Measuring Measurement Error in Economic Time Series
提出一种谱分解方法，用于估计时间序列中测量误差的上限，并应用于M1b、实际GNP和CPI数据，揭示这些序列的测量误差大小及其随机特征。
A spectral decomposition method is described for obtaining an upper bound on the amount of measurement error in a time series. The method is applied to generated data and to M1b, real GNP, and the CPI. The bounds provide insight into both the amount of measurement error in these series and the stochastic specification of the errors.