Private prisons are a useful case study. First, they are a growth industry, having progressed from humble beginnings in the late ’70s and early ’80s to now house about one in sixteen prison inmates nationwide. Second, the opponents of private prisons commonly make the political influence argument.

For example, in a recent Duke Law Journal article, Sharon Dolovich writes that “the legitimacy of punishment” is threatened “whenever parties with a financial interest in increased incarceration are in a position to exert influence over the nature and extent of criminal sentencing. If this concern is real”—and she suggests that it may well be—prisons should not be privatized because “the state ought not to foster yet another potentially influential industry that could seek to compromise further the possibility of legitimate punishment to promote that industry’s own financial interests.”

David Shichor, a prominent contributor to the prison privatization literature, opposes prison privatization in part because:

Through political lobbying, PACs, campaign contributions, and the provision of perks to politicians (as industrial and business corporations do), corporations are likely to continue to support and even accelerate incapacitation-oriented legislation and policies by which more people will spend longer periods of time in correctional institutions. Conversely, this trend may diminish the emphasis on alternative programs and will result in the pursuance of the “Hilton Inn mentality,” that is, trying to maintain high occupancy rates for profit purposes.

And Brigette Sarabi and Edwin Bender’s thesis is clear from the title of their report, The Prison Payoff: The Role of Politics and Private Prisons in the Incarceration Boom, in which they argue that prison privatization should be resisted in part because private prison firms have a “vested financial interest[] in increasing rates of imprisonment.” This is only a small sample of the literature. Click here (mw112) for a sample of the art.

* * *

This is just some set-up for the question. In later posts, I'll discuss my actual thesis, lay out my economic model, and discuss evidence.

This is the simplest form of the story, but one can also tell more complicated versions of the story in which privatization doesn’t necessarily decrease industry-expanding political advocacy. After presenting my main model, I introduce a few realistic complications into the model to show how the effect of privatization could be ambiguous—increasing private-sector advocacy but also decreasing public-sector advocacy. If those extensions of the model are closer to the truth, then total advocacy may rise—but it may also fall, depending on which effect dominates. We cannot determine the net effect a priori.

There is thus no reason to believe an argument against prison privatization based on the possibility of self-interested pro-incarceration advocacy—unless the argument takes a position on how lobbying, political contributions, and advocacy work, and why (for instance) increases in private-sector advocacy would outweigh the decrease in public-sector advocacy. Either this argument against prison privatization is clearly false, or it is only true under certain conditions that the critics of privatization have not shown exist.

The analysis here not only sheds light on the prison privatization debate but also provides a roadmap for analyzing military contracting and other privatization contexts. Because privatization can affect the incentives of both the private and public sectors to wield political influence, one shouldn’t conclude that privatization distorts substantive policy unless one can tell a story, based on a plausible view of government agents’ behavior, in which private-sector advocacy rises more than public-sector advocacy falls. In the end, each industry has its own idiosyncrasies, so I do not make a strong claim about the use of the argument outside of prisons. But, at the very least, the use of the political influence argument is often theoretically unsound to the extent it ignores this comparative analysis.

* * *

In later posts, I'll set forth my main model. In this model, the sector with the greatest benefit from the expansion of the industry does all the advocacy, and the sector with the smaller benefit free-rides off the larger one. Thus, privatization will reduce industry-expanding advocacy if, after privatization, the public sector remains the sector with the greatest benefit.

Then, I'll apply the theoretical model to prisons and suggest, based on an informal calculation, that the actors that would benefit the most from increased incarceration are indeed the public-sector corrections officers’ unions. Thus, we should expect the public sector to do all the pro-incarceration lobbying (though less than it would have done without privatization). This simple model, despite its stark result, may be quite close to the truth, as there is a wealth of evidence that public corrections officers’ unions advocate incarceration, and no such hard evidence on the private side.

I'll explain why it is appropriate to focus on public corrections officers’ unions and private prison firms as the relevant actors, and how cooperation within the prison industry affects the results. Finally, I'll complicates the model in various ways. Some of these complications don’t change the basic result of the simple model. Other complications make the result muddier, so that instead of unambiguously reducing advocacy, privatization has a theoretically ambiguous effect on the amount of industry-expanding advocacy.

Suppose you are, as economists say, a rational, risk-neutral expected utility maximizer. One may dispute how much of life this assumption can explain, but on balance it seems to be at least a good starting point for predicting the behavior of business organizations. You are faced with the choice of whether or not to spend a dollar on political advocacy—donating to the campaign of a politician or voter initiative, contributing to your trade association’s lobbying expenses, or running an ad—in favor of some reform that could increase the size of your market. We may assume that this dollar has some influence in the world, whether appropriate or inappropriate—it could corrupt a legislator, raise the chance of his election, contribute to the passage of the initiative, or change popular opinion.

The benefit of this dollar is the value of the increased probability of getting your desired policy change. It is reasonable to think that spending money on advocacy is subject to decreasing marginal returns, so each additional dollar gets you less and less benefit. The cost of a dollar, on the other hand, is and remains $1, no matter how many of them you spend. As long as the benefit of an advocacy dollar is greater than $1, you continue spending; and you stop as soon as that benefit reaches $1. Finally, you settle on the optimal total amount of advocacy spending—say $1 million.

The picture below illustrates the situation. The expected benefit—that is, the probability of success times the benefit—is represented by the curved line below: The more you spend, the greater the probability of success, but the less you get for each extra dollar; and because a probability can’t get any higher than 1, the curve is bounded above by the dashed line representing the total benefit of the policy. The cost of advocacy is represented by the straight line below: Each dollar of advocacy costs $1. Your problem is to maximize the vertical distance between the expected benefit curve and the cost line. In the figure, that maximum distance occurs at a spending level of $1 million.

