Trust and Parameter Heterogeneity in the Neoclassical Growth Model
Dearmon, Jacob, The Journal of Developing Areas
Previous research has shown that trust alters input efficiencies across countries, which is suggestive of parameter heterogeneity. In this paper, trust's role in parameter heterogeneity is further explored within the context of the neoclassical growth model. This heterogeneity creates a nonlinear regression specification, which is estimated using a Metropolis within Gibbs algorithm. Results show that country-level trust differences cause human capital's exponent to vary by 43% and physical capital's exponent to vary by 29% across countries. This trust-induced parameter heterogeneity has important implications for various aspects of the development process. Under higher trust levels, the responsiveness of output per worker to changes in savings is increased by as much as 52%. In addition, this parameter heterogeneity also serves to mitigate untenably high implied rates of rate return differences between the US and other countries.
JEL Classifications: O11, O50, Z13
Keywords: Trust, Development, Parameter Heterogeneity
Corresponding Author's Email Address: email@example.com
The neoclassical growth model (NGM) has long been the workhorse of macroeconomic development research. A key assumption of the NGM and its associated linear specification is that the exponents of the production function, and consequently, the estimated parameters, do not vary across countries. In other words, the same production function characterizes each country, whether it is the United States or Zimbabwe. Recent empirical research, however, suggests that this assumption of parameter homogeneity might be too strong.
While parameter heterogeneity has been investigated in a wide variety of settings using a diverse set of empirical techniques, no one to date has investigated trust's role in the process. Recently, Dearmon and Grier (2009) have shown that trust increases the efficiency of macroeconomic factor inputs. This trust-efficiency relationship should generate parameter heterogeneity by causing the exponents of the traditional Cobb Douglas production function to vary across countries on the basis of trust. From an empirical standpoint, each country's NGM parameter estimate will also be different due to trust differences. Thus, this paper furthers the literature by examining the relationship between trust and parameter heterogeneity.
I find that trust does indeed generate parameter heterogeneity across countries, and this heterogeneity has significant consequences for macroeconomic development. Inter-country variations in trust cause human capital's exponent to vary by 43% and physical capital's exponent to vary by 29%. Consequently, increasing trust also raises the elasticity of output with respect to savings implying that an increase in savings will increase output per worker more for a high trust country than for a low trust country. Trust-induced parameter heterogeneity also serves to mitigate implied rates of return differences that are much too large under the traditional assumption of parameter homogeneity.
The paper is organized as follows. In Section 2, the literature on parameter heterogeneity and trust is examined. Section 3 has a brief description of the model. Section 4 discusses the data sources and variables used in the analysis. Section 5 contains the empirical modeling strategy. The analysis is provided in Section 6. Some closing remarks are made in Section 7.
The assumption of parameter homogeneity within the traditional NGM framework is far from trivial. While it may be a good approximation for the developed countries that it was originally designed to describe, its extension to the developing world is far more tenuous. Beginning with Grier and Tullock's (1989) finding that the NGM does not pool around the world, there has been an extensive literature examining this topic. Researchers, using a diverse set of statistical methodologies (ranging from quantile regression to regression tree analysis), have reached a similar conclusion: significant parameter heterogeneity exists across countries. …