Industrial Policy Is a Very Old, New Idea

By Matthew D. Mitchell and Adam Thierer

There is a hot new idea bouncing around the corridors of power—or at least its boosters think it is hot and new. It’s actually a very old and very bad idea.

It takes different forms, depending on who’s talking about it. On the political left, for instance, it involves the federal government selectively throwing its weight behind green industry. On the right, the federal government does the same, but to promote domestic manufacturing. And in its state-based variant, state governments promote sexy industries such as film production, technology or even paint (the latter being Ohio’s definition of sexy).

Depending on where you sit on the political spectrum, this idea has different names. Some call it “industrial policy.” Others call it “economic nationalism.” Whatever you call it, the defining characteristic is the same: selective discrimination in favor of some industries or firms over others.

As we’ve noted, this “new” idea has actually been around a long time. It was not new in the 1980s and 1990s when policymakers thought industrial policy was the way to counter the economic rise of Japan. Nor was it new in 1936 when Mississippi’s governor Hugh White announced that Mississippi would jump-start economic development through manufacturing subsidies. (State per capita GDP figures suggest that his plan has yet to yield its promised results.) It was not new in the 1830s when several states bankrupted themselves subsidizing canals, turnpikes, railroads and banks. Nor, alas, was it new in 1791 when Alexander Hamilton proposed selective subsidies in his “Report on the Subject of Manufactures.”

Indeed, the idea of selectively helping special interests in the name of economic growth goes back much further than the founding of the American republic—to the ancient world. If anything, the American novelty is that industrial policy has tended to be limited. At times, it has been limited by the democratic process. Hamilton’s subsidies were rejected by Congress (even as they accepted his tariffs), and Donald Trump’s new nationalism was not as popular as many thought it would be. At other times, this sort of government favoritism has been limited by institutional constraints. For instance, the federal constitution outlaws state regulatory privileges, and most states have anti-aid provisions that—at least in their plain reading—seem to make subsidies illegal.

The Costs and Benefits of Industrial Policy

But despite a long, checkered history, there has recently been an explosion of bipartisan advocacy for expansive national industrial policy, and this tracks with a widespread continued interest in economic development efforts at the state level. Many policymakers—following the advice of a wide variety of political advisers and industry representatives—suggest that governments can calibrate economic development through carefully planned policy interventions. The goal is usually to develop specific technological capabilities or sectors through generous state support. Although not phrased as such in the United States, the aim is to essentially create what Chinese and European policymakers unabashedly refer to as “national champions”—firms that will become global leaders in “strategic” sectors, and which will then help grow the base of jobs and technological capabilities that policymakers desire.

The political benefits of industrial policy are easy enough to spot. Politicians can stand in front of a subsidized factory and proclaim that it would not be there but for their intervention. They can point to the new capital and the new jobs created. They can even proclaim that these benefits will ripple out to the broader economy, thanks to the vaunted multiplier effect.

Targeted privilege has costs, however. Indeed, it often fails altogether. The history of industrial policy and state-based economic development efforts is littered with a long string of costly boondoggles that ultimately did little to benefit growth, jobs, competitive advantage or consumer welfare. Proponents will point to a few “wins” without mentioning the many “losses,” let alone providing a fuller account of all the costs—both direct and indirect—of their planning and spending efforts.

The primary problem with the targeted approach to economic development is that it depends on the use of government-granted privileges—discriminatory tax or regulatory relief, cash subsidies, loans and loan guarantees, in-kind donations or the provision of other valuable goods and services. In other words, governments are trying to pick winners.

Advocates of the targeted approach insist they are not doing this, but in practice that is precisely what the targeted approach is about. Because targeting operates through selective privileges for certain firms, industries or regions, it comes at the expense of others—taxpayers who foot the bill, competitors who lose the subsidy race or consumers who are left with lower-quality products and services provided by sheltered firms.

And even when there are “winners,” they usually don’t really win anything of lasting importance. In many cases, states or localities end up getting stuck with the price tag for failed development efforts. Or, the nation absorbs the costs associated with expensive industrial policy bets that end up being boondoggles, such as solar energy firm, Solyndra.

To better understand the costs and benefits of industrial policy, consider the case of Wisconsin’s Foxconn subsidy, analyzed at length in this Mercatus study. In the summer of 2017, in exchange for up to $3.6 billion in direct payments and other privileges from Wisconsin taxpayers, Foxconn Technology Group promised to make a $10 billion, 13,000 employee investment in Southeast Wisconsin. At this point, it seems clear that the company has no intention of following through with this promise and is likely to develop a much smaller facility.

But even under the original terms, the deal was bad. You wouldn’t know that if you listened to Foxconn or the politicians throwing money at it. Foxconn commissioned a study predicting that, due to multiplier effects, the $10 billion investment would add as much as $62.4 billion to state GDP over the course of 15 years. That is quite a rate of return.

Unfortunately, it doesn’t stand up to scrutiny. As shown in the Mercatus study, this extraordinary rate of return is only realized if one makes a number of heroic assumptions:

1. First, don’t think about the costs. While the economic benefits of the new plant were widely touted, Foxconn’s boosters said nothing of the economic effects of the taxes needed to fund the subsidy. In fact, decades of economic research suggest that taxation is costly. By the best estimate, the higher taxes needed to fund the Foxconn subsidy would entail an economic cost (what economists call “deadweight loss”) of around $29 billion.

