Abstract
We test the rational inattention theory of exchange rates and proxy attention for different economic fundamentals by the search volume of related search queries on Google. We demonstrate that the higher the attention to a certain fundamental, the better its ability to forecast exchange rate movements. Our forecasts based on fundamental model selection by Google Trends significantly outperform the random walk, both statistically and economically. The best forecasts and the highest investment returns are systematically delivered by models that select the most salient fundamental. Both the survey expectations and the predicted currency returns are more persistent than their actual counterparts, consistent with the underreaction to economic news hypothesis. Investment returns are substantially higher during periods of elevated uncertainty, as a result of increased attention. Our findings provide strong support in favor of the rational inattention theory of exchange rates and the similar behaviour of the attention-based forecasts, and survey expectations suggest that the rational inattention operates via the expectation-formation channel.
![](http://media.springernature.com/m312/springer-static/image/art%3A10.1057%2Fs41308-022-00195-3/MediaObjects/41308_2022_195_Fig1_HTML.png)
![](http://media.springernature.com/m312/springer-static/image/art%3A10.1057%2Fs41308-022-00195-3/MediaObjects/41308_2022_195_Fig2_HTML.png)
![](http://media.springernature.com/m312/springer-static/image/art%3A10.1057%2Fs41308-022-00195-3/MediaObjects/41308_2022_195_Fig3_HTML.png)
Similar content being viewed by others
Notes
See Rossi (2013) for an excellent review.
By economic decisions we mean here investment and hedging decisions, central banks’ interventions, goods and services trading decisions, etc.
Da et al. (2011) use Google search volume to proxy for attention for individual stocks and find that it is a more timely measure than other measures of attention. Search engine data has recently been applied in several forms of so-called nowcasting of macro-economic variables as well (Choi and Varian 2009). Goel et al. (2010) show that search behaviour also has predictive content further in the future.
The Google Trends tool is freely accessible via www.google.com/trends.
We left out a few common exchange rate models, because of their limited applicability in combination with the Google Trends Index. An example is the Taylor rule-based model, which is difficult to translate into search queries, because of its complexity and involvement of multiple economic fundamentals.
The results are qualitatively insensitive to the choice of maturities and not depending on taking an average of both forecasts or selecting one of them.
See for instance (Chen et al. 2010).
The reason could be that even in Japan most investors read international news in English and hence search for queries in English.
We include an example of the between model correlations for a specific query, because a full correlations matrix is too sizeable with 35 search queries.
See Rossi (2013) for description of in-sample versus out-of-sample fit of exchange rate models.
The sample period could not be extended further in history, because Google Trends data only dates back to January 2004.
By including all the currency pair combinations, our results are not dependent on the choice of the base currency.
More precisely, the selected model might not be the model with the highest average GTI for both countries. Consider the following example for illustration: the monetary model has an average GTI for its search queries of 70 for the USA and 80 for the UK and the PPP-model has an average GTI of 50 for the USA and 90 for the UK. In that case, the averages for the USD/GBP exchange rate are 75 and 70, respectively. Hence, in this case we prefer the monetary model above the PPP-model, while the PPP-model would be preferred for the UK on an individual basis.
For simplicity, we ignore transaction costs when calculating strategy returns. Given the liquidity of foreign exchange markets and the relatively low frequency of our analysis, however, we do not expect transaction costs to affect the results significantly.
We present figures for these five countries separately, because Google Trends can only show search data for either the entire world or for individual countries and geographical areas within countries. We are therefore not able to aggregate search data over the Eurozone.
Typical concern with using survey data is whether they truly reflects market participants’ expectations. The literature has shown that survey data cannot always adequately capture unobservable expectations, partly because respondents have incentives to provide biased or strategic forecasts. Laster et al. (1999) find, for example, that forecasters issue forecasts more extreme than their true expectations. Campbell and Sharpe (2009) provide evidence of an anchoring bias in consensus forecasts.
The number of currency pairs is restricted to five (GBP, JPY, AUD, CAD and CHF versus the USD), while our GTI-based model selection procedure includes 15 bilateral exchange rates.
Because \(\tau\) = 6 in Eq. (17), the first six forecasts are needed to construct the first forecast weights. As such, out-of-sample forecasts start from July 2004.
Strictly speaking, the outcomes in Table 13 are not directly comparable between GTI-based forecasts and actual returns on the hand, versus survey forecasts on the other hand, because the numbers for survey forecasts are calculated on a sub-sample (2008M7–2016M12) of the sample period (2004M1–2016M12).
The empirical evidence rather suggests that the carry trade performed worse during the Great Recession. Still, we want to rule out the possibility that we capture it instead of attention shifts.
The results for the carry trade strategies are not reported to save space, but are available upon request. Also, returns on carry trades reported in the literature are usually higher what we find. This is most likely explained by the sample period, which is relatively short and includes the break-down of carry trades during the financial crisis.
