Economics 312

Spring 2012
Econometric Project #8

Due 6am, Tuesday, April 17

This project consists of two parts:

• A time-series exploration of the relationship between industrial production and consumer prices in the United States
• A panel-data examination of the Project STAR dataset that you used earlier.

Part I: Time-Series Relationship Between Output and Prices

The dataset USMacro_Monthly.xlsx is an Excel spreadsheet (with description file) taken from the collection accompanying the Stock and Watson text. It contains monthly data on four macroeconomic variables for the United States from January 1947 to December 2009. You may delete the PCED and Oil variables because this project focuses on the bivariate relationship between industrial production (IP) and the consumer price index (CPI). You should take logs of these variables so that differences can be interpreted as growth rates and inflation rates. In interpreting your data, remember that monthly changes in the log of production or prices would need to be multiplied by 12 to be interpreted as "annual rates."

• You are to explore the dynamic relationship between the logs of IP and CPI over the sample period (or whatever sub-sample you think is appropriate). Your exploration should consist of assessment of stationarity or degree of integration of the variables, determination of the possibility of cointegration, and estimation of an appropriate VAR or VEC model. Because the data are monthly, you should be very liberal with your lag lengths, including the possibility of 12, 24, or even 48 lags.
• Perform and interpret appropriate Granger causality tests. Estimate and interpret the impulse-response function over a 48-month horizon for your model. Is the nature of the IRF dependent on your identifying assumption about whether IP affects CPI contemporaneously or vice versa?
• Use your preferred model to forecast the two variables for the 48 months after the sample. Discuss your forecasts in detail: Does your model forecast a robust recovery or a slow one, a resurgence of inflation or stable prices? If you are ambitious, obtain data on these variables for 2010 and 2011 and compare your forecasts to the actual outcome.
• You may find the following Stata commands useful, but be sure to research them carefully to determine exactly what they do and decide what options are appropriate for your model: dfuller, pperron, varsoc, vargranger, var, vecrank, vec, vecstable, irf, and fcast. You are expected to justify your choice of options.

Part II: Based on HGL Problem 15.7

Problem 15.7 uses panel-data methods to look for effects of schools in the Project STAR dataset that you used in an early project. You are to explore models of test scores using panel-data methods. Let the sub-parts of HGL's problem guide your exporation, but do not restrict yourself to their limitations. Once again, your response should be a discussion and assessment of your preferred model and important alternative models.

Datasets

US Macro Data: USMacro_Monthly.xlsx USMacro_Monthly.pdf
Problem 15.7: star.dta star.def

Project Teams

Project teams for this assignment are below, with partners shown in the rows of the table.