Blog o’ Burns

Prescott, Edward C. “Theory ahead of business cycle measurement.” Federal Reserve Bank of Minneapolis Quarterly Review (1986) Fall, pp9-11.

Just as a disclaimer (I know this is lame!): Since Prescott spends a great portion of his paper testing theoretical predictions of statistical properties, it was a bit difficult to follow how (and why) the statistical methodology played a major role in his theory. I apologize in advance if I interpreted anything incorrectly.

Introduction

By deriving aggregate models from micro observations of behavior, new-classical economists argue that fluctuations in output and employment despite small changes in the marginal product of labor should not be surprising. Given people’s ability to substitute consumption for leisure in a world of changing production possibilities, it would be odd for fluctuations not to occur. By using standard theory, Prescott argues that the new-classical real business cycle model has the power to predict the amplitude, serial correlation properties, and volatility of investment relative to consumption with respect to these fluctuations. He argues that the growth model should serve as the basic model describing macroeconomics just as the theory of supply and demand describes price theory. By incorporating uncertainty in the rate of technological change into the growth model, Prescott seeks to demonstrate that it displays the business cycle phenomenon. As the title of his work implies, he emphasizes the importance of theory guiding empirical analysis rather than the reverse. However, he presents a very abstract framework and acknowledges that it should only serve as a foundation for future research that must incorporate elements such as public finance, the foreign sector, and monetary factors. I can’t help but feel that Prescott is proposing a “Theory of Everything” for economics, at least more so than any prior model we have looked at. (For those who are interested, the “Theory of Everything” is kind of the Holy Grail of physics– I saw a documentary on PBS… crazy stuff!)

The Business Cycle Phenomenon

Prescott’s inquiry begins with the question: Do the the stochastic difference equations that are the equilibrium laws of motion for the stochastic growth display the business cycle phenomenon? He uses Lucas’ definition of a business cycle phenomenon as consisting of recurrent fluctuations of output about trend and the co-movements of among other aggregate time series. Prescott shows that when you plot the percentage deviation from trend of output and hours of market employment, output and hours move together at nearly the same amplitude. From this analysis, you can also see that consumption appears less variable and investment more variable relative to output. In addition, the average product of labor is procyclical but does not vary as much as output or hours.

The Growth Model

Prescott builds on the growth models of Solow and Swan, which assumes an aggregate production function with constant returns to scale, two inputs (labor and capital), constant supply of labor, and output that can be used for consumption or investment. He argues that this structure is not adequate to study business cycles because neither the employment or savings rate varies. To derive the determinants of employment level and the savings rate, Prescott looks at the household. In a world of uncertainty, the household attempts to maximize its expected discounted utility. The social optimum will be at the competitive equilibrium, assuming no externalities. Both the wage and the rental price of capital that households receive for their factors of production are a function of technology and the stock of capital.

On the other hand, firms seek profit maximization. To produce, firms try to rent labor and capital at low prices and sell their goods at higher prices. Knowing that household decisions depend on expectations of future prices with relation to the economy’s state (in terms of capital and technology) and that firms depend on the state of the economy as well, for equilibrium to occur: 1) the firm’s policy functions must be optimal given the pricing functions; 2) the household’s policy functions must be optimal given the pricing functions and level of future capital per capita; 3) all markets clear; and 4) expectations are rational. Leisure, and thus the number of hours of employment, will vary in equilibrium depending on expectations of future prices and wages.

As a last modification to his model. Prescott relaxes the assumption that technology shocks are identically and independently distributed random variables. This is necessary is technology shocks demonstrate high serial correlation.

Using Data to Restrict the Growth Model

The strength of this model rests in the fact that it uses both growth and micro observations to determine production and utility functions. Its key parameters are the intertemporal and intratemporal elasticities of substitution, with the thesis that measures from aggregate series are consistent with those from panel data. One constraint on the growth model with respect to the production function is the fact that capital and labor shares of U.S. output have been relatively constant since the Korean War. An observation that restricts the the utility function is the fact that leisure per capita displays no trend despite real wages increasing steadily. With respect to technology, it has been found that developing countries experiencing strong technological shocks have also experienced similar fluctuations in their economies to those with no technological shocks.

The Nature of Technological Change

Solow defined technological change as changes in output minus the sum of changes in labor’s input times labor share plus the change in capital input times capital share. Prescott supplements this approach by accounting for measurement error.

Some Thoughts…

Not to fudge the rest of Prescott’s article, but he spends most of the remainder of his article comparing computer generated statistical results to the actual values of the post-Korean War U.S. economy. The predicted values of the simulations were found to be surprisingly close to the actual values.

I feel like I missed a major point in this article, but I don’t think I’m smart enough to make the connection. Something seems really unsatisfying about this article, because despite all of the claims that new-classical theory incorporates micro observations, I am still not sure what Prescott is saying about how rational actors react to technology shocks. What is a technology shock? Prescott doesn’t provide any examples of what this would look like. It seems too convenient to have all fluctuations in output be caused by changes in technology, a variable that cannot reliably be measured by any current data available (correct me if I’m wrong). Maybe I misunderstood the definition of technological change, but it seems like the definition itself seems to indicate that it reflect fluctuations in the economy. Is it so surprising for x to cause y, when by definition it does so? I know I have to be wrong. Correct me please!