Economics 113 Problem Set No.

2: Solutions

Problem 2.3 from Chapter 2 Wooldridges Book

The ACT and GPA (grade point average) is given for 8 students: Student 1 2 3 4 5 6 7 8 GPA 2.8 3.4 3.0 3.5 3.6 3.0 2.7 3.7 ACT 21 24 26 27 29 25 25 30

1. the relationship between GPA and ACT using OLS GP A = 0 + 1 ACT. Comment on the direction of the relationship. Does the interpret have a useful interpretation here? Explain. How much higher is the GPA predicted to be if the ACT score is increased by 5 point? The direction of the relationship implies that the causality is such that variations in the ACT score can be used to explain variations in the GPA. A non-zero intercept implies that the student can score a zero on the ACT but still have some non-zero GPA (it is assumed that neither scores can go negative). If the intercept term is statistically signicant, it may provide some information with regards to grade ination at dierent institutions (depending on the sample) since the ACT is a standardized test, while GPA is specic to the institution.

To solve for the estimates, we will use equations (2.17) and (2.49) from the text. According to equation (2.49), 1 = =
n i=1 (xi x)yi n (xi x)2 i=1

8 i=1 (xi 25.875)yi 8 2 i=1 (xi 25.875)

(21 25.875)(2.8) + + (30 25.875)(3.7) (21 25.875)2 + + (30 25.875)2 5.8125 = 56.87500 = 0.102197802.

Substituting this value into equation (2.17), we get 0 = y 1 x = 3.2125 (0.102197802)(25.875) = 0.568131873. This implies that our estimation becomes GP A = 0.568131873 + 0.102197802ACT. So, if ACT increases by 5 points, GPA is predicted to increase by 0.102197802(5) = 0.51098901 points. 2. Compute the tted values and residuals for each observation, and verify that the residuals (approximately) sum to zero. We want to ensure that
8

u 0.
i=1

Proceeding,
8 8

u=
i=1 i=1

(yi yi )

= (2.8 2.71428571) + (3.4 3.02087912) + (3.0 3.22527473) + (3.5 3.32747253)+ + (3.6 3.53186813) + (3.0 3.12307692) + (2.7 3.12307692) + (3.7 3.63406593) = 9.99999994 109 0. 3. What is the predicted value of GPA when ACT = 20? We can nd this by simply substituting in the value for ACT: 0.102197802(20) + 0.568131873 = 2.61208791. 2

4. How much of the variation in GPA for these eight students is explained by ACT? Explain. Well want to calculate the R2 statistic for this model. We know that R2 = 1 We know
n

SSR . SST

SSR =
i=1

u2 i

and
n

SST =
i=1

(yi y)2 .

Hence, SSR = (2.8 2.71428571)2 + (3.4 3.02087912)2 + (3.0 3.22527473)2 + (3.5 3.32747253)2 + + (3.6 3.53186813)2 + (3.0 3.12307692)2 + (2.7 3.12307692)2 + (3.7 3.63406593)2 = 0.434725275 and SST = = (2.8 3.2125)2 + (3.4 3.2125)2 + (3.0 3.2125)2 + (3.5 3.2125)2 + + (3.6 3.2125)2 + (3.0 3.2125)2 + (2.7 3.2125)2 + (3.7 3.2125)2 = 1.02875. Therefore, R2 = 1 0.434725275 1.02875 = 0.577423791,

which implies that approximately 58% of the variation in GPA is explained by the variation in ACT.

Problem 2.4 from Chapter 2 Wooldridges Book

Given the following regression where bwght is the infant birth weight in ounces and cigs is the average number of cigarettes the mother smoked per day during pregnancy bwght = 119.77 0.514cigs 3

1. what is the predicted birth weight when cigs = 0 What about when cigs = 20 (one pack per day)? Comment on the dierence. The predicted birth weight when cigs = 0 is the intercept, 119.77. When cigs = 20, the predicted birth weight is 119.77 0.514(20) = 109.49. 2. Does this simple regression necessarily capture a casual relationship between the childs birth weight and mothers smoking habits? Explain Not necessarily. The variance in birthweight of children is notoriously large, so there is the distinct possibility that there is more to this picture than this simple regression. For example, there is a causal relationship between malnurishment and birthweight. There is also a high level of correlation between malnurishment and poverty. A great proportion of smokers can be categorized as living at poverty levels, so it perhaps may be a coincidence that an increase in cigarette consumption is related to a decrease in birthweight. We cant be 100% sure without controlling for other factors, such as income/caloric intake. 3. To predict a birth weight of 124 ounce, what would cigs have to be? Coment. According to this regression, cigs would have to be negative, which is absurd. 4. The proportion of women in the sample who do not smoke while pregnant is about .85. Does this hep reconcile your nding from part (iii)? This proportion implies that only 15% of the sample is causing uctuations in the independent variable. Given the naturally high variance in the dependent variable, and more weight in the sample placed towards children birthed from non-smoking mothers, this helps reconcile the nding from part (iii).

