Monthly Archives: June 2014


Our expectation is that knowledge follows a diffusion process through geographic, institutional and technological spaces. Thus, researchers that are “nearby” along each of these dimensions would be particularly likely to benefit disproportionately in the time period immediately after the antecedent innovation occurs. We expect, however, that this “localization effect” will tend to fade over time, so that eventually the probability of an antecedent benefiting a remote descendant may be no lower than the probability of benefiting one nearby.


We have three goals in this paper. First, we demonstrate how an econometric model can be used to make citations a potentially useful measure of knowledge flows, by controlling for the effects of truncation, changes in citation patterns, and technology field effects. Second, we explore for the first time the citation patterns among all combinations of the G-5 countries, the U.S., Great Britain, France, Germany and Japan. This gives a much richer picture of the geographic dimension of citation diffusion, by examining the extent and speed of diffusion of citations within and among all combinations of these countries. This permits us to estimate the extent and nature of “localization” of citations within each of these countries, to examine differences among the countries in their apparent absorption of foreign technology, and to identify some interesting pairwise interactions. Finally, we add the dimensions of “institutional localization” and “technological localization” to the modeling, and examine the interactions between localization in these dimensions and in geography.


i (2)
Knowledge is inherently nonrival in its use, and hence its creation and diffusion are likely to lead to spillovers and increasing returns; it is this nonrival property of knowledge that is at the theoretical heart of models that produce endogenous growth from research. But to the extent that the knowledge or technology flow is embodied in a purchased piece of equipment, it may not produce a spillover, or, if it does, the spillover may take the form of a pricing or pecuniary externality rather than a technological one (Griliches, 1979).
Knowledge spillovers are much harder to measure than technology transfer, precisely because they tend to be disembodied. In previous work (Jaffe and Trajtenberg, 1996; Jaffe, Henderson and Trajtenberg, 1993), we have looked at citations made by patents to previous patents as a “window” on the process of knowledge flow. Jaffe, Henderson and Trajtenberg, 1993, showed that patent citations do appear to be somewhat localized geographically, implying that a region or country does utilize knowledge created within it somewhat more readily than do more remote regions. In Jaffe and Trajtenberg, 1996, we went further, looking in detail at citations from other countries’ patents to those of the U.S. We showed there that there is a clear time path to the diffusion of knowledge, in which domestic inventors’ citation probabilities are particularly high in the early years after an invention is made.


The rate at which knowledge diffuses outward from the institutional setting and geographic location in which it is created has important implications for the modeling of technological change and economic growth, and for science and technology policy. Models of endogenous economic growth, such as Romer (1990) or Grossman and Helpman (1991), typically treat knowledge as completely diffused within an economy, but implicitly or explicitly assume that knowledge does not diffuse across economies. In the policy arena, ultimate economic benefits are increasingly seen as the primary policy motivation for public support of scientific research. Obviously, the economic benefits to the U.S. economy of domestic research depend on the fruits of that research being more easily or more quickly harvested by domestic firms than by foreign firms. Thus for both modeling and policy-making purposes it is crucial to understand the institutional, geographic and temporal dimensions of the spread of newly created knowledge.


A test of the quality of labor market information regarding male and female employees suggests that employers have better information about male workers, which may explain the lower starting wages paid to women, although again the evidence is not strong.


This paper attempts to test whether information problems in labor markets help to explain why minority or female workers are sometimes paid less than equally-qualified white male workers. In particular, the relationship between starting wages, current performance, and race and sex is studied. OLS regressions of starting wages on current performance—which is measured some time after the beginning of employment—indicate that minority workers are paid lower starting wages than white workers with the same eventual performance, among both men and women. This could reflect taste discrimination.


i (3)
Columns (1) and (2) look at sex differences, grouping workers of each race and ethnicity, of the same sex, together, and including race and ethnicity dummy variables in the regression. For both the linear and log specifications, the ratio aoLS/aIV is considerably higher for men (.12-. 13) than for women (.02-.03). The lower estimate of a0LS for women implies that women’s starting wages are much more weakly related to their performance rating—which is measured after they have accumulated some time with the employer—than are men’s. On the other hand, the estimates of aIV are if anything higher for women, implying that their starting wages are at least as strongly related to expected productivity as are men’s. Together, this evidence suggests that employers have considerably worse information about new female employees than about new male employees. However, the standard errors of the estimated ratios of a0LS/aIV are relatively large, so the t-statistics for testing the null hypothesis of equality of these ratios for men and women are in the 1.4-1.5 range, implying that the evidence of a lower ratio for women is not statistically strong.


Is Labor Market Information Better for Some Demographic Groups?

The evidence from the preceding sections suggests that labor market information problems may partially account for the lower wages paid to minority workers, among both women and men. This subsection turns to the question of whether employer information is better about some demographic groups than others. If mismatches lower productivity, then worse information about women or minorities can lower that group’s average wage, providing another channel for labor market information to lead to lower wages for such groups. In addition, even if the type of test from the preceding section does not point to information problems as a source of unexplained wage differences between equally productive workers in different demographic groups, the test in this subsection can. This is potentially most pertinent to sex differences in wages, which cannot be explained as stemming from simple statistical discrimination, given that women’s performance ratings are on average at least as high as men’s.


Another way this result should manifest itself is in the changes in the estimated race gaps in wages for the alternative subsamples. In particular, when statistical discrimination is likely to be more important (i.e., for the probationary workers) the estimates should indicate that a larger fraction of estimated race gaps in wages are attributable to statistical discrimination. Table 5 reports these results, providing the estimated race coefficients corresponding to the specifications in Table 4 for the male-only sample.35 The results for both blacks and Hispanics are consistent with expectations. Simply using the point estimates, the estimated proportion of the wage gap attributable to statistical discrimination is higher for the probationary workers. Indeed, for the probationary workers the IV estimates of the race/ethnicity gaps in wages are non-negative, whereas for the non-probationary workers the IV estimates of these gaps are about three-fourths as large as the OLS estimates. In results for women not reported in the table, the same conclusion emerged.


Results for the non-probationary and probationary subsamples are reported in Table 4.

Columns (1) and (2) present results for men and women combined, using age and education as instruments, and columns (3) and (4) present results for men only, using age as the instrument. The proportion subject to probation is testing is .79, with the proportions very close among men and women.34 The results in Table 4 suggest that labor market information drives the differences between the OLS and IV estimates of a. For both samples, the ratio aoLS/aIV is considerably higher for the non-probationary sample for which initial labor market information should be better, although this ratio is estimated imprecisely for the smaller non-probationary subsamples.

Pages: 1 2 Next