This thesis investigates the articulation of ~he incentives to perform Research and
Development of profit seeking firms. Throughout the thesis, the dynamic evolution
of the distribution of these incentives across firms is the engine of industry
transformation and growth. Thus, in order to a.'5SesS the impact of different industry
characteristics on the market structure, we need a faithful picture of the
context where firms make their R&D choices.
Chapter.one exposes more in detail the motivation to pursue the analysis
developed in each chapter independently, and how they combine to build up' the
search for the understanding of the interactions between R&D, appropriability
and market structure.
Chapter two presents a dynamic model of the firm size distribution. Empirical
studies of the firm size distribution often compare its moments to those of
a log-normal distribution, as implied by Gibrat's Law, and note important deviations.
Thus, the first and basic questions addressed in the first chapter are
how well does the dynamic industry model reproduce Gibrat's Law and how well
does it match the deviations uncovered in the literature. We show that the model
reproduces these results when testing the simulated output USing the techniques,
of the empirical literature. We then use the model to study how structural parameters
affect the firm size distribution. We find that, among other things, fixed and sunk costs increase both the mean and variance of the firm size distribution
while generally decreasing the skewness and kurtosis. The rate of growth in an
industry also raises the mean and variance, but has non-monotonic effects on the
higher moments.
In the third chapter we explore the implications of different degrees of R&D
appropriability on market structure and welfare. We propose a framework to
--
pursue this analysis by extending the ¥arkov-Perfect dynamic industry model
proposed by Ericson and Pakes (E-P, henceforth) (1995) through the introdu~tion
of a non-proprietary productivity component to R&D as part of a dynamic,
stochastic process. We first assume that spillovers are costlessly absorbed and
exploited by firms in the industry, and find that, in this case, a free rider problem
arises, thereby decreasing the incentives for investment. This leads to a lower
amount of innovations being developed in the industry, which in turn, implies
lower consuriier welfare while leaving the degree of concentration in the industry
fairly unchanged. We then model a settmg where it is assumed that in order to
build its absorptive capacity the firm has to engage in some R&D of its own.
In this case, we find that an increase in spillovers will enhance both consumer
and producer welfare substantially, and increases the likelihood of neck-and-neck
competition, therefore reduciJ;lg the level of concentration in the industry. These
results arise from the fact that having absorptive capacity being.built as a byproduct
ofR&D enhances the productivity ofR&D investment, compensating for
the free rider effect associated with the lack of appropriability.
The frameworks used in the two first chapters suffer from the 'curse of dimensionality',
such that the industries under analysis are limited in terms of the
number of agents simultaneously active. In order to overcome this problem, in
chapter four we move away from oligopolistic market structures and propose a model of monopolistic competition, where firms are sufficiently large to generate
a firm size distribution with a certain degree of asYIDlIletry, although each firm
is too small to affect the industry's outcome. Furthermore, we account for industry
growth by having the industry's output increasing over time as a result
ofknowledge externalities. The rich micro set-up of this model is analogous to
.that of E-P (1995), as it is composed by heterogeneous :firms making their in-
.~
vestment decisions in a world of uncertainty, but we abstract from entry and exit
and instead of an oligopolistic market structure we model a monopolistic competition
environment with many, heterogeneous firms. In thiS setting, :firms are
asYmmetric in terms of the technology they use to produce a given commodity,
and they are able to increase the likelihood of decreasing their marginal costs of
production by investing in Cost Reducing R&D. In order to evaluate their future
stream of profits and make their investment decisions :firms only care about the
eV9lution of their efficiency and the long-run efficiency index in the industry. Cutting
down the oligopolistic interactions present in the E-P framework, and having
firms looking at the long-run average industry state, allows us to overcome the
curse of dimensionality usually associated with dynamic models with agent heterogeneity.
Therefore, we are able to simulate the model with a large number
of firms competing in the industry and we show that, contrary to most existing
endogenouS growth models, this model is able to deliver a firm size distribution
with a substantial degree of heterogeneity.
Chapter 6 presents the final remarks to the investigation carried out in this
thesis.