Cost–benefit analysis almost always requires the analyst to predict the future. Whether it is efficient to begin a project depends on what one expects will happen after the project has begun. Yet, analysts can rarely make precise predictions about the future. Indeed, in some cases, analysts can reasonably assume that uncontrollable factors, such as epidemics, floods, bumper crops, or fluctuations in international oil prices, will affect the benefits and costs that would be realized from proposed policies. How can analysts reasonably take account of these uncertainties in CBA?

Imagine that you completed the first three steps in CBA set out in Chapter 1. You have identified the alternative policies, determined standing, and catalogued the relevant impacts with appropriate units of measure. Next you must predict the impacts of each alternative and monetize them. As mentioned in Chapter 1, sometimes prediction and monetization can be done together (for example, when a demand curve is available), but often they must be done separately (for example, when a policy potentially affects health or crime rates).

The Washington State legislature asked the Washington State Institute for Public Policy (WSIPP) to identify programs that could be implemented in Washington to reduce the number of children entering and remaining in the child welfare system. We briefly review the approaches to prediction and monetization used by the analysts in conducting a CBA of a particular intervention, the Nurse–Family Partnership for Low-Income Families (NFP).

In conducting CBAs of government policies, there is a natural tendency to list as many effects of the policies as one’s imagination permits. For example, an improvement in public transportation in a particular city may increase bus usage and reduce car usage. It may also reduce downtown pollution and congestion. Further, it may subsequently reduce the demand for automobile repairs, parking places, and gasoline.

Chapter 3 presents the basic microeconomic foundations of cost–benefit analysis. As discussed there, the change in allocative efficiency (i.e., the change in social surplus) due to a new project or a change in government policy depends on changes in consumer surplus, producer surplus, and net government revenues. This chapter and the next two illustrate how changes in these variables could be estimated if the pertinent market demand and supply curves were known.

Both private and public investment decisions can have important consequences that extend over time. When consumers purchase houses, automobiles, or education, they generally expect to derive benefits and incur costs that extend over a number of years. When the government builds a dam, subsidizes job training, regulates carbon dioxide emissions, or leases the outer continental shelf for oil exploration, it also sets in motion impacts that extend over many years. In order to evaluate such projects, analysts discount future costs and benefits so that all costs and benefits are in a common metric – the present value. By aggregating the present values of the costs and benefits of each policy alternative that occur in each time period, analysts compute the net present value of each alternative. Typically, analysts recommend the alternative with the largest net present value.

A key concept for valuing policy impacts is change in social surplus. As discussed in Chapter 3, changes in social surplus are represented by areas, often as triangles or trapezoids, bounded by supply and demand schedules (curves). Measurement of changes in social surplus is relatively straightforward when we know the shapes (functional forms) and positions of the supply and demand curves in the relevant primary market, before and after the policy change. In practice, however, these curves are usually not known. Analysts have to estimate them from available data or find alternative ways to measure benefits and costs. In this chapter, we consider estimation of demand curves.

Chapter 1 emphasizes that CBA can almost always usefully inform public-sector decision-making. In practice, its usefulness depends on its accuracy. One way to examine the accuracy of CBA is to perform analyses of the same project at different times and to compare the accuracy of the results. In Chapter 1 we called such studies ex ante/ex post comparisons or ex ante/in medias res comparisons. We now return to this topic in more detail.1

Public policies usually require resources (i.e., inputs) that could be used to produce other goods or services instead. Public works projects such as dams, bridges, highways, and subway systems, for example, require labor, materials, land, and equipment. Similarly, social service programs typically require professional employees, computers, telephones, and office space; wilderness preserves, recreation areas, and parks require at least land. Once resources are devoted to these purposes, they obviously are no longer available to produce other goods and services. As a result, almost all public policies incur opportunity costs. Conceptually, these costs equal the value of the goods and services that would have been produced had the resources used in carrying them out been used instead in the best alternative way.

Revealed preference methods facilitate inferences about individuals’ valuations of goods by observing their behaviors in markets or analogous situations in which they must make trade-offs between things they value. Analysts generally prefer inferences from revealed preference methods to inferences based on stated preference methods, which are normally elicited through surveys. However, there are usually few observable “behavioral traces” relevant to public goods and externalities with missing markets. This is particularly the case for the passive use goods discussed in Chapter 13. In the absence of behavioral traces, analysts very rarely have a viable alternative to asking samples of people about their preferences.

Microeconomic theory provides the foundations for CBA. This chapter begins with a review of the major concepts of microeconomic theory that are relevant to the measurement of social costs and benefits. Most of them should be somewhat familiar from your previous exposure to economics. After that, the chapter moves to welfare economics, which concerns the normative evaluation of markets and of policies. This part of the chapter is particularly concerned with the treatment in CBA of the taxes required to fund government projects.

Policy analysts typically face time pressure and resource constraints. They naturally wish to do cost–benefit analysis as efficiently as possible and without getting into estimation issues beyond their competence. Anything that legitimately lowers the cost of doing CBA increases the likelihood that any particular CBA will be worth doing, because it increases the chance that a CBA of doing a CBA will be positive.

Sometimes the benefits and costs of program interventions can be estimated through experimental and quasi-experimental designs.1 This chapter first describes experimental and quasi-experimental designs, indicating how they are used in estimating the impacts of social programs in policy areas such as health, education, training, employment, housing, and welfare. It then describes how these impacts are incorporated into CBAs of employment and training programs, illustrating how concepts developed in earlier chapters can be used in actual cost–benefit analyses. The case study that follows the chapter examines CBAs of a number of employment and training programs that were targeted at welfare recipients.

Table 14C.1, which appears at the end of this case, presents summary results from CBAs of 26 welfare-to-work programs.1 These programs were all targeted at single parents who participated in the Aid for Families with Dependent Children (AFDC) program, which at the time the CBAs were conducted was the major cash welfare program in the United States. Most of the programs were mandatory in the sense that AFDC benefits could be reduced or even terminated if a parent did not cooperate. The CBAs were all conducted by MDRC, a well-known non-profit research firm, and used a similar framework including the classical experimental design with random assignment to treatment and control groups.