|
on Forecasting |
| By: | Tae-Hwy Lee (Department of Economics, University of California Riverside); Eric Hillebrand (CREATES, Aarhus University); Huiyu Huang (ICBC Credit Suisse Asset Management); Canlin Li (Federal Reserve Board) |
| Abstract: | Following Diebold and Li (2006), we use the Nelson-Siegel (NS, 1987) yield curve factors. However the NS yield curve factors are not supervised for a specific forecast target in the sense that the same factors are used for forecasting different variables, e.g., output growth or inflation. We propose a modifed NS factor model, where the new NS yield curve factors are supervised for a specific target variable to forecast. We show that it outperforms the conventional (non-supervised) NS factor model in out-of-sample forecasting of monthly US output growth and inflation. The original NS yield factor model is to combine information (CI) of predictors and uses factors of predictors (the entire yield curve). The new supervised NS factor model is to combine forecasts (CF) and uses factors of forecasts of output growth or inflation conditional on each point of the yield curve. We formalize the concept of supervision, and demonstrate, both analytically and numerically, how supervision works. For both CF and CI schemes, principal components (PC) may also be used in place of the NS factors. In out-of-sample forecasting of U.S. monthly output growth and inflation, we find that supervised CF-factor models (CF-NS and CF-PC) are substantially better than unsupervised CI-factor models (CI-NS and CI-PC), especially at longer forecast horizons. |
| Keywords: | Level, slope, and curvature of the yield curve; Nelson-Siegel factors; Supervised factor models; Combining forecasts; Principal components. |
| JEL: | C5 E4 G1 |
| Date: | 2018–08 |
| URL: | http://d.repec.org/n?u=RePEc:ucr:wpaper:201903&r=all |
| By: | Tae-Hwy Lee (Department of Economics, University of California Riverside); Yundong Tu (Peking University) |
| Abstract: | This paper examines the theoretical and empirical properties of a supervised factor model based on combining forecasts using principal components (CFPC), in comparison with two other supervised factor models (partial least squares regression, PLS, and principal covariate regression, PCovR) and with the unsupervised principal component regression, PCR. The supervision refers to training the predictors for a variable to forecast. We compare the performance of the three supervised factor models and the unsupervised factor model in forecasting of U.S. CPI inflation. The main finding is that the predictive ability of the supervised factor models is much better than the unsupervised factor model. The computation of the factors can be doubly supervised together with variable selection, which can further improve the forecasting performance of the supervised factor models. Among the three supervised factor models, the CFPC best performs and is also most stable. While PCovR also performs well and is stable, the performance of PLS is less stable over different out-of-sample forecasting periods. The effect of supervision gets even larger as forecast horizon increases. Supervision helps to reduce the number of factors and lags needed in modelling economic structure, achieving more parsimony. |
| Keywords: | Combining forecasts; Principal components; Supervision matrix; Fixed point; Principal covariate regression; Partial least squares. |
| JEL: | C53 C55 E31 |
| Date: | 2018–12 |
| URL: | http://d.repec.org/n?u=RePEc:ucr:wpaper:201909&r=all |
| By: | Gergely Ganics; Florens Odendahl |
| Abstract: | We incorporate external information extracted from the European Central Bank's Survey of Professional Forecasters into the predictions of a Bayesian VAR, using entropic tilting and soft conditioning. The resulting conditional forecasts significantly improve the plain BVAR point and density forecasts. Importantly, we do not restrict the forecasts at a specific quarterly horizon, but their possible paths over several horizons jointly, as the survey information comes in the form of one- and two-year-ahead expectations. Besides improving the accuracy of the variable that we target, the spillover effects to ``other-than-targeted'' variables are relevant in size and statistically significant. We document that the baseline BVAR exhibits an upward bias for GDP growth after the financial crisis and our results provide evidence that survey forecasts can help mitigate the effects of structural breaks on the forecasting performance of a popular macroeconometric model. |
| Keywords: | : Survey of Professional Forecasters, Density forecasts, Entropic tilting, Soft conditioning. |
| JEL: | C53 C32 E37 |
| Date: | 2019 |
| URL: | http://d.repec.org/n?u=RePEc:bfr:banfra:733&r=all |
| By: | Mawuli Segnon (Department of Economics, Institute for Econometric and Economic Statistics, University of Münster, Germany); Rangan Gupta (Department of Economics, University of Pretoria, Pretoria, 0002, South Africa); Keