8.4County Density Quotients and Labor Productivity (Model 3)

This section discusses the results of density quotient regressions, as defined in section 7.3. The unit of observation is still the county, and I want to analyze the effects of the density of employment in the two types of industry on county labor productivity. Elasticities estimation results appear in Table 8.3 [regressions (8.11) and (8.12)], which consider the density quotients independently. Indeed, strong multicollinearity rendered the simultaneous use of these two measures meaningless. Furthermore, after a mathematical simplification similar to the one used in deriving equation 8.2, the ratio of the density quotients ends up being equal to the ratio of the location quotients. Thus, the estimates of regressions using either ratio should be the same.

The coefficient for the IT employment density quotient is small but significantly positive at the 0.01 level as shown in regression (8.11). The coefficient for this measure of density is however not significant regarding non-IT employment, as indicated in regression (8.12). These results mean that, at the county level, the density of IT employment significantly affects labor productivity in a positive way, whereas the density of traditional employment does not seem to play any role regarding labor productivity differences. It follows from regression (8.11) that if the number of IT employees per acre doubles in a given county, labor productivity may increase by half a percentage point in this county.