NBER WORKING PAPER SERIES BUBBLES AND CAPITAL FLOWS Jaume Ventura Working Paper 9304 http://www.nber.org/papers/w9304 NATIONAL BUREAU OF ECONOMIC RESEARCH 1050 Massachusetts Avenue Cambridge, MA 02138 October 2002 Comments are welcome at [email protected] or [email protected]. I thank Pol Antràs for providing excellent research assistance. I am also grateful to the members of the Faculty Macro Lunch at MIT for their useful comments. Of course, none of them is responsible for any error or omission in the paper. The views expressed herein are those of the authors and not necessarily those of the National Bureau of Economic Research. © 2002 by Jaume Ventura. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including © notice, is given to the source. Bubbles and Capital Flows Jaume Ventura NBER Working Paper No. 9304 October 2002 JEL No. F15, F36, F43 ABSTRACT This paper presents a stylized model of international trade and asset price bubbles. Its central insight is that bubbles tend to appear and expand in countries where productivity is low relative to the rest of the world. These bubbles absorb local savings, eliminating inefficient investments and liberating resources that are in part used to invest in high productivity countries. Through this channel, bubbles act as a substitute for international capital flows, improving the international allocation of investment and reducing rate-of-return differentials across countries. This view of asset price bubbles has important implications for the way we think about economic growth and fluctuations. It also provides a simple account of some real world phenomenae that have been difficult to model before, such as the recurrence and depth of financial crises or their puzzling tendency to propagate across countries. Jaume Ventura Department of Economics MIT 50 Memorial Drive Cambridge, MA 02139 and NBER [email protected] This paper presents a stylized model of international trade and asset price bubbles.1 Its central insight is that bubbles tend to appear and expand in countries where productivity is low relative to the rest of the world. These bubbles absorb local savings, eliminating inefficient investments and liberating resources that are in part used to invest in high productivity countries. Through this channel, bubbles act as a substitute for international capital flows, improving the international allocation of investment and reducing rate-of-return differentials across countries. This view of asset price bubbles has important implications for the way we think about economic growth and fluctuations. It also provides a simple account of some real world phenomenae that have been difficult to model before, such as the recurrence and depth of financial crises or their puzzling tendency to propagate across countries. Tirole [1985] has argued that markets create asset price bubbles to eliminate inefficient investments.2 His argument goes as follows: Consider an economy where the growth rate exceeds the rate of return to capital. In this economy, a bubble can create its own demand without outgrowing savings by offering a rate of price appreciation above the rate of return but below the growth rate. The bubble absorbs part of the economy’s savings, crowding out investment and reducing the capital stock and output. Since the resources devoted to investment (roughly growth times the capital stock) exceed the resources obtained from such activity (roughly the rate of return times the capital stock), the bubble raises consumption and improves welfare. The key insight of Tirole’s theory is that the bubble takes away resources from inefficient investors and puts them in the hands of consumers. An implicit assumption in Tirole’s argument is that all investors face the same rate of return. Assume instead that, as a result of frictions in financial markets, the economy contains efficient investors that enjoy rates of return in excess of the growth 1 By a bubble or asset price bubble, I refer to the difference between an asset price and the net present value of its dividend flow or fundamental value. 2 Tirole’s paper builds on the path-breaking work of Samuelson [1958], who was the first to note that useless assets might be valued in a competitive economy and that this would be Pareto improving. 1 rate, and also inefficient investors that do not. In this economy, a bubble can create its own demand within the group of inefficient investors without outgrowing their savings by offering a rate of price appreciation above their rate of return but below the growth rate. The bubble crowds out inefficient investments and liberates resources that can be used not only to raise consumption, but also to increase efficient investments. Through this channel, the bubble can now lead to an increase in the capital stock and output. Since inefficient investors devote more resources to investment than they obtain from it, the bubble still raises welfare. The key insight now is that the bubble takes away resources from inefficient investors and puts them in the hands of both consumers and efficient investors. The goal of this paper is to examine the consequences of this view of asset price bubbles for the theories of economic growth and fluctuations. To do this, I keep two maintained hypotheses throughout the paper. The first one is that international goods markets are sufficiently integrated that long-run rates of economic growth are positively linked across countries. This implies that countries that save more and have better technologies and policies are richer but do not grow faster.3 The second hypothesis is that international financial markets have limited ability to arbitrage cross-country differences in rates of return. This might be due to a variety of frictions such as policy-induced barriers, transaction costs, information asymmetries and sovereign risk.4 If trade in goods ensures that all countries share the same long-run rate of economic growth while frictions in asset trade allow countries to have different rates of return, asset price bubbles naturally arise in those countries where the rate of return is below the common or world growth rate.5 3 Leaving a few miracles and disasters aside, there is ample evidence in support of this view. Despite large cross-country differences in economic policies, saving rates and technology, the world income distribution has been relatively stable in the second half of the twentieth century. Howitt [2000] has argued that this stability might be due to technology spillovers, while Acemoglu and I [2002] have argued that it might be due to terms-of-trade effects. Through any of these channels, countries with bad characteristics are able to grow fast and keep up with the rest of the world. 