The following is part of an e-mail that I opened earlier today from Aaron the Aaron. Mr. Aaron remains unusually perturbed at Russ and me for suggesting that minimum-wage legislation might have some ill, unintended consequences that escape empirical detection:
You [Boudreaux] and Russ … are too anxious to substitute your own theorizing for measured, verifiable data. I do not trust your and Russ’s theorizing any more than I trust anybody else’s theorizing…. Your way of going about economics is cheating….
You would both be taken more seriously by people who matter if you both take empirical work more seriously.
Neither this forum nor this economist is properly equipped to launch into a lengthy discussion of scientific method. But I can’t resist writing a few words here in response.
First, I do take empirical work – including, by the way, non-quantitative history – very seriously. I emphatically do not believe that reality can be divined merely from armchair, a priori theorizing. (I’m sure that the same holds true for Russ, but I’ll speak here formally only for myself.) But to take empirical work seriously is not at all the same practice as naively swallowing asserted ‘facts’ and quantitative relationships whole and uncritically. Quite the opposite is true. To take empirical work seriously requires thinking seriously about methodology and epistemology – a practice that leads to mature evaluation of empirical claims. In turn, a mature evaluation of empirical claims often leads to a rejection of many such claims (for a variety of possible reasons). Such rejections of empirical claims reflect not a rejection of the importance of empirical information but, rather, a recognition of the inescapability of evaluating any and all empirical claims through theoretical lenses.
Second, the nature of economics and other social sciences means that the proper role of abstract theory in these sciences is greater even than it is in the ‘natural’ sciences. Not only is ceteris almost never paribus in social and economic reality, but many of the ‘predicted’ effects of postulated causes are often too fine and too small to detect with any practically useable tools of observation and measurement – even when ceteris is paribus.
Here’s an example, one that I first used, back in 2006, in another context : Suppose you’re standing beside a swimming pool that’s 1,000 times larger than an Olympic-sized swimming pool. This gigantic pool is full of water. The surface of the pool is calm. You drop into the pool a grain of sand. Question: what happens, as a result of dropping the grain of sand into the pool, to the pool’s water level?
Answer: it has risen. (Or, more precisely, the water level of the pool with the grain of sand in it is higher than that level would have been had you not dropped the grain of sand into the pool.) How do we know this fact? Reason tells us. By “reason” I mean here only – and perhaps in a way annoying to professional philosophers – that amalgam of human thought processes made up of experience with basic, everyday physical relationships as well as of ‘common-sense’ logic.
But no practical method for empirically measuring this pool’s water level will detect that that level has risen in response to your dropping into the pool a single grain of sand. Surely, though, you would be justified in dismissing as mistaken some sophomore who, standing beside you, insists that because the pool’s measured water level is unchanged, you are engaged in unjustified and anti-empirical theorizing when you conclude that the grain of sand dropped into pool actually raised the water level. Likewise if some empirical study found that, after the grain of sand was dropped into the pool, the water level fell.
The sophomore, believing only what is empirically detectable, concludes that the dropping of grains of sand, one by one, into a huge pool does not raise the pool’s water level (or that it actually causes the water level to fall). You – not rejecting empirical studies but also not rejecting what you know reasonably well about cause and effect – reach a conclusion opposite to that reached by the sophomore. And no protestations by the sophomore about your alleged scientific Neanderthalism and his or her devotion only to ‘the facts’ will, or should, change your mind.
Who, in this case, is the better scientist? You or the sophomore?
My pool example is intentionally extreme. But it’s perfectly valid for justifying why those of us who insist that a higher legislated minimum-wage will have some negative effect on low-skilled-workers’ employment options continue to accept this proposition despite failures of some measurement attempts to detect those effects.
Of course, like all such scientific propositions, the one defended here – about the inevitable negative consequences of a modest legislated minimum wage – must be held not as a dogma but, rather, as a scientific proposition that is the result of theoretical and empirical analysis. The possibility that this proposition is mistaken must never be forgotten. But because the case against the validity of this proposition challenges the very foundations of well-accepted economic science, the burden of proof borne by those who would have us reject this proposition is enormously heavy. A few empirical studies that find no negative consequences, or that find positive consequences, hardly suffice to meet this burden.