This paper adapts the non-parametric dynamic time warping (DTW) technique in an application to examine the temporal alignment and similarity across economic time series. DTW has important advantages over existing measures in economics as it alleviates concerns regarding a pre-defined fixed temporal alignment of series. For example, in contrast to current methods, DTW can capture alternations between leading and lagging relationships of series. We illustrate DTW in a study of US states’ business cycles around the Great Recession, and find considerable evidence that temporal alignments across states dynamic. Trough cluster analysis, we further document state-varying recoveries from the recession.

Additional Metadata
Keywords Business cycles, Dynamic time warping, Non-parametric method
JEL Semiparametric and Nonparametric Methods (jel C14), Econometric Modeling: General (jel C50), Large datasets: Modelling and Analysis (jel C55), Econometric Software (jel C87), Business Fluctuations; Cycles (jel E32)
Persistent URL dx.doi.org/10.1007/s10614-020-09986-0, hdl.handle.net/1765/126617
Series Econometric Institute Reprint Series
Journal Computational Economics
Citation
Franses, Ph.H.B.F, & Wiemann, T. (2020). Intertemporal Similarity of Economic Time Series: An Application of Dynamic Time Warping. Computational Economics. doi:10.1007/s10614-020-09986-0