No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Using Out-of-Sample Mean Squared Prediction Errors to Test the Martingale Difference Hypothesis
... To account for potential seasonality in these data, I will adjust the prices, output, and money by averaging the log 4 levels over four quarters prior to any empirical analysis. 6 Forecasting evaluation I employed the recursive approach to produce predictions. The paper utilizes data spanning from 1973:1 to 2021:4 to estimate factors and their loadings, and to constructF it for i = 1, ldots, 17. ...
... We present one-sided hypothesis tests at the 10 percent significance level with H0: RM-SPE(our model) = RMSPE(random walk) against HA: RMSPE(our model) ¡ RMSPE(random walk). These tests follow the methodology of Clark and West (2006), who developed a procedure to address the potential inflation of the factor model's RMSPE. Given the numerous currencies (17 in our initial sample), it's likely that some test statistics will be significant even if predictions aren't better than a random walk. ...
... Therefore, I decided not to report it to limit the number of figures presented. 6 ...
Forecasting the exchange rate using factor analysis, factors based on cross section exchange rates of different economies in terms of trade, would beat random walk model in longer out of sample horizon forecasting. In this research project, we will replicate the NBER working paper Factor Model Forecasts of Exchange Rates" by Charles En-gel, Nelson C. Mark & Kenneth D. West (2012) which has used 17 OECD countries exchange rates in getting the potential factors but with extended time period from 1973:1-2021:4. We will report early sample forecasting statistics for a 1987-1998 sample. Results for late samples (1999-2022) were promising, at least for horizons of 8,12 or 16 quarters and early sample with factor 1 and 3 for shorter horizon 1 / 4. 1 Research question Is it possible to improve exchange rate forecasting through factor analysis, where the factors are derived from the cross-sectional exchange rates of various countries? Numerous studies and research efforts have attempted to develop exchange rate forecasting models, yet most have failed to surpass the random walk model, which essentially predicts that the nominal exchange rate will remain unchanged (i.e., the forecast implies no variation in the exchange rate). Numerous models attempt to forecast exchange rates on a theoretical basis, linking exchange rates to key fundamentals such as money supply, economic output, inflation, productivity,
... As we will see in the next section, in the particular case in which the null model is a simple martingale in difference process and parameter estimates are updated in rolling windows, expression (2.17) will converge in probability to 1 when the null hypothesis is true. This, Slutsky's theorem plus asymptotic normality of the test by Clark and West (2006) ensures asymptotic normality for our approach. In the case of the test in Clark and West (2007) which is not normal, we rely on the good behavior of the normal approximation described by simulations in that paper, and many others, to use normal critical values for our test as well. ...
... Here we provide a formal asymptotic analysis for our new test in the particular case in which the null model is a simple martingale in difference process and parameter estimates are updated in rolling windows, as in Clark and West (2006). This means that in (2.1) and (2.2) we are considering the special case β = 0. ...
... DGP 1: Here we focus on the case where the null is a martingale model. DGP 1 is fairly similar to the first DGP in Pincheira and West (2016) and to those used in Clark and West (2006), Mankiw and Shapiro (1986), Nelson and Kim (1993), Stambaugh (1999), Campbell (2001), Tauchen (2001) and Pincheira (2013). This DGP is designed to match exchange rate series for which the martingale difference is a plausible null hypothesis and a model based on uncovered interest parity is a plausible alternative. ...
In this paper we introduce a “power booster factor” for out-of-sample tests of predictability. The relevant econometric environment is one in which the econometrician wants to compare the population Mean Squared Prediction Errors (MSPE) of two models: one big nesting model, and another smaller nested model. Although our factor can be used to improve finite sample properties of several out-of-sample tests of predictability, in this paper we focus on the widely used test developed by Clark and West (2006, 2007). Our new test multiplies the Clark and West t-statistic by a factor that should be close to one under the null hypothesis that the short nested model is the true model, but that should be greater than one under the alternative hypothesis that the big nesting model is more adequate. We use Monte Carlo simulations to explore the size and power of our approach. Our simulations reveal that the new test is well sized and powerful. In particular, it tends to be less undersized and more powerful than the test by Clark and West (2006, 2007). Although most of the gains in power are associated to size improvements, we also obtain gains in size-adjusted-power. Finally we illustrate the use of our approach when evaluating the ability that an international core inflation factor has to predict core inflation in a sample of 30 OECD economies. With our “power booster factor” more rejections of the null hypothesis are obtained, indicating a strong influence of global inflation in a selected group of these OECD countries.
... We use equation (7) as our benchmark setup in calculating h-horizon-ahead out-of-sample forecasting intervals. According to equation (8), the matrix X 1,t in equation (1) includes economic variables q t ≡ s t + p * t − p t , ...
... As a result, the interval obtained from this method is semi-parametric. Wu (2007) shows that under some weak regularity conditions, this method always consistently estimates the forecast distribution, 8 and hence the forecast intervals, of s τ +h − s τ conditional on X m,τ , regardless of the quality of model m. That is, the forecast intervals are robust. ...
... 7 We choose b using the method of Hall, Wolff, and Yao (1999). 8 It is consistent in the sense of convergence in probability as the estimation sample size goes to infinity. ...
This paper attacks the Meese-Rogoff (exchange rate disconnect) puzzle from a different perspective: out-of-sample interval forecasting. Most studies in the literature focus on point forecasts. In this paper, we apply Robust Semi-parametric (RS) interval forecasting to a group of Taylor rule models. Forecast intervals for twelve OECD exchange rates are generated and modified tests of Giacomini and White (2006) are conducted to compare the performance of Taylor rule models and the random walk. Our contribution is twofold. First, we find that in general, Taylor rule models generate tighter forecast intervals than the random walk, given that their intervals cover out-of-sample exchange rate realizations equally well. This result is more pronounced at longer horizons. Our results suggest a connection between exchange rates and economic fundamentals: economic variables contain information useful in forecasting the distributions of exchange rates. The benchmark Taylor rule model is also found to perform better than the monetary and PPP models. Second, the inference framework proposed in this paper for forecast-interval evaluation, can be applied in a broader context, such as inflation forecasting, not just to the models and interval forecasting methods used in this paper.
... A positive R 2 implies that the predictive regression produces lower average mean-squared prediction error than the historical average. To compare the performance across models I use the test statistic from Clark and West (2006) and Clark and West (2007). ...
