Comments

Miguel Urquiola: Cowan, Micco, and Pagés make the case that the rise of Chilean unemployment after 1998 is in line with what one would expect given the interaction of an external demand shock with wage rigidity. They further argue that minimum wage policies and wage rigidity explain the historically high, slowly declining unemployment in Chile. The authors assemble an impressive amount of data and perform a large number of exercises to back up these claims, and the paper is a stimulating and worthwhile contribution. While it contains useful direct evidence on the issues raised, many of its results should be viewed as offering solid, but nonetheless circumstantial evidence for the above points. My comments on the paper raise a number of concerns that I hope will point to avenues for further research.

The paper's first claim is that the rise in unemployment that Chile experienced after 1998 is in line with what one would expect—that it was merely a fall in output not compensated by a reduction in wages. The authors contrast this assertion with previous analyses in Chile, which suggest that it might reflect labor-saving technological change. The evidence they cite comes from a labor demand model with out-of-sample predictions for 1999-2001, together with a time series analysis that finds no evidence of structural change. These results show that the slowdown in employment growth is consistent with predictions based on the behavior of wages and output, but they do not establish that the outcome would have been different if only wages had been more flexible. For instance, up to mid-2004 the United States experienced a recovery (one that was much more robust than Chile's recovery in the period considered in the paper) with lower-than-expected employment growth. In this case, the most common explanations concerned not wage rigidity, but rather stubbornly weak business confidence leading to lower hiring than past experience suggested. Might employment demand in Chile have similarly stagnated even if wages had fallen?

Moreover, the above set of exercises in some sense simply assumes that wages in Chile are rigid. To establish this directly, the authors present two additional findings. First, they point out that throughout the slowdown, real wage growth was never negative, either overall or for different occupational groups. It would be useful to see the evolution of productivity among these groups, as well, particularly since overall productivity continues to rise throughout the period. Second, the paper appeals to cross-country evidence on the relation between changes in real wages and changes in employment, presented in figure 5. Based on this figure, the authors conclude that "real wages in Chile are among the most rigid in the group of emerging economies." While suggestive, such international comparisons are not sufficient for establishing this assertion—and in any case the statement could be more precise. Could it be, for instance, that other countries have exactly the same nominal wage rigidity as Chile, but are simply characterized by higher inflation? Could adjustment be occurring along other margins (such as the number of hours worked), and could these play a bigger role in Chile than elsewhere?

The authors then turn to how policy might have aggravated this situation. In particular, after 1998 the average public sector wage and the minimum wage began growing faster than average and unskilled wages, whereas before they had moved roughly in lockstep. This partially reflects preestablished minimum wage increases legislated in 1997 for the 1997- 2000 period. This is interesting (and potentially counterproductive) policy. It could be further exploited in future work, since it would seem to provide interesting variation. For instance, presumably the increase in the minimum wage was more binding and therefore had larger effects in some regions of Chile than in others. The paper does show some evidence of regional variation, and this type of finding is very useful in establishing that the policy was indeed potentially detrimental in its interaction with falling demand. Future work could advance in this area by controlling, for instance, for preexisting trends. One might also uncover announcement effects beginning in 1997, as well as region-specific...