代做125.785 Research and Analytics in Economics and Finance exam 2代做Processing

日期：2024-04-29 11:32

125.785 Research and Analytics in Economics and Finance

Answer all 8 questions, marks are shown with each question.

You are required to answer ALL questions.

Click here to access Formula Sheet

Click here to download Excel Spreadsheet

At the end of the exam click 'Finish attempt', then confi rm by clicking 'Submit all and Finish'.

Total: 100 marks

Question 1

Please concisely answer the following:

a. List and describe the benefi ts of larger sample size in a regression model estimation.  (3 marks)

b. Defi ne and describe residual sum of squares. Briefl y explain how it is related to the goodness of fi t statistics, R2.  (3 marks)

c. Defi ne the continuously compounded returns, which are regularly used in financial research. Please explain why they are particularly useful in studying time series properties of fi nancial returns.  (3 marks)

d. Explain why it is important to include control variables as explanatory variables in a regression.  (4 marks)

Question 2

(Please concisely answer the following questions)

A fellow researcher estimates a two-variable regression model (i.e., one single explanatory variable) by regressing fi rm value (dependent variable proxied by Tobin’s Q) on firm’s corporate governance quality (the only explanatory variable proxied by a corporate governance quality index) using Thai data. Unexpectedly, the regression coeffi cient is negative but highly signifi cant (with t-stat of -3.97). He/she claims this finding as interesting evidence on unusual perceptions among Thai investors (e.g. the more twisted the fi rm, the higher its value).

a. Generally, we believe that fi rm value is determined by a long list of factors. Explain what the major problem in this two-variable regression model is.  (4 marks)

b. If there is a such problem, is the signifi cant relation shown in the T-test valid?  (4 marks)

c. Is it possible that the two variables in the model have a signifi cantly negative relation? Why?  (5 marks)

Question 3

In theory we have a model which explains a company’s debt ratio:

DebtRatio= α+β1*NDT + β2*FixedAsset + β3*MBRatio

                            + β4*ROA + β5*SIZE + β6*RISK +ε

(NDT is the ratio of depreciation costs to total assets)

Now we have run the OLS regression and produced the result: β1 = 1.315, and its T_value is 3.693 and P_value is <0.0002 (The degree of freedom: T-k=153).

a. Describe in detail how to test the following hypotheses:

H0: β1≤ 1

H1: β1 ≠0

(In your answer, you should state the assumptions for the T-test, and provide the meaning of the P_value). (4 marks)

b. What are the T-test result and conclusion?  (4 marks)

c. If we are also interested to know the result of the second set of the following hypotheses: 

H0: β1≤1

H1: β1 >1

What is your conclusion of the hypothesis testing?  (4 marks)

Question 4

In the year 2002, a group of financial researchers aimed to search for the determinants of a fi rm’s extra liquidity (e.g., current ratio of a firm beyond its industry’s ratio) in an exotic market. Without referring to theoretical models, they came up with a list of 25 accounting variables and ran a Fama and MacBeth (1973) type regression using ten-year annual data from 1991 to 2000 (with the average number of companies in the sample per year of 45.). Regression estimated coeffi cients of none of the variables consistently show signifi cance, except for this one (continuous) variable called “political connection (POLTC)”. These authors are excited and developed a conjecture around this fi nding. Following are the regression coeffi cients and t statistics on the independent variable, POLTC from running a regression model on each year separately.

a. Consider the cross-sectional regression setting. Does political connection of a fi rm enhance its extra liquidity? (4 marks)

b. Consider the Fama and Macbeth procedure. Does political connection of a fi rm enhance its extra liquidity? (4 marks)

Fama and Macbeth method formulae:

c. Provide critiques on the fi ndings of these authors.  (4 marks)

Question 5

Qualitative Analysis:

The logit regression between dividend paying and the profi t margin is as follows.

Dividend paying = b0 + b1*profi t_margin + e

The “Dividend paying” is a dummy variable with the value one for the fi rm of making a payment at the end of the year and zero if not paying anything. “profi t_margin” is the fi rm’s profi t margin.

a. Compute the difference in the probability of dividend paying when the profi t margin increases from 0.20 to 0.30. Note that the cumulative probability of the logit function is 1/(1+exp(-Z)), where Z is the quantile value. Show the calculation in detail.  (6 marks)

b. When does the predicted probability of the dividend paying from the logit regression substantially differ from the predicted probability from the OLS linear probability regression?  (6 marks)

Question 6

Time-series analysis

a. Univariate time-series

Is the autoregressive model below stationary? Show the evidence to support your answer.  (6 marks)

Model: Y(t) = 2.2 – 0.71 Y(t-1) - 0.23 Y(t-2) + u(t)

b. Multivariate time-series

The theoretical model of the determinant of oil demand and supply in equilibrium is as follows.

Demand Model: Quantity(t) = d0 + d1*OilPrice(t) + d2*GDP(t) + u(t)

Supply Model: Quantity(t) = s0 + s1*OilPrice(t) + s2*Tech(t) + v(t)

Quantity is the quantity sale of oil.

OilPrice is the price of oil.

GDP is the gross domestic product.

Tech is oil extraction technology.

Assume GDP and Tech are exogenous variables.

Why are the OLS estimates of the demand and models above biased? What is the reduced form. model that we can use the OLS to estimate?  (6 marks)

Question 7

Endogeneity

a. Consider an event study of the impact of the stock split announcement on the stock return. Explain how to calculate the CAAR(-1,+1) which is the cumulative cross-sectional average abnormal return from one day before the announcement to one day after the announcement.  (6 marks)

b. Under the propensity score matching (PSM) model, discuss in detail how to show that the sample of the treatment group is comparable to the control group.  (7 marks)

Question 8

Panel regression and Forecasting

a. Panel regression:

Using the fi xed effect regression, discuss when the fi rm fixed effect is appropriate and when the industry fi xed effect is appropriate.  (6 marks)

b. Forecasting:

The overfi tting model has poor forecasting performance. What should be done to avoid overfi tting the model? Similarly, the underfi tting model also has poor forecasting performance. What should be done to avoid underfi tting the model?  (7 marks)