The next picture is an equivalent way of seeing the same problem. The curve below represents the marginal expected benefit—that is, the benefit of an extra dollar of spending, which is just equal to the total benefit times the extra probability that a dollar buys you. As noted above, the marginal benefit is decreasing. The straight line is the marginal cost of advocacy: An extra dollar of advocacy always costs $1. If the marginal expected benefit is above $1, you’re not spending enough; if it’s below $1, you should cut back. At a spending level of $1 million, an additional dollar of spending gives you exactly $1 of expected benefit.

Now suppose the Department of Justice’s Antitrust Division comes in and splits you up, so that you now have 90% of the market and are faced with a competitor with the other 10%. Your previous optimal amount, $1 million, is no longer optimal for you: The cost of that last dollar was $1, and while the benefit of the dollar is $1 for the whole industry, you, who now represent only 90% of the industry, only see 90¢ of that benefit. All your benefits are now lower by 10% because you have to share them with your competitor; for our purposes, the split-up thus has the same effect as a 10% tax on your benefit.

Because your spending on advocacy—an investment in the growth of your industry—is only 90% as productive, you do less of it. You start cutting back on your spending, because a dollar saved puts $1 back in your pocket and only reduces your benefits by 90¢. As you cut back more, the benefit of the last dollar rises; you stop cutting back as soon as the benefit of your last dollar to the industry reaches about $1.11 (which is a $1 benefit to you). Call the new amount $900,000.

This new situation is illustrated in the next pictures. In the next one, the top curve is the expected benefit to the whole industry (as before), and the second curve is your expected benefit, newly reduced now that you only have 90% of the industry. The bottom curve is your 10% competitor’s expected benefit. The maximum vertical distance between your curve and the cost line now occurs at $900,000, and the maximum distance between your competitor’s curve and the cost line occurs at $300,000.

In the next picture, the equivalent graph that shows marginal quantities instead of total quantities, you want to find the point where the marginal expected benefit to the industry is $1.11. This is equivalent to finding the point where 90% of the marginal expected benefit (i.e., the benefit to you) is $1. That point is again $900,000. And your competitor wants to find the point where the marginal expected benefit to the industry is $10, which gives him $1; this is again $300,000.

This story is incomplete. You don’t want the amount spent to be exactly $900,000; obviously, you would be thrilled if other people happened to contribute more. It’s just that you’re not personally willing to put any dollar into the pot after the 900,000th. You want the total amount spent to be at least $900,000, and are willing to contribute money until that point is reached, but you stop being willing to personally contribute once you’re holding the 900,001st dollar.

This is because, in this model, the benefit of a dollar only depends on the total amount of money spent, and if the 900,000th dollar had a benefit to the industry worth $1.11 (and thus a benefit to you worth $1), then the 900,001st dollar has a benefit worth slightly less than $1.11.

Your new competitor, who represents the remaining 10% of the industry, and who is equally interested in this reform that will increase the size of the pie, wants the total amount spent to be at least $300,000, and won’t put a 300,001st dollar into the pot.

All this leads to two conclusions.

First, the total amount spent should not exceed $900,000. If it did, you would want to take some money out of the pot, since the dollars beyond the 900,000th are giving the industry a benefit below $1.11, and therefore giving you a benefit below $1.

Second, there’s no reason for your competitor to spend anything. He’s unwilling to spend any dollar beyond the 300,000th, since its marginal benefit to the industry is under $10, and so its marginal benefit to him is under $1. Thus, suppose you were going to spend $600,000 and he was going to spend $300,000. These actions would not be an equilibrium, since he would prefer to keep his $300,000. Why should he spend any extra dollar beyond the $600,000 you’re already spending, if the 300,001st dollar already isn’t worthwhile to him? The only equilibrium is where you give $900,000 and he gives $0. Because he’s the smaller actor, he totally freerides off you. The result is what Mancur Olson calls the “systematic tendency for ‘exploitation’ of the great by the small.”

How far can we continue to privatize before advocacy stops falling? As privatization increases, the second step always holds—by definition, privatization shrinks the size of the public sector. The first step, however, does not always hold for large enough levels of privatization. Obviously, at a certain point, the private sector can come to dominate the public sector. Then the private sector will do all the advocacy, with the public sector free-riding, and from then on privatization might increase advocacy. The level of privatization at which advocacy stops falling is a threshold that we may call an “advocacy-minimizing privatization level.”

For instance—going back to the graphical model above—suppose our two firms benefit identically from having a given proportion of the industry. Then the advocacy-minimizing breakup is a 50–50 split (or 33–33–33 if the breakup is into three firms, and so on)—because under such a split, the dominant firm is as small as it can possibly get.

If a split creates a splinter firm that is twice as profitable as the incumbent firm, or perhaps twice as slick, then the advocacy-minimizing split is 67–33, again equalizing the firms’ stake in the system. The splinter firm is half as large but twice as profitable, so again the dominant firm (in terms of “real” share) is as small as it can possibly get. Conversely, if the splinter firm is only half as profitable as the incumbent firm, the advocacy-minimizing split is 33–67.

This concept becomes useful in later posts. I will argue that for prisons, the public-sector share of total benefit is much larger than the private-sector share—first, because the public sector still has a larger industry share, and second, because private firms are subject to a more competitive regime, so their profits are fairly low. Thus, the advocacy-minimizing level of privatization is probably quite high. Within this model, it would take a lot more privatization before pro-incarceration advocacy stops falling.

This basic result—that, by fragmenting an industry, one can reduce that industry’s political advocacy to increase its market—is also consistent with empirical studies on the relationship between industry concentration and lobbying.

