct ;bt+1
log (bt+1 ) (8)
s.t. ct + st = et and bt+1 =
18
t+1
(st )
5.3
Technology
We extend the model assuming that entrepreneurs’ technology uses capital and labor as inputs. A type i entrepreneur (i = (h; l)) who invest an amount K of capital and hires L workers reaps an output
Yi = Bi K i L1
(9)
i
Capital is provided by the entrepreneur managing the project, this may be its own funds or its borrowing from domestic and/or foreign …nanciers. Labor is provided by domestic …nanciers who play the dual role of providing domestic capital and labor to entrepreneurs. Domestic …nanciers are therefore like workers who have a labor supply but no physical investment opportunity. They hence need to lend their capital to entrepreneurs since capital depreciates totally in one period. Importantly we assume that the low productivity technology is more labor intensive and the high productivity technology is more capital intensive, i.e. h
>
i.e.
l.
h
To …x ideas, we assume that high productivity entrepreneurs hence do not need to hire workers,
= 1 and
l
< 1. Finally to simplify exposition and further computations, we assume entrepreneurs’
technology is AK, i.e. total factor productivity Bi veri…es Bi = Ai K=L
1
i
, K (L) being the aggregate
amount of capital (labor) type i entrepreneurs invest (hire). Finally the assumption Ah > Al ensures that high productivity entrepreneurs e¤ectively have access to a more productive technology.
5.4
The optimal consumption/saving decision
Markets being competitive, type i entrepreneurs all feature the same capital to labor ratio, i.e. K=L = K=L. The entrepreneurs’cash ‡ow is therefore simply proportional to the capital K invested
Ri K =
i Ai K
(10)
A type i entrepreneur who save si , borrows Ld from domestic …nanciers and Lf foreign …nanciers in period t therefore reaps in period t + 1 an income
t+1
(si ) = (si + Ld + Lf ) Ri
19
rd Ld
rf Lf . Since the borrowing
limit binds at the optimum, this expression can be simpli…ed as
t+1
(si ) = [(1 +
i ) Ri
i rf ] si
+ [ri
rd ] Ld
(11)
If type i entrepreneurs borrow from domestic …nanciers, then rd = ri while if type i entrepreneurs do not borrow from domestic …nanciers, then Ld = 0. Hence irrespective of the equilibrium of the domestic capital market and the type of entrepreneurs borrowing from domestic …nanciers, a type i entrepreneur reaps a period t + 1 income which veri…es
t+1
Entrepreneurs’income in period t + 1,
(si ) = [(1 +
t+1
i ) Ri
i rf ] si
(12)
(si ) is therefore directly proportional to their own savings si in
period t. This implies that high and low productivity entrepreneurs all take the same savings decision si which consists in savings a fraction of their initial endowment et : si =
1+
et . Hence, denoting Et agents’
aggregate initial capital endowment in period t, type i entrepreneurs’aggregate savings is
Si = S =
1+
Et
(13)
so that the dynamics of entrepreneurs’aggregate initial endowment writes as
Et+1 = [(1 +
i ) Ri
i rf ]
1+
Et
(14)
Let us now turn to the case of a domestic …nanciers/worker. This agent saves and lends sd to entrepreneurs in period t . It also works for entrepreneurs. Domestic …nanciers/workers therefore reap in period t + 1 a capital income rd sd and a labor income denoted w for now. A domestic …nancier/worker hence reaps a total income sd =
t+1
et wt+1 =rd . 1+
= rd sd + wt+1 in period t + 1. Domestic …nanciers/workers savings decision is then
Denoting Wt+1 the aggregate labor income in period t + 1, Domestic …nanciers/workers’
20
aggregate savings then writes as Sd =
Et
Wt+1 =rd 1+
(15)
Let us now determine the aggregate labor income Wt+1 . The wage bill a type i entrepreneur distributes to its workers is simply the di¤erence between total output Yi and the entrepreneur’s cash ‡ow Ri K. For a high productivity entrepreneur, this is zero because high productivity entrepreneurs do not use capital. By contrast, low productivity entrepreneurs do use labor as in input and the wage bill of a low productivity entrepreneur writes as Yl
