Interests, Incentives and Institutions
By Joseph Stiglitz

An economist’s experience of government failure

As a long-time student of the public sector, I welcomed the opportunity to come to Washington as a member of the Council of Economic Advi-sers and later to become the Chairman of the Council, partly because it gave me an opportunity to study at first hand this immensely important part of our economy and society and to test my ideas against the reality of government in action.

To be sure, I came also as an activist, if not with a fully articulated agenda, at least with a view about what it was that government should, and should not be doing. My reference point was the fundamental theorems of welfare economics which, as some describe them, proved that the market left to itself would produce efficient allocations. Many saw in these theorems the vindication of Adam Smith’s faith in the invisible hand leading the self-interested decisions of each person to maximise the well-being of the nation as a whole. Today, many of us look at the fundamental theorem not as a description of the world, but as an explication of the conditions under which a market equilibrium will be Pareto efficient.

These conditions are quite strong. The importance of some of the more explicit assumptions – like the lack of externalities and the completeness of markets – has long been known. In the last two decades, we have explored much more seriously the consequences of the informational assumptions implicit in the belief that markets are efficient. In particular, it has been shown that in the presence of imperfect information or incomplete markets, the economy will not be Pareto efficient; in other words, there will always be some intervention by which the government can make everyone better off (Greenwald and Stiglitz, 1986).

Some of my friends who had spent a spell in Washington suggested that I would return a bit wiser for the experience, a bit more jaundiced about the role of government. That seems a shared experience: a better understanding of government failures to counterbalance the market failures that have occupied so much of my thinking as a professional economist.

In this article I want to share with you some of my thoughts about the possibilities and limitations of government. These thoughts are focused around a simple question: Why is it so difficult to implement even Pareto improvements?

Four reasons why potential Pareto improvements fail

The list of failures is a long one – ranging from the more obvious ones in trade and agriculture policy to those in social, health, and environmental policy. In reality, many of the policies I describe are near-Pareto improvements, which bring diffuse benefits to a large group but at a well-defined cost for a narrowly defined group. In some of these cases, feasible compensation was proposed to make it a strict Pareto improvement; in other cases, it was not.

The traditional answer to these questions is Mancur Olson’s Logic of Collective Action (1971) which argued that it is hard to overcome the free rider problem in organising a large group to defend itself against a concentrated interest. This answer, however, is incomplete because it does not explain why policies that harm no one or involve compensation fail to be adopted. My four hypotheses, in attempting to understand the unique nature of government’s powers and limitations, address this issue.

1. The inability of government to make commitments

Policy-making is a dynamic process, with today’s decisions shaping options and coalitions in the future. In the naive view, a Pareto improvement is a one-shot, static policy change. In reality, it is part of a sequence of policies, and although a reform may be favorable to all groups in earlier stages of that process, it may undermine one or a few groups’ interests in later stages. These disadvantaged groups, of course, are often far--sighted enough to anticipate that in the long run they will be worse off and thus act accordingly to oppose a seeming Pareto improvement.

Early on in the Clinton administration, we put forward a National Action Plan to address the problems of global warming. Among the myriad of actions was a small one which represented a Pareto improvement: making better use of the nation’s hydroelectric sites. The government enterprises that currently run many public dams sell electricity at far below the fair market price. The plants set prices at or slightly above their cost of production but, because hydro-electricity is much cheaper to produce than the more common coal-generated electricity, the prices of the government enterprises are lower than the marginal cost of electricity produced by the entire generating system. In many cases, the government enterprises lack the budget to finance additional investment which would allow them to produce more electricity, still at a price far below the cost of coal-generated power.

Our proposal addressed these concerns by allowing private firms to bid for the right to make these incremental investments and sell the additional electricity at market prices. We designed the proposal to be a Pareto improvement, ensuring that current recipients of ‘subsidised’ electricity would continue to receive the amount of electricity previously produced at the existing prices. The proposal was good for our budget (substantial revenues were anticipated from selling the rights); it was good for the economy (marginal cost pricing combined with replacing expensive electricity with cheap electricity); and it was good for the environment (increasing the non-carbon-emitting production of electricity). And since those who were already getting electricity at subsidised prices from the government would be able to continue doing so, no one would be hurt.

