Random Walk Benchmark Research Articles

Forecasting transportation demand for the U.S. market

In this paper we forecast air, road and train transportation demand for the U.S. domestic market based on econometric and machine learning methodologies. More specifically, we forecast transportation demand for various horizons up to 18 months ahead, for the period 2000:1–2015:03, employing, from the domain of machine learning, a Support Vector Regression (SVR) and from econometrics, the Least Absolute Shrinkage and Selection Operator and the Ordinary Least Squares regression. In doing so, we follow the relevant literature and consider the contribution of selected variables as potential regressors in forecasting. Our empirical findings suggest that while all models outperform the Random Walk benchmark, the machine learning applications adhere more closely to the data generating process, producing more accurate out-of-sample forecasts than the classical econometric models. In most cases, we find that the transportation demand is driven by fuel costs, except for road transportation where macroeconomic conditions affect transportation volumes only for specific forecasting horizons. This finding deviates from the existing literature, given the support of previous studies to macroeconomic conditions are driving factors of transportation demand. Our work relates directly to decisions on transport infrastructure improvement, while it can also be used as a forecasting tool in shaping transportation-oriented policies.

Metropolitan Hotel Sector Forecast Accuracy in El Paso

This study evaluates the accuracy of previously published econometric forecasts for seven lodging sector variables that measure hotel activity in El Paso, Texas. The hotel forecasts have been generated annually using an econometric model of the El Paso metropolitan economy from 2006 forward. Predictive accuracy is evaluated relative to random walk benchmarks. Assessment is completed using both descriptive forecast error summary statistics as well as formal statistical tests. The econometric model outperforms the random walk benchmarks for a majority of the variables analyzed. However, statistical tests of forecast error differentials do not yield conclusive evidence in favor of the econometric historical track record. Tests of directional forecast accuracy also produce mixed results. Although the structural econometric model of hotel business conditions appears to provide useful predictive information, analysts and planners should also monitor recent history closely.

Bank Net Interest Margin Forecasting and Capital Adequacy Stress Testing by Machine Learning Techniques

The 2007-09 financial crisis revealed that the investors in the financial market were more concerned about the future as opposed to the current capital adequacy for banks. Stress testing promises to complement the regulatory capital adequacy regimes, which assess a bank's current capital adequacy, with the ability to assess its future capital adequacy based on the projected asset-losses and incomes from the forecasting models from regulators and banks. The effectiveness of stress-test rests on its ability to inform the financial market, which depends on whether or not the market has confidence in the model-projected asset-losses and incomes for banks. Post-crisis studies found that the stress-test results are uninformative and receive insignificant market reactions; others question its validity on the grounds of the poor forecast accuracy using linear regression models which forecast the banking-industry incomes measured by Aggregate Net Interest Margin. Instead, our study focuses on NIM forecasting at an individual bank's level and employs both linear regression and non-linear Machine Learning techniques. First, we present both the linear and non-linear Machine Learning regression techniques used in our study. Then, based on out-of-sample tests and literature-recommended forecasting techniques, we compare the NIM forecast accuracy by 162 models based on 11 different regression techniques, finding that some Machine Learning techniques as well as some linear ones can achieve significantly higher accuracy than the random-walk benchmark, which invalidates the grounds used by the literature to challenge the validity of stress-test. Last, our results from forecast accuracy comparisons are either consistent with or complement those from existing forecasting literature. We believe that the paper is the first systematic study on forecasting bank-specific NIM by Machine Learning Techniques; also, it is a first systematic study on forecast accuracy comparison including both linear and non-linear Machine Learning techniques using financial data for a critical real-world problem; it is a multi-step forecasting example involving iterative forecasting, rolling-origins, recalibration with forecast accuracy measure being scale-independent; robust regression proved to be beneficial for forecasting in presence of outliers. It concludes with policy suggestions and future research directions.

Non-Linear Interactions and Exchange Rate Prediction: Empirical Evidence Using Support Vector Regression

ABSTRACTThis paper analysed the prediction of the spot exchange rate of 10 currency pairs using support vector regression (SVR) based on a fundamentalist model composed of 13 explanatory variables. Different structures of non-linear dependence introduced by nine different Kernel functions were tested and the predictions were compared to the Random Walk benchmark. We checked the explanatory power gain of SVR models over the Random Walk by applying White’s Reality Check Test. The results showed that the majority of SVR models achieved better out-of-sample performance than the Random Walk, but in overall they failed to achieve statistical significance of predictive superiority. Furthermore, we observed that non-mainstream Kernel functions performed better than the ones commonly used in the machine-learning literature, a finding that can provide new insights regarding machine-learning methods applications and the predictability of exchange rates using non-linear interactions between the predictors.

Forecasting Energy Commodity Prices: A Large Global Dataset Sparse Approach

This paper focuses on forecasting quarterly energy prices of commodities, such as oil, gas and coal, using the Global VAR dataset proposed by Mohaddes and Raissi (2018). This dataset includes a number of potentially informative quarterly macroeconomic variables for the 33 largest economies, overall accounting for more than 80% of the global GDP. To deal with the information in this large database, we apply a dynamic factor model based on a penalized maximum likelihood approach that allows us to shrink parameters to zero and to estimate sparse factor loadings. The estimated latent factors show considerable sparsity and heterogeneity in the selected loadings across variables. When the model is extended to predict energy commodity prices up to four periods ahead, results indicate larger predictability relative to the benchmark random walk model for 1-quarter ahead for all energy commodities. In our application, the largest improvement in terms of prediction accuracy is observed when predicting gas prices from 1 to 4 quarters ahead.

Liquidity and the convergence to market efficiency

ABSTRACTThe aim of this study is to investigate the relationship between market liquidity changes on the Johannesburg Stock Exchange (JSE) and the market’s degree of efficiency. Market efficiency is characterised in terms of two philosophies: Fama’s (1970) Efficient Markets Hypothesis (EMH), and Shiller’s (1981; 2003) informational efficiency designation. Efficiency was tested using measures of return predictability, a random walk benchmark and price volatility; liquidity was measured using market turnover. The tests were conducted on JSE Top 40 shares across three regimes, spanning January 2012–June 2016. The regimes are demarcated by two structural breaks in the JSE’s microstructure: the 2012 trading platform upgrade and the 2014 colocation centre launch. The results show that past order imbalances are a significant predictor of daily returns, although the significance of this predictability has dissipated over time. Return predictability is not influenced by liquidity. In fact, there is evidence that illiquidity weakens return predictability. Prices were closer to random walk benchmarks during the third regime. In consideration of informational efficiency, during the latter two regimes price volatility is greater during trading versus non-trading hours. This is coupled with an emergence of nonlinear return dependence, which is indicative of greater mispricing. Thus, over the three regimes, market efficiency improved in the sense of the EMH, but informational efficiency deteriorated. This study contributes to the field by: introducing an inverse measure of market efficiency; providing insight into the measure’s time variation and relation to liquidity; and demonstrating that market efficiency tests should incorporate the dual meanings of the term, enabling richer understanding of how such meanings intersect.

Relative option liquidity and price efficiency

Options trading can stimulate price efficiency in underlying stock markets by providing a platform for informed trades, increasing the production of information, and mitigating momentum arbitrage strategies and short-sale constraints. Using a large sample of 8146 firms with option trading from 1996 through 2014, this study examines the extent to which liquidity in option markets relates to the ability for stock prices to reflect all publicly available information. Variance ratio tests document that price efficiency monotonically increases across relative option liquidity deciles, proxied by the option-to-stock volume ratio, progressively becoming closer to random walk benchmarks. Evidence suggests that enhanced price efficiency is pronounced among small company stocks.

The RMB central parity formation mechanism: August 2015 to December 2016

We study the renminbi (RMB) central parity formation mechanism following the August 2015 reform using statistical models. We identify the roles of the onshore and offshore RMB exchange rates and the US dollar index in determining the central parity in a linear regression framework. The effect of the RMB currency basket index, however, is revealed after controlling for multiplicative offshore RMB volatility effects. The offshore RMB volatility exerts a dampening effect on the links between the central parity and its determinants. In the prediction comparison exercise, the three selected models statistically outperform the random walk benchmark. Among these four models, the selected multiplicative specification yields the smallest root-mean squared prediction error and mean absolute prediction error.

Predicting crypto‐currencies using sparse non‐Gaussian state space models

AbstractIn this paper we forecast daily returns of crypto‐currencies using a wide variety of different econometric models. To capture salient features commonly observed in financial time series like rapid changes in the conditional variance, non‐normality of the measurement errors and sharply increasing trends, we develop a time‐varying parameter VAR witht‐distributed measurement errors and stochastic volatility. To control for overparametrization, we rely on the Bayesian literature on shrinkage priors, which enables us to shrink coefficients associated with irrelevant predictors and/or perform model specification in a flexible manner. Using around one year of daily data, we perform a real‐time forecasting exercise and investigate whether any of the proposed models is able to outperform the naive random walk benchmark. To assess the economic relevance of the forecasting gains produced by the proposed models we, moreover, run a simple trading exercise.