In general, industry concentration can have two opposing effects on lobbying. The first effect is the one discussed above: A concentrated industry can more easily overcome its collective action problems, so we should expect it to lobby more. But not all lobbying is industry-expanding; some lobbying is simply anti-competitive, seeking to regulate the market (for instance, through entry restrictions) to allow existing firms to charge above-market prices. Firms in a more concentrated industry can more easily cooperate to suppress competition in the product market, so they have less need for anti-competitive lobbying: They can raise prices above market levels all by themselves by directly using anti-competitive methods.

Thus, studies that don’t disentangle these effects can come up with results in either direction. One study found a positive effect of concentration on industry contributions, while another found that the percentage of firms in an industry that had political action committees rises and then falls as concentration goes up.

Here, though, I focus only on advocacy for reforms that increase the size of the industry, and not on advocacy for reforms that would squelch competition in the industry—since it is primarily the first sort of advocacy that privatization critics urge is illegitimate. So only the first of these forces comes into play here.

Now, I know what you're thinking: Look, Volokh, you economists spin out your simple, highly stylized models, but can you really expect the private sector to do zero advocacy?

Whatever the general merits of such skepticism, in this case the simple model might actually be true.

The next post or two will document what we know about prison industry advocacy. In brief, there is a lot of hard evidence of pro-incarceration advocacy by public corrections officers’ unions (though a small part of union advocacy also cuts the other way). (There is also hard evidence that most Departments of Corrections advocate the other way—in favor of alternatives to incarceration.) But there is virtually no hard evidence of private-sector pro-incarceration advocacy. This may simply mean that the private sector advocates secretly. But, in light of the theory, it is more plausible that the private sector simply free-rides, saving its political advocacy for policy areas where the public good aspect is less severe—pro-privatization advocacy.

Even if one disagrees with the preceding sentence, this model needn’t be realistic in a literal sense. Advocacy needn’t be an entirely public good, and the actors in the industry needn’t totally free-ride. The point is merely that these assumptions are plausible, perhaps even likely. Advocacy has some public-good aspects, and free riding happens to some extent in the world. If people act enough like this model, privatization will still reduce total pro-incarceration advocacy.

This plausible scenario rebuts to the simple anti-privatization claim that privatization does increase pro-incarceration advocacy. (The extended models presented in even later posts, in which the effect of privatization on advocacy is ambiguous, further rebut the claim.) This scenario also points up a potential irony in the position of some incarceration opponents who, so as to avoid “reinforc[ing] the incarceration boom by introducing the profit motive into incarceration,” would make common cause with public corrections officers’ unions, who concededly are active lobbyists for incarceration.

Private prison firms depend, for their livelihood, on two policies: privatization and incarceration. Indeed, they admit as much to the world, in their annual reports filed with the SEC. As to privatization, The GEO Group, the second largest private prison firm, explains that “[p]ublic resistance to privatization of correctional and detention facilities could result in our inability to obtain new contracts or the loss of existing contracts, which could have a material adverse effect on our business.” As to incarceration, GEO candidly remarks:

[A]ny changes with respect to the decriminalization of drugs and controlled substances or a loosening of immigration laws could affect the number of persons arrested, convicted, sentenced and incarcerated, thereby potentially reducing demand for correctional facilities to house them. Similarly, reductions in crime rates could lead to reductions in arrests, convictions and sentences requiring incarceration at correctional facilities.

Similar statements are easily available in prison firms’ public filings. It is thus natural to suspect that prison firms may advocate both privatization and incarceration in the public square. Their political advocacy—which is extensive—mainly takes the forms of contributions to politicians and participation in the American Legislative Exchange Council (a conservative organization that drafts modern legislation), though they also testify before Congress and present arguments in the popular press. But, while it is clear that these firms advocate privatization, it is unclear that they advocate incarceration to any significant extent.

Most of the evidence of advocacy specifically in favor of incarceration has been speculative. Some writers state that it doesn’t happen, while others who are concerned about the prospect hedge their statements with terms like “may” or “are likely to.”

I noted above that the general contributions of corrections officers’ unions can’t be traced back to any specific goal, like pro-incarceration advocacy. Similarly, some commentators note private prison firms’ advocacy but don’t distinguish between pro-privatization and pro-incarceration advocacy. This blanket approach is a mistake, unless one is attacking all political involvement by prison firms. Generalized contributions to candidates, unlike targeted activities like contributions to single-issue voter initiative campaigns, are mute. Some of the industry’s contributions to politicians may be multi-purpose, for privatization as well as for incarceration. Merely advocating increased privatization raises quite different concerns than advocating changes in the criminal law itself, and certainly does not implicate the same sorts of “legitimacy” values.

Since the industry’s public statements virtually all relate to favoring privatization, there is little hard evidence on the basis of which to attribute part of their political contributions to a pro-incarceration motive. Indeed, the Association of Private Correctional and Treatment Organizations (APCTO), the industry’s trade group, speaking for its member firms, denies that the industry lobbies for increased penalties:

Individually and as an Association, we do not lobby in favor of longer sentences, so-called “three-strikes” laws, or other legislation which could result in an increase in the jail or prison population. To the contrary, the Association and its member companies encourage the use of appropriate alternatives to incarceration; provide inmates with treatment, education and rehabilitative services designed to positively impact and reduce recidivism rates; and encourage effective transitional programs for offenders upon release.

APCTO frequently endorses alternatives to incarceration, treatment programs, and other measures to reduce recidivism. Its executive director recently suggested in the Denver Post that to alleviate prison overcrowding, Colorado should “[l]ook to alternatives to incarceration that can provide treatment and rehabilitative programs to first-time, nonviolent drug and alcohol offenders,” “[r]educe recidivism by investing in the treatment, education and rehabilitation that offenders need to be successful when they leave prison,” and “[i]ncrease the likelihood that released inmates will not re-offend by providing substantive transitional programs to help released inmates adjust to the community outside the walls of prison.” (He made similar recommendations to Ohio in the Cincinnati Post.) He also suggested in the Fort Pierce Tribune and the Palm Beach Post that Florida should invest more in juvenile justice services in order to reduce the adult prison population in the long run. (He noted that APCTO’s member companies mostly provide adult incarceration services, though some would like to expand their juvenile programs.)

