Thursday, July 29, 2010

Economist and minimum wages

From an article on Brazil:
How much of the credit does Lula [the current president] deserve for all this [growth]? ... He also raised the minimum wage by two-and-a-half times since 2003, taking its purchasing power to its highest level since 1979. This has not destroyed jobs: some 13m new jobs in the formal (ie, legally registered) economy have been created since 2003."
Oh dear! The implicit estimator here is the before-after estimator or, if you are a more expensive consultant, the "interrupted time series design." It requires, to produce a consistent estimate of a causal effect, that absent the change in the minimum wage, the change in the number of jobs would have been zero.

Given that Brazil increased its minimum wage in the midst of a boom, this assumption seems highly unlikely. Thus, the correct comparison would be between the number of jobs created in a boom with minimum wage increases and the number created in a boom without minimum wage increases. Both will be large positive numbers. The fact that the first is a large positive number is uninformative about the sign of the difference.

How big the difference would be depends on many factors, including how binding it was to begin with, something the Economist piece is silent about.

One expects this sort of basic error in counterfactual reasoning from the NYT or the WaPo, not the Economist.

No comments: