1. Introduction
1.1 Measuring human capital
The measurement of intangible capital in general, and human capital in particular, is an intrinsically difficult task for economists, sociologists and statisticians alike. We are, after all, dealing with conceptually defined quantities that often cannot be directly observed. How does one measure things like creativity, ambition or loyalty? And are these valid factors to include in a measure of human capital? In the wake of the vast body of scientific literature on this topic, a decidedly incoherent taxonomy has emerged. There will, however , be no attempt in this paper to disentangle the multitude of definitions of human capital and the methods of its approximation. Instead the neoclassical framework of human capital will be adopted, principally developed by economists Jacob Mincer (1958) and David Becker (1962), leaving the issue of nomenclature to be dealt with on an as-we-go basis. In the neoclassical framework, human capital encompasses all conditions that are tied to the individual that affect her capacity to produce regardless of how or why those conditions have originated. Examples of those conditions include the accumulation of knowledge and experience, but also things like mental and physical health along with cognitive and motor skills.
On another note, I would like to emphasize that the ambition of this study is merely exploratory. No attempts will be made to test the causal directions of the associations that are estimated. However , since our theoretical framework is built upon assumptions of these causal directions, interpretations of the results will necessarily be made in relation to these assumptions so as to give the estimates economic meaning. But since the causal directions of the associations will not be explicitly tested for , those interpretations should be read as humble suggestions.
1.2 Background, purpose and motivation
The evolutionary process of economic progression has transformed the modern economy from previous stages of industrial and agrarian production into what is often referred to as a “knowledge economy”. Characterized by its enormous diffusion of information and knowledge, the knowledge economy carries with it a new economic functionality. In 1997 , Nobel Laureate Edward Prescott declared that cross-country differences in productivity are mainly explained by the ability of the workforce to adopt innovative knowledge (Prescott, 1997). It has since been shown that the need to understand how intangible inputs affect economic output in modern economies often surpasses the need to measure the contribution of buildings and equipment (Corrado et al. 2005) (World Bank, 2006) (Corrado et al. 2009) (Edquist, 2010).
In light of this context, the aim of this study is to estimate the distribution of human capital in the Swedish economy. The estimates will be derived from market valuations of human capital by disentangling the influence of observable as well as unobservable components of human capital on the wage of individuals. This will be achieved by means of an estimation strategy made possible by the rare availability of panel data in this study. Hopefully, the novel nature of the methodology employed, as well as the estimates themselves, will contribute in a significant way to the scientific field.
The approach of using variations in longitudinal data at the level of the individual to estimate human capital through wage equations has been applied before (Abowd, Kramarz and Margolis, 1999) (Abowd, Lengermann and McKinney, 2003) (Abowd et al. 2005) but is nevertheless quite rare in the scientific literature due to its critical dependence on the availability of data. The estimation strategy in this thesis will resemble those of the studies referenced above, and I clearly stand on the shoulders of these economists in this work. Though this is certainly not a replication study, a fair amount of methodological cherry picking will occur to enable comparisons between the studies. This will in turn offer an opportunity to confirm or reject previous findings in studies of this kind. Nevertheless , the method in this paper has to my knowledge never before been applied on Swedish data to obtain a measure of human capital.
Since the human capital measures are longitudinal, thereby making it possible to study temporal processes, they have the potential to become important tools of policy analysis as well as instruments for testing economic theory. For example, because the individuals in the panel can be traced to their registered residence, it is possible to explore the distribution of human capital across different regions to explain differences in regional growth or unemployment. Hypothetically, regions with high amounts of human capital should benefit (employment-wise) from the relatively compressed wage distribution in Sweden.
Alternatively, following the spirit of the literature on economies of agglomeration (see for example: Sörensson, 2010), one can investigate whether threshold amounts of different kinds of human capital can be identified in relation to regional economic variables such as business start-ups or commuting patterns. Another way of using the estimates is presented in Abowd et al. (2005) where the relation between human capital, productivity and market value is examined. One of their most intriguing findings is that the unobservable component of human capital is important in predicting the market value of firms, which indicates that firms who manage to attract workers with high levels of unobservable ability are recognized by the market as undervalued compared to otherwise similar firms.
Though the possibilities to gather new knowledge about the functioning of human capital are indeed plentiful, delimitations necessitated by time and resources inhibit an exhaustive effort to do so in this paper. Instead the focus will be restricted to the estimation procedure and the properties of the estimates themselves. The only exception to this in the paper is found in section 5.3, where human capital and productivity growth in different branches of industry are briefly compared to exemplify the use of the obtained human capital measures.
1.3 Disposition
The paper will proceed as follows: in chapter 2 a population model will be specified and the whole strategy for estimating the parameters presented. Specifically, it will present to the reader, in a general context, a solution to the problem of estimating the effect of something that is unobservable. It will then guide the reader through the process of using these estimates to obtain a measure of human capital. Chapter 3 will contain information about the data used to estimate the human capital, including a description of the variables in the model and some summary statistics.
Chapter 4 will begin with an overarching discussion of the affinities of the experience component of human capital and the sources of variation that need to be controlled for in the estimation model. This will be followed by some results from the estimation procedure. Chapter 5 will present novel statistics on the distribution of human capital in Sweden. It will also include tests to assess the estimation strategy, followed by a simple application of the human capital estimates where they are related to productivity growth in different branches of industry. Finally, chapter 6 will conclude with a summary of the study and its results, as well as some suggestions for future research.
