Analyzing Policy:

Introduction to the Analysis of Choice and Management

Michael C. Munger

Duke University

Table of Contents



Section I: Policy Happens When We Decide to Decide

1. An Overview of Policy Analysis as

a Profession and a Process

2. Deciding How to Decide: "experts,"

"the people," and "the market"

Section II: Why is Policy Analysis Important?

3. A Benchmark for Performance: What is a "Market?"

Case #1: A Simple Exchange Economy

4. Evaluation and "Market Failure":

Criteria for Intervention

5. Experts and "Advocacy":

The Limits of Policy Analysis

6. Democracy and "Government Failure":

The Limits of Choice by the People

Case #2: A Simple Command Polity

Section II: How is Policy Analysis Conducted?

  1. The Welfare Economics Paradigm:



    Introduction to Analytics

    8. Choice of Regulatory Form:

    Efficiency, Equity, or Politics?

    Case #3: Social Security Crisis

    Section III: Cost-Benefit Analysis

  3. Discounting I: Probability, Expected
Values, and Risk

10. Discounting II: Time

11. Cost-Benefit Analysis

12. Conclusion



Human societies seem complex. People who study them spend a lot of time comparing differences, and trying to figure out why people do what they do. This is the task of the social sciences, including psychology, sociology, political science, economics, and history. Even though the social sciences, in their current form, have only existed for about a century, we have made lots of progress toward understanding many pieces of the puzzle of human behavior.

There is another social science profession which draws on some of the best features of the main disciplines listed above. This profession is called policy analysis. In addition to studying what people actually do, and why, policy analysts establish benchmarks for evaluating those actions. More simply, policy analysis tries to answer two practical questions: First, what would be the best thing to do? Second, what is the best result that can actually be achieved? This book is an introduction to both questions. We will examine the way policy analysts conceive of the "best" policy choice, using the concepts of efficiency and equity. We will also look at the limits and constraints that restrict real policy choices.

As with any nascent science, some of the news I will be reporting will be bad. At times, it may appear to the student that real policy analysis is impossible. But donít be dismayed, because good policy analysis is not impossible. It is just hard. I hope the difference will become apparent. In any case, let me be clear from the outset: there is no more important task, in any of the social sciences, than identifying what a good policy would look like. With some idea of what is good, we have some hope of knowing bad, or mediocre, when we see it.

We can start with some broad alternatives. Though governments, and nations, are complicated, the fact is that there are only three ways for humans to organize activity:

It appears that something like the market form of organization dominates almost every society in the world. Most activity, after all, takes place in a setting where government acts to limit, rather than to direct, human choices. Markets happen almost spontaneously, requiring only a rudimentary set of property rights definitions, some reasonably accepted medium of exchange, and a legal system for adjudicating disputes.

In the absence of government provision of this institutional "infrastructure" for markets, something mysterious can happen, something not well understood from a scientific perspective. Without any central plan or direction, a dizzying array of conventions, informal rules, or other means of reducing the costs of transactions emerge spontaneously. These institutions for facilitating exchange may be imperfect, but they are neither haphazard nor ineffective. The market places of the ancient Egyptians or Aztecs, the "law merchants" of the middle ages, and the present day diamond markets of Antwerp and New York are all examples of the spontaneous emergence of institutions which reduced transactions costs and permitted an wide variety of welfare-enhancing exchanges which would have been impossible without markets.

On the other hand, the fact that effective "private" institutions might emerge spontaneously, if we wait and have faith, is hardly proof that we should always count on the deus ex machina of the marketís invisible hand. Sometimes, as in Italy in the 8th and 9th centuries, or Russia in the late 20th century, the organizations that emerge in the absence of government control are far from optimal. Incentives are pathological, outcomes are disastrous, and people lose faith in markets.

To avoid such missteps, every government in the world manages its markets. In deciding how to set and monitor these regulations, decision-makers rely for guidance on either authorities (the "experts" in all but the most primitive societies) or democracy in choosing the form that management should take. Some societies manage markets very intensively, taking over many of the marketís functions in broad sectors of citizensí activities. Others rely on something much closer to the laissez faire, or "let (the people) do (as they will)," ideal of classical liberalism. Few nations stay with precisely the same set of policies regarding the management of markets for very long, adjusting and "reforming" their regulatory policies almost every year.

Policy analysis helps explain why are there so many differences in the way we organize productive activity. It also helps us judge if some policies are good, and others are bad, by creating benchmarks for comparing the results we observe in one country, or at one point in time, with others. These benchmarks are hypothetical, because perfect policies are never observed, but without a standard of comparison evaluation is impossible.


My goal is for the reader of this book to take away two, apparently conflicting ideas about policy analysis: optimism and skepticism. The conflict between these two perspectives drives much of the debate we see in Congress, on talk shows, and in the newspapers.

Optimism can mean one of two things. Market-oriented optimists tend to believe that that the world we live in is the best of all possible worlds (a variant of the view espoused by Liebniz, but mocked by Voltaire in Candide). Regulation-oriented optimists tend to believe that all will turn out well, and expect the best, rather than the worst, outcome from any attempt to redirect market activities in a direction that will make everyone better off.

Neither of these two extreme forms of optimism is satisfactory, as a guide to policy choice. As Voltaire said through his artless naïf, Candide, optimism of the laissez-faire sort "is the mania of maintaining everything is well when we are wretched." (1977; p. 73). On the other hand, many government policies harm, rather than help, but optimists keep trying anyway to reform and improve policies. As Hegel(1832) said, "What experience and history teach is thisóthat people and governments never have learned anything from history, or acted on principles deduced from it."

So, the first theme of this book is simple: the best policy analysts are skeptical optimists. They believe, in principle, that better policies are possible, but any particular proposal with a strong presumption of doubt. More simply, a good policy analyst hopes for the best, but assumes the worst unless her skepticism is disproved. The use of analytic techniques described in this book, including models of incentives, discounting with respect to time, and discounting to account for uncertainty, are all crucial tools for someone who wants to evaluate, and usually reject, the claims of optimists.

Another theme, obvious from the title as well as the sequence of chapters, is that I believe there are three conflicting sources of power and legitimacy in policy processes. The power bases are (1) markets, (2) politics, and (3) expertise/authority. Policy conflicts tend to be dyadic, with some pair of these decision bases clashing over the right course of action. The result is that a policy which answers a conflict between politics and markets may appear irrational or harmful when viewed by policy experts. Policies that reconcile expert-market conflicts may be politically infeasible. Consequently, the first step in understanding a particular policy is to understand what sort of conflict gave rise to the policy in the first place. We will consider each of the three power bases at some length in later chapters.

Policies and Institutions

Suppose we could establish that some kinds of results, the "outcomes" of choosing policies for managing markets and directing resources, are better than others. Then we have to face the fundamental dilemma: how should we choose how to choose? Are good outcomes more likely if we rely on the institutionalized judgments of experts, or the aggregated voices of citizens, expressed through votes or surveys? This book describes advantages and drawbacks in using expertise and democracy in choosing outcomes, as well as giving the reader some important tools for deciding whether a particular attempt at management, or "policy," is good or bad.

If it sounds like "policy analysis" has too many meaningsÖwell, it may! Policy analysis defies easy definition, because it is hard to pin down as belonging to any one social science discipline. The simple reason is that no one discipline has all the answers. In fact, no one discipline even has all the right questions, and asking the right questions is the heart of good policy analysis. Policy analysts get blamed for asking hard questions, because they are skeptical. They worry about how much a policy will cost, who will pay those costs, and whether the benefits promised by the policy will ever be delivered. They worry about unintended, and unanticipated, consequences of policy choices.

On the other hand, being a policy analyst is one of the best jobs there is. Providing advice to key decision-makers is scary, but it is also exhilarating. Rather than just sitting in a room thinking about theories, or analyzing data, policy analysts have to know about every aspect of the policy being considered, from the theory which got it started to the history of similar programs in other states, or other countries, to the process of implementation and evaluation of the nuts and bolts of the policy itself. At the end of the day, elected officials and appointed bureaucratic leaders have to make decisions. If you the policy analyst donít know the answers, or have the information they need, they are going to make decisions anyway. Worse, decision-makers under pressure may end up listening to some crack-pot who has even less information, but lots more confidence. Everything you know, and everything you think, adds up to make you a better policy analyst. Good luck!

Aims of the book

This book is targeted at undergraduates in public policy programs or political science departments, as well as masters programs in public policy or public administration. programs. There are three distinct goals in these courses:

Acknowledgements: I tried out many of the arguments, and examples, in this book over more than a decade of teaching policy analysis and public administration classes at the University of Texas at Austin and the University of North Carolina at Chapel Hill. While there were far too many students with helpful suggestions to name, I should acknowledge just how useful their responses and comments have been in teaching me about policy analysis.

Several faculty colleagues have been particularly helpful in guiding me toward useful ideas, and away from foolish ones. Perhaps the most influential are Duncan MacRae, at UNC, and Murray Weidenbaum, at Washington University. Their views of what constitutes good policy are very different, but each in his own way has a level of insight that leaves me feeling both admiration and envy. Further, each conducts himself in a way that sets a high standard for intellectual discourse. Michael Ensley and Melissa Harris provided first rate research assistance at several parts of the process of writing and rewriting, as well as finding obscure information that had eluded me completely. Rebecca Lewis read several chapters, and helped me clarify the ideas I was trying to get across. Brooke Barton read the whole manuscript, and made many useful suggestions for clarification. Jay Hamilton and Brian Sala were the most helpful readers of all, carefully criticizing the whole project as well as many of its parts. Thanks to all.

Finally, for Kevin and Brian Munger: Yes, we can go outside now, and play baseball. I get to pitch first, though.

1. An Overview of Policy Analysis as a Profession and a Process

Policy: (1) Political wisdom or cunning; diplomacy; prudence; artfulness; (2) wise, expedient, or crafty conduct or management; (3) any governing principle, plan, or course of action.

Analysis: (1) A separating or breaking up of any whole into its parts so as to find out their nature, proportion, function, relationship, etc.; (2) A statement of the results of this process; (3) An examination of the relations among variables; (4) The tracing of things to their source; the resolving of knowledge into its original principles.

(Websterís New Universal Unabridged Dictionary, New York: Dorset and Baber, 1983)

Policy analysis is as old as policy. Everyone has an opinion on policies; everyone wants to believe their conclusions are the result of careful and objective analysis. But they are often wrong. As William Fulbright said, "We are handicapped by policies based on old myths rather than current realities." Fulbright was referring to U.S. foreign policy, but the same could be said about welfare policies, prison reform, or dozens of other problems. You may have noticed the problem: on most policies, most people think the answer is obvious. But they sometimes disagree about what the "obvious" answer is.

One wonders, of course, if Fulbrightís "myth" is really just what someone else believes. Oneís own beliefs, by contrast, are "the truth." Clearly, that is no basis for having a debate: we have to be able to say more than you have your opinion and I have mine. We have to say what is right, and what is wrong, or at least make a good guess. Are there good policy analysts? How would we know one if we saw one? Shakespeare praised Henry V as many things, but not least was old Henry a policy analyst of the first rank:

Hear him debate of commonwealth affairs,

You would say it hath been in all his study:

List his discourse of war, and you shall hear

A fearful battle rendered in music:

Turn him to any cause of policy,

The Gordian knot of it he will unloose,

Familiar as his garter; that, when he speaks

The air, a chartered libertine, is still."

"Unloosing the Gordian knot" is a perfect metaphor for the practice of policy analysis. After all, "analysis" comes from a Greek word meaning to loosen, or to break up a complex whole into parts that can be understood.

The problem is that, for most causes of policy, the Gordian knot resists loosening. In Greek legend, Gordius was a King of Phrygia who tied an intricate, incomprehensible knot. But the knot was of great importance, because the Oracle at Delphi revealed that whoever untied the knot would be the future king of all Asia. Alexander the Great tried, for a time, to untie the knot. Before long, though, he gave up and angrily whacked at the thing with his sword. The rope fell in pieces, Alexander went on to become king of the whole region, and now we all rush out to elect politicians who grossly oversimplify our policy problems. Blame Alexander when Ross Perot says, "Listen! Itís simple!"

The point is that we admire those who can "cut the Gordian knot." This is just a colorful way of saying such a person can come up with a quick, simple solution for what seems to be a complex problem. As we shall see, however, this is a fundamental problem in a democracy, because the same sword appears to "work" for all knots: just cut the darned thing, and get on with it. But every knot is a little different, and if you want to untie the knot, to understand it, you have to have a process for simplifying complex problems, for focusing on parts to get an idea of the nature of the whole.

Unfortunately, many of our difficulties, as a city, a state, or a nation, will not be solved with by impatient knot-whacking. Shakespeare was giving Alexander a not so subtle jibe, because King Henry "unloosens" the knot, actually solving the problem, rather than just slicing through it, which gets rid of the knot but ruins the rope.

And there you have the essence of policy analysis: we try to understand the knotty problems of policy, rather than propose simple, universal solutions. This means that policy analysts are not very popular with politicians. Political leaders often feel they were elected to make problems go away, fast, and get on to the next task. A politician elected for a two year term is not likely to have to patience to want to appreciate the uniqueness, depth, and difficulty of the policy being analyzed.

Consequently, there is a natural conflict between elected officials, who (probably rightly) see themselves as the delegates of the people, and policy experts who think they know what is right because of an abstract, and in some cases highly developed, theory of cause and effect. All politicians, not just Ross Perot, would like to apply a simple, universal principle, because politics rewards the ability to offer consistent reasons and explanations. It is tempting to think that one should ignore all the complex "analyses" of experts, and get back to underlying fairly simple reality.

On the other hand, expert policy analysts often think of problems in odd ways, because they are reconceiving the problem as they analyze it. Someone familiar with the problem, or with the politics of coalition-building, may find the recommendations of policy analysts ludicrously naive. For example, in the mid-1980s, some analysts who worked for the Social Security Administration, and some Congressional staff analysts, realized that many elderly people would not be able to afford hospital stays in the event of a long-term illness. Because of the federal budget deficit, it was decided that the government could not help for free; some charge for the "service" would have to be levied.

For most people, with annual incomes under $30,000, the charge would have amounted to about $4 per month. For those who had high incomes, a maximum fee of $67 per month would be charged. The "advantage" of the program (or so the analysts thought) was that mandatory membership among the elderly would make the "catastrophic coverage," as it was called, a very effective, inexpensive insurance program with benefits that far exceeded the fees paid by most seniors.

Well, as soon as the Medicare Catastrophic Coverage Act of 1988 was passed, there was a firestorm of protest. Surprisingly, the protest did not come from young people, whose Social Security taxes would have been subsidizing the below-cost benefits received by the elderly. Instead, it was the supposed beneficiaries, the people the policy analysts thought they were going to help, who raised a ruckus. Seniors were not much concerned with hospital stays; they wanted the government to help pay for long-term care, in nursing homes. Further, the fact that the program was (a) mandatory, and (b) not free, was very upsetting to senior citizens on very tight, fixed budgets.

