Prescott’s article begins by saying that for a long time, economists have not really known what to make of large fluctuations in output and employment over relatively short periods of time. Movements as much as ten percent are observed while movements in labor’s marginal product are small. He says that these observations should not be puzzling and they are in fact what should be predicted by economic theory. He said given the conditions in the United States, it would be puzzling if these fluctuations did not take place. He also points out that standard theory also correctly predicts the amplitude of these fluctuations as well as the fact that the investment component is about six times as volatile as the consumption component. Prescott, with the aid of several others, looked at the American economy since the Korean War to see if it displayed fluctuations with developments in technology. Sure enough, it did. He found that when the rate of technological change was uncertain, the business cycle phenomena was both dramatic and unanticipated. The models are highly abstract and therefore hypothesis testing will reject them but they are viewed more as the promising beginning of a larger research program than the definite end and correct answer.
He criticizes the use of the term “business cycle” because it tags on the time series components and also because the term does not accurately represent the cycle aspect of the idea. He therefore classifies it as business cycle phenomena. He uses Lucas’ definition for the business cycle as being the “recurrent fluctuations of output about trend and the co-movements among other aggregate time series. The trend is neither a measure nor an estimate of the unconditional mean of some stochastic process and it would actually be problematic if it was. It is the computational process of fitting a smooth curve to the data set. He then gets into the technical aspects of fitting a line using logarithms which I do not completely understand. He does speak about using the same points on the two different data sets in order for the findings to be meaningful.
He next moves on to every macro student’s favorite model: the Solow-Swan Growth Model. He breaks down the different aspects of the growth model and explains some of the math behind it. *Since I am fairly sure all of us have had either macro or development I will not bore you with a recap. If you need to see it again it is on pages 370-373. He does criticize that this model should not be used to look at the business cycle because one of the restrictions is that neither employment nor savings vary. He moves on to show that one approach used to look at this problem is the Pareto optima approach. “Given a single agent and the convexity, there is a unique optimum and that optimum is the unique competitive equilibrium allocation. He next introduces expectations into the growth model by making uncertainty the household’s expected discounted utility.
He next shows how data can be used to restrict the growth model and two of the most important parameters are intertemporal and intratemporal elasticities of substitution. One thing that must be looked at is the Cobb-Douglas function and if it supports the fact that the American real wage has increased more than 100 times since the Korean War. Again he gets into math and equations I don’t quite understand but I think the main point is that traditionally the elasticity of substitution between consumption and leisure has been close to 1. Also that the real wage has increased steadily in the US.
The nature of technological change is the next thing Prescott explores. He cites how it can be calculated in the Solow model by the changes in output minus the sum of the changes in labor’s input times labor share and the changes in capital’s input times capital share. Measuring variables in logs, this is the percentage change in the technology parameter of the Cobb-Douglas production function. He points out that Solow’s model overestimates the standard deviation of technological change and therefore the variability of that parameter. He says there are non-negligible errors in measuring the inputs especially with regards to labor. He summarizes this section by saying that technological shocks are highly persistent and that tying down the standard deviation of the technology change shocks is very difficult.
Next Prescott goes on to describe the statistical behavior of the growth models and how theory provides an equilibrium stochastic process for the growth economy studied. Again, he gets technical and I get lost. The basic growth model has a standard deviation of the technology shock equal to 0.763. Theory implies that the standard deviation should be 1.48 percent; in fact it is 1.76 for the post-Korean War American economy. The difference appears to be due to errors in measuring aggregate hours and output.
The Kyland-Prescott economy modified the growth model in two ways. First, they assumed that a distributed lag of leisure and the market-produced good combine to produce the composite commodity good valued by the household. The second modification is to permit the workweek of capital to vary proportionally to the workweek of the household. The Hansen indivisible labor economy shows that when movements between employment and nonemployment are considered and secondary workers are included, elasticities of labor supply are much larger. He also finds that most of the variation in aggregate hours arises from variation in the number employed rather than in the hours worked per employed person. He finds that if the technology shock standard deviation is 0.71, then fluctuations in output for his economy are as large as those for the US economy. Also, variability in hours is 77 percent as large as variability in output. Empirical labor elasticity is found to larger than the true elasticity. One reason this model does not accurately display the numbers observed is because of cyclical measurement errors in output.
This is a very difficult article I think. There is a lot of very technical information with regards to models and that makes it kind of tough to follow. I understand what he means about errors in measurement affecting the results and the discrepancies between their models and the observed values however, I do not quite follow some of the logic he uses within the article.
[...] [...]