I keep wondering what the deal is with Megan McArdle. A more or less self-taught libertarian economics blogger who turned that into some big-league writing gigs, she shows much more facility with the technical aspects of the issues she discusses than most bloggers, which still leaves a lot of room for improvement. My impression has been that her posts are generally coherent (in the sense of being understandable – the first hurdle for most bloggers and many libertarians), but not especially well thought-out. In particular, I think she pretends to a degree of technical sophistication that she doesn’t fully command.
Today’s post – arguing for a “scientific” approach to the bailout debate – seems to me illustrative of that phenomenon. She throws around some technical terminology, misapplies technical concepts, and adopts an artificial neutrality in places where it isn’t justified, in order to position herself as rational and disinterested. I think there’s a lot less there there than it looks.
Here’s my position on what ended the Great Depression: I don’t know. There are a whole lot of theories out there, but with an “n” of 1, no overwhelming evidence in favor of any of them.
Misapplication of Theory The size of your “n” – the number of data points measured in a study – is relevant to statistical analysis. A larger “n” usually gives a smaller standard deviation and thus more reliable conclusions from whatever statistical tests you perform. Having an “n” of 1 is the classic mistake of assuming one example proves some sort of rule or trend. But this concept applies only to statistical analysis. The events of the Great Depression are historical, and reasonably well-documented. To get the answer we need, we merely have to trace out the events in question – what happened, and what consequences it had, and what happened from those consequences. The story is complex, and that’s why people still argue over it, but the size of the “n” (there was only one Great Depression) has nothing to do with it. To treat distinct historical events as unanalyzable statistical problems is essentially to say that we can never know what caused any historical event. It doesn’t even make sense to treat this as a problem of sample size: sample size merely affects the mathematics of the statistics (standard deviation and confidence interval) – but this isn’t a mathematical problem. She’s borrowing the supposed sophistication of mathematical economics to bolster her conclusion about a purely descriptive study. And she’s not being metaphorical – she really means it (see below). But it’s nonsense.
There are a few that I think most economists agree are not true, like that the spending portion of the New Deal ended the Great Depression through the magic of fiscal stimulus; a few percent of GDP in stimulus are not, at any reasonable multiplier, enough to produce high single-digit economic growth, which is why economists from Friedman to Tobin generally concluded that the decisive moment was either the monetary expansion of the late 1930s, while others credit the massive fiscal stimulus of World War II.
False Example, Artificial Neutrality I suppose most economists would agree on that if you asked them, but since that forms no part of the pro-FDR side of the argument, there’s little point in asking them. This is classic straw-opponent argumentation: holding up a thesis that nobody entertains, and then demonstrating that the debate has to be conducted on her terms by demolishing the stupid claim while implicitly letting it stand in for the reasonable claim her opponents really do hold. (Nobody says that New Deal spending alone ended the Great Depression – especially since the Depression is generally held to have lasted at least until some time into WWII. But many people claim that the New Deal helped greatly ease the effects of the Depression, and that the problem with the New Deal was that spending was not high enough, not that the plan in concept was wrong-headed.) McArdle moves on to two alternate theories that leave the New Deal out of the picture after grandly dismissing an absurdly simplistic cartoon version of the pro-New-Deal theory; but she does so while pretending that she is being mainstream and uncontroversial, and that all economists agree with her. She thus positions herself as both neutral and among the anointed of economic theorists, when in fact she is, simultaneously, radical, revisionist, and wrong.
But which of those two theories is correct? No idea. The stimulus story and the monetary story both track the time frame reasonably well, and much depends on a counterfactual we can’t test about what would have happened if America hadn’t gotten into the war in 1941.
Misapplication of Theory Here she’s back to the faux-science thing again. This time it’s not statistics, but “experiemental method”, that provides the cloak of sophistication she’s hoping nobody will notice she’s not actually wearing. She wants to set up a “controlled experiment” – have a whole bunch of Great Depressions in which we could try different economic theories (and presumably a control in which we’d do nothing at all and simply watch people starve). That would be the “scientific” way – because obviously you test a theory of global economics the same way you test a new drug, by setting up parallel test cases and observing them remotely. Without that, we just have “no idea” which economic theory might be best. Somehow, it never seems to occur to her that there might be other ways of validating economic theories besides full-scale testing to destruction. It also seems never to have occurred to her that it has never occurred to any other economists – including the ones she quotes – that that is how you’d do it. Nobody has performed multiple-Depression comparative studies under controlled conditions (or even imagined that it was sane to consider doing so), yet that did not stop her fellow Ayn-Randite Milton Friedman from developing his own theory on what ended the Depression – which she quotes just one line above! She claims that the New Deal hypothesis has been decisively eliminated without the benefit of comparative testing, and she also notes that celebrated economists, including ones she admires, have not considered such testing was necessary for their own theories, but this suggests nothing to her about whether her own “scientific” approach to macroeconomics makes any sense.
[When a recession ends, it can take some time for complete recovery, because if] your economy drops by 33%, it takes four and a half years at 10% annual growth just to return to where you were before the crisis; meanwhile, your labor force and productive capacity have presumably grown somewhat, so you need to go even further to close the output gap.
Pretended Sophistication She knows math and everything! Sort of. The growth-rate calculation is correct as far as it goes; the “output gap” remark is wrong. If “output” means only total production, then obviously a growth rate of . . . whatever . . . in output will generate exactly as much of an increase in total output over a given time as the same growth rate in “the economy” produces an increase in “the economy”, whether or not the population, or total manufacturing capacity, have increased. If it is true that the population and manufacturing capacity have both increased in that time, then after the recovery back to pre-depression levels you will have the same total output being generated by a larger number of workers, using more equipment. This will mean you have, on average, lower output per worker and per machine. But there are technical terms for output per worker and output compared to productive capacity: they are “productivity” and “capacity utilization”, respectively. It’s true that you would have to grow above the pre-Depression output level to accommodate increased workforce and capacity at their former levels of productivity and capacity utilization, but that is not what she said. It is not true simply to achieve the former level of GDP, or whatever your measure of raw output may be, which is what she did say. This is McArdle trying to be too clever by half. But the real point is that it’s totally unnecessary: the time to recovery has virtually nothing to do with the rest of her article, and the output/productivity point was entirely out of left field – it has nothing to do with anything. She throws it in gratuitously, just to sound like a real economist – and gets it wrong.
I post about these things as, hopefully, something close to a matter of science.
Thank you, Megan! Thank you for showing us the light of . . . science!
It is an empirical question whether the multiplier for government spending is greater or less than one. It is an empirical question whether the multiplier for the spending we just did is greater or less than one. It is an empirical question whether the sorts of crises we are now in produce liquidity traps such that fiscal shock therapy can result in a permanently higher level of growth.
I do not say that we will know the answer to these empirical questions; I daresay we won’t, at least not to a certainty. But there *is* an answer out there in the ether, yet most of the public debate about these questions is not much tied to empirics. They are being debated as emotionally as if the topic were the relative virtues of the debaters’ spouses.
Once again, I am driven to quote the immortal Charles Murtaugh: the universe is not here to please you
Misapplication of Theory So if you thought it was, or you thought that related in any way to economics . . . you’re wrong. She makes a good point about the mistake of treating factual issues ideologically. That point needs to be made again and again, in so many circumstances. But not by Megan McArdle. In repeating over and over that “it is an empirical question” how certain economics phenomena take place, she underscores her previous “scientific” view of them. But this seems strangely incompatible with her original claim that scientific methods can’t be brought to bear on phenomena where “n = 1″. It is also oblivious to any notion that “empirical” and “scientific” are not really the same thing. We use scientific methods to investigate empirical phenomena, but there are other ways of doing so, especially when those phenomena – though “empirical” – are not in their most significant aspects material or procedural, but rather historical or sociological. This article just takes a weird swing that makes it hard to draw out any coherent point: she starts out by emphasizing how opague economic events are, then cites authorities who have pretty definite points of view on the same events she says cannot be rigorously characterized, then emphasizes “scientific” methods of studying them, and then insists repeatedly that they are “empirical” and the answers to questions about them are “out there” (“in the ether” – ironically, a quintessential modern example of a theoretical phenomenon science has abandoned; I wonder if she thinks the stimulus multiplier effect is “in the phlogiston”?). In the end she decides again that “we may never know” most of these answers.