Even if one ignores the industry association’s official statement as self-serving and dismisses their anti-incarceration positions as PR, at most political contributions are “soft evidence” of pro-incarceration advocacy. The most we can say empirically based on such evidence is that maybe pro-incarceration lobbying happens and maybe it doesn’t. Perhaps the hard evidence is missing because the industry covers its tracks; or perhaps the hard evidence is missing because there is nothing to cover up.

Prison firms also participate in the American Legislative Exchange Council, an influential conservative organization that drafts model legislation. Both CCA and the former Wackenhut Corp. (now called the GEO Group) have been members of ALEC (and they and Sodexho Marriott, a major CCA stockholder, are prominent corporate funders of ALEC), and, over the years, at least CCA has participated in (and two of its executives have chaired) ALEC’s Criminal Justice Task Force, which drafted, among other things, a “Truth in Sentencing Act” and a “Habitual Violent Offender Incarceration Act.”

The inner workings of ALEC are hazy, and indeed, some commentators argue that the private prison industry expressly seeks out channels that are “conveniently out of public view” and “behind closed doors” to promote its pro-incarceration agenda. Presumably, prison firms work within ALEC on privatization issues: Prison privatization is one of the “major issues” of ALEC’s Criminal Justice Task Force; the Task Force has a Subcommittee on Private Prisons and has a model “Housing Out-of-State Prisoners in a Private Prison Act”; and CCA is known to have talked to the Task Force on the subject.

But this, too, is “soft” evidence; we do not know that they also work on sentencing or incarceration issues. Indeed, CCA asserts that it has not participated in, voted on, or endorsed any stand on model legislation for sentencing or crime policies within ALEC. Apparently, the only CCA official to have ever publicly taken a stand on sentencing is J. Michael Quinlan, formerly Director of the Federal Bureau of Prisons and now a CCA Senior Vice President, who, after he joined CCA in 1993, told a House subcommittee that mandatory minimum sentences “are unnecessary for non-violent, non-serious offenses” and “pose[] a severe threat to prison discipline and management.”

So far, I have found a single piece of evidence of arguable pro-incarceration advocacy by a private firm. In 1995, Wackenhut chairman Timothy P. Cole testified in favor of certain amendments to the Violent Crime Control and Law Enforcement Act of 1994. The main point of his testimony was to propose additional provisions (1) making clear that prison grants under the 1994 Act would “help pay for the entire range of correctional services states can provide in-house or under contract” (not merely for “alternative correctional facilities”), (2) requiring states to “show that they have all the necessary legislative authority to embark upon a comprehensive, integrated program and that they will employ the best technology at the lowest cost” (presumably to boost privatization), (3) directing the Attorney General to “give top priority to the construction of larger, ‘harder’ [i.e., higher-level security] facilities,” and (4) directing the Attorney General to give priority to states with “an executive body dedicated to the review and consideration of privatization.” During this testimony, he said the following:

• “Our proposed amendment . . . would help to assure that these grants will help the states incarcerate more violent criminals and not make the state governments more dependent on federal tax dollars in the long term.”

• “By passing ‘truth-in-sentencing’ laws, states have begun to restore a fundamental sense of justice and fairness to our system of crime and punishment.”

• “The new grant program [under the 1994 Act, without the proposed amendments] is available for ‘alternative correctional facilities’ and does not recognize the urgent need for more cells in secure facilities.”

• “Current law encourages billions to be spent on new or retrofitted facilities that are not large enough, secure enough or efficient enough to keep the maximum number of violent criminals in prison for the least cost.”

This isn’t great evidence—Cole was really only advocating funding priorities and privatization-friendly decisionmaking. Cole’s request to divert money from alternative facilities, his kind words for truth-in-sentencing laws, and his positive attitude toward locking up violent criminals are hardly a pro-incarceration smoking gun. But this is the best I’ve found. Private prison firms may have made other statements and taken other public positions that are arguably pro-incarceration, but I haven’t found any, and to my knowledge, privatization critics have not brought them to light.

* * *

Now, I'll elaborate on two curlicues of the theory. First, I'll explain why I have focused only on public-sector unions and private firms. (In short, private-sector workers aren’t unionized, and public Departments of Corrections actually want fewer prisoners.)

Next, I'll explain why I have assumed that the private firms act as a bloc instead of competing with each other or, at the opposite extreme, cooperating with the public sector. (In short, cooperation within a fairly concentrated oligopoly isn’t that difficult, because firms interact with each other a lot and have ample opportunity to punish each other for non-cooperative behavior. And private-sector firms interact with each other more than they do with the public sector, so enforcing cooperation across the whole prison industry would be more difficult. However, it turns out that how the industry cooperates, or whether it cooperates at all, doesn’t make much of a difference for the main result.)

* * *

One might ask, at this point, why I have focused primarily on two apparently asymmetrical groups: the private-sector firms and the public-sector employees. What about the employees of those firms, or the employers of those public guards?

In principle, it’s unclear a priori who would want to lobby; a case-by-case analysis of the incentives of the various parties is necessary. In this case, my choice of actors was inspired by the state of the evidence and the debate: Public corrections officers’ unions, especially in California, are known to engage in pro-incarceration advocacy; and private prison firms are alleged to do so. But let us think about this theoretically anyway.

