|
on Forecasting |
| By: | Juan Ramón Hernández |
| Abstract: | The neutral band is the interval where deviations from Covered Interest Parity (CIP) are not considered meaningful arbitrage opportunities. The band is determined by transaction costs and risk associated to arbitrage. Seemingly large deviations from CIP in the foreign exchange markets for the US Dollar crosses with Sterling, Euro and Mexican Peso have been the norm since the Global Financial Crisis. The topic has attracted a lot of attention in the literature. There are no estimates of the neutral band to assess whether deviations from CIP reflect actual arbitrage opportunities, however. This paper proposes an estimate of the neutral band based on the one-step-ahead density forecast obtained from a stochastic volatility model. Comparison across models is made using the log-score statistic and the probability integral transformation. The stochastic volatility models have the best fit and forecasting performance, hence superior neutral band estimates. |
| Keywords: | Covered interest parity; stochastic volatility; forward filtering backward smoothing; auxiliary particle filter; density forecast |
| JEL: | C53 C58 F31 F37 |
| Date: | 2020–03 |
| URL: | http://d.repec.org/n?u=RePEc:bdm:wpaper:2020-02&r=all |
| By: | Yifei Lyu; Jun Nie; Shu-Kuei X. Yang |
| Abstract: | To examine whether including economic data on other countries could improve the forecast of U.S. GDP growth, we construct a large data set of 77 countries representing over 90 percent of global GDP. Our benchmark model is a dynamic factor model using U.S. data only, which we extend to include data from other countries. We show that using cross-country data produces more accurate forecasts during the global financial crisis period. Based on the latest vintage data on August 6, 2020, the benchmark model forecasts U.S. real GDP growth in 2020:Q3 to be −6.9 percent (year-over-year rate) or 14.9 percent (quarter-over-quarter annualized rate), whereas the forecast is revised upward to −6.1 percent (year-over-year) or 19.1 percent (quarter-over-quarter) when cross-country data are used. These examples suggest that U.S. data alone may fail to capture the spillover effects of other countries in downturns. However, we find that foreign variables are much less useful in normal times. |
| Keywords: | Forecasting; Dynamic factor model; GDP growth; Cross-country data; Global financial crisis; COVID-19 |
| JEL: | C32 C38 C53 C55 E32 E37 |
| Date: | 2020–08–20 |
| URL: | http://d.repec.org/n?u=RePEc:fip:fedkrw:88691&r=all |
| By: | Marius Lux; Wolfgang Karl H\"ardle; Stefan Lessmann |
| Abstract: | Appropriate risk management is crucial to ensure the competitiveness of financial institutions and the stability of the economy. One widely used financial risk measure is Value-at-Risk (VaR). VaR estimates based on linear and parametric models can lead to biased results or even underestimation of risk due to time varying volatility, skewness and leptokurtosis of financial return series. The paper proposes a nonlinear and nonparametric framework to forecast VaR that is motivated by overcoming the disadvantages of parametric models with a purely data driven approach. Mean and volatility are modeled via support vector regression (SVR) where the volatility model is motivated by the standard generalized autoregressive conditional heteroscedasticity (GARCH) formulation. Based on this, VaR is derived by applying kernel density estimation (KDE). This approach allows for flexible tail shapes of the profit and loss distribution, adapts for a wide class of tail events and is able to capture complex structures regarding mean and volatility. The SVR-GARCH-KDE hybrid is compared to standard, exponential and threshold GARCH models coupled with different error distributions. To examine the performance in different markets, one-day-ahead and ten-days-ahead forecasts are produced for different financial indices. Model evaluation using a likelihood ratio based test framework for interval forecasts and a test for superior predictive ability indicates that the SVR-GARCH-KDE hybrid performs competitive to benchmark models and reduces potential losses especially for ten-days-ahead forecasts significantly. Especially models that are coupled with a normal distribution are systematically outperformed. |
| Date: | 2020–09 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2009.06910&r=all |
| By: | Behrens, Christoph |
| Abstract: | I analyze the joint efficiency of export and import forecasts by leading economic research institutes for the years 1970 to 2017 for Germany in a multivariate setting. To this end, I compute, in a first step, multivariate random forests in order to model links between forecast errors and a forecaster's information set, consisting of several trade and other macroeconomic predictor variables. I use the Mahalanobis distance as performance criterion and, in a second step, permutation tests to check whether the Mahalanobis distance between the predicted forecast errors for the trade forecasts and actual forecast errors is significantly smaller than under the null hypothesis of forecast efficiency. I find evidence for joint forecast inefficiency for two forecasters, however, for one forecaster I cannot reject joint forecast efficiency. For the other forecasters, joint forecast efficiency depends on the examined forecast horizon. I find evidence that real macroeconomic variables as opposed to trade variables are inefficiently included in the analyzed trade forecasts. Finally, I compile a joint efficiency ranking of the forecasters. |
| Keywords: | Trade forecasts,German economic research institutes,Forecast efficiency,Multivariate random forests |
