The franc shock and Swiss GDP: how long does it take to start feeling the pain?

ABSTRACT

The article addresses the question on what is the typical time horizon over which a full transmission of movements in the real exchange rate takes place into real economy. We base our analysis on the mixed-frequency small-scale dynamic factor model (DFM) proposed by Siliverstovs in 2012 fitted to the Swiss data. In this article, we augment the benchmark model with the real exchange rate of the Swiss franc vis-a-vis currencies of its 24 trading partners, while keeping the rest of model specification intact. We are interested in investigating the relationship between the common latent factor, representing the Swiss business cycle, and the real exchange rate. We explore the temporal relationship between these two variables by varying the time lag with which the real exchange rate enters the factor model by recording the magnitude and statistical significance of the factor loading coefficient in the equation pertaining to the real exchange rate variable. Our main conclusion is that the fluctuations in the exchange rate start influencing real economy after 1 month and their effect is practically over after 13 months. The largest effect is recorded at the time horizon of about 6 to 9 months.

KEYWORDS:

• Factor model
• mixed-frequency data
• exchange rate pass-through
• GDP growth

JEL CLASSIFICATION:

• C32
• E32

Acknowledgements

We are grateful for useful comments and feedback following the presentation of the article at the KOF Brown Bag seminar, Finance and Economics Conference 2015 (Frankfurt am Main, Germany), and DIW Berlin Macroeconometric Workshop (Berlin, Germany). All computations and graphics were carried out in OX version 6.30 (Doornik 2007) and in R version 3.0.2. (R Core Team 2013). The usual disclaimer applies.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ http://www.schweizer-franken.eu/2015/01/ploetzliches-eurchf-mindestkurs-ende.html.

² Camacho and Doménech (2012) also differentiate between the leading properties of the auxiliary indicators. In their small-scale DFM, they selectively allow some financial variables to lead the common factor.

³ For the sake of brevity, the complete state-space form of the DFM is relegated to the Appendix.

⁴ Recall that 1995M1 is the earliest period for which the PMI is available.

⁵ We make the complete model estimation results available upon the request.

⁶ Please observe that the colours refer to the online version of the article. Since in the printed version of the article the figure is in black and white, the significance at 5%, 10% and 20% levels is indicated by areas corresponding to the monochrome scala on the right, respectively.

Additional information

Funding

This article is a part of a larger project ‘Real-time forecasting with mixed-frequency data’ that benefited from the financial support of the MTEC Foundation at ETH Zurich.