The Act was repealed, in its entirety, in 1989. The very people who appeared to be helped by Catastrophic Coverage were the ones who killed it. Not surprisingly, the politicians caught in this politically misguided policy imbroglio were furious at the analysts. It seemed as if the analysts had no idea what real people actually wanted or thought. They just used abstract theory as a guide to what should be done, based on what a "rational" person should want.

The Process of Policy Analysis

Policy analysis is a process. More specifically, policy analysis is the process of assessing, and deciding among, alternatives based on their usefulness in satisfying one or more goals or values. Generally, the policy recommendation is made by one person, a decision is made by another entity, and enforcement of the policy actually chosen is left to yet a third person or group.

For the sake of clarity, letís consider an example. Suppose you are a policy analyst for a governor of a state in the U.S. Your state has a problem: There are nearly twice as many prisoners as there are normal beds in the state penitentiaries. Up until now, the situation has been handled by putting in temporary cots, so that prison rooms designed for two inmates hold three, or even four, people. But that has to change: A federal court has ordered that prison overcrowding in your state has to be reduced within 30 days.

The governor asked you to do an analysis, and make recommendations. It is useful to think of the process of analysis as having five parts:

(I) Problem Formulation

(II) Selection of Criteria

(III) Comparison of Alternatives

(IV) Consider Political and Organizational Constraints

(V) Evaluation

Letís consider each briefly.

I. Problem formulation: The analyst must create a statement of the policy problem to be addressed. The "problem" often includes, at least implicitly, a theoretical model of causation. This stage includes the listing of available alternatives. One important task for the analyst is the redefinition or creative restatement of the problem in such a way that new alternatives become available. Often, the context that gives rise to the analysis is different from the analytical problem formulation, if the analyst is doing her job right.

In our prison-overcrowding example, the context is simply too many prisoners per room. But the problem formulation may be elusive, and controversial. Are there too few prisons? If so, more should be built. Are there too many prisoners? If so, what is needed are programs to improve economic opportunities, educate under-privileged children, and rehabilitate prisoners to reduce recidivism. There are other possibilities, of course, including a mismatch of sentencing practices and type of crime, or prisons that are too comfortable and attractive to provide deterrence.

The point is that the analyst has to be careful. By saying, "The problem is that ____", some alternatives are advantaged, others impaired, and still others ruled out completely. Psychologists call this "framing," and it has been shown that the way one frames the question may determine the type of answer you get. Later, we will consider the difference between objective analysis and advocacy; for now, just remember that the problem formulation, or the way the policy is conceived or framed by the analyst, has a lot to do with the direction the analysis takes.

II. Selection of criteria: "Criteria" are the bases for judging or choosing. The word comes from the Greek krites, a judge. Once you have selected a way of framing the problem, this immediately suggests some alternative solutions. Criteria are the premises for analysis, for saying that one alternative is better than another. The difficulty, of course, is that one alternative may be superior in satisfying one criterion, but another alternative is better in other ways. Consequently, in the criteria stage the analyst must first specify the list of criteria that are relevant, and then define the trade-offs, or relative importance, of those criteria.

This is in many ways an ethical problem, rather than a purely analytic one. Criteria, and their relative importance, are statements about the ethical world-view of the analyst, or of the society the analyst serves. There have been many attempts to guide analysts in choosing criteria. The clearest, yet most useful, guidelines come from MacRae (1993), and MacRae and Whittington (1997). They argue that criteria should themselves satisfy five meta-criteria, which can be paraphrased as follows:

To continue our example of prison overcrowding, imagine that we select four criteria:

(a) cost (where less is better, all else equal)

(b) justice (the sense that the incarcerated person is neither mistreated nor coddled, but is being appropriately punished)

(c) space (immediate prison beds freed up by the policy choice)

(c) recidivism (fewer repeat offenders is better)

III. Comparison of alternatives. The notion of "alternatives" implies, by definition, at least two courses of action. In terms of our example, we could: This conception of alternatives implies that the choice must be mutually exclusive: the selection of one course of action rules out the "alternatives." One immediate difficulty with this idea of discrete choice is that the actual choice could be some combination of these policies. For example, we could build one small new prison, rent a little prison space, and focus most of our attention on the rehabilitation program.

The problem, of course, is that the choice among alternatives (whether the choice is mutually exclusive, or some combination) requires some assumptions about the trade-offs between criteria. No question about it: this is the hard part. The framing of the problem, identification of alternatives, and the choice of criteria, are all functions that trained analysts can handle. But the choice of one alternative, one policy, requires a knowledge of the values of the polity. Tradeoffs now have to be explicit, because it often happens that one alternative does well on one criterion, but some other alternative is preferable on other grounds. How can we know how much better is enough?

There is no magic answer, but a good place to start is the "criteria/alternatives matrix," or CAM. As you can see in Table 1.1, this matrix is a way of organizing the process of analysis. The CAM sets up a comparison of the performance of alternatives in satisfying different criteria. Alternatives are columns, and criteria are rows.

[Table 1.1 about here] What goes into the cells in the CAM? It depends on the kind of information you can get. More specifically, the CAM cell entries depend on the types of measures you can find about the way that the alternatives satisfy the criteria. In general, there are three levels of measurement: To compare, you have to be able to aggregate the good and bad aspects of each alternative, and determine which is best overall. If there are n different criteria, what you are looking for is the alternative with the highest rating on the criteria, given the relative importance you have assigned those criteria. In equation form, that goes something like this:


a terms: These are weights. They give the relative importance of the criteria in the decision process. That is, if ai =0, then criterion "i" is not important. If ai <0, then the criterion is a "bad" in the decision. An example of such a criterion would be cost: usually, more cost means the alternative is less preferred. If ai >0, the criterion is a "good." In our example, justice would be such a criterion.

V terms: These are the values of the alternatives measured by satisfaction of the criteria. It is worth saying again: these values must have some mathematical meaning for the CAM approach to work. Categorical values are not "measures;" they may be important, but they cannot be shoehorned into this approach. Donít do it, and donít let other people do it in your presence!

To clarify the abstract content of equation (1), letís go back to our prison overcrowding example. Suppose you had filled in Table 1.1 as shown in Table 1.2, using estimates from published studies from other states, interviews with experts, and some of your own educated "guesstimates."

[Table 1.2 about here]

Now, before you can continue, you have to assign weights to the criteria. You could assign equal weights (e.g., ai=0.25 for all four criteria, though with a negative sign for "cost"). You could assign all weight to one criterion ("space" is the only concern, so that aspace=1, and all the other weights are zero). But what is the right thing to do?

This is not a technical question, but an ethical one. For decisions to be accepted as legitimate (at least in democracies), the idea of trade-offs needs to capture the relative weights placed on the criteria by the public. The dilemma of choosing a set of weights is a problem of democracy, or "social choice," which will be considered at length in Chapter 6. For now, I will simply point out that the choice of weights in the CAM is almost exactly the same problem as the use of a voting procedure to select policies. In both cases, information is required about voter preferences and values. This information might be acquired through surveys, public meetings, voting, etc., but it has to come from somewhere.

Suppose you decide on equal weights. Then the results of your analysis might look like the first panel in Table 1.3. Note that all the "ratings" are negative, because cost is large, and has a negative weight (i.e., more costs are bad). But we can still judge the highest rating: it is for the "Early Release" program. Early release is not very costly (at least in terms of the governorís budget!), and frees up lots of beds immediately. There are recidivism problems, of course (in this example, nearly three times as high as the other alternatives), but it looks like early release is the way to go. You can go to the governor now, and make your recommendation.

[Table 1.3 about here]

Or can you? Recidivism is a real problem; maybe you should give it more weight in the decision analysis. How about if we try a weight of 40% on recidivism, and 20% each on the other criteria? That set of weights, [.2, .2, .2, .4], is not so different from the "equal weights" [.25, .25, .25, .25] used earlier. Disturbingly, we get a different answer. "Rent prison space" is now the best alternative, by quite a bit.

Letís make some observations about the choice of weights.

  1. The decision about the best alternative may depend on the choice of weights on the criteria. To the extent that this is true (and it always is, in any decision process where the answer is not so obvious that analysis is a waste of time), policy analysis using the CAM is weak. More precisely, you are only as sure about your conclusion as you are about the weight assignments. You had better have good reasons for choosing one set of weights over another.
  2. The units of the measures are extremely important. Without warning you, I have used "millions" as the units for costs, "thousands" as the units for beds, an ordinal scale for "justice," and "percentages" as the units for recidivism. This amounts to a weighting scheme, all by itself! For example, notice that the "justice" scale only can change 2 units, as it goes from 3 ("best") to 1 ("poor"). That means that, in terms of units, justice "counts" (at most!) as much as a $2 million difference in costs. The implied substantive justice difference between "best" and "poor" may be large, but a $2 million cost difference is relatively small. Why? Why should they count the same? They shouldnít! We should have been more careful to standardize units, so that the criteria are counted on a more equal basis. The problem is that it is not clear how to "standardize," unless we know the weights!
  3. The CAM approach is much better conceived as a means of organizing a decision, rather than being a decision itself. The discipline imposed by laying out alternatives, coming up with criteria, and then deciding on weights to describe the trade-offs among criteria, is very useful. But if we had stopped after the first round of measurement in Table 3.1, we would have been making a mistake.
I donít mean to be too negative in my description of the CAM approach, because it really is useful. Often, analysts go something like the procedure I have discussed without calling it a "criteria-alternatives" matrix, or even realizing they are doing a formalized analysis at all. The main value of writing down the matrix, and thinking hard about the cell contents, is that one can identify the sensitivity of the "answer" the analyst comes up to small changes in assumptions, or to difficulties in quantification. Further, if you can get good, accurate measures of the levels of satisfaction of the criteria for each alternative, the CAM approach is a genuinely scientific approach to decision analysis. I have seen groups of people, ranging from faculty deciding on curriculum reforms to a collection of scientists debating technical aspects of hazardous waste reform, change their minds about the best course of action after doing a CAM analysis. It really works, because of the structure it imposes on the decision process, and the discipline it imposes on the analyst to reveal and justify assumptions about trade-offs in values.

IV. Consideration of political and organizational constraints.Once the analyst has decided on a policy recommendation, there are two kinds of "constraints" that come into play. These constraints have to do with the acceptability of the policy to other participants in the process: (a) Political feasibility--Will elected officials vote for the proposal and make it law? (b) Organizational feasibility--Will appointed officials support the law and implement it in a way that makes its success possible?

It is tempting to merge step IV into step III, adding political and organizational feasibility as new criteria in the CAM. That would be a mistake, however, for two reasons. First, these are not criteria for judgment, in the sense that more or less of the quality is a "good" for the policy. Remember: policy analysis starts with a benchmark for the best policy, and then considers how real policy may be different.

Political and organizational feasibility are constraints, not criteria. If the legislature, city council, or other relevant elected body wonít convert the policy proposal into law, then it is time to stick a fork in the thing: itís done! Likewise, bureaucrats charged with implementing the policy are fully capable of foiling an otherwise well-designed initiative, either by a lack of enthusiasm or actual sabotage.

Second, the analyst generally cannot get accurate information on the likely reaction to a proposal until it is proposed. Predicting the reaction of elected officials, or bureaucrats, to a proposed policy is at best an inexact science. As we will see in Chapter 6, there is some basis for predicting how legislators may vote, but lots of things can happen to change the mind of policy-makers.

In general, there is only one way to ensure that politicians and bureaucrats are more likely to favor, or at least not oppose, a policy. It seems obvious, but it is often overlooked: get them involved from the beginning. There are at least two places where their participation is most important, at the stage of framing the problem, and at the stage of selecting alternatives. Since politicians and bureaucrats have different kinds of veto power over many policies, it is important to get their views on each alternative before you go too far down the road toward selecting that alternative.

For example, if one of your alternatives is the elimination of a cabinet level agency, with the associated budget transferred to another agency, you have to recognize that you are activating powerful forces whose interest is in maintaining the status quo. You will need to anticipate their response, and offer the affected employees, managers, interest groups, and their elected supporters in Congress some compensation. It doesnít matter if your policy is the "right thing to do" from an analytic perspective, because the costs of transition to your ideal policy will be all the affected parties can see.

On the other hand, even if you do everything right, you may run afoul of the feasibility constraints. Even if you asked questions, invited comment, and thought everyone had said they would support you, when the policy proposal is actually introduced, it is as if you are starting over. Carefully crafted compromises, and detailed, interdependent portions of the proposal may be selectively changed, modified completely, or reworded in such a way that your goals for reform are no longer met. Once the proposal leaves the analystís desk, it is a whole new world.

Consider the misadventures of the Clinton administration with "health care reform." Health care and the economy had been the two central issues of the 1992 U.S. presidential election. After the election, the administration began a series of meetings focused on developing a proposal for reform. The goals of the reform effort were complex, and the kind of information available for real decisions was just not available. The first lady, Hillary Rodham Clinton, was appointed by the president to head the so-called "Health Care Reform Task Force," a large group of analysts and departmental representatives with a stake in the outcome of the reform process. Another central figure in the process of developing a policy proposal was Ira Magaziner, a Rhodes Scholar turned business consultant.

What happened over the next 18 months was a case study in how to not to do policy analysis. Listen to Woodwardís (1995) description, which is worth quoting at length.

Magazinerís strategy on health care reform had been to run two tracks of briefings. First, he held a series of briefings attended by [the entire Task Force]. For the second track, Magaziner met with just the president and Hillary aloneÖOne day at one of the larger group meetings on the fourth floor of the Old Executive Office Building, Bob Boorstin [another staff member] was summoned awayÖMagaziner said not to worry. The meeting was not that significant. He was having regular private, confidential meetings with the ClintonsÖA regular series of secret meetings was explosive, Boorstin felt. "If they find out, youíre a dead man. Youíre digging your own grave. Not only will the policy be dead, but it will be dead for less than a good reason. It will be because these people [from eight different federal agencies, all affected by health care reform] feel cut out." Magaziner ignored Boorstinís warning. The private briefings for the Clintons continued. (pp. 188-189).

[In a meeting a year later, in August, 1994, Treasury Secretary Lloyd Bentson] was disturbed that health had not been subjected to the collegial deliberative process of the economic plan [now passed into law, in modified form], but was handled back channel with Magaziner trying to keep all the information to himself. He argued that the resulting plan was not politically attainable in CongressÖMagaziner felt that by proposing a plan that was left of center, Clinton would eventually win something in the center. But Magaziner did not fully grasp that the health plan, no matter how wonderfully configured or academically sound, would get dropped in the same caldron as the economic plan. The Congress, the media, and the interest groups would all have their spoons in the brew stirring at his creationÖ(pp. 372-373, emphasis mine).

[President Clinton said "I am] like the captain of a ship," grasping for the metaphor of a very old ship with oars. "That is, I can steer it, but a storm can still come up and sink it. And the people that are supposed to be rowing can refuse to rowÖ." In the fall of 1994ÖCongress killed any and all proposed health care reformóno compromise, no half measures, none of the meagerest changes. Nearly two years of work went down the legislative drainÖ.Clinton, his wife and Magaziner had designed an unwieldy monster of a planÖ(p. 389, p. 398, emphasis mine).