Rl K = (1
l ) Al K,
K being the aggregate capital stock invested in low
productivity projects which is simply the sum of low productivity entrepreneurs’ savings and borrowing. Low productivity entrepreneurs’ savings is …xed and given by (13) but low productivity entrepreneurs’ borrowing depends on the equilibrium of the domestic capital market. Denoting ! l = (1
l ) Al ,
domestic
…nanciers/workers aggregate labor income hence writes as:
Wt+1 = ! l
8 > >
> : (1 +
l) S
l) S
+ (1 +
if rh l ) Sd
rl if rh
(16) rl
This expression shows that the aggregate labor income depends on domestic …nanciers/workers aggregate savings Sd . This is because more savings raise the capital stock invested which increases the wage bill distributed. This however holds only in the case where domestic capital is invested in low productivity projects because high productivity projects do not involve any labor. Using (15) and (16), the expression for domestic …nanciers/workers savings Sd can therefore be simpli…ed as Sd = (rd ) S with
(rd ) =
(1 + ) rd (1 + l ) ! l (1 + ) rd + (1 + l ) ! l 1[rl >rh ]
(17)
Expression (17) shows that domestic …nanciers/workers save less than entrepreneurs. The reason is that domestic …nanciers consume in period t part of the labor income which they reap only in period t + 1. In the absence of a labor income in period t + 1, i.e. ! l = 0, domestic …nanciers/workers would save exactly as entrepreneurs, i.e. a fraction
1+
of their initial endowment Et . The dynamics of domestic
21
…nanciers/workers’aggregate initial endowment writes as
Et+1 = [(rd + (1 +
5.5
l ) !l )
(rd ) + ! l (1 +
l )]
1+
Et
(18)
Capital ‡ows and investment
Now that we have determined agents consumption and savings decision, we can turn to determining capital ‡ows, investment and growth. When high productivity entrepreneurs are able to borrow domestically, i.e. rh > rl , aggregate foreign borrowing Fp is the sum of high and low productivity entrepreneurs’borrowing from foreign …nanciers, while aggregate investment Ip is the sum of high productivity entrepreneurs’savings and domestic and foreign borrowing:
Fp = Ip =
Then, denoting
i
= (1 +
i ) Ai
1+
i rf ,
[
1+
[1 +
(rh )
h
l+1+
+
h
h + (1
and ah = (1
l ] Et
(19)
h ) (rh )] Et
h ) Ah +
h rf ,
aggregate growth in the initial capital
endowment is gp =
Et+1 = Et 1+
[
+
h
l
+ ah (rh )]
(20)
Similarly, when low productivity entrepreneurs are able to borrow domestically, aggregate foreign borrowing Fb and aggregate investment Ib is
Fb = Ib =
Finally, denoting al = (1 +
l ) Al
1+
[1 +
l rf ,
gb =
[
1+ l
h
+
l
+1+
+ (rl )
h
+ (1 +
l ] Et
(21)
l ) (rl )] Et
aggregate growth in the initial capital endowment write as
Et+1 = Et 1+
[
22
h
+
l
+ al (rl )]
(22)
From these two expressions, it is straightforward to note that foreign borrowing is lower when high productivity entrepreneurs are borrow domestically, i.e. Fp < Fb . The mechanism is as previously, when high productivity entrepreneurs borrow from domestic …nanciers, the substituability between domestic and foreign borrowing reduces borrowing from abroad. On the contrary, when low productivity entrepreneurs borrow domestically, the complementarity between domestic and foreign borrowing actually raises borrowing from abroad. Let us now turn to aggregate investment. Given that aggregate foreign borrowing is lower when high productivity entrepreneurs borrow from domestic …nanciers, aggregate investment correlates negatively with foreign borrowing if and only if (1 +
l)
(rl ) < (1
h)
(rh )
(23)
We can then derive the following result. Proposition 1 There exists an lower bound ! such that when the labor share ! l of low productivity projects veri…es ! l
!, then aggregate foreign borrowing correlates negatively with aggregate investment and growth.
