Support for industrial policy is growing. Thankfully, no one is talking yet about the full-blown central planning popular in the 1930s. Instead, what’s now in vogue is a 1970s and ‘80s-era version of industrial policy allegedly meant to improve upon the market. In this incarnation, planning treats government as an actor much like any other market participant, but devoted to, among other things, directing R&D in ways supposedly superior to how it would be directed by market forces.
This idea failed enough when it was tried three or four decades ago that federal spending in R&D lost a lot of support. And there’s no cause for thinking that it will work better today. For one thing, this institution is the same government that is currently $23 trillion in debt. It’s also the same government that can’t run Amtrak competently or the postal service innovatively, and hasn’t been able to for decades. It’s engaged in useless wars. It doesn’t seem to learn from past mistakes.
In addition, the federal government is hardly a normal market participant. This is not new to even some of those who advocate for this position but part of the problem is that government employees who would direct these investments – no matter how brilliant and well-meaning they are – don’t spend their own money.
Even when private entrepreneurs aren’t barred from the field (which of course is a better situation than when government faces no competition), government officials remain largely shielded from the discipline of profits and losses that markets impose on people spending their own money. These officials will inevitably make many of their investment decisions based on political factors rather than on strictly economic considerations. In addition are the corruption and straight incompetence that, sadly, persist in government longer than in the private sector. And, obviously, government officials have no greater access to knowledge of market opportunities or of the future than do private entrepreneurs!
You don’t have to believe that market participants never make mistakes to understand that the government is nothing like private market participants. Indeed, the very same free-rider problems and externalities that allegedly justify government intervention loom much larger among political decision-makers. The ability to spend other people’s money, and to compel them to act as you command, destroys nearly all motivations for you to act only in ways that are mutually beneficial.
But maybe it is worth looking at past experiments in this area. Consider the detailed case studies by economists at the Brookings Institution. In 1991 Linda R. Cohen (with help from Roger G. Noll, Jeffrey S. Banks, Susan A. Edelman and William M. Pegram) wrote an entire book called The Technology Pork Barrel on the federal government as an investor in R&D for the sake of technological advancement.
What’s interesting about this book is that the authors are sympathetic to the usual arguments presented in favor of government intervention in R&D. They believe that the market in some cases fails to invest enough in R&D—an argument made across the political spectrum, including many libertarians. The authors also believe this failure keeps economic growth artificially low, and that government should correct this problem. And still, in the end, they conclude that people should be skeptical about federal government support for R&D.
As they explain, there are many types of R&D projects, and a wide range of methods to support these. Surveying them all, the authors find for government programs a mixed record. They did identify some successes in all categories and methods of support, but they note that failures are so prevalent that R&D was rightly regarded, at least when they wrote the book, as chiefly political pork.
For the purpose of their book, however, they focused on federal government projects to develop new technology mostly for the benefit of the private sector. They picked six major commercial R&D programs that had been in place for a while and assessed the results. They found that even programs that could be labelled as successful suffered from unsustainable annual budgets, which made them impossible to sustain. Meanwhile, most programs they studied were failures made worse by the fact that they continued to receive support long past the time they should have been terminated. And in every case, politics made things worse.
The authors conclude from their detailed analysis that “The case studies obviously justify skepticism about the wisdom of government programs that seek to bring new technologies to commercial practice. But the case provides far more insight than that. They [the cases studied] identify how and why the government programs go wrong, and hence the problems in the federal decision-making process that needs to be solved if performances are to be improved.”
These scholars then offered suggestions for reform. But they warned that “ We are not sanguine about the prospect that this or any other recommendations about structuring the process by which the decisions are made will dramatically raise the batting average of R&D commercialization.”
They weren’t the first (or the last) to warn against the soundness of federal government’s R&D projects. That said, economists usually feel differently about federal funding for basic research—roughly a quarter of total federal R&D funding. Chris Edwards at the Cato Institute, for instance, writes in 2017:
“The federal government spent $147 billion on research and development in 2016, including $77 billion on defense and $70 billion on nondefense. Federal R&D spending has risen in recent decades on a constant-dollar basis, but has dipped as a share of gross domestic product. The AAAS has the data here.