4 These frictions do not preclude all international capital flows, although the latter are much smaller than what theories based on frictionless markets predict. See Kraay et al. [2000] for a review of the evidence. 5 In an influential paper, Abel et al. [1989] noticed that the capital share exceeds investment in industrial countries. Since this implies that the average rate of return is above the growth rate, many have used 2 To study the implications of this observation, I construct a stylized world equilibrium model in which the cost of trading goods is negligible while the cost of trading assets is prohibitive. This is a crude but effective way to capture the two maintained hypotheses discussed above. The model has equilibria with bubbles, in addition to the bubbleless equilibrium that we always take for granted. These are country bubbles, since they can be sold only within the country.6 Bubbles appear in countries with low productivity, eliminating domestic investment and raising domestic consumption. This shift in demand lowers the price of investment goods relative to consumption goods all over the world, raising investment in countries with high productivity. Since the transfer of resources from low- to high-productivity countries is done via prices and without any actual or recorded capital flow, this could be aptly described as a theory of capital flows with zero current accounts. Since bubbles act as a substitute for international capital flows, many of their effects are akin to those that one would expect from financial integration. By improving the average efficiency of investment, asset price bubbles tend to raise the world growth rate. By shifting investments towards countries with high productivity, asset price bubbles tend to make the world economy more sensitive to shocks in these countries and less sensitive to shocks in other countries. By providing low- productivity countries with a better savings vehicle, asset price bubbles also tend to improve the world income distribution. By expanding at the end of booms and contracting at the end of recessions, bubbles tend to amplify the effects of productivity shocks on investment while dampening their effects on consumption. All of these effects, which are typically associated with financial integration, arise here as this observation to question the empirical validity of Tirole’s model. But this conclusion rests on the assumption that financial markets are frictionless. Even if the average rate of return exceeds the growth rate, the economy might contain pockets of investors with low rates of returns that are willing to buy a bubble. Hence it is not possible in general to rule out the presence of bubbles by comparing the aggregate capital share and investment. 6 In Ventura [2002], I consider the case in which financial markets are well integrated in the sense that domestic and international transactions are subject to the same sort of frictions. In this case, bubbles are traded across countries and I refer to them as global bubbles. 3 the sole result of asset price bubbles since capital flows are not possible and trade is always balanced by assumption. Bubbles also have some effects that are different from those that one would expect from financial integration. For instance, domestic and foreign shocks lead to movements in the size of the bubble that generate potentially large wealth effects. Through this channel, the presence of bubbles magnifies the effects of productivity and other shocks on aggregate activity. Another effect of bubbles is to open the door to shocks to expectations as a new source of macroeconomic fluctuations. Bubbles are inherently unstable, since they arise in environments where agents need to coordinate to one among a variety of possible equilibria. Since bubbles magnify the effects of productivity shocks and open the door to expectational shocks, low productivity countries that have large bubbles tend to be more volatile than high productivity countries that have small bubbles. But the relationship between volatility and productivity is nonlinear. Countries with intermediate levels of productivity tend to have smaller but more volatile bubbles. In particular, I find that in these countries external shocks generate movements in the size of their bubbles that have potentially large wealth effects. This observation can be used to provide a theoretical justification to the notion that middle-income countries are very vulnerable to small economic fluctuations in high-income countries. It can also be used as key building block in an attempt to sketch a theory of the international contagion of financial crises. The paper is organized as follows: Section one presents a stylized model of trade and growth and describes the world equilibrium without bubbles. Section two shows that there are additional world equilibria with bubbles and formally describes them. Sections three to seven present five examples designed to illustrate the macroeconomic effects of bubbles. The first two emphasize the role of bubbles as a substitute for international capital flows. The remaining three deal with the concepts of financial fragility, vulnerability to external shocks and contagion. Section eight concludes. 