Proof. The condition
(rl ) (1 + "
1
l)
< (rh ) (1
rl 1+
l
+
!l 1+
rh 1
h
#
h)
1+ 1
can be simpli…ed as
l h
"
rl 1
!l < 1
rl 1+
l
h
+
!l 1+
#
(24)
The left hand side of this inequality is always negative since the assumption Ah > Al directly implies that rh 1
h
>
rl 1+
l
+
!l 1+
. Now the right hand side of (24) can be either positive or negative. When it is positive,
i.e. when the labor share of low productivity projects ! l veri…es
!l 1+
>
1 1
h
23
1 1+
rl l
(25)
then (24) always holds. On the contrary when the right hand side is negative then condition (24) can simplify as 1+ 1
rl l
!l >
h
1
h
rl 1+
h
rl 1+
rh 1
l
l
+
!l 1+
+
!l 1+
rh 1 rl 1+
l
h
+
(26)
!l 1+
This condition holds if and only if the labor share ! l of low productivity projects is su¢ ciently large since its left hand side is increasing in ! l while it right hand side is decreasing in ! l . Combining (25) and (26) shows that provided the labor share ! l of low productivity projects is above some threshold !, aggregate investment is larger when aggregate foreign borrowing is lower, i.e. when high productivity entrepreneurs borrow from domestic …nanciers. Finally that the condition under which investment and foreign borrowing correlate negatively is a su¢ cient condition to ensure that growth and foreign borrowing also correlate negatively. Indeed the expressions (19) and (21) show that growth is higher when foreign borrowing is lower if and only if
((1 +
l ) Al
+
l rf )
(rl ) < ((1
h ) Ah
+
h rf )
(rh )
(27)
which always holds when investment and foreign borrowing correlate negatively, i.e. when condition (23) holds. The intuition for this result is relatively straightforward: when the labor share of the low productivity technology ! l is su¢ ciently large, this reduces domestic …nanciers/workers’savings and the more so, when domestic capital goes to low productivity entrepreneurs. When domestic capital is invested in low productivity projects, domestic …nanciers/workers enjoy a larger labor income and hence raise current consumption in anticipation of a larger future labor income. Hence the larger the labor share of low productivity projects the larger future income and hence the larger the reduction in current savings.
24
5.6
Contrasting …nancial openness and autarky
In the previous paragraph, we have compared two open economies with di¤erent characteristics which has lead to the conclusion that higher foreign borrowing is associated with lower investment and lower growth. Yet, does this result imply that reforms promoting …nancial openness are growth reducing? To answer this question, let us …rst consider the case of an economy under …nancial autarky. In this economy, type i entrepreneurs savings Si and domestic …nanciers/workers’savings respectively write as
Si =
1+
Wt and Sd =
Wt with
1+
=
(1 + ) Ah ! l (1 + ) Ah
(28)
Indeed in the closed economy, high productivity entrepreneurs can borrow on the domestic capital market since they can o¤er a return equal to Rh which is larger than the return Rl low productivity entrepreneurs can pay to domestic …nanciers. Then, aggregate growth in the initial capital endowment is
g=
Et+1 = Et 1+
[(1 + ) Ah + Al ]
(29)
while average total factor productivity A simply writes as
A=
(1 + ) Ah + Al 2+
(30)
Turning now to the case of an open economy, we need to look at two di¤erent cases.
5.6.1
The case where productivity di¤erences dominate.
When productivity di¤erences dominate, high productivity entrepreneurs borrow domestically. Then the domestic cost of capital is equal to rh and entrepreneurs’savings S and domestic …nanciers/workers’savings Sd respectively write as
S =
1+
Et and Sd =
h
1+
Et with
25
h
=
(1 + ) rh (1 + (1 + ) rh
l ) !l
(31)
Two remarks can be made here. First entrepreneurs’savings are unchanged because agents’preferences are such savings are independent of the return on capital. Second, domestic …nanciers/workers’savings go down compared to the case of the closed economy,
h
< . This is because the interest rate on the domestic capital
market goes down from Rh to rh and also because workers’future labor income -per unit of entrepreneurs’ own funds- increases from ! l to (1 +
l ) !l
which provides workers with incentives to anticipate on future
income by raising current consumption and reducing current savings. Then, denoting and al = (1 +
l ) Al
l rf ,
i
= (1 +
i ) Ai
i rf ,
aggregate growth in the initial capital endowment gf is
gf =
Et+1 = Et 1+
[
h
+
l
+ ah
h]
(32)
while average total factor productivity Af is as previously a weighted average of high and low productivity entrepreneurs’TFP: Af =
(1 +
+ 2+ h
h h ) Ah
h h
+(
h
h
+ (1 + l ) Al h + l)
(33)
When high productivity entrepreneurs borrow domestically, …nancial openness has the aforementioned negative e¤ect on savings exempli…ed in the drop from
to
h.
