خود اشتغالی، آموزش و پرورش و اعتبار محدودیت: یک مدل تصمیم گیری سهمیه بندی اعتباری به هم وابسته

An OLG model with occupational choice and endogenous credit constraints is developed. Heterogeneous agents decide whether to become educated when young; this requires borrowing. In the second period, middle-aged educated agents decide whether to become self employed in the following period, which again requires borrowing. Credit may be rationed because of a moral hazard problem in lending. In a macroeconomic framework with capital, we derive a number of comparative statics results; we consider the effects of changes that relax credit constraints on the real wage and the real interest rate and discuss issues relating to the interdependence of credit constraints.

There is now much evidence of the importance of financial factors for economic growth and development (for a review of the evidence see Levine, 1997); however, the process of financial liberalization and development is far from unproblematic, as certain East Asian economies have discovered in recent years.1 One reason may be that credit markets are notoriously imperfect; it is thus important to analyse the role of credit market imperfections in a macroeconomic context and consider the effects of financial market development in such an environment, and it is intended that this paper should make a contribution in this area. There is a large literature in economics on credit rationing, which occurs when investment projects with an expected return greater than the cost of funds are not financed.2 The existence of such rationing is compatible with rational behavior by lenders, who may be unwilling to lend more than they do, even at a higher interest rate, because if they do so, there is a higher probability that borrowers default on loans. Two common explanations for increased default under these circumstances are adverse selection and moral hazard. Adverse selection means that the mix of borrowers becomes less favorable as interest rate rise; moral hazard means that the behavior of given borrowers changes in such a way as to reduce expected loan repayments after an increase in interest rates. One factor in many borrowing decisions is the likelihood of being able to borrow again. There are many examples of this interdependence: a firm’s return on borrowing to build a factory may depend crucially on whether it will be able, later on, to borrow to finance working capital. It might seem that the existence of credit constraints in one sector of the economy may exacerbate credit constraints in other areas; a related point is that: “Once the assumption of perfect capital markets is dropped, every lender must take into account not only the real characteristics of the firm or individual to whom he lends but also the constraints on [his] future borrowing” (Gale, 1982, p. 161). The main innovative feature of this paper is to build a macroeconomic model that incorporates this interdependence of lending decisions. Our model has an overlapping generations (OLG) structure where agents live for three periods and population is constant. In the first period (youth) agents decide whether to attempt to borrow to finance education; if they do so and are successful they become educated; if not, they remain uneducated and receive the unskilled wage for all three periods. In the second period (middle age), an educated agent works at the skilled wage and can apply for a loan to finance an investment that enables him to become self-employed in the third period (old age).3 If unsuccessful, he continues to work as an educated worker in his old age. Those who apply for loans may be denied credit because of the legitimate fear by lenders that if they do lend, the borrowers will “take the money and run” (abscond). We hence adopt a simple moral-hazard explanation of credit rationing; similar accounts appear in Banerjee and Newman, 1993, Banerjee and Newman, 1998 and Fender and Wang, 2003 and Lloyd-Ellis and Bernhardt (2000). If lending is to take place, an Incentive Compatibility Constraint (ICC) must be satisfied: the borrower must be at least as well off borrowing the money and using it to purchase either education or self-employment inputs as he is absconding. There is a cost to absconding, which may represent the costs of avoiding penalties, or the expected value of penalties, or some appropriate mixture of the two. In equilibrium, agents do not abscond; lenders do not make loans to agents which would give them an incentive to do so. Increases in the cost of default will typically reduce credit rationing, as absconding is now more costly. Financial development, which may include measures designed to increase the ability of creditors to enforce repayment, may be interpreted as increasing these costs. This framework is embedded in a macroeconomic model where educated workers combine with capital to produce output. Whether credit rationing occurs depends on parameter values; if the cost of default is sufficiently high, then the ICC will always be satisfied when borrowers borrow the amount they would wish to on the assumption they repay the loan, so credit rationing will not occur. Alternatively, for a sufficiently low cost of default, borrowers will always prefer to abscond if they borrow the efficient amount, so there will be credit rationing. So there are four possibilities: there could be rationing for both types of loan, for neither, or for just one but not the other. A further complication is caused by our assumption that education is necessary for self employment; it could be the case that all those educated become self employed, or perhaps more realistically, that there