A new approach to detect persistence change in fractionally integrated models based on recursive forward and backward estimation of regression-based Lagrange Multiplier tests is proposed. This procedure generalizes approaches for conventional integrated processes to the fractional integration context. Asymptotic results are derived and the performance of the new tests evaluated in a Monte Carlo exercise. In particular, analytical and simulation results are provided for cases where the order of fractional integration is both known and unknown and needs to be estimated. The finite sample size and power performance of the statistics are encouraging and compare favorably to other recently proposed tests in the literature. The test statistics introduced are also applied to several world inflation rates and evidence of persistence change is found in most series

--

LM tests; Nonstationarity; Fractional integration; Persistence change; Inflation

• Matemáticas - Ciências Naturais
• Ciências da Computação e da Informação - Ciências Naturais