In devising a health care proposal, analyst Magaziner seemed to believe that there were only three steps in the policy process, ending with the proposal itself. By denying a role in the process to others with a stake in the outcome, he had doomed the whole enterprise. The Clintons, by allowing an analyst to dominate the process of drafting the proposal, seemed to forget that all policies eventually have to satisfy the legislature who must vote on the program, and the bureaucrats who must implement the program. The key to success (and I am not saying itís easy!) is to get information about political and organizational feasibility built into the proposal itself.

V. Evaluation of the program as it is passed by the legislature and implemented by the bureaucratic agency, given steps1-4. MacRae and Wilde (1985, pp. 266-268) point out that there are three types of accountability in social processes:

1. Market accountabilityóCitizens decide whether they like the goods and services provided by firms. Firms that fail the "market test" (i.e., do not provide quality service, deliver inferior quality goods, or charge exorbitant prices) will not survive, as citizens express their dissatisfaction by taking their dollars elsewhere. Market accountability is a decentralized process, with significant subjective elements, because it relies on the aggregation of the tastes of consumers choosing products.

2. Political accountabilityóCitizens decide whether they like a particular policy or activity. If citizens oppose the policy, they can use the "direct" democracy (such as a referendum) or "indirect" democracy (vote in elections where representatives are chosen) to express this opposition. If enough (often, but not always, a majority of the citizens) oppose a policy, it will be changed. Political accountability may be either decentralized (if primarily a grass-roots, citizen-led phenomenon) or centralized (if organized by the president, or leaders of the legislature). Furthermore, it can be just as subjective as market accountability, as voters decide whether they like, or donít like, a policy. That is, the reasons why a voter doesnít like a policy donít matter: if enough oppose the policy, for whatever reason, the policy is dead.

3. Expert analysisóExperts use some objective, scientifically grounded, set of criteria to decide if the policy or activity is a success. Many of the tools we will consider later in this book, such as cost-benefit analysis, discounting to account for time, and indirect valuation, are intellectual devices for analysis by experts. However, unlike both market accountability and political accountability, expert analysis purports to be objective. It may be centralized, in a "blue ribbon" panel or commission. Or it can be decentralized, as experts debate policy consequences in refereed professional journals and at conferences.

Consider how these three types of social accountability would work in an everyday example. Are "Beanie Babies" a good product?

Now, it should be pointed out immediately that there has been no "Ban Beanies!" movement among policy experts. Further, if there were, such a recommendation would clearly fail the political accountability criterion above (i.e., the experts would be overruled by politicians, for fear of being trampled to death by waves of Beanie-loving moppets).

But we evaluate things all the time, by some combination of market test, political test, and expert analysis. Most of the time, we donít think much about it, but sometimes it matters. The choice not to evaluate using expert analysis is tantamount to selecting the status quo (usually the market outcome), or risking subjective and perhaps poorly informed change (resulting from the vagaries of democratic politics.) Still, I will use evaluation in the narrower sense of expert analysis, because that is the usual sense of the word in the policy analysis literature.

In short, evaluation is a different process from the policy analysis performed in problem formulation, selection of criteria, and comparison of alternatives. Evaluation is an analysis of whether the program or policy that was implemented has achieved its stated goals. The analyst may use some of the same information, and ask some of the same questions, as in the original analysis, but in evaluation the set of alternatives, and the scope for creative problem redefinition, are sharply circumscribed.

There have been many descriptions of "types" of evaluation within the policy analysis literature. The one presented here is adapted from the discussions of Trisco and League (1978) and Bingham and Felbinger (1989). Before giving the general format for an evaluation, it is useful to offer a qualification: There are many types of programs, and many dimensions of programs that can be evaluated. You have to know what kind of evaluation you are performing if you are going to complete the job successfully. To clarify these differences, consider Trisco and Leagueís typology of "levels of evaluation," which have to do with the types of objectives being evaluated

Types of Evaluations:

Purely Formal Evaluation: monitoring daily or routine tasks. Are contracts met? Are budgets balanced? Are procedures followed? An answer of "yes" doesnít mean that the program is a good one, but "no" almost certainly means trouble.

Client Satisfaction Evaluation: performance of primary functions. Do employees understand who is the client, or customer, for the program? Are clients satisfied? Again, passing this type of evaluation is not necessarily a sign that the program is successful, but failing this test is a bad sign.

Outcomes Checklist: satisfaction of list of measurable desired outcomes. How many of the programís objectives have been met?

Expense and Effectiveness: cost-benefit analysis, measuring both the costs and impact of the program. Was the program cost effective? What would have happened to the target population in the absence of the program?

Long-term Consequences: impact on the core problem, rather than symptoms. Has the program, in the long run, achieved its goals of curing the social problem it was designed to address? If the cure was not achieved, is the program worth continuing? Notice that a program might "pass" most of the other types of evaluation, yet still be found wanting in the long run.

All of these types of evaluation are important, but the first three are relatively straightforward to carry out. These evaluations (formal rules, client satisfaction, and outcomes checklist) can be accomplished by the careful collection of data using surveys and interviews, and a study of the rules and procedures that are supposed to govern the programís functioning.

It is with the fourth and fifth types of evaluations (cost-benefit analysis, and the long-term evaluation of program effectiveness) that we will be concerned later in this book. These are the types of evaluation where analysis, of both the theoretical and statistical types, are most important. And it is for these types of analysis that the following process of evaluation is most appropriate.

Process of Evaluation:

1. Identify the goals and objectives of the program or policy in a way that makes measurement, and evaluation, possible.

2. Construct an analytic model of what the policy or program is expected to accomplish. This model will embody a set of theoretical propositions about means-end relationships, based on the accumulation of knowledge from research on the subject.

3. Develop a research design capable of distinguishing (a) what is expected from the program (goal), (b) what is actually observed (data), and (c) the range of outcomes that might be observed if the variation of outcomes is simply random, or unaffected by the policy (null hypothesis).

4. Collect the data, or actual measurements that describe the phenomena of interest.

5. Analyze and interpret the results. In particular, do the data imply that actual performance is at or above the goal? For example, suppose the goal was the improvement in some measurable indicator (such as prison recidivism). Is there a difference observable in the data? Is this difference statistically significant, in the sense that it is unlikely to have been generated by chance variation?

Review of the Process of Analysis:

At the beginning of this section, I claimed that the "process of analysis" can usefully be divided into five parts:

(I) Problem Formulation

(II) Selection of Criteria

(III) Comparison of Alternatives

(IV) Consider Political and Organizational Constraints

(V) Evaluation

We considered a specific example for each of the first three steps; that example was a decision about making a recommendation to your boss, the governor of a state facing a court order to make its prisons less crowded. The political and organizational constraints may simply overwhelm any policy recommendation you make, of course. For example, if the court order says that the prisons have to be less crowded by tomorrow, then "early release" might be your only option. You would have to stay up all night coming up with criteria for the classes of criminals, or types of crimes, that would indicate releasing that inmate is likely to do the least damage (on average) to the stateís citizens. And you wonít be able to perform step V, "evaluation," until enough time has passed to decide what the effects of your decision have been.

Most of the rest of this book will be devoted to elaborating parts of this process of analysis and evaluation. Donít worry if portions seem unclear now; this chapter has served only to give an overview of what will be developed at length in later chapters. This chapter will close with a little perspective on the profession of policy analysis in the broader context of expert advice to decision-makers generally.

Policy Analysis as Part of a Broader Profession: Expert Advice on Policy

Because policy can be analyzed in many ways, from a variety of theoretical and ethical perspectives, it is useful to consider policy analysis as part of a broader profession: experts providing advice to decision-makers on policy. In some cases the experts may themselves be decision-makers, but the two functions can usefully be analytically separated. Expert advice to policy-makers is concerned with the development of a community of experts in a variety of fields to aid public policy choice.

This field of policy analysis involves research regarding the design of policy alternatives and the decision to choose one among several possible alternatives. Broadly construed, "advice" includes both policy relevant research such as basic scientific or social science research (theory about causal connections, empirical research about the magnitudes of the cell entries in a particular CAM) and research designed to improve the way policy is designed, either in terms of its ethical or procedural basis (metapolicy analysis) and specific institutional design questions.

Consequently, expert advice on policy could encompass both an political scientistís research on committee decision making and an environmental scientistís research on the implications of lead levels in sumac leaves on a gradient away from an interstate highway. Clearly, the nature of "expertise" is difficult to define, because the term is used in many ways. The discussion given by MacRae and Whittington (1997) may be the most useful for present purposes.

The quality of "expertness" here refers to a claim by a specially trained group to contribute knowledge or advice. an expert groupís claim of authority can be based both on their collective training and on group membersí quality control over one anotherís work. Disciplinary definitions of expertise are not, however, transferred to EAP because most policy problems extend beyond the domain of any one discipline. Moreover, the extension of analytic capacity to citizens and the media can blur the boundaries of "expertise." (MacRae and Whittington, 1997, p. 12). In this view, then, an "expert" is someone whose advice is likely (at least on average) to improve the outcomes of a policy choice, from the perspective of the society. This is clearly a much broader definition than would be implied in a discussion of "policy analysis" as conceived in this book. In the broader sense MacRae and Whittington intend, an expert is anyone who has something useful to say, because of their high level of training or experience.

I will go with a narrower definition: An "expert" in policy analysis is someone who has mastered the techniques presented in this book. Now, mastery is quite different from the level of understanding one can achieve from reading an introductory text, but the goal of the remaining chapters will be to expose the reader to the basics of policy analysis. Policy analysis is that portion of expert advice devoted to the choice among specific policy alternatives.

The process of policy analysis is primarily the gathering of data and the measurement of various values achieved by different alternatives. Policy analysis is ideally performed when a problem is first recognized, but is more commonly done only when the problem becomes acute. Expert advice, in the broadest sense, may be sought at any time, and may be performed quite independently from any specific "problem" or policy concern. Ideally, EAP assists decision-makers by providing courses of action that prevent the more crisis-oriented kind of policy analysis from ever happening!

Plan of the Book

The rest of the book develops and extends the themes we have glimpsed in this first chapter. Chapter 2 describes some sources of conflict between the three institutional sources of accountability already described: markets, politics, and expert analysis. The next section examines the performance of each of these institutions separately. Chapters 3 and 4 consider the performance, and potential failures, of markets. Chapter 5 outlines the difficulties, and advantages, of focusing on expertise as a basis for policy decisions. Chapter 6 is a brief introduction to the performance of governments and the limits of collective choice.

We then move to the nuts and bolts of how policy analysis is practiced. The "welfare economics paradigm," in the many ways the methodological center of policy analysis, is the subject of chapter 7. Chapter 8 identifies three conflicting metacriteria in the choice of regulatory form: efficiency, equity, and politics. Chapters 9, 10, and 11 introduce the concept of "cost-benefit analysis, first giving a description of discounting for risk (in Chapter 9), discounting for the rate of time preference (in Chapter 10), and then presenting examples and techniques of cost-benefit analysis in Chapter 11. Chapter 12 then offers some caveats, and other perspectives, on cost-benefit analysis in particular and policy analysis in general.

As I noted earlier, this introduction will hardly make you an expert in policy analysis. But for students interested in finding out what policy analysis really is, or for experts in other fields who need to learn how policy analysis is performed, this book will prove very useful. And I promise that we will try to have some fun along the way.

Key Concepts

1. Problem formulation 2. Selection of criteria 3. Comparison of alternatives

4. Political and organizational constraints

5. Evaluation

1. Categorical 2. Ordinal 3. Interval

4. Ratio

1. Formal 2. Client Satisfaction 3. Outcomes Checklist

4. Expense/Effectiveness

5. Long-term Consequences

Homework Questions
  1. Choose a "problem" that concerns policy makers. State the problem situation (for example, "Our prisons are overcrowded") clearly, and then give at least three different problem formulations (for example, "Criminals donít fear imprisonment enough," or "Underprivileged people have no economic opportunities, and so turn to crime." Notice that the problem situation simply claims that we have a problem, while the problem formulation makes an implicit argument about why we have a problem. How would you go about deciding which of your problem formulations is better?
2. Consider the following criteria/alternatives matrix, which describe the problem of which health plan to choose at your new job:
A. Aaronís 



Fly-Buy Knightís

Health Service




no deductibles

on hospital stay


$1,000 deductible on each hospital stay

no fixed fee; if you get sick, you pay $250 per office visit, plus $5,000 per hospital stay
Waiting Period

For Appointments

1 week
6 weeks
No waiting

Suppose you expect to get sick enough to require an office visit to the doctor 2 or 3 times per year, and have never had to go to the hospital. What additional information would you need to choose a health plan? In particular, what weights would you assign to the three criteria in the CAM? Choose a set of weights that would imply that A. Aaron is better, and another that makes Fee-for-Service preferable. Which set of weights seems more plausible to you? Why?

  1. Give an example of a policy variable that can best be measured categorically. What sort of analysis can be performed on such variables? If we assigned each category a numerical value, what would this mean? For example, suppose you find the following information in a report: Trees in a public park were counted and recorded, with an oak tree assigned a "1", a pine tree assigned a "2", and a birch tree assigned a "3." The average value of all trees in the park is a 2.05; if you took a walk in that park, would you expect to see lots of pines, or no pines and lots of oaks and birches?
2. Deciding How to Decide: "experts," "the people," and "the market"
  Government is a contrivance of human wisdom to provide for human wants. Men have a right that these wants should be provided for by this wisdom.

(Edmund Burke, Reflections on the Revolution in France, 1790).

What should government do? What do you want it to do? What does everyone else want? Do the desires of the people have anything to do with the "wisdom" that Edmund Burke cites? How should we discover what is wise?

Policy analysis is mostly about the last question: Which policy is wise? That is, given a proposed policy, or a policy actually in operation, what tools of analysis can guide us in identifying and evaluating the effects of the policy? How can experts improve what government does? What policies can experts propose that will improve the functioning of markets?

Of course, policy analysis does not take place in isolation. Instead, there are many groups in the political process who think that they know what wisdom is. These groups may have little use for "experts" or "evidence." Furthermore, as was argued in the previous chapter, the market has a logic, and a kind of wisdom, of its own. These three kinds of wisdomómarkets, democracy, and expertsóare constantly in conflict. This chapter considers the sources of this conflict, and how it affects the practice of policy analysis by experts.

To understand the nature of this conflict, it is useful to consider the overview depicted in Figure 2.1. To the casual observer, it often seems that policies are in conflict with one another, or that policy-makers are just confused. This may be true, but the panoply of polices we see can only be understood as ways of reconciling the various conflicts depicted in the figure.

[Figure 2.1 about here]

We can consider each pairwise conflict separately:

Experts v. MarketsóAs we will see in later chapters, markets have an internal logic of operation. However, markets can fail, sometimes spectacularly, without management and support of experts. While there is debate about the extent of intervention required, there is substantial agreement about the categories of intervention that can improve market performance. Although in principle an expert could have any sort of goal one can imagine, I am going to use the term "expert" in this chapter to mean someone who has the goal of improving the functioning of politics or markets. What is meant by "improve" should become clear as we proceed, but I am intentionally leaving the goals of experts unspecified for now.

These categories of "market failure" are (1) "market structure" regulation, including management of natural monopolies (such as utilities), or anti-trust policies to control concentrations of economic power, (2) policies to control externalities, such as pollution, through regulation or internalizing costs, (3) the systematic undersupply of public goods, due to the related problems of "free riding" (if privately provided) or "demand revelation" (if publicly provided, but financed by tax shares revealed by the individual), and (4) information problems, such as drug approvals or physician licensing, to reduce the costs of fraud or simple confusion.

The conflict between markets and experts will be analyzed in Chapters 3 and 4. We can call these policies "efficiency" policies, because the guiding value is efficiency. While we will consider efficiency much more deeply later on, it is worth pointing out that efficiency, for the policy analyst, is defined a little differently than you might expect. The dictionary definition of "efficient" includes: "producing the desired effect or result with a minimum of effort, expense, or waste; working well; competent; able; capable." We will use the following definition, applying the concept of efficiency to the allocation of resources:

A particular matching of resources to uses is efficient if and only if there exists no alternative allocation of those same resources which results in a more desirable result.

One obvious problem with this definition is that "more desirable" is hard to specify precisely, when there are two or more people involved. Suppose person 1 likes allocation A better, and person 2 likes allocation B better; which allocation is more desirable? To "solve" this problem, we will simply ignore it: if there is disagreement, we will not be able to apply our concept of efficiency! More precisely, we will apply the so-called "Pareto criterion," which requires unanimous agreement.

To see how the concept of efficiency based on the Pareto criterion would work, imagine that both allocation A and allocation B are technically feasible, in the sense that society can marshall the resources and apply them to the uses required in those two allocations. Now, suppose person 1 likes A better, and then imagine that person 2 likes A better also. That is, all people agree that A is better than B.

Then forcing allocation B on our two person society would be inefficient. What is meant by "efficient" in this context is a lot like "wasteful." It would be wasteful to choose allocation B, because A is feasible and because both person 1 and person 2 prefer A. In this case, we say that A is "Pareto superior", or unanimously preferred, to B.

It is worth remembering, however, that efficiency is not the only consideration in expert management of markets. It is as if the analyst, in focusing efficiency, is saying, "For a given distribution of incomeÖ" That is, the distribution of gains is an important, but separate, question. As the dotted line in Figure 2.1 indicates, there may be substantial influence from political actors. In some sense, this influence is secondary, because it does not directly involve the management of markets. But political considerations may distort, or even rearrange, attempts by experts to manage markets.

Politics v. MarketsóThe nature of politics in a democracy inevitably creates a tension between markets and collective decision-making. Markets are decentralized, operate with little central direction, and (most importantly) recognize power based on wealth. If you have more dollars than someone else, you have more "votes" in the market. Democratic politics operates on a much more egalitarian basis: each person gets only one vote, so that in the ballot box, at least, the poor person and the rich person have the same power.

Consequently, conflicts between politics and markets often take the form of disagreements over the outcomes of market processes, such as the redistribution of income or the use of publicly held resources such as national forests or fisheries. In a larger sense, since markets are the engines of growth in capitalist economies, but politics is the process by which property rights, tax rates, and social programs are decided, this conflict may be the most fundamental of all. There are those who portray the conflict between collective decision-making and market processes as a battle of good versus evil, but this view is simplistic. The fact is that both markets and politics provide us with a useful way of deciding; the conflict has to be managed, because it isnít going to go away. This type of conflict can be called equity policy.

Another word for equity is "fairness." One meaning of "fair," of course, is average, as in the following scale you might find on a survey about food service on your campus:

The food at the cafeteria was ____.

Pick one: (a) poor (b) fair (c) very good

The other meaning is more complimentary: fair means honest, just, or unprejudiced. It is the second meaning that makes fair a synonym for equitable, but the first meaning sometimes creeps in when we are thinking of politics. As we shall see, some people argue that the distribution of income implied by the use of markets is not "fair," and requires political intervention. But the standard for fairness seems to be that we all get the average, with no deviations allowed. This conflict between the two notions of "fair" as definitions for equity is a microcosm of the policy conflicts between politics and markets. People who favor markets tend to think that "equitable" and "equal" are unrelated concepts, at least for descriptions of the distribution of wealth among the members of society. People who favor political redress for the inequities they perceive in markets appear to think "equitable" and "equal" are very nearly the same word.

Once again, the source of wisdom not directly involved in the conflict may have an influence. Experts may try to affect the debate in equity policy by proposing solutions, or new problem definitions. We will consider expert analysis of equity policies in Chapters 5, 6 and 7.

Experts v. PoliticsóThe way that we decide influences what is decided. As we will see later in this book, the particular "institutions" of choice, given public opinion, make a big difference. Consequently, experts often have advice for politicians on how the political process itself should be reformed. However, these reforms may not be popular with elected officials. The reason may seem obvious, but it is worth remembering: whatever else politicians may differ on, they share incumbency. Incumbents were all elected under the current system, every one of them! Consequently, efforts at "improvement" by experts may be met with skepticism, or open hostility. This final type of conflict, between experts and popular politics, can be called institutional reform policy.

This brings us to the third dotted line in Figure 2.1, the one where markets may influence institutional reforms. Depending on your point of view, this influence can be fairly benign, or completely evil. The choice of institutional form for public choices is fundamental to the functioning of any democracy. Most importantly, whatever the institutions of collective choice, it is crucial that the public perceives the government selected by this process to be legitimate and fair.

Markets, by injecting considerations of wealth and economic power into political choice, may distort choices and threaten the legitimacy on which the whole system depends. In some ways, this problem is outside the scope of policy analysis, but we will consider it briefly in Chapters 6, 8, and 9.

Letís summarize what has been said so far in this chapter: there are three bases of wisdom, or sources of accountability, in policy processes. These are markets, politics, and experts. In choosing the wise policy, there is inevitably conflict about which of these to follow. If the primary conflict is between markets and experts, the result is an efficiency policy. If the primary conflict is between markets and politics, the result is an equity policy. Finally, if the primary conflict is between experts and politics, the result is an institutional reform policy.

The State of Nature: No Markets, No Politics, No Experts

In the previous section, I argued that the context of a policy debate is the key feature in determining what kind of policy results. Different policies seem to be contradictory, but that is because they resulted from different conflicts. Efficiency policies often seem in conflict with equity policies; institutional reforms may seem to serve neither efficiency nor equity very well. Each policy arena has its own logic, where policies may seem rational given the problem the participants think they are trying to solve. From a larger perspective, of course, the whole thing may seem messy and incoherent.

Before we start thinking about trees, we should take a step back and think about the forest, and where it came from. This step back must take the form of a thought experiment: What would policy look like without any political, market, or expert context. Precisely because the context of policy analysis is so important, it is useful to imagine what things would be like if we remove context, at least as far as is possible.

In the U.S., we live in a democracy, in an advanced industrial society, with well-developed norms of choice and an elaborate system of expertise in almost every field. Suppose that none of this were true. Imagine we had only rudimentary role definitions and rules of behavior to guide us, but that we face problems of deciding what to do. How would we conduct make decisions?

Philosophers and social scientists call this idea the "state of nature," a thought experiment considering what life would be like without institutionalized markets, politics, or experts. Different people have had very different ideas about what "natural" life would be like. One of the most famous is Thomas Hobbes, who was not (to say the least) optimistic:

Öthe nature of War, consisteth not in actual fighting; but in the known disposition thereto, during all the time there is no assurance to the contraryÖ Whatsoever therefore is consequent to a time of War, where every man is Enemy to every man; the same is consequent to the time, wherein men live without other security, than what their own strength, and their own invention shall furnish them withall. In such condition, there is no place for Industry; because the fruit thereof is uncertain; and consequently no Culture of the Earth; no Navigation, nor use of the commodities that may be imported by Sea; no commodious Buildings; no Instruments of moving, and removing such things as require much force; no Knowledge of the face of the Earth; no account of Time; no Arts; no Letters; no Society; and which is worst of all, continual feare, and danger of violent death; And the life of man, solitary, poore, nasty, brutish, and short. (Chapter 13, p. 186). Hobbes recognizes that his dark portrayal of the condition of humankind in the state of nature may not be realistic ("It may peradventure be thought, there was never such a timeÖ" p. 187). But that is not his point. What Hobbes is analyzing are the conditions or conventions that enable a society to avoid the cataclysmic "war of every man against every man." (p. 188).

In Hobbesí view, societies are able to avoid the state of nature by vesting power, and the legitimate ability to focus force, in the person of the sovereign. As democratic theory has progressed, we have adapted Hobbesí notion of sovereignty into something more abstract, and at the same time more concrete. That "something" is the will of the people. This is really quite an astonishing intellectual achievement: we start with (1) a state of nature, move to (2) the person of the sovereign, who embodies the power and legitimacy of the state, and then (3) mentally divorce the literal person (king, sultan, or chieftain) from the function of that office, which is to carry out the will of the people. In this construction, the sovereign is the will of the people, which (in theory, at least) is the anthropomorphised ruler of the society.

Not everyone would accept this formulation, not by a long shot. For some, it is not the state of nature that is to be avoided, but rather the rule by the general will we should fear. Edmund Burke, speaking perhaps ironically, makes this argument most clearly:

In vain you tell me that Artificial Government is good, but that I fall out only with the Abuse. The Thing! The Thing itself is the abuse! Observe, my Lord, I pray you, that grand Error upon which all artificial legislative power is founded. It was observed, that Men had ungovernable Passions, which made it necessary to guard against the Violence they might offer to each other. They appointed governors over them for this Reason; but a worse and more perplexing Difficulty arises, how to be defended against the Governors? (Burke (1982 [1756]), pp. 64-65). In this passage, Burke would appear to argue that nature, or "natural society," may be preferable to government, since there is no way to ensure that the will of the people is obeyed. "The thing" is government; saying "the thing itself is the abuse" means that Hobbes had it all wrong, and humans in the natural state would be just fine. It is the power of unjust government we should fear, and guard against.

I am not going to pretend to offer an answer in this debate, but it is useful to think about where governments, and societies, come from. Consider Aristotleís account, from Book I, chapter 2 of the Politics:

When several villages are united in a single complete community, large enough to be nearly or quite self-sufficing, the state comes into existence, originating in the bare needs of life, and continuing in existence for the sake of a good life. And therefore, if the earlier forms of society are natural, so is the stateÖHence it is evident that the state is a creation of nature, and that man by nature is a political animal. And he who by nature and not by mere accident is without a state, is either a beast or a god. (emphasis added). This passage leads to a hard question: Is it true that "man by nature is a political animal"? If the answer is yes, then institutions of collective decision are required. More simply, humans will have to make decisions as a group that somehow embody, or at least account for, the desires of the individuals who make up the group.

Many people have worked on the problem of how to make collective choices out of individual desires. One of the most important early efforts was by Jean-Jacques Rousseau. Rousseauís thought was complex, so any attempt at brief summary will not do it justice. However, it is clearly true that Rousseau believed in the superiority of some idealized "natural" condition of humanity, where people are free to delight in liberty and nature.

On the other hand, he recognized that some mechanism is required for generating binding collective choices, and for governing the otherwise unavoidable tendency for inequalities among citizens to arise. These two ideas (complete freedom of the individual, the need for submission of the individual to the general will) are not fully reconciled in Rousseau, but he certainly recognizes the problem. In a celebrated and controversial passage, Rousseau makes his argument:

As long as several men in assembly regard themselves as a single body, they have only a single will which is concerned with their common preservation and general well being. . . .

A State so governed needs very few laws; and, as it becomes necessary to issue new ones, the necessity is universally seen. The first man to propose them merely says what all have already felt . . . .

There is but one law which, from its nature, needs unanimous consent. This is the social compact, for civil association is the most voluntary act in the world. Since every man is born free and master of himself, no one can, under any pretext whatsoever, subjugate him without his consent.. . . . Apart from this primitive contract, the vote of the majority is always binding on all the others; this is a consequence of the contract itself. But it may be asked how a man can be free while he is forced to conform to wills that are not his own. How are the opponents free while they are bound by laws to which they have not consented?

I reply that the question is not properly. The citizen consents to all the laws, even to those that pass against his will, and even to those which punish him when he dares violate any of them. The unchanging will of all the members of the state is the general will; through it they are citizens and free. When a law is proposed in the assembly of the people, what they are being asked is not precisely whether they approve or reject the proposal, but whether or not it is consistent with the general will that is their own; each man expresses his opinion on this point by casting his vote, and the declaration of the general will is derived from the counting of the votes. When, therefore, the opinion that is contrary to my own prevails, this proves neither more nor less than that I was mistaken, and that what I thought to be the general will was not so. If my private opinion had prevailed, I would have done something other than what I had willed; it is then that I would not have been free. (Rousseau, 1973, pp. 150-151).

This reasoning may seem a little tortuous, but the argument is important. As a citizen, you want government to do the right thing. However, citizens may disagree about what the right thing is. We could vote, as a way of deciding what "we" think, as opposed to insisting that we must all think the same thing. Provided each of us renders a judgment about what is best for society, rather than just what is in our self-interest, this process of voting is a means of discovering the collective wisdom, or general will. Each member of a society must agree, in the abstract, to accept specific decisions we may not agree with. Rousseau is arguing, then, that this submission to the general will is the price of freedom.

In mentally creating the general will out of some combination of individual desires, we have also created something else, the "society." In fact, the very essence of the general will is the notion that there is some group larger than the individual whose welfare we can measure, or at least compare across different policy choices.

The comparison of the welfare of the individual and the welfare of some larger entity, whether it is a rural community, a city, or the entire nation, is at the heart of many policy questions. You have probably seen media accounts of attempts by cities to expand a road system, where one old house (owned in our example by Grandma Filinda Blank) stands in the way. The city offers to pay the "market value" of the house, but the property and the family farm is worth much more than that to Ms. Blank. What should the city do? It can use its right of "eminent domain," and force Grandma Blank out of the house. Or it can spend millions of extra dollars to reroute the road, delaying completion of the highway by six months. How should we decide? What is the wise policy, in a situation where the welfare of one individual is clearly in conflict with the welfare of most, or even all, other citizens?

We will consider, in Chapters 10, 11, and 12, some potential ways to address the question, accounting for risk, time, and the use of cost-benefit analysis. While these techniques arenít perfect, they offer at least a starting point for measuring values and organizing our thinking. The set of techniques called cost-benefit analysis are some of the most widely used, and valuable, tools a policy analyst can possess. But like any potent tool, cost-benefit analysis can be misleading or even dangerous if it is used improperly.

One word of caution, however. My mother, who grew up on a farm, always said "You canít make chicken salad out of chicken feathers." (Yes, I am paraphrasing; she didnít say "feathers"). Cost-benefit analysis is a way of coming up with measures of the values of certain actions, but it canít tell you what the fundamental values of a society are, or should be. If you believe that the society has interests, and rights, that are superior to those of any one individual, or even groups of individuals, then some very aggressive actions by government on behalf of that collective interest may be justified. On the other hand, if you think that the rights of individuals are paramount, the ability of government to function in the face of disagreement may be severely curtailed. One canít rely on cost-benefit analysis to answer the really fundamental questions, or to create consensus where there is profound disagreement. Policy analysis generally works best for societies that have already agreed, among themselves, on most of the hard questions of collective choice and values.