First, the workers. No single worker has enough of a stake in the system to benefit from spending resources on advocacy to help his industry. We should only expect workers to be a significant political force if they can enforce some sort of collective action by punishing their own free-riders. The easiest way for them to do this is to require membership in or contribution to a union, which then lobbies out of members’ dues. Private corrections officers aren’t unionized in most states, and this is a sufficient explanation for why they haven’t been observed lobbying.

As I explain above, I assume that unions represent their members and seek to maximize total union rents—the difference between union and non-union wages, times the size of the public sector. The prediction that such unions would seek to increase the size of their sector is straightforward: A larger sector may mean a more powerful union and therefore potentially higher wages, benefits, or job security down the road (and perhaps—to introduce agency costs for a moment—perks for union officials). It’s possible that unions may sometimes oppose expansion of their industry—for instance, if increases in prisoners made corrections officers worse off because they weren’t accompanied by compensating wage or staff increases. This may occur in some industries, but apparently not in the prison industry. The public corrections officers’ unions seem, so far, to have been strong enough to make sure that an increased flow of prisoners has not made them worse off (even if budgets have been tight elsewhere).

Now, let’s consider the employers. Some private prison firms also run alternatives to incarceration, so it isn’t obvious that they would advocate an increased emphasis on imprisonment. Still, they may benefit from the other elements I have included in the term “incarceration”: increased illegalization, increased law enforcement, and longer sentences (once the imprisonment decision has already been made). Though increased incarceration may also increase private firms’ costs, private firms have a built-in protection against too much deterioration in their position: They don’t have to bid on a contract unless they anticipate making enough profit. So it isn’t implausible that private firms would benefit from incarceration; though of course their willingness to spend money to increase incarceration depends on how profitable they are.

What about the public employers, the Departments of Corrections? They are not players in the pro-incarceration advocacy game for a simple reason: Generally, they favor alternatives to incarceration.

The Alabama DOC commissioner has advocated sentencing reform, community correction programs, and other measures to “reverse the prison population growth trend.” The head of the Illinois DOC advocates reentry programs that would lower the prison population by countering “the awful, vicious cycle” by which recidivist parolees are re-incarcerated “before the ink is dry on their parole papers.” The Michigan DOC director concerns herself with measures to reduce the prison population and thus delay the day the state runs out of funded capacity for prison beds. The Montana DOC director candidly tells crowds that “[p]rison isn’t working,” and his department considers measures to reduce the prison population and increase community corrections. The New Mexico Corrections Department is focusing on using early parole to control its prison population. The North Carolina DOC advocates redirecting non-trafficking drug users from prison to “intermediate programs.” Ohio corrections officials complain about the high costs of mandatory minimum sentences. The Pennsylvania DOC is implementing programs “aimed at diverting less serious offenders from prison” to “free-up prison space needed for more serious offenders.” The Washington DOC secretary “is a big believer in work-release programs.” And the Wisconsin DOC secretary advocates focusing on “prevention and treatment in addition to effective law enforcement.”

This makes some sense: While it is commonly thought that agencies want to aggrandize themselves, that intuition is only a special case of a more general belief that agency officials act in their own self-interest and that their self-interest tends to be aligned with the size and power of their agencies. And increasing prisoners without corresponding budget increases to match the increasing cost of incarceration (a cost that of course includes corrections officers’ salaries, as well as health care and other factors) can easily make prison officials worse off. Thus, the interests of Departments of Corrections may not be aligned with those of corrections officers and their unions. Moreover, DOCs run both prisons and many alternative programs, so even if more inmates means more power for the DOC, it makes sense that the DOC would want to handle those inmates in cheaper ways than incarceration. Thus, it is not surprising to find prison systems arguing for alternatives to incarceration in a time of tight budgets.

* * *

All this talk of how the 10% firm acts and the profits of “the industry” assumes that the private sector, in deciding how much to spend, acts as a bloc: The private firms all cooperate with each other, but don’t cooperate with the public sector. This is possible, but it’s not the only possible story. I could have made either of two other, more extreme assumptions. First, there could be no cooperation at all—all the firms could be acting independently. Second, there could be total cooperation—all the firms could be cooperating with each other and with the public sector. This section explores the implications of these alternative assumptions and tentatively defends my decision to adopt the intermediate assumption of cooperation within the private sector but not with the public sector. But in the end, the different assumptions don’t significantly alter the bottom line.

If all firms act independently, the relevant shares are even less than indicated above. In 1999, CCA had a bit over half the market, Wackenhut (now the GEO Group) had about a quarter, Management & Training Corp. had about 5–8%, Cornell Corrections Inc. and Correctional Services Corp. each had about 5–6%, CiviGenics, Inc. had about 2–3%, and a handful of other firms had under 1%.

So, while the average private-sector share in State X may be 10%, that number is irrelevant if all firms act independently. The relevant shares may be, for instance, 6% for CCA, 2% for the GEO Group, 1% for Management & Training Corp., and 1% for Cornell Corrections . . . and 90% for the public sector. The assumption of independent firms would make it even more likely that the public sector is the dominant sector.

Now consider the opposite assumption—that everyone cooperates. A single prison industry bloc would choose an optimal total amount to maximize total industry benefit. Because the actors are still formally separate, they would also choose some way to allocate the contributions among themselves.

If the private industry had the same benefit per prison as the public sector, then total cooperation would be indistinguishable from monopoly: Because total industry benefit would be the same before and after privatization, the strategy that chooses contribution amounts to maximize that benefit would likewise be the same.

However, as I argue above, private firms aren’t terribly profitable, while public-sector unions have significant public-sector wage premiums to protect. By replacing part of the public sector with a relatively unprofitable private sector, privatization actually decreases the industry’s total benefit. Therefore, even under total cooperation, there is less to maximize; expenditures on pro-incarceration advocacy are thus less productive (just as if there were a tax rate on industry revenues); and so expenditures on advocacy still go down under privatization.

