TS sproj GDP Growth 0221100714040106031607
[NEAS: The candidate examines Gross Domestic Product from January 1950 through January 2006. Several candidates use GDP for their student projects, and some errors occur repeatedly. Other NEAS postings provide more guidance on GDP, time periods, second differences, and moving average models. The NEAS guidance and past student projects help you avoid common errors and complete your project efficiently.
This student project illustrates a common error in the student projects. The error seems small at first, and you should catch it by examining the graph of the first differences or its correlogram. But the textbook is not clear, and use the wrong correction. This student project is clearly written, and you can see the progression of the error.
~ The candidate combines time periods with different trends (different GDP growth rates).
~ The resulting correlogram has positive sample autocorrelations for many lags, followed by negative sample autocorrelations for many lags, with a slow climb afterward to zero.
~ He incorrectly takes a second difference to make the time series stationary.
~ He observes a negative sample autocorrelation of lag 1.
~ He incorrectly assumes a moving average process.
You may learn more from following an inadvertent error and seeing its effects on ARIMA modeling than from memorizing a list of rules. Checking for changes in the ARIMA process saves you much time in your student project.
Several student projects on GDP, CPI, and interest rates make this error. This student project is well written, and you can trace the error through each step of the analysis.]