A Theory of Falling Growth and Rising Rents
GAICIL Insights
What if the efficiency of superstar firms is the cause of slow growth?1 A Theory of Falling Growth and Rising Rents, by Philippe Aghion, Antonin Bergeaud, Timo Boppart, Peter J. Klenow and Huiyu Li, offers an elegant answer to a central macroeconomic puzzle: US growth slowed while firm concentration rose. Market share shifted toward low labor share firms, while within-firm labor shares rose. The authors link these trends through one driving force: falling overhead costs of spanning multiple products, or a rising efficiency advantage of large firms.
The mechanism unfolds like a story in three acts. When overhead costs fall or leading firms' productivity edge widens, the most efficient firms, which enjoy higher markups, expand into new product lines. Concentration rises, the aggregate labor share falls and a temporary burst of growth follows. Eventually, because any firm is more likely to face highly productive competitors, within-firm markups fall for high and low productivity firms alike. Incentives to innovate decline, and growth slows. The boom contains the seeds of the slump.
The model, solved analytically for the balanced growth path and numerically for transition dynamics, can account for part of the US experience over the past 30 years: a productivity slowdown after a temporary burst, rising national concentration and opposing between-firm and within-firm movements in labor's share. Despite the temporary growth burst, welfare fell in the authors' calculation, modestly or more significantly depending on whether overhead costs fell or the productivity edge widened. A boom that ends with lower welfare is a strange kind of success.
The suspected trigger is familiar. Information technologies, linked to computer hardware and software, may have reduced overhead costs or widened productivity gaps. The authors cite evidence that bigger firms invest a higher share of sales in intangibles and IT, in the United States and in France respectively. Future work could document whether, how much and why these changes occurred.
Now substitute artificial intelligence for IT. AI can lower the costs of managing many products and markets and can widen the productivity advantage of firms deploying it effectively. If the mechanism applies, AI could produce a growth burst as leading firms expand, followed by a slowdown as markups and innovation incentives fall. The productivity statistics of the early AI years may flatter the long run.
The welfare result is the sharpest edge. A technology boosting growth temporarily can reduce welfare if it ultimately weakens incentives to innovate. That challenges the assumption that efficiency gains at leading firms necessarily benefit the economy over time. Competition dynamics and innovation incentives matter as much as initial productivity gains.
Policy has to be subtle. The mechanism runs through efficient firms expanding into new markets, itself a productive process. The trouble is that expansion reduces markups and innovation incentives for all firms over time. Policies sustaining incentives to innovate, through support for research, protection of intellectual property or measures keeping markets contestable, might help counteract the slowdown. Punishing efficiency would be the wrong lesson; protecting the incentive to keep inventing is the right one.
For AI, firm boundaries deserve monitoring. If AI lets leading firms expand across many product lines, concentration may rise and the competitive environment may shift. Tracking these changes and their effects on innovation is important for policy.
Parsimony is the method's strength. A single change in technology generates patterns in growth, concentration and labor shares that might otherwise look unrelated. Unified explanations like this are valuable for guiding policy.
Taken together, technological change affects growth through market structure and incentives. Lower overhead costs or wider productivity gaps can raise concentration, produce a temporary growth burst and ultimately reduce innovation and welfare. As AI transforms the economics of scale and scope, this theory offers a rigorous framework for anticipating long-run consequences and the importance of preserving incentives to innovate.
A hard question for AI optimists: an early productivity surge should be treated as the opening act. The real test of AI's economic value comes later, in whether challengers still find it worth inventing against the giants.
Introduction
Growth has fallen in the U.S. amid a rise in firm concentration. Market share has shifted to low labour share firms, while within-firm labour shares have actually risen. We propose a theory linking these trends in which the driving force is falling overhead costs of spanning multiple products or a rising efficiency advantage of large firms. In response, the most efficient firms (with higher markups) spread into new product lines, thereby increasing concentration and generating a temporary burst of growth. Eventually, due to greater competition from efficient firms, within-firm markups and incentives to innovate fall. Thus our simple model can generate qualitative patterns in line with the observed trends.
Takeaway
We developed a model of innovation-led growth with intrinsic firm heterogeneity. We solved analytically for the balanced growth path and its qualitative properties. We then solved for transition dynamics numerically to gauge whether a fall in overhead costs or an increase in the productivity edge of large firms could account for a portion of the U.S. growth experience over the past 30 years, in particular (i) a productivity growth slowdown (after a temporary burst in growth); (ii) rising concentration at the national level; and (iii) opposing between- and within-firm changes in labor’s share of income. According to our model, lower overhead costs and higher productivity at leading firms enticed such firms to expand their product boundaries. Because the most efficient firms enjoyed higher markups, when they expanded their reach into more markets this pushed down the aggregate labor share. And it led to a temporary surge of productivity growth.1
Within-firm markups eventually fell for both high and low productivity firms, as any given firm became more likely to face high productivity competitors.
This force ultimately reduced within-firm markups and dragged down innovation and growth. Despite the temporary burst of growth, welfare fell in our calculation (either modestly or more significantly, depending on whether overhead costs fell or the productivity edge of leading firms widened). Future work could try to document whether, how much, and why overhead costs fell or productivity gaps widened.
Information Technologies (IT) — linked to computer hardware and software — may have played a role. Crouzet and Eberly (2019) and Lashkari, Bauer and Boussard (2021) document that bigger firms invest a higher share of their sales in intangibles and IT, respectively.2 The former evidence is for U.S. firms and the latter for French firms.
Notes
- Philippe Aghion et al., A Theory of Falling Growth and Rising Rents, 90 Rev. Econ. Stud. 2675 (2023), https://doi.org/10.1093/restud/rdad016. Takeaway source: Conclusion, reproduced from a free author or repository copy (working-paper version). ↩
- Crouzet, Nicolas and Janice C Eberly, “Understanding weak capital investment: The role of market concentration and intangibles,” Proceedings of the 2018 Jackson Hole symposium, May 2019, pp. 87–148.. ↩
Cite as: Philippe Aghion et al., A Theory of Falling Growth and Rising Rents, 90 Rev. Econ. Stud. 2675 (2023), https://doi.org/10.1093/restud/rdad016
Header image: Aerial photo of Port of Tacoma.jpg by D Coetzee from Berkeley, CA, USA, CC0, via Wikimedia Commons.
