Financial Fragility, Patterns Of Firms' Entry And Exit And Aggregate Dynamics

Recent literature stresses the drawbacks of the representative agent ap-proach. A growing number of contributions construct model which allows for agents heterogeneity [1, 2] while others claim that aggregate phenomena could be explained by the failure of the law of large numbers rather than by exogenous shocks (see [3] for instance). Another strand of literature stresses the importance of financial factors in determining the behavior of real vari-ables [4, 5, 6]. A few recent papers combine the two concepts. They show that a measure of dispersion of the distribution, other than its average value, matters for the dynamics of the aggregate variable, i.e., one should consider the whole (statistical) distribution of individual variables [7]. Of course, this distribution is heavy in uenced by the "entry-exit" process: bankruptcy eliminates the "worse" tail of the distribution, while the characteristics of the entrants modify it. Some analytical results has been already reached within such a framework [8].The goal of this paper is to verify if and how information availability on the credit market affects the economic dynamics and the distribution of firms with respect to their financial position and size. Moreover, ows into and out of the industry could be affected by credit availability. Our results show that the mean level and the variance of indebtedness of the system varies with the availability of information. In particular, if information is asymmetric the firmsÆ leverage ratio (as proxied by the ratio of corporate debt to capital) is higher and varies very smoothly. Aggregate output is also higher when firms are very leveraged and uctuations are amplified. Since the share of more leveraged firms is procyclical, cyclical downturns are driven by the exit process. We may therefore state that, according to this model, firmsÆ financial distribution movements drive the business cycle. Our results are corroborated by an econometric analysis of the model (European data 1960:I-1997:IV).We construct an agent based model on the basis of [9]. Differently from it, we introduce a credit supply curve from the profit maximization problem of the bank. This allows us to determinate the credit equilibrium conditions for each firm. The "bank problem" deserves some explanation since it in-troduces us to the "spatial structure" of the model. Because of information imperfections, banks do not know the "true" capital level of each firm and subscribe a standard debt contract with debtors: i.e., when debt is greater than capital, the bank takes all the capital. "Lemon" firms are allowed to reduce their capital to zero if the bank asks its money back. In such a case, a bank which finances a lemon firm, may lost all the loan. We assume that the bank activity is concentrated in a local zone and we endow a bank with a square lattice filled with firms to finance. By means of a "research center" the bank knows the average level of the surrounding firmsÆ capital: i.e. it knows only the average level of the financial indicator. We also assume that the share of the lemon firms is a positive function of the mean leverage ratio of the zone. It is possible now to deal with the heterogeneity of the financial position of the firms and interaction among agents (in the case in point, there is indirect interaction through banks.)The entry exit rules are modelled as follows. The exit rules are straight-forward in this framework. First, if a firm is very leveraged, it can judge that the advantage to remain on the credit market is lower than the bur-den of debt commitments (these are the lemon firms) and leaves the market. Secondly, if a firm demands for too much credit, the bank may adopt credit rationing policies. Even this kind of firm leaves the market, even if it is able to repay all the credit. In this case the banks avoids the formation of new lemons.For the entry process we exploit our spatial structure drawn from physics field, in particular from the model of magnetization. We endow each site of our lattice with a structure enclosing its four nearest neighbors. If a site is full, nothing happens and the firm goes ahead with its activity. If the site is empty, itÆs filled according to a Gibbs probability distribution with a nearest neighbors potential with anti-ferromagnetic characteristics. This means that the probability of entry rises with the number of empty neighbors. Such a rule has proven to be very useful when we consider more banks, i.e. more zones. In this situation we can analyze the case in which the firm care about credit conditions in the entry decision. If more zones are present, we let the exponent of our Gibbs distribution to depend positively on the financial situation: all in all, the probability of a birth is higher in the zones where credit conditions are favorable.