Archive for the ‘e488-rbc’ Category

A Response to Summers’ “Some skeptical observations on real business cycle theory”

Thursday, March 13th, 2008

Lawrence Summers starts out by comparing the foundations of Keynesian macroeconomic theory to that of astrological science.  That is, both are “premised on the relevance of variables that are in fact irrelevant.”  According to real business cycle economists, Keynesian economics didn’t explain the macro economy because it wasn’t based on microfoundations.  As we have learned in class, RBC models are based off of utility maximization and profit maximization principles, which evolve at the microeconomic level.  Summers brings Prescott’s “Theory Ahead of Business Cycle Measurement” into his critique of RBC models.  Prescott’s article is essentially asserting the claim that the theory cannot be currently tested because there does not yet exist measurement tools capable of testing the theory.  Nevertheless, Summers critiques Prescott’s article on four levels–the parameter estimation, the shocks present in the model, price levels, and exchange failures.  Summers goes on to say that throughout history theories have been developed that seemed plausible (or at least a good starting point) because they “mimic” or approximate reality well enough for that period in time.  However, as measurement tools improve and people are made more aware of their surroundings, theories will change (i.e. the Earth was at one point considered the center).  This is what this critique attempts to accomplish–that is, to determine whether or not Prescott’s theory mimics the economy in its current state coincidentally or if it captures the observed business cycle trends.

With respect to the parameter estimates, Summer can find no evidence to support Prescott’s claim that the one-third of all household time is devoted to market activities.  Most other studies, such as Martin Eichenbaum, Lars Hansen, and Kenneth Singleton (1986) have estimated that to be only one-sixth since 1956.  In addition, Prescott’s model assumes the average real interest rate to be four percent.  Over his thirty year study, however, the real interest rate only averaged out to one percent.  Summers’ last critique with regards to parameter estimates is Prescott’s inability to display evidence that supports his claim on the elasticity measurements of labor supply.  According to Summers’ reading of many studies, labor supply is only minimally effected by changes to the real wage.

Like many RBC models, the observed cyclical behavior is caused by external/exogenous shocks, known as technological shocks.  However, the critique makes the claim that he doesn’t have nay evidence to support the business cycle movements.  Even the oil shocks of the early 1970s haven’t contributed to “large movements in measured total factor productivity.”  In small sectors of the economy, however, such as in the mining and construction sectors, negative productivity growth has been observed.  In addition, technology shocks may not be as large as originally thought.  Studies have shown, especially in Jon Fay and James Medoff (1985) that the reason for cyclical behavior is due to firms holding more labor than necessary during troughs.  Known as “labor hoarding,” firms may hold labor in excess of regular production requirements during recessions because it is deemed more costly than the wage rate to hire or fire employees.  Therefore, when the economy is experiencing a peak, the labor force is being fully utilized and shown as productive.  However, when a recession occurs, those excess laborers are kept on rather than fired, which makes it appear as each labor unit is now less productive and factor productivity will decrease.  While Summers doesn’t think of “labor hoarding” as a technological shock, many RBC economists do think of any deviation from the long-run potential output as a technological shock.

The third argument that Summers makes is the absence of price data in his model.  I don’t understand this paragraph because I do not comprehend what Summers means when he argues that Prescott’s model was tested without price data.  To me, it is unfathomable as to how any economics can be empirically tested without paying attention to price.  I will have to do some exploration as to what a “price-free economic analysis” means in the eyes of Summers.  The last objection to the Prescott model is the inattention to exchange breakdowns.  From the critique, it is mentioned that studies that analyzed the Great Depression made it clear that firms had output to sell and workers wanted to exhange their labor for those products, but there was a breakdown in the exchange mechanism of labor for products and the exchange never transpired.  This “breakdown” caused U.S. GNP to decline fifty percent over the years 1929 to 1933.  The best explanation for these exchange mechanism failures is due to the credit market breakdown during the same time period.

Summers sums up the critique to say that economists will continue to be better at explaining behaviors of individual economic agents than explaining the equilibrium in markets when many economic agents interact.  With that being said, Summers stresses that the importance of being able to explain why exchange markets breakdown, and if that can be accomplished, then these macroeconomic models will be able to help forecast economic fluctuations.  What I was hoping for the critique to address was the parameterization issue that was discussed in class on 3/12.  If models are constantly changing to fit the data, how can we be sure that it’s the best model out there and not one that simply “mimics” the available data–which is something that Summers starts out discussing?

Source: Summers, Lawrence H.  1986.  Some skeptical observations on real business cycle theory.  Federal Reserve Bank of Minneapolis Quarterly Review (Fall): 23-27.  (This can be found in the Reader.)

A Response to Stockman’s “Real Business Cycle Theory: A Guide, an Evaluation, and New Directions”

Thursday, February 28th, 2008

Alan Stockman comes right out and states the purpose of real business cycle (RBC) models–that is, these models are used to “explain aggregate fluctuations in business cycles without reference to monetary policy.”  In fact, he goes on to make four assertions as to why the real business cycle model is important.  From evidence gathered, monetary policy does not affect the real output as much as economists once believed.  Second, even if it does affect the real output of the economy, it isn’t the driving force behind the business cycle, supply shocks and non-monetary occurrences usually influence the aggregate fluctuations to a greater degree than monetary policy, and lastly, real business cycle models can be used to determine how disturbances affect different sectors of the economy.  These real business cycle models incorporate the following: GDP, consumption, fixed assets (i.e. investment, nonresidential, structures, equipment), average nonfarm employment, and capital stocks.  Stockman notes that others have used variations of the real business cycle model to include cross-sector analyses to track output and production across the aggregate economy in order to trace the disturbances from one sector to another.