Let us now turn back to the notion of a state of nature, and see how policies may have originated in the simplest sorts of human cultures.

The Hun-gats: Choosing and the Wisdom of the Group

Consider the simplest case of collective choice, and see if it has some lessons for larger societies. Suppose we lived in a small extended family group, or clan, of hunter-gatherers. Imagine we had only rudimentary role definitions and rules of behavior to guide us, but that we face problems of deciding what to do. How would we conduct the business of society? How would we make decisions?

Each personís membership in the clan is initially conferred by birth, or marriage. But as life continues, each person has the option to leave, to strike out on their own and abandon the "we" to live as "I." The fact that tribes stay together, and try to decide as a group, tells us the first thing we need to know about policy analysis. Policies are "public": once a policy is chosen, all of us have to live with it, whether we agree with it or not. Consequently, policy analysis is different from any other problem you are likely to encounter when making decisions. You arenít just deciding what you want, or what is best for you. Somehow, you have to figure out what is best for other people.

Letís make the example more specific. Suppose you are a member of a hunter-gatherer tribe (I will call them the "Hun-gats," as in my earlier book, Hinich and Munger, 1997) You, and the clan, have hunted (and gathered) all of the food in the area that you know is safe to eat. Now, everyone is hungry, and everyone is starting to look at other plants, strange animals, and maybe even Cousin Gombog, in a whole new way. These are all things that the tribe does not normally consider to be food, but hunger has made them define potential food sources more broadly. What "policy" should the tribe adopt about eating?

The first thing the Hun-gats would have to decide is how to decide. It is by no means obvious that the "what do we eat?" question is a collective one. So, step one is to decide whether we need a policy at all, or whether it is okay to let a more private or "market" oriented solution (i.e., each of us tries whatever we want) work itself out.

[Figure 2.2 about here]

Letís suppose the Hun-gats decide they need a policy, for two reasons. The first is the chance that lots of members of the tribe will eat something poisonous, and then die. This is a waste of effort, because more people are exposed to risk than is necessary to find out if the food is safe. This is a problem of too much risk. The Hun-gatsí ancestral enemies, the fierce Raouli tribe, would likely find out about widespread food poisoning, and use the Hun-gatsí weakness as an opportunity to attack.

Second, there is a "free rider" problem: someone found an oyster, opened it up, and then looked at his companion: "Hey, this looks like food. OoohÖwhy donít you try it?" Free riding creates a problem of too little risk: some foods would not be sampled, because everyone is waiting for someone else to try it first.

The two problems (too much risk from uncontrolled experimentation, and too little risk from free-riding) are not likely to cancel out. Instead, the two problems are both going to hurt the effort to find new foods that will relieve the impending famine. The tribe has two choices: (1) it can try to use theory, or (2) it can use empirical experimentation to determine what is safe. The sort of "theory" they might use is not based on knowledge of germs and poisons, of course. Rather, for primitive peoples theory is more likely to be religious law, applied by the use of analogy.

Suppose someone finds a beaver, and brings it back to camp. The shaman, the Hun-Gatís chief religious official, makes a policy pronouncement based on theory:

Our laws say that pigs are unclean, and you may not eat them. Beavers look like pigs, albeit with large teeth and funny, flat tails. Therefore, beavers (like pigs) are unclean. Obviously, theory may have problems as a tool for policy analysis, especially if the theory was really designed to explain something else.

The alternative might be an experiment, based on empirical investigation:

Letís roast up this bad boy, and see how he tastes. Here, you take the first bite.

If the person who takes the first bite says it tastes good, and shows no ill effects in the next day or so, then the policy is decided: Beavers can be added to the "acceptable foods" list for the tribe. In fact, the shaman may even update his theory, using the newly acquired empirical knowledge:

Animals that look a lot like pigs are okay to eat, provided that (unlike pigs) those animals have big teeth and funny flat tails. My example has been rather silly, of course, but theory and analogy can play an important role, even when cause and effect relationships are poorly understood. In fact, it may be precisely in those instances that little is known about the mechanics of cause and effect that people are likely to use "theory." This is a very conservative (in the sense of avoiding risk) approach to policy analysis, but it can work pretty well. Consider Table 2.1, which gives a comparison between the rules for handling meat in a "kosher" butcher shop, and the 1998 version of USDA procedures for doing the same tasks. [Table 2.1 about here] The similarities are striking, yet the approach to creating the "policy" in each case was dramatically different. Kosher rules were "designed," but not by scientists. Kosher rules are a set of religious practices; "keeping kosher" is a moral imperative, not a health concern. Yet it seems obvious that the kosher rules have a strong basis in science, or at least in health. In many ways, the "religious" kosher food handling rules approximate the "scientific" USDA guidelines. The USDA requirements (presumably) are based on science; their sole goal is to promote health. The USDA has the benefit of significant experience, and a well-developed germ-based theory of contamination of food products.

In a primitive society, adopting a policy like orthodox Jewish rules for keeping kosher has important health benefits. You could imagine how the policy might be adopted: people notice that if one eats chicken prepared according to the rules, one is less likely to get sick. Without a germ theory of disease, or an understanding of food parasites, the people may misattribute this sickness to the anger of spirits or deities that control our health. Consequently, "policies" about food might start out as religious rules. If the rules adapt over time to record safe, and outlaw unsafe, foods and practices, the power of the shamans will increase, because they really do appear to speak for the gods.

For example, suppose the shamans say,

Donít eat pork, as it is unclean. If you eat pork, you will die; demons will possess your stomach. You flout the policy, and eat pork in a hot environment where cleanliness and cooking procedures are (at best) inconsistent. Before long, parasitic trichinae from an undercooked piece of pork enter your intestines. Untreated, the trichinosis becomes acute, affecting your viscera and then even some of your voluntary muscle groups. You die, screaming. The shamans, shaken by the power of God to punish the wicked, redouble their efforts to ensure that Godís will is both universally known and obeyed.

This leads us to a problem: how are we to evaluate the opinions of experts, or (alternatively) the democratic decision processes that use majority rule? Our Hun-gats donít know what foods are safe to eat; they are very uncertain, and afraid. Their votes would be meaningless on this question, because they have no basis for making good choices. The shamans only have laws based on accumulated experience, codified as religious dietary restrictions. These rules do not apply to new foods, because it is not clear where new comestibles fit into the existing system of safe and unsafe foods. Without a scientific understanding of what causes disease, or of what foods are safe, we have no means of making a decision. Neither expertise, nor democracy, is equipped to handle this problem.

But being a policy analyst means you never get to say "I donít know." The Hun-gats have to do something, because their children are crying from hunger. And Gombog, who is a bulky man, is starting to get nervous at the way people stop talking, but keep staring, when he walks up. Suppose you are a consultant, whose job it is to help the Hun-gats decide what policy they should adopt for new foods. What would you recommend? What are the alternatives?

There is a general process of policy decision, which you need to understand to get through the rest of this book. We will spend quite a bit of time fleshing out this process later, but for now letís just give the outline, as depicted in Figure 2.3.

Step 1: Is there a problem? What is it?

Step 2: Should we decide collectively, or privately, how to solve the problem?

Step 3: If a collective decision is required, should we decide using democracy, or delegate policy-making authority to experts?

[Figure 2.3 about here]

The outcomes of this process have three broad categories, the figure shows. In the broadest sense, the alternatives are those listed in Figure 2.3:

(a) Allow the problem to remain outside the scope of collective control. This might occur because no one cares, or because the issue is not deemed as important. Alternatively, this "nondecision" may be an aspect of civic or religious culture: it never occurs to citizens to have government intervene. In practical terms, citizens may choose whatever action they like, subject to other restrictions the society has placed on individual action. In the case of our Hun-gats, if no explicit policy is chosen, each person may eat, or not eat, new foods.

(b) Decide the problem collectively, but choose to allow citizens to make their own choices. This seems a lot like option (a), but there is an important difference. In the first case, there is no decision, or else it just appears that the problem is beyond the appropriate scope of government action. For option (b), there is an affirmative act, an actual decision that citizens must be allowed to make their own choices, even if the collective may not like those choices. This means there has been a collective choice that the problem will be treated as if it were private. So, though in practical policy terms these two seem the same (in each case, people choose), in fact there is a world of difference (in the second case, the choice was collective, so it is much more likely to change in the future).

(c) Collective choice. Once a decision has been made to decide collectively, the collective choice is often made by democratic means, such as voting in a referendum or choosing representatives. Within the prescribed limits of the collective choice, the will of the majority (or whatever other decision rule is in use) is binding on the other citizens.

(d) Delegate authority to make a collective choice to experts. This type of outcome has different properties than the option (c), democratic choice. In some ways, it may be better, because the choice is made by well-informed, highly trained people. On the other hand, the decision may appear less legitimate, because it seems imposed. The people did not consent to the choice, as in majority rule; they delegated the choice. What this all means is that the basis of the grant of "expertise" is crucial. If expertise comes from knowledge of reading goat entrails or star portents, there are likely to be problems. The biggest problem, of course, is that non-experts have a hard time judging expertise. Experts have to judge themselves, with peer review and licensing requirements.

For most of the rest of this book, we will consider some of the implications of delegation to experts, and look at problems with markets and democracy. The technical tools we consider in the later chapters are all designed to ensure that the reader, once he or she has become an expert policy analyst, will get things right, or will recognize when others have things wrong. Before we can do that, however, it is useful to consider the biggest question that societies have to face, one that we already mentioned but did not really address in any fundamental way. That question is the choice of the public or private arenas for decision.

Public Decisions and Collective Decisions

So far, I have talked about public and private decisions as if there were clear and obvious differences. But this is hardly the case. In fact, the rhetoric of many policy conflicts is about just the problem of whether the public has any business telling individuals what to do. While I canít offer a general solution to that problem, I can point out a subtlety that is often missed in policy debate: there is a difference between public decisions and collective decisions.

The easiest way to appreciate this distinction is to consider Figure 2.4.

This "affect welfare" standard is subjective, of course. It may "affect my welfare" that you wear an ugly (in my opinion) tie, or use racial epithets, or enjoy pornographic films and books. In the extreme, "you have offended me" could be an almost universal excuse for forcing my will on others. Alternatively, the idea of "private" actions is hard to define, or sustain. I have said that marriage is a private act. Yet the potential that gay men or lesbian women may want to marry is an important political issue, because at present that right is not established. In some ways, "public" decisions are just those that other people care a lot about, which is simply subjective. Still, some objective criteria for "externalities" can be devised, as we will see.

[Figure 2.4 about here]

The difference between individual and collective decisions, by contrast, can be defined in practical, measurable terms, as a question of the power to choose.

As can be seen, there is a difference between the public-private and individual-collective conceptions of choice. Certainly, many of the decisions a society faces fall along the main diagonal (top left and bottom right boxes) in Figure 2.4: private decisions are made by individuals, and public decisions are made collectively. But that is not always true. There is nothing about the machinery of democracy that prevents private decisions from being made collectively. This result is usually identified as "tyranny of the majority" over an individual, or smaller group.

Conversely, public decisions may be made by individuals: I may choose to pollute common pool resources such as air or water, or I may undertake activities that affect others positively (such as seeking an education) without being able to capture the full gains from the activity. These "negative externalities" (if the activity is harmful to others) can effectively result in theft of value from the rest of society. "Positive externalities" (when the activity results in uncompensated benefits to others), may explain why there is underinvestment, or too little of such activities undertaken privately.

The final point worth noting about Figure 2.4 is the fact that it identifies rhetorical strategies in policy debates. In most cases, the default "policy" is the top left box: Individual decisions made on private matters. Earlier, in Figure 2.3, we saw that the first two steps in a public policy process are identification of a problem, and then a decision about whether the problem is public. The rhetoric of the debate will focus on the question of whether the problem is really more appropriately in the bottom right box, or public decisions, collectively reached. So the point of the debate is whether the decision should be made collectively, but the words of the debate will be a contest over whether the decision has public implications. It turns out that the concept of "externality" is a powerful rhetorical tool, as well as a legitimate analytical concept.


I began this chapter with a quote from Edmund Burke: "Government is a contrivance of human wisdom to provide for human wants. Men have a right that these wants should be provided for by this wisdom." Using human wisdom to provide for human wants is not easy, but it is what policy analysis is all about. In this chapter, there were three separate themes of introduction to policy analysis. It is useful to summarize each so the reader can see how they fit together.

First, as the title of the book suggests, policy outcomes are the results of policy conflicts. I argued that there are three main dyadic conflicts in the policy world, with three different types of associated policies. The reason that policy may seem contradictory or incoherent is that the results a particular policy conflict produces is an answer to a very specific question. If the goals or interests of the observer are different, then the result may be hard to explain.

The three main policy conflicts are as follows:

Second, the origins of many policies may be obscure. We tend to use thing that work, even if we donít understand quite why. Just as dietary laws had to be based on accumulated trial-and-error knowledge from the past rather than a scientific theory of epidemiology and disease, our understanding of policy from a theoretical perspective is primitive. This makes policy analysis frustrating, since so little is known about cause and effect. It also makes policy analysis exciting, because it is possible to make lots of progress very quickly on fundamental problems.

Third, it is important to recognize that there may be "policies" even when we havenít decided anything, or even when we havenít decided whether to decide. I proposed a three-stage process for conceiving the policy process:

Step 1: Is there a problem? What is it?

Step 2: Should we decide collectively, or privately, how to solve the problem?

Step 3: If a collective decision is required, should we decide using democracy, or delegate policy-making authority to experts?

In many cases, the fact that there is no formal policy is not the result of any failures in steps 2 or 3. Rather, we never got past step 1, realizing that there is a problem or agreeing on its nature. The notion that a problem exists is logically antecedent to the conclusion that something should be done. Generally, we hope that "something should be done" about what I have called "public" problems, where the actions of one person affect the welfare of others. The difficulty is that what is decided is not whether the problem is public; rather, what is decided is whether the authority to delegate power is held collectively. This contest, over what is legitimately within the purview of the society to decide for its citizens, may be the biggest policy conflict of all.

Key Concepts

Homework Questions

1. In Figure 2.4, two kinds of potential "tyranny" were identified. One of these occurred when a group decides a private matter for an individual, as when I am told how I must act, dress, eat, or think. This is "tyranny of the majority." The other type occurred when an individual imposes his or her will on the group, as when one person pollutes the atmosphere or steals public property. This is "tyranny of the individual." Obviously, societies face an important problem in deciding how to balance these two types of tyranny: giving the collective more power helps control tyranny of the individual, but encourages tyranny of the majority. Write an essay to answer each of the following questions.

a. In this chapter, we saw some philosophers (e.g., Rousseau) who thought that the power of the collective is most important, and others (e.g., Burke, at least in the quoted passage) who that that freedom of the individual is paramount. Who is right?

b. More importantly, how should society divide the public and private spheres of our lives, identifying what should be decided collectively and what should be chosen by individuals without government interference?

c. Finally, should the answer to this question be taken from economic theory, and the notion of public goods, or from theories of justice, and the good society? If neither tells us everything we need to know, then how should we approach the problem?

