A NOTE ON SPURIOUS BREAK

摘要: Recently, Nunes, Kuan, and Newbold (1995) (henceforth NKN) pointed out that when the disturbances of a regression model follow an I(1) process there is tendency to estimate break point in middle sample, even though does not actually exist. This phenomenon called "spurious break" by authors was discovered simulation experiment. In this note, we provide mathematical proof for phenomenon. It interest ask following question, it is, fact, often asked. Given with no point, supposing entertained estimation, how estimated behave? Let k denote T sample size. Define AT = kIT. Thus, denotes fraction. For I(O) disturbances, can be shown fraction converges boundary (i.e., either 0 or 1). Because values imply conforms true break. However, different emerges disturbances. NKN found will stay suggesting existence point. The call spurious analogy regression. Although problem much better understood well documented (see Granger Newbold, 1974; Phillips, 1986; Durlauf 1988), less studied. note takes up issue. particular, shall deliver converge 1, corroborating confirming findings NKN. should emphasized our analysis assumes absence data-generating process. result suggests caution exercised estimating are I(1). Diagnostic testing performed prior es-