This work is motivated by the following empirical regularities about the relationship between credit, growth and volatility. First, there is a positive cross–country relation between economic growth and credit market development as measured, for instance, by the ratio of private credit to GDP. While it is undisputed that financial development and growth go hand in hand, their causal relationship is a much debated issue in the empirical literature. Second, there is a negative cross–country relation between the volatility of GDP growth and the level of economic and financial development. Along a similar vein, in many developed countries aggregate output volatility has declined considerably together with an expansion of the financial sector during the last decades. And third, the fall in macroeconomic volatility has been accompanied by a rise in microeconomic (firm–level) volatility.
The purpose of this paper is to account for these observations in a model in which both economic growth and credit market development are endogenous. A more developed credit market improves the efficiency of resource allocation, contributing thus to higher growth. Conversely, a growth push makes credit markets more valuable, improves financial development and reinforces the initial growth effect. Thus, the model is consistent with the first stylized fact, incorporating a two–sided linkage between finance and growth. Our model is also consistent with the other two facts. An expansion of the credit market goes hand in hand with a decline in aggregate volatility. Moreover, there is a hump–shaped relation between credit market development and idiosyncratic (firm level) volatility. Thus, a credit expansion may easily induce a decline in aggregate volatility together with a rise in idiosyncratic volatility.
We consider a model of linear endogenous growth where a continuum of infinitely lived producers draw their idiosyncratic capital productivities from a distribution that itself varies with an aggregate state. More productive agents wish to borrow from less productive ones, but borrowers are constrained in their demand for loans because only a fraction of their assets can be seized in the event of default. We assume that there is an exogenous institutional parameter m ? [0,1], termed “creditor rights”, which is the fraction of principal and interest owed that a bankrupt lender must pay to his creditors. After this penalty, defaulters are perpetually excluded from future borrowing while their assets are protected against former creditors. The potential threat of exclusion from credit specifies endogenous borrowing limits which are just tight enough to prevent default. These borrowing limits depend on the creditor rights parameter m and also on economic fundamentals which determine how much producers value participation in credit markets.4 Thereby, credit market development becomes endogenous and responds both to changes in institutions and to changes in economic fundamentals. The special case m = 1 corresponds to the standard SDGE environment of perfect capital mobility where borrowing limits are irrelevant. Whenever m < 1 however, we show that capital mobility is imperfect so that some agents must be rationed (Proposition 1). In some extreme circumstances (particularly, weak creditor rights and impatient producers), the credit market shuts down completely (Proposition 2). Nevertheless, even in the sovereign default environment of Bulow and Rogoff (1989) with m = 0, positive borrowing limits may be sustained, as is the case in Hellwig and Lorenzoni (2003).
The paper has four main results. First, economies with low growth and under developed financial markets tend to have higher aggregate volatility than economies with perfect capital mobility (Proposition 3). This result follows whenever there are fluctuations in the productivity distribution which are uncorrelated to fluctuations of the technology frontier. A credit expansion shifts more funds to the technology frontier so that aggregate output becomes less volatile. We also explore the relation between credit and aggregate volatility quantitatively in a calibrated version of the model in Section 6.
Second, the economy can have multiple balanced growth paths. A high–grow the quilibrium where debt constraints are loose may coexist with two low–growth equilibria with tighter constraints (Proposition 4). An implication of this result is that small changes in institutions or in economic fundamentals can trigger a take–off of growth and financial development. The reason why there are multiple equilibria is a dynamic complementarity in endogenous borrowing limits. Agents’ expectations of future credit market conditions affect their incentives to default, and this in turn takes an impact on their current borrowing limits. If future constraints are tight, agents value participation in credit markets only little and their incentives to default are high. Consequently, borrowing agents face currently tight constraints. But there may also be an equilibrium where agents expect loose credit limits in the future so that participation in credit markets is desirable; in turn, agents do not default easily and current borrowing limits are loose. We also find that some of the low–growth equilibria are indeterminate, so that there exists an infinity of stochastic (sunspot)equilibria (Proposition 6).
Third, when innovations push the technological frontier upwards, credit markets become more valuable to borrowers who want to make use of the leading technology. Thus they get punished more severely if they are excluded from borrowing which reduces their incentives to default. Hence borrowing limits relax, the volume of credit goes up, and funds are more efficiently allocated. Besides a direct growth effect of the technological innovation, there is an indirect growth effect that results from improved financial development. Similarly, enhanced access to better technologies (e.g. because of better education) makes credit market participation more valuable, improves credit market development and takes both a direct and an indirect impact on growth (Proposition 5).
And finally, there is a hump–shaped relation between financial development and idiosyncratic volatility, as measured by the volatility of equity returns or firm growth. At low level of financial development, volatility of the equity return merely reflects the volatility of the firm’s capital return, but when financial development improves, the leverage effect raises the spread between equity returns in high and low productivity periods. At very high levels of credit market development, however, the leverage effect disappears together with firm–level volatility. The positive effect of leverage on idiosyncratic volatility has also been explored in the empirical finance literature, see e.g. Dennis and Strickland (2005).
There is a substantial literature on the role of credit market frictions for economic growth (e.g. Greenwood and Jovanovic (1990), Bencivenga and Smith (1991), Marcet and Marimon (1992), Galor and Zeira (1993), Azariadis and Chakraborty (1999)). Like these papers, our model is compatible with the view that a higher level of financial activity spurs economic growth. In contrast to most of the existing literature, however, this paper assumes perfect information and there are no exogenous frictions in the process of financial intermediation. The only friction is the inability of borrowers to perfectly commit to repay their loans. Credit constraints arise endogenously without explicit costs of financial intermediation. Similar to our paper, Acemoglu and Zilibotti (1997) develop a theory of financial development which improves endogenously in the growth process. Their model is based on the idea that, due to project–size indivisibilities, security markets in less developed countries may be incomplete, preventing investment in riskier profitable projects, whereas security markets improve as the economy reaches higher stages of development. In this paper, there are no such indivisibilities, and market incompleteness is not decisive for the results.
As in this paper, there are multiple steady states in the growth models of Galor and Zeira (1993), Acemoglu and Zilibotti (1997) and Azariadis and Chakraborty (1999). In the first of these papers, the initial distribution of wealth is the sole determinant of the long–run growth performance, whereas it is the initial stock of capital in the other two papers. In our model, in contrast, neither the distribution of wealth, nor the initial capital endowment play a decisive role for the long–run growth performance. Instead, financial development can be purely a matter of coordination in the credit market. When financial markets are expected to work well in the future, credit markets are in a better condition today since agents have stronger incentives not to default.
The paper is organized as follows. Section 2 lays out the general environment where firms draw productivities from an arbitrary distribution. Sections 3 and 4 characterize and analyze equilibria in this general environment, establishing results on imperfect capital mobility, autarky equilibria and volatility of growth. Section 5 focuses on the special case of a two–point productivity distribution, discussing multiplicity and indeterminacy of equilibria, as well as idiosyncratic volatility. Section 6 contains a numerical example that is calibrated to the US economy to show how much aggregate volatility can vary with creditor rights. Variations of the model are discussed in Section 7, and proofs which are not included in the main text are contained in the Appendix.
Download
PDF Credit Market Development, Growth and Volatility