2. Define, and give an example of, each of the following categories of policy. Describe also the policy conflict that gives rise to such policies.

a. Efficiency policies

b. Equity policies

    1. Institutional reform policies


3. A Benchmark for Performance: What is a "Market?"

Overview of Markets

The first two chapters have described the profession of policy analysis, given a little background on the problem of making collective and private decisions, and identified the key tensions or conflicts that shape real policy debates. Now it is time to turn to the form of organization most commonly used to direct human activity: the market. This chapter, and the next, are devoted to an outline of basic economic concepts. Let there be no mistake, however: this introduction is no substitute for a real course of study in economics. Economics is central to the study of markets, and to the analysis of policy in general, so if you are serious about becoming an analyst you will have lots more work to do.

On the other hand, it is possible to give a quick, and hopefully intuitive, overview that will actually take you quite a long way toward appreciating the value of economic reasoning. Markets arenít like politics, which takes place in a particular institutional context created by rules and procedures, or expert analysis, which follows professional norms and guidelines. A "market" is a set of institutions, rules, or informal norms, that promote exchange. Interestingly, it is not unusual for complex institutions for the facilitation of exchange among owners of resources to emerge spontaneously, without any central plan or conscious act of collective creation. More simply, markets happen.

That doesnít mean that what happens is always good. Precisely because a market is likely to break out in almost any context, sometimes markets perform well and sometimes they perform poorly. Any one of the following conditions is likely to give rise to the set of exchange institutions we think of as a "market."

Preconditions for the existence of a "market"

1. Differences in goals, tastes, or desires (diverse preferences)

2. Differences in endowments of productive resources and personal talents (diverse endowments)

3. Declining average costs as more output is produced (economies of scale)

4. Declining average costs as the scope of action of one producer is decreased (specialization)

If people were clones in preferences and endowments, and production processes were linear in scale and scope, then markets would be irrelevant. It would be just as easy (efficient) to produce everything we wanted on our own. This "go it alone" kind of economic organization is called autarky, or a "Robinson Crusoe" economy, after the famous Daniel Defoe character.

The fact that markets are widely observed is a hint about their function: markets make human beings better off. Not all human beings are better off, perhaps, and some of us are helped more than others, but by and large markets improve the human condition. There are three aspects to the benefits of markets. Any action by experts, or by politics, to suppress or distort the functioning of markets can cause harm by denying citizens these three advantages. That doesnít mean that such action may not be necessary, or even beneficial. Rather, identifying the beneficial aspects of markets helps us keep in mind the nature of the costs of using regulation, or alternative means of organization.

To see how these aspects of markets work, it is useful to consider some thinkers who have studied market processes. Nearly 2,500 years ago, the Greek philosopher Aristotle made some accurate and important observations about the origins of money, and of markets, that are no less true today. Of everything which we possess there are two uses: both belong to the thing as such, but not in the same manner, for one is the proper, and the other the improper or secondary use of it. For example, a shoe is used for wear, and is used for exchange; both are uses of the shoe. He who gives a shoe in exchange for money or food to him who wants one, does indeed use the shoe as a shoe, but this is not its proper or primary purpose, for a shoe is not made to be an object of barter. The same may be said of all possessions, for the art of exchange extends to all of them, and it arises at first from what is natural, from the circumstance that some have too little, others too much. Hence we may infer that retail trade is not a natural part of the art of getting wealth; had it been so, men would have ceased to exchange when they had enough. In the first community, indeed, which is the family, this art is obviously of no use, but it begins to be useful when the society increases. For the members of the family originally had all things in common; later, when the family divided into parts, the parts shared in many things, and different parts in different things, which they had to give in exchange for what they wanted, a kind of barter which is still practiced among barbarous nations who exchange with one another the necessaries of life and nothing more; giving and receiving wine, for example, in exchange for coin, and the like.

This sort of barter is not part of the wealth-getting art and is not contrary to nature, but is needed for the satisfaction of men's natural wants. The other or more complex form of exchange grew, as might have been inferred, out of the simpler. When the inhabitants of one country became more dependent on those of another, and they imported what they needed, and exported what they had too much of, money necessarily came into use. For the various necessaries of life are not easily carried about, and hence men agreed to employ in their dealings with each other something which was intrinsically useful and easily applicable to the purposes of life, for example, iron, silver, and the like. Of this the value was at first measured simply by size and weight, but in process of time they put a stamp upon it, to save the trouble of weighing and to mark the valueÖ

When the use of coin had once been discovered, out of the barter of necessary articles arose the other art of wealth getting, namely, retail trade; which was at first probably a simple matter, but became more complicated as soon as men learned by experience whence and by what exchanges the greatest profit might be made. Originating in the use of coin, the art of getting wealth is generally thought to be chiefly concerned with it, and to be the art which produces riches and wealth; having to consider how they may be accumulated. Indeed, riches is assumed by many to be only a quantity of coin, because the arts of getting wealth and retail trade are concerned with coin. Others maintain that coined money is a mere sham, a thing not natural, but conventional only, because, if the users substitute another commodity for it, it is worthless, and because it is not useful as a means to any of the necessities of life, and, indeed, he who is rich in coin may often be in want of necessary food. But how can that be wealth of which a man may have a great abundance and yet perish with hunger, like Midas in the fable, whose insatiable prayer turned everything that was set before him into gold? (Aristotleís Politics, Book I, Section 9).

The paradox that Aristotle points out is interesting. Money is clearly not wealth, because a person with nothing but money would starve. The answer, of course, is that money represents all other commodities. The medium of exchange is a measure of value, or the unit of account, of the barters people would negotiate if money were not available. Instead of me trading you two bottles of wine for a sheep, we can exchange money. I give you $10 for a sheep. You give me $5 each for bottles of wine, and end up buying two bottles. The $10 went back and forth, to no net effect; for this reason, some economists say that money is a "veil," disguising but not really affecting the contours of trade. In our example, money was an artifice: the real exchange was the trade of wine for mutton.

Having a currency, however, reduces the frictions or "transactions costs" of the exchange, making it easier for both of us. Furthermore, if we have money the exchange need not be directly a barter between two people. If my only option is trading wine for sheep, but you donít like wine, I canít get any sheep, even if each sheep is "worth" two bottles of wine! Money breaks the dyadic relations of barter into separate exchanges, allowing me to obtain abstract command over goods and services (i.e., units of money) instead of having to take physical possession of a commodity I donít value or canít use. Money allows us to focus on an important aspect of scarcity, the "opportunity cost" of a commodity. It is worth pausing to consider opportunity cost more carefully.

Opportunity Costs and Scarcity

Using a resourceótime, gold, wood, a drill press, or laborófor one purpose means you can not use it for something else. We all want more things (more leisure, more consumption, more savings), so there couldnít possibly be enough to go around. Consequently, societies have to find ways to allocate limited resources in the face of unlimited wants. This is what economists mean when they talk about the problem of "scarcity": How do we allocate limited resources to satisfy unlimited human wants?

To understand scarcity better, I have to define a related concept, "opportunity cost."

Opportunity cost: the cost of foregone alternatives, or uses given up if a resource is used in a particular way. If a resources can be used for activity A or activity B, the opportunity cost of using it for B is the value of its use for A. In an idealized, perfectly functioning market system, the opportunity cost of a resource is greater than or equal to its price. In practice, however, the relation between opportunity cost and price has no necessary sign or magnitude. Opportunity cost can be illustrated fairly simply, but it is a subtle concept. Consider this: If you found a valuable diamond, and I wanted it, would you give it to me? Suppose you donít know me, and that I am not a member of your family, and so have no claim on your affections. Then I suspect you would not give me the diamond for one penny less than its market value, despite the fact that the diamond was "free." The price you paid to obtain a commodity is completely irrelevant to its value to you, or to the opportunity cost of using it or giving it away.

As an illustration, ponder a test question I asked in an intermediate microeconomics class while teaching at Dartmouth College. I was surprised then how many people missed it; see if you get the "right" answer.

Mental exercise on opportunity cost: What would you do?

Suppose you are a really big fan of the rock star, Mickey Martin, especially his number one hit about root canal surgery without anesthesia, "Living, But Need a Local!" You hear he is coming to town to give a concert in two weeks. Seating at the concert is "festival" style, so that there are no reserved seats: one ticket is just like another, since you have to stand in line at the gate to get seats. Tickets cost $25.00, and you want two, so you can take your friend along.

Arriving to buy tickets when they go on sale a week before the show, you see that you have underestimated how popular Rartin is in your city. The line stretches around the block. You wait in line, and move slowly toward the ticket window. When you are only about 30 people from the front of the line, with hundreds now standing behind you, you hear the ticket window slam shut, and hear a loud murmur rippling back along the line: "Sold out? SOLD OUT!" Oh, no.

Days later, back at home, you read that "scalpers" (assume re-selling tickets, or scalping, is legal in your state) are asking, and getting, $200 or more for a ticket to the show. As you walk out to your car, you hear snatches of Martinís songs from a nearby fraternity house. You think to yourself how much you want to see that show. You have more than $1,000 in your checking account, and still have no plans for the night of the concert. But there is just no way that it is worth $400 for two tickets! Heck, for that much money, you can buy a CD player, and every Mickey Martin CD ever made!

As you start to get in the car, you notice a scuffed envelope on the sidewalk. Looking around, you donít see anyone. You pick up the envelope and inside, mirabile dictu, you find two tickets to the Bingsteen show. You make a legitimate effort to see if anyone lost the tickets, but of course you canít just go ask random people, because they would say, "Yeah, sure, hand Ďem over. I lost those. You bet I did!" Deciding the tickets really are yours now, here is the question:

Do you go to the concert?

The answer, as any economist will tell you, is "no." After all, you can scalp the tickets, which means you should value the tickets at their "opportunity cost": $200 each. But we have already established that the concert was not worth $200 per ticket, at least not to you, because you had a chance to pay that price to a scalper and decided against it.

This turned out to be a bad test question: all my students missed it. It is not that I am a bad teacher: the student seemed to understand opportunity cost as an abstract concept, and could recite the formal definition without a hitch. But when it came time to apply the concept, they had no intuition. They all had some version of the same answer: "Of course I would go. Free tickets: cool!" No, no, NO! The tickets are not free, because using them costs you the $400 you could have obtained by selling them. It doesnít matter what you pay for something; the "price" is what you give up by using the resource. The real price of something is its opportunity cost. Obtaining something for free is very different from valuing that thing at zero. If you found a diamond, you would not give it away.

Having defined opportunity cost, it is immediately possible to define scarcity:

Scarce resource: A resource is scarce if its effective opportunity cost exceeds zero. Generally, this means that any resource with a price greater than zero is scarce, because opportunity cost is at least as great as price. If a resource has a zero effective price it may still be scarce, however, because property rights to the resource may be imperfectly specified. Notice that "scarce," by this definition, and another adjective, "valuable," are certainly not synonyms. Air to breathe is not scarce, for most of us, because it can be obtained for free. But it is of paramount value. If property rights to air are poorly specified (if the air is what is sometimes called a "common pool resource"), the market price of air might be zero. Yet the opportunity cost of using the air as a dumping ground for industrial waste may be far higher than zero.

One of the main goals of experts working in policy analysis is to find ways to make the price to an individual of using a resource equal to its true social opportunity cost. It is useful to state the reason explicitly, in a form I will call "Pigouís Conjecture," after A.C. Pigouís famous argument in The Economics of Welfare.

Pigouís Conjecture: Markets are inefficient, as a means of allocating resources, to the extent that the (social) opportunity cost of a resource diverges from its price, or private cost to the user. The inefficiency derives from the overuse (if social cost is greater than price) or underuse (if social cost is less than price) of the resource. To return to the air pollution example, if the cost to me of using the air to dump wastes is small, but the costs to society are large, then I will overuse (in terms of efficiency, to say nothing of justice) the atmosphere as a trash can.

The policy prescription that arises from this conjecture is the need to use a system of taxes or subsidies to make the opportunity cost of the resource equal the effective price as nearly as possible. That is, if the amount I pay is equal to the full social cost, I will take all of those social costs into account when making decisions on resource use. Consequently, I will use the resource only if the advantages to me exceed the full social costs.

There is a counterclaim that questions Pigouís Conjecture; I will call it Coaseís Counter-Conjecture," from R. H. Coaseís (1960) "The Problem of Social Cost."

Coasesís Counter-Conjecture: If property rights are clearly and exclusively defined, and the cost of writing and enforcing contracts is not too high, market forces will make the opportunity cost and the price of a resource converge. Why are these two claims "conjectures?" Because neither is proven! In fact, each claim is "true," in some deductive sense. The debate turns on the question of which approach makes better public policy: an expert-managed approach (Pigou) or a decentralized market-oriented approach (Coase). In terms of the model of policy conflict developed in the previous chapter, we are in the arena of markets vs. experts, with attempts by political actors to manipulate the outcome to their benefit. In later chapters, the bases of this debate will be considered in more detail. At this point, I will turn to an introduction to the logic and function of a market system, so that we can begin to evaluate the two conjectures.

Allocating Scarce Resources

Markets exist for a reason. Our interest in market processes derives from their importance as a mechanism for addressing scarcity. But it is worth remembering that markets are not the only means available. As a matter of institutional technology, there are, there are four major ways to allocate resources in the face of scarcity:

1. Price System (market): Resources are directed to their highest-valued use, so that whoever is willing to pay the most (either in terms of other valuable goods, or in currency) gets to control the resource. Big winners: People with lots of money, or with talents or resources the society values highly. Disadvantages: There are two. (1) Poor people may get too little, creating ethical problems of equity. (2) Independently of their basis in justice, market allocations may be politically untenable, if democratically based authority is in a position to impose redistributive or confiscatory taxes.

2. Queuing: A queue is a line. Queuing means a system of allocation based on waiting your turn. So, first in line is first in priority. If all the resource is used up before your turn, you lose out. Big winners: People with lots of time (actually, a low opportunity cost of time spent waiting in line). Disadvantages: There are two. (1) People standing in line incur lots of "deadweight losses," or time wasted, for no gain in consumption or productivity. (2) There is no reason to believe that resources are directed to their highest valued uses. As evidence, consider that queue-based allocations often evoke secondary, or "black", markets, where allocations initially dictated by queueing are reallocation by prices.

3. Chance: Lotteries, drawings, or other random selection processes mean everyone has an equal chance of winning. Big winners: No individual is a winner from the process, because in terms of expected value everyone is treated the same. From an ethical perspective, however, this may be an advantage. Disadvantage: By definition, allocation is random. The person who actually gets the resource may value it at only a fraction of its worth to someone else. Opportunity cost is explicitly ignored in random processes. Consequently, chance allocations evoke secondary markets for reallocating by price.

4. Authority/Discretion: Allocations can be made by experts, party officials, elected leaders, or central planners. This sort of allocation process is also called a "command" system. Big winners: Guess who: the party officials, their friends, and family! Alternatively, the beneficiaries of the policy may be those targeted by the policy, if discretion is used to avoid corruption and follow the rules. Disadvantage: There are two. First, it is very difficult to obtain the information required to make accurate judgments about scarcity and opportunity cost, without some mechanism for measuring intensity. In effect, command systems often fall back on queuing, as people are forced to stand in long lines for products "sold" far below their opportunity cost values. Second, command systems present almost irresistable opportunities for corruption and favoritism, because the central authority canít monitor all the officials with the ability to hold up products in exchange for bribes or in-kind payoffs.

