关键词 > ECMT6002/6702
ECMT 6002/6702: Econometric Applications
发布时间:2023-06-07
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
ECMT 6002/6702: Econometric Applications
1 Practice problems
1. Consider the following stationary AR(1) prediction model:
xt = ϕxt − 1 + ut .
The Yule-Walder equation of the above model is given by
V1 = E [xtxt − 1] = ϕE[xt(2)− 1] = ϕV1 (1.1)
Throughout this problem, the sample autocovariances k is given by
k =
工 xtxt −k .
t=k+1 (1.2)
T
and suppose that 1 < 0.
(a) Obtain the OLS estimator and the Yule-Walker estimator of ϕ . Which one is bigger? (b) Suppose that the estimated model is
xt = −0.7xt − 1 + t .
Obtain forecast of xT1+k as a function of xT1 and k .
(c) Suppose that xT1 − 1 = 1, xT1 = −0.9, xT1+1 = 0.52, xT1+2 = −0.58, xT1+3 = 0.45. Compute the MSE of the forecasts of xT1+1 , . . . , xT1+3
(d) Suppose that we considered AR(2) model instead and the estimation results are given as follows
xt = −0.8xt − 1 − 0.2xt −2 + t .
Compute forecasts of xT1+1 ,xT1+2 and xT1+3, and compute their MSE. Is the AR(2) model is preferred in terms of forecasting accuracy?
2 Empirical application
We will consider the following regression models:
yt = β 1 + β2 x2t + β3 x3t + β4 x4t + ut ,
where x2t is expected to be an endogenous regressor. The following two may be considered as an IV:
(i) Mother’s years of education (x5t)
(ii) Father’s years of education (x6t)
Instructions:
1. Compute the IV estimate of β2 using father’s education as an IV. Compare this result with the previous result obtained by using mother’s education as an IV (Week 6 Tutorial).
- result = ivreg(wage∼educ+exp+exp2 |feduc+exp+exp2); report=summary(result), where edu : education, feduc = father’s education
- Note : In Week 6 Tutorial, we already obtain the results when mother’s education is used as an IV. With only a slight modification, the results can be easily obtained.
- The result must be similar to
2,IV = 0.397. (2.1)
2. Compute the standard error of 2 .
- The result must be similar to
E (
2,IV) = 0.16. (2.2)
3. Compute the 2SLS estimator using all the suggested IVs, and then compute its standard error.
- result = ivreg(wage∼educ+exp+exp2 |meduc+feduc+exp+exp2); report=summary(result) . where meduc = mother’s education
- The result must be similar to
2,2SLS = 0.328 and
E (
2,2SLS) = 0.146. (2.3)
4. Check how the results change if “wage” is replaced by “log-wage”.
5. I recommend you to directly compute the IV/2SLS estimator and the standard error by constructing data matrix y , X and Z as in the lecture.
6. This computing exercise is not mandatory.