EC306 时序预计

EC306 Econometrics II Time Series 2023:Vacation Assignment

Data The data file contains monthly U.S. observations on

– the Unemployment Rate

– The All-Items CPI level

– The Core-CPI level

The first observation is Jan 1957.

1. Compare and contrast the time-series behaviour of the all-items and core inflation levels, Jan-1985 to Dec-

2010. You could comment on the order of integration, stationarity, first and second moment behaviour

of the series. (30 marks)

2. King, Stock and Watson (1995) find that the inflation became harder to forecast using a bivariate model

including unemployment after 1973. Does their result hold with this data on their sample (to Dec-1994)?

What happens if you consider the sample Jan-1995 to Dec-2019? Which shows greater forecastability in

the bivariate context, unemployment or inflation? (30 marks)

3. Using an interpretation of the theory that is appropriate to the time-series properties of the data, test

the hypothesis that there is a one for one relationship between inflation and core inflation in:

(a) The sample from Jan-1957 to Dec-1982. (20 marks)

(b) The sample from Jan-1983 to Jan-2023 (20 marks)

Generic guidance and notes for non economists:

Inflation is the percentage change in the Consumer Price Index (CPI). Log approximations are appropri-

ate.

The Core-CPI index excludes certain items such as food and energy.

You should inspect data by plotting it; it is expected that you make basic transformations to ensure you

work with variables suitable for linear models

Graphs can be included in an answer e.g. if inspection of the graph helps you decide how to make

modelling decisions. Unless otherwise stated in the question, graphical analysis alone is not sufficient.

When you report results of a statistical test, state the test regression, the parameter(s) of interest, the

null hypothesis, the test statistic, the value of the test statistic you find, the critical value and the decision

you make. Bullet points are fine for such discussion

Try to summarize the important parts of the regression output in a table, rather than screen-grabbing

whole chunks from the raw stata output; this makes you think about the output and demonstrates

understanding of what is important.

WX：codehelp mailto: thinkita@qq.com