Estimating a static Phillips curve

by vimothy

Unemployment + boredom + Stata = this

To estimate a “classic” static Phillips curve, I analysed annual UK data for the period 1971-2010 (ONS series BCJE for inflation and series CZBH for unemployment). I estimated an equation of the form,

Where RPI is the annual year-on-year percentage change in the RPI index (i.e. the annual rate of inflation) and unemp is the rate of unemployment as a percentage of the population.

To examine the trade off between unemployment and inflation, I test,

Using the OLS t-statistic. Under the null, t has a  t-distribution with 40 df. Therefore, reject at the 5% level if .

Using the ONS data, I estimate the following model:

(1.883)   (0.3090)

Since , this suggests a trade-off between unemployment and inflation. The t-statistic is approximately -1.47. The p-value against a two-sided alternative is 0.149. Therefore, we fail to reject the null at the 5% level.

Unfortunately, the Classical Linear Model assumptions that OLS relies on are violated in this case. I may address this in a future post…