ICMA CENTRE

UNIVERSITY of READING

Quantitative Methods for Finance

Exercise 5: Assumptions of the CLRM and Structural Stability
1. What might we use Ramseys RESET test for? What could we do if we find that
we have failed the RESET test?
2.
(i) Why do we need to assume that the disturbances of a regression model
are normally distributed?
(ii) In a practical econometric modelling situation, how might we get around
the problem of residuals that are not normal?
3. A researcher is attempting to form an econometric model to explain daily
movements of stock returns. A colleague suggests that she might want to see
whether her data are influenced by daily seasonality.
(i) How might she go about doing this?
(ii) The researcher estimates a model with the dependent variable as the daily
returns on a given share traded on the London stock exchange, and various
macroeconomic variables and accounting rations as independent variables. She
attempts to estimate this model, together with five daily dummy variables (one for
each day of the week), and a constant term, using EViews. EViews then tells her
that it cannot estimate the parameters of the model. Explain what has probably
happened, and how she can fix it.
(iii) The final model for asset returns, rt is as follows (with standard errors in
parentheses):
rt = 0.0034 - 0.0183 D1t + 0.01554 D2t -0.0007 D3t - 0.0272 D4t+ other variables
(0.0146) (0.0068)
(0.0231)
(0.0179)
(0.0193)
The model is estimated using 500 observations. Is there significant evidence of any
day of the week effects? Assume that there are 3 other variables.

4.(a) What do you understand by the term parameter structural stability?

(b) A financial econometrician thinks that the stock market crash of October 1987
fundamentally changed the risk-return relationship given by the CAPM-type
equation. He decides to test this hypothesis using a Chow test. The model is
estimated using monthly data from January 1980 - December 1995, and then two
separate regressions are run for the sub-periods corresponding to data before and
after the crash. The model is
rt = + Rmt + ut
so that the return on security i at time t is regressed upon the return on the market at
time t. The results for the 3 models estimated for shares in British Telecom are as
follows
1981M1-1995M12
rt = 0.0215 + 1.491 Rmt
RSS=0.189 T=180
1981M1-1987M10
rt = 0.0163 + 1.308 Rmt
RSS=0.079 T=82
1987M11-1995M12
rt = 0.0360 + 1.613 Rmt
RSS=0.082 T=98
(i) What are the null and alternative hypotheses that are being tested here, in terms
of and ?
(ii) Perform the test. What is your conclusion?
5. Another way to test whether the regression parameters are structurally stable is to
perform a predictive failure test.
For the same model as above, and given the following results, do a forward and
backward predictive failure test.
1981M1-1995M12
rt = 0.0215 + 1.491 Rmt
RSS=0.189 T=180
1981M1-1994M12
rt = 0.0212 + 1.478 Rmt
RSS=0.148 T=168
1982M1-1995M12
rt = 0.0217 + 1.523 Rmt
RSS=0.182 T=168
What are your conclusions?