Debates over public policy are full of what Ambrose Bierce called “vagrant opinions, without visible means of support.” Those of us who try to do policy analysis as a serious intellectual pursuit continually face the charge that the analysis we present is merely a fig-leaf to hide our otherwise naked interest or prejudice. And of course advocates for specific policies (or the interests they are designed to serve) have learned to dress up their lobbying efforts in analytic clothing.
When I teach courses in the methods of policy analysis, my first problem is to try to convince students that such a topic exists, without inducing the false belief that there exists some set of methods which, if properly applied, can provide “the right answer” to any problem.
Here’s the introductory text from that syllabus:
Public policy analysis is the discipline that considers what actions would best serve the public interest in a given situation, and how those actions can be implemented through actual institutions. Like any discipline, it seeks the truth, but the truth sought is pragmatic, in the original sense of that term: truth about what is to be done, rather than merely about how the world is to be described. That differentiates policy analysis from social science.
Policy analysis differs from policy advocacy in that it regards policy choices as instrumental: it reaches a policy conclusion at the end of its argument, rather than starting with a preferred policy and seeking arguments in its support.
A full policy analysis, if one were feasible, would answer the following questions:
1. What is the situation to be considered?
2. What suggests that this situation calls for a specific policy intervention, rather than being left to individual choices, markets, and ordinary formal and informal social controls?
3. What is at stake? What are the positively or negatively valuable outcomes of alternative policies, or of inaction?
4. On what basis should those outcome dimensions be weighed against one another?
5. What specific options exist, or can be invented? (Your answer to #2 ought to provide some hints. Remember that “doing nothing” and “continuing current policies” are always options.)
6. What is the probability distribution of the outcomes of each option on each of the identified outcome dimensions?
7. Which option has the best distribution of possible outcomes, appropriately weighted?
8. What actions would need to be undertaken to implement that option? What agencies exist that could undertake such actions? What changes would need to be made in those agencies if they were to do so? Are those changes feasible, and if so what would the costs be of bringing them about? Would the agencies in question then be more or less capable of producing public value in other ways?
9. Alternatively, what new institutions would need to be created, and what is the likelihood of doing so successfully? What would be the side-effects of creating such agencies?
10. What are the predictable implementation problems, both operational and political? How can they be overcome, or at least ameliorated?
11. What decision-makers would need to agree to the proposed course of action? What are the prospects of their doing so?
12. What social and political dynamics would the proposed policy set in motion? In particular, would the policy itself evolve appropriately over time, or would it instead tend either to degrade toward a less desirable policy or to fossilize as the conditions that make it desirable now change?
13. Are there second-best options with greater administrative or political feasibility, or more desirable likely evolutionary paths, than the chosen policy?
Any actual analysis will fall short of giving a full answer to those questions, and is to be judged, among other things, by how carefully it indicates the questions it has omitted and the associated ranges of uncertainty.
Notice that item #4 involves the application of some set of values, and that, as as result, people with different values will often reach different conclusions even if they agree on the facts. Benefit-cost analysis, which uses the willingness-to-pay principle, is a useful heuristic for many policy choices, but applying it uncritically means accepting the existing distribution of income and wealth as optimal and assuming that what people (say or think) they want is invariably identical with what’s actually good for them.
So we should not expect the application of policy-analytic methods to lead everyone to agree. What we should expect it to do is to eliminate options that don’t make sense under any reasonable set of values, and to help identify the factual and normative bases of actual disagreements. It also forces it practitioners, at least while they are doing it, to step back from individual or group self-interest, and from received opinion and factional loyalty, and try to reason in the public interest.