Max R. P. Grossmann

What determines policy? The question appears simple, but it spans the normative and positive realms.

In a previous blog post, I emphasized that empirical findings cannot be interpreted without a substantial qualitative understanding of institutions and the exchange within them. Moreover, I mentioned that policymakers can endogenously select which facts (i.e., “outcome variables”) they find important.

Contrary to the market, there is no forcing mechanism that requires policymakers to do anything in particular. (I will write more on this issue at a later point.) If we want to consider the feedback that policymakers obtain, then we must unavoidably look at the institutions that confer their authority. A dictator or absolute monarch claims ultimate ownership of the entire country. Public choice analysts have long concerned themselves with the political economy of dictatorship. In democratic countries, policymakers in a legislature are elected. Policymakers in bureaucracies are typically appointed by the executive. The executive in turn is derived from some democratic mechanism.

Once elected, a legislator is able to whatever they want within the confines of a constitution. However, the degree to which a constitution truly binds a legislator varies. Crucially, there is no constitution on earth that requires politicians to follow scientific findings. Such a requirement would be impossible to fulfill. Who determines what the science says? And, again, which outcome variables are to fall under the purview of science? Let us address these questions on normative and positive grounds.

Let us consider the incentives faced by politicians. Many of the pathologies of our age are a result of politicians doing what rewards them most. Pandering to voters' irrational instincts and showering them with financial bonbons is sensible policy. Even though these policies are unscientific, they tend to be implemented. Why? Because politicians depend on voters' approval. There are substantial literatures on these issues.

Science simply is also not unambiguous. There are enough scientists out there so that a policymaker who prefers policy X can always find a literature arguing that X is good. Even if there is large consensus, differing opinions exist and they will be consciously used to drive policy agendas. Don't trust me? Look who was selected as Director of the National Trade Council under a previous U.S. president. There are simply no constraints on who gets to speak for science. Scientists, like outcome variables, are endogenously selected by policymakers.

The incentives of public choice are not alleviated by science. There is no direct mechanism for science to play a role in democratic policy. However, there are indirect mechanisms. A fundamental idea in public choice is the median voter theorem. It is often misunderstood as a normative argument for majority rule, but in fact it is a positive statement about whose opinion will prevail if every voter has single-peaked preferences over a policy space. (Needless to say, the median voter theorem hints at attractive normative features of majority rule.)

The conclusion then, is clear. If the median voter is swayed by science, so will elected politicians be swayed. It is not for me to judge whether voters are or should be more concerned about scientific findings rather than their own material benefit if the two are in conflict. The same principle applies to bureaucrats: If the executive branch of a country were to decide that science is what should determine policy, bureaucrats will be able to effectuate more scientific policies.

The point here is a very simple one: There is no automaticity between science (even if uncontested) and policy, but science can matter under specific conditions.

Now that we have resolved this positive aspect of policymaking, what about normative considerations? Should science determine policy?

My single-minded focus in life is to get governments to adopt pro-growth policies. Generally speaking, science is conducive to growth. Modern technologies rely on semiconductors, battery technologies and research on machine learning.

As shown above, legislators will not automatically implement science, but they rely on voters to demand that of them. Normative judgments are required to translate science into policy. Even if we all agreed on what science says on any given topic, we would be unable to objectively select those policy areas where science should rule. The discrete bits of information delivered by science are not unambiguously translated to policy.

For example, it is a well-known scientific fact that healthy human beings require only one kidney to function. It does not follow, however, that people's spare kidney should be redistribute to those who require one. It does not follow, either, that human beings should be allowed to buy and sell kidneys on the market. The fact that one kidney is enough for most is just that: a fact. Policy recommendations can come about only from the combination of real-life facts and normative judgments.

That those normative judgments cannot be delivered by science itself should be obvious. In democratic societies, normative judgments are in the purview of the people. While it is technically true that facts cannot be subject to opinion, making this argument in the pursuit of policy involves a double-barrelled fallacy. First, facts are in dispute; they are often highly context-dependent, constructed, extrapolated, methodologically questionable or irrelevant for policy. Yes, ironclad facts are facts. Who gets to decide what is ironclad? Second, the application of scientific principles to policymaking involves a category error. Science and policy are two different spheres of reality. Scientific principles cannot prove or disprove religious judgments because the latter are not testable. Similary, the normative judgments by voters are not subject to scientific tests. They are, in that sense, sovereign. Scientists must remain silent about them.

Science matters profoundly for policy and it should matter—but only in combination with normative judgments. Who should give us those? Not scientists. Let's stop trying to read policy into science. Science as a truth-seeking mechanism does not directly influence policy (because policy is governed by the incentives of public choice) and it represents a category separate from the untestable, personal, sovereign and free value judgments of voters and policymakers. To assume away the latter is to destroy freedom and democracy.