... The R OOS is computed for the posterior mean. Clark and West test by Clark and West (2007) and Clark and West (2006) is used to compare the outof-sample mean squared prediction errors produced by the models relative with those based on the historical mean under one-sided hypothesis. The stars flag levels of significance: * * * p < 0.01, * * p < 0.05, * p < 0.10. ...
... Let me turn to the out-of-sample results now. The statistical measure of forecast performance relative to the naive benchmark out-of-sample R 2 OOS is reported in -0.08 -0.14 -0.09 -0.08 Clark and West test by Clark and West (2007) and Clark and West (2006) is used to compare the out-of-sample mean squared prediction errors produced by the models relative with those based on the historical mean under one-sided hypothesis. This table reports the CERs for an investor with a power utility function with a coefficient of relative risk aversion A = 5. ...
In this dissertation, I revisit two problems in empirical asset pricing. In Chapter 1, I propose a methodology to evaluate the validity of linear asset pricing factor models under short sale restrictions using a regression-based test. The test is based on the revised null hypothesis that intercepts obtained from regressing excess returns of test assets on factor returns, usually referred to as alphas, are non-positive. I show that under short sale restrictions a much larger set of models is supported by the data than without restrictions. In particular, the Fama-French five-factor model augmented with the momentum factor is rejected less often than other models. In Chapter 2, I investigate patterns of equity premium predictability in international capital markets and explore the robustness of common predictive variables. In particular, I focus on predictive regressions with multiple predictors: dividend-price ratio, four interest rate variables, and inflation. To obtain precise estimates, two estimation methods are employed. First, I consider all capital markets jointly as a system of regressions. Second, I take into account uncertainty about which potential predictors forecast excess returns by employing spike-and-slab prior. My results suggest evidence in favor of predictability is weak both in- and out-of-sample and limited to a few countries. The strong predictability observed on the U.S. market is rather exceptional. In addition, my analysis shows that considering model uncertainty is essential as it leads to a statistically significant increase of investors’ welfare both in- and out-of-sample. On the other hand, the welfare increase associated with considering capital markets jointly is relatively modest. However, it leads to reconsider the relative importance of predictive variables because the variables that are statistically significant predictors in the country-specific regressions are insignificant when the capital markets are studied jointly. In particular, my results suggest that the in-sample evidence in favor of the interest rate variables, that are believed to be among the most robust predictors by the literature, is spurious and is mostly driven by ignoring the cross-country information. Conversely, the dividend-price ratio emerges as the only robust predictor of future stock returns.
... Then, equal predictive accuracy is tested by checking whether the population mean of is zero or not. We follow the West's (2006 and inference procedure that tests the predictive ability of an econometric model against a martingale difference model. In this approach, under the null, we have the following: ...
... Hence, under the null, MSPE of the structural model is higher by construction. Therefore, the adjustment proposed by Clark and West (2006) corrects for this bias so that the CW test statistics is normally distributed. Clark and West (2006) adjustment to the loss-differential function ( ) is equal to (̂−̂) 2 wherê is the prediction obtained from the null model and̂is the prediction of the econometric model. ...
... Therefore, the adjustment proposed by Clark and West (2006) corrects for this bias so that the CW test statistics is normally distributed. Clark and West (2006) adjustment to the loss-differential function ( ) is equal to (̂−̂) 2 wherê is the prediction obtained from the null model and̂is the prediction of the econometric model. A statistically significant positive CW test is read as better performing structural model i over the benchmark model b. ...
In this paper, we use Google Trends data to proxy macro fundamentals
that are related to two conventional structural determination of exchange
rate models: purchasing power parity model and the monetary exchange
rate determination model. We assess forecasting performance of Google
Trends based models against random walk null on Turkish Lira-US Dollar
exchange rate for the period of January 2004 to August 2015. We offer a
three-step methodology for query selection for macro fundamentals in
Turkey and the US. In out-of-sample forecasting, results show better
performance against no-change random walk predictions for specifications
both when we use Google Trends data as the only exchange rate predictor
or augment it with exchange rate fundamentals. We also find that Google
Trends data has limited predictive power when used in year-on-year
growth rate format.
... Also, the use of bootstrapped or finite-sample critical values tends to reduce the rejection rate and, hence, reinforces the low rejection results reported in the following sections. The case of using the adjusted MSPE statistic (Clark and West, 2006) is discussed in Section 4.4. ...
... 25 It is known that the use of the Diebold-Mariano statistic may yield a conservation test against the random walk specification. To enhance the test power, Clark and West (2006) proposes an adjusted mean squared prediction error (MSPE) statistic. For brevity, we present in Appendix 3 a description of the Clark-West statistic, and the forecast comparison results based on this statistic in Appendix 4 in a layout similar to that of Table 1. ...
... To evaluate the forecasting accuracy of the different structural models, we use the adjusted mean squared prediction error (MSPE) statistic proposed by Clark and West (2006). Under the null hypothesis, the MSPE of a zero mean process is the same as the MSPE of the linear alternative. ...
... Empirical findings from studies that use the interest differential as predictor are not very positive. Clark and West (2006) report predictability only for one out of four FX rates considered and only for the short horizon (one month ahead). ...
... . So the supF test is then based upon the following statistic: is also used in Clark and West (2006), where the null model is represented by a zero mean martingale difference model. In that case, the one-step-ahead prediction of the null model is always 0, changing the forecast errors and the adjusted terms accordingly. ...
... withCheung et al. (2005) who find more positive evidence for the shorter horizons,58 whileClark and West (2006) andMolodtsova and Papell (2009) find no predictability of the USD/GBP depreciation rate for the 1-month-ahead horizon using linear regressions. The term structure of forward premia seem to be significantly (at the 10% level) good predictors, but only for the 1-quarter-ahead horizon. ...
... In Monte Carlo experiment with simulated data, we show that as a point forecast criterion, the Clark and West's (2006) unconditional test of mean squared prediction errors does not reflect the relative performance of a superior model over a relatively weaker one. The simulation results show that even though the mean squared prediction errors of a constructed superior model is far below a weaker alternative, the Clark-West test does not reflect this in their test statistics. ...
... For that purpose, Diebold and Mariano (1995) and West (1996) proposed a test of equal predictability of two non-nested models. Moreover, Clark and West (2006) introduced the test of equal predictability for nested models. In this approach, a quadratic loss function is defined as the square of the prediction error. ...
