If you can’t get what you want, what’s the next best thing? This is pretty much the deepest question in economics, being that preferences are king in a world of scarcity.

It lives also in a more complicated way in a relatively obscure piece of economics called the theory of the second best (which stems from a paper all the way back in 1956 by Richard Lipsey and Kelvin Lancaster). I’ve been thinking about this today after going back and forth with Megan McArdle on Twitter about the funding of policy research, ending up here. The question at hand is whether “relevant” policy research is done in academia, with funding therefore from the higher education system.

For some background, my favorite non-technical piece about it is from 2007 on The Economist’s Free Exchange blog.

Consider a frivolous analogy to cookie-baking. If the optimal cookie contains chocolate chips and coconut flakes, but you have no chocolate chips, chances are you don’t need the coconut either. The second-best cookie may be the gingersnap. If ingredients (or logical conditions) do their work through a certain combination or complementarity, you may have to aim for something completely different even if you’re missing just one of them.

The “textbook” theory of the second best is an abstract story like any other economic model, of course. The mathy story is there to contrast with the optimality conditions at the heart of the general equilibrium story (the kind that have the analogies across the political theory spectrum that I talked about recently). It’s stories all the way down!

So what’s the moral of the story?

Lipsey recommends verbal or “appreciative” theorizing in the style of Adam Smith, Thomas Schelling, and Milton Friedman informed by formal theory, but not bound by it.

Good policy advice in the real world is at least as much applied moral philosophy, psychology, and anthropology as it is applied economics. And if piecemeal policy advising is an art, then a useful wonk is a sort of artist, but in the way a good surgeon is an artist, not the way a novelist is. The theorists are novelists, illuminating the real world by inventing false ones. The best policy-oriented economists, both left and right, are second best economists in the sense that they grasp the lesson of their fictions, but aim at truly feasible ideals, not blackboard utopias.

Am I on board? I wouldn’t be as quick as this to sweep away the Rubinstein “pointless fables” reading of economic theory, and yet the point about feasibility is well taken. There is a fundamental difference in my mind between a “blackboard utopian” economist, lost inside the desert of the real that is the Model, and a “fable telling” economist, self-conscious and unashamed just to spin yarns.

But maybe what’s going on here is a workable distinction between the wonk-economist and either the fable-economist or utopian-economist. It’s right to say that both the wonk and the fable types need to use their creativity, but perhaps also fair to say that the creativity is not quite of the same kind.

I think—despite the caricature of economists as the utopian type, blindly wedded to hopelessly unrealistic models—the majority of academic economists are wonk types or fable types. I also think that economists in both categories are explicitly interested in promoting good policy thinking, albeit with different means and methods. Ideology haunts utopians, but how many utopians do you know?

Anyway, back to the point. Where does policy-relevant research happen? I think that lots and lots of policy-engaged research really does get done in academic economics, largely in the wonky category. Economists are deeply engaged with grounded, unsexy minutiae and explicitly consider the effects of actual policy and the structure of hypothetical policy.

McArdle’s point, though, was something slightly more challenging:

One of the problems with “relevant” is “has an interest group powerful & interested enough to move it forward”

This gave me pause. I had been defining policy relevance as focusing on the feasible, rather than the achievable. How significant is the difference between the two? I think no-one would argue that we shouldn’t think about things that seem hard or even impossible to achieve. Certainly we need to identify the best feasible policies, including interrogating where our concept of “best” comes from.But I think also that we can nudge our research in a good direction at low cost.

The first-best world for the wonky economist might be one in which the “best” feasible policies become achievable once identified. But we certainly don’t live in that world. Given that we are in a second-best world (at best… wait), what should the wonky economist do?

The theory of the second best says that if you are restricted to be suboptimal in one dimension, it’s generally not the right idea to keep doing what would have been optimal in all of the other dimensions. Thinking must be holistic. Taking that lesson literally in this case would then say: if you want to change policy, you cannot continue to “optimize” by making the identification of the best feasible policy the “conclusions” section of your research paper.

Most economists I know are fairly savvy about political, fiscal, and institutional constraints. We are quite capable of making an honest stab at identifying the achievable. I don’t think we need to radically rethink our modeling approach—we don’t need to start explicitly writing political constraints into our mathematical fables—in order to give the question some attention. In our research papers, we should unashamedly give the powerful and influential something they can realistically sell, our eyes open to the challenges they will face. We should explicitly engage—right there in our research papers for publication—with the achievability question. This should not be looked down on as “soft” or “fuzzy” or “unscientific”.

Maybe this would be a way for us to wrest a small sliver of influence away from pay-to-play think-tanks and back to academia.

It’s true that the predominant currency of the academic world is not the op-ed. Policy advocacy of that type lives largely outside of our incentive structure of peer-reviewed publications and citation counts. Why not compromise, then, with a gesture inside of our incentive structure? Our academic research could itself engage with the question of what is achievable, to the mutual benefit of the economist who seeks influence and the interest group that seeks expert support.