The distinguishing feature of a market system of allocation is the reliance on prices. You are used to prices; you have dealt with them all your life. Some prices are stated in terms of goods, as when you decide whether to trade baseball cards: Is a "Ken Griffey" rookie card worth a "Mark Lemke" plus a "Dave Stieb," or would it take a third card to make the values equal? (Hint, non-sports fans: it would take a third card!) Some prices are quoted in terms of dollars, like when you go into a store and see a candy bar in a display rack. The price is on a sticker on the candy, or on a sign above the rack: "75 cents." Have you every thought about what this price means, what it really means?

The price is what you give up to get something else. If you spend the 75 cents on the candy, you wonít have it anymore. But there are other prices operating here, in ways that are less obvious but just as important. The storeowner decided to display that candy bar, and not some other; another display choice might have resulted in increased sales. The candy maker uses her machines, buildings, and distribution network to put that candy bar, and not some other one, in the stores. Just as price affects your decision to buy the candy bar, price influences the store owner and the manufacturer in their choices about what to display, and what to produce.

More broadly, in a market system, the prices of resources serve two very different, and very important functions. First, prices are signals about the relative scarcities of resources. Second, prices determine wealth by giving a concrete value to the resources owned by individuals. Letís consider each of these functions in turn.

Prices are scarcity signalsóSuppose you arenít just worried about candy bars anymore. Suppose you were put in charge of planning a very large number of activities. In fact, you must make all production decisions and consumption choices in an entire nation. What is the right set of choices? You must first allocate all of your productive resources among myriad alternative uses. Then you have to decide who gets to consume all these different outputs, and how much they get.

Start with farmland: should you grow corn, wheat, or soybeans? Should you grow cotton? Should you use the land for cows, or pigs for food, or sheep for wool to make warm clothing? After all these production decisions are made, your storehouses are full of clothes, food, and machines; how do you dole them out? What allocation rule would give you the "best" result? You could give everyone an equal share, or you could use some other rule to assess each personís needs.

Obviously, in a market system this is not the way things work, because there is no central planner who directs resources to their highest valued use. A farmer in central Illinois doesnít wait for a letter from the government to decide how to use his land. Instead, he makes forecasts of the results of different courses of action, and then chooses the one that results in the largest excess of revenue over costs. What information is available to make these forecasts? How does the farmer know what to do, if he is not given instructions from a central planner?

The farmer uses prices as a signal. Notice how cheaply, and yet elegantly, price signals the value of different resources for the society. The farmer looks at a newspaper, and sees that soybean prices (letís keep this simple, and ignore futures markets) have risen by more than $2.00 per bushel, from $6.20 per bushel to $8.25 per bushel. If he is a reflective man, he may wonder why. If he is a compassionate man, he may feel sorrow for the hardships that the scarcity of soybeans is imposing on consumers. But suppose he is neither reflective nor compassionate. All this farmer is is a terrible, selfish man, not at all given to altruistic impulses. What does he do? He immediately drives to the seed store, and plants every inch of land he owns in soybeans!

So the selfish, narrow-minded, "only out for himself" farmer hurries to do what an omniscient, benevolent central planner would have wanted the farmer to do from the perspective of benefiting the entire society. This should give you goosebumps: the selfish farmer (1) didnít need to know the source of the scarcity, and (2) doesnít need to care about the welfare of those harmed by the scarcity. All he needs to do is care about himself, and the scarcity is very effectively addressed. This feature of markets has been extensively analyzed, by F. A. Hayek (in his 1945 article "The Use of Knowledge in Society") and others, but the best-known statement of the function of prices is the earliest. In his 1776 The Wealth of Nations, Adam Smith resolved the problem of how resources are directed to their highest valued use without resort to any central direction whatsoever:

Every individual is continually exerting himself to find out the most advantageous employment for whatever capital he can command. It is his own advantage, indeed, and not that of the society, which he has in view. But the study of his own advantage naturally, or rather necessarily leads him to prefer that employment which is most advantageous to the society.

ÖAs every individual, therefore, endeavours as much as he can both to employ his capital in the support of domestic industry, and so to direct that industry that its produce may be of the greatest value; every individual necessarily labours to render the annual revenue of the society as great as he can. He generally, indeed, neither intends to promote the public interest, nor knows how much he is promoting itÖBy preferring that industry in such a manner as its produce may be of the greatest value, he intends only his own gain, and he is in this, as in many other cases, led by an invisible hand to promote an end which was no part of his intention. By pursuing his own interest he frequently promotes that of the society more effectually than when he really intends to promote itÖ

What is the species of domestic industry which his capital can employ, and of which the produce is likely to be of the greatest value, every individual, it is evident, can, in his local situation, judge much better than any statesman or lawgiver can do for him. The statesman, who should attempt to direct private people in what manner they ought to employ their capitals, would not only load himself with a most unnecessary attention, but assume an authority which could safely be trusted, not only to no single person, but to no council or senate whatever, and which would nowhere be so dangerous as in the hands of a man who had folly and presumption enough to fancy himself fit to exercise it. (Smith, 1994, pp. 484-5).

In these passages, Smith is making two rather astonishing claims: (1) In a properly functioning market system, actions based on self-interests are identical with those actions implied by altruism (ignoring problems of the distribution of wealth). (2) The net value (i.e., the benefits minus the costs of operation and transmission of information) of a price system for signaling scarcity, and directing resources, is so great that no expert-driven central planning regime could possibly duplicate it. With hindsight, and two hundred years of economic analysis of markets, we now know that there are a number of important qualifications and corrections that weaken these claims. In some cases, as we will see in later chapters, one or both claims are simply not true.

Still, the basic structure of the reasoning of Smith has stood the test of time: markets are uniquely suited to foster the organization of large numbers of decentralized production and consumption decisions. Not only is no central plan required to create this happy result, but even the most draconian repression of market processes is unlikely to snuff out a "black market" system of exchanges based on the self-interest of the participants. This is the "efficiency" argument for market processes, and it is a powerful argument indeed.

Prices Determine Wealth in a Market System

As I noted above, prices have two functions in a market system. The first, directing economic activity in production and consumption by producing signals of relative scarcity, is the one proponents of the market point to when they advocate the use of markets. What about the second function, the determination of wealth?

Wealth has two very different definitions, and it is important to keep them distinct. In a material sense, wealth might simply mean possession of a surplus of consumption goods. Robinson Crusoe, alone on his island, might have been wealthy in this sense, if he were able to provide for his own needs very comfortably by farming and by using the flotsam washed up from the wreck of his ship.

But we usually mean something different by "wealth," because we live in a society where there are markets, with production and exchange going on all around us. In this setting, wealth means the possession of command over lots of resources, and the ability to sell these resources to others, to direct their use, or to store them for use in the future. So one can no longer measure wealth by how much stuff one has. Instead, wealth is determined by the monetary value of all that stuff. And value means that we need prices.

Suppose that there are "n" different products or resources one might own, and let us use "i" as an index for some unspecified resource. Let "x" represent resources, and let "p" represent prices. Then, for example, x7 might be acres of land (suppose the person has four acres), and p7 is the price of one acre (each acre is worth $15,000). The value of land the person owns would then be x7 ´ p7 (in this case, 4 ´ $15,000 = $60,000) so that part of the personís wealth could be measured this way. Formally, total wealth can be defined as follows:

This may seem disturbing, when you think about it. If prices change, your wealth

changes, even though the amount of stuff you own hasnít changed at all. To put it

another way, you arenít wealthy unless the society you live in places a high value on the

resources you control. Most disturbing of all, a significant part of the wealth of most people is their own labor, which they offer in the market in exchange for wages. If the value of your labor is very low, you are poor. This may happen because your skills, though significant, are obsolete, as in the case of a buggy whip maker, or are in excess supply, as with a migrant farm field worker.

The reason this creates difficulty is that, in a market system, the "demand" each of us has for consumer products is based not on how much we want the product, subjectively, but rather on how much our want is made effective by wealth. For example, consider this problem: how much does a starving person want a sandwich? The answer is hard to swallow: if the starving person doesnít have any money, he doesnít want the sandwich at all, not one bit! From a practical perspective, this is nonsense, of course: the starving man is desperate for food. But market systems operate on the premise that human desires have to be made effective by an offer of wealth, of one form or another.

If the working of market processes leaves some people with too little wealth to survive, society may decide that the distribution of wealth needs to be changed. There are many ways of effecting this redistribution, but the premise is always the same: markets may operate efficiently, in the sense of directing resources to their highest-valued use. But that conception of "value" relies on the prices of resources possessed by individuals, and on the demands made effective by the value of those resources. To the extent that this definition of monetary value (based on scarcity) does not correspond to the ethical values (in terms of a theory of justice) of the society, then the distribution of wealth needs to be adjusted. The particular value which is generally invoked to justify income redistribution, from the wealthy to the poor is "equity," or the sense that all citizens are entitled to a certain standard of living.

The performance of markets in terms of scarcity-signaling (efficiency) is well-established; the performance of markets in terms of ethical values (equity) is much more hotly debated. Furthermore, just to make everything a little more interesting, it turns out that expert-driven regulatory policies to improve performance on one dimension of performance may hurt performance on the other dimension.

For example, an income tax designed to redistribute wealth from the rich to the poor can hurt efficiency, because it distorts the signal sent by effective (after tax) income. The conflict between efficiency arguments and equity arguments is one of the central themes in policy analysis, as we will see.

For the remainder of this chapter, I will focus on four specific questions regarding the functioning of markets.

Market Origins

Earlier in this chapter, I claimed that markets are likely to arise, spontaneously and without conscious decision on the part of any central authority, under any one of four conditions. It is useful to repeat those four conditions, because it is easy to forget just how general and flexible markets can be. The four conditions were:


EFFICIENCY GAINS It is the maxim of every prudent master of a family, never to attempt to make at home what it will cost him more to make than buy. The tailor does not attempt to make his own shoes, but buys them of the shoemaker. The shoemaker does not attempt to make his own clothes, but employs a tailor. The farmer attempts to make neither the one nor the other, but employs those different artificers. All of them find it for their interest to employ their whole industry in a way in which they have some advantage over their neighbours, and to purchase with a part of its produce, or what is the same thing, with the price of a part of it, whatever else they have occasion for. (p. 485). As a market system becomes more highly developed, enterprises spring up that can realize enormous economies of scale. Consider the cost per automobile if each automobile, or personal computer, had to be made separately, from a unique plan. If each of us had to rely on our own resources, we would enjoy very few of the technological marvels that now fill our homes. The reason we see these amenities, and are able to buy them at low prices, is that markets have developed to exploit the profit potential of economies of scale. To take an example, therefore, from a very trifling manufacture; but one in which the division of labour has been very often taken notice of, the trade of the pin-maker; a workman not educated to this business (which the division of labour has rendered a distinct trade), nor acquainted with the use of the machinery employed in it (to the invention of which the same division of labour has probably given occasion), could scarce, perhaps, with his utmost industry, make one pin in a day, and certainly could not make twenty. But in the way in which this business is now carried on, not only the whole work is a peculiar trade, but it is divided into a number of branches, of which the greater part are likewise peculiar trades. One man draws out the wire, another straights it, a third cuts it, a fourth points it, a fifth grinds it at the top for receiving the head; the making of the head requires two or three distinct operations; to put it on, is a peculiar business, to whiten the pin is another; it is even a trade in itself to put them into the paper; and the important business of making a pin is, in this manner, divided into about eighteen distinct operation, which, in some manufactories, are all performed by distinct hands, though in others the same man will sometimes perform two or three of them. I have seen a small manufactory of this kind where ten men only were employed, and where some of them consequently performed two or three distinct operations. But though they were very poor, and therefore but indifferently accommodated with the necessary machinery, they could, when they exerted themselves, make among them about twelve pounds of pins in a day. There are in a pound upwards of four thousand pins of middling size. Those ten persons, therefore, could make among them upwards of forty eight thousand pins in a day. But if they had all wrought separately and independently, and without any of them having been educated to this peculiar business, they certainly could not each of them have made twenty, perhaps not one pin in a day. (Smith, 1994, pp. 4-5). It is important to understand the distinction between the two types of productive efficiencies fostered by markets. Economies of scale mean products in capital intensive industries are cheaper, as more output is produced. The gains to specialization derive from the division of labor into a larger number of smaller, discrete tasks. But both economies of scale and the division of labor are limited by the extent of the demand for the product. By rewarding increases in scale, and specialization, markets tie us all closer together, and create links, first across families, then across communities, and eventually across nations, as entrepreneurs seek out larger markets.

Barter and Pure Exchange

A "barter" system is created, or may just emerge spontaneously, as a means to capture gains from trade. As was pointed out in the apples, oranges, and blueberries examples earlier in this chapter, diversity in either preferences or endowments create the surprising opportunity for doing magic: Exchange improves everyoneís welfare, just by rearranging consumption bundles. That is, allowing trade, without increasing the total quantity available for consumption, can increase the value each person places on his or her consumption.

We could start at a simpler level, of course. A true "Robinson Crusoe" economy, with just one person, is still an interesting problem. Our Robinson has complex analysis to perform, as he decides how to allocate his effort in growing, hunting, and gathering food, finding water, and building shelter. Each minute spent in one activity "costs" Robinson whatever he could have done with that minute in some other activity. Suppose he spends a month working on a garden, instead of fishing and preserving the fish he catches by salting and drying them. If the garden fails, Mr. Crusoe may starve.

But the fact that there are costs doesnít mean that there is a market. The distinctive feature of markets is their use of a price system to allocate resources and send signals of relative scarcity. As F.A. Hayek (1945) argued,

The economic problem of society isÖhow to secure the best use of resources known to any members of society, for ends whose relative importance only these individuals knowÖthe knowledge of the particular circumstances of time and placeÖTo know of and put to use a machine not fully employed, or somebodyís skill which could be better utilized, or to be aware of a surplus stock which can be drawn upon during an interruption of supplies, is socially quite as useful as the knowledge of better alternative techniques. (pp. 520-522).

ÖIt follows from this that central planning based on statistical information by its nature cannot take direct account of these circumstances of time and placeÖ(p. 524).

Assume that somewhere in the world a new opportunity for the use of some raw material, say tin, has arisen, or that one of the sources of supply of tin has been eliminated. It does not matteróand it is very significant that it does not matterówhich of these two causes has made tin more scarce. All that the users of tin need to know is that some of the tin they used to consume is now more profitably employed elsewhere, and that in consequence they must economize tinÖThe most significant fact about [market processes] is the economy of knowledge with which it operates, or how little the individual participants need to know in order to be able to take the right action. In abbreviated form, by a kind of symbol, only the most essential information is passed on, and passed on only to those concerned. It is more than a metaphor to describe the price system as a kind of machinery for registering change, or a system of telecommunications which enables individual producers to watch merely the movement of a few [prices]Öin order to adjust their activities to changes of which they may never know more than is reflected in the price movementÖThe marvel is that in a case like that of a scarcity of one raw material, without an order being issued, without more than perhaps a handful of people knowing the cause, tens of thousands of people whose identity could not be ascertained by months of investigation, are made to use the material or its products more sparingly; i.e., they move in the right direction. (pp. 526-527).

