Econ 505: Economic Models and Forecasting Homework Assignment #2 jj espinoza Problem 1
Problem 2: Consider the following simple error correction model
Download 0.67 Mb.
|
- Navigate this page:
- A priori, what sign do we expect for the coefficient
- Do you think that the magnitude of will be different for low income groups from that of high income groups Explain.
- Estimate this error correction model and comment on results.
- Model ACF Behavior PACF Behavior
Problem 2: Consider the following simple error correction model: 
Here  is the 1st difference of the PCE,  is the 1st difference of the PDI, is the residual of the regression of PCE on PDI, and is the error term.
A priori, what sign do we expect for the coefficient? I would expect the coefficient to be positive because if there is an decrease income today then consumption will not adjust automatically, but there will be a lag in the impact of a decrease in personal disposable income (PDI) on personal consumption expenditure (PCE). The same argument applies to an increase in personal disposable income, consumption will not automatically increase, so this lag can be captured with a negative coefficient in the lagged error term. 
Do you think that the magnitude of will be different for low income groups from that of high income groups? Explain. High income groups consumption as a proportion of their Personal Disposable income tends to be lower, where as Low income groups tend to consume a large portion of any increase in their Personal Disposable income. Therefore, the magnitude of the regression coefficient of the lagged error term should be should smaller for the high income group and higher for the low income group Estimate this error correction model and comment on results. To get the first difference of the variables in the error correction model the EViews commands where: 
and the lag in the error term will be included in the regression as u(-1) in the Eviews commands; the estimation of the equation will be input into the computer in this way,  
As you can see from the equation above the coefficient of the lagged error term in the United States does have a negative coefficient which corresponds to the increasing amounts of disposable income in the United States and the lags in consumer response. Problem 3: Collect the Quarterly Industrial Production (IP) of the United States from 1980:I to 2009:II. I collected the Industrial Production Index, but I emailed the professor a question about using this measure since the one he provided (IP) did not explicitly appear in Federal Reserve Bank of St. Louis Economic Data website. Which ARIMA model does the IP series fall into? The first step is to check for the stationary property of the Industrial Production time series. Graphing the time series we see that there is an intercept with a positive trend in the time series. 
We will need to take the first difference to see if that looks more like a unit root process. Typing in the Eviews command for the first difference (series dindpro=indpro-indpro(-1)), we get the following graph of the first difference, which looks stationary. 
Despite the stationary look of the first difference of industrial production, the statistical validation of our perception that the first difference is stationary. Using a Phillip-Perron test we see that we can reject the null hypothesis that the first difference of Industrial production is a unit root process; therefore the first difference is stationary see below. 
Now that we have a stationary process we must calculate the Autocorrelation and the Partial Autocorrelation Functions and fit the model using the Box-Jenkins I.D. Table:
Directory: 2009 2009 -> The Truth about the Rwandan Genocide 2009 -> Guide to learn how to install, set up and access the music you need today! 2009 -> Lesson #11: Words with 2009 -> Terry L. Root 2009 -> The environment in the news 2009 -> Town of Nags Head Annual Report 2009 -> Cemir international Download 0.67 Mb. Share with your friends:
|