PROBLEM SESSION #4 Suggested Solutions in Red Application of Regression Analysis to Managerial Economics Q.1 You are the CEO of a regional telephone company . You picked up the October 6 th edition of your local newspaper and began reading on page D1: The Federal Government completed the biggest auction in history today, selling off part of the nation’s airwaves for $7 billion to a handful of giant companies that plan to blanket the nation with new wireless communications networks for telephone and computers….. You read the article with interest because your firm is scrambling to secure loans to purchase one of the licenses the FCC plans to auction off in his region next year. The region serviced by his firm has a population that is 7% greater than the average where the licenses have been sold before, yet the FCC plans to auction the same number of licenses. . This troubled you, since in the most recent auction 99 bidders coughed up a total of $7billion – an average of $70.7 million for a single license. Fortunately for you, the newspaper article contained a table summarizing the price paid per license in 10 different regions, as well as the number of licenses sold and the population of each region. You quickly entered this data into your spreadsheet, clicked the regression tool button, and found the following relation between price of a license, the quantity of licenses available and regional population size (price and population figures are in millions) ln P = 2.23 – 1.2 ln Q + 1.25 ln Pop Based on your analysis, how much money do you expect your company will need to buy a license? Since this is a log-log model, the coefficients are elasticities. The coefficient of ln Pop is 1.25 the percentage change in price resulting from a one percentage change in population. Since the population in the relevant region is 7% higher than the average, this means price will have to change by 1.25(7%) = 8.75%. The CEO should expect to pay a price which is 8.75% higher than the average price paid in the auction. Since the price paid was $70.7m, the expected price need to win the auction in his region = 70.7(1.085) = $76.9 million Q.2 You manage a brewing company that sells its microbrew in seven provinces. However, you are interested in the demand and pricing strategy for your brew in British Columbia. The company’s marketing department has collected data from its distributors in BC. The data consists of quantity of your microbrew (Qx), price (per case) of your microbrew (Px), the price of product Y (Py), the price of product Z (Pz), and income levels. Your boss wants you to develop a pricing strategy (reduce price, increase price or maintain price) because your competitors drop their prices which resulted in a loss of revenue for your firm. Using the data, you obtained the regression results in Exhibit A below a. Evaluate the overall fit of the estimated regression equation. (2 lines) Roughly 80% of the variation in our quantity sold can be explained by changes in Px, Py and Pz and Incomes. If we base our conclusions solely on this, then the model is a good fit b. c. d. e. Which of the explanatory variables have real effects on your demand? Why? (2 lines) Px and Pz since their t-ratios fall in the critical regions and lead to a rejection of the null of no significant relationship with Qx. Suppose you expect price of Y to fall by 10% and price of Z to fall by 10% as well, what is the predicted change in the quantity demand of your product X. (3 lines) [(-0.58*10) + (1.20*10)]% What pricing strategy would you propose (be specific) to counteract the actions taken by the producers of product Y and product Z in question (c) above so that the quantity demanded of your product X remains the same as before their actions? (3 lines) Change your price by -{[(-0.58*10) + (1.20*10)]%/2.02} Should you be concerned if macroeconomic forecasts predict a recession? Why or why not? Explain. (2 lines). No really, because incomes are insignificant (|t ratio| less than 1.96) so a recession will not have any significant effect plus incomes elasticity<1. SUMMARY OUTPUT Regression Statistics Multiple R 0.90850899 R Square 0.82538859 Adjusted R Square 0.79745077 Standard Error 0.05997329 Observations 30 ANOVA df Regression Residual Total 4 25 29 Coefficients Intercept lnPx lnPy lnPz lnincome -3.24323769 -2.02041877 -0.58293365 1.20954497 0.32286375 SS 0.42505154 0.08991989 0.51497143 MS 0.106263 0.003597 F 29.54377 Standard Error 3.74299952 0.43904185 0.56015006 0.17969263 0.41551418 t Stat P-value -0.86648 -4.60188 -1.04067 6.731188 0.761016 0.394467 0.000248 0.307987 0.014422 0.035639 Q.3 You gave your intern data on quantity demanded (Qx), price of your product (Px), price of a substitute (Py), price of a complement (Pz), average Income levels (Income) and your advertising budget (Advertising) to estimate a demand function for your company and present the results to your regional manager. A week after her presentation, you received the following note from your boss. “One of your MBA interns sent this spreadsheet in response to the question I asked regarding the potential impact (on our sales) of the following simultaneous events predicted for next year: 7% drop in our competitor’s prices, 12% increase in our collaborator’s prices, 9% decline in incomes and a 5% cut in our advertising budget. This is absolutely nuts – do they not teach people how to speak in words anymore? Can you help me out?” All I need to know are a. Summarize her findings and tell me what these numbers mean. The spreadsheet is absolute gibberish to me. EXPLAIN THE COEFFICIENTS. CHECK THE SIGNS, EXPLAIN R-SQUARED. EXPLAIN SIGNIFICANCE OF EACH VARIABLE TO VARIATION IN OUR QTY. b. In very simple words and numbers, tell me what will be the total effect of the events I described above on our quantity sold next year. (-7%*1.583)+(12%*-1.212)+(-9%*0.332)+(-5%*1.45) And next time, please do your job rather than send an intern to me. Thank you SUMMARY OUTPUT Regression Statistics R Square 0.825 Adjusted R Square 0.797 Observations 300 Intercept ln Px ln Py ln Pz ln Income ln Advertising Coefficients -3.24 -2.020 1.583 -1.212 0.332 1.452 Standard Error 3.74 0.44 1.56 0.250 0.042 0.867 t Stat -0.87 -4.60 1.012 -4.848 7.904 1.67 P-value 0.394 0.000 0.330 0.000 0.000 0.095