Problem 2. (8 marks) For a problem, 50 observations for (X,Y) are available where X is equidistant on the interval (10,80) and the response Y is randomly generated using an error term and the linear component as E[YX]=20+10X. I decided to compare the Ordinary Least Squares estimate to the Weighted Least Squares (WLS) estimate because I had some suspicion that the constant variance assumption may not hold. I estimate ^0 and ^1 using both OLS and WLS estimation methods. Below picture shows the histograms I obtain for each of the 4 estimates when the above data generation procedure is repeated 200 times, each time with errors (and, as a result, response) is being randomly generated. Which estimators among the OLS and WLS will you prefer for (a) 0, and (b) 1 ? Explain your conclusions