Customer Accumulation, Firm Size, and Aggregate Productivity (Job Market Paper)

Abstract: Modern economies exhibit striking variation in firm size. While this fact is often attributed to productivity differences, recent literature has emphasized the role of customer accumulation. In this paper, I study the possibility of mismatch between these two objects: productive firms with few customers and unproductive firms with many. I argue that this is an important channel for understanding firm heterogeneity and has strong policy implications. To do so, I develop a dynamic macroeconomic model where firms endogenously differ in both productivity and customer access. Customer accumulation in the model is gradual and uncertain, naturally resulting in productivity-customer mismatch. I show that the firm's static optimality conditions distinguish customer access from productivity, and use this fact to calibrate the model to administrative data covering the universe of Canadian firms. The calibrated model features prominent mismatch, allowing it to match salient empirical patterns missed by alternative frameworks. This mismatch also leads to a surprising policy result. A tax on high revenue firms — clearly detrimental in standard models — can increase aggregate productivity and welfare. This is because the tax shifts production away from unproductive large firms, reducing misallocation by enough that the net effect is positive. I also find substantial welfare gains from subsidizing customer accumulation, highlighting an understudied margin for policy intervention. 

Growth, Spillovers, and Declining Entry, with Renato Zimmermann

Abstract: The rate of innovation in a market is closely related to its level of competition. As such, endogenous growth models increasingly incorporate rich interactions between innovation and market structure. In this paper, we develop a model of endogenous growth and oligopolistic competition that distinguishes between two margins of dynamic investment: i) innovation and ii) demand (branding, marketing, etc.). Demand investment increases the distance between incumbents and new firms, naturally deterring entry. This gives rise to negative spillovers, in stark contrast to the positive spillovers of innovation. If governments cannot fully distinguish true innovation from demand investment, this can dampen the welfare gains from R&D subsidies. Previous oligopolistic growth models have abstracted from the demand-innovation distinction in large part due to technical infeasibility (i.e., the curse of dimensionality). We show that recent computational methods can be used to overcome this barrier. Among other techniques, we represent the state and operator spaces using sparse matrices. This allows us to make use of modern linear algebra libraries and graphics processing units (GPUs), enabling computation even with an exceptionally large state space.

The Rise of Winner-Takes-All Markets

Abstract: Recent literature has found that leading firms have higher market shares, profits, and markups than they did 40 years ago. I add to this literature by presenting new evidence focused on the persistence of market leadership. I find that persistence is high and rising, meaning that firms who take hold of the high rewards associated with market leadership are increasingly difficult to dislodge. I then propose a dynamic model of oligopolistic competition and endogenous growth that captures this fact. The key novelty of the model is a “learning by doing” mechanism, whereby productivity is determined in part by past levels of production. I calibrate the model to match firm-level US data in two time periods: Early (1976-1990) and Late (2000-2020). I find that the learning by doing mechanism is much more prominent in the later time period, and this quantitatively explains 54% of the rise in leadership persistence.

Protectionist Tariffs With Third Country Effects, with Abdelrahman Amer and Daniel Trefler

Slides

Abstract: It is well known that import competition can have strong negative effects on domestic workers. Shielding workers from these shocks is among the primary stated objectives of tariffs. In practice, however, tariffs often fail to increase domestic production. For example, the 2018 US tariffs on China benefited countries such as Thailand and Vietnam far more than domestic producers. These "third country effects" cannot be explained by workhorse trade models. We build on recent methodological advancements to assess the effect of bilateral tariff changes on sectoral employment while incorporating third country effects. Our framework combines flexible substitution patterns with an otherwise standard model of labor market frictions. The effects of tariffs in this environment are heterogeneous across workers and products, providing a rich laboratory for counterfactual analyses. We use our model to quantify the efficacy of tariffs in protecting domestic workers and reversing job losses from free trade.

Income Dynamics Before, During, and After Entrepreneurship: Evidence and Theory, with Gueorgui Kambourov, Burhan Kuruscu, Sergio Ocampo, and Baxter Robinson

Abstract: New firms play a key role in shaping aggregate productivity, innovation, and wealth accumulation. Many entrepreneurs behind these new firms also work in other firms, participating in the labor market before, during, or after operating their business. How are their entrepreneurial and labor market outcomes connected? We document new facts about these outcomes using rich Canadian data covering the universe of workers and entrepreneurs from 2001-2019. We find high income workers are more likely to enter entrepreneurship, their businesses survive longer, and, on average, earn higher profits. However, after business exit, entrepreneurs return to their pre-business-entry levels of labor income, while workers without entrepreneurial activity experience income growth. This forgone growth is part of the cost of entrepreneurship. Finally, we present evidence of a positive but low correlation between labor and entrepreneurial ability among individuals who we observe as both workers and entrepreneurs. We develop a dynamic quantitative model consistent with these facts and use it to recover the distribution of abilities behind the selection into entrepreneurship and the labor and firm outcomes we observe.