If we are going to use data from this year to predict unemployment next year, why not use this year's unemployment to predict next year's unemployment? A model like this, in which previous values of a variable are used to predict future values of the same variable, is called an autoregressive model. The following table presents the data needed to fit this model. 10. Compute the least-squares line for predicting next year's unemployment from this year's unemployment. 11. Predict next year's unemployment if this year's unemployment is 4.0%. 12. Compute the correlation coefficient between this year's unemployment and next year's unemployment. 13. What proportion of the variance in next year's unemployment is explained by this year's unemployment? 14. Which of the three models do you think provides the best prediction of unemployment, the one using inflation in the same year, the one using inflation in the previous year, or the one using unemployment in the previous year? Explain. 15. The U.S. economy was in recession during part of 1991, most of 2001, and from December 2007 through June 2009. Compute the residuals for the least-squares line found in Exercise 10. During times of recession, does the least-squares line predict the following year's unemployment well? Does it tend to overpredict or underpredict?

Fig: 1