Posts Tagged Analogies

The Real Problem with the Broken Window Fallacy

John Quiggin recently posted on the “Broken Window Fallacy” (BWF), a parable beloved by libertarians, originating from Frederic Bastiat but finding its most modern exposition in Henry Hazlitt. The basic idea is that while breaking a window will seem to stimulate spending by providing work for a glazier, the money used to employ him could have been spent elsewhere, say by employing a tailor to make a new suit. Therefore, as a result of the broken window the community has only a window (what they started with), rather than both the original window and a new suit. We must look at the “unseen” in order to understand the true economic effects of smashing the window.

Quiggin tries to refute the fallacy thusly:

Implicit in the crowd’s reaction is the assumption that glaziers are short of work. If (as sometimes happens) glaziers have more jobs than they can handle, then there is no extra window – at best, the shopkeepers order simply displaces some other, less urgent, repair. Similarly, for Hazlitt’s riposte about the tailor to work, there must exist unemployed resources in the tailoring industry, so that the shopkeeper’s suit represents an addition to output. If not, the additional demand from the shopkeeper will raise the price of suits marginally, just enough to lead some other customer to buy one less suit. So, the story seems to imply that the economy is in recession, with unemployment across a wide range of industries.

Yet “rais[ing] the price of suits marginally” – such that the person most willing to pay receives the suit – is precisely what libertarians have in mind when they envision a market economy functioning nicely. Under the assumption of full employment and no broken window, the shopkeeper purchases a suit while the glazier is put to work elsewhere creating a new window. Under the assumption of full employment and a broken window, the shopkeeper employs the glazier, meaning that somebody who previously would have employed the glazier goes without, while the tailor is put to work for somebody who likes the suit slightly less than the shopkeeper. Aggregate welfare and wealth is decreased, even if the flow of production is the same.

Quiggin attempts to introduce the assumption of unemployment to counter the standard story:

With these facts in mind, we can tell a different story. Suppose that the glazier, having been out of work for some time, has worn out his clothes. Having fixed the window and been paid, he may take his $50 and buy a new suit. To make the story stop here, we’ll suppose that the tailor is a miser (a vice traditionally associated with the clothing industry, as with Silas Marner), and puts the money under his mattress. So, in this version of the story, the glazier and the tailor are both paid, and the social product is increased by a new window and a new suit.

But the social product is not increased. If the window were not broken, we’d have a window and a new suit. When the window is broken, we have a window and a new suit. The allocation of the suit has changed, but not the total product. Quiggin will never refute the BWF like this, on its own terms, because if you start with the premise that a window gets broken, you will inevitably end up at the conclusion that the world is worse off than before. Once the window has been broken, we have lost $50 worth of window and will have to replace it. Depending on your ethical presuppositions, you might view the redistribution as desirable, and the employment of the glazier as an end in itself, but this is another debate.

And this is the real problem with the BWF: it’s a complete straw man. Noone, anywhere, ever, has claimed that ‘breaking windows’ is a desirable economic strategy, or that it will somehow add to wealth or welfare. True, you can pick and choose your own auxiliary assumptions to add ‘silver lining’ to the story. Given that the window is broken, the fact that the glazier then wants to buy a new suit is better than if he just hoarded the money. Perhaps the new window is slightly nicer than the old one. Perhaps, as a commenter suggested, the shopkeeper has an emergency fund which is “psychologically separate” from his other money, so he still buys both the window and the suit. Or maybe the glazier has an apprentice who benefits from the training when he otherwise wouldn’t, while the tailor does not. We can do this all day but ultimately, the broken window devotes resources which could have been used – even if they were previously idle – to increasing wealth and welfare.

Should we utilise idle resources? The answer to this question needn’t have anything to do with breaking windows. Quiggin, like most critics of the BWF, implicitly recognises this, which is why he stresses that none of what he says “means that it’s a good idea to go around smashing windows during recessions.” So why start with the assumption of a broken window? We could instead tell an alternative story where there is no broken window. The tailor is unemployed, and there is a kid who wants a shirt but cannot afford it. The government prints $50 and gives it to the kid, who buys a shirt from the tailor, increasing the social product without any broken windows. This story is similarly arbitrary, demonstrating the ease with which we can formulate a parable to come to the conclusion we like. But the question of which story’s assumptions (in particular unemployment) are true or not is an empirical matter.

Quiggin, in trying to refute an abstract parable built on arbitrary assumptions by introducing his own, slightly different arbitrary assumptions, is fighting a losing battle. The BWF may or may not be useful for demonstrating a certain point, but it is not a model of the economy and it is not always and everywhere applicable to economic problems. If you are arguing with somebody who thinks repeating ‘Broken Window Fallacy’ at you will settle the debate, you aren’t going to convince them by telling them the ‘Broken Window Fallacy, version 2’. You simply need to stop having the debate in terms of Broken Windows, and start having it in terms of what is actually going on in the economy. Otherwise you’ll be forever trapped at a useless level of abstraction.

, , , ,


Against Analogies

Bryan Caplan offers a typical libertarian thought experiment in defense of the position that redistribution is bad. As with with many libertarian thought experiments, this one is on an island:

Suppose there are ten people on a desert island.  One, named Able Abel, is extremely able.  With a hard day’s work, Able can produce enough to feed all ten people on the island.  Eight islanders are marginally able.  With a hard day’s work, each can produce enough to feed one person.  The last person, Hapless Harry, is extremely unable.  Harry can’t produce any food at all.

He then goes on to the obvious conclusion that forcing Abel to work is not fair just because the others can’t fend for themselves.

Here’s the problem: what does this have to do with income distribution in a modern capitalist economy? The answer is as follows: nothing. Absolutely nothing. In this island there is no state; people do not cooperate (actually they appear not to have any relationship whatsoever); there is no injustice; there is not even trade. So the thought experiment is worthless.

But the fact is that analogies are generally pretty awful, for the simple reason that they aren’t, by definition, the same as whatever they are purported to represent. They are often said to illuminate a few important aspects of a situation, but this is an illusion. I will demonstrate it by offering two oft-used and reasonable sounding analogies for the economy, one suggesting that monetary policy is impotent at the zero bound, one suggesting that it isn’t:

(1) You can’t push on a string.

(2) Money is like a hot potato that people pass around until the supply matches the demand.

All these analogies help us do is come to the conclusions that we already had in mind. When building analogies people sift through various images until they find one that satisfies the story they wanted to tell. There are an infinite number of feasible sounding metaphors that can go either way: maybe the economy is like a car, and fiscal stimulus is a push from a few friendly passers by; maybe it’s a dog race and the fed sets the speed of the rabbit; maybe it’s a babysitting coop. Or maybe it’s none of those things.

This has important implications for economics. In her 1986 book ‘The Rhetoric of Economics‘, Deirdre Mckloskey argues, among other things, that economic models are basically just metaphors. This is true. As economists like to point out, they seem to care little about whether a model adequately represents the structural mechanics of a system and instead they (supposedly) look at its conclusions. In other words, we don’t have a model of the economy – we’ve got a metaphor for what the economy could look like if it were something different. This is not useful.

P.S. The blogging hiatus didn’t last

, ,