How can we tell which form of cooperation is most likely? Not being able to find explicit cooperation doesn’t mean anything: The cooperation may just be tacit. Observing the private industry’s trade association, APCTO, also doesn’t answer the question: While trade groups may provide a forum for discussing common lobbying strategies, but talk is cheap and many trade groups are ineffective; in particular, APCTO doesn’t seem to fulfill much of a coordinating function, as firms do their own lobbying and most of their own advocacy.

Even observing some actual lobbying by the major firms doesn’t answer the question: As I noted above, they may all be lobbying for privatization, which has a strong private-good component, since a firm’s contributions may increase the probability that it gets a project in the future.

On theoretical grounds, it seems at least plausible that the private firms would cooperate among themselves. They’re repeat players in a long-term process, which includes both political advocacy and their concern of bidding on prison projects; so there is ample opportunity for private firms to enforce a regime of cooperative behavior. If firms free-ride off each other in their advocacy expenditures, their fellows could punish them in the future in any number of ways—for instance, by never cooperating on campaign spending anymore, by bidding aggressively in prison auctions, or by bidding aggressively only in certain markets.

By contrast, public corrections officers’ unions may have fewer ways of punishing private firms. They don’t bid against each other in the underlying auctions, so they can’t threaten to end any cooperative behavior there. They’re bitter political adversaries in the privatization advocacy world, so again there seems to be no preexisting cooperation that can be terminated. They can threaten to not cooperate any more in pro-incarceration advocacy or to step up their anti-privatization advocacy, but this may not be as effective a threat.

For these reasons, I believe that cooperation among private prison firms is more likely than either, on the one hand, totally non-cooperative behavior or, on the other hand, totally cooperative behavior between the public and private sectors. However, because the ultimate results under any of the assumptions don’t differ that much, which assumption we choose isn’t terribly important.

Allowing Money to Change Candidates’ Positions.

So far, I have taken the political agenda as given: I didn’t explain where the proposed reform came from. Thus, I’ve assumed that money is important because it buys victory—for instance, by persuading voters of the benefits of the policy or the merit of the candidate. But money can also affect the agenda—by changing candidates’ positions, by inducing the sponsors of voter initiatives to propose a different initiative, and so on.

When money can affect the agenda (but the other assumptions are unchanged), the analysis is essentially the same. Suppose you are considering whether to contribute to place an initiative on the ballot. The initiative is supported by some group or other, but for specificity, let’s say it’s being sponsored by a politician. This politician may be fairly pro-incarceration himself, but he’s limited in how strict an initiative he can propose: He won’t prevail unless the median voter, whose views control the outcome of the election, prefers his proposal over the status quo. However, before the substance of the initiative is set in stone, you can move him in a more pro-incarceration direction if—by offering him money to pay for persuasive advertising—you offer him the possibility to also move the median voter.

A monetary contribution has the following effects:

1. Electoral influence.

   • As before, you benefit because, by paying for persuasion, your contribution directly increases the probability that the initiative prevails.

   • But the contribution also moves the initiative in a more pro-incarceration direction, which cuts against the effect above.

2. Substantive influence. Finally, you benefit if the initiative prevails, because the policy is better for you than it would have been if you hadn’t contributed.

It turns out that this complication to the model doesn’t change the underlying result. As a prison provider thinking about how much to contribute, you follow the same framework as before: You contribute until the benefit of an extra dollar is worth $1 to you. The benefit of an extra dollar is more complicated than it was in the earlier model—in addition to encompassing the positive electoral influence effects, it now adds the negative electoral influence effect, as well as the benefit of substantive influence—but the basic idea is the same.

Now suppose, again, that the industry is split up into a 90% sector (you) and a 10% sector (them). As before, your benefits go down to 90% of their previous level, so you now want to contribute until the benefit of an extra dollar to the industry is worth $1.11. As before, you contribute less than before the split, because having only 90% of the industry is like facing a 10% tax on benefits. And as before, your competitor free-rides off you, because when he takes your contribution level into account, an extra dollar in the pot is no longer worthwhile to him.

* * *

Anti-Incarceration Advocacy.

This model focused only on the pro-incarceration side’s advocacy, taking the anti-incarceration side’s advocacy as given. But clearly anti-incarceration advocacy exists, and it is plausible that the pro- and anti-incarceration forces respond strategically to each other’s expenditures. This suggests two questions.

First, one might wonder whether the existence of anti-incarceration advocacy changes what we have already said about the effect of privatization on pro-incarceration advocacy. It turns out that it does not: Just as in the simple case, privatization makes pro-incarceration advocacy go down, even when we consider interactions with anti-incarceration advocacy.

Second, one might wonder how privatization changes anti-incarceration advocacy. After all, some anti-incarceration advocacy is as plausibly self-interested as the prison providers’ pro-incarceration advocacy. For instance, Proposition 66, which would have limited California’s Three Strikes Law, was partly funded by “Sacramento businessman Jerry Keenan whose son Richard is serving time for manslaughter after crashing his car while driving drunk and killing two passengers.” Proposition 36, the drug treatment diversion initiative, was supported by dozens of drug treatment providers and 16 medical and public health organizations, including the California Association of Alcoholism and Drug Abuse Counselors and the County Alcohol and Drug Program Administrators Association of California. And, as I showed above, state DOCs generally advocate against incarceration. If there exists all this self-interested anti-incarceration advocacy, perhaps those who are concerned about self-interest coloring people’s positions on criminal justice should be concerned about this as well.

