3. (Multivariate Regression) In order to study the effect of change in hourly wage (WPH) on the supply of labor (hours worked, HRS), we have data on the 38 demographic groups. There are other factors that will affect the supply of labor, which include spouse's annual income (ERSP) and the number of years of education (SCL). We have data saved in file PS2Wage.txt. You are asked to estimate the supply equation. (a) Assume the labor supply depends on both the wage paid and how much the spouse makes as in the following multivariate regression model, HRS = B1 + 32WPH + BERSP + u. What sign do you expect for 32 and 63? Are the corresponding coefficient estimates significant and have the right sign? What does the intercept term 31 mean, and can you make sense of the estimate? (b) What if you delete the 19th observation, and rerun the multivariate regression, what do you observe when comparing with those of (a)? (c) If a spouse's income provides a substitution, we can study whether there is a complete substi tution effect of spouse by testing the hypothesis of 32 * WPH = -B3 * ERSP, where WPH and ERSP are the averages of WPH and ERSP, respectively. Based on estimates in (a), is the complete substitution hypothesis true? (d) Finally, since both HRS and WPH are endogenous, we can study what determines labor income instead to get around with the endogenous issue. In particular, we estimate the following multivariate regression model, (WPH * HRS) = 1 + 72SCL + 13NEIN + VERSP + v, where NEIN is non-wage income. Interpret your coefficient estimates, and test their signifi- cance. Does the sign of coefficient estimates make sense