代做BUSI4528 QUANTITATIVE RESEARCH METHODS FOR FINANCE AND ACCOUNTING AUTUMN SEMESTER 2018-2019代写留学生Ma

BUSI4528-E1

A LEVEL 4 MODULE, AUTUMN SEMESTER 2018-2019

QUANTITATIVE RESEARCH METHODS FOR FINANCE AND ACCOUNTING

1.    a) A researcher is interested in the factors that affect women’s decision to participate in the labour force. He interviews a total of 753 women in 2000 and collects data on a number of variables as described below:

fempart        : Variable taking value 1 if a woman participated in the labour force, 0 otherwise

faminc          : Gross family income in the year 2000, measured in US $

educ            : years of schooling, in years

age              : age, in years

exper           : years of working experience, in years

urban           : Variable taking value 1 if the woman lives in the urban area, 0 otherwise

kidsunder6   : number of kids under 6 years old

(i)    Explain how the Logit model is set up and how the researcher should use it to calculate the probability of participating in the labour force for each individual in the data set.   [30 marks]

(ii)   The actual estimation of the Logit model reveals the following output. Calculate the probability of participating in the labour force for a woman who is 36 years old and lives in the urban area. She has completed 14 years of schooling and 5 years working experience. She currently has 2 kids under 6 years old. It is also known that her total family income is US$ 60,000 per annum. Show all your workings.  [10 marks]

Iteration 0:   log likelihood = -514 .8732

Iteration 1:   log likelihood =-411 .62085

Iteration 2:   log likelihood =-407 .17257

Iteration 3:   log likelihood = -407 .1063

Iteration 4:   log likelihood =-407 .10628

Logistic regression	Number of obs LR chi2(6) Prob > chi2	= = =	753 215 .53 0 .0000
Log likelihood = -407 .10628	Pseudo R2	=	0 .2093

fempart	Coef .	Std . Err .	z	P>|z|	[95% Conf . Interval]
faminc	.0000169	8 .06e-06	2 .10	0 .035	1 .16e-06 .0000327
educ	.1596136	.0429795	3 .71	0 .000	.0753753 .243852
age	- .1017045	.0134487	-7 .56	0 .000	- .1280634 - .0753456
exper	.126727	.0133507	9 .49	0 .000	.1005601 .152894
urban	- .1671848	.1881382	-0 .89	0 .374	- .5359288 .2015592
kidsunder6	-1 .421205	.1974545	-7 .20	0 .000	-1 .808209 -1 .034201
_cons	1 .431475	.7389594	1 .94	0 .053	- .0168589 2 .879809

(iii)  Using information contained in part (ii) calculate the effect on the probability of participating in the labour force if a woman lives in the rural area. Show all your workings.   [15 marks]

b) Explain two ways of evaluating the goodness of fit of the Logit regression model?   [20 marks]

c)  Discuss the consequences of Heteroskedasticity for the OLS estimator.   [25 marks]

Total [100 marks]

2.  a) i)  Discuss, using  basic mathematical notations, the assumptions for a simple  linear regression model : y = β1  + β2x   +  e.   [30 marks]

ii) Briefly outline the Gauss-Markov Theorem and the Central Limit Theorem.   [20 marks]

b) What are the likely consequences of using non-stationary time series data in a linear regression?   [15 marks]

c)  Consider the following AR(1) model with drift:

yt  = μ + Pyt−1  + vt , with |P| < 1

what is the de-meaned series of the AR(1) model? What is the mean of the de-meaned series?   [15 marks]

d) Explain the circumstances under which the Augmented Dickey- Fuller (ADF) test is used instead of the standard Dickey-Fuller (DF) test.   [20 marks]

Total [100 marks]

3.  a) What are the main limitations of the Linear Probability Model (LPM) as compared to the Logit model?   [20 marks]

b) Consider the following time series model with serially correlated error term :

yt  = α + βxt  + εt,  with εt   = P1 εt−1  + P2 εt−2  + vt   where var(vt ) = σv(2),  and  Cov (vt, vs ) = 0  for  t  ≠ S

show the above model can be rewritten into a new model that has an error term uncorrelated over time. Discuss on the merits of doing so.   [30 marks]

c) Explain the term “Heteroskedasticity” in cross-sectional multivariate regression.   [15 marks]

d) How might Heteroskedasticity be detected and addressed in regression analysis?   [35 marks]

Total [100 marks]

4.  a) Discuss how the fixed-effects model might be adopted to reduce endogeneity concerns resulting from the unobserved heterogeneity associated with firms and years in a panel dataset. Specify the regression model in your discussions.   [50 marks]

b) What is meant by saying that a time series is stationary?  Is the following time series stationary or non-stationary?

yt  = yt−1 + vt

where error terms vt  are i.i.d. with mean zero and variance σv(2) , and t denotes the t-th time period Justify your answer.   [25 marks]

c)  Explain the Engle-Granger test for cointegration and write down the steps involved in the test.   [25 marks]

Total [100 marks]

5.  a) Discuss how the difference-in-difference (DID) estimator (specify the DID regression model) might be used to test for a potential treatment effect of a policy reform. Outline the key assumptions for the DID estimation. Use graphs where necessary.   [75 marks]

b) Consider the following time series model :

yt  =  β0  + β1xt  + β2xt−1 + yyt−1 + εt

It is suspected that the model suffers from serial correlation in the error term of the form.

εt  = Pεt−1  + ut

where ut   is an identically independently distributed error term and t denotes the t-th time period. Describe in detail a test to detect serial correlation in the above form of the model. [25 marks]

Total [100 marks]