The French economist Frédéric Bastiat, writing in 1850, proposed what is known as the Broken Windows Fallacy, the idea that naïve observers, examining a scene where something useful has been wasted or destroyed, consider the beneficial visible economic effects, (increased demand, to replace what is wasted or destroyed) while ignoring the indirect detrimental effects (reduced demand for other products). To put it in M. Bastiat’s own words – well, his translated words, anyhow, courtesy of the Library of Economics and Liberty:
Have you ever been witness to the fury of that solid citizen, Jacques Bonhomme, when his incorrigible son has happened to break a pane of glass? If you have been present at this spectacle, certainly you must also have observed that the onlookers, even if there are as many as thirty of them, seem with one accord to offer the unfortunate owner the selfsame consolation: "It's an ill wind that blows nobody some good. Such accidents keep industry going. Everybody has to make a living. What would become of the glaziers if no one ever broke a window?"
Now, this formula of condolence contains a whole theory that it is a good idea for us to expose, flagrante delicto, in this very simple case, since it is exactly the same as that which, unfortunately, underlies most of our economic institutions.
Suppose that it will cost six francs to repair the damage. If you mean that the accident gives six francs' worth of encouragement to the aforesaid industry, I agree. I do not contest it in any way; your reasoning is correct. The glazier will come, do his job, receive six francs, congratulate himself, and bless in his heart the careless child. That is what is seen.
But if, by way of deduction, you conclude, as happens only too often, that it is good to break windows, that it helps to circulate money, that it results in encouraging industry in general, I am obliged to cry out: That will never do! Your theory stops at what is seen. It does not take account of what is not seen.
It is not seen that, since our citizen has spent six francs for one thing, he will not be able to spend them for another. It is not seen that if he had not had a windowpane to replace, he would have replaced, for example, his worn-out shoes or added another book to his library. In brief, he would have put his six francs to some use or other for which he will not now have them.
I’ve always contended that M. Bastiat was wrong. Based on my own introspection, what would happen if someone broke one of my windows? Would I reduce my spending on something else in order to pay for the broken window? Not at all: I would “save” less, in the sense that, at the end of the month, the increase in my bank balance would be less than it otherwise would have been, but I would not feel a need to reduce my standard of living temporarily in order to pay for the glass.
Moreover, when I was younger and didn’t make enough money to have savings left over at the end of the month, I would merely have let my credit card balance run up a little to pay for the glass. (If such losses proved frequent, I would – and in fact did, back in my poorer days – take out a home equity loan to pay off the credit card balances.) I imagine (perhaps wrongly) that most people, or at least many people, are like me in this respect.
This is old news, and M. Bastiat’s defenders have several counterarguments, none of which I find convincing. First they argue that, even if I don’t reduce my current consumption of other products to pay for the glass, I will have to reduce my future consumption because I will have less savings. This logic seems to presume that I intend to go to my grave with a net worth of zero, or with some net worth that I will decide beforehand independent of the window-breaking incident.
That presumption is wrong. I intend to go to my grave with a positive net worth, because I am uncertain about when I will die, and I am averse to the possibility of running out of wealth before I die if I underestimate my lifespan. Nor is my intended net worth at my expected time of death a fixed number. As long as I expect that number to be positive enough to leave minimal risk of outliving my wealth, I will consume what I consume and not worry about the exact number. So no, I won’t reduce my future consumption.
Perhaps so, they may argue, but surely your heirs will then have to reduce their consumption. The problem there is that I expect my heirs to have a predisposition similar to mine, and to continue their lifestyle indifferently to a one-time bequest. The Bastiatites may then take the argument one step further and talk about the heirs of my heirs. And we can keep this argument going all day as I apply the same logic to each successive generation. At the end of time, perhaps, some distant heir will have to reduce their consumption – but perhaps not, because the usual rules of trade don’t apply when people believe that the end of the world is imminent. And if it comes as a surprise, it will obviously not reduce the heir’s consumption.
All of which is a red herring, because the real argument is that, even if I don’t reduce my consumption, someone else will have to reduce their consumption. If I don’t reduce my consumption, I will (as I acknowledged) have to reduce my bank balance by the end of the month, and there will be less for the bank to lend to others, who will thus have to reduce their consumption (or investment). Not really, though. If the bank has a disposition similar to mine, it will merely reduce its excess reserves in the same way that I reduce my saving. Or the Fed, which has a policy of targeting the interest rate at which banks borrow, will replenish the bank’s reserves in order to prevent that interest rate from rising.
But, the Bastiatites may contend, eventually, if enough windows get broken and enough people reduce their savings, the Fed will have to raise interest rates to reduce the additional demand (as the demand for glass increases while the demand for other things initially remains constant), lest it stress the economy’s resources and produce inflation. My window could, as likely as any other, be the straw that breaks the camel’s back. (I’m trying to imagine a window made of straw.) The general principle is, Fed or no Fed, and no matter how profligate I may be personally, the economy has limited resources, and if some of those resources are diverted to produce more glass, fewer resources will be available to produce everything else.
But here they are wrong again. (I will skip the argument about inventories here, since I probably lose that one.) The logic of limited resources only applies when the economy is using most of those limited resources. If there are slack resources, we need merely mobilize some of the slack resources. If the economy is operating below full employment, as is often the case, then there is no need for the Fed to raise interest rates. The window-breaking incident will indeed create additional net employment, just as the naïve onlookers thought.
