Single step estimation of ARMA roots for nonfundamental nonstationary fractional models
提出一种单步估计方法，用于非平稳分数型ARMA过程的自回归和移动平均根，无需施加因果性或可逆性限制，并应用于道琼斯工业平均指数成分股交易量，发现非基础性证据且非因果性比非可逆性更常见。
Summary We propose a single step estimator for the autoregressive and moving average roots (without imposing causality or invertibility restrictions) of a nonstationary Fractional ARMA process. These estimators employ an efficient tapering procedure, which allows for a long memory component in the process, but avoids estimating the nonstationarity component, which can be stochastic and/or deterministic. After selecting automatically the order of the model, we robustly estimate the AR and MA roots for trading volume for the thirty stocks in the Dow Jones Industrial Average Index in the last decade. Two empirical results are found. First, there is strong evidence that stock market trading volume exhibits nonfundamentalness. Second, noncausality is more common than noninvertibility.