4 1. A Productivity-Based View of the Growth Process In this section I present a simple model of trade and growth that will serve as a tool or vehicle to study the conditions under which asset price bubbles can exist and their macroeconomic effects. The model has some unrealistic features such as the prediction that factor prices are equalized across countries or the assumption that the law of one price applies to all goods. These aspects of the model are not important however for the results of this paper, and have been chosen only to streamline the discussion. The two critical features of the model have already been discussed in the introduction: while trade in goods ensures that long-run rates of economic growth are equalized across countries, frictions in asset trade preclude the elimination of rate-of- return differentials across countries. There are various models of trade and growth that share these two critical features without predicting that goods and factor prices are equalized across countries. For instance, Acemoglu and I [2002] wrote a model with these characteristics recently. Consider a world economy with J countries, indexed by j=1,2,…,J. There are two factors of production: labor and capital, and they are used in the production of two intermediates: K- and L-products. These intermediates are in turn used to produce two final goods: consumption and investment. The costs of transporting factors are prohibitive, while the costs of transporting goods are negligible. In all countries there are competitive firms with access to a common technology to produce intermediates. To produce one unit of the K-product, these firms require one unit of capital. To produce one unit of the L-product, they need one unit of labor. Since final goods production only requires intermediates, full employment of factors implies that both intermediates are produced in all countries. Perfect competition ensures that the rental and the wage are equal to the prices of K- and L-products in all countries, while international trade ensures that the prices of these intermediates are equalized across countries. It follows that factor prices are 5 equalized across countries as well.7 Define r and w as the common rental (or price of t t K-products) and wage (or price of L-products) at date t. In all countries there are also competitive firms with access to a common technology to produce final goods. To produce one unit of the consumption good, firms have a Cobb-Douglas technology with a L-product (or labor) share equal to α, with 0<α<1. To produce one unit of the investment good, firms have a technology that requires one unit of the K-product. Since all firms in all countries face the same price of intermediates, the costs of producing final goods are also the same in all countries. Perfect competition then ensures that goods prices equal these production costs: (1) 1=r1−α ⋅wα t t (2) q =r t t where q is the price of investment and consumption is the numeraire. t The demographic structure of countries is that of a two-period overlapping generations model with constant population. All generations have size one. In each date, a new generation of consumers is born that lives two periods: young and old. The consumers’ goal in life is to maximize the expected value of old age consumption. When young, consumers work and save their wage. The only decision in their life is what to do with these savings. When old, consumers use the return to their savings to purchase consumption goods. This formulation is nothing but a stark version of the popular life-cycle model of savings. 7 This result is due to Samuelson [1948] and applies to a wide range of models (See Krugman and Helpman [1985]). What is perhaps surprising here is that factor prices are equalized for any cross- country distribution of capital-labor ratios. This is a special but very convenient property of this specific production structure that I borrow from Ventura [1997]. 6 In each generation, some of the young create and operate firms that purchase investment goods in their youth and convert them into capital that can be used in their old age. For simplicity, I assume that this entrepreneurial activity requires no effort and that there is enough talent in the country so as to drive the wage of entrepreneurs to zero. The entrepreneurs of country j at date t are capable of producing πj units of capital with each investment good. I refer to πj as the t+1 t+1 productivity of country j and assume that it can vary stochastically over time within a support that is strictly positive and bounded above. Let Ij and Kj be the investment t t and capital stock of the country. Assuming that capital fully depreciates in one generation, we have that: (3) Kj =πj ⋅Ij t+1 t+1 t Given these assumptions, each unit of income invested in country j yields a r ⋅πj rate of return equal to t+1 t+1 . To finance their purchases of investment goods, q t firms issue shares and sell them in the domestic stock market. After output has been produced and distributed in the form of dividend, the firm has no assets and the value of its shares drops to zero. A key assumption is that the costs of trading in the domestic stock market are negligible, while the costs of trading in foreign stock markets are prohibitive. This asymmetry in costs could be due to a policy-induced barrier such as prohibitive capital controls, or it could be due to the inability of countries to commit not to expropriate foreign investments. Whatever the reason, the young are forced to invest all of their savings in the domestic stock market, and this implies the following: (4) q ⋅Ij = w t t t (5) Cj =r ⋅Kj t t t 7 Equation (4) states that investment is equal to labor income, since the young save the entire wage. Equation (5) shows that consumption is equal to capital income, since the old have no bequest motive. Define world averages by omitting the country subscript. For instance, the 1 ∑ world average capital stock is K = ⋅ Kj . Applying Shepard’s lemma to Equations t J t j (1) and (2), we find the world average demands for K and L-products as 1 1−α α ⋅q ⋅I + ⋅C and ⋅C , respectively. Since the average supplies of these r t t r t w t t r t products are K and 1, international commodity markets clear if and only if: t (6) r ⋅K =q ⋅I +(1−α)⋅C t t t t t (7) w =α⋅C t t The competitive equilibrium of this world economy is a set of prices and quantities such that consumers optimize and markets clear. Together with an initial distribution of capital stocks, Equations (1)-(7) provide a complete description of this equilibrium. By Walras’ law, one among Equations (4)-(7) is redundant and can be dropped. I show next some properties of this world equilibrium. Let Rj be the rate of return to the portfolio of country j. Since this portfolio t contains only domestic investment, we can use Equations (1), (2) and (7), to find that: −α (8) Rj = g1−α ⋅πj t t t C where g is the growth rate of world consumption, i.e. g ≡ t+1 . Consider the case in t t C t which this growth rate is positive. Despite the continuous decline in the rental rate, 8