and computer codes have greatly facilitated the ±tting of these models to data, and solving for the endogenous variables both backward and forward in time. These models are generally more acceptable for capturing funda- mental features of economic growth in which a country is in transition to long-run equilibrium. In my view, Prof. Patrick Kehoe and associates have advanced the Prescott methodology making this class of theory to be very competitive with the salt water approach (at least faculty at Princeton°s the Dept. of Economics must think so since Kehoe is now a member of their faculty). Must our structural models capture-explain the data? Yes and no. Sev- eral papers posit a phenomena seen in the data and/or previous literature that is paradoxical. Example 1 : based on trade theory, trade liberalization should increase factor payments to a country°s relatively most abundant resources (say un- skilled workers) relative to lessor abundant resources (e.g., skilled labor). A model is ±t to data, simulations performed, and ...sorry...results are not ob- tained consistent with this observation, suggesting theory is incorrect and missing of some key component. The author then changes the production technology so that unskilled and skilled workers are compliments, not sub- stitutes. Now, the theory explains about 50 percent of the variation in the data and thus serves as a partial explanation of the phenomenon. Example 2: the resource curse comes from the observation that coun- tries whose natural (exhaustible) resource share in the value of total exports to GDP grow more slowly than other countries. Some studies ±nd that even after controlling of political economy, this phenomenon still "seems" to exist, it even exists among states in the U.S.; a recent paper in JEEM, using state of the art econometrics, says no, this is a "red herring." A very re- cent paper argues that the JEEM paper, lacking sound theory and therefore not knowing what variables to condition the analysis upon, is wrong. The new paper shows for a two country world with plausible parameters that the resource curse can exist, but less likely so for countries that are already among developed countries. However, only about 1/3 of the growth gap is explained, leaving 2/3 to other explanations, such as institutional structure and political economy. Example 3: applied general equilibrium models ±t to country data were used extensively to help explain the likely outcome of NAFTA, and more recently world-wide reform of agricultural policies as proposed under the DOHA round of trade negotiations. An expost review of the NAFTA models 20
to see how well they predicted found signi±cant errors. The errors were traced to the models°inability to account for di/erences among ±rms in an industry. The data show that only selected ±rms expand trade after reform, while others in the same industry may actually contract, and new more export oriented ±rms seem to emerge. Trade at the extensive margin (meaning new export oriented ±rms) were not captured by the old models.