Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit

ECON60622

Further Econometrics

Mid-term Exam

March 10 2020

1. (a)  Let z and u be two random variables and E[u|z] = 0.  Setting z = (1,z), show that E[zu] = 0.                                                                                                        [4 marks]

1. (b)  Consider the simple linear regression model

y  =  β0  + β1x + u,

where E[u|z] = 0, Var[u|z] = σ 2 , z is an observable random variable.  Let βˆ be the Instrumental Variables (IV) estimator of β = (β0 , β1 )/ based on the population moment condition E[zu] = 0 and a random sample {xi ,zi ,yi ; i = 1, 2, . . . ,n}. It can be shown that the formula for the IV estimator of β1  can be written as follows:

βˆ1   =  β1  +  ni=1 (zi  )(ui  )

(1)

where ,  and  are the sample means on u, x and z respectively. Assuming E[zu] = 0 is true:

(i) Explain how the argument for the consistency of βˆ1  depends on the population correlation between x and z .                                                                  [4 marks]

(ii) Are any other features of the sampling distribution of βˆ1  affected by ρx,z ? Briefly explain.                                                                                                    [4 marks]

2.  Consider the following simple linear regression model

y  =  β0  + xβ1  + u,                                               (2)

Assume that equation (2) holds in the population and E[u|x] = 0.

(a) Using this model, briefly explain the difference between endogenous sample selec- tion and exogenous sample selection.                                                        [4 marks]

(b) Do endogenous sample selection and exogenous sample selection have the same consequences for the OLS estimator of the parameters of the population model in equation (2)? Justify your answer briefly but no derivations are required.

[8 marks]

3. (a)  Explain the difference between repeated cross-section data and panel data.  [2 marks]

3. (b) Explain what is meant by the term unobserved heterogeneity.                      [2 marks]

4. Kiel & McClain (1995) investigate whether the location of a new garbage incinerator had an effect on housing prices in North Andover, MA in the USA. The rumour that the incinerator would be built began spreading in 1979,  and construction began in

1981.  Using data on house prices for the years 1978 and 1981, the following model is estimated:

log(price)  =  β0  + 60 y81 + β1 ∗ nearinc + 61 ∗ y81 ∗ nearinc + error,      (3)

where price is the house, y81 is a dummy variable that takes the value one if the observation is for year 1981 and is zero otherwise, and nearinc is a dummy variable that takes the value one if the house is located within three miles of the incinerator site. The output from this estimation is presented on the next page.

(a) What is the estimated impact of the construction of the incinerator on the prices of houses close to the incinerator site (that is, houses within three miles of the site)?                                                                                                        [4 marks]

(b) Test whether the construction of the incinerator has led to a reduction in the prices of the houses located close to the incinerator site? Be sure to specify clearly your null and alternative hypotheses, and your decision rule.                       [8 marks]

4. The Stata output for the estimation of (3) is as follows, with some parts of the output deliberately omitted.

Source  |              SS                      df              MS

-------------+----------------------------------

Model  |    11 .8433224                  3    3 .94777412

Residual  |    36 .3058203              317    . 114529402

-------------+----------------------------------

Total  |    48 . 1491427              320    . 150466071

Number of  obs F(3 ,  317)        Prob  >  F

R-squared        Adj R-squared Root MSE

=

=

=

=

=

=

321 34 .47 0 .0000 0 .2460 0 .2388 .33842


 

lrprice

|

+

|

Coef .

Std . Err .            t        P>|t |          [95%  Conf .  Interval]

y81

. 193094

.0453208

nearinc

|

- .3399216

.0545555

y81nrinc

|

- .0626498

.0834408

_cons

|

11 .28542

.0305145

 

 

where lrprice and y81nrinc denote respectively log(price) and y81 ∗ nearinc.