... Section 2 summarizes the Diebold and Mariano (1995) and West (1996) and Clark and West (2006) approach to predictive accuracy testing. Section 3 introduces simulation practice where two series that have a non-linear relation by construction are created to look at the in-sample and out-of-sample performance of OLS and non-parametric model. ...
In Monte Carlo experiment with simulated data, we show that as a point forecast criterion, the Clark and West's (2006) unconditional test of mean squared prediction errors does not reflect the relative performance of a superior model over a relatively weaker one. The simulation results show that even though the mean squared prediction errors of a constructed superior model is far below a weaker alternative, the Clark- West test does not reflect this in their test statistics. Therefore, studies that use this statistic in testing the predictive accuracy of alternative exchange rate models, stock return predictability, inflation forecasting, and unemployment forecasting should not weight too much on the magnitude of the statistically significant Clark-West tests statistics.
... As a result, the response of aggregate consumption and output is unambiguously stronger than in the RA model. 23 In addition to the presence of Non-Ricardian household, di¤erent estimates for parameters and shock distributions determine asymmetries in the dynamic performance of the RA and LAMP models. To better understand the role of the Non-Ricardian households group, we also investigate the counterfactual responses of key macroeconomic variables to a stochastic simulation of the LAMP model where we impose = 0 (see Table 8). ...
... 24 With the notable 22 The speci…c role of LAMP in explaining the co-movements of consumption with investment and output, observed in the data, was …rst discussed in Furlanetto et al. (2013). 23 After a slight initial fall, Ricardian households consumption rises well above the levels observed for the RA model. This is due to the favourable redistributive e¤ect associated to the fall of the labor income share. ...
We estimate a medium‐scale dynamic stochastic general equilibrium model for the Euro area with limited asset market participation (LAMP). Our results suggest that in the recent European Monetary Union years LAMP is particularly sizable (39% during 1993–2012) and important to understand business cycle features. The Bayes factor and the forecasting performance show that the LAMP model is preferred to its representative household counterpart. In the representative agent model the risk premium shock is the main driver of output volatility in order to match consumption correlation with output. In the LAMP model this role is played by the investment‐specific shock, because non‐Ricardian households introduce a Keynesian multiplier effect and raise the correlation between consumption and investments. We also detect the contractionary role of monetary policy shocks during the post‐2007 years. In this period consumption of non‐Ricardian households fell dramatically, but this outcome might have been avoided by a more aggressive policy stance. (JEL C11, C13, C32, E21, E32, E37)
... Since then, exchange rate predictability has become an obsession in the literature with a number of articles trying to overturn the seminal results of Rogoff (1983a, 1983b) or simply trying to address the problem from another perspective, using a new dataset, theory or econometric technique. See for instance Cheung, Chinn, and Pascual (2005), Clark and West (2006), Engel and West (2005), Engel, Mark and West (2015), Molodtsova and Papell (2009) and Ince and Molodtsova (2016) just to mention a few. While in the last years, some papers have shown to outperform the DRW, according to the review in Rossi (2013a), "…Meese and Rogoff's (1983aRogoff's ( , 1983b finding does not seem to be entirely and convincingly overturned." ...
... The forecast error when forecasting with the ./0 is given by using a one-sided DMW test as before 5 . Table 2 summarizes our analysis with this more competitive benchmark in the shorter sub- sample period for which we know the exact day in which the survey was released to the public. ...
We examine the accuracy of survey-based expectations of the Chilean exchange rate relative to the US dollar. Our out-of-sample analysis reveals that survey-based forecasts outperform the Driftless Random Walk (DRW) in terms of Mean Squared Prediction Error at several forecasting horizons. This result holds true even when comparing the survey to a more competitive benchmark based on a refined information set. A similar result is found when precision is measured in terms of Directional Accuracy: survey-based forecasts outperform a "pure luck" benchmark at several forecasting horizons. Differing from the traditional "no predictability" result reported in the literature for many exchange rates, our findings suggest that the Chilean peso is indeed predictable.
... We investigated whether the mean squared prediction errors (MSPEs) between two models had statistically significant differences. There are many statistical tests for testing out-of-sample prediction, such as the Diebold and Mariano test (Diebold and Mariano (1995)), the West test (West 1996), and the Clark and West test (Clark and West 2006). The models in this study were the nested model of random walk, following Engel et al. (2007), and we used the Clark and West test proposed by Clark and West (2006). ...
... There are many statistical tests for testing out-of-sample prediction, such as the Diebold and Mariano test (Diebold and Mariano (1995)), the West test (West 1996), and the Clark and West test (Clark and West 2006). The models in this study were the nested model of random walk, following Engel et al. (2007), and we used the Clark and West test proposed by Clark and West (2006). This test investigated whether the MSPEs of two models (Model 1 and Model 2) were equivalent or not. ...
This paper investigates the predictability of exchange rate changes by extracting the factors from the three-, four-, and five-factor model of the relative Nelson–Siegel class. Our empirical analysis shows that the relative spread factors are important for predicting future exchange rate changes, and our extended model improves the model fitting statistically. The regression model based on the three-factor relative Nelson–Siegel model is the superior model of the extended models for three-month-ahead out-of-sample predictions, and the prediction accuracy is statistically significant from the perspective of the Clark and West statistic. For 6- and 12-month-ahead predictions, although the five-factor model is superior to the other models, the prediction accuracy is not statistically significant.
... In order to arrive at a 2 week forecast for volatility, after transforming the forecasted, daily, log-volatility values The table lists the root mean-squared error (RMSE) multiplied by 100 of the in-sample predictions of the models as well as the R 2 of a Mincer-Zarnowitz regression Mincer and Zarnowitz (1969) in percentages. Using the forecast evaluation test of Clark and West (2006) for nested models, the RMSE of Models 1-5 can be tested whether on whether they result in smaller RMSE than the one of Model 0, our benchmark model. ...
... Patton 2011). Using the forecast evaluation test of Clark and West (2006) for nested models, the RMSE of Models 1-5 can be tested whether they result in a smaller RMSE than Model 0, our benchmark model. One star indicates that the null hypothesis (that the RMSE of the benchmark Model is smaller) can be rejected on a 10% significance level, two stars signify rejection on the 5% significance level. ...
Available at SSRN: https://ssrn.com/abstract=3126324
Published version: https://doi.org/10.1016/j.irfa.2019.03.003
We evaluate the usefulness of Google search volume to predict returns and volatility of multiple cryptocurrencies. The analysis is based on a new algorithm which allows to construct mulit-annual, consistent time series of Google search volume indices (SVIs) on various frequencies. As cryptocurrencies are actively traded on a continuous basis and react very fast to new information, we conduct the analysis initially on a daily basis, lifting the data imposed restriction faced by previous research. In line with the literature on financial markets, we find that returns are not predictable while volatility is predictable to some extent. We discuss a number of reasons why the predictive power is poor. One aspect is the observational frequency which is therefore varied. The results of unpredictable cryptocurrency returns holds on higher (hourly) and lower (weekly) frequencies. Volatility, in contrast, is predictable on all frequencies and we document an increasing accuracy of the forecast when the sampling frequency is lowered.
... estimator Ω m,h , such as Newey and West (1987). 11 The test statistic for coverage test is defined as: ...
... Define the length loss as: 11 We use Newey and West (1987) for our empirical work, with a window width of 12. ...
... In the online Appendix, we use economic metrics for forecast evaluation. To assess the statistical significance of the differences in the forecasts, many papers employ the Diebold and Mariano (1995) and West (1996) tests (hereafter DMW) and/or the Clark andWest (2006, 2007) test (hereafter CW). The DMW tests whether two competing forecasts are identical under general conditions ( Diebold, 2015). ...
... The CW tests whether the benchmark model is equivalent to the competing model in population. However, Clark and West (2006) show that when comparing nested models, the DMW test is undersized; hence, the RMSFE differential should be adjusted by a term that accounts for the bias introduced by the larger model. On the other hand, Rogoff and Stavrakeva (2008) make the case for using the bootstrapped DMW test, instead of the CW test, arguing that the latter does not always test for a minimum mean squared forecast error. ...
In a unified framework, we examine four sources of uncertainty in exchange rate forecasting models: (i) random variations in the data, (ii) estimation uncertainty, (iii) uncertainty about the degree of time-variation in coefficients, and (iv) uncertainty regarding the choice of the predictor. We find that models which embed a high-degree of coefficient variability yield forecast improvements at horizons beyond 1-month. At the 1-month horizon, and apart from the standard variance implied by unpredictable fluctuations in the data, the second and third sources of uncertainty listed above are key obstructions to predictive ability. The uncertainty regarding the choice of the predictors is negligible. This article is protected by copyright. All rights reserved
... average R 2 from these estimations across all years. In the spirit of Campbell and Thompson (2008) and Clark and West (2006), if the realized growth rate series is truly unpredictable, then in a finite sample the predictive regression will on average have a higher mean squared prediction error. Therefore, the expected R 2 under the null of unpredictability is negative, and a 0 or positive R 2 can be interpreted as evidence of predictability. ...
Long-term growth expectations are central to investment analysis and corporate valuation. Despite a dominant effect on firm value, the academic literature and practitioner conventions provide little guidance on determining this long-term growth rate. This article takes a step in addressing this gap: we estimate the relationship between long-term growth and an extensive selection of firm, industry, and market characteristics. Market prices do not seem to fully capture long-term growth information. Cross sectional tests yield substantial positive abnormal returns for firms with high expected long-term growth.
... However, these effects typically diminish over the long run. Furthermore, Clark and West (2005) provide evidence that volatility tends to diminish over time, raising questions about its lasting influence on economic results. Thus, it can be observed that fluctuations in exchange rates can have a notable impact on economic activity in the short run. ...
This study investigates the influence of exchange rate fluctuations on the informal economy within the BRICS countries from 2010 to 2019. Employing linear and nonlinear analytical methods, this research utilizes linear autoregressive distributed lag (ARDL) models to analyze the relationship between exchange rate variations and the shadow economy in each BRICS nation. The study consistently reveals significant evidence of the impact of exchange rate fluctuations on the informal economy across both short-term and long-term timeframes. Moreover, employing a nonlinear ARDL model uncovers distinct and uneven effects of exchange rate volatility among the BRICS nations, emphasizing the varied characteristics of currency fluctuations in the clandestine economy. The findings underscore the need for BRICS member states to develop and implement tailored fiscal strategies to mitigate the risks associated with exchange rate volatility. Recognizing these dynamics is crucial for policymakers and stakeholders to effectively address the unequal ramifications of the informal sector within each country. Considering these findings, policymakers must devise economic policies that account for the diverse attributes of each nation’s informal economy. Implementing measures to reduce and manage the impacts of exchange rate fluctuations can foster stability and resilience within the informal sector, ultimately contributing to broader economic development objectives. This study contributes to the existing literature by employing linear and nonlinear methodologies to explore the relationship between currency exchange rates and the informal economy within the BRICS context. The identification of varied impacts across nations underscores the importance of tailored policy responses to address the unique challenges of exchange rate volatility on informal economic activities.
... Nikolsko-Rzhevskyy & Prodan (2012) examined the recent success of modern macroeconomic models in forecasting nominal exchange rates by evaluating the Clark and West (2006) inference procedure. They model the drift term using the two-state Markov-switching stochastic segmented trend. ...
This study investigates the Impact of Exchange Rate Regime Change on Non-oil Export in Nigeria from 1985Q1-2018Q4. The paper employs Markov Regime-Switching Approach to capture the exchange rate regime change. Results of the study reveal that exchange rate regimes have a positive and statistically significant impact on the non-oil export in Nigeria. The results also show that the degree of the impact in the two regimes are not the same, it is higher in regime 2. This is as a result of the fact that prior to the 1980s; Nigeria adopted the fixed exchange rate regime which does not allow for domestic currency devaluation hence, the relatively low non-oil export. However, in the 1980s, especially at the introduction of the structural adjustment programme of the 1986 which saw most economies of the world switch over to the floating exchange rates regime, the Naira is often time devalued and hence the higher increase in the non-oil exports. Therefore, the study recommends that there is need to JOURNAL OF ACADEMIC RESEARCH IN ECONOMICS 108 VOLUME 16 NUMBER 1 MARCH 2024 achieve a stable exchange rate that when combined with the export-oriented policy will promote non-oil exports in Nigeria.
... This is also known as the Theil's U statistic. The second method is the statistic produced by Clark and West (2006) and Clark and West (2007), which allows for the comparison of forecasts produce by nested models in terms of whether the difference between two forecasts for the same forecasting period is statistically significant and whether or not the improvement is statistically significant (one forecast being "better" than another). ...
