When the disturbances of a regression model follow an 1(1) process the
re is a tendency to estimate a break point in the middle of the sample
, even though a break point does not actually exist. In this note, we
provide a mathematical proof for this phenomenon.