Long run models in economics.ppt

Long run models in economics Professor Bill Mitchell Director, Centre of Full Employment and Equity School of Economics University of Newcastle Australia Objectives To introduce the concept of a long-run (steady-state) model in economics. To demonstrate the hazards in using econometrics to estimate the steady-state. To distinguish types of non-stationarity. To examine impulse responses and stability. To consider cointegration. Long run relations Much of economic theory is comparative static. That means it considers equilibrium or steady-state relationships. These are also called long-run relations. Usually these are cast in terms of relations between levels. What does this mean? What are the problems in estimating these models? Figure 1 Z1 and Z2 Question 1: Describe the pattern you observe and speculate a priori on whether you think there would be a relationship between these two variables and whether it would be a positive or negative relationship. Levels and Differences - Z1 and Z2 Question 3: Interpret results and confirm “eye balling” Question 4: Interpret as a money demand function? Question 6: Assume all right hand side variables take their mean values in perpetuity? Is there a unique steady-state value for Z1*? Mean values: Z1= 9.369454 Z2 = 8.435816 Z4 = -9.904742 We can thus compute Z1* from the regression. Steady-state Means: Z1= 9.369454; Z2 = 8.435816, Z4 = -9.904742. As long as there are no changes in Z2 and Z4 then Z1 will remain stable and only be subject to random shocks (with mean zero). Question 7: The alarm bells Concept of stationarity Classical inference is based on strict assumptions about the residuals. They must be white noise. These assumptions are typically violated when we use non-stationary regressors. Spurious regression problem arises – a relationship appears to exist but in fact it is just an artifact of contemporaneous correlation between the variables. Two types of non-stationarity How were Z1 and Z2 generated? They were sim