It turns out that the privatization-induced decrease in pro-incarceration advocacy has an indirect effect on anti-incarceration advocacy. Unfortunately, we cannot say anything a priori about the direction of this effect. On the one hand, pro-incarceration advocacy decreases the effectiveness of anti-incarceration advocacy by counteracting it. So a decrease in pro-incarceration advocacy makes anti-incarceration advocacy more effective, which would tend to increase it. On the other hand, a decrease in pro-incarceration advocacy also makes anti-incarceration advocacy less necessary, which would tend to decrease it. There is no theoretical way to know how these conflicting effects would balance out; but in principle future research could answer the question empirically.

What this means normatively depends on one’s attitude toward anti-incarceration advocacy. If one opposes pro-incarceration advocacy because there’s already too much incarceration, then there’s nothing wrong, and perhaps everything right, with advocacy the other way.

On the other hand, if one opposes pro-incarceration advocacy because it is assumed to be self-interested, then perhaps anti-incarceration advocacy is just as bad if it comes from boot camps, halfway houses, drug treatment providers, and other presumptively self-interested parties. This model, which says nothing specific about total advocacy (either its amount or its effect), is thus normatively ambiguous.

* * *

Relaxing the Assumption of Fungible Money.

Recall the main model presented in the third post, in particular the last two pictures. A monopoly provider would have spent $1 million on advocacy, but under a 90–10 split, the 90% provider is unwilling to spend beyond the 900,000th dollar and the 10% provider is unwilling to spend beyond the 300,000th dollar; and so total advocacy falls to $900,000, with the dominant provider spending everything and the other one spending nothing.

The result that the smaller-share-of-total-benefit sector totally free-rides off the efforts of the dominant sector was driven by the assumption that the probability of getting the change in policy only depended on the total amount of money in the pot. All advocacy was fungible. A dollar from a public actor had the same effect as a dollar from a private firm. This is not an implausible assumption. For instance, dollars are fungible in buying advertising, which increases the probability of a change. A politician may adopt the view of whatever “policy position” contributed the most to his war chest.

On the other hand, some alternate assumptions may also be plausible. For example, one group might be attractive only to Democrats, while another might be attractive only to Republicans. More generally, perhaps politicians are just sensitive to the variety of voices in a coalition, feeling (rightly or wrongly) that having a wide variety of groups shows that a policy has wide support. Then neither group’s contributions totally “crowd out” the other’s. Your 500,001st dollar still has less benefit than your 500,000th dollar—there are still decreasing marginal returns—but (unlike in the previous model) it does not have the same benefit as your first dollar added on to your competitor’s 500,000th. Therefore, the total free-riding effect from the simple model above no longer occurs.

In this context, privatization has two effects. First, it increases the private-sector share, so private-sector advocacy goes up. Second, it decreases the share of the public sector, so public-sector advocacy goes down. We can’t say anything a priori about whether the first effect outweighs the second. If we know some facts about public- or private-sector advocacy—for instance, if one sector is completely unpersuasive, while the other sector is slick and sympathetic—then we can hazard some predictions, but we can’t say anything without such empirical facts. Because the empirical effect of privatization is ambiguous, the normative effect of privatization is also ambiguous if one opposes pro-incarceration advocacy. Unless we can be specific about how different groups’ advocacy has different effects and how effective the groups are, it is impossible to say whether prison privatization increases or decreases self-interested pro-incarceration advocacy.

* * *

Strong and Weak Unions and Industries.

Let us return to the point I made above that an industry’s effectiveness at advocacy is relevant to its “real” share for purposes of this analysis. For instance, if your competitor, with a 10% share, is twice as slick a lobbyist as you, meaning that his marginal dollars produce twice the benefit of yours, he will act as though his share is 20%.

Which way this cuts isn’t clear, as we don’t know which sector is more effective at lobbying in favor of incarceration. The CCPOA, as we’ve seen, is highly effective, but corrections officers’ unions are much less active outside of California, and perhaps this is because they are less effective. It’s hard to say how effective private prison firms would be at lobbying in favor of incarceration, since, as we’ve seen, there’s little evidence that they do this at all; and if they do it secretly, it’s likewise hard to gauge how effective they are.

But let us suppose that one’s effectiveness at lobbying for incarceration is correlated with one’s effectiveness at lobbying for (or against) privatization. For simplicity’s sake, let us suppose that they are perfectly correlated. Consider the states with high levels of privatization. We may conclude that those states have high privatization because their corrections officers’ unions were not effective at opposing privatization; the private industry was just too strong for them. When that relatively “weak” public sector was partly displaced by a relatively “strong” private sector, a weak pro-incarceration voice was similarly displaced by a strong pro-incarceration voice. Pro-incarceration advocacy, then, may plausibly have increased.

Similarly, consider the states with low levels of prison privatization, like California (1.8% private in 2004), or no privatization at all, like New York or Rhode Island. The unions in those states, on this view, must have been stronger than the industry, or else we would see privatization there now. If privatization were introduced, total advocacy would go down; but privatization is unlikely to be introduced there, so we won’t see that happen.

This is a story where—contrary to my implicit assumption so far—privatization is endogenous: The states where privatization has gained a foothold aren’t randomly chosen; rather, privatization emerges where corrections officers’ unions are weak and fails to emerge where the unions are strong. Thus, past privatization may have, on balance, increased pro-incarceration advocacy. If one could somehow eliminate prison privatization (despite the confluence of powerful political forces that established it to begin with), one would reestablish the rule of the ineffective corrections officers’ unions in those states where they were ineffective—to the benefit of those who oppose pro-incarceration advocacy. By a similar logic, one should introduce privatization where it is currently absent: If it is currently absent, it is because it wasn’t a powerful enough political force to win on its own, which means it will also be an ineffective political force in fighting for incarceration.

