We model innovation investments as real options and explore the implications of ambiguity—Knightian uncertainty—and risk for innovation decisions. Our model provides predictions for creating options to invest and options to wait. The ensuing empirical analysis uses a risk measure and a new outcome-independent measure of ambiguity. We find a consistently significant negative effect of ambiguity on R\&D, patents, and citations, supporting our theoretical predictions. We also find a significant positive effect of risk on R\&D, but the effect of risk on patents and citations is negative and significant. Ambiguity matters more for high-tech firms, consistent with intuition.

We study the relationship between patents and actual product innovation in the market, and how this relationship varies with firms’ market share. We use textual analysis to create a new data set that links patents to products of firms in the consumer goods sector. We find that patent filings are positively associated with subsequent product innovation by firms, but at least half of product innovation and growth comes from firms that never patent. We also find that market leaders use patents differently from followers. Market leaders have lower product innovation rates, though they rely on patents more. Patents of market leaders relate to higher future sales above and beyond their effect on product innovation, and these patents are associated with declining product introduction on the part of competitors, which is consistent with the notion that market leaders use their patents to limit competition. We then use a model to analyze the firms’ patenting and product innovation decisions. We show that the private value of a patent is particularly high for large firms as patents protect large market shares of existing products.

How do children affect women in science? We investigate this question using rich biographical data, linked with patents and publications, for 83,000 American scientists in 1956 at the height of the baby boom. Our analyses reveal a unique life-cycle pattern of productivity for mothers. While other scientists peak in their mid-thirties, mothers become more productive after age 35 and maintain high productivity in their 40s and 50s. Event studies show that the output of mothers increases after 15 years of marriage, while other scientists peak in the first 10 years. Differences in the timing of productivity have important implications for tenure and participation. Just 27% of mothers who are academic scientists get tenure, compared with 48% of fathers and 46% of women without children. Mothers face comparable tenure rates to other assistant professors for the first six years but fall behind afterwards, suggesting that they face higher standards of early productivity. Mothers who survive in science are extremely positively selected: Compared with other married women, mothers patent (publish) 2.5 (1.4) times more before the median age at marriage. Compared with men, female scientists are more educated, half as likely to marry, onethird as likely to have children, but half as likely to survive in science. Employment records indicate that a generation of baby boom mothers was lost to science.

Venture capitalists suggest that incumbent internet platforms create a kill zone around themselves, where any competing entrant is acquired quickly. Consequently, financing new startups becomes unprofitable. We construct a simple model that rationalizes the existence of a kill zone. The price at which an acquisition is done depends on the number of customers the entrant platform can attract if it remains independent, which in turn depends on the number of apps that have adapted to the platform. The prospect of a quick acquisition by the incumbent platform, however, reduces the app designers’ benefits from adaptation, making it harder for a technological superior entrant to acquire customers. This reduces the stand-alone price of the new entrant, decreasing the price at which they will be acquired, and thus reducing the incentives of VCs to finance their entry. We discuss the policy implications of this model.

Using unique City of Oakland data during COVID-19, we document that small business survival capabilities vary by firm size as a function of revenue resiliency, labor flexibility, and committed costs. Nonemployer businesses rely on low cost structures to survive 73% declines in own-store foot traffic. Microbusinesses (1-to-5 employees) depend on 14% greater revenue resiliency. Enterprises (6-to-50 employees) have twice-as-much labor flexibility, but face 11%-to-22% higher residual closure risk from committed costs. Finally, inconsistent with the spirit of ChettyFriedman-Hendren-Sterner (2020) and Granja-Makridis-Yannelis-Zwick (2020), PPP application success increased medium-run survival probability by 20.5%, but only for microbusinesses, arguing for size-targeting of policies.

Policymakers fear artificial intelligence (AI) will disrupt labor markets, especially for high-skilled workers. We investigate this concern using novel, task-specific data for security analysts. Exploiting variation in AI’s power across stocks, we show analysts with portfolios that are more exposed to AI are more likely to reallocate efforts to soft skills, shift coverage towards low AI stocks, and even leave the profession. Analyst departures disproportionately occur among highly accurate analysts, leaving for non-research jobs. Reallocating efforts toward tasks that rely on social skills improve consensus forecasts. However, increased exposure to AI reduces the novelty in analysts’ research which reduces compensation.