University of Sydney School of Economics ECMT 1020

Tutorial #5 – Week 6 Kadir Atalay

Q1
Consider the distribution given in the table below. For example, its first entry means that
Pr[X = 1| Y = 0] = 0.15.
X\Y 0 1
-1 0.15 0.10
0 0.20 0.30
1 0.15 0.10

(a) Compute the marginal distribution of Y ; that is, the probabilities Pr [Y = 0] and
Pr [Y = 1]. Also compute E [Y ] and Var [Y ].
(b) Compute the conditional distribution of Y given X = 1; that is, the probabilities
Pr [Y = 0|X = 1] and Pr [Y = 1 |X = 1]. Also compute E [Y|X = 1] and Var [Y |X =1].
(c) Compute the covariance and correlation between X and Y.
(d) Are X and Y independent?
Q2
The following Table gives the (joint) probabilities for a 5000 student passing the course (P)
or failing (F) and for the same student doing well on the mid-term (H) or not so well (L): so
that the probability of doing well on the mid-term and failing the course H and F, for
example, is 0.1.

P F
H 0.60 0.10

L 0.15 α

a) Find α.
b) Find the probability of passing the course. Complete the marginal distributions.
c) Find the (conditional) probability that a student will fail given that she did well in the
mid-term.
d) Show that how well a student does on the course is not independent of how well she
does in the mid-term. One explanation of this dependence is that the final mark depends in
part on the marks in the mid-term. Suggest another possible reason
Q3
The output below shows the result of regressing the weight of the respondent in 1985,
measured in pounds, on his or her height, measured in inches, using US data. Provide an
interpretation of the coefficients.

. reg WEIGHT85 HEIGHT

Source | SS df MS Number of obs = 540

-------------+------------------------------ F( 1, 538) = 355.97
Model | 261111.383 1 261111.383 Prob > F = 0.0000
Residual | 394632.365 538 733.517407 R-squared = 0.3982
-------------+------------------------------ Adj R-squared = 0.3971
Total | 655743.748 539 1216.59322 Root MSE = 27.084

------------------------------------------------------------------------------
WEIGHT85 | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
HEIGHT | 5.192973 .275238 18.87 0.000 4.6523 5.733646
_cons | -194.6815 18.6629 -10.43 0.000 -231.3426 -158.0204
------------------------------------------------------------------------------

Q4
The output shows the result of regressing the weight of the respondent, in pounds, in 2002 on
the weight in 1985. Provide an interpretation of the coefficients. Summary statistics for the
data are also provided.

. reg WEIGHT02 WEIGHT85

Source | SS df MS Number of obs = 540

-------------+------------------------------ F( 1, 538) = 1149.83
Model | 620662.43 1 620662.43 Prob > F = 0.0000
Residual | 290406.035 538 539.788169 R-squared = 0.6812
-------------+------------------------------ Adj R-squared = 0.6807
Total | 911068.465 539 1690.294 Root MSE = 23.233

------------------------------------------------------------------------------
WEIGHT02 | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
WEIGHT85 | 1.013353 .0298844 33.91 0.000 .9546483 1.072057
_cons | 23.61869 4.760179 4.96 0.000 14.26788 32.96951
------------------------------------------------------------------------------

. sum WEIGHT85 WEIGHT02

Variable | Obs Mean Std. Dev. Min Max

-------------+--------------------------------------------------------
WEIGHT85 | 540 155.7333 33.48673 89 300
WEIGHT02 | 540 181.4315 41.11319 103 400
Q5
The output shows the result of regressing the hourly earnings of the respondent, in dollars, in
2002 on height in 1985, measured in inches. Provide an interpretation of the coefficients,
comment on the plausibility of the interpretation, and attempt to give an explanation.
. reg EARNINGS HEIGHT

Source | SS df MS Number of obs = 540

-------------+------------------------------ F( 1, 538) = 29.39
Model | 6242.37244 1 6242.37244 Prob > F = 0.0000
Residual | 114276.055 538 212.409025 R-squared = 0.0518
-------------+------------------------------ Adj R-squared = 0.0500
Total | 120518.428 539 223.596341 Root MSE = 14.574

------------------------------------------------------------------------------
EARNINGS | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
HEIGHT | .8583752 .1583393 5.42 0.000 .5473361 1.169414
_cons | -38.05756 10.70014 -3.56 0.000 -59.07674 -17.03837
------------------------------------------------------------------------------ 

Q6
We are interested in estimating the model so we got a data set, which
has the following characteristics. There are n = 6 observations, with means ̅ 10
7 variances 100 , 25 40

(a) Compute the least squares estimates

(b) Compute the sample correlation and the of the regression.

(c) Compute the sums of squares TSS, ExpSS, and RSS.

(d) Compute
(e) Compute a 95% confidence interval for . Use the fact that , . 2.776 .

Q7
For all fifty American states, we have data on the labour supply (the percentage of people
who are either employed or actively looking for employment) and the average income
taxation rate. Assume that the sample summary statistics are as follows. Labour supply has a
mean of 65 and a standard deviation of 20, while the tax rate has a mean of 25 and a standard
deviation of 10. The correlation coefficient between the two variables is –0.4.
(a) Compute the coefficients in the regression line
(b) Compute the coefficients in the regression line
(c) Are both estimated regression lines equal? Why?
(d) Which of the two regressions makes the most sense?
Q8
We are interested in understanding the effect of the urban population on economic growth In
the below figure , we plot the % urban population and GNP of 185countries.

100
80 60
urban2005
40 20
0

0 20000 40000 60000 80000 100000

gnp

a) Consider the regression of % urban population on per capita GNP. What functional
form you think is suitable for this regression? Explain.
b) We have run following regression
Urban ln
where Urban variable measures % urban population. Interpret the coefficient of ).

Source | SS df MS Number of obs = 185

-------------+------------------------------ F( 1, 183) = 259.95
Model | 59707.8535 1 59707.8535 Prob > F = 0.0000
Residual | 42033.5519 183 229.69154 R-squared = 0.5869
-------------+------------------------------ Adj R-squared = 0.5846
Total | 101741.405 184 552.942421 Root MSE = 15.156

------------------------------------------------------------------------------
urban2005 | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
lngnp | 14.47963 .8980784 16.12 0.000 12.70771 16.25155
_cons | -73.23552 7.996436 -9.16 0.000 -89.01258 -57.45846
------------------------------------------------------------------------------

c) If we reverse X and Y from the above example, so that we regress the log of GNP/capita
on the %urban. In this case, the logarithmically transformed variable is the Y variable.
Interpret the coefficients.
Source | SS df MS Number of obs = 185
-------------+------------------------------ F( 1, 183) = 259.95
Model | 167.12837 1 167.12837 Prob > F = 0.0000
Residual | 117.656198 183 .642930041 R-squared = 0.5869
-------------+------------------------------ Adj R-squared = 0.5846
Total | 284.784568 184 1.54774221 Root MSE = .80183

------------------------------------------------------------------------------
lngnp | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
urban2005 | .04053 .0025138 16.12 0.000 .0355702 .0454897
_cons | 6.610927 .1489918 44.37 0.000 6.316965 6.90489
------------------------------------------------------------------------------