Assessment Instructions MODELLING AND FORECASTING CONSUMERS’ EXPENDITURE IN THE UK This assignment gives you a chance

Assessment Instructions
 
MODELLING AND FORECASTING CONSUMERS’ EXPENDITURE IN THE UK
 
This assignment gives you a chance to apply the econometric theory you have learned in Introductory Econometrics to the challenging task of modelling and forecasting consumers’ expenditure.  Please note the marks allocated to each section.
 
Section 1 (10 marks)
Using Stata and the data file Resit Cons 1976-2019.xlsx, estimate the following initial model for the subperiod 1976–2008:
                                         Ct = α β1Yt  β2Ht  β3Ct-1  εt                                                      (1)
 
where  Ct is households’ real final consumption expenditure (£ million) in year t
Yt is households’ real disposable income (£ million) in year t
            Ht is the Nationwide real house price index in quarter 2 of year t (£ per house, 2019Q2 prices)
εt is a random error term with an expected value of zero in year t
Using your log, paste in your initial results from Stata, along with a correlation matrix and summary table with means, standard deviations, etc. 
Check that you are using only the required variables and that your output relates to the specified time period.
Section 2 (35 marks)
Offer a brief economic rationale for the inclusion of the explanatory variables in model (1).  Comment on your findings, referring particularly to the signs, values and statistical significance of the estimated coefficients.  Be sure to explain how to interpret the values of the regression coefficients, specifying the relevant units.  How satisfactory is your initial model?  You should play close attention to any problems of serial correlation, multicollinearity and heteroscedasticity, as well as to possible specification errors.  In considering multicollinearity, you should include VIFs from a Stata command as well as calculating one VIF derived from an appropriate regression.
Check that you have addressed all parts of this question.
Section 3 (40 marks)
Now consider how you might refine your initial model.  You should present the results for four models in a summary table (including your initial one in the first column) and explain, in each case, how and why you refined your initial model, including an economic rationale for any additional variable.  You should follow the ‘general-to-specific’ (g-to-s) modelling methodology, taking care that any multicollinearity does not lead you to dismiss a theoretically justified variable. Use the Resit Cons 1976-2019.xlsx data file as the source of data on additional variables.  For consistency, all models must be fitted to data for 1976–2008.  In your summary table, show standard errors in brackets beneath the regression coefficients, followed by t ratios in bold type and then p values.  For each model, briefly comment on and report R2, , average VIF value as well as D.W. and χ2 test statistics for serial correlation and heteroscedasticity (along with p values).
Check that you have addressed all parts of this question.
Section 4 (10 marks)
Using your preferred model from Section 3, forecast consumers’ expenditure from 2009 to 2012.  Provide a plot of actual and fitted values of Ct for the whole period, 1976-2012.  Provide a table of actual, fitted, absolute and relative errors for 2009-2012.  How accurate are the forecasts?  Attempt to provide an economic explanation for discrepancies between observed and forecasted expenditure?
Section 5 (5 marks)
Outline what improvements you think could be made, if you had more time, to your preferred model used in Section 4.