Vertical Integration and Competition
GAICIL Insights
Competition sets prices, and it also shapes how firms are built.1 It decides whether a firm buys its supplier or leaves it independent, and through that, who innovates. Vertical Integration and Competition, by Philippe Aghion, Rachel Griffith and Peter Howitt, belongs to a research program analyzing how competition affects aggregate innovative activity through firms' organization. The focus: whether firms integrate vertically with their suppliers.
The evidence points to a U-shaped relationship between competition and vertical integration. A moderate increase in product market competition reduces a producer's incentive to integrate, because it improves non-integrated suppliers' outside options and raises their incentive to innovate. Too much competition raises the producer's incentive to integrate, because it lets non-integrated suppliers capture most of the innovation surplus. Integration falls, then rises, as rivalry intensifies.
The result connects to earlier work. The authors previously found an inverted-U relationship between competition and innovation, explained by competition's composition effect on the distribution of technological gaps across industries. Organizational considerations may partly explain that inverted U. How firms organize, integrated or not, affects incentives to innovate. The famous inverted U may have an org chart hidden inside it.
A theoretical debate is engaged. Transaction Cost Economics, pioneered by Oliver Williamson, and Property Right Theory, developed by Sanford Grossman, Oliver Hart and John Moore, offer different explanations for integration. The results fit the property rights approach better, which emphasizes how ownership affects incentives to invest and innovate.
The authors describe the evidence as preliminary, presenting it as suggestive of a nonlinear relationship between competition and the propensity to integrate. Further work would deepen understanding of how organizational choices mediate competition's effects on innovation. Honest framing of early evidence is itself a contribution.
Bring the result forward to the present. Artificial intelligence supply chains span many layers, from chips and computing infrastructure to models and applications. Firms decide whether to integrate across these layers or rely on independent suppliers. Competition shapes those decisions, and the decisions in turn shape incentives to innovate. The structure of the AI stack is partly a by-product of how hard its firms compete.
Property rights thinking is especially pertinent. Ownership of key assets, such as data, models or computing capacity, affects who captures innovation returns. Independent suppliers with strong outside options have incentives to innovate. Vertical integration can change those incentives, sometimes reducing supplier innovation. An independent chip designer and an owned one may not try equally hard.
The U shape matters for AI competition policy. Moderate competition may encourage a vertically disaggregated industry with innovative independent suppliers. Very intense competition may push firms toward integration. Knowing where an industry sits on the curve helps assess policy changes' effects. Pushing competition ever higher may, past a point, produce the very integration regulators worry about.
Firm strategy is implicated too. AI firms must decide whether to build capabilities in-house or partner with suppliers. These choices depend on competitive conditions and affect all parties' innovation incentives.
Translated into policy, competition affects innovation through multiple channels, including firm organization. Evaluating vertical mergers and integration in AI supply chains requires attention to how they change incentives to innovate at each stage.
Data on organizational choices across industries with different competitive conditions are central, and richer data would allow sharper tests of competing theories of integration.
Read as a whole, competition shapes how firms organize, and organization shapes innovation. A U-shaped relationship between competition and vertical integration, consistent with property rights theory, helps explain why competition and innovation relate nonlinearly. As AI supply chains evolve, attention to these organizational dynamics will help foster innovation across every layer of the technology stack.
Here is the uncomfortable test for AI regulators: before concluding that vertical integration in AI reflects abuse, check where the industry sits on the competition curve. Integration can be a rational answer to very fierce competition, and the policy response should depend on which side of the U the market is on.
Introduction
No abstract was published with this article. The text below is the opening of the article, quoted verbatim.
This paper is part of a research program analyzing how competition affects aggregate innovative activity through its effects on firms’ organization. In previous work (Aghion et al., 2005a), we found an inverted-U shaped relationship between competition and innovation. Our explanation emphasized the “composition effect” of competition on the steady-state distribution of technological gaps across industries. Our focus here is on firms’ decisions whether or not to integrate vertically with their suppliers. We provide evidence of a U-shaped relationship between competition and vertical integration. Our explanation is based on the following idea: a moderate increase in product market competition will reduce a producer’s incentive to integrate by improving the outside options of her nonintegrated suppliers and hence raising their incentive to innovate. Too much competition will raise the producer’s incentive to integrate, however, by allowing nonintegrated suppliers to capture most of the innovation surplus. Finding a U-shaped relationship between competition and vertical integration sheds light on the debate over the “Transaction Cost Economics” (TCE) approach to vertical integration pioneered by Oliver Williamson (1975, 1985) versus the “Property Right Theory” (PRT) approach developed by Sanford Grossman and Oliver Hart (1986) and by Hart and John Moore (1990). According to the TCE approach, vertical integration is a way for contracting parties involved in a specific relationship to limit ex post bargaining inefficiencies due to holdup and thereby minimize the loss in ex ante investment that would result from it. This approach thus predicts a positive correlation between vertical integration and the degree of relation specificity. According to the PRT approach, the ownership structure will affect not so much the ex post bargaining efficiency as the relative bargaining powers of the (two) contracting parties, and therefore their relative ex ante investment incentives. Thus, while vertical integration should enhance both parties’ investments positively in the TCE approach by reducing the extent of ex post inefficiency, in the PRT approach ownership by one party, say the buyer, will enhance the buyer’s ex ante incentives at the expense of the seller’s, as it enhances the buyer’s bargaining power ex post at the expense of the seller’s. Thus, the TCE approach predicts that increased competition on the producer’s (or supplier’s) market, which reduces the overall degree of asset specificity, should therefore reduce the need for vertical integration in order to preserve ex ante investment incentives by either party. On the other hand, as we show below, the PRT approach allows the U-shaped relationship between vertical integration and competition that we find empirically.
Takeaway
“In this paper, we have provided some preliminary evidence to suggest that there is a nonlinear relationship between competition and the propensity of firms to integrate vertically. These results seem to be more consistent with the PRT approach of vertical integration than with the TCE approach and they also suggest that organizational considerations may partly explain the observed inverted-U relationship between competition and innovation.”1
Notes
- Philippe Aghion, Rachel Griffith & Peter Howitt, Vertical Integration and Competition, 96 Am. Econ. Rev. 97 (2006), https://doi.org/10.1257/000282806777211595. Takeaway source: Philippe Aghion, Rachel Griffith, Peter Howitt, Vertical Integration and Competition, Conclusion. ↩
Cite as: Philippe Aghion, Rachel Griffith & Peter Howitt, Vertical Integration and Competition, 96 Am. Econ. Rev. 97 (2006), https://doi.org/10.1257/000282806777211595
Header image: Library Display.jpg by Andrewmemela70, CC0, via Wikimedia Commons.