To the extent that domestic …nanciers/workers’
savings are invested in high productivity projects, the reduction in domestic savings translates into a reduction in average TFP and growth. This is the negative e¤ect of …nancial openness. However, …nancial openness also has a positive e¤ect on average TFP and growth since it allows high and low productivity entrepreneurs to borrow from abroad. Growth under …nancial openness is always larger because this last positive e¤ect always dominates the negative e¤ect related to the drop in domestic savings:
gf > g , (Rh
rf ) (
h
h)
+ Rl
rf +
!l 1+
l
>0
(34)
Turning to average TFP, things are some di¤erent: Average TFP is larger under …nancial openness if and only if high productivity entrepreneurs’borrowing from abroad compensates for the drop in domestic investor’s
26
savings and low productivity entrepreneurs’borrowing from abroad:
h
Af > A ,
5.6.2
h h
1+
+
1+ 1+
h
>1+
(35)
l
The case where borrowing constraint di¤erences dominate.
Let us now turn to the case where borrowing constraint di¤erences dominate. Low productivity entrepreneurs then borrow domestically. entrepreneurs’savings S and domestic …nanciers/workers’savings Sd respectively write as S =
1+
Wt and Sd =
l
1+
Wt with
l
=
(1 + ) rl (1 + (1 + ) rl + (1 +
l ) !l l ) !l
(36)
As in the previous case, entrepreneurs’savings are unchanged compared with the case of …nancial autarky. However in contrast with the previous case, domestic …nanciers/workers’savings faces opposite forces. On the one hand, entrepreneurs’foreign borrowing raises future labor income and hence reduces current savings. Moreover when domestic capital goes to low productivity entrepreneurs, the presence of a complementarity between domestic and foreign borrowing in low productivity entrepreneurs borrowing constraint, further contributes to reduce domestic …nanciers’ savings. On the other hand, the interest rate on the domestic capital market moves from Rh to rl and it is possible that …nancial openness actually contributes to raise the cost of domestic capital if complementarity
l
between domestic and foreign borrowing is su¢ ciently large
degree. It can however be shown that this second e¤ect is always dominated, i.e. domestic …nanciers/workers’ savings go down when the economy is …nancially opened:
l
< . Hence no matter whether domestic capital
actually goes to high or low productivity entrepreneurs, …nancial openness reduces domestic savings compared to …nancial autarky. Then, denoting
i
= (1 +
i ) Ai
i rf ,
and al = (1 +
l ) Al
l rf ,
aggregate growth
in the initial capital endowment is
gf =
Et+1 = Et 1+
[
27
h
+
l
+ al (rl )]
(37)
Turning to average total factor productivity gl , it writes as
Af =
(1 +
h ) Ah
+ (1 + l + l + 1+ h+1+ l+ l+
l l ) Al
(38)
l l
Total factor productivity and growth under …nancial openness di¤er from the case of the …nancial autarky for three reasons. First foreign borrowing allows high and low productivity entrepreneurs to increase investment. Second domestic savings which are invested in high productivity projects under …nancial autarky are now allocated to low productivity projects, which contributes to reduce total factor productivity. Last, domestic savings decrease under …nancial openness which has a negative impact on aggregate growth but a positive one on average TFP given that domestic savings are now invested in low productivity projects. Similar to the previous case, it can be shown that these …rst positive e¤ect related to the ability to borrow from abroad under …nancial openness dominates the two other e¤ects so that aggregate growth is always larger under …nancial openness compared with …nancial autarky:
gf > g , (Ah
rf )
h
"
+ Rl
rl !l rf + (1+ l )! l 1 + rl + 1+
#
l
>
!l ! l (1 + l ) rl 1+ 1+ rl + (1+1+l )!l
However turning to average TFP, things are di¤erent since …nancial openness raises average TFP if and only if high productivity entrepreneurs’ borrowing from abroad compensates for the misallocation of domestic capital in low productivity projects as well as low productivity entrepreneurs’borrowing from abroad:
Af > A ,
1+ 1+
h
>1+
l
+
l
+
l l
(39)
High productivity entrepreneurs’ borrowing from abroad hence needs to compensate for low productivity entrepreneurs’borrowing from abroad as well as for domestic savings whose allocation moves from high to low productivity projects.
