This is the first in a series of posts serial-blogging a draft chapter I’ve written on privatization and competition policy for a forthcoming Stanford University Press book on Competition and the Role of the State. What follows represents the current draft, though sans footnotes for the blog format. Sorry, I don’t have it posted on SSRN yet, so I can’t link to the full document, but drop me a line if you’d like the Word document.
Why a discussion on privatization, in a volume on competition policy? Because the scope of government and the role of the state are central to broader issues of competition. Privatization of monopolies without sufficient antitrust protection has been a recurring complaint among critics of privatization. So, on the other side, has been anti-competitive behavior by state-owned enterprises that often are not subject to antitrust at all. More theoretically, antitrust law may focus on certain problems and choose to ignore others as unrealistic because of an assumption that firms behave as profit-maximizers—but, given the mass of public choice literature exploring what governments supposedly maximize, as well as empirical evidence that privatization often increases productivity and profitability, this may be a bad assumption when firms are owned (or even heavily regulated) by government.
This chapter, therefore, seeks to explore the links between privatization and competition policy. First, I explain why privatization is even relevant at all—a question economists paid surprisingly little attention to before a quarter-century ago; I survey theoretical models and empirical results, discuss the relevance (if any) of theory to policy, and show how privatization can be a catalyst for various forms of social change. Many policies are complementary to privatization, in the sense that the design of those policies systematically affects the effectiveness of privatization; two of the most important complementary policy areas are corporate governance and competition policy. Moreover, since failures in both of these areas lead to distortions relative to the imaginary “perfectly competitive” outcome, good corporate governance and good competition policy are to some extent substitutes.
Privatization and Its Possible Irrelevance
Privatization has a long history. Peter Drucker famously advocated “reprivatization” in 1969; but the term “privatize” and its derivatives were used even earlier, to refer to the selloff of government assets in West Germany in the 1950s and in Nazi Germany in the 1930s and 1940s—and even before that, we had the term “denationalization” as far back as the 1920s. And asset sales (or giveaways) and contracting out somehow managed to exist even before there was a word for them; vast public lands were privatized in the United States under the preemption and homestead acts of the mid-19th and early 20th centuries, Jeremy Bentham envisioned privately managed prisons in the late 18th century, and we can trace contracting out—in tax collection and a vast array of other public services—back to ancient Greece and the Roman Republic.
The history of the theory of privatization, on the other hand, is quite short. For much of its history, privatization was “a policy in search of a rationale.” Early debates over privatization were far from rigorous, and it’s probably fair to say that the era of economically sophisticated discussion of privatization began in 1987, when David Sappington and Joseph Stiglitz pointed out that, in a simple model, whether assets are owned publicly or privately is irrelevant.
To simplify slightly: suppose the government is considering whether it should own an asset. The asset could produce a quantity Q, whose valuation is V(Q). This valuation could encompass any objectives, including distributional ones. The government could induce any private provider to produce the optimal quantity by simply paying it a price P(Q)=V(Q). (Of course, many other contracts will do just as well, for instance, “Produce the optimal quantity Q* in exchange for a price that guarantees you zero economic profits, or we boil you in oil”; but the Sappington-Stiglitz proposal has the advantage that the government doesn’t need to know costs.)
The private provider would thus get paid the entire social benefit of his production; but the government can extract this rent by first auctioning off the right to produce among potential non-colluding risk-neutral suppliers with symmetric beliefs about the least-cost technology. (This idea had already been anticipated in 1979 by Martin Loeb and Wesley Magat, though not in the context of privatization.)
One might add that a government manager can be incentivized the same way with a wage W(Q)=V(Q), so all ownership modes are equivalent. Generalizing still further, Q could be multi-dimensional and indicate not just quantity but any attribute of how the business is run. Any rules that a government manager is required to follow can be worked into a contract, or can be written into regulations; government enterprise, contracting out, and a regulated market are thus potentially equivalent in every way.
From this perspective, much of the conventional wisdom about privatization appears, if not wrong, at least undertheorized. Private firms may have better monitoring by the capital market—but capital market monitoring has problems of its own, and why can’t the government retain any advantages of such monitoring by only owning partial shares in firms? Private firms may have a harder budget constraint, but why can’t the government shut down failing agencies or, alternatively—as recent events have reminded us—bail out failing private firms? Governments might undermine the investment incentives of public agencies by redirecting their profits to other uses—but can’t shareholders do the same? Public agencies may have goals that are hard to specify—but isn’t this also a problem when the government regulates private firms? Governments may be subject to interest-group lobbying—but can’t that lobbying also affect contracts or regulation of private firms? Governments are better positioned to accomplish social goals—but can’t they also accomplish those goals through contract or regulation?
One can easily apply these points to various concrete examples. For instance, why should the Post Office be public, rather than contracted out to FedEx or UPS, or left to a regulated private market? The Post Office takes all comers and charges uniform postal rates, but that could be written into regulations as a condition of carrying on postal business. The Post Office loses money, but FedEx could charge the same to its customers that the Post Office charges now and collect its deficit in contract fees from the government; or the government could regulate prices and subsidize entrants in a private market. The quality and speed of postal delivery can be easily checked, and made the subject of rewards and punishments, by sending test letters and packages. On this last point, one could make similar arguments about the Transportation Security Administration, the nationalization of which was rushed through post-9/11 with very little appreciation of Sappington-Stiglitz.
Of course, the proper treatment of any equivalence theorem is as a checklist to see how differences emerge when the assumptions of the theorem don’t hold. One issue is that the rent-extraction auction might not work well because of, say, collusion, asymmetric beliefs, risk aversion, or the winner’s curse. To the extent that the winning bidder retains some rents, P(Q)=V(Q) privatization potentially implicates distributional concerns—and requires raising government revenue, with the associated deadweight loss of taxation.
But the literature has mostly focused on the most obvious line of inquiry: V(Q) might practically be indescribable because of the sheer number of possible contingencies (really, V(Q) is V(Q,ϑ), where ϑ, a parameter describing the state of the world, can just take too many values); V(Q,ϑ) might even be unknown until ϑ occurs (which isn’t until after the contracting stage); the level of production Q might be unverifiable; the contract P(Q)=V(Q) might be imperfectly enforceable; and so on.
In short, the P(Q)=V(Q) contract is trickier than it might at first seem. This idea has a name: the theory of incomplete contracts, pioneered by Sanford Grossman, Oliver Hart, and John Moore in a theory of the firm context and now the centerpiece of privatization theory as well. It’s now second nature to privatization theorists that “any organizational mode can be copied by any other organizational mode through a complete contingent contract. Therefore, if there is any difference, it must be due to the fact that only incomplete contracts are feasible at the stage of privatization.”
Next time: Why privatization might end up being relevant after all.