Time-Frequency Analysis of Foreign Exchange Rate Periodicities

First published: 2011

Brief summary

Applies bilinear time-frequency distributions (including the Page distribution) to detect hidden periodic components in high-frequency USD/EUR and USD/JPY exchange rate data, an unusual application of spectral methods in a field dominated by ARIMA/GARCH modeling.

Article

Time-Frequency Analysis of Foreign Exchange Rate Periodicities is a preprint published by Academia.edu (working paper) in 2011. It applies bilinear time-frequency distributions (including the Page distribution) to detect hidden periodic components in high-frequency USD/EUR and USD/JPY exchange rate data, an unusual application of spectral methods in a field dominated by ARIMA/GARCH modelling.

The study compares median filter vs. Hodrick-Prescott filter for identifying significant cycle lengths in USD/EUR and USD/JPY exchange rates. It also considers methodology paper on time-frequency periodicity detection. This gives the cycle claim a specific numerical and evidential setting rather than presenting periodicity only as a visual impression.

The article reports the following result: The median filter outperformed the HP filter in identifying significant cycle lengths in both USD/EUR and USD/JPY exchange rates; using spectral analysis is very common in technical areas but rather unusual in economics and finance. The interpretation is strongest when volatility, non-stationarity, market-regime changes and multiple-frequency testing are treated explicitly.

For cycles researchers, the article brings together foreign exchange periodicity, time-frequency analysis, usd-eur, usd-jpy. It is relevant to exchange-rate cycle research because periodic components must be separated from volatility, trend, market microstructure and changing monetary regimes.

Because it is a preprint, the work should be read alongside later peer-reviewed publications and independent replications. It remains useful because the proposed cycle, dataset and analytical approach are stated clearly enough to be scrutinised.

Source details and credits

• Source / publisher: Academia.edu (working paper)
• Source type: Preprint
• URL type: WWW
• Credits: Academia.edu (working paper)
• URL: https://www.academia.edu/916238/Time_frequency