This paper contributes to the literature as the first work of its kind to examine the role and importance of Divisia monetary aggregates and concomitant User Cost Price indices as superior monetary policy forecasting tools in a negative interest rate environment. We compare the performance of Di-visia monetary aggregates with traditional simple-sum aggregates in several theoretical models and in a Bayesian VAR to forecast the exchange rates between the euro, the dollar and yuan at various horizons using quarterly data. We evaluate their performance against that of a random walk using two criteria: Root Mean Square Error ratios and the Clark-West statistic. We find that, under a free-floating exchange regime, superior Divisia monetary aggregates outperform their simple sum counterparts and the benchmark random walk in a negative interest rate environment, consistently. *
... In line with Clark and West (2006) and Meng and Liu (2019), we also employ other wildly used forecasting evaluation method, out-of-sample R 2 (R oos 2 ), which scholars and policy makers usually apply for financial forecasting. The related statistic R oos 2 is given as follows: Table 6 The results for the long-term forecasting. ...
Based on the previous studies that Markov-type GARCH models exhibit inconsistent predictive ability over different horizons, we conduct the improvement of predictive power of renewable energy stock volatility by developing Markov switching GARCH-MIDAS models both in short- and long-terms. By using various out-of-sample tests, the models allowing for regime-switching in the short- and long-volatility components simultaneously outperform other competing models for short-term forecasting. However, the empirical results show that the long-term Markov regime-switching plays a more significant role on the predictive accuracy at longer horizon. Our novel findings indicate that it is necessary to adopt the appropriate predictive models that include short-term, long-term, or both of the above terms in regime-switching. Meanwhile, our extended models indeed provide a more detailed picture of the dynamic behavior over time in renewable energy stock market. Finally, our findings reveals that the governments should adopt a combination of short- and long-term policies when considering the different role of regime shift over different horizons on volatility prediction of the renewable energy stock.
... b σ 2 is a HAC estimator of σ 2 and qðPÞ is a bandwidth that grows with P (Newey & West, 1987). Following Giacomini and Rossi's propose, we substitute F t;m with the test statistic suggested by Clark and West (2006). The null hypothesis is rejected against the one-sided ...
We investigate the impact of the global economic policy uncertainty (GEPU) on stock volatility for nine emerging economies (Brazil, Russia, India, China, South Africa, Mexico, Indonesia, South Korea, and Turkey). We employ an expanded GARCH-MIDAS approach to connect low-frequency GEPU data and high-frequency stock data, assuming that GEPU affects stock volatility via the long-run component of total volatility. We not only use DM and SPA tests to statically evaluate the out-of-sample forecasting ability of the extended model, taking traditional GARCH model and GARCH-MIDAS model as benchmarks, but also use the Fluctuation test to examine the time-varyingly relative forecasting performance in the presence of potential instability. From the in-sample estimation results, we find that GEPU has empirically significant impact on stock volatility for the nine emerging economies. The out-of-sample forecasting results show that the GEPU-based model can improve forecasting performance of stock volatility for emerging markets, especially in unstable environments.
... Taking guidance from Forni and Gambetti (2014), we constructed a factor by extracting the first principal component from FRED-QD, and tested whether this extracted factor Granger cause any of the 23 BVAR equations in an out-of-sample forecasting exercise. Using the procedure described by Clark and West (2006) to test for predictability in nested models, we did not find evidence that the extracted factor from the FRED-QD dataset Granger causes any of our VAR variables in an out-of-sample forecasting exercise, suggesting our 23 variable BVAR system is informational sufficient. ...
... The precision of forecasting models can be validated using the model precision analysis (MPA). 66 The statistical criteria such as the mean absolute error (MAE), the mean absolute percentage error (MAPE), and the root mean square error (RMSE) are evaluated to measure the prediction accuracy of forecasting models. 67 A brief description of these performance indexes are described in Section 1.1, Appendix. ...
The ever‐increasing demand for green energy sources encourages wind farms to participate in the energy market through long‐term contracts. This paper discusses the existing frameworks of the wind energy market and presents the strategies to integrate wind energy into the grid through the valuable power purchase agreement (PPA). Critical observations reveal that the opportunities of the short‐term energy market remain unexplored for wind farms till date. To promote more investments and to enhance the operating economy, the wind energy sector must adapt to the sporadic nature of generation in the day‐ahead energy market. The selection of accurate PPA for wind farms is a potent area of research and has encouraged the authors to explore this area vividly. In view of this, the main objective of the article is to develop a day‐ahead sleeved PPA model and to evaluate its effectiveness in the Indian energy market. To correlate the proposed model with practicality aspects, seasonal wind speed scenarios are forecasted using the sARIMA model for a practical wind farm. The performance of the employed sARIMA model is evaluated through proper comparative assessment with other models. Thereafter to validate the importance of the developed PPA, a detailed comparative analysis is carried out with the existing long‐term PPA. Further, the potentiality of the proposed sleeved PPA is compared with its other counterparts. Result analysis confirms the usefulness of the proposed PPA model in maximizing the economy of the renewable energy market entities.
... This variable gives 6 In terms of one period returns, we have 212 monthly observations and 70 quarterly observations. 7 Examples in the exchange rate literature are given by Meese and Rogo¤ (1983) and Clark and West (2006). When forecasting commodity prices, Chen, Rogo¤ (2010, 2014) consider the RW and an AR(1) as benchmarks, Lof and Nyberg (2017) consider both a causal and noncausal AR(1) whereas Groen and Pesenti (2011) use autoregressions with more lags, but autoregressions in the end. ...
In this paper we show that the Chilean exchange rate has the ability to predict the returns of the London Metal Exchange Index and of the six primary non-ferrous metals that are part of the index: aluminum, copper, lead, nickel, tin and zinc. The economic relationship hinges on the present-value theory for exchange rates, a floating exchange rate regime and the fact that copper represents about a half of Chilean exports and nearly 45% of Foreign Direct Investment. Consequently, the Chilean peso is heavily affected by fluctuations in the copper price. As all six base metal prices show an important comovement, we test whether the relationship between copper prices and Chilean exchange rates also holds true when it comes to the six primary non-ferrous metals. We find interesting evidence of predictability both in-sample and out-of-sample with traditional statistical measures. We also show that the information contained in the Chilean peso can successfully be used to obtain positive returns when trading base metals. Our paper is part of a growing literature that in the recent years has evaluated the ability of commodity currencies to forecast commodity prices.