28
6
Conclusion
The main idea of this paper consists in showing that with imperfect capital markets, foreign capital in‡ows can be negatively associated with investment and growth. When domestic and foreign sources of …nance are complements for low productivity entrepreneurs but substitutes for high productivity entrepreneurs, then foreign capital in‡ows move in opposite direction to total factor productivity. Moreover, when labor income from low productivity projects is su¢ ciently large, then foreign capital in‡ows also move in opposite direction with investment and savings. This imperfect capital market framework can hence provide a intuitive accounting of recent empirical evidence on the negative relationship between current account de…cits on one hand and savings, investment and growth on the other hand. Recent trends of uphill international capital ‡ows can therefore be rationalized on the grounds that low income countries -which su¤er from low …nancial development- have incentives from a growth point of view to limit current account de…cits while high income countries -which bene…t from high …nancial developmenthave no incentives from a growth point of view to restrict their current account de…cits. A general equilibrium model with two economies and an endogenous international cost of capital is however needed to properly determine the growth implications of capital in‡ows.
References [1] Aghion, Ph., Comin, D., Howitt, P., 2006. When Does Domestic Savings Matter for Economic Growth? NBER Working Paper 12275. [2] Aizenman, J., Pinto, B., Radziwill, A., 2004. Sources of …nancing domestic capital -Is foreign savings a viable option? forthcoming Journal of International Money and Finance. [3] Bacchetta, Ph., 1992. Liberalization of Capital Movements and of the Domestic Financial Sytem. Economica 59, 465-74.
29
[4] Bekaert, G., Harvey C.R., Lundblad, C., 2001. Does Financial Liberalization Spur Growth? NBER Working Paper 8245. [5] Borensztein E., De Gregorio J., Lee, J.-W., 1998. How Does Foreign Direct Investment A¤ect Growth? Journal of International Economics 45, 115-35. [6] Bosworth, B., Collins, S., 1999. Capital Flows to Developing Economies: Implications for Saving and Investment. Brookings Papers on Economic Activity 30,143-80. [7] Broner, F. J. Ventura, 2006. Globalization and Risk Sharing. CEPR Discussion Papers 5820. [8] Caballero R. J. A. Krishnamurthy, 2001. International and Domestic Collateral Constraint in a Model of Emerging Market Crises. Journal of Monetary Economics 48, 513-48. [9] Detragiache, E. A. Demirgüç-Kunt, 1999. Financial Liberalization and Financial Fragility , in Pleskovic, B., Stiglitz, J.E. (Eds.), Annual World Bank Conference on Development Economics. [10] Edwards S., 2001. Capital Mobility & Economic Performance: Are Emerging Economies Di¤erent? in Siebert, H. (Ed.), The World’s New Financial Landscape: Challenges for Economic Policy. Heidelberg and New York: Springer, pp. 219-44. [11] Feldstein, M.
C. Horioka, 1980. Domestic Savings and International Capital Flows. Economic
Journal 90, 314-29. [12] French, K. J. Poterba (1991. Investor Diversi…cation and International Equity Markets. American Economic Review 81, pp. 222-26. [13] Gourinchas, P.O. O. Jeanne, 2006. The Elusive Gains from International Financial Integration. The Review of Economic Studies 73, 715-741. [14] Gourinchas, P.O. O. Jeanne, 2007. Capital ‡ows to developing countries: The allocation puzzle. CEPR Discussion Papers 6561.
30
[15] Guiso, L., P. Sapienza L. Zingales, 2004. Does Local Financial Development Matter? The Quarterly Journal of Economics 119, 929-969. [16] Kose, A., Prasad, E., Rogo¤, K., Wei, S.-J., 2003. E¤ects of Financial Globalization on Developing countries: some Empirical Evidence. IMF Occasional Paper 220. [17] Kose, A., Prasad, E., Terrones, M., 2003. Financial Integration & Macroeconomic Volatility. IMF Working Paper 50. [18] Lucas, R., 1990. Why Doesn’t Capital Flow from Rich to Poor Countries? American Economic Review 80, 92-96. [19] Mian, A., 2006. Distance Constraints: The Limits of Foreign Lending in Poor Economies. Journal of Finance, vol. 61(3), pp. 1465-1505. [20] Obstfeld, M., 1998. The Global Capital Market: Benefactor or Menace? Journal of Economic Perspective, vol. 12 (4), pp. 9-30. [21] Obstfeld, M., 1999. Foreign Resources In‡ows, Saving, and Growth. in: Scmidt-Hebbel, K., Servèn, L. (Eds.), The Economics of Savings and Growth: Theory and Implications for policy. Washington and Cambridge University Press. [22] Obstfeld, M., Taylor, A., 2004. Global Capital Markets: Integration, Crisis, and Growth. Cambridge University Press, Cambridge. [23] Obstfeld, M., Rogo¤, K., 2000. The Six Major Puzzles in International Macroeconomics: is there a common Cause? NBER Macroeconomics Annual 15. [24] Prasad, E., Rajan, R., Subramanian, A., 2007. Foreign Capital and Economic Growth. Brookings Papers on Economic Activity 38, 153-230. [25] Razin A., Sadka, E., Yuen, C.-W., 1999. Implications of the Home Bias: A Pecking Order of Capital In‡ows and Corrective Taxation. in: Razin, A., Sadka, E. (Eds.), The Economics of Globalization: Policy Perspectives from Public Economics, Cambridge University Press. 31
[26] Rodrik, D., 2000. Savings Transition. The World Bank Economic Review 14, 481-507. [27] Song Z., K. Storesletten F. Zilibotti, 2011. Growing Like China. American Economic Review 101, 202-241.