But of course, the answer depended on what implicit property rights people thought they already had and how this initiative might affect existing property rights. If those who currently had access to subsidised electricity thought that eventually they would be able to get an increased supply at the subsidised price by public investments in upgrades of existing sites, then the change was, from their perspective, a change for the worse.

Probably more important was the principle: once the principle of opportunity cost pricing was accepted, once it became clear that, in effect, those who were obtaining hydro--electricity at ‘cost’ – but far below market price – were in effect being subsidised, it would be only a matter of time before the subsidy was eliminated. Our modest initiative was viewed as the ‘thin edge of the wedge,’ a ‘slippery slope’ down which those who benefited from the current setup simply did not want to risk going.

There were two problems. The first was that the reform would make the effective subsidy transparent and, in doing so, would undermine its political viability. The second was that we could not make a credible commitment that these subsidies would be continued. The two problems are related: the increased transparency made it less likely that the subsidies would be continued and put further pressure on our inability to make commitments.

The problem of commitment stems from the inherent nature of government itself. Government is the primary enforcer of contracts. It uses its monopoly on the legal use of force to create the possibility of private commitment. There is no one, however, whose job it is to guard the guardian. The government cannot make commitments because it always has the possibility of changing its mind, and earlier ‘agreements’ cannot be enforced.

The inability to make commitments causes another set of inefficiencies: the cost of creating next-best credibility-enhancing mechanisms. While those in government at one date cannot commit future governments, they can affect the transaction costs of reversing the policies.¹ Public choice scholars such as James Buchanan (1975, 1991) have argued that the Constitution represents a form of commitment, since it increases the costs of some policy reversals such as those pertaining to civil rights. My analysis suggests three extensions to these arguments: first, there are a range of forms of actions which affect transactions costs and thereby make change difficult; second, these transaction costs, by making change difficult, can impede what would appear to be Pareto improvements; and third, while some transaction costs may have been desirable for those who created them, changes in the world may make them obstacles to efficient reform.

The issue of commitment is especially important in establishing the compensations which are frequently associated with Pareto-improving policies. For example, in the United States an elaborate set of government arrangements keeps the price of milk well above its competitive market price. An attempt to expand this legally sanctioned cartel--like arrangement was pushed forward under the euphemism of ‘self-help,’ but the cartel arrangement’s lack of budgetary cost and nice-sounding title did not make it any less objectionable. A cartel by any other name is just as odious. The annual welfare cost is huge and would have been even greater under the proposed expansion. Poor children are hit especially hard – the higher price of milk, for instance, seriously erodes a substantial fraction of the value of the government subsidy for Women, Infants, and Children (WIC) and for school lunches.

It should be possible to eliminate this cartel and still be able to compensate dairy producers with direct non-distortionary payments that leave them better off. Dairy farmers would be reluctant to agree to this change because the direct compensation is more visible than price fixing, and thus more vulnerable to political pressure for cuts later on. Even if the government promised to maintain the dairy payments, in the absence of a commitment mechanism it is unlikely that the dairy industry would believe those payments would be as politically secure as price fixing, especially if they can get away with a sweet-sounding name such as ‘self-help.’ Capitalising the value of the cartel profits and receiving that amount as lump-sum payments at the termination of the program eliminates that risk but introduces a new one: the industry cannot commit itself not to try to reinstate the cartel at some later date.

Another example is provided by our attempt to rationalise the U.S. air traffic control system, and to institute congestion pricing. The failures of our current system have moved from amusing stories – government computers that still use vacuum tubes that have to be purchased in Poland – to real fears of an inability to handle the demands of the coming decades. We devised an effective set of reforms that included user fees more closely reflecting ‘market’ prices. But the owners of corporate jets and small planes –  who currently get close to a free ride – were an effective lobby in stopping these reforms, because they realised that the move to a more market-based system would inevitably entail their having to pay their fair share. There was simply no commitment that we could make that had any credibility.