2. Second, assume the subsidy made all the difference. We can’t know the thinking of Foxconn executives for certain, but it’s worth noting that the vast majority of subsidies (75%-98%) do not tip firm location decisions. It’s also worth noting that, in choosing Wisconsin, Foxconn eschewed a larger subsidy offered by Michigan. Also of note: The site was home to the U.S. Speaker of the House and the White House chief of staff, and also happened to be the preferred location of the president of the United States. These facts suggest that the subsidy was probably not decisive in encouraging Foxconn to locate in Wisconsin. But even if there were a 50% chance that the subsidy actually tipped the balance in the Foxconn case, then the expected gross value of the subsidy was $31 billion—half of the $62 billion rather than all of it. This, combined with the $29 billion cost of taxation implies a 15-year net benefit from the subsidy of $2 billion, not $62 billion. Moreover, if the subsidy had been decisive with only a 25% probability, then the net value of the subsidy was negative $13 billion (the calculation also turns negative if the costs of taxation end up being greater).

3. Third, ignore all the other costs. We have some (uncertain) idea of how to measure spending and taxing multipliers and the likelihood of a subsidy inducing a certain location decision. There are, however, other costs of a subsidy that are much harder—in some cases impossible—to measure. For example, subsidies encourage unwise investments by allowing firms to ignore real world costs and by protecting them from the discipline of competition. They also tend to reward existing firms with established political connections rather than new and innovative firms. Worse still, privileges encourage firms to invest time and effort wooing politicians. This process, known as rent-seeking, wastes valuable resources that could instead be devoted to pleasing customers or improving operations. Though these sorts of costs are hard to measure, we do know that rent-seeking societies tend to be low-income societies.

At the national level, past industrial policy fiascos present a similar story. A 1991 Brookings Institution book entitled The Technology Pork Barrel surveyed the U.S. government’s track record in promoting high-tech initiatives in six major commercial R&D programs, including supersonic aircraft and synthetic fuels. “The case studies obviously justify skepticism about the wisdom of government programs that seek to bring new technologies to commercial practice,” the editors noted, after documenting how “American political institutions introduce predictable, systematic biases into R&D programs so that, on balance, government projects will be susceptible to performance underruns and cost overruns.”

Americans can at least be thankful that we haven’t gotten caught up in the sort of foolhardy schemes that European countries have backed recently to expand their digital tech base. Europe’s efforts to produce home-grown tech champions have instead mostly resulted in a long string of costly failures. European governments have tried replicating American success in Internet access, search engines, GPS, cloud services, and even Netflix and Airbnb. The efforts yielded endless delays, bloated budgets and almost instant obsolescence once services launched, if they did at all.

A Better Way

Europe’s experience helps explain why the U.S. didn’t need massive industrial policy to develop national champions in the digital marketplace. The Department of Defense did have a hand in helping lay the early groundwork for what eventually became today’s internet. But the reason digital innovation exploded during the mid-1990s in the U.S. had nothing to do with grandiose industrial planning schemes.

Instead, the U.S. was pursuing the superior general approach to economic development, creating an environment conducive to economic growth without offering targeted assistance to particular firms or industries. The key was an innovation culture that fostered entrepreneurialism and risk-taking by giving creative minds the ability to explore bold new ideas in a relatively unfettered fashion.

The generalized approach the U.S. adopted for computing, the internet and digital technology has sometimes been called “permissionless innovation.” The Clinton administration and a Republican-led Congress worked together throughout the mid-1990s to lay the policy groundwork for a modern technological revolution. They allowed the commercial development of the internet, rejected analog-era telecom and media regulations, discouraged discriminatory taxation of online services, and left dispute resolution to voluntary bodies and the courts.

The Clinton administration went so far as to release a “Framework for Global Electronic Commerce” recommending that “governments should avoid undue restrictions on electronic commerce” and “parties should be able to enter into legitimate agreements to buy and sell products and services across the Internet with minimal government involvement or intervention.”

As a result of this generalized policy approach, many U.S. technology companies became household names across the globe. Firms, investors and enormously talented immigrants flocked to American shores to take advantage of the opportunity afforded by the simple but effective approach to tech policy.

The building blocks of the general approach—a mix of broadly applicable tax, spending, regulatory and legal rules—are often rejected because they seem less exciting than targeting specific companies or industries for help. Pundits and policymakers are fond of using machine-like metaphors to suggest they can “fine-tune” innovation or “dial-in” economic development according to a precise formula they believe they have concocted. They also savor the attention that goes along with ribbon-cutting ceremonies and the big headlines often generated by political targeting efforts.

But the boring general model of development reaps bigger longer-term rewards, even if it isn’t as flashy in the short-term. The Nobel economist F. A. Hayek once suggested that policymakers should aim to “cultivate a growth by providing the appropriate environment, in the manner in which the gardener does this for his plants.” That’s the essence of the generalized approach. By contrast, to extend the metaphor, with targeted efforts the gardener is fertilizing some plants by composting others.

The empirical case for cultivating growth through generalized economic freedom—low taxes, minimal regulation and protection of property—is overwhelming. There have now been nearly 200 peer-reviewed studies assessing the link between economic freedom and prosperity, and the vast majority find a positive association.

As the chart above shows, there are a host of options available to the government. The predicament for policymakers is that, while it is wiser to focus on the generalized approaches, the temptation will remain strong to jump to targeted gambles that may grab headlines but are far more risky and costly.

It’s true that if governments roll the proverbial industrial policy dice enough times, a few bets are bound to pay off. As Deirdre N. McCloskey and Alberto Mingardi note in their new book, The Myth of the Entrepreneurial State, “in view of the gigantic increase in public spending since 1900 it would be strange indeed if none of the dollars didn’t finance something technologically relevant.” While we can recognize these payoffs, we must also acknowledge that we will never know what private initiatives didn’t happen because resources were siphoned off to pick these winners. Moreover, this doesn’t excuse the fact that governments are gambling with taxpayer dollars and making a lot of manifestly bad bets on risky ventures. These bets look even worse when we consider the empirically supported alternative: generalized economic freedom.