| JEL: | C53 F17 F47 |
| Date: | 2019 |
| URL: | http://d.repec.org/n?u=RePEc:zbw:pp1859:9&r=all |
| By: | Calvin Price (MUFG Bank) |
| Abstract: | Capturing interdependencies among many variables is a crucial part of economic forecasting. We show how multiple estimated equations can be solved simultaneously with the Stata forecast command and how to simulate the system through time to produce forecasts. This can be combined with user-defined exogenous variables, so that different assumptions can be used to create forecasts under different scenarios. Techniques for assessing the quality of both ex post and ex ante forecasts are shown, along with a simple example model of the U.S. economy. |
| Date: | 2020–08–20 |
| URL: | http://d.repec.org/n?u=RePEc:boc:scon20:5&r=all |
| By: | Alejandra Lelo-de-Larrea |
| Abstract: | Four specifications of an affine model with risk aversion and no arbitrage conditions are estimated for the Mexican Term Structure of Interest Rates, contrasting their empirical properties and the accuracy of their in and out of sample forecasts. The traditional models are extended by adding macroeconomic variables to analyze if the latter provide sufficient information to improve the adjustment and the forecast of interest rates. Using monthly data of the Zero Coupon Bonds, VIX, WTI, exchange rate, inflation and growth in the period 2002-2017, it is found that, although there is no superiority of a single model for the in and/or out of sample forecast of the yield curve, adding macroeconomic variables helps to improve the short and medium term forecasts independently of the type of factors used. |
| Keywords: | Afin Model, Yield Curve Forecast, Principal Components, Kalman Filter, No-Arbitrage Condition. |
| JEL: | C12 C32 C53 E43 E47 G12 |
| Date: | 2020–03 |
| URL: | http://d.repec.org/n?u=RePEc:bdm:wpaper:2020-01&r=all |
| By: | Ayush Jain; Smit Marvaniya; Shantanu Godbole; Vitobha Munigala |
| Abstract: | Accuracy of crop price forecasting techniques is important because it enables the supply chain planners and government bodies to take appropriate actions by estimating market factors such as demand and supply. In emerging economies such as India, the crop prices at marketplaces are manually entered every day, which can be prone to human-induced errors like the entry of incorrect data or entry of no data for many days. In addition to such human prone errors, the fluctuations in the prices itself make the creation of stable and robust forecasting solution a challenging task. Considering such complexities in crop price forecasting, in this paper, we present techniques to build robust crop price prediction models considering various features such as (i) historical price and market arrival quantity of crops, (ii) historical weather data that influence crop production and transportation, (iii) data quality-related features obtained by performing statistical analysis. We additionally propose a framework for context-based model selection and retraining considering factors such as model stability, data quality metrics, and trend analysis of crop prices. To show the efficacy of the proposed approach, we show experimental results on two crops - Tomato and Maize for 14 marketplaces in India and demonstrate that the proposed approach not only improves accuracy metrics significantly when compared against the standard forecasting techniques but also provides robust models. |
| Date: | 2020–09 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2009.04171&r=all |
| By: | Alkema, Leontine |
| Abstract: | BACKGROUND The Global Burden of Disease (GBD) project’s forecasts up to 2100 suggest fertility drops will be even greater in sub-Saharan Africa than the UN Population Division (UNPD) has predicted. OBJECTIVE This reflection summarizes the main assumptions used in the GBD fertility forecasts. I assess the methods used, focusing on high fertility countries and the use of met need for contraceptives as a predictor. RESULTS Based on GBD’s forecasting method, I draw two conclusions. Firstly, GBD fertility forecasts are based on unvalidated assumptions about increasing met need for contraception and may overestimate decreases in fertility in countries with low levels of modern contraceptive use. Secondly, the GBD forecast model for fertility is not a causal model for predicting changes. Claims GBD researchers make about the effect of changing access to family planning on fertility are not informative for guiding policy. Based on the GBD validation exercise, I conclude that the GBD study did not check the performance of the method for predicting left-out fertility data. Also the approach used to compare the predictive performance of UNPD and GBD forecasting methods may give the GBD method an inherent advantage. CONCLUSIONS Communication regarding the GBD method and its findings must avoid causal language and acknowledge the method’s limitations. Future research should examine the performance of the method, especially for countries with low modern contraceptive use. CONTRIBUTION This paper summarizes the GBD fertility forecasting method and indicates three areas of concern about it and its use. |
| Date: | 2020–08–20 |
| URL: | http://d.repec.org/n?u=RePEc:osf:osfxxx:3m6va&r=all |
| By: | Qiao Zhou; Ningning Liu |