The limitations on the ability to make commitments are reflected too in how Congress goes about its business. Congress recognises, for instance, that gridlock caused by local interests might prevent military base closings, although the country as a whole would benefit from these closures. It set up a base closure commission, committing itself to vote the entire set of recommendations of the commission up or down. Similarly, Congress has recognised that special interest pressures would, under normal procedures, make ratification of trade treaties as negotiated all but impossible – and that foreigners’ recognition of this would make trade negotiations all but impossible. Accordingly, Congress has repeatedly passed ‘fast track’ legislation, committing itself to vote trade agreements up or down, without amendments.

2. Coalition formation and bargai-ning

The second hypothesis for the failure to gain near-Pareto improvements is based on the theory of coalition formation and bargaining. One of the strongest objections to those who believe that Coase-style bargaining will reach efficient outcomes is that with imperfect information, such bargaining often results in suboptimal outcomes (Farrell, 1987). To convey information about bargaining resolve, fallback positions, and so on, there is often recourse to inefficient signals. Bargaining in life is also not a one-shot episode. Each round affects the fallback position for the next. Although the two sides in, say, a labor dispute may not fully realise it, each is solving a complicated dynamic program problem with uncertainty and imperfect information. Strikes are a manifestation of ‘bargaining’ inefficiency in the private sector; the failure to enact Pareto improvements is a manifestation in the public.

One vivid example comes from the attempts to reform one of the most glaring inefficiencies in our environmental laws, the legal framework for dealing with toxic wastes: Superfund, as it is commonly called. For toxic waste sites with more than one responsible party – the vast majority of targeted sites – transaction costs represent 20 percent or more of the total cost to responsible parties (Probst et al., 1995). For insurers the numbers are even higher. According to a Rand study, only 12 percent of total insurance company outlays have been spent cleaning up toxic waste sites, while 80 percent has been spent on their clients and their own legal fees (Acton and Dixon, 1992). These seem like unnecessarily large transaction costs. Surely there must be an alternative which can benefit the environment, provide strong incentives not to pollute in the future, and have economic benefits today, with only the lawyers being worse off.

We were convinced that such an alternative existed, and carefully crafted an approach that incorporated a new legal framework and clean-up standards, combined with an effective way of disposing with the myriad of outstanding insurance cases. We worked hard and long with environmentalists, the affected businesses, and the insurance industry to forge a consensus that was not just sound and fair, but perceived that way by the affected parties. That consensus helped the measure sail through the House Committee on a 40-3 vote, a rare display of bipartisan unity on such a controversial subject. It failed, however, to be enacted before the midterm elections in 1994.

When the new Republican Congress arrived, the grand coalition fell apart. The business community believed that they could get a better bargain; in fact, all that they got was continued stalemate. They realised that they were in a new dynamic bargaining game and that the solution that worked for the previous game might be improved upon. It was, they hoped, no longer an equilibrium. Each side was holding out – very much like a strike, where massive amounts of resources are wasted while waiting for a resolution of the bargaining problem –  in part to demonstrate resolve, in part in the hope that the political dice would, in the coming years, roll in their favor.

The awareness of the dynamic nature of the bargaining game has further repercussions. Legislation can help crystallise some groups, and attenuate the strength of others. It affects the coalitions which are formed, and thereby the outcomes of political processes. Participants in the political game today realise this, and hence actions which in the short run might look like a Pareto improvement can look far riskier from a long-term, dynamic perspective.

3. Destructive competition

Superfund also illustrates another explanation for the failure of Pareto improvements to be enacted. In market economies, we are used to extolling the virtues of competition. Yet we recognise that in the absence of perfect competition, sometimes competition can be destructive. In imperfectly competitive markets, firms can get ahead not just by producing a better product at lower costs, but also by raising the costs of their rivals (Salop and Scheffman, 1983). Destructive competition is most prevalent in zero-sum games where the gains of one are at the expense of another. Political games, with position to be won or lost, are particularly prone to this kind of behavior. Competition in political markets is far from perfect, and the scope for destructive competition is therefore all the greater.