Results on individual models are not reported, but are available upon request.
References
Andrei, D., and M. Hassler. 2014. Investor attention and stock market volatility. Review of Financial Studies 28: 33–72.
Bacchetta, P., and E. van Wincoop. 2004. A scapegoat model of exchange rate determination. American Economic Review, Papers and Proceedings 94: 114–18.
Bacchetta, P., and E. van Wincoop. 2006. Can information heterogeneity explain the exchange rate determination puzzle? American Economic Review 96: 552–76.
Bacchetta, P., and E. van Wincoop. 2013. On the unstable relationship between exchange rates and macroeconomic fundamentals. Journal of International Economics 91: 18–26.
Bank for International Settlements. 2019. Triennial Central Bank Survey: foreign exchange turnover in April 2019. Basel: Monetary and Economic Department.
Barber, B.M., and T. Odean. 2008. All that glitters: the effect of attention and news on the buying behavior of individual and institutional investors. Review of Financial Studies 21: 785–818.
Ben-Rephael, A., Z. Da, and R.D. Israelsen. 2017. It depends on where you search: institutional investor attention and underreaction to news. The Review of Financial Studies 30: 3009–3047.
Burnside, C., M. Eichebaum, I. Kleshchelski, and S. Rebelo. 2011. Do peso problems explain the returns to the carry trade? Review of Financial Studies 24: 853–891.
Campbell, S., and S. Sharpe. 2009. Anchoring bias in consensus forecasts and its effects on market prices. Journal of Financial and Quantitative Analysis 44 (2): 369–390.
Chen, Y., K. Rogoff, and B. Rossi. 2010. Can exchange rates forecast commodity prices? The Quarterly Journal of Economics 125: 1145–1194.
Cheung, Y., and M.D. Chinn. 2001. Currency traders and exchange rate dynamics: a survey of the US market. Journal of International Money and Finance 20 (4): 439–471.
Cheung, Y., M.D. Chinn, and A.G. Pascual. 2005. Empirical exchange rate models of the nineties: are any fit to survive? Journal of International Money and Finance 24: 1150–1175.
Clark, T., and K.D. West. 2007. Approximately normal tests for equal predictive accuracy in nested models. Journal of Econometrics 138: 291–311.
Choi, H., and H. Varian. 2009. Predicting the present with Google Trends. Economic Record 88: 2–9.
Coibion, O., and Y. Gorodnichenko. 2012. What can survey forecasts tell us about information rigidities? Journal of Political Economy 120 (1): 116–159.
Coibion, O., and Y. Gorodnichenko. 2015. Information rigidity and the expectations formation process: a simple framework and new facts. American Economic Review 105 (8): 2644–2678.
Da, Z., J. Engelberg, and P. Gao. 2011. In search of attention. Journal of Finance 66: 1461–1499.
Della Corte, P., L. Sarno, and I. Tsiakas. 2009. An economic evaluation of empirical exchange rate models. Review of Financial Studies 22: 3491–3530.
Della Corte, P., and I. Tsiakas. 2012. Statistical and economic methods for evaluating exchange rate predictability. In Handbook of Exchange Rates, ed. J. James, I.W. Marsh, and L. Sarno. Hoboken, NJ: Wiley.
Engel, C., and K.D. West. 2005. Exchange rate and fundamentals. Journal of Political Economy 113: 485–517.
Fratzscher, M., D. Rime, L. Sarno, and G. Zinna. 2015. The scapegoat theory of exchange rates: the first tests. Journal of Monetary Economics 70: 1–21.
Goel, S., J.M. Hofman, S. Lahaie, D.M. Pennock, and D.J. Watts. 2010. Predicting consumer behavior with web search. Proceedings of the National Academy of Sciences 7: 17486–17490.
Hattori, M., and H.S. Shin. 2009. Yen carry trade and the subprime crisis. IMF Staff Papers 56: 384–409.
Hirshleifer, D., and S.H. Teoh. 2003. Limited Attention, Information Disclosure, and Financial Reporting. Journal of Accounting and Economics 36: 337–386.
Kouwenberg, R., A. Markiewicz, R.C. Verhoeks, and R.C.J. Zwinkels. 2017. Model uncertainty and exchange rate forecasting. Journal of Financial and Quantitative Analysis 52: 341–363.
Laster, D., P. Bennett, and I.S. Geoum. 1999. Rational bias in macroeconomic forecasts. Quarterly Journal of Economics 114 (1): 293–318.
Mackowiak, B., F. Matejka, and M. Wiederholt. 2020. Rational inattention: a review. CEPR working paper DP15408.
Meese, R.A., and K. Rogoff. 1983. Empirical exchange rate models of the seventies: do they fit out of sample? Journal of International Economics 14: 345–373.
Molodsova, T., and D.H. Papell. 2010. Out-of-sample exchange rate predictability with Taylor rule fundamentals. Journal of International Economics 77: 167–180.
Mondria, J., T. Wu, and Y. Zhang. 2010. The determinants of international investment and attention allocation: using internet search query data. Journal of International Economics 82 (1): 85–95.
Pashler, H.E. 1998. The Psychology of Inattention. Cambridge: MIT Press.
Pashler, H., and J.C. Johnston. 1998. Attentional limitations in dual-task performance. In Attention, ed. H. Pashler. Psychology Press/Erlbaum (UK) Taylor & Francis.
Peng, L., and W. **ong. 2006. Investor attention, overconfidence and category learning. Journal of Financial Economics 80: 563–602.
Pesaran, M.H., and A. Timmermann. 1992. A simple nonparametric test of predictive performance. Journal of Business and Economic Statistics 10: 461–465.
Rensink, R.A., J.K. O’Regan, and J.J. Clark. 1997. To see or not to see: the need for attention to perceive changes in scenes. Psychological Science 8: 368–373.
Rossi, B. 2013. Exchange rate predictabibility. Journal of Economic Literature 51 (4): 1063–1119.
Sarno, L., and G. Valente. 2009. Exchange rates and fundamentals: footloose or evolving relationship? Journal of European Economic Association 7: 786–830.
Schinasi, G.J., and P.A.V.B. Swamy. 1989. The out-of-sample forecasting performance of exchange rate models when coefficients are allowed to change. Journal of International Money and Finance 8: 375–390.
Sicherman, N., G. Loewenstein, D.J. Seppi, and S.P. Utkus. 2016. Financial attention. Review of Financial Studies 29: 863–897.
Simons, D.J., and C. Chabris. 1999. Gorillas in our midst: sustained inattentional blindness for dynamic events. Perception 28: 1059–1074.
Sims, C.A. 2003. Implications of rational inattention. Journal of Monetary Economics 50 (3): 665–690.
Stroop, J. 1935. Studies of interference in serial verbal reactions. Journal of Experimental Psychology 28: 643–662.
Timmermann, A. 2006. Forecast combinations. In Handbook of Economic Forecasting, ed. G. Elliot, C.W.J. Granger, and A. Timmermann. Elsevier.
Wright, J. 2008. Bayesian model averaging and exchange rate forecasting. Journal of Econometrics 146: 329–341.
Yuan, Y. 2015. Market-wide attention, trading and stock returns. Journal of Financial Economics 116: 548–564.
Acknowledgements
We thank participants of the Australasian Finance and Banking Conference 2017; the Society for Nonlinear Dynamics and Econometrics Conference 2018; the European Economic Association Annual Meeting 2018; the Research in Behavioral Finance Conference 2018; and anonymous referees for their helpful comments and suggestions. A previous version of this paper circulated under the title “The Winner Takes It All: Predicting Exchange Rates with Google Trends”.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix—How Google Trends Data Are Used
Appendix—How Google Trends Data Are Used
In this appendix we describe in more detail how we use the tool of Google Trends to extract the data series of the Google Trends Index, our proxy for investor attention for the different exchange rate fundamentals. First, it is important to specify the search query correctly. Google Trends gives different possibilities to use punctuation in searches to filter the results. Table 16 below shows how the use of punctuation influences the filtering of results by means of an example query from the monetary model.
For all different searches above, it should be noted that no misspellings, spelling variations, synonyms, or plural or singular versions of the search query are included. Based on the consequences of the different uses of punctuation, we decided to always apply the double quotation marks in our queries (the second line in the table above). This option most appropriately filters the outcomes and hinders unintended queries to influence the results. On the other hand, using the double quotation marks is relatively strict: if we enter money supply, the results will not include queries like supply of money, because the order of words is different. However, to be as objective as possible, we apply strict filters to the Google Trends tool, such that the Index only identifies phenomena we are interested in.
The punctuation options in the table above can also be combined. For instance, if we would enter “money supply” + “money demand” the results will include all searches containing money supply or money demand. We have used the option to combine the punctuation options in the case of Japan. In many cases, the English version of our search queries did not have enough coverage for Google to calculate the index (which is automatically indicated by Google). In that case we entered both the English and the Japanese term for the query involved (using Google Translate). For instance, the English query “balance of trade” did not result in sufficient coverage for Japan. Therefore, we add the Japanese translation to our search query. The use of translations of English search queries is restricted to exchange rates where the Japanese Yen was involved. Extending our analysis to other currencies is problematic, in particular in case of smaller countries, due to limited Google coverage. In the case of our search queries, it was in most cases possible to calculate country specific GTIs by restricting the data to a specific country. However, unreported analysis shows that this would be more challenging for other countries, particularly when Google is not intensely used in the country involved. In the case of the Euro currency, the use of Google Trends data is also a bit more problematic as the Eurozone encloses different languages.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.