... Similar results are found by Cheung et al. (2005) and Alquist and Chinn (2008). Slightly more positive findings have been reported by Clark and West (2006) over short-term horizons, and Molodtsova and Papell (2009). According to McCallum (1994), one reason for these rejections could be that the monetary policy behavior is inconsistent with UIP. ...
... Similar results are found by Cheung et al. (2005) and Alquist and Chinn (2008). Slightly more positive findings have been reported by Clark and West (2006) over short-term horizons, and Molodtsova and Papell (2009). According to McCallum (1994), one reason for these rejections could be that the monetary policy behavior is inconsistent with UIP. ...
The concept of factor investing emerged at the end of the 2000s and has completely changed the landscape of equity investing. Today, institutional investors structure their strategic asset allocation around five risk factors: size, value, low beta, momentum and quality. This approach has been extended to multi-asset portfolios and is known as the alternative risk premia model. This framework recognizes that the construction of diversified portfolios cannot only be reduced to the allocation policy between asset classes, such as stocks and bonds. Indeed, diversification is multifaceted and must also consider alternative risk factors. More recently, factor investing has gained popularity in the fixed income universe, even though the use of risk factors is an old topic for modeling the yield curve and pricing interest rate contingent claims. Factor investing is now implemented for managing portfolios of corporate bonds or emerging bonds. In this paper, we focus on currency markets. The dynamics of foreign exchange rates are generally explained by several theoretical economic models that are commonly presented as competing approaches. In our opinion, they are more complementary and they can be the backbone of a Fama-French-Carhart risk factor model for currencies. In particular, we show that these risk factors may explain a significant part of time-series and cross-section returns in foreign exchange markets. Therefore, this result helps us to better understand the management of forex portfolios. To illustrate this point, we provide some applications concerning basket hedging, overlay management and the construction of alpha strategies.
... The rolling window scheme allows the test to have a non-time variant critical value. For more details about the test implemented in case of Clark and West (2006), the null hypothesis, as well as the critical values obtained using a Monte Carlo experiment, see Giacomini and Rossi (2010). 15 In the figures we report only the upper bound since the statistics never cross the lower bound. ...
We assess the importance of parameter instabilities from a forecasting viewpoint in a set of medium-scale DSGE models with and without financial frictions using US real-time data. We find that, first, failing to update DSGE model parameter estimates with new data arrival deteriorates point forecasts due to the estimated parameters variation. And second, the presence of financial frictions helps to better address, city, state, ZIP code, province, country with country codes for all authors.
... , change is possibly non monotonic but not necessarily symmetric. Giacomini and Rossi's (2010) Fluctuation test statistics (in absolute value) implemented using the Clark and West's (2006) statistics for comparing forecasts of the Non-linear Taylor rule exchange rate model relative to the Linear Taylor rule exchange rate models as the benchmark (solid line). The dashed line denotes the one-sided 5% critical value of the Fluctuation test statistic. ...
... , change is possibly non monotonic but not necessarily symmetric. Giacomini and Rossi's (2010) Fluctuation test statistics (in absolute value) implemented using the Clark and West's (2006) statistics for comparing forecasts of the Non-linear Taylor rule exchange rate model relative to the Linear Taylor rule exchange rate models as the benchmark (solid line). The dashed line denotes the one-sided 5% critical value of the Fluctuation test statistic. ...
This research utilises a non-linear Smooth Transition Regression (STR) approach to modelling and forecasting the exchange rate, based on the Taylor rule model of exchange rate determination. The separate literatures on exchange rate models and the Taylor rule have already shown that the non-linear specification can outperform the equivalent linear one. In addition the Taylor rule based exchange rate model used here has been augmented with a wealth effect to reflect the increasing importance of the asset markets in monetary policy. Using STR models, the results offer evidence of non-linearity in the variables used and that the interest rate differential is the most appropriate transition variable. We conduct the conventional out-of-sample forecasting performance test, which indicates that the non-linear models outperform their linear equivalents as well as the non-linear UIP model and random walk.
... The values of R 2 oos are calculated as the percent reduction of mean squared predictive error (MSPE) of the combination strategy of interest relative to the MSPE of the benchmark model of historical average. p-values are for Clark and West (2006) test for the null hypothesis that the MSPE of the strategy of interest is higher than or equal to the benchmark model. Notes: This table reports the annualized average returns of the buy/sell trading rules constructed based on the sign of return forecasts obtained from combination strategies. ...
Testing the out-of-sample return predictability is of great interest among academics. A wide range of studies have shown the predictability of stock returns, but fail to test the statistical significance of economic gains from the predictability. In this paper, we develop a new statistical test for the directional accuracy of stock returns. Monte Carlo experiments reveal that our bootstrap-based tests have more correct size and better power than the existing tests. We use the forecast combinations and find that the stock return predictability is statistically significant in terms of reduction of mean squared predictive error relative to the benchmark of historical average forecasts. However, the results from our tests show that the predictability is not economically significant. We conclude that there will be still a long way to go for forecasting stock returns for market participants.
... 6 In terms of one period returns, we have 189 monthly observations. 7 Examples include the articles by Meese and Rogo¤ (1983), Clark and West (2006), Chenn, Rossi and Rogo¤ (2010, 2014), Lof and Nyberg (2017), Groen and Pesenti (2011), Buncic and Moretto (2015) and Goyal and Welch (2008) to mention a few. ...
In this paper we show that expectations about the future evolution of the Chilean exchange rate have the ability to predict the returns of the six primary non-ferrous metals: aluminum, copper, lead, nickel, tin and zinc. Predictability is also found for returns of the London Metal Exchange Index. Previous studies have shown that the Chilean exchange rate has the ability to predict copper returns, a world commodity index and base metal prices. Nevertheless, our results indicate that expectations about the Chilean peso have stronger predictive ability relative to the Chilean currency. This is shown both in-sample and out-of-sample. By focusing on expectations of a commodity currency, and not on the currency itself, our paper provides indirect but new evidence of the ability that commodity currencies have to forecast commodity prices.
... Nikolsko-Rzhevskyy and Prodan argue that the drift term and the fact that a simple two state Markov switching random walk (MSRW) model is a good representation of the nominal exchange rate leads to the relative success of macroeconomic fundamental models [13]. In order to evaluate the out-of-sample performance of the Markov switching model versus the random walk, they apply the inference procedure proposed by Clark and West for testing the null hypothesis of equal predictive ability for the two nested models [14]. They model the drift term using a two state Markov switching stochastic segmented trend model and present evidence of both short run, one month and long run, up to one year, predictability for monthly exchange rates over the post-Bretton Woods periods. ...
