Determinants of demand for biomedical sciences labor in the post-NIH-doubling environment, 2004–2010.

Standard errors robust to arbitrary heteroskedasticity are presented in parentheses below each coefficient estimate.Both models above were estimated using seven years of statistical (microdata-based) and macroeconomic time series variables, years ranging from 2004 through 2010. The outcome variable for the inverse demand model in column (1) is the first-difference of the log ratio of weighted average salary for biomedical sciences occupations, versus that for alternative S&E careers. The outcome variable for model (2) is the first-difference of the log ratio of biomedical sciences employment to total employment in S&E bachelor's-degree fields. All explanatory variables in equations (1) and (2) are also first-differenced log ratios, with NIH R&D obligations and biopharmaceutical industry R&D expenditures (BEA composite measure) divided by U.S. GDP. Both models employ as instrumental variables the second through fourth first-differenced lags of the logged ratios of biopharmaceutical industry R&D expenditures (PhRMA measure) to U.S. GDP.