are some educated workers who do not wish to become self employed. So for each assumption about whether credit rationing does or does not occur in each period, there are these two possibilities, and thus there are eight different types of equilibria (or Regimes), each of which generates a fair number of comparative statics results. (Regimes are hence defined by (i) whether or not there is rationing of self-employment loans; (ii) whether or not there is rationing of education loans; (iii) whether or not all the educated become self employed.) Presenting results for all eight Regimes would take up too much space and not be necessary, as many results are similar across Regimes. We concentrate on the two Regimes that are of greatest interest, namely those with credit rationing in both periods and show that, in general, credit rationing is associated with a lower real interest rate and higher skilled wage. (However, we also discover circumstances under which the opposite will happen.) Financial development will then reduce the skilled wage but raise the real interest rate; it will also promote occupational mobility as more agents are able to purchase education and/or borrow to become self employed. Unskilled workers who do not purchase education are better off because of the higher real interest rate, whereas the self employed, who borrow twice, first to finance education and then to become self employed, are worse off both because of the higher real interest rate and because of the lower skilled wage. So financial development reduces inequality. (It is sometimes argued that financial development is accompanied by increased inequality, as in Greenwood and Jovanovic, 1990). We would stress, however, that the financial development we are considering concerns increased access to loans for human capital improvement purposes, whereas most analyses consider borrowing to finance physical investment. One issue we were interested in exploring was whether relaxing credit constraints could have multiplier-type effects—that is, relaxing one constraint relaxes another, and this feeds back on the first, relaxing it further, and so on, producing a much more powerful “credit multiplier” than might be expected from studying just one market. We do indeed find this is possible; however, we also find that “crowding out” effects can occur as well—if one credit constraint is relaxed, then there may be general-equilibrium repercussions that mean that the other credit constraint is strengthened. There is some related literature. Stiglitz and Weiss (1983) and Scharfstein and Bolton (1990) develop the idea that lenders may use the threat of denying credit in the future as a way of inducing borrowers to repay current loans. Stiglitz and Weiss’s main concern is to establish the possibility that lenders may use the threat of rationing credit as an inducement to encourage borrowers to repay their current loans. Scharfstein and Bolton build a model of predation around such an idea. Kiyotaki and Moore (1997a) present a model of short-term trade credit; a firm’s ability to repay a loan afforded it by a supplier may depend on whether a firm to which it is supplying goods repays the loan it has been granted. Gale (1982) explains how inflation can exacerbate liquidity constraints as follows: a firm with a given nominal indebtedness is forced to contract its real indebtedness at a faster rate when inflation rises. Why cannot the firm raise its nominal indebtedness in these circumstances? Lenders may not be willing to lend to enable a firm to make interest payments (a “rule of thumb” which has evolved under non-inflationary conditions). But why does one particular lender not find it profitable to lend to a firm for such a purpose? An explanation is that he only wishes to lend for a limited period of time, and does not believe that other lenders will be willing to take over when he wishes to withdraw his money. So a lender may not lend because he, quite rationally, believes that other lenders will be unwilling to lend in the future. Since all lenders behave this way, no lending in these circumstances is carried out. But none of these papers is at all similar to ours, which looks at the longer term, macroeconomic effects of such interdependence. There are a number of other papers in the literature that incorporate credit market imperfections in a general-equilibrium framework; examples are Azariadis and Smith, 1993 and Azariadis and Smith, 1998 and Kiyotaki and Moore (1997b). Our model extends Fender and Wang (2003), which is a general-equilibrium model with just one borrowing decision and one occupational choice, to incorporate an interdependence of lending decisions. Other examples of what might be called general-equilibrium models with credit constraints, but more explicitly directed towards development issues, are Banerjee and Newman, 1993 and Banerjee and Newman, 1998 and Acemoglu and Zilibotti (1999); in these papers a moral-hazard problem afflicting lenders is a crucial part of the story. Galor and Zeira, 1993 and Aghion and Bolton, 1997 and Piketty (1997) all consider the dynamics of wealth distribution with credit constraints, and wealth redistribution can improve the efficiency of the economy by relaxing these constraints. Our model differs from those presented in these papers by, inter alia, abstracting from wealth effects and bequests; it also has two sequential occupational choice decisions and examines the effects of relaxing the credit market constraint. The plan of the rest of the paper is as follows: Section 2 presents the basic model; Section 3 gives and discusses the comparative statics results and Section 4 concludes.