This paper extends the traditional monetary model and the random walk model with Markov-switching method and proposes two new forecasting models called Markov switching monetary model (MSMM) and Markov switching random walk model (MSRW). Compared with the traditional models, the evidence shows the two hybrid models perform better on forecasting the exchange rate.
... ,Taylor et al. (2001),Kilian and Taylor (2003),Clark and West (2006), Engel et al. (2007),Alquist and Chinn (2008),Molodtsova and Papell (2009), and Wang and ...
... Our in-sample and out-of-sample analyses at the monthly frequency are based on the following simple speci…cations: 5 In terms of one period returns, we have 212 monthly observations and 70 quarterly observations. 6 Examples in the exchange rate literature are given by Meese and Rogo¤ (1983) and Clark and West (2006) for instance. When forecasting commodity prices, Chenn, Rossi and Rogo¤ (2010, 2014) consider the RW and an AR(1) as benchmarks, Lof and Nyberg (2017) consider both a causal and noncausal AR(1) whereas Groen and Pesenti (2011) uses autoregressions with more lags, but autoregressions in the end. ...
In this paper we show that the Chilean exchange rate has the ability to predict the returns of the London Metal Exchange Index and of the six primary non-ferrous metals that are part of the index: aluminum, copper, lead, nickel, tin and zinc. The economic relationship hinges on the present-value theory for exchange rates, a floating exchange rate regime and the fact that copper represents about a half of Chilean exports and nearly 45% of Foreign Direct Investment. Consequently, the Chilean peso is heavily affected by fluctuations in the copper price. As all six base metal prices show an important comovement, we test whether the relationship between copper prices and Chilean exchange rates also holds true when it comes to the six primary non-ferrous metals. We find interesting evidence of predictability both in-sample and out-of-sample. Our paper is part of a growing literature that in the recent years has evaluated the ability of commodity currencies to forecast commodity prices.
... Table 10 present Monte Carlo simulations of the out-of-sample adjusted MSPE (Mean Squared Prediction Error), also known as the Clark-West statistic. Clark and West (2006) provide an adjustment to the difference in MSPE between two encompassing models that leads to increased power. The adjustment is the mean difference between the forecast error squared between two models (Y F 1-Y F 2) 2 /N, where Y F 1 denotes the forecast of model one (the full model), and Y F 2 the forecast of model two (the restricted model). ...
... In contrast, however, others suggest that models can outdo the random walk method over long horizons, but not for short horizons (Clark & West, 2006;Peel, & Sarno, 2001;Molodtsova & Papell, 2009;Mark & Sul, 2001;Taylor, Kilian & Taylor, 2003;Groen, 2000;Mark, 1995;Chinn & Meese, 1998;La Cour & MacDonald, 2000;Alquist & Chinn, 2007;Wang, 2012). ...
Numerous researchers have studied the connection between exchange rate fluctuations and macroeconomic variables for various market economies. Few studies, however, have addressed whether these relationships may differ based on the market classification of the given economy. This study examined the impact on exchange rates for Japan (a proxy for developed economies) and South Korea (a proxy for emerging economies) yielding from the macroeconomic variables of the sticky-price monetary model between February 1, 1989 and February 1, 2015. The results show that money supply and inflation constituted a significant, but small, influence on South Korean exchange rate movements, whereas no macroeconomic variable within the model had a significant impact on Japanese exchange rates fluctuations. The results of the autoregressive error analyses suggest small variances in the affect that macroeconomic variables may have on developed versus emerging market economies. This may provide evidence that firms may use similar forecasting techniques for emerging market currencies as used with developed market currencies.
... Inference on the statistical significance of the U-stat is based on the usual asymptotic Clark and West (2006) test. This is a test of the null hypothesis of equal RMSFEs between our unrestricted FbFM and the restricted RW, after adjusting the MSFE of the FbFM to account for the noise that it introduces into the forecasting process ( Clark & West, 2006). The direction of change statistic, which is also known as the success or hit rate, captures the ability of the FbFM to time the market; we use it as a complement to the Ustat metric. ...
Research shows that the predictive ability of economic fundamentals for exchange rates varies over time; it may be detected in some periods and disappear in others. This paper uses bootstrap-based methods to select time-specific conditioning information for the prediction of exchange rates. By employing measures of the predictive ability over time, along with statistical and economic evaluation criteria, we find that our approach based on pre-selecting and validating fundamentals across bootstrap replications leads to significant forecast improvements and economic gains relative to the random walk. The approach, known as bumping, selects parsimonious models that have out-of-sample predictive power at the one-month horizon; it is found to outperform various alternative methods, including Bayesian, bagging, and standard forecast combinations.
El propósito de este artículo es evaluar la capacidad predictiva que tiene la herramienta de Google Trends para predecir los tipos de cambio de los países que componen el mercado MILA: Chile, Perú, Colombia y México. El argumento económico y financiero detrás de esta investigación es que la frecuencia de búsqueda que entrega Google Trends es un proxy de las expectativas que tienen los agentes en el mercado sobre el rendimiento en la economía. Se utiliza el Diccionario de Harvard para recono-cer palabras con connotación positiva y negativa, respecto al estado de la economía de cada país. Se realizan análisis dentro y fuera de la muestra para evaluar la capacidad predictiva de Google Trends. Además, existe una reciente literatura que utiliza esta herramienta para pronósticos de tipo de cambio en los países de Japón y Estados Unidos, pero poco se conoce en economías latinoamericanas de países MILA. Consistente con evidencia previa, los resultados de predictibilidad son inestables y poco robustos. Este estudio es importante para inversionistas, portfolio manager o agentes que deben ges-tionar el riesgo del tipo de cambio.
In this paper we show that the Chilean exchange rate has the ability to predict the returns of oil and of three additional oil-related products: gasoline, propane and heating oil. We show this using both in- and out-of sample exercises at multiple horizons. Natural explanations for our findings rely on the well know “dollar effect” and on the present-value theory for exchange rate determination in combination with the strong co-movement displayed by fuel and metal prices. Given that the Chilean economy is heavily influenced by copper, which represents nearly 50% of total national exports, the floating Chilean Peso is importantly affected by price fluctuations in this metal. As oil-related products display an important co-movement with base metal prices, it is reasonable to expect evidence of Granger causality from the Chilean peso to these oil-related products. Interestingly, we provide sound evidence indicating that the predictive ability of the Chilean Peso goes beyond these natural explanations. In particular, we show another plausible predictive channel: volatility in combination with a negative contemporaneous leverage effect in fuel returns. Finally, we compare the Chilean peso with other commodity-currencies in their ability to predict fuel returns. The Chilean peso fares extremely well in this competition, especially at short horizons of one, three and six months.
