School of Business,Economics and
Management Studies.
ECF480- Assignment
Advanced Econometrics Assignment
August 2, 2024
Answer all the questions. Submit a clear hand written document .
1
Question 1: (25 marks)
A research team at a marketing firm is analyzing the impact of several factors
on whether customers purchase a product (Y ). They use a linear probability
model to estimate the probability of purchase based on a set of explana-
tory variables X, which includes factors such as age, income, and marketing
exposure. The model is specified as:
Y = Xβ + U
where Y is an n × 1 vector of binary outcomes, X is the n × k matrix of
explanatory variables, β is a k × 1 vector of coefficients, and U is an n × 1
vector of residuals.
(a) The research team uses Ordinary Least Squares (OLS) to estimate the
coefficients β. Derive the OLS estimator for β. Discuss the implications
of applying OLS to this binary dependent variable model, particularly
focusing on issues related to heteroscedasticity and non-normality of
residuals. Consider how these issues might impact the validity of the
estimates and any subsequent decisions based on these results. (10
marks)
(b) The team wants to ensure that their OLS estimator β̂ is reliable. Prove
that the OLS estimator β̂ is unbiased under the Gauss-Markov assump-
tions. Then, identify which of these assumptions are likely violated
when using a linear probability model for binary outcomes. Explain
the implications of these violations for the accuracy and reliability of
the OLS estimates in this context. (10 marks)
(c) Given the limitations of the linear probability model, suggest a more
appropriate estimation technique for modeling binary dependent vari-
ables. Explain why this alternative method is better suited for the
analysis and how it addresses the issues identified in the linear proba-
bility model. (5 marks)
2
Question 2: (25 marks)
A company is analyzing the relationship between employee productivity (yi )
and years of training (xi ) using a linear regression model. The model is
specified as:
yi = β1 + β2 xi + ui
where ui represents the error term with heteroscedastic variances σi2 that
are known.
Tasks:
(a) The company decides to use the Generalized Least Squares (GLS)
method to account for heteroscedasticity. Explain the GLS method,
derive the GLS estimator, and show how it adjusts for heteroscedastic-
ity in this context. (10 marks)
(b) Analyze how heteroscedasticity might affect the efficiency and unbi-
asedness of the Ordinary Least Squares (OLS) estimators in the given
model. (5 marks)
(c) The company is considering different tests for detecting heteroscedas-
ticity. Compare and contrast the White test, Goldfeld-Quandt test,
and Breusch-Pagan-Godfrey test. Additionally, explain the benefits of
using robust standard errors when heteroscedasticity is present. (10
marks)
3
Question 3: (25 marks)
A company is analyzing the relationship between the number of customer
complaints (x) and the amount of time (y) a customer spends on hold. They
have identified that the joint distribution of these two variables can be mod-
eled as:
θe−(β+θ)y (βy)x
f (x, y) =
x!
where β and θ are positive parameters, y ≥ 0, and x = 1, 2, . . . with x!
denoting 1 × 2 × 3 × · · · × x.
(a) The company collected a dataset consisting of observations on x and
y. Using this dataset, derive the maximum likelihood estimators for
β and θ. Discuss the properties of these estimators, including their
consistency, efficiency, and asymptotic normality. (15 marks)
θ
(b) Calculate the maximum likelihood estimator for the ratio β+θ . Provide
an interpretation of this estimator within the context of the model and
how it might be useful for the company’s analysis. (5 marks)
(c) Critically assess the limitations and potential biases of the maximum
likelihood estimation method used in this case. Consider the implica-
tions of these limitations for the company’s decision-making process.
(5 marks)
4
Question 4 (25 marks):
.
A company is conducting a study to understand the factors influencing em-
ployee wages. They have collected data on employee wages and their asso-
ciated characteristics, including schooling (s), tenure, and experience (expr).
The goal is to analyze how these factors affect the natural logarithm of the
wage (lw).
The company has performed an Ordinary Least Squares (OLS) regression
with the following model:
. reg lw s expr tenure
Source | SS df MS Number of obs = 758
-------------+------------------------------ F( 3, 754) = 117.64
Model | 44.4080258 3 14.8026753 Prob > F = 0.0000
Residual | 94.878124 754 .125833056 R-squared = 0.3188
-------------+------------------------------ Adj R-squared = 0.3161
Total | 139.28615 757 .183997556 Root MSE = .35473
------------------------------------------------------------------------------
lw | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
s | .1066109 .0059536 t1 0.000 .0949233 .1182984
expr | .0375699 .0064777 t2 0.000 .0248534 .0502864
tenure | .0381141 .0079174 t3 0.000 .0225714 .0536568
_cons | 4.122627 .0849018 t4 0.000 3.955955 4.289299
------------------------------------------------------------------------------
(a) Specify the estimated regression model. Discuss the interpretation of
each coefficient, including the intercept. (5 marks)
(b) Evaluate the overall significance of the model using the F-statistic.
What does the F-statistic indicate about the model? (5 marks)
(c) Calculate the t-statistics t1 , t2 , t3 , and t4 . Identify which coefficients are
statistically significant and discuss the economic implications of these
findings. (5 marks)
(d) Discuss the goodness-of-fit of the model as indicated by R2 and Ad-
justed R2 . Explain the limitations of using R2 as a measure of model
fit. (5 marks)
5
(e) Critically analyze the role of the Root Mean Squared Error (Root MSE)
in evaluating model performance. (5 marks)
