You want to study the effect of education on earnings (yit = earnings). You have panel data on annual earnings for many workers. You think earnings will depend on age, education, and union status and you run that regression. Then the thought occurs to you: many workers receive percentage increases in wages each year. For example, all assembly line workers at a factory might receive a 2% baseline salary adjustment one year to the next. So, this year’s wage will depend on last year’s wage. Can you add last year’s wage (that is, yit-1) as a regressor and still satisfy Fixed Effects Assumption #1 for panel data regression (see section 10.5 in Stock & Watson)?