Productivity, Innovation, and Entrepreneurship Meeting

March 16, 2012
Nick Bloom of Stanford University and Josh Lerner of the Harvard Business School, Organizers

Erik Brynjolfsson, MIT and NBER, and Heekyung H. Kim, MIT

CEO Pay and Information Technology

Compensation for CEOs and other top executives has increased dramatically in recent decades, drawing increasing scrutiny from policymakers, researchers, and the broader public. Brynjolfsson and Kim find that information technology (IT) intensity strongly predicts compensation of top executives. Their examination of panel data from 2,507 publicly traded firms over 15 years controls for other types of capital, number of employees, market capitalization, median worker wages, industry turbulence, firm or industry fixed effects, and other factors. Their interpretation of this result builds on earlier work that found a correlation between CEO pay and firm size. They hypothesize that IT increases the information available to the top executives for decisionmaking, magnifies their ability to propagate instructions throughout the firm, and improves the monitoring and enforcement of those instructions. When a top executive's instructions are implemented with higher fidelity, the fortunes of the firm will more closely mirror her performance. From the perspective of the top executive, this increases "effective size" of the firm that she controls and thus her marginal productivity. In turn, in an efficient market, this will increase overall executive compensation.


Ajay K. Agrawal, University of Toronto and NBER; Iain M. Cockburn, Boston University and NBER; Alberto Galasso, University of Toronto; and Alexander Oettl, Georgia Institute of Technology

Why Are Some Regions More Innovative Than Others? The Role of Firm Size Diversity

Large labs may spawn spin-outs caused by innovations deemed unrelated to the firm's overall business. Small labs generate demand for specialized services that lower entry costs for others. Agrawal, Cockburn, Galasso, and Oettl develop a theoretical framework to study the interplay of these two localized externalities and their impact on regional innovation. They examine MSA-level patent data during the period 1975-2000 and find that innovation output is higher where large and small labs coexist. The finding is robust to across-region as well as within-region analysis, IV analysis, and the effect is stronger in certain subsamples consistent with their explanation but not the plausible alternatives.


Ufuk Akcigit, University of Pennsylvania and NBER, and William R. Kerr, Harvard University and NBER

Growth through Heterogeneous Innovations

Akcigit and Kerr study how exploration versus exploitation innovations affect economic growth. They use a tractable endogenous growth framework that contains multiple innovation sizes, multi-product firms, and entry/exit. Firms invest in exploration R&D to acquire new product lines and exploitation R&D to improve their existing product lines. The authores model and show empirically that exploration R&D does not scale as strongly with firm size as exploitation R&D. The resulting framework conforms to many regularities regarding innovation and growth differences across the firm size distribution. They also incorporate patent citations into the theoretical framework. The framework generates a simple test using patent citations that indicates that entrants and small firms have relatively higher growth spillover effects.

Leonid Kogan, MIT and NBER; Dimitris Papanikolaou, Northwestern University; Amit Seru, University of Chicago and NBER; and Noah Stoffman, Indiana University

Technological Innovation, Resource Allocation, and Growth

Kogan, Papanikolaou, Seru,and Stoffman explore the role of technological innovation as a source of economic growth by constructing direct measures of innovation at the firm level. They combine patent data for U.S. firms from 1926 to 2010 with the stock market response to news about patents to assess the economic importance of each innovation. Their innovation measure predicts productivity and output at the firm, industry, and aggregate level. Furthermore, capital and labor flow away from non-innovating firms towards innovating firms within an industry. There exists a similar, though weaker, pattern across industries. Cross-industry differences in technological innovation are strongly related to subsequent differences in industry output growth.


Serguey Braguinsky, Carnegie Mellon University; Sergey V. Mityakov, Clemson University; and Andrei Liskovich, John F. Kennedy School of Government

Direct Estimation of Hidden Earnings: Evidence from Administrative Data

Braguinsky, Mityakov, and Liskovich estimate hidden earnings by matching car registries to employers' records of paid earnings for a panel of individuals and households in Moscow. The identification strategy is based on the idea that reported earnings may be falsified, but car registries are accurate. Hidden earnings comprise over 75 percent of actual earnings of the large majority of car owners, at least twice as high as estimated in previous studies using less direct methods. There is also a lot of heterogeneity across employers. Foreign-owned firms, large firms, and state-owned firms in capital-intensive industries report earnings more transparently than do small firms and firms in labor-intensive industries, where actual earnings may be more than five times higher than reported earnings. Differentials of similar magnitude are found in public services, especially among educators. These findings shed new light on the perceived links between firm ownership, size, and productivity in countries with large hidden economies.


Fabian Waldinger, University of Warwick

Bombs, Brains, and Science - The Role of Human and Physical Capital for the Creation of Scientific Knowledge

Waldinger analyzes the effects of human capital (HC) and physical capital (PC) on the productivity of science departments. To address the endogeneity of input choices, he uses two extensive but temporary shocks: as the HC shock, he uses the dismissal of mostly Jewish scientists in Nazi German; as the PC shock, he uses the destruction of facilities by Allied bombings during WWII. In the short run, a 10 percent shock to HC lowered departmental productivity by about 0.21sd. A 10 percent shock to PC lowered departmental productivity by about 0.05sd in the short run. While the HC shock persisted until the end of the sample period (1980), departments experiencing a PC shock recovered very quickly (by 1961). Additional results show that the dismissal of "star scientists" was particularly detrimental, and that a fall in the quality of hires was an important mechanism for the persistence of the HC shock.