Disasters and Growth

Former student Curt James sent along this link to an article by Frank Hollenbeck about a topic that we’d discussed in class: are disasters good for the economy?

Outside of economics, it’s relatively common to say “oh … boo-hoo about that disaster … but at least it increased growth rates”.

This is known as the broken window fallacy, and goes back to Bastiat.

In contemporary times, people fall for this because of our focus on GDP (a flow variable), and our lack of decent information on national wealth (a stock variable).

The destruction of disasters is a problem because what is wrecked is what is already built. This is part of national wealth, but because we measure that badly, and ignore it most of the time, people are willing to forget about it.

And … it is theoretically true (and empirically confirmed) that disasters increase growth rates. The reason is that all the rebuilding gets counted in GDP.

This all makes sense in the context of a growth model: a disaster takes us further away from the steady state (which is bad), but the further we are from the steady state that faster we move towards it (this is the improved growth rates that people harp on).

Obviously, if you count one part and not the other, you might conclude that disasters are good.

The linked piece reviews some of the literature, but I think it misses an important conclusion. It focuses on the 2002 (not 2007) conclusion of Skidmore and Toya that sometimes disasters do seem to make things better. This result has been cited over 300 times:

… They found a positive relationship between climate disasters (e.g., hurricanes and cyclones), and growth. The authors explain this finding by invoking what might be called Mother Nature’s contribution to what economist Joseph Schumpeter famously called capitalism’s "creative destruction.” By destroying old factories and roads, airports, and bridges, disasters allow new and more efficient infrastructure to be rebuilt, forcing the transition to a sleeker, more productive economy. Disasters perform the economic service of clearing out outdated infrastructure to make way for more efficient replacements.

Hollenbeck argues that this violates Bastiat’s argument.

But I see an argument that he misses. When we calibrate a growth model, we come to the startling conclusion that the lemonade stand story of growth under capitalism (sell some lemonade, and reinvest in the stand to sell even more lemonade) should have played itself out about 2 centuries ago. The story has theoretical support, but that same theoretical support also limits it to explaining a lot less growth than we’ve actually seen. Where does the rest come from?

The answer is labor-enhancing technology, that allows individual workers to control more capital. But then we run into the problem that high tech innovations are relatively easily transmitted across regions, and thus can’t explain regional differences in well-being. So it’s got to be low tech stuff, like culture and institutions, that makes the difference.

How can low tech be growth incompatible? The usual explanation is that it ossifies in place inefficient uses of capital. Typically there’s a political explanation behind this: someone benefits from discouraging more efficient uses, and they use the political system to keep restrictions in place that limit efficiency.

If this is the case, then I see some merit in the results of Skidmore and Toyo that Hollenbeck has missed. Perhaps disasters help out backwards placed because they are most incident on the capital that has passed its useful lifetime. Essentially, corruption preserves capital that stinks, and a disaster can destroy it and create a void that can be filled with more efficient capital.