Large Firm Dynamics and the Business Cycle
GAICIL Insights
What if recessions need no villain?1 Large Firm Dynamics and the Business Cycle by Vasco M. Carvalho and Basile Grassi asks whether the usual suspects of aggregate shocks, monetary policy, oil prices, collective mood swings, are needed at all. The authors build a quantitative theory of aggregate fluctuations driven by firm-level disturbances alone, and show that a standard heterogeneous firm dynamics setup already contains a theory of the business cycle.
The premise is an empirical regularity everyone knows and few models respect: a small number of firms accounts for a substantial share of aggregate activity. When the biggest firms are big enough, their individual fortunes refuse to average out. The authors characterize analytically the law of motion of the firm size distribution, the aggregate state in this class of models, and show that aggregate output and productivity display persistence, volatility and time-varying second moments. A large fraction of aggregate dynamics can be rationalized by large firm dynamics.
Translate that into the age of artificial intelligence and the result becomes a forecast. AI development and deployment sit in a small number of very large firms. Their investment decisions, product launches and productivity gains move their own markets and aggregate output together. On this framework, the fortunes of a handful of AI leaders are a macroeconomic variable. A breakthrough at a leading firm could lift aggregate productivity. A stumble could show up in the business cycle. Earnings calls start to look like monetary policy meetings.
The practical consequence is a new monitoring task. Forecasting that ignores the firm size distribution misses an important source of fluctuations. Statistical agencies and central banks may need to track the dynamics of the largest firms as closely as aggregate indicators. In an economy where AI capabilities are concentrated, firm-level news becomes macro news.
The paper also exposes a blind spot in standard models. In the framework, a firm does not internalize its own effect on aggregate prices and factor costs. When the law of large numbers holds, treating firms as infinitesimal price takers is justified. The authors show that in a standard firm dynamics setting, large firms have a non-negligible effect on aggregates. Extending the analysis to firms that recognize their influence is an open question with obvious AI resonance: leading firms' demand for compute, energy and talent can move prices in entire factor markets.
A second extension concerns small firms. A full account of firm dynamics over the business cycle must also match small-firm dynamics, which requires leaving the frictionless world of the paper to study adjustment costs, credit constraints and other frictions. The authors note that Hopenhayn's framework is tractable enough to handle them. For AI this is pressing: cheaper AI tools may reshape small-firm dynamics, while credit constraints may limit small firms' ability to invest in adoption.
How should policy read all this? One reading treats size itself as a macroeconomic hazard. Another, closer to the paper's emphasis on firm dynamics, notices that large firms often become large through productivity growth, and that their growth fuels expansions as much as their setbacks fuel contractions. The first reading leads to shrinking the giants. The second leads to keeping the whole firm distribution dynamic, with entry, growth and exit working smoothly, so the economy enjoys large-firm success while new firms keep challenging incumbents. Breaking up a growth engine to calm the cycle would be a strange trade.
The method is the deeper lesson. By taking the firm size distribution seriously, the paper derives aggregate fluctuations from the behavior of individual firms. Macroeconomics built from the ground up sees things that representative-agent models cannot.
In the AI era, when a handful of firms may drive a disproportionate share of innovation, that micro-to-macro view becomes essential for understanding growth, volatility and the policies shaping both. The business cycle may increasingly have names, logos and product roadmaps. Policymakers would do well to learn them.
Introduction
Do large firm dynamics drive the business cycle? We answer this question by developing a quantitative theory of aggregate fluctuations caused by firm-level disturbances alone. We show that a standard heterogeneous firm dynamics setup already contains in it a theory of the business cycle, without appealing to aggregate shocks. We offer an analytical characterization of the law of motion of the aggregate state in this class of models, the firm size distribution, and show that aggregate output and productivity dynamics display: (i ) persistence, (ii ) volatility, and (iii ) time-varying second moments. We explore the key role of moments of the firm size distribution, and, in particular, the role of large firm dynamics, in shaping aggregate fluctuations, theoretically, quantitatively, and in the data. (JEL D21, D22, D24, E32, L11)
Takeaway
A small number of firms accounts for a substantial share of aggregate economic activity. This opens the possibility of doing away with aggregate shocks, instead tracing back the origins of aggregate fluctuations to large firm dynamics. We build a quantitative firm dynamics model in which we cast this hypothesis. The first part of our analysis characterizes, analytically, the law of motion of the firm size distribution and shows that the implied aggregate output and productivity dynamics are persistent, volatile and exhibit time-varying second moments. In the second part of the paper, we explore quantitatively and in the data, the role of the firm size distribution – and, in particular, that of large firm dynamics – in shaping aggregate fluctuations. Taken together, our results imply that a large fraction of aggregate dynamics can be rationalized by large firm dynamics. The results in this paper suggest at least two fruitful ways of extending our analysis. First, while we have intentionally focused our attention on large firm dynamics as drivers of aggregate fluctuations, a complete analysis of firm dynamics over the business cycle should also match those of small firms. This surely implies moving away from the frictionless environment presented here and understanding how small firm dynamics are distorted by adjustment costs, credit constraints and other frictions. The fact that Hopenhayn’s (1992) framework is tractable enough to handle such frictions, render our analysis easily extensible to such environments.1
Second, in our framework, a firm does not internalize its own effect on aggregate prices and factor costs. In other models of firms dynamics, the assumption that the law of large numbers holds, justifies thinking about firms as infinitesimal price takers. However, we have shown that, in a standard firm dynamics framework, large firms have a non-negligible effect on aggregates. Thus, the price taker assumption should be taken carefully. This points to generalizations of our setup that do away with this assumption and take on market structure and market power as further determinants of aggregate dynamics.
Notes
- Vasco M. Carvalho & Basile Grassi, Large Firm Dynamics and the Business Cycle, 109 Am. Econ. Rev. 1375 (2019), https://doi.org/10.1257/aer.20151317. Takeaway source: Conclusion, reproduced from a free author or repository copy. ↩
Cite as: Vasco M. Carvalho & Basile Grassi, Large Firm Dynamics and the Business Cycle, 109 Am. Econ. Rev. 1375 (2019), https://doi.org/10.1257/aer.20151317
Header image: Big Textile Factory (12671454823).jpg by darkday from Brisbane, Australia, CC BY 2.0, via Wikimedia Commons.
.jpg){kind=link}