ASSESSMENT OUTLINE

You are required to specify and estimate a multiple regression model that can be

used for generating forecasts of some variable that is of interest to you.

Broad Overview of the Assessment

Your first task is to identify a variable of interest. You may wish to search through

the Office for National Statistics web site (http://www.ons.gov.uk/) , the databases

contained on the UK Data Service site (http://ukdataservice.ac.uk/), particularly the

OECD Main Economic Indicators dataset (a guide to accessing and downloading

these data can be found at

http://esds80.mcc.ac.uk/wds_oecd/TableViewer/document.aspx?ReportId=725), or

the various databases referred to on the Biz/ed web site

(http://www.bized.co.uk/dataserv/freedata.htm) in order to identify a relevant

variable. In each case you will need to focus on searching for annual time-series

data. You may also consult the statistical collection in the Library, or any other

library to which you have access, or any other database to which you have access.

Alternatively, you may already have a variable of interest derived from the other

modules that you are studying or previous work/study experience.

In any event, the variable should be economics/finance/business/sociological in

nature, and you should obtain annual observations only (that is, you should not

use daily, weekly, monthly or quarterly data). You will then be required to specify

and estimate a regression model to be used for forecasting purposes, which should

contain at least two but not more than four independent variables. You must

obtain a dependent variable with at least 40 annual observations, and the

time-period should extend to at least 2013. That is, the start date of your

data series must be no later than 1974.

Assessment Details

The details of the assessment are as follows (your assessment should clearly

indicate your answers to each of the following 5 parts, each of which should be

labelled/headed accordingly):

1. Provide a description of the dependent variable you have selected, and

provide a detailed discussion as to why you consider this variable to be of

interest. You should collect at least 40 annual observations on this

variable, and the time-period should extend to at least 2013.

You must provide details of the source(s) from which you obtained your data,

in addition to presenting a table of your data in an appendix, which should

also include the data and sources on your independent variables detailed in

Parts 2 and 3 below (if more than 40 observations are available you should

use all of the observations). FAILURE TO USE A DATA SERIES MEETING

THESE REQUIREMENTS WILL RESULT IN A REDUCTION OF UP TO 20

MARKS FROM THE FINAL GRADE AWARDED TO THE ASSESSMENT.

You should place an emphasis on deriving an ‘interesting’ dependent variable

that exhibits considerable variability and would therefore be challenging to

model. For example, should your selected variable exhibit very little year to

year variability, and hence be of little interest for modelling and forecasting

purposes then you should consider transforming this variable into growth rate

form – that is, transform the variable so that it measures the percentage

change from year to year – and use this variable as your dependent variable.

In general a variable expressed in growth rate form, rather than levels form,

presents a more interesting forecasting challenge. (See the following

paragraph and the appendix for a more detailed discussion of what

constitutes an appropriate data series for the purposes of this assessment.)

Present a graph of the data on your dependent variable, and place your

discussion within the context of this graph, providing an overview of the

broad movements in the data, and if appropriate, some tentative explanations

for some of these movements. If your data are measured in monetary terms,

be clear as to whether the data are measured in current or constant prices,

and why you consider the price base you are using to be appropriate. You

must not use any textbooks as a data source, nor should you use the dependent variables that have been used in examples that have

been covered in lectures, seminars and handouts In particular, you

should NOT develop any models of aggregate consumers’

expenditure, either for the UK or any other country. If you are in any

doubt as to the appropriateness of your selected data series you should

consult the module leader. (10 marks)

(The Appendix to these assessment details provides graphs of

unacceptable and acceptable dependent variable data series. Thus

Figure 1 presents a data series that would NOT be acceptable for the

purposes of this assessment as it exhibits very predictable year to

year variation, and therefore can be forecast very easily by simple

extrapolation, rather than requiring an econometric model. Figure 2

presents a data series exhibiting much more irregular year to year

variation than is the case with the data in Figure 1, and hence would

be an acceptable dependent variable for the purposes of this

assessment. Figure 3 presents the annual percentage change of the

data series in Figure 1 – simply derived as the percentage change in

the series from year to year – and also would be an acceptable data

series for the purposes of this assessment. That is, if your selected

data series is similar in form to that shown in Figure 1, but you still

consider the data series to be of some intrinsic interest, then you

should transform this series to a growth rate series, as in Figure 3,

and then use this growth rate series as your dependent variable.

But note that if you adopt this approach you should give careful

consideration to the appropriate form of the independent variables

in your model.)

2. Specify a single equation econometric model that you consider should provide

an adequate explanation for the annual variation in the dependent variable

you have identified under Part(1) above. Your model should contain at least

2 but not more than 4 independent variables. Provide a detailed discussion of

the expected relevance of the variables that you have selected, and the

manner in which you would expect these variables to influence your

dependent variable. At this stage, you should not be concerned about the

availability of data on your proposed independent variables, but rather you

should place an emphasis on the structure of your ideal model.

(25 marks)

3. Collect sample data on the independent variables you identified under Part 2.

above, indicating your data source(s) clearly (which again should not be a

textbook nor derived from lectures, seminars, handouts). You should include

these data in a table in an appendix. Again, if any of your independent

variables are measured in monetary terms, be clear as to whether the data

are measured in current or constant prices, and why you consider the price base you are using to be appropriate.

If you cannot locate an appropriate data series for one or more of your

proposed independent variables, feel free to use appropriate proxy variables –

that is, variables that you consider should exhibit similar variability to your

‘ideal’ variables that you discussed in Part (2).

You may find that you identify appropriate independent variables, but that

data are not available over the full 40 (or more)-year period corresponding to

the dependent variable. You should make whatever compromises that you

consider appropriate, and justify these compromises.

Using EViews, estimate the initial version of your model, but drop the last 5

years from your data set (that is, the years 2009 to 2013 – this period will be

used to test the forecasting performance of your model). Present the EViews

output, and provide a discussion of the main features of your estimated

model, using the appropriate diagnostic testing procedures in EViews.

In the light of your regression output, discuss any inadequacies in your

model. Amend your model appropriately, in terms of re-specifying the form in

which your independent variables enter the model, disturbance term

specifications, etc. You should not spend too much time finding data

series on new independent variables, but rather indicate additional

or replacement variables that you might explore, given the time.

Re-estimate your model in the light of this discussion, again presenting and

discussing the EViews output. Provide a clear statement of your finally

selected model, and provide a clear justification for this finally selected model.

The objective of this part of the assessment is for you to provide a detailed

discussion of the process you went through to decide upon the final version of

your model. (40 marks)

4. Using your finally selected model, generate forecasts over the 5 year forecast

period, and discuss the forecasting performance of your model in the light of

these forecasts, and in comparison to the actual data values for this 5 year

period. You should use the various procedures in EViews for evaluating

forecasting performance. Does this forecasting performance suggest any

further improvements that could be made to your model? If so, what

adjustments would you consider making to your model? (15 marks)

5. Provide a critical evaluation of the econometric approach to model building

and forecasting in the light of your answers to Parts 1 to 4 above.

(10 marks)

Not to exceed 2000 words, excluding computer output, graphs and appendices.

Is this the question you were looking for? If so, **place your order here** to get started!