Working Paper: CEPR ID: DP15758
Authors: Martin Almuzara; Gabriele Fiorentini; Enrique Sentana
Abstract: We analyze a model for N different measurements of a persistent latent time series when measurement errors are mean-reverting, which implies a common trend among measurements. We study the consequences of overdifferencing, finding potentially large biases in maximum likelihood estimators of the dynamics parameters and reductions in the precision of smoothed estimates of the latent variable, especially for multiperiod objects such as quinquennial growth rates. We also develop an R2 measure of common trend observability that determines the severity of misspecification. Finally, we apply our framework to US quarterly data on GDP and GDI, obtaining an improved aggregate output measure.
Keywords: Cointegration; GDP; GDI; Overdifferencing; Signal Extraction
JEL Codes: C32; E01
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| measurement errors in persistent latent time series (C22) | common trend among measurements (C29) |
| ignoring the common trend (C29) | large biases in the estimation of autocorrelation parameters of the latent variable (C20) |
| misspecification of the measurement error dynamics (C32) | large biases in the estimation of autocorrelation parameters of the latent variable (C20) |
| common trend model (C32) | lower signal extraction uncertainty (D89) |
| absence of a common trend (C22) | inflate the uncertainty of estimates (C51) |