Monetary policy changes and unit root statistics

Main points

• Conventional unit root tests are biased towards non-rejection of the nuIl when applied to variables characterised by a one-time change in their mean. Suggested modifications allowing for a break point require identification of an exogenous change initiated at the time of the break and not subsequently reversed. This paper suggests that modified tests are particularly useful in dealing with legislative changes. Two UK examples are given; the abolition of exchange controls in 1979 and the changes in monetary policy in 1980. Results suggest that allowing for a break changes the apparent order of integration of portfolio shares and real interest rates.