Exchange rate prediction redux: New models, new data, new currencies

Previous assessments of nominal exchange rate determination, following Meese and Rogoff (1983) have focused upon a narrow set of models. Cheung et al. (2005) augmented the usual suspects with productivity based models, and “behavioral equilibrium exchange rate” models, and assessed performance at horizons of up to 5 years. In this paper, we further expand the set of models to include Taylor rule fundamentals, yield curve factors, and incorporate shadow rates and risk and liquidity factors. The performance of these models is compared against the random walk benchmark. The models are estimated in error correction and first-difference specifications. We examine model performance at various forecast horizons (1 quarter, 4 quarters, 20 quarters) using differing metrics (mean squared error, direction of change), as well as the “consistency” test of Cheung and Chinn (1998). The purchasing power parity estimated in the error correction form delivers the best performance at long horizons by a mean squared error measure. Moreover, along a direction-of-change dimension, certain structural models do outperform a random walk with statistical significance. While one finds that these forecasts are cointegrated with the actual values of exchange rates, in most cases, the elasticity of the forecasts with respect to the actual values is different from unity. Overall, model/specification/currency combinations that work well in one period and one performance metric will not necessarily work well in another period and alternative performance metric.

A Synergistic Forecasting Model for High-Frequency Foreign Exchange Data

In this study, we develop a synergistic forecasting model using the information fusion approach. By using high frequency (one-minute) foreign exchange (FX) data, the model fuses two standalone models, namely the technical analysis structural model and the intra-market model. Subsequently, the outputs are fed into a unique modified extended Kalman filter whose functional parameters are estimated dynamically by using an artificial neural network. The synergistic model is tested on four currency pairs that dominate the FX market. In terms of forecasting performance, both root mean squared error and correct directional change performance results show that the synergistic model is statistically outperform and superior to each of the both standalone models as well as to the benchmark random walk model in the literature.

EXCHANGE RATE FORECAST FUTILITY FOR THE TAKA

An autoregressive distributed lag (ARDL) version of an error correction model based on a balance of payments approach is used to forecast the nominal exchange rate for the Bangladeshi taka. Based on existing trade volumes and trade practices, the bilateral exchange rate of the taka with the dollar is analyzed. Annual frequency data for a four decade period from 1976 to 2015 are utilized for the study. Descriptive statistics, formal inferential tests, and directional accuracy tests are used to assess out-of-sample forecast accuracy. Results indicate that, in spite of good insample estimation diagnostics, the model forecasts do not fare well against random walk benchmarks.

On accuracy of survey forecasts of US mortgage spread

AbstractThe cyclical variation behavior of the mortgage spread has motivated some studies to investigate its relationship to economic activity. Indeed, recent empirical findings indicate that the mortgage spread is a determinant/predictor of economic activity. We define the mortgage spread as the difference between the 30-year mortgage and 10-year Treasury rates and ask whether the Blue Chip (consensus) forecasts of these series are accurate for 1988–2015. Our findings indicate that the Blue Chip forecasts of both the 30-year mortgage and 10-year Treasury rates are biased, fail to outperform the random walk benchmark, and are directionally inaccurate. However, the Blue Chip forecasts of the mortgage spread are generally unbiased, outperform the random walk benchmark, and are directionally accurate—thus of value to a user. Given such evidence, a natural extension for future research is to explore whether the predictive information content of Blue Chip forecasts of the mortgage spread is a better predictor ...

Assessing the predictive ability of sovereign default risk on exchange rate returns

Increased sovereign credit risk is often associated with sharp currency movements. Therefore, expectations of the probability of a sovereign default event can convey important information regarding future movements of exchange rates. In this paper, we investigate the possible pass-through of risk in the sovereign debt markets to currency markets by proposing a new risk premium factor for predicting exchange rate returns based on sovereign default risk. We compute it from the term structure at different maturities of sovereign credit default swaps and conduct an out-of-sample forecasting exercise to test whether we can improve upon the benchmark random walk model. Our results show that the inclusion of the default risk factor improves the forecasting accuracy upon the random walk model and alternative models provide less accurate predictions.

OIL PRICE FORECASTS FOR THE LONG TERM: EXPERT OUTLOOKS, MODELS, OR BOTH?

Little is known about the accuracy of expert outlooks, so heavily relied upon by industry participants and policy makers, regarding the future path of oil prices. Using the regular publications by the Energy Information Administration (EIA), we examine the accuracy of annual recursive oil price forecasts generated by the National Energy Modeling System model of the Agency for forecast horizons of up to 15 years. Our results reveal that the EIA model outperforms the benchmark random walk model around the two ends of the forecast horizon spectrum. Additionally, at the longer horizons, simple econometric forecasting models often produce similar, if not better accuracy than the EIA model. Time varying such specifications generally also exhibit stability in their forecast performance. Finally, although combining forecasts does not change the overall patterns, some additional accuracy gains are obtained at intermediate horizons, and in some cases, forecast performance stability is also achieved.