Several Organisation for Economic Co‐operation and Development countries have recently implemented an automatic link between the statutory retirement age and life expectancy for the total population to ensure sustainability in their pension systems due to increasing life expectancy. As significant mortality differentials are observed across socio‐economic groups, future changes in these differentials will determine whether some socio‐economic groups drive increases in the retirement age, leaving other groups with fewer pensionable years. We forecast life expectancy by socio‐economic groups and compare the forecast performance of competing models by using Danish mortality data and find that the most accurate model assumes a common mortality trend. Life expectancy forecasts are used to analyse the consequences of a pension system where the statutory retirement age is increased when total life expectancy is increasing.
Motivated by the common finding that linear autoregressive models forecast better than models that incorporate additional information, this paper presents analytical, Monte Carlo, and empirical evidence on the effectiveness of combining forecasts from nested models. In our analytics, the unrestricted model is true, but as the sample size grows, the data generating process converges to the restricted model. This approach captures the practical reality that the predictive content of variables of interest is often low. We derive MSE-minimizing weights for combining the restricted and unrestricted forecasts. Monte Carlo and empirical analyses verify the practical effectiveness of our combination approach.
Precision agriculture is the new tendency of agricultural development all around the world today. During the implementation process, precision agriculture is been required to collect information on crop diseases, pests and fertilizers at any time. The acquisition of this information depends on the precise position information. Considering the problem that the positioning accuracy of the traditional GPS system is low or the other signal needs to be sent to correct the position, a GPS position based on neural network is designed to correct the precision problem. The GPS module was used to receive GPS signal, and the position information was extracted by the MCU. Then, the real-time location information is displayed on the screen. The back propagation neural network was used to generate a prediction model of error value between the measured data and real data. This model can predict and compensate the errors of measured values. Finally, the measured data and the corrected data are shown on the screen. The precision of GPS positioning designed in this paper is 10 times higher than that of traditional GPS, meeting requirement with high precision of the information-based agriculture.
This study explores the possibilities of applying Google Trends to exchange rate forecasting. Specifically, we use Google Trends to capture market sentiment in Japan and the United States and construct a sentiment index. We forecast the one-month-ahead USD/JPY rates using three structural models and two autoregressive models and examine whether our sentiment index can improve the predictive power of these models. The data we use run from as the forecast sample. We find that the addition of the sentiment index into the autoregressive models decreases the mean squared prediction error. We also test the sentiment indices of different word numbers and find that the 25-and 30-word indices perform best; in particular, the 30-word index improves all the models tested in this study.
This paper examines short‐horizon exchange rate predictability and investigates whether stock returns contain information for forecasting daily exchange rate movements. Inspired by the uncovered equity parity condition, we show that stock return differentials have in‐ sample and out‐of‐sample predictive power for nominal exchange rates with short horizons (one‐day‐ahead predictions). That is, stock markets inform us about exchange rate move‐ ments, at least in the case of high‐frequency data.
This paper investigates robust model rankings in out‐of‐sample, short‐horizon forecasting. We provide strong evidence that rolling window averaging consistently produces robust model rankings while improving the forecasting performance of both individual models and model averaging. The rolling window averaging outperforms the (ex post) “optimal” window forecasts in more than 50% of the times across all rolling windows.
This paper examines the usefulness of the labor market conditions index (LMCI) in forecasting key labor market variables, particularly unemployment rates. Using a number of models, we compare out-of-sample forecasts of the unemployment rate with the LMCI to those without the LMCI. We also estimate models of the disaggregated unemployment rates by gender, race, and race by gender, with and without the LMCI, to identify disparities in the predictive power of the LMCI for different subgroups. Last, to determine how the LMCI performs in the presence of labor market shocks, we compare the forecasting performance of the LMCI during recessionary periods and expansionary periods. Our results confirm the potential usefulness of the LMCI as a parsimonious forecasting tool; we find that the LMCI generally improves unemployment forecasts. But, disparities exist in the predictive power of the index across subpopulations and the index forecasts slightly better during recessionary periods than expansionary periods.
We show that high-frequency private payroll microdata can help forecast labor market conditions. Payroll employment is perhaps the most reliable real-time indicator of the business cycle and is therefore closely followed by policymakers, academia, and financial markets. Government statistical agencies have long served as the primary suppliers of information on the labor market and will continue to do so for the foreseeable future. That said, sources of “big data” are becoming increasingly available through collaborations with private businesses engaged in commercial activities that record economic activity on a granular, frequent, and timely basis. One such data source is generated by the firm ADP, which processes payrolls for about one fifth of the U.S. private sector workforce. We evaluate the efficacy of these data to create new statistics that complement existing measures. In particular, we develop a set of weekly aggregate employment indexes from 2000 to 2017, which allows us to measure employment at a higher frequency than is currently possible. The extensive coverage of the ADP data—similar in terms of private employment to the BLS CES sample—implies potentially high information value of these data, and our results confirm this conjecture. Indeed, the timeliness and frequency of the ADP payroll microdata substantially improves forecast accuracy for both current-month employment and revisions to the BLS CES data.
In this paper, we use Google Trends data for exchange rate forecasting in the context of a broad literature review that ties the exchange rate movements with macroeconomic fundamentals. The sample covers 11 OECD countries’ exchange rates for the period from January 2004 to June 2014. In out-of-sample forecasting of monthly returns on exchange rates, our findings indicate that the Google Trends search query data do a better job than the structural models in predicting the true direction of changes in nominal exchange rates. We also observed that Google Trends-based forecasts are better at picking up the direction of the changes in the monthly nominal exchange rates after the Great Recession era (2008–2009). Based on the Clark and West inference procedure of equal predictive accuracy testing, we found that the relative performance of Google Trends-based exchange rate predictions against the null of a random walk model is no worse than the purchasing power parity model. On the other hand, although the monetary model fundamentals could beat the random walk null only in one out of 11 currency pairs, with Google Trends predictors we found evidence of better performance for five currency pairs. We believe that these findings necessitate further research in this area to investigate the extravalue one can get from Google search query data.
ResearchGate has not been able to resolve any references for this publication.