A simple enquiry on heterogeneous lending rates and lending behaviour

respect to the opportunity cost seems to be lower. The relevance of geographic areas is again emphasized
by Finaldi Russo and Rossi (1999), who find localization in industrial districts to be a relevant and
significant variable.

The focus of most empirical contributions in this research field is on agents’ heterogeneity and its
implications for the transmission of monetary shocks and for the process of aggregation of different
observable variables. Obviously, all this does not seem to match with the “new econometrics”
methodological approach, which postulates that the macroeconomic relations to be estimated have to be
microfounded on the basis of a representative agent model. Since it is very hard to capture any
heterogeneity within the representative agent framework, at least three theoretical attitudes seem to be
possible in this regard. The first one consists of not taking too literally the “new econometrics” approach
and still performing empirical analyses with aggregate time series based on theoretical assumptions only
defined in aggregate terms. A good argument in this regard is provided by Blinder (1986) who claims that
the use of specific properties associated to the representative agent’s utility function may cause serious
bias in the estimates, since “for many goods, the primary reason for a downward sloping market demand
curve may be that more people drop out of the market as the price rises, not that each individual
consumer reduces his purchases”¹. Another argument lies in the fact that the utility function of a
representative agent is actually a “non-microfounded macroeconomic function”: it could be theoretically
obtained by integration on the basis of an aggregate consumption, and its analitycal form is simply based
on “ad hoc” assumptions².

Of course, a second (and very popular) theoretical attitude consists of only performing empirical
studies with individual microdata, in order to properly capture the different and asymmetric effects that
might be generated by heterogeneous agents and - at the same time - avoid any Lucas’ critique objection.
There is no doubt that this approach is very appropriate, but some residual (although relevant) problems
might arise from sampling bias, since it is very difficult to exclude that some specific categories of
individuals can be underrepresented (for instance, very small firms, in case one wishes to study the impact
of monetary policy on the industrial sector of an economy). For all these reasons, we follow a third
approach and perform empirical analyses with an intermediate or “mesoeconomic” level of data
aggregation might provide a relevant source of information. Some goods arguments in this regard might
also be provided by the recent literature on the statistical properties of aggregated and disaggregated data.
Forni and Lippi (1997), for instance, show that the conventional mainstream econometric assumption on

¹ Blinder, cit., p. 76

² In addition Benartzi and Thaler (1995), Kahneman (1994), Shafir, Diamond and Tversky (1997) report a very extended
experimental evidence showing that the actual behaviour of individuals choosing in conditions of uncertainty is characterized
by the so-called “status quo bias” and can be modelled with a “kinked utility function”, extremely different from any analytical
form commonly employed to model a “nicely behaved” utility function.

<<   1   2   3   4   5   6   7   >>

More intriguing information