“Devastating storm may ultimately boost US GDP” read the headline on CNBC’s Market Insider. Much like the debate around price gouging (addressed here), every storm or natural disaster seems to bring with it a discussion of whether physical destruction, or at least the aftermath and reconstruction arising from it, is somehow “good for the economy.”

Now, the particular CNBC headline above may prove to be right or it may prove to be wrong for a given period, as I’ll explain below. It’s purely an empirical matter. But examining the economic impact of hurricanes such as Harvey through assessing movements in short-term GDP alone is clearly a very partial account of the legacy of such a storm.

Firstly, the obvious point. Hurricane Harvey has destroyed property and hence destroyed wealth. This is unequivocally bad for the economy. Studies from Goldman Sachs and others estimate the economic destruction at anywhere between $30–40 billion.

Yet wealth is a “stock” concept, whereas measured GDP is a flow of activity in a given period. In principle then it is less obvious what the short, medium and long-term impacts of a hurricane on measured GDP (a measure of economic value-added at market prices) would be.

The very short-term impact of such a storm on GDP is almost certain to be negative. Hurricanes destroy productive capacity, disabling factories and, in the case of Texas, curtailing the oil refining sector. This acts as a negative supply-shock to both the local and national economy, with higher gasoline prices (an input to travel and production processes) filtering through to raise input prices and hence production costs.

Of course, as the effects of the storm wade, rebuilding activity and construction will begin. This will count towards GDP, and as much of GDP relates to voluntary activity that has real value-added, so it should. People will genuinely want to replace destroyed or damage homes, cars, fences, and the like, and this is valuable economic activity. In Houston this reconstruction and redevelopment is likely to be quicker than if it occurred in many other areas, due to the less restrictive zoning laws and hence lower transaction costs associated with new building.

But will all this increase GDP overall, relative to a counter-factual in which the storm had not taken place?

As my colleague David Boaz has previously written, to look at this activity alone would be to fall for what Frederic Bastiat described as the “broken window fallacy”. If a window is smashed, one observes the window being repaired and the associated spin-off ventures as economic activity. What is unseen is the economic activity that would have existed had people made decisions to save for a college degree, buy a new suit or invest in a start-up with the resources which they have now have to put towards rebuilding their house or replacing their car.

Whether the counterfactual path of GDP would be exactly the same or even higher than the post-storm economy is difficult to say. If there are unused resources, then it is theoretically possible that observed GDP could be temporarily higher for a period than it would otherwise have been absent a hurricane whilst construction is taking place. But that does not appear to be the empirical record. In fact, the same CNBC article shows that the New York economy performed substantially worse in the 12 quarters after Hurricane Sandy relative to the national economy.

And in the longer-term the consequences are certainly negative, whether represented in measured GDP or some broader conception of economic welfare (see this study, for example). Even if there were a case when GDP did look higher than expected for a while, this would not mean we were “richer” or somehow better off in an economic sense.

We would have expended a bunch of resources to get back to where we were before the storm, and lost out on much other valuable activity that we would have preferred to have freely chosen to engage in. Circumstances would have altered our choices, and to the extent this meant we would have adjusted our preferences to buy certain repair goods and services, this would show up as economic activity. But overall, whatever measured GDP shows in the next few months, we would still be economically poorer for the destruction wrought by the storm, due the opportunities and decisions we were unable to make as a consequence of it.

In fact, in the very long run the only way a super-damaging storm such as this could improve the economic performance of an economy is if its effects led us to reassess damaging policies such as subsidizing flood insurance.