| Abstract: | The prediction of a stock price has always been a challenging issue, as its volatility can be affected by many factors such as national policies, company financial reports, industry performance, and investor sentiment etc.. In this paper, we present a prediction model based on deep CNN and the candle charts, the continuous time stock information is processed. According to different information richness, prediction time interval and classification method, the original data is divided into multiple categories as the training set of CNN. In addition, the convolutional neural network is used to predict the stock market and analyze the difference in accuracy under different classification methods. The results show that the method has the best performance when the forecast time interval is 20 days. Moreover, the Moving Average Convergence Divergence and three kinds of moving average are added as input. This method can accurately predict the stock trend of the US NDAQ exchange for 92.2%. Meanwhile, this article distinguishes three conventional classification methods to provide guidance for future research. |
| Date: | 2020–09 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2009.03239&r=all |
| By: | Yuhyeon Bak (Department of Economics, Korea University, 145 Anamro, Seongbuk-gu, Seoul, Korea 02841); Cheolbeom Park (Department of Economics, Korea University, 145 Anamro, Seongbukgu, Seoul, Korea 02841) |
| Abstract: | Uncovered interest rate parity is known to perform poorly in forecasting exchange rate movements, especially in the short run. One possible reason for this failure is the existence of unobservable risk premium. We estimate the unobservable risk premium with a predictive system using the implied volatility of at-the-money currency options as an imperfect predictor. We find that expected exchange rate changes, constructed from forward-spot differentials and estimated risk premiums, track actual exchange rate changes more closely than do the fitted values of the Fama regression. When we add the estimated risk premium from the predictive system in the Fama regression, the UIP puzzle becomes weakened. An out-of-sample analysis reveals that adding the estimated risk premium greatly improves the short-run predictability of exchange rates. |
| Keywords: | exchange rate, Bayesian approach, predictive system, risk premium |
| JEL: | F31 F47 |
| Date: | 2020 |
| URL: | http://d.repec.org/n?u=RePEc:iek:wpaper:2006&r=all |
| By: | A. Fronzetti Colladon; S. Grassi; F. Ravazzolo; F. Violante |
| Abstract: | This paper uses a new textual data index for predicting stock market data. The index is applied to a large set of news to evaluate the importance of one or more general economic related keywords appearing in the text. The index assesses the importance of the economic related keywords, based on their frequency of use and semantic network position. We apply it to the Italian press and construct indices to predict Italian stock and bond market returns and volatilities in a recent sample period, including the COVID-19 crisis. The evidence shows that the index captures well the different phases of financial time series. Moreover, results indicate strong evidence of predictability for bond market data, both returns and volatilities, short and long maturities, and stock market volatility. |
| Date: | 2020–09 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2009.04975&r=all |
| By: | Bony Josaphat; Khreshna Syuhada |
| Abstract: | Risk measure forecast and model have been developed in order to not only provide better forecast but also preserve its (empirical) property especially coherent property. Whilst the widely used risk measure of Value-at-Risk (VaR) has shown its performance and benefit in many applications, it is in fact not a coherent risk measure. Conditional VaR (CoVaR), defined as mean of losses beyond VaR, is one of alternative risk measures that satisfies coherent property. There has been several extensions of CoVaR such as Modified CoVaR (MCoVaR) and Copula CoVaR (CCoVaR). In this paper, we propose another risk measure, called Dependent CoVaR (DCoVaR), for a target loss that depends on another random loss, including model parameter treated as random loss. It is found that our DCoVaR outperforms than both MCoVaR and CCoVaR. Numerical simulation is carried out to illustrate the proposed DCoVaR. In addition, we do an empirical study of financial returns data to compute the DCoVaR forecast for heteroscedastic process. |
| Date: | 2020–09 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2009.02904&r=all |
| By: | L. Ingber |
| Date: | 2020 |
| URL: | http://d.repec.org/n?u=RePEc:lei:ingber:20fi&r=all |
| By: | Paul Allison (Statistical Horizons LLC) |
| Abstract: | Although logistic regression is the most popular method for regression analysis of binary outcomes, there are still many attractions to using least-squares regression to estimate a linear probability model. A major downside, however, is that predicted “probabilities” from a linear model are often greater than 1 or less than 0. That can be problematic for many real-world applications. As a solution, we propose to generate predicted probabilities based on a linear discriminant model, which Haggstrom (1983) showed could be obtained by rescaling coefficients from OLS regression. We offer a new Stata command, predict_ldm, that can be used after the regress command to generate predicted values that always fall within the (0,1) interval. We show that, for many applications, these values are very close to those produced by logistic regression. We also explore applications where there are substantial differences between logistic predictions and those produced by predict_ldm. Finally, we show that the linear discriminant method can be used to substantially improve multiple imputations of categorical data based on the multivariate normal model. We are currently developing a new mi impute command to implement this method. |
| Date: | 2020–08–20 |
| URL: | http://d.repec.org/n?u=RePEc:boc:scon20:1&r=all |