I doubt I’m the only one who is wearied by the way so many of the participants in the debate seem to already know the answer they want, and are merely looking for a set of questions that will get them there most expeditiously. Was there ever a time when people didn’t think that tricky economic conundrums could, or should, be used to “prove” that their personal values about the level of taxation and spending are a scientific fact?
Artificial Neutrality This also is a reasonable point, though hardly an original one. But I think that, again, here, she’s adopting a tortured stance of neutrality on what is far from a neutral question. Of course there are intellectually dishonest arguments on all sides of any controversial issue, and of course you can find liberal examples as well as others. But it’s far more than a mere matter of opinion where those arguments are mostly found, and which side of many debates they mostly characterize. Whether it is flattering to their reputation or not, it is no more than simple truth to say that conservatives make a political tool of distorted falsehoods, in technical or scientific matters as much as purely ideological ones. Never mind the stupid artificial controversies they constantly drum up: does Obama wear a flag pin?; how much does John Edwards pay for a haircut?; what color are Al Gore’s suits? That surely detracts from reasoned debate as much as their outright lies, but we can put that down to a basically childish mentality. But beyond that, they cannot, and by choice and by temperament will not, present basic facts in a straightforward, honest, and accurate manner. There is no neutrality to be had on the question of distorting facts to fit pre-determined conclusions: that is utterly and overwhelmingly a conservative practice. It is not even an error or mistake on their part – it is a deliberate strategy. Conservatives actively and knowingly embrace factual falsehood as a tactic and a general approach to fact-based issues. Conservatism is the root of creationism, of all things – an explicit denial of the scientific and empirical approach to knowledge that McArdle demands; some of its practitioners have explicit creeds openly stating that they reject facts in favor of pre-ordained dogma. It is the source of persistent, obsessive attempts to remove books from school (and often public) libraries, ban words, plays, movies, or music in broadcast media, restrict what can be shown in movies and TV, and to meddle in virtually every aspect of education from a religious and ideological point of view – in short, it is a movement that approaches education and knowledge as things to be prohibited and restricted. Conservatism is the source of sex education that doesn’t actually discuss bodily functions and prohibits any reference to contraception or abortion. And in public policy discourse, conservative lies and distortions are legion, and well-documented. (There have been, literally, entire books devoted to the basic factual errors of Rush Limbaugh, Ann Coulter, and even so-called “scholars” like Dinesh D’Souza, Thomas Sowell, and the “Bell Curve” clowns.) The right-wing treatment of global warming, oil availability, conservation policy, and any other factual issue you care to name is simply thoroughly dishonest. And all this not even to mention the incredible dishonesty of their political campaigns and policy debates. There are mistakes, or even distortions, on all sides, but the right wing practices deliberate and open lying persistently, obsessively, and as a standard part of their political repertoire. It is another exercise in artificial neutrality to pretend otherwise – in this case perhaps not directly self-serving on McArdle’s part, but indirectly so, as evidence of the supposed independence of ideology she enjoys as a libertarian rather than one of those ideologues on the political extremes. In fact, she makes herself partisan by ignoring the gross imbalance in partisanship along the left-right continuum.
In the end, this is all very confused and overwrought. If her only point is that we should treat debates over economic policy rationally, and with as much rigor as we can muster, that goes without saying. Elevating that to some kind of high-level theory about the nature of science and economics is just grandstanding. Larding it with all sorts of technical terminology, and even mathematical calculations, that add nothing whatsoever to the actual substance of the point being made, is self-aggrandizing bluster – putting on a show of technical knowledge to make herself seem sophisticated and authoritative. And when she gets the terminology wrong, bungles the calculations, and cites inapplicable statistical and experimental methods as necessary to the knowledge she seeks, her whole self-satisfied dramatic display goes hilariously off the rails.