To put the point simply, then, markets are useful to societies because they can organize the activities of members of society, even in the absence of any central plan or direction. That is why even the most primitive market requires at least two people. Hayek is claiming that, especially in huge markets with thousands or millions of people, the organizing and directing function of prices is more important than one might think.

People have tastes that drive their consumption choices; exchange is perhaps the most important way for people to satisfy these tastes. To see how this works, letís consider a simple example. Everybody talks about the trade-offs between guns and butter; letís not be so boring. We will analyze the trade-off between guns (more broadly, defense) and roses (representing the arts and other cultural amenities). Consider a representative citizen in the society where roses and guns are the only products available for consumption, and assume the following four statements are true:

If these statements are accepted, we can depict a "map" of the value the citizen places on different combinations of roses and guns. More technically, we can represent the citizenís preferences, using what are called indifference curves. Pick any point in the nonnegative quadrant of consumption space, depicted in Figure 3.1. Restricting ourselves to the nonnegative quadrant simply means that the citizen must consume quantities of roses, and guns, of at least zero: there can be no negative consumption.

[Figure 3.1 about here]

Suppose, when I said "pick any point," that you picked "B." There are then three types of alternative consumption bundles in the picture. Some, like "D", with fewer roses and fewer guns, the citizen likes less than B. On the other hand, if there are more guns and more roses, the citizen likes it better, as with point "E." If there are some points the citizen likes better, and some the citizen likes less, than B, there must be some boundary between these two sets. This boundary is the "indifference curve," or the set of consumption bundles the citizens likes just as well as B. All we know for sure is that there are some points with more guns and fewer roses, and others with more roses and fewer guns, such that the citizen is indifferent between these points and B. For the sake of example, I have arbitrarily chosen points "A" and "C", and drawn the indifference curve implied by this preference profile.

The notion of an indifference curve is very useful for analyzing bargaining and the gains from trade. The fact that every point "above" the indifference curve (i.e., all consumption bundles with no less of both, and more of at least one, of the commodities compared to the points on the curve) is strictly preferred to any point on the curve, and that points "below" the curve are strictly inferior, allows us to depict problems of commodity exchange very precisely.

To see this, consider Figure 3.2. Panel A of the figure depicts an exchange problem for two people, Mr. 1 and Ms. 2. Suppose there is a total of 40 guns, and 30 roses, available for allocation for consumption for the two people, and that Mr. 1 starts out with thirty guns, five roses (letís write this "(30, 5)"), and that Ms. 2 starts with (10, 25).

[Figure 3.2 about here]

The box in Figure 3.2 is called an Edgeworth Box, after its creator, Francis Ysidro Edgeworth. The clever thing about an Edgeworth Box is that its dimensions depend on the total amount of the resources available (the box is 30 roses high, and 40 guns wide). But each position in the Edgeworth Box depicts the way this total quantity of resources is divided up. That is, any point in the interior of the box actually has four coordinates: an allocation for Mr. 1 (in this case, (30, 5)), and an allocation for Ms. 2 (here, (10, 25)). Notice that the allocations are assumed to have no waste, or leakage, so that 30+10=40, and 5+25=30. The top right corner of the box is the "origin", or (0,0) point, for Ms. 2; it is also the point where Mr. 1 gets everything: (40, 30).

The question the Edgeworth Box provokes is obvious, but it is important: Are there any allocations that are stable, in the sense that if such an allocation is ever achieved, at least one party will refuse to exchange any of the commodities they possess for a quantity of some other commodity their trading partner is willing to give. A stable allocation of this sort is called an "equilibrium," because it does not change. Is the initial allocation, where Mr. 1 has (30, 5) and Ms. 2 has (10, 25), an equilibrium?

The answer depends, of course, on whether there are any feasible trades that make both parties strictly better off. As you recall, any allocation "above" the relevant indifference curve is strictly preferred by the citizen to any point on the curve. This is true for both people, so the question of whether there are any mutually beneficial trades amounts to a simple graphical question: is there a nonempty intersection of the sets of consumption bundles preferred to the initial allocation?

You bet there is! The hatched area, called the "lens" for its resemblance to an optical glass in side view, is the set of points both parties prefer to the initial allocation. The edges of the setóthe two indifference curvesórepresent the boundaries of the set of mutually beneficial trading outcomes. So, the initial allocation is not an equilibrium, and we might conclude (rightly) that any division of the commodities is not an equilibrium if the intersection of preferred bundles is nonempty. But then it seems like you could draw a figure analogous to Panel A over and over, without ever establishing what we really want to know: what allocations could be equilibria?

The answer is actually fairly easy, when you think about it. Panel B shows an example of an equilibrium consistent with trading starting at the initial allocation of (30, 5), (10, 25). That equilibrium occurs at (22, 16) for Mr. 1, and (18, 14) for Ms. 2. How do we know that this is an equilibrium? Consider the set of points Mr. 1 prefers to this "final allocation": the points above, and to the right of the indifference curve through (22,16). Likewise, for Ms. 2: she likes points below, and to the left, of (18, 14) (where the coordinates are measured out from her origin, at the top right corner). What is the overlap of the two sets of preferred bundles? Nothing, nothing at all, except the allocation itself: the indifference curves are tangent to each other. Tangency means, by definition, that the curves intersect at only point: the equilibrium itself.

There are many such points of tangency for different pairs of indifference curves, of course. Depending on the preferences of the traders, the points of tangency can lie near the straight 45° line connecting the two origins, or these points can wander around, as shown in the heavy dotted line in Panel B. Regardless of the shape of the line connecting the points of tangency, this set of potential equilibria is called the "contract curve." The reason is that we expect that fully informed trading partners will always ultimately write a contract that puts them somewhere on this curve. Later in this chapter, I will introduce some measures for evaluating performance, but it is worth pointing out that we have already discovered one, the "Pareto optimum."

Pareto optimum: An allocation of resources such that any trade or exchange makes at least one party to the trade worse off. It follows immediately that all points on the contract curve are Pareto optima. It also follows immediately that the concept of Pareto optimum is not a very strong criterion for evaluation. Notice that the following allocation (one endpoint of the contract curve) is a Pareto optimum: Mr. 1 gets everything, Ms. 2 gets nothing. So is the opposite endpoint on the contract curve, where Mr.1 gets nothing and Ms. takes it all. In a broader sense, we might want to say that neither of these is terribly "optimal," from the perspective of society.

Still, it is clearly true that points off the contract curve (i.e., allocations that are not Pareto optima) are poor candidates for selection as policy "solutions." By definition, points off the contract curve allow reallocations such that both parties can be made better off. So, the intuition of the Pareto optimum really has more to do with efficiency (the absence of waste), than with any normative criterion of equity in distribution.

One final point: what about price? I have argued that the essence of the market system is the use of price to allocate resources. Yet it doesnít seem like prices appear anywhere in the analysis of exchange I have presented. Well, thatís wrong: prices are lurk at each point on the contract curve in the Edgeworth Box. We are all accustomed to thinking of prices in terms of money, but the example in Figure 3.2 involves only pure exchange. That is how prices started out: to get something you wanted, you had to give up something in exchange. Consequently, the price of a gun (in our example) is the number of roses you have to pay. In our example, it appears that (at the equilibrium) two guns exchange for three roses; that means that the price of a gun is 1.5 roses. Graphically, the price is the slope of the line through the point of tangency which intersects the indifference curves only at that point.

The Functions of Money

Why do we need money at all? What is wrong with barter? After all, as we saw from the Edgeworth Box in the previous section, it is possible to arrive at a Pareto optimum through barter alone, and the price implied by that equilibrium reflects the relative scarcity of the commodities available for consumption, given the preferences of the market participants. If I have to give up one apple to get half an orange, the opportunity cost of consuming that half orange is clear: I donít get to eat the apple.

But prices are usually stated in terms of an abstraction: money. You may never have given money much thought (unless you were trying to get more!), but money is an astonishing creation. Without money, societies as we know them could not possibly exist. Letís start with a definition.

money: anything generally accepted as a medium of exchange, meaning that money represents abstract command over any goods and services available in the market. But that doesnít help very much, and is very nearly circular: if I will accept something in exchange for my goods, that means I must believe it has value as a medium of exchange. Why would I believe that it has such value? Because I think other people will accept it in exchange for their goods. This is all rather upsetting, because we would like to have a better definition. But there really isnít one; all that can be done is list the characteristics that make for "good" money.

Characteristics of a Good Currency

1. Widely accepted as a medium of exchange, for any kind of transaction

2. Clear, consistent unit of account (one unit is identical with another).

3. Store of value (does not deteriorate if unspent)

4. Durable (does not wear out)

5. Difficult to counterfeit or create (not easily debased)

6. Divisible into small units (so you can make change)

Now we can answer the "why not barter?" question. Suppose that we had not just two, but three people, and three commodities. This is still a pretty simple society, but the added complexity in negotiating exchanges is considerable. The problem with three people and three commodities is that you may not be able to exchange your goods directly for what you want. Instead, you may have to conduct intermediate exchanges, and visit both other traders, to reach a Pareto optimal distribution of resources.

Suppose Mr. 1 has apples, doesnít like oranges, and wants to get some bananas. Mr. 1 goes to Mr. 3, who has lots of bananas. However, Mr. 3 doesnít like apples; he will only trade his bananas for oranges. So Mr. 1has to go over to Ms. 2, the orange farmer, and trade apples for oranges. Then, but only then, can Mr. 1 go back to Mr. 3 and effect the trade 1 wanted to make all along, securing bananas for his supper.

Notice that this is true even if there is universal agreement (or a law) that the "price" of any one fruit is any other fruit. That is, an apple exchanges for an orange exchanges for a banana exchanges for an apple. If I have an apple, and you have a banana, I would like to trade. But you hate apples, and insist on an orange. I could walk over to the neighborís house, a mile away, and trade my apple for an orange, but that is too much trouble. More simply, the difference in my satisfaction between the banana I love and the apple I like is not enough to make me walk two miles. The essence of barter is then not "haggling", or the separate negotiation of price, but rather the act of physically exchanging one commodity for an agreed-on amount of another commodity.

The inconvenience and wasted time in making trades through barter dissipates much, or all, of the value of the trade. Economists call this friction in making exchanges "transactions costs." Transactions costs are pure, deadweight losses, because no one gets any value or advantage out of the waste of time and energy. Reducing transactions costs is Pareto superior, because you are (in effect) getting something for nothing.

And so, we have our answer: people use money, instead of barter, because money reduces the transactions cost of exchange. Our three fruit lovers might agree to create a "fruit note," with the understanding that one unit of fruit note exchanges for an apple, a banana, or an orange. Then the holder of a note has abstract command over fruit of any type; Mr. 1 can immediately obtain his bananas, without the intermediate exchanges.

There are other advantages, of course. Mr. 1, who has apple trees, could plausibly borrow against his nearly ripe, but not yet edible, apple crop, obtaining fruit notes on loan from Mr. 2, with the understanding that the fruit notes will be paid back when it becomes possible to sell the apples. In effect, Mr. 1 is just exchanging apples in the future for bananas now, but the institution of a currency facilitates this exchange.

In nearly all human societies, most economic activity is concerned with the making and spending of money incomes. Historically, many different objects have served as money, including stones, shells, ivory, wampum beads, tobacco, furs, and dried fish, but it appears that even from the earliest times precious metals (especially gold and silver) have been preferred because they satisfy the characteristics of "good" money listed above.

To a larger and larger extent, however, money does not depend on its value as a commodity. Paper currency was first issued about 300 years ago, but until recently was almost universally "backed" by some standard commodity of intrinsic value into which the currency could be freely converted on demand. By contrast, fiat money is inconvertible money made legal tender by the decree of the government. The Australian government has introduced a plastic money, lighter and cheaper than metal, and more durable than paper.

But most money in the world does not exist at all, at least not in any physical sense. More than 90% of the worldís "money" exists only as a binary code on some electromagnetic medium, with an electronic tag identifying who owns it. The increasing use of credit cards, ATM-debit cards, and other "paperless" forms of transacting appear destined to eliminate the last 10% of money held as currency or commodities. Why? Because eliminating the physical aspect of money reduces transactions cost even further. Arenít markets wonderful?

Well, yes, markets are wonderful. Immediately following this chapter is a case study, excerpted from a fascinating account of the spontaneous emergence of a set of institutions of exchange, and how a "market" based on a cigarette currency allowed a group of prisoners of war to improve their lives. But the market system of organization is not perfect, and sometimes it is downright bad. In Chapter 4, following the case study, we will consider failures of markets, and difficulties in evaluating market processes.

Key Concepts

A. Gains from Trade

1. Diverse Preferences

2. Diverse Endowments

B. Technical Cost Conditions

1. Economies of Scale

2. Division of Labor

A. Uses of Money 1. Medium of Exchange

2. Unit of Account

3. Store of Value

B. Characteristics of "Good" Money

1. Widely Accepted

2. Consistency, Interchangeability

3. Nonperishable

4. Durable 5. Hard to Counterfeit

6. Divisible

A. Prices

B. Queues

C. Lotteries

D. Authority/Discretion

A. Scarcity Signals

B. Determinants of Wealth

Indifference Curves

Edgeworth Box


Contract Curve


Pareto Optimum

Homework Questions

1. Suppose that a society has a total of two people (Xerxes and Zenobia), and the following quantities of two commodities: 50 pomegranates and 30 wheels of goat cheese. Draw the Edgeworth Box for the society, assuming the origin for Xerxes is at the bottom left and the origin for Zenobia is at the top right.

2. In this same diagram, draw a set of indifference curves for Zenobia, and a set for Xerxes. (Remember that higher levels of utility are down and to the left for Zenobia!) Make sure that no two indifference curves representing the same person cross each other, and be careful that you donít violate the "more of both is better" rule.

3. Given the indifference maps you have constructed for Xerxes and Zenobia, draw in the "contract curve" with a heavy black line. By definition, every point on the contract curve is a "Pareto optimum"; every point off the curve is therefore "inefficient," because it is "Pareto inferior" to at least one point on the contract curve. Make an argument for why any point not on the contract curve is wasteful, in the sense that it is not the best use of resources available to the society.

4. Identify the point (call it P1) in the box where Xerxes gets 40 pomegranates and 5 cheeses, and Zenobia gets 10 pomegranates and 25 cheeses. Is this point on the contract curve, as you have drawn it? (Hint: The answer will be no, unless you have drawn some very unusual indifference curves).

5. Redraw the section of the diagram you have already constructed to identify the "lens", or bargaining space, associated with P1. Be careful to place P1 first, then draw one indifference curve that passes through P1 for Xerxes, and one for Zenobia. The lens will then be the overlap of (a) the points preferred to P1 by Xerxes, and (b) the points preferred to P1 by Zenobia.

6. Consider the four means for allocating scarce resources (price, queues, lotteries, and authority). Give at least one example in the real world where this allocation scheme is used, and offer an explanation for why. Is one of these allocation schemes inherently better than the others, or is each useful in the right situation?