Here the argument becomes more subtle. M. Bastiat’s defenders will argue that there is no fixed point of full employment. Rather, one can merely say that one state of employment is, as it were, “fuller” than another. In the terminology of contemporary Keynesians, there is a “Phillips curve” or an “aggregate supply curve,” which is not flat, and which, econometricians typically assume, is roughly a straight line.
Moreover, there is a certain point on that curve that corresponds to the non-accelerating inflation rate of unemployment [NAIRU]. If the unemployment rate is higher than the NAIRU (which, they will argue, is what I must mean by “slack resources”), then the inflation rate will decline, and the Fed (assuming its long-run inflation target is unchanged) will then allow unemployment to fall below the NAIRU in the future, to bring us back to the original inflation rate. If my glass purchase reduces unemployment, then the initial decline in the inflation rate will be smaller. There will therefore be less room to increase the inflation rate in the future. Thus, in the future, the Fed will not allow the unemployment rate to go down as far as it would have if the glass had not been broken. The additional employment today will be offset by reduced employment in the future.
But yet again they are wrong. If the Phillips curve were actually a straight line, their argument would be valid. But the Phillips curve is not a straight line. That’s pretty obvious. The unemployment rate can’t go below zero, so either the curve has a kink at zero (where it becomes vertical), or it has some convexity as it approaches zero. The latter possibility seems infinitely more likely to me, since I cannot imagine that a decline in the unemployment rate from 10% to 9.01% has the same effect on inflation as a decline from 1% to 0.01%. When the number of available workers becomes extremely small, the difficulty in obtaining workers becomes extremely large, and even a little bit of additional demand will necessitate a huge increase in prices, if only to cover the incredibly large recruitment costs.
I contend that the Phillips curve has considerable convexity throughout. Does one really believe that reducing the unemployment rate from 20% to 19% would involve any noticeable change at all in the pattern of prices, let alone a change as large as, say, reducing the rate from 5% to 4%? Does it make any sense that the curve would be vertical on one end, almost horizontal on the other end, but a straight line on the range in between? Not to me.
In particular, I am convinced that, when the unemployment rate is below the NAIRU, the Phillips curve is steeper than when the unemployment rate is above the NAIRU. When the unemployment rate is above the NAIRU (as it surely is now, for example), a certain increase in employment (due, for example, to broken windows) will be associated with a smaller – by a certain amount – decline in the inflation rate. When it comes time to make up that decline by allowing unemployment to fall below the NAIRU, the necessary decline in the unemployment rate will also be smaller – but by a smaller amount.
Perhaps an example will make this a little clearer. Suppose that the unemployment rate is 10% and that the inflation rate falls from 5% to 2% – in accordance with a hypothetical Phillips curve – over the course of a year, and let’s say 2% is the target. Then the unemployment rate falls gradually but remains “too high” for some time, and the inflation rate continues to fall. Suppose inflation falls to zero by the time the economy gets back to equilibrium, and let’s say this corresponds to an unemployment rate of 5%. To bring inflation back up to the target, the Fed now allows unemployment to fall from 5% to 2.5% for a year, and let’s say this is just sufficient. That is our baseline case.
Now, going back to the beginning of the example, suppose an epidemic of broken windows causes the unemployment rate to be 9% instead of 10%. Inflation will fall more slowly, so let’s say it falls from 5% to 2.5% over the course of a year. Subsequently, following a path roughly parallel to the baseline case, the inflation rate falls to 0.5% instead of 0.0%. Again, the Fed wants to bring inflation back to the 2% target by allowing the unemployment rate to fall below 5% for a year. How far below? Not as far as in the baseline case. Maybe to 3% this time instead of 2.5%.
The broken windows originally caused an additional 1% of the labor force – about 1,500,000 people – to be employed for a year. The reversal process has caused an additional 0.5% of the labor force – about 750,000 people – to be unemployed for a year. Jacques Bonhomme’s son and his fellow vandals have created a net 750,000 jobs.
(I have left some I’s undotted and some T’s uncrossed in the example above. But you see it is already getting complicated and not so easy to follow. If I’m going to avoid writing a whole book here, you’ll have to take my word for it: qualitatively, the argument works, as long as you believe in a Philips curve that is convex in the region of the NAIRU.)
I’m not advocating that the stimulus package include funding for slingshots. The window-breaking solution does involve some net loss of production (750,000 person-years' worth, in my example – without dotting the I’s). There are certainly more productive ways to stimulate the economy. If you can create 750,000 jobs without foregoing the additional production, that’s obviously better. That is, in fact, one sense in which the so-called Broken Windows Fallacy is indeed fallacious: While the observers may be right to conclude that the broken window will increase employment, they would clearly be wrong if they thought it would create a quantity of employment that could not be created more profitably by other means.
Under normal circumstances, anyhow. When serious deflation becomes an issue, there is a case to be made (which I did make) that the emergency might call for breaking windows as a preferred stimulus compared to something more productive. I don’t think that’s where we are right now. But I do think, when putting together an economic stimulus, there is no great need to worry about how many windows get broken in the process. If someone insists on building a bridge to nowhere, I say build it.
DISCLOSURE: Through my investment and management role in a Treasury directional pooled investment vehicle and through my role as Chief Economist at Atlantic Asset Management, which generally manages fixed income portfolios for its clients, I have direct or indirect interests in various fixed income instruments, which may be impacted by the issues discussed herein. The views expressed herein are entirely my own opinions and may not represent the views of Atlantic Asset Management.