7 7.1
Appendix. List of countries in the sample
Argentina, Australia, Austria, Belgium, Burkina Faso, Bangladesh, Bulgaria, Bahrain, Bolivia, Brazil, Botswana, Central African Republic, Canada, Switzerland, Chile, China, Cote d’Ivoire, Cameroon, Congo, Rep., Colombia, Comoros, Costa Rica, Cyprus, Germany, Denmark, Dominican Republic, Algeria, Ecuador, Egypt, Arab Rep., Spain, Finland, France, Gabon, United Kingdom, Ghana, Gambia, The, Greece, Guatemala, Hong Kong, China, Honduras, Haiti, Hungary, Indonesia, India, Ireland, Iran, Islamic Rep., Iceland, Italy, Jamaica, Jordan, Japan, Kenya, Korea, Rep., Kuwait, Sri Lanka, Luxembourg, Morocco, Madagascar, Mexico, Mali, Mauritania, Mauritius, Malawi, Malaysia, Niger, Nigeria, Nicaragua, Netherlands, Norway, New Zealand, Oman, Pakistan, Panama, Peru, Philippines, Papua New Guinea, Poland, Portugal, Paraguay, Rwanda, Saudi Arabia, Sudan, Senegal, Singapore, Sierra Leone, El Salvador, Suriname, Sweden, Syrian Arab Republic, Togo, Thailand, Trinidad and Tobago, Tunisia, Turkey, Uruguay, United States, Venezuela, RB, South Africa, Zambia, Zimbabwe.
7.2
A micro-foundation for borrowing constraints
Consider a type i entrepreneur (i = h; l) with w units of own capital who borrows Ld from domestic …nanciers at a gross interest rate rd , Lf units of capital from foreign …nanciers at a gross interest rate rf . This entrepreneur can hence invest w + Ld + Lf and reap a pro…t
= (w + Ld + Lf ) Ri
32
rd Ld
rf Lf
(40)
Now the entrepreneur can decide to default on her liabilities. She then pays back only a fraction of her liabilities to …nanciers but her return drops from Ri to (Ri
). The entrepreneur’s pro…t in this case
writes as = (w + Ld + Lf ) (Ri
)
qd;i rd Ld
qf;i Lf
(41)
where qd;i (qf;i ) is the share of claims, domestic (foreign) …nanciers are able to recoup on a type i entrepreneur who defaults. Let us assume a type j investor needs to pay
cj;i ln (1
q) L to recover a fraction q of a loan of size L
extended to a type i entrepreneur who defaults. Then type j …nanciers choose the fraction qj;i of the loans they want to recoup in order to maximize their repayment net of recovering costs:
qj;i = arg maxqrj Lj
[ cj;i ln (1
p
The optimal fraction qj;i should therefore verify (1 and solving the incentive constraint
q) Lj ]
(42)
qj;i ) rj = cj;i . Plugging this expression for qj;i in (41),
, the amount of capital Lf a type i entrepreneur can borrow from
foreign …nanciers veri…es Lf
cf;i
cd;i cf;i
w
Now assuming the parameters cj;i verify cf;l > cf;h >
Ld
and cd;h >
> cd;l , the borrowing constraint for
high productivity entrepreneurs simpli…es as
Lf
with
h
=
cf;h
and
h
=
cd;h cf;h
and
h;
hw
h
h Ld
(43)
> 0. Turning to low productivity entrepreneurs, their
borrowing constraint simpli…es as Lf
with
l
=
cf;l
and
h
=
cd;l cf;l
and
l;
l
lw
> 0.
33
+
l Ld
(44)
