Politics does not lead to a broadly shared consensus. It has to yield a decision, whether or not a consensus prevails. As a result, political institutions create incentives for participants to exaggerate disagreements between factions. Words that are evocative and ambiguous better serve factional interests than words that are analytical and precise.
Science is a process that does lead to a broadly shared consensus. It is arguably the only social process that does. Consensus forms around theoretical and empirical statements that are true. Tight links between words from natural language and symbols from the formal language of mathematics encourage the use of words that are analytical and precise.
For the last two decades, growth theory has made no scientific progress toward a consensus. The challenge is how to model the scale effects introduced by nonrival ideas...The question posed here is why the methods of science have failed to resolve the disagreement between these two groups.
Economists usually stick to science. Robert Solow (1956) was engaged in science when he developed his mathematical theory of growth. But they can get drawn into academic politics. Joan Robinson (1956) was engaged in academic politics when she waged her campaign against capital and the aggregate production function.
Academic politics, like any other type of politics, is better served by words that are evocative and ambiguous, but if an argument is transparently political, economists interested in science will simply ignore it. The style that I am calling mathiness lets academic politics masquerade as science. Like mathematical theory, mathiness uses a mixture of words and symbols, but instead of making tight links, it leaves ample room for slippage between statements in natural versus formal language and between statements with theoretical as opposed to empirical content. Paul M. Romer (2015) "Mathiness in the Theory of Economic Growth" American Economic Review.
Last week I discussed a recent instance of the recurring complaint about mathematical economics (and more indirectly here.) Romer's piece was cited in a recent Aeon blog post. On, reading Romer's piece, I learned that Romer is not against mathematical economics as such (he is a leading practitioner), but rather against certain abuses of mathematics. In particular, his is a call to reduce the abuse in order to "make faster scientific progress if we can continue to rely on the clarity and precision that math brings to our shared vocabulary, and if, in our analysis of data and observations, we keep using and refining the powerful abstractions that mathematical theory highlights—abstractions like physical capital human capital, and nonrivalry." (90) Regular readers know that I am very interested in the distinction between science, politics, and academic politics and I return to it below. But first I want to set the stage
The abuse he has in mind is [i] the "ample room for slippage between statements in natural versus formal language and between statements with theoretical as opposed to empirical content." In what follows I am going to pretend as if such slippage is avoidable when one is using abstractions in theory, despite the fact that there is considerable work in the philosophy of scientific (including the philosophy of mathematical economics [see here]) practice that suggests some such slippage is always a live possibility. We can charitably suggest on Romer's behalf that in mathematical economics we want to avoid slippage that facilitates what he calls "academic politics.”
It is not always clear if we can blame “academic politics” for the slippage. For example, elsewhere in his piece, Romer describes yet another abuse: [i*] “Because incentives in these models motivate neither discovery nor diffusion, agents exchange ideas in the same way that gas molecules exchange energy—involuntarily, through random encounters. Given the sharp limits imposed by the mathematics of their formal framework, it is no surprise that traditionalists were attracted to the extra degrees of freedom that come from letting the words slip free of the math.” (91) Let’s stipulate Romer is right here about what drives the ‘traditionalists’ in ‘growth theory.’ What he describes is compatible with a move in academic politics, but need not be so. Some such ‘extra degrees of freedom’ may be thought an advantage on legitimate scientific grounds—it may facilitate (for example) finding solutions (to solving equations) or the search for empirical instantiations (when working with aggregations). To be sure, as regular readers of my blogging posts about the treatment of uncertainty in economics know, I am no fan of the widespread use of devices that purport to model mathematical randomness within mathematical economics. Because these mathematical devices have lots of tacit assumptions about distributions built into them, even superb practitioners tend to miss that they are making non-trivial bets on the structure of modal reality (leaving aside the fact that humans are intentional agents not merely extensional objects).
