UNIVERSITY OF CALIFORNIA, LOS ANGELES Department of Policy Studies
Policy Studies 208 (Cameron) - Policy Research and Analysis
Computing Lab Session #8: Serially Correlated Errors

Goals for this Lab: For this lab, we will be using the data (n:failn.dat and n:mfgprod.dat) and the initial program file (n:fail.sha) employed in the classroom handout on serially correlated errors. The dependent variable is number of business failures per month in the US (from January 1984 through October 1997; CITIBASE variable FAILN). The main explanatory variable being considered is total manufacturing production, 1987=100, not seasonally adjusted; CITIBASE variable IPMFG6).

1. Introduce access to the CITIBASE inventory of time-series data
2. Explore use of DWPVALUE and ORDER= options on OLS command
3. Explore consequences of failing to recognize AR(1) errors in regression
4. Investigate higher-order correlation patterns in regression errors
• Save residuals from initial naive OLS
• Create lagged residuals
• Regress current on lagged residuals
• Implications for ORDER= option on AUTO command
5. Use of DLAG option if first explanatory variable is lagged dependent variable (a dynamic model)
6. Review of the process of iterations to convergence on the rho parameter(s)

---

COURSE OUTLINE	LECTURE OUTLINES	PROBLEM SETS	PROBLEM SOLUTIONS	COMPUTER LABS
SHAZAM EXAMPLES	DATA SETS	ONLINE QUIZZES	GRAPHICS	HANDOUTS

Update date: March 10, 1998
Prepared by: Trudy Ann Cameron

---