Sweet Science

ENGINEERING A NEW APPROACH TO ECONOMIC POLICY

Economic policy does not follow from economic theory. Instead, policy needs to be drawn from a complicated blend of judgments about ambiguous empirical evidence, normative judgements, and sensibilities that may be framed, but are not determined, by scientific theory. Put another way, economic policy is a blend of engineering and judgment — an "art and craft," not a scientific endeavor that follows from economic theory. Debates about policy are best treated as debates about the art and craft of economics, using a methodology appropriate for an art and craft. Policy debates should not be treated as debates about science, and should not be governed by a methodology more appropriate to science.

Unfortunately, modern economics doesn't treat policy in this way. Instead, it conceives of policy as an applied science, and uses the methodology of science to study policy issues. To some degree, that makes sense. Clearly, one wants evidenced-based, objective analysis of policy. An art and craft methodology uses theory and science whenever it can, meaning to the extent it is appropriate. But when dealing with the messy issues of policy, an art and craft methodology takes seriously the fact that statistical evidence needs interpretation and that one's views and analysis are inevitably influenced by one's normative judgments. An art and craft methodology recognizes that policy is intrinsically entangled and must be dealt with using a methodology designed to guide in such ambiguous, messy, and uncertain situations. To pretend otherwise undermines both the science of economics and economic policy discussions.

Early on, economists struggled with this policy/science divide. Advocates of various policy positions all claimed to have science and theory on their side. They inevitably attacked their opponents for being non-scientific. This struggle led Classical Liberal economists to embrace a methodological tradition that interpreted the science of economics narrowly and created a firewall between scientific pursuits and policy endeavors. This tradition is best found in the policy methodology of Classical Liberals such as John Stuart Mill and his followers. That methodology recognized the messiness of policy compared to the elegance of the theory underlying science.

To deal with that messiness, the policy methodology needed a branch of economics that was free of scientific certainty. One way to handle that problem would be to accept that no part of economics was a science. The second way — the path adopted by Classical Liberals — was to divide economics into different branches — a scientific branch concerned with agreed-upon empirical facts and logical implications of assumptions, in which normative values played as minimal a role as possible, and a policy branch in which values were seen as essential elements of the analysis. The policy branch of economics would use a different methodology than the scientific branch, and its conclusions would not be considered scientific conclusions.

Since our goal in this book is to talk about the methodology appropriate for applied policy, we do not distinguish between the "economics is not a science" and the "economic policy branch of economics is not a science" alternatives. The reason is that our interest is in applied policy, not the science of economics. We interpret the "no economic science" alternative, such as that proposed by some philosophers, for example, Hillary Putnam (2002) or Alexander Rosenberg (Rosenberg 1992), as being included in the Classical "separate branch" alternative. The "no economic science" alternative simply makes the further assumption that the science branch of economics is empty, an assumption we do not accept. But for our purposes these come to the same conclusion since, if there is no part of economics that is a science, then the applied policy branch of economics will not be guided by scientific methodology. If one believes, as Putnam and Rosenberg believe, that no part of economics is a science, it does not change our argument that applied policy should not be thought of as applied science — it strengthens it, since if there is no economic science, our argument — that applied policy economics should not be seen as applied science — becomes tautologically true.

Most classical economists accepted that there was a scientific branch of economics. But they also believed that policy did not follow from scientific theory, and they built that belief into their applied policy methodology. By doing so, Classical Liberalism sought to discourage the conflicting advocates of any policy issue from claiming the authority of scientific justification. Only a powerful firewall between theory and policy could accomplish that. For John Stuart Mill, policy was not based on science, and science did not concern policy. Instead, science was about a search for the truth. In pursuit of that truth, in order to see the scientific truth more clearly, one should ideally harbor no policy considerations whatsoever. Since that was practically impossible, one should attempt to guard, as much as possible, against a tendency to claim too much from science. Policy construction was meant to be about the search for answers to specific policy questions. That search required one to go far beyond the limits imposed by science. The objective was to integrate into the argument judgments that had no scientific basis, but that might have a philosophical basis. To keep the two separate, an economist needed to always lean over backward to confess that, in his or her role as an economic scientist, he or she had nothing to say about policy. That doesn't prevent him or her, when operating in a "statesman" role, from expressing views, and if he or she has expertise in that area, from offering them as the views of a specialist. But that expertise has to be broader than that of an economic scientist, and it must involve knowledge that goes far beyond science. Given the complexity of the economy it is an expertise that will emphasize its fallibility and view that often the best we can do in a complex world is to muddle through without definitive answers (Colander 2003).

The Abandonment of Classical Liberal Methodology

In the 1930s, the economics profession began to abandon the policy methodology of Classical Liberalism by removing the firewall between economic science and policy. This book is an exploration of that abandonment and a call for the profession to...