How much should the federal government spend on R&D? AAAS data show that 23 percent of federal spending is for “basic” research, 25 percent is for “applied” research, and 52 percent is for “development.” Most economists would support the basic part, but would be more skeptical of the applied and development parts because the private sector handles those activities.”
Yet, even in the case of funding for basic research, there are reasons to be careful. For one thing, that economists are generally more sympathetic to the federal government supporting basic R&D and yet the majority of funds go toward applied and development R&D is, perhaps, itself indicative of the special-interest nature of federal R&D policy.
What’s more, the late William Niskanen, economist and former Chairman of the Cato Institute looked at this issue in his 1997 paper “R&D and Economic Growth: Cautionary Thoughts.” He argued that the idea that R&D drives economic growth isn’t as sound as it seems on the surface. What we know, Niskanen writes, is that there is “a strong relationship between real expenditures for research and development (R&D) and the level of national output—but little relationship with the rate of economic growth. This record is more consistent with a hypothesis that R&D is an income-elastic consumption good, something that rich people and rich nations do, rather than an investment that will increase future economic growth.”
But let’s assume, for argument’s sake, that R&D will increase future economic growth. Would this fact imply that the federal government should invest in R&D? To answer this question, Niskanen looked at the implicit assumptions behind the belief that government should indeed finance R&D. The argument, which he calls Bacon’s Chain, is this:
· government financing is necessary to provide the adequate level of
· basic research, which is necessary to provide the scientific foundation for
· advanced technology, which accounts for a large part of economic growth.
Working backwards from the last bullet point to the first, Niskanen exposed the analytic and empirical problems in each. The exercise is quite interesting. He first warns that most economists who study economic growth tend to call “technology” pretty much anything they don’t understand. That, he says, makes it hard to trust any asserted relationship between technology and economic growth. That is a big hole in the last bullet point.
One of the most interesting aspects of Niskanen’s paper is his finding that “In the short-term, …, most technological innovation is based on other advances in technology, with little contribution from recent advances in basic research. ” In other words, what may be ever more important than investment in basic research is the existence of other innovations to build on. He adds that “The long-term contribution of basic research is substantially higher, but with lags of 20 to 30 years. And basic research is quite dependent on both industrial support and recent technological innovations and challenges.”
Finally, Niskanen turned to the first bullet point: the need for government to augment a lack of private-sector investment. Here Niskanen explains that that while economists have treated this claim as self-evident, it doesn’t hold empirically. For instance, he makes the case that basic science is not a pure public good, since in the absence of government spending there was (“private finance was the largest source of support [of basic research] until the 1950s,” he writes, and that firms’ profits depend on their own investments in basic science), is, and still would be a lot of private funding of basic science.
Further, even if private investments remain too small, getting government involved might still be undesirable. First, like Cohen and Noll, Niskanen notes that government incentives and information limitations are often a big problem. Another risk is the fact that in the presence of federal funding, firms try first to get the money from the government, rather than spend their own money as they should or would otherwise. Thankfully, at the time of his writing, the crowding out didn’t appear to be a big problem; mostly a majority of the outlays for R&D was for defense and space technology. Finally, funding from the government doesn’t fall from the sky; it comes from higher taxes, which have their own negative consequences.
Niskanen concludes that “Each of the links in the bacon’s chain is plausible but empirically weak. Neither the theory nor the available evidence provides a satisfactory guide to federal science policy.” Others have questioned the Bacon’s Chain arguments more recently. Niskanen then looks at the relationship between productivity growth, real federal R&D, and total R&D outlays. Like with the Bacon’s chain argument, he comes out underwhelmed by the evidence.
His conclusion is that when it comes to civilian R&D “private organizations have surely better information and incentives than the government that most contributes to growth. What’s more, underinvestment from the private sector and sub-optimal level of R&D doesn’t imply that the government can improve on that outcome.” That, he explains, is because the government is unduly influenced by special interests and R&D investments are no exception. He concludes that for all of these reasons the spending and the allocation should be left to the private sector. He is, however, open to using the tax code to create incentives to increase private spending in R&D, but again those tax breaks ought to be generic as to not affect the allocation of resources.
Looking at the federal government today tells me that the problems surrounding R&D programs in the past continue today, and will continue tomorrow, because they are simply a consequence of the normal functioning of government. It is hard to wish these problems away, even in the face of the private sector’s “imperfections.” Those arguing for more funding in R&D should proceed with caution.