RICERCA ITALIANA     

FINANCIAL FRAGILITY AND TECHNICAL PROGRESS WITH HETEROGENEOUS AGENTS AND SOCIAL INTERACTION: MODELS, SIMULATIONS, EMPIRICAL ANALYSIS

Università Cattolica del Sacro Cuore

Abstract

The national research project is centered around a framework (described in section 2.1) for the analysis of financing, production, investment and technology adoption decisions of interacting, heterogeneous firms. Heterogeneity concerns size, expectations, state of technology and degree of financial fragility. Therefore, the framework is characterized, as to methodology, by the emphasis on heterogeneous interacting agents and, as to the contents, by the focus on the nexus of firms' capital structure, production, investment and technology adoption. This nexus is deeply affected by monetary policy. The local units will contribute to the specification, completion and development of this framework (on the base of the division of labor outlined at 2.4) and of complementary research areas. The empirical research will be carried out developing and using specific databases for econometric studies. <<<

Principal Investigator

Domenico DELLI GATTI Università Cattolica del Sacro Cuore

Research Objectives

The national research group will focus on a framework (outlined in Delli Gatti et al. (2004) centered around the financing, production investment and technology adoption decisions of firms in the presence of "financial frictions". In this context, characterized by a financing hierarchy and the risk of bankruptcy, macroeconomic outcomes are the results of social interaction - both strategic and non-strategic (mean field effects) - among heterogeneous agents. The dynamics of macro-variables are affected by the distribution of firms by size, state of technology and financial conditions, captured by indicators of financial fragility such as the leverage ratio (in the above mentioned framework we use the equity ratio, an indicator of financial robustness inversely correlated to the leverage ratio) and are affected, in this setting, by the adoption and diffusion of new technologies in which a relevant role is played by financing of R&D expenditure and technological learning.
We will develop, therefore, models of agent-based economies in which the financial constraints of heterogeneous interacting agents affect real activity (production, investment, technology adoption) and financial developments (Stock price dynamics, volume and allocation of credit extended). These models can be built starting from the individual laws of motion, obtaining the dynamics of the aggregates by means of summation and averaging (bottom-up procedure). They usually yield complex dynamics both at the individual level and in the aggregate. Moreover, the distributions of firms'size generated by simulations are asymmetric and generally follow a power law. These two outcomes are correlated: the emergence of scaling behavior (power law distributions) is a symptom of self-organized criticality.
Monetary policy exerts a remarkable influence on this nexus of relationships, not only by means of aggregate shocks - such as a change of the interest rate - but also by means of its effects on the structure of the economy i.e. on the distribution of firms by size, net worth, etc. <<<

First Results

First of all, we will write essays and position papers which outline specific lines of research in fields which are covered by more than one research unit. However, also in this first stage there will be original results obtained building, developing and extending the models more familiar to some of the research groups. The empirical work will be mostly preliminary to the econometric research which will be carried out mainly in the second stage.In this second stage, we aim at producing mainly innovative papers to be presented in meetings, seminars, workshops and to be published possibly in the major journals. Some of these workshops will be jointly organized by the local research units. <<<