In fact, the assumption here—that the effectiveness of pro-incarceration advocacy is perfectly correlated with the effectiveness of pro- or anti-privatization advocacy—implies that pro-incarceration advocacy is already as high as it can get, because the slick advocates, who were already slick enough to establish themselves in the industry, are now plying their slickness in the incarceration policy field. Adding a thumb to the privatization scales in either direction would tend to support the victory of the less persuasive party and would therefore reduce the total amount of pro-incarceration advocacy.

This story may be plausible, but it requires more fleshing out. For one thing, the assumption may not be right. Low-privatization states need not be high-union-strength states. While antipathy to privatization and the strength of public-sector unions are probably correlated, a very Democratic state may plausibly oppose privatization even if, for whatever reason, its unions are weak. Actors in the prison industry may not be similarly effective in the privatization debate as in the incarceration debate.

While one’s effectiveness at advocacy probably depends on one’s general characteristics, like goodwill, persuasiveness, and slickness, the specific subject matter of the advocacy the correlation also plays a big role. The incarceration debate is peopled by different interest groups than the privatization debate. For instance, prosecutors, police officers, victims’ rights groups, and rural communities are interested in incarceration policy but not so much in privatization policy. Conversely, prison privatization is a matter of interest even to interest groups without a direct interest in prisons, like, on one side, generalized public employee unions, and, on the other side, small-government advocates, who assume (probably sensibly enough) that a victory for privatization in any field is a victory for the general privatization movement. Moreover, the appeal of incarceration arguments, which connect to fears of drugs and crime and concerns over civil liberties, seems to have a very different source than the appeal of privatization arguments, which relate to taxes, spending, and the effectiveness of government services.

We are back, then, to a general state-by-state analysis. In the first set of models—where the effectiveness of advocacy only depended on the total amount of money in the pot—everything was driven by the dominant actor, where “dominant” also takes effectiveness into account. I have given arguments above as to why the private sector is currently probably the smaller actor. The “slickness adjustment” described here might change that in some places, but it is an empirical question. As is by now familiar, privatization still increases the private-sector share but decreases the public-sector share. This “slickness adjustment” may change the de facto shares of the different sectors, but it doesn’t change the qualitative result. The effect of privatization is theoretically ambiguous.

One further note. If one opposes self-interested pro-incarceration advocacy, one may object at this point that I have not exonerated private prisons with my economistic legerdemain. Rather, I have only shown that the entire system is corrupt, and perhaps I have unwittingly demonstrated that the only way out of this mess is to reject the “interest group model of politics” entirely as it applies to criminal justice policy.

Fair enough. If self-interested pro-incarceration lobbying is indeed undesirable, then perhaps the system is corrupt. But how does this translate into an argument against prison privatization? It’s not enough to show that private prisons are part of the problem: Removing one problem isn’t guaranteed to make things better when there are other problems around. As the models above have suggested, even if all this political advocacy is illegitimate, the existence of the private sector reduces public-sector advocacy and may reduce total advocacy; eliminating the private sector may thus exacerbate the problem.

Nor is it just economists who oppose making the best the enemy of the good: As Rawls (no economist he) teaches, the analyst who makes specific policy recommendations in our fallen world—not in the idealized world of “strict compliance” with the principles of justice that characterizes a “well-ordered society”—is acting in the realm of “nonideal theory,” which asks how the “long-term goal” dictated by ideal theory “might be achieved, or worked toward, usually in gradual steps. It looks for policies and courses of action that are morally permissible and politically possible as well as likely to be effective.”

Because nonideal theory requires that we ask about the real-world effectiveness of any reform, merely observing undesirable lobbying by the private sector will not support an argument against prison privatization unless, say, privatization actually increases “the danger of . . . corrupting influence” or “compromise[s] further the possibility of legitimate punishment.”

If it turns out that privatization actually reduces pro-incarceration lobbying—if, with privatization, prisoners’ sentences are less influenced by improper factors than they otherwise would be—it is unclear that there is any “tension between the state’s use of private prisons and the demands of” liberal legitimacy. If “private prisons are by no means unique,” and if any prison provider, public or private, will lobby for incarceration, any “tension” has nothing to do with private prisons and everything to do with the crooked timber of humanity.

* * *

The same sort of analysis that I have conducted here on the prison industry can also be used to evaluate the claim that, say, buying weapons from defense contractors (rather than having the military make them in-house) will exacerbate pro-war lobbying. Since governmental providers of defense services—i.e., the military leadership—have, on some accounts, been notorious pro-war lobbyists throughout history, such a claim is not credible unless one can tell a plausible story about why any defense contractor lobbying won’t crowd out some lobbying by the military itself; and doing this requires taking a position on what motivates the people at the Pentagon.

The same goes for private attorneys general, private redevelopment corporations, private landfill operators, and the like. The result won’t always be the same, and the political influence argument may turn out to be correct in some of these cases and incorrect in others. But this should be the structure of the argument.

The surprising moral of this story should not be that surprising. Indeed, the central insight here was also an important argument in favor of the antitrust laws. William Howard Taft wrote, shortly after the enactment of those laws, that “business methods and plans . . . directed to . . . suppressing competition . . . had resulted in the building of great and powerful corporations which had, many of them, intervened in politics and through use of corrupt machines and bosses threatened us with a plutocracy.” The argument is plausible, and it is likewise plausible that privatization, by fragmenting an industry into at least two chunks (and more if private firms don’t cooperate on advocacy), may similarly reduce that industry’s political power.

In a roundabout way, then, privatization is a form of antitrust, and antitrust is a form of campaign finance regulation. It may not be worthwhile to privatize industries—or break up large corporations—merely to reduce their political advocacy, but at the very least this may count as an unintended—and possibly happy—side effect of privatization that, if real, should be taken into account in future analysis.