Stockman starts the discussion on real business cycle (RBC) models with two assumptions–people maximize their utility with different combinations of leisure and consumption and there’s a technology coefficient that allows for the transformation of capital and labor into output that can be either consumed by households or reinvested back into capital stocks for period t+1.  Stockman traces some early, prototype models that had their roots with RBCs.  One of the first was Kydland and Prescott, who used backwards induction to come up with an abstract model that looked at utilization of capital, lagged effects of leisure on utility, and imperfect information about productivity.  Hansen furthered the Kydland-Prescott 1982 model by adding a “lottery on employment,” in which people are assumed to either work full time or not at all, without looking at part-time work.  This lottery of employment assumes that people that choose to work and those who choose not to work are randomly selected.  The Greenwood et al. model looks at current and future investment.  It is noted that the model shows increases in consumption, labor supply, output, and investment are due to current economic conditions.  The model discusses technology shocks as one factor, but that would only increase future output resulting from increases in future capital.  Current output, however, is affected by current economic conditions.  Kydland-Prescott’s 1988 model incorporates that the cost of a greater utilization of capital is a greater utilization of labor.  This variable work week (a longer work week) predicts the variability in the U.S. inventories.  Parkin (1988) tried to calculate the parameters in the Cobb-Douglas production function using labor data from the GDP.  He determined that these parameters varied over time and as a result, was able to calculate the technology shock.  Because his model showed the share of leisure s(preference shock) relatively stable over time, this implied that preference shocks can be viewed as unimportant to RBC models.  The last model mentioned is that of Christiano and Eichenbaum (1988), which looked at government shocks as affecting shifts in the labor supply curve.  Coupled with technology shifts, these two movements could induce changes to the real wage.

Stockman starts his discussion on RBC models by listing parameters often included in these models.  They include the fraction of total time spent working and the time spent in leisure, the psychological factor, the rate of capital depreciation, the marginal rate of substitution in consumption, the marginal rate of substitution in leisure, labor as a percentage of GDP, and the variance of productivity shocks.  However, the one common criticism is how to use the RBC models to explain periods in which real output falls using the logic that negative technology shocks exist.  (This is most commonly observed in smaller sectors of the economy.)  What Robert Hall (1988) deems the most important is the ability to differentiate between temporary reductions (aggregate output) from permanent reductions (measured output).  However, I don’t understand the difference bewteen the above logic dealing with negative technological shocks and the difference between measured output and total output (page 32).

An interesting note when discussing the criticisms, Stockman observes that econometric tests may reject the RBC model.  This is because the models may be wrong due to large influences in measurement errors.  Even if they prove to be wrong, RBC models have been shown to give better advice for policymakers than other incorrect theories.  Another criticism is that involuntary unemployment isn’t explained by RBC models.  The example looks at two individuals–both of whom have the same tastes and preferences and characteristics, but the only difference is that one is employed and one is unemployed.  However, the model will not differentiate between the two; the only explanation is that unemployment is the result of a random fluctuation in productivity within the economy.  Going back to the technology criticism, Stockman proposes the question that technology shocks may be more influential to a group of industries rather than the nation as a whole.  Thus, this brings into question that though technology shocks are important, nation-specific disturbances play at least as large of a role in output fluctuations as these technology shocks.

To see if the RBC model explains the random fluctuations of the international economies, there are two “tests.”  One is to see if the goodness of fit improves with the addition of more variables, but this requires the addition of more equations and more parameters.  The other is to test the RBC model with the same criteria and parameters to a different set of macroeconomic facts.  These RBC models should work fairly well because exchange rates look at the relationship between currencies (relative and nominal) due to “real shocks,” which are what RBCs are trying to model and explain.  However, the one obstacle that will be encountered is that various economies have different parameters and disturbances.

The ultimate question is whether these real business cycle models should be used for optimal policy decisions.  If the assumptions of RBC models are true, then monetary policy will have no effects on the real output of the economy.  To achieve optimal policy responses, as was suggested in our reading of Kydland and Prescott, there can be no fiscal and monetary interventions because policies are made based on expectations and conditions in time t.  The policy is implemented assuming the status quo, but once the policy is implemented, in time t+1, t+2, … , t+n, expectations will change and the policy will not achieve the optimal response.  Another thing to consider, which goes along with the Kydland-Prescott model is that the responses of the economy (to changes in regimes/administrations) bring about these suboptimal conditions because of market failures.  Stockman ends the discussion by stating that fluctuations in the economy are most likely optimal resopnses to uncertainty, rather than the failure of markets to clear, which was the general thinking of Keynesian economists and monetarists.  Essentially, the RBC should be concerned with long-run rates of technological change and low inflation instead of large fluctuations in GDP because those fluctuations are random and shouldn’t be “massaged” by either monetary or fiscal intervention.

Source: Stockman, Alan C.  1988.  Real business cycle: A guide, an evaluation, and new directions.  Economic Review 24 no. 4 (Quarter 4): 24-43.