Timescale

24 months

National and international background

The macroeconomic impact of capital market imperfections has been an active international research field (with a basically New Keynesian flavour) since the early ‘80s. A growing body of results shows that these "financial frictions" (Cooley and Quadrini, 2001) may have far-reaching consequences on the dynamics of aggregate income, policy effectiveness, the sources and propagation of business fluctuations. The foundations of this macroeconomics of imperfect capital markets rest on the violation of the Modigliani-Miller irrelevance theorem, which is at the root of three well known phenomena: (i) financing hierarchy: different ways of raising funds have different costs for the borrower; (ii) bank dependence: bank debt is the second most important source of finance (after internal funds) and for many medium-small borrowers it is the sole external source of finance; (iii) financial constraints: some classes of borrowers have serious difficulties in accessing capital markets; hence their ability to invest is constrained by the availability of internal funds.
As a consequence firms' spending capacity depends on their internal assets (net worth or equity base) - as emphasized by the literature on the balance sheet channel - and on banks' lending policy (lending channel). Bank dependence has a major implication in that levered firms face bankruptcy risk. In the Greenwald-Stiglitz approach, from the attempt to internalize the risk of bankruptcy follows a positive relationship between net worth and production which is the key to the the supply-side effects of capital market imperfections (Greenwald and Stiglitz 1988a,b; 1990, 1993,2003).
The dependence of firms' spending on net worth is the focus of models which emphasize financial factors as amplifying and propagation mechanism in business flucttuations (financial accelerator: Bernanke and Gertler, 1989, 1990; Bernanke, Gertler and Gilchrist 1996, 1999, Kiyotaki and Moore, 1997, 2002). Apart from some differences, the financial accelerator macro-models share several features: (i) the informational frictions on capital markets insert a wedge (the premium for external funds) between the cost of internal and external funds; (ii) this premium depends negatively on the borrower's balance sheet; (iii) there exists a positive relationship between the individual borrowers' balance sheets and aggregate activity.
The financial accelerator approach implicitly or explicitly assumes that agents are heterogeneous. Matter of factly firms are different from two points of view: access to capital markets and degree of "financial fragility" as originally defined by Minsky. In the financial accelerator approach, however, the dynamics of the financial positions of heterogeneous firms is not thoroughly analyzed, and there is no interaction among agents. This is particularly troublesome since heterogeneous agents' interaction seems to be the ultimate cause of the several scaling laws that the literature has detected as universal features of many economic phenomena (Amaral et al. 1997, 1998a,b; Gabaix, 2002; Gaffeo et al, 2003a,b).
The statistical properties that the evolution of firms must satisfy in order to reproduce the empirical evidence were originally analyzed by Gibrat (1931). He traced back the right skewed distribution of firms' size to the fact that the growth rate of a firm does not depend on its size. Gibrat's law of proportional effect has not been(always) corroborated by the evidence but the right skewed distribution of firms'size is a well established fact in the empirical literature (Sutton, 1997; Caves, 1998). The specific functional form of the distribution, however, is controversial. Gibrat's conjecture according to which firms' size follows a log-normal distribution has been rejected in the more recent literature, according to which the best fit is obtained by a power law of the Zipf type (Axtell, 2001; Gaffeo et al. 2003b).
The theoretical problem consists in explaining why and how these power law distributions emerge. One of the possible answers is that they are the typical outcome of systems composed by a large number of heterogeneous interacting subsystems. In fact several agent-based models show this power law property.
The methodology of agent-based modeling has grown substantially in the last few years, spreading to many disciplines. In Economics it is adopted mainly to analyze markets characterized by non-price interactions. In this way the economy con be viewed as a Complex Adaptive System (Arthur, Durlauf and Lane, 1997; Mansky 2000; Brock and Durlauf, 2000). These techniques are particularly suitable for contexts in which the representative agent assumption is blatanly unrealistic. The representative agent hypothesis rules out, by construction, the persistence of heterogeneity (and the occurrence of power laws). Moreover, as Kirman (1992) and Forni-Lippi (1998)stressed, this assumption can lead to wrong conclusions when analyzing aggregate quantities. In a context of heterogeneous agents -such as agent-based economies - aggregate behavior depends on the whole distribution and on the type of interaction (e.g. local or global) involving agents. Therefore, to understand macroeconomic fluctuations one has to analyze industrial dynamics, i.e. the entry-exit process and the process of internal growth of surviving firms. As a consequence, studies in industrial organization and macro-dynamics naturally converge to the same field. From the macroeconomic point of view, for instance, the emergence of power law distributions suggests a link between the business cycle and scaling behavior. Gabaix (2002) points out that if the distribution of firms'size has fat tails, small idiosynchratic shock may lead to large aggregate fluctuations. <<<