4. Uncertainty about the consequences of change

Finally, imperfect information can create an impediment to mutually productive bargains. In some models of the stock market, no trading takes place because of information asymmetries. By indicating one’s willingness to sell at x, an informed seller shows that the stock is really worth less than x (ignoring risk aversion); buyers, knowing that the seller would only sell if they were overpaying, thus refuse to trade. These information asymmetries limit trade even when differences in risk preferences and circumstances might, with symmetric information, lead to mutually advantageous exchanges (Akerlof, 1970). The reason is simple – the buyer is never sure whether the seller is willing to sell because of inside information which lets the seller know that the buyer is overpaying, or whether there are grounds for a mutually beneficial exchange.

Similarly in politics, there is often a generalised scepticism about proposals offered by an adversary that leads politicians to think that anytime an adversary makes a proposal, it must involve the adversary benefiting at their own expense. This scepticism derives not just from the standard asymmetric information in economic models, but also from the fact that many people lack the training or patience to understand the consequences of policies.

There is a certain sense in which the ‘zero sum’ view of the world is true. Competition for electoral votes or House seats is truly zero sum: if the Democrats gain, then the Republicans lose. If your objective is simply vote maximisation, then you will not find any opportunities to cooperate. Furthermore, if you define your gains in relative terms – how much I won compared to how much another group won – then any game turns into a zero-sum game: my relative gains are your relative losses. The fact that the political game – with a winner and a loser – is zero-sum leads many politicians to see all of the world in a similar vein. The fact that information is imperfect and the games we play are often not transparent means that there is always uncertainty. Your gain may indeed be at my expense.

But policy, as opposed to politics, is not zero sum. Some policies are Pareto improvements. One of the hardest tasks of economists is to explain this – a task made all the more difficult by much of the political rhetoric. Nowhere is the problem greater than in the area of international trade, ironically an area where economists have the most developed and convincing arguments for mutual gains. The main political argument for free trade is that it creates jobs. The Administration even had an official number that was used to convert the value of exports into a number of jobs. Unfortunately, this rhetorical justification could be easily turned against us, and it was. At one Congressional hearing, a Senator asked me if each $1 billion of exports created 20,000 to 25,000 jobs, then would it not also be true that each $1 billion of imports cost America 20,000 to 25,000 jobs. I had to restrain myself from pointing out that our imports are probably more labor- intensive than our exports and thus that each $1 billion dollars of imports probably ‘cost’ even more than 25,000 jobs.

What I did point out, however, is that the economic justification for free trade is not that it creates jobs – this is a matter for macroeconomic policy combined with flexible labor and product markets – but that it allows us to take advantage of our comparative advantage, resulting in higher wages and lower prices (Congress of the United States, 1997). In contrast, the rhetorical link between trade and jobs puts the discussion squarely back in the zero sum framework. Similarly, when we look to our trade deficit as a measure of our success, we are again forced back into the zero sum framework – for the world’s trade surpluses by definition must be zero. It is perhaps ironic that the one area in which economists have true opportunities for Pareto improvements is the one which the political process most often looks at from a zero-sum perspective.

The uncertainty about the consequences of policies has an important implication: complicated policies and arguments have little place in political discourse. The public has neither the background nor the patience to digest a complicated message, so this ‘simplicity constraint’ makes it more difficult to put together politically appealing reforms which are Pareto improvements. For academics, this is a hard pill to swallow: we pride ourselves in the subtlety of our arguments, not in their obviousness, in the cleverness of our solutions, not necessarily in their simplicity. For analysts of government behavior, this ‘simplicity constraint’ is hard to model, but this makes it no less real. We note, however, that the simplicity constraint can move: a few years ago, public discourse involving extended discussions of adverse selection and moral hazard effects would be unthinkable, yet today they are commonplace.

While political rhetoric may contribute to our problems, and the necessity of keeping it simple may make finding Pareto improvements all the more difficult, the manner in which so much decision-making occurs – the secrecy, the midnight committee meetings –  exacerbates our problems. In the next section I will discuss steps we can take to build a climate of openness and transparency that allows good policies to be recognised and promoted.

