Practical Exercises – Two-variable Regression Model

1) The following Table gives data on indexes (1977=100) of real compensation per
hour (Y) and out put per hour of all persons(X) for the U.S business sector for the
period 1973-1987.
a) Theory postulates a positive relationship between earnings and productivity.
Plot a scattergram of Y against X and see if the postulated relationship holds.
b) If it does, do an OLS regression of Y on X.
c) How do you decide whether earnings is a function of productivity or productivity
is a function of earnings?

Year Y X Year Y X
1973 96.8 95.9 1981 95.8 100.7
1974 95.4 93.9 1982 97.3 100.3
1975 96.0 95.7 1983 98.2 103.0
1976 98.8 98.3 1984 97.9 105.5
1977 100.0 100.0 1985 98.8 107.7
1978 100.9 100.8 1986 101.2 110.1
1979 99.4 99.6 1987 101.5 111.0
1980 96.7 99.3
Y - Indexes of real compensation per hour
X- Output per hour
Note: Data file name is “Earning productivity relation”

2) The following Table provides details about per capita consumption expenditure
(PCE) and per capita disposable personal income (PDPI) in USA during 1980-1995
(in 1992 dollars)

Years PCE PDPI

1980 13216 14813
1981 13245 15009
1982 13270 14999
1983 13829 15277
1984 14415 16252
1985 14954 16597
1986 15409 16981
1987 15740 17106
1988 16211 17621
1989 16430 17801
1990 16532 17941
1991 16249 17756
1992 16520 18062
1993 16809 18078
1994 17159 18330
1995 17400 18799
Note: Data file name is “PCE”
Find out how PCE behaves in relation to PDPL using OLS regression results and
interpret the results.

3) The following Table gives data on X [net profits after tax in U.S manufacturing
industries ($, in millions)] and Y [cash dividend paid quarterly in manufacturing
industries ($,in millions)] for years 1974-1986
a) What relationship, if any, do you expect between cash dividend and after tax
profits?
b) Plot the scattergram between Y and X.
c) Does the scattergram support your expectation in part (a)?
d) If so, do an OLS regression of Y and X and interpret them.
e) Obtain a 99% confidence interval for the true slope and test the hypothesis that
the true slope coefficient is zero; that is no relationship between the dividend and
the after-tax profit.

Year Y($,in millions) X Year Y($,in millions) X

1974 19,467 58,747 1981 40,317 101,302
1975 19,968 49,135 1982 41,259 71,028
1976 22,763 64,519 1983 41,624 85,834
1977 26,585 70,366 1984 45,102 107,648
1978 28,932 81,148 1985 45,517 87,648
1979 32,491 98,698 1986 46,044 83,121
1980 36,495 92,579
Note: Data file name is “Dividend payment”

4) The following table gives data on average starting pay (ASP), GPA scores (on a
scale of 1-4), GMAT scores, and annual tuition fee for MBA graduates of 30 well-
regarded business schools in the United States for the year 1994.
a) Using a two-variable regression model find out if GPA has any effect on ASP.
b) Using a suitable regression model, find out if GMAT scores have any relationship
to ASP.
c) Does annual tuition have any relationship with ASP? How do you know? If there
is a positive relationship between the two, does that mean it pays to go to the
highest-cost business school?
d) Can you argue that a high-tuition business school means a high-quality MBA
program? Why or why not?
 Average starting salary of MBA graduates, 1994
School ASP($) GPA GMAT Tuition ($)
Harvard 102630 3.4 650 23894
Stanford 100800 3.3 665 21189
Columbia 100480 3.3 640 21400
Dartmouth 95410 3.4 660 21225
Wharton 89930 3.4 650 21050
Northwestern 84640 3.3 640 20634
Chicago 83210 3.3 650 21656
MIT 80500 3.5 650 21690
Virginia 74280 3.2 643 17839
UCLA 74010 3.5 640 14496
Berkeley 71970 3.2 647 14361
Cornell 71970 3.2 630 20400
NYU 70660 3.2 630 20276
Duke 70490 3.3 623 21910
Carnegie Mellon 69890 3.2 635 20600
North Carolina 69890 3.2 621 10132
Michigan 67820 3.2 630 20960
Texas 61890 3.3 625 8580
Indiana 58520 3.2 615 14036
Purdue 54720 3.2 581 9556
Case Western 57200 3.1 591 17600
Georgetown 69830 3.2 619 19584
Michigan State 41820 3.2 590 16057
Penn State 49120 3.2 580 11400
Southern Methodist 60910 3.1 600 18034
Tulane 44080 3.1 600 19550
Illinois 47130 3.2 616 12628
Iowa 41620 3.2 590 9361
Minnesota 48250 3.2 600 12618
Washington 44140 3.3 617 11436
Note: Data file name is “MBA Salary”

5) Suppose we have the following regression results for our consideration

Where Y = the coffee consumption in India (cups per person per day)
X= the retail price of coffee (in Rs.)
t = time period
(a) Is this a time series regression or a cross-sectional regression?
(b) What is the interpretation of the intercept in this example? Does it make economic
sense?
(c) How would you interpret the slope coefficient?