Explaining the Modern Productivity Paradox Using Stochastic Frontier Analysis

<p dir="ltr">This dissertation tests alternative hypotheses for the “modern productivity paradox.'' This productivity paradox comes from the slowdown in measured productivity growth in the United States and other OECD countries in the last two decades despite the perception of fast paced innovation across technological fields during this time. The academic literature includes several competing explanations for this paradox. Understanding the nature of this disconnect can have serious implications for long-run growth and the policies needed to address the slowdown. Both chapters use the empirical framework of stochastic frontier analysis to decompose estimated productivity trends into alternative channels in order to assess the different explanations. <br></p><p dir="ltr">The first chapter uses two functional forms to incorporate patent stocks into SFA models. It shows that adding these measures significantly changes the scale and trend of estimated technological change at the frontier without affecting estimates of a country's efficiency, measured by a country's proximity to the frontier, as greatly. It further provides suggestive evidence that new patents are having a diminished impact on shifting out the technological frontier but that other unidentified sources of innovation may be driving further change. This supports recent findings in the literature that the quality of new patents may be declining and that the benefits of new innovation are becoming increasingly concentrated. It suggests that lagging producers are finding it more difficult to keep up with these innovations.</p><p dir="ltr">The second chapter then tests the specific hypothesis that research efforts are becoming less productive at producing ``new ideas." It uses both the methods laid out in previous literature (Bloom et al. 2020) and estimates idea production using stochastic frontier analysis to find falling research productivity for the United States as well as a broader sample of 44 countries from approximately 1980 to 2015. Across the models, SFA estimates that average research productivity in the U.S. is declining by about 4.3 percent per year with most of the decline coming from a fall in the frontier. By controlling for other channels, this suggests a semi-endogenous growth model effect. This average decline also applies to idea production across the full sample but is less steep.</p>