Suppose you Department of Motor Vehicles (DMV) locations. The program is instituted at a randomly chosen 250 of the 500 DMV offices in the state at the beginning of the year. You have data for all visits to all the offices for the whole year. This data includes these variable: are asked to evaluate a new staff training program to reduce wait times at the y: Appointment duration in minutes (including wait time) Ti: 1 if the office was assigned to the new program (treatment), 0 otherwise age, Individual's age urban,: 1 if the office is in an urban area, 0 otherwise a) You start by estimating this equation Assuming the program had no effect on the duration of the non-waiting portions of the appoint- ment, is an unbiased estimate of the effect of the program on wait times? Why or why not? b) Which parameter does identify if DMV offices in the Treatment group can choose whether to implement the new program the case in which there is full compliance with the Treatment, so that all DMV offices assigned to treatment actually implement it? rain, and only 70% of them decided to implement it? What about in c) Suppose you believe the treatment effect is different in urban and n-urban locations. You estimate the following model What is the difference in expected wait time between a treated and un-treated location? d) Suppose you estimate the above model and get the following estimates What is the effect of the program in (i) an urban location? (ii) a non-urban location? e) Suppose that instead of rolling out this program randomly, the central DMV management had asked local offices to come forward and apply for this program, and the program had been rolleod out on a first come, first serve basis. Would ß1 in the above regression still identify the causal effect of the program on wait times? Why or why not?