commentary

(Financial Times)

August 31, 2011

All Models Fail in Certain Situations

by Krishna B. Kumar

Sir, In his article "A realm dismal in its rituals of rigour" (Analysis, August 26), John Kay provides a useful critique of economic modelling.

A model is like a map. It provides a schematic representation of the path from point A to point B. If one follows a map, but finds an overnight storm has felled a tree that blocks the path, one just takes a detour until the path can be restored or lays a new road if the damage is irreparable.

Models are metaphors, and all of them fail in certain situations, but there is much to be learnt from why and how exactly they fail. The best of economists — and you can include the originators of the dynamic stochastic general equilibrium paradigm among them — readily see the fallen tree. It appears one cannot fault them for not foreseeing all the ways in which their paradigm could have run into trouble, just as one cannot fault the meteorologist for not seeing which particular tree would fall in a perfect storm, or the epidemiologist for not seeing the Aids virus would jump from monkeys to human beings.

Prof Kay also provides a useful defence of induction over deduction in economics. In that vein, one does not need the deductive powers of "Ricardian equivalence" to see that fiscal stimulus and its sister policy, monetary stimulus, are not all they are touted to be. Japan tried both to little effect in the 1990s. And, with all the spending by governments and money injected by the central banks, the US and Europe should be booming by now. Instead the mechanical effects of government spending on the economy have faded and most of the injected liquidity is being held by banks as excess reserves.

Perhaps fiscal and monetary policies have not been effective in the US because the underlying problems may be structural. A key problem facing the technology-laden globalised US economy is the gap between skills that employers demand and the unemployed have. Ironically, the DSGE paradigm can be quite useful in providing guidance on solving this problem. It can capture the deep factors affecting the supply and demand of labour. And it is amenable to conducting counterfactual experiments on training and education policies to assess their relative effectiveness.

Krishna B. Kumar, Senior Economist, RAND, Santa Monica, CA, US

This commentary appeared on Financial Times on September 1, 2011