Economics 312

Spring 2012

Econometric Project #5

Due 6am, March 27

Based on Hill, Griffiths, and Lim, Problems 9.14 and 9.22

This project builds on two HGL problems related to stationary time-series models. Note that there are additional aspects of these model that I would like you to explore in addition to the questions in HGL.

Problem 9.14 explores the response of crop planting to prices using 34 annual observations from Bangladesh. Perform all of the analyses that the authors require in parts (a) through (e), plus the additional tasks below:

• When estimating the original (unrestricted) model in part (a), test whether the error term is autocorrelated using a Breusch-Godfrey test or a Box-Ljung Q test (or both). If appropriate, use Newey-West (HAC) standard errors in all of your analysis to correct for autocorrelation.
• When you find the estimates for the restricted β_s parameters from the estimated α parameters, find and discuss their standard errors and t-statistics as well.
• Test the hypothesis that the linear restrictions on the lag-coefficients hold. (If you find that you need the Newey-West standard errors, you should use the test command after running the newey command in order to use the robust SEs in the test. The familiar test based on SSE_r - SSE_u assumes that there is no heteroskedasticity or autocorrelation.)
• If there are alternative specifications that suggest themselves (and you have time), go ahead and explore them.
• Note that you will need to create a "year" variable in Stata in order to define the dataset as time series. Since we don't know the year numbers, just use gen year = _n to create one starting at 1. Once you have done this, use tsset year to declare year to be your time variable.
• Write up a summary analysis of your results, including in it the usual validity assessment.

 Problem 9.22 considers the dynamic relationship between income and consumption using quarterly U.S. data. 

• For this problem, analyze using appropriate methods the time-series relationship between income growth and consumption growth, considering consumption growth to be the dependent variable. Use parts (a) through (g) as suggestions for possible ways to proceed, but you don't need to answer them explicitly. (You may also wish to take a glance at problems 9.23 and 9.24, which use the same dataset.)
• The do-file time.do can be downloaded and run. As part of your data introduction, look at this file and explain how it creates the Stata variable time that corresponds to the appropriate sample period.
• Write a full description of your analysis, emphasizing the logic that you used to decide on specifications and estimators. Your validity assessment should include discussion of whether or not income growth is exogenous.

Links to Datasets

Problem 9.14: bangla.def bangla.dta
Problem 9.22: consumptn.def consumptn.dta

Project Teams