Secrecy vs. openness in decision-making

Secrecy aggravates the government failures identified above. First, it makes it harder to establish credible commitments. Those excluded by secrecy from the process feel fully justified in trying to change the outcomes. Moreover, proposals arrived at in secret are less likely to have paid due attention to the concerns of those left out, and in doing so they have increased the incentives of those groups to overturn the results should an occasion to do so arise in the future. Second, secrecy aggravates the problem of positional goods and destructive competition. It short-circuits the consensus process and makes it more likely that outcomes will lead to a greater divergence between winners and losers. Third, by making information scarce, it contributes both to the perception and reality of asymmetrical information, and puts into play a dynamic which is more likely to lead to biased and unrealistic information.

In a world of secrecy, you will always suspect that some interest group is taking advantage of the secrecy to advance their causes over yours, to steal, if not directly from you personally, more broadly from the public. Why else the secrecy? There is plenty of evidence to support these anxieties: the special tax provisions put into every tax bill at the last moment are perhaps the most glaring example.

This penchant for secrecy extended into the Clinton administration. Why, I wondered, was there such a focus on secrecy in a Democratic administration, in a democratic society? We were no worse than previous administrations, and I suspect we did better than most, but why couldn’t we do still better? Whatever our position, eventually it would have to be debated in the open, in Congress. If our positions were well thought out, then surely we would be able to withstand pressures from the special interest groups that would, in any case, eventually be mobilised. Should a committed democrat (or Democrat) believe that it requires stealth to advance policies that are in the national interest?

The one argument that may have some merit is that hiding information may sometimes provide a tactical advantage in the political bargaining game. But my own experience is that all too often, secrecy is neither justified by national security interests, nor as a prerequisite for rational and thoughtful debate, nor even as a tactical necessity in a broader strategy, but rather, secrecy serves as a cloak behind which special interests can most effectively advance their interests, outside of public scrutiny. There is an old expression that sunshine is the most powerful antiseptic.

Incentives for secrecy

Other forces besides special interests and the fear of exposing policy mistakes help maintain the climate of secrecy. Secrecy creates rents, because the hidden information is potentially valuable. Whenever a valuable commodity exists, markets are created. Those on both sides of the market have an incentive for continuing with the artificially created scarcity. In this market, there are at least two active parties: government officials and the press. Their exchanges do not directly involve money, but they are just as real. A reporter who gives good coverage, walking carefully the line between pandering and honest reporting, gets access to ‘leaks.’ Part of the exchange was an occasional puff piece. We could tell which reporter was in the hands of which administration official.

If all that was at stake was an occasional puff piece, these would be innocent gift exchanges. But, at least from my vantage point, all too often the reporting was distorted, the world seen through a particular lens, not the balanced kind of reporting that is required for informed public decision-making. Less secrecy would not only increase the flow of information; it would reduce the rent-creating and rent-seeking activities which lead to a distorted flow of information.

Ironically, one of the most powerful arguments for secrecy was premised on the continued existence of secrecy itself. When the press got wind of an intra-administration dispute over a policy issue, it would often turn the issue into a big story, writing about it as if the administration were confused and divided. Although these were mostly big stories only for the cognoscenti inside the Beltway, they were enough to worry many White House officials. Their answer was ever tighter control to ensure that further stories did not appear. Since it was impossible to entirely eliminate articles about intra-administration disputes – the incentives to leak were too powerful –  this approach just increased the cost of the stories that did appear. If instead they had increased the number of stories – increased transparency – they would have found that the press and public would come to a better understanding of the deliberative process or, more likely, simply become too bored to raise much of a problem.

Expertise and democratic values

There is another arena in which democratic processes and rational decision-making seemingly come into conflict, in which the resolution is not so apparent. For a large number of issues, expertise is required. This is, of course, the case in the running of any complicated business. Good managers either have the expertise themselves, or know how to hire it. It is not apparent, however, that the political process sorts well for those who have these abilities. Indeed, I was struck by the non-scientific tone of political discourse; since ‘expert’ arguments could not be well evaluated by the electorate, they had little play among those whose focus was on the electorate.