Calibrating the Predictability of Exchange Rates: A Model-Independent Approach

This paper proposes a nonparametric model-independent methodology to calibrate the predictability of exchange rates. In order to predict the exchange rates, the predictors should contain enough information about the future return, regardless of the specification of the model. The information transfer from the predictors to future return can be measured by their mutual information. According to Shannon's channel coding theorem, it is the measure of the statistical dependence, linear and nonlinear, and not just the linear dependence as the correlation coefficient measures. The information transfer also acts as the upper bound of the predictive power of any model based on these predictors. Empirically, we evaluate the predictability at the hourly, daily and monthly frequency, and find that exchange rates are systematically predictable intraday, but not on other frequencies. Moreover, the linear model is suboptimal and fails to capture most of the information, which explains why it is so hard for the traditional linear models to outperform the random walk benchmark.

Improving Forecast Accuracy of Financial Vulnerability: Partial Least Squares Factor Model Approach

We present a factor augmented forecasting model for assessing the financial vulnerability in Korea. Dynamic factor models often extract latent common factors from a large panel of time series data via the method of the principal components (PC). Instead, we employ the partial least squares (PLS) method that estimates target specific common factors, utilizing covariances between predictors and the target variable. Applying PLS to 198 monthly frequency macroeconomic time series variables and the Bank of Korea's Financial Stress Index (KFSTI), our PLS factor augmented forecasting models consistently outperformed the random walk benchmark model in out-of-sample prediction exercises in all forecast horizons we considered. Our models also outperformed the autoregressive benchmark model in short-term forecast horizons. We expect our models would provide useful early warning signs of the emergence of systemic risks in Korea's financial markets.

Conditionally Optimal Weights and Forward-Looking Approaches to Combining Forecasts

In applied forecasting, there is a trade-off between in-sample fit and out-of-sample forecast accuracy. Parsimonious model specifications typically outperform richer model specifications. Consequently, there is often predictable information in forecast errors that is difficult to exploit. However, we show how this predictable information can be exploited in forecast combinations. In this case, optimal combination weights should minimize conditional mean squared error, or a conditional loss function, rather than the unconditional variance as in the commonly used framework of Bates and Granger (1969). We prove that our conditionally optimal weights lead to better forecast performance. The conditionally optimal weights support other forward-looking approaches to combining forecasts, where the forecast weights depend on the expected model performance. We show that forward-looking approaches can robustly outperform the random walk benchmark and many of the commonly used forecast combination strategies, including equal weights, in real-time out-of-sample forecasting exercises of inflation.

Cross-sectional variation of market efficiency

PurposeThis paper aims to provide evidence that market efficiency varies greatly across individual stock, and across market exchanges.Design/methodology/approachThree approaches, partial adjustment model, Dimson beta model and variance ratio test, are used on a large sample of US stocks.FindingsThis paper finds prices are closer to random walk benchmarks (i.e. more efficient) for stocks with better liquidity provision, frequent trading, greater return volatility, higher prices, larger market capitalizations and smaller trade sizes. These findings suggest that liquidity stimulates arbitrage activity, which, in turn, enhances market efficiency. Market efficiency also varies with information environment. The results show that stocks with greater information-based trading exhibit higher level of efficiency. Finally, market structure influences market efficiency. New York Stock Exchange stocks achieve higher level of efficiency than NASDAQ stocks do. The empirical results are robust and not driven by differences in stock attributes between the two markets.Research limitations/implicationsOverall, these results indicate that liquidity provision, stock attributes and market structure exert a significant impact on the realization of market efficiency.Practical implicationsIn addition, this paper is also relevant to both stock exchanges facing increased competition and to market regulators.Originality/valuePrior studies offer little evidence on the speed at which new information is impounded into the price. There is also limited evidence regarding how liquidity provision and market structure affect market efficiency. Using a transformation of the speed of price adjustment and other measurements as proxies for individual stock efficiency, this study may shed further lights on our understanding of market efficiency.

Exchange Rate Prediction Redux: New Models, New Data, New Currencies

Previous assessments of nominal exchange determination, following Meese and Rogoff (1983) have focused upon a narrow set of models. Cheung et al. (2005) augmented the usual suspects with productivity based models, and behavioral equilibrium exchange rate models, and assessed performance at horizons of up to 5 years. In this paper, we further expand the set of models to include Taylor rule fundamentals, yield curve factors, and incorporate shadow rates and risk and liquidity factors. The performance of these models is compared against the random walk benchmark. The models are estimated in error correction and first-difference specifications. We examine model performance at various forecast horizons (1 quarter, 4 quarters, 20 quarters) using differing metrics (mean squared error, direction of change), as well as the “consistency” test of Cheung and Chinn (1998). No model consistently outperforms a random walk, by a mean squared error measure, although purchasing power parity does fairly well. Moreover, along a direction-of-change dimension, certain structural models do outperform a random walk with statistical significance. While one finds that these forecasts are cointegrated with the actual values of exchange rates, in most cases, the elasticity of the forecasts with respect to the actual values is different from unity. Overall, model/specification/currency combinations that work well in one period will not necessarily work well in another period