Problem 2.5 from Chapter 2 Wooldridges Book

In the linear consumption function cons = 0 + 1 inc the estimated marginal propensity to consume (MPC) out of income is simply the slope, while the average propensity to consume is cons/inc = 0 /inc + 1 . Using observations for 100 families on annual income and consumption (both measured in dollars), the following equation is obtained: cons = 124.84 + 0.853inc where n = 100, R2 = 0.692 1. Interpret the intercept in this equation, and comment on its sign and magnitude. 4

The intercept in this equation indicates that at an income of 0, a family is estimated to have negative consumption which is absurd. Economic intuition should tell us that the intercept should be positive at the minimum subsistence level of consumption for survival, autonomous consumption. The large negative value indicates the model performs poorly in estimating the consumption (or mean consumption) of poor people. 2. What is the predicted consumption when family income is $30000? The predicted consumption when family income is $30000 is 124.84 + 0.853(30000) = 25465.16 3. with inc on the x-axis, draw a graph of the estimated MPC and APC. Graph is below.
MPC and APC versus Income 10 0 -10 APC, MPC -20 -30 -40 -50 -60 0 20 40 60 80 Income 100 120 140 APC MPC

Figure 1: Graph for problem 3, part 3. Note: max income on graph is 150. As income increases, the lines become unintelligible.

Problem C2.4 from Chapter 2 Wooldridges Book

USe the data WAGE.RAW to estimate a simple regression explaining monthly salary (wage) in terms of IQ score (IQ). 5

1. Find the average salary and average IQ in the sample. What is the sample standard deviation of IQ? (IQ scores are standardized so that the average in the population is 100 with a standard deviation equal to 15) We can use the following code in R. Results included (truncated): > w = read.table("http://people.ucsc.edu/~archeng/teach/econ113/WAGE2.txt", header = T) > summary(w) wage hours IQ Min. : 115 Min. :20.00 Min. : 50.0 1st Qu.: 669 1st Qu.:40.00 1st Qu.: 92.0 Median : 905 Median :40.00 Median :102.0 Mean : 958 Mean :43.93 Mean :101.3 3rd Qu.:1160 3rd Qu.:48.00 3rd Qu.:112.0 Max. :3078 Max. :80.00 Max. :145.0 > sd(w$IQ) [1] 15.05264 2. Estimate a simple regression model where a one-point increase in IQ changes wage by a constant dollar amount. Use this model to nd the predicted increase in wage for an increase in IQ of 15 points. Does IQ explain most of the variation in wage? To answer this question, we will use the following code (results included): > results = lm(wage ~ IQ, w) > summary(results) Call: lm(formula = wage ~ IQ, data = w) Residuals: Min 1Q Median -898.7 -256.5 -47.3

3Q Max 201.1 2072.6

Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 116.9916 85.6415 1.366 0.172 IQ 8.3031 0.8364 9.927 <2e-16 *** --Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1 Residual standard error: 384.8 on 933 degrees of freedom Multiple R-squared: 0.09554, Adjusted R-squared: 0.09457 F-statistic: 98.55 on 1 and 933 DF, p-value: < 2.2e-16

An increase in IQ of 15 points is predicted to increase wage by 8.3031 20 = 124.5465 dollars. Since the R2 value is approximately 0.10, most of the variation in wage is not explained. 3. Now, estimate a model where each one-point increase in IQ has the same precent-age eect on wage. If IQ increases by 15 points, what is the approximate percentage increase in predicted wage? To answer this question, we will use the following code (results included): > results = lm(lwage ~ IQ, w) > summary(results) Call: lm(formula = lwage ~ IQ, data = w) Residuals: Min 1Q -2.09324 -0.25547 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 5.8869942 0.0890206 66.13 <2e-16 *** IQ 0.0088072 0.0008694 10.13 <2e-16 *** --Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1 Residual standard error: 0.3999 on 933 degrees of freedom Multiple R-squared: 0.09909, Adjusted R-squared: 0.09813 F-statistic: 102.6 on 1 and 933 DF, p-value: < 2.2e-16 If IQ increases by 15 points, the percentage change in wage will approximately be 0.0088072(15) 100 = 13.2108%.

Median 0.02261

3Q 0.27544

Max 1.21486

Problem C2.5 from Chapter 2 Wooldridges Book

For the population of rms in the chemical industry, let rd denote annual expenditures on research and development, and let sales denote annual sales (both are in millions of dollars) 1. Write down a model (not an estimated equation) that implies a constant elasticity between rd and sales. Which parameter is the elasticity? A constant elasticity model for rd and sales is log(rd) = 0 + 1 log(sales) + u. 7

2. Now, estimate the model using the data in RDCHEM.RAW.Write out the estimated equation in the usual form. What is the estimated elasticity of rd with respect to sales? Explain in words what this elasticity means. To answer this question, we will use the following code (results included): > r = read.table("http://people.ucsc.edu/~archeng/teach/econ113/RDCHEM.txt", header = T) > results = lm(lrd ~ lsales, r) > summary(results) Call: lm(formula = lrd ~ lsales, data = r) Residuals: Min 1Q Median -0.90406 -0.40086 -0.02178

3Q 0.40562

Max 1.10439

Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) -4.10472 0.45277 -9.066 4.27e-10 *** lsales 1.07573 0.06183 17.399 < 2e-16 *** --Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1 Residual standard error: 0.5294 on 30 degrees of freedom Multiple R-squared: 0.9098, Adjusted R-squared: 0.9068 F-statistic: 302.7 on 1 and 30 DF, p-value: < 2.2e-16 The estimated equation in usual form is log(rd) = 4.10472 + 1.07573 log(sales). The estimated elasticity of rd with respect to sales is the coecient on log(sales), 1.07573, which implies that a 1% increase in rm sales increases rd by approximately 1.08%.