In recognition of this problem, we have established independent agencies in many areas to move critical parts of decision-making at least slightly further from the political scene. Ultimately, there is political responsibility for the performance of these agencies, such as the Federal Trade Commission or the Securities and Exchange Commission. Deciding how far to remove what decisions is a key issue in a world of increasing complexity.

Conclusion

Having spent so much time looking at the implications of information problems for market failure, it was natural for me to address the issue of explaining government failures through an analysis of information problems there. In this article, I have stressed the difficulties of achieving Pareto improvements. What I have really shown is how hard it is to construct these Pareto improvements amidst the problems of commitment and the dynamic bargaining games that characterise the political process.

Those who said that I would leave the White House with a more jaundiced view of the role of government were only partly correct. While special interests do often dominate over the general interests and while seeming near-Pareto improvements are often resisted, these failures do not undo the great achievements of the public sector, from mass education to a cleaner environment. These failures should focus our attention on re-examining both how and what the government should do.

Making government processes more open, transparent, and democratic, with more participation and more efforts at consensus formation is likely to result not only in a process that is fairer, but one with outcomes that are more likely to be in accord with the general interests. Maybe eventually we will be able to bring Coase to the public sector, so that Pareto improvements will actually be adopted.

References

Acton, Jan Paul, and Lloyd Dixon 1992, Superfund and Transaction Costs, RAND Institute, Santa Monica, California.

Akerlof, George 1970, ‘The Market for Lemons: Quality Uncertainty and the Market Mechanism,’ Quarterly Journal of Economics 84(3): 488-500.

Buchanan, James M. 1975, The Limits of Liberty, University of Chicago Press, Chicago.

Buchanan, James M. 1991, Constitutional Economics, Basil Blackwell, Oxford.

Congress of the United States 1997, ‘Hearing before the Joint Economic Committee: The 1997 Economic Report of the President,’ February 10, Government Printing Office, Washington.

Farrell, Joseph 1987, ‘Information and the Coase Theorem,’ Journal of Economic Perspectives 1(2): 113-29.

Greenwald, B., and J.E. Stiglitz 1986, ‘Externalities in Markets with Imperfect Information and Incomplete Markets,’ Quarterly Journal of Economics 101(2): 229-64.

Olson, Mancur 1971, The Logic of Collective Action, Harvard University Press, Cambridge, Mass.

Probst, Katherine, Don Fullerton, Robert Litan, and Paul Portney 1995, Footing the Bill for Superfund Cleanup, The Brookings Institution and Resources for the Future, Washington. 

Salop, Steve, and D. Scheffman 1993, ‘Raising Rivals’ Costs,’ American Economic Review 73: 267-71.

Vickers, John, and George Yarrow 1988, Privatisation: An Economic Analysis, MIT Press, Cambridge, Mass.
 
Notes:

1 One example of this effect in action comes from Vickers and Yarrow (1988).  They argue that the British government deliberately sold shares in some privatised enterprises at below market prices to a broad spectrum of the population in order to create a forceful constituency which would resist renationalisation.

About the Author:
Joseph Stiglitz is Senior Vice President and Chief Economist, World Bank, Washington, D.C. He was a member of the Council of Economic Advisers from 1993-95, and chairman of the CEA from 1995 through February 1997. He is on leave from Stanford University. This paper is an extract from the Distinguished Lecture on Economics in Government delivered to a joint session of the Society of Government Economists and the American Economic Association in Chicago, Illinois, on January 4, 1998, and published in the Journal of Economic Perspectives for Spring 1998; reprinted by permission.

Policy is the quarterly review of The Centre for Independent Studies. For more information on subscribing to Policy, click HERE

If you are interested in the Centre's activities and publications, why not subscribe to e-PreCIS, our regular email update on the latest news and events.

(e-PreCIS requires html capable email facilities, such as Microsoft Outlook Express or Netscape Messenger)