Image Source: PexelsIf the government is considering an industrial policy to subsidize a certain industry (for example, as the US is subsidizing domestic semiconductor manufacturing), what’s the best way to do it? I am not typically a fan of such subsidies, but it’s possible to make an economic case that for certain industries that provide social as well as private benefits (say, reducing carbon emissions or strengthening the domestic supply chain in produce important to national security), some form of government subsidy can be production. But what form?I ran across an example of some issues that can arise in “The Political Economy of Industrial Policy,” by Réka Juhász and Nathan Lane (Journal of Economic Perspectives, Fall 2024, 38:4, 27–54). They describe a situation in which the appropriate form of subsidy would not be for the firms that succeed in producing the desired product, but instead for firms that fail. The mechanism would be a “repayable advance” from the government, where successful firms need to repay the advance, but firms that fail to compete in the market do not need to repay. They write:

Consider a green industrial policy, where a public agency subsidizes risky projects that, if successful, would generate both private and social benefits. How should the agency design conditional subsidies? Meunier and Ponssard (2024) show that when firms and public agencies have symmetric information about the probability of a project’s success, rewarding success is optimal. However, under asymmetric information, where only the firm knows its probability of success, failure should be rewarded (!)—as it mitigates the windfall profit that arises when an agency subsidizes projects that would have received financing absent the subsidy. This insight speaks directly to the experience of the French Agency for Ecological Transition (ADEME), a public agency monitoring innovative activities for the energy transition funded by the Investments for the Future Programme. At the outset, ADEME used flat subsidies, but evidence of windfall profits quickly emerged in some projects. Therefore, the agency introduced “repayable advances,” which are subsidies that need to be paid back in the case of success—that is, they are subsidies for failure.

The article that Juhász and Lane cite, by Guy Meunier and  Jean-Pierre Ponssard, is titled “Green Industrial Policy, information asymmetry, and Repayable Advance” (Journal of Public Economic Theory, February 2024, e12668). The article will be math-heavy for the uninitiated reader. One danger of government subsidies for successful firms is that they reward those that would have already been successful in producing the desired good: in fact, they may reward firms for projects that were already financed and underway. In effect, they reward firms that would have already been profitable with even higher profits. In contrast, a repayable advance is aimed at encouraging more firms to try to produce the desired product, because the repayable (and forgivable) advance means that losses from failure are reduced.On the other side, a danger of repayable advances is that, well, they are paid to firms that don’t succeed. One can imagine a situation where a firm, knowing that it is somewhat protected against losing money, fails to make an optimal effort to succeed. In addition, politicians will find it easier to explain subsidizing success–which also offers photo opportunities!–than subsidizing failure.In the specific setting of the Meunier-Ponsard model, it turns out that the key difference between subsidizing the winning firms or the losing firms relies on what information is available. For example, consider a situation in which there are a number of possible firms that might be ready to invest in trying to produce the desired product, but neither the government nor the firm themselves have any clear idea of who is likely to be most successful. In this setting, it might make sense to promise additional rewards for the winners. But now consider an alternative situation, where the government does not know which firms are likely to succeed, but the firms themselves have a pretty good idea. In that situation, the firms that are likely to succeed are also more likely to be going ahead without the subsidy–and the government is only rewarding what would have happened anyway.At a broader level, the key point is that providing a broad overall justification for an industry subsidy is only a first step. The actual design of a specific policy can matter a great deal for the incentives that are created. Meunier and Ponsard write:

More generally our analysis justifies the importance of the empirical recommendations made by Rodrik (2014) for an agency in charge of monitoring a green policy. Let us review three of them briefly. Discipline: clarify ex ante objectives for the agency, build an evaluation protocol on what should or could be observed, making this an important aspect of the contracting process. Embeddedness: introduce reasonably simple conditional incentives along the life of the project but be aware of the risk of manipulation. Gaming: it plays an important role in our analysis, we formalize this issue and show how repayable advance may be used in some circumstances to mitigate its impact.

Of course, these guidelines for attempting to assure that industry subsidies are not wasteful or ineffective apply not just to green energy, but to any type of industrial subsidy. For those interested in learning more, Dani Rodrik’s (2014) paper is “Green Industrial Policy (Oxford Review of Economic Policy, 30(3), 469–491).More By This Author:An Economist Chews Over ThanksgivingRobots RisingThe Middle Income Trap