In fact, Romer’s piece also refers to other practices by other economists that fall under mathiness that come in for critical scrutiny: "The authors [McGrattan and Prescott] chose a word that had already been given a precise meaning by mathematical theories of product differentiation and economic geography, but their formal equations are completely different, so neither of those meanings carries over." (89) Here [ii] mathiness facilitates a species of equivocation by drawing on verbal familiarity with the term (not the math). It’s hard to see why this is due to mathiness; when (inspired by the economists David Levy and Sandra Peart), I use the term ‘methodological analytical egalitarianism’ as a technical term of art (as I do), I know that I risk confusion with my fellow philosophers (whose minds are trained to think of a different literature).*
In addition, ‘mathiness’ is used to refer to [iii] the practice of using words ‘that are disconnected from the formal results and a mathematical model that is not well specified.” (92) For, Romer also refers to a more general decline in standards in which ‘mathiness’ stands for [iv] “Neither colleagues who read working papers, nor reviewers, nor journal editors, are paying attention to the math,”(92) in which mathematical “theory” is “entertainment.” The earlier standards involved [this should be familiar to analytical philosophers] “clarity, precision, and rigor.” (92) I distinguish between different practices, [i-iv, including i*], not to make fun of Romer’s tendency to lump distinct phenomena under ‘mathiness.’
Now, let’s stipulate, for the sake of argument, that there has been a decline in standards (rather than the data-poor kind of nostalgia frequently exhibited by senior figures in a field). It is worth noting that Romer’s explanation for the decline, which is based on incentives, is very partial. For, it pretends as if some economists are driven by non-epistemic motives while others are more pure (conveniently the folk Romer agrees with). Yet, it fails to treat all the participants equally and so fails at explanation—what we would like to know [again stipulating that Romer is right about motives], is how come a group of practitioners subject to same incentives fail to main similar standards.
It is worth noting that Romer does not consider two obvious other hypotheses (again stipulating that he is right about the purported ‘decline’): (a) size of group: economics is vastly larger than half a century ago, and it is much harder to enforce a shared esprit de corps, including standards, so we should expect ‘drift’ in these standards; (b) perhaps, as suggested above, those that have embraced a new set of standards have done so because of alternative epistemic advantages that Romer is not giving sufficient credit—what looks like ‘academic politics’ is really a different understanding of what the science is about or supposed to do. Because he fails to consider alternative hypotheses, and assumes purity on his side, he makes an inference that is by no means fair (albeit) ‘natural:” “The natural inference is that their use of mathiness signals a shift from science to academic politics, presumably because they were losing the scientific debate. If so, the paralysis and polarization in the theory of growth is not sign of a problem with science. It is the expected outcome in politics.” (90)
With that (especially (b)) in place, we can return to the three-fold distinction between science, scientific politics, and politics. In it, science is incentive free (the agents are pure truth-seekers), while scientific politics is governed by non-epistemic incentives that can generate ‘equilibrium’ (but that falls short of the one that would obtain if economists were pure truth-seekers), and politics is governed by incentives that entail non-disagreement. The idea that science is about consensus and politics not is an idea that, within economics, goes back to Sidgwick and Keynes Sr. in the late nineteenth century; it is part of the founding narrative of the image of science that economics gave itself (see for details of the story here). It may have been taught to Romer’s generation by way of Robbins’ influential (LSE) statement about methodology at the height of the Great Depression. Within philosophy of science, Thomas Kuhn articulated the most famous version of it. Of course, eventually political philosophers decided that politics could also be a site of consensus, and they (Rawls, Habermas, etc.) re-imagined the political as amenable to technocratic reinterpretation.
Because science is understood as an instance of consensus seeking, deviations from the norm are treated as ‘unscientific’ (because political). It is well known (and within economics this has been known since Stigler in the 1970s) that this conception of science exhibits lack of tolerance for dissent and disagreement. This image of science (science = consensus) is especially surprising because economics is about agents that operate in reflexive and complex (data rich) environments. It would be astounding if only one model of reality would suffice, and if only one interpretation of that model would be available. In fact, here’s a useful heuristic: if consensus obtains about an empirically complicated topic (yes, that’s a weasel phrase) in economics, one ought to look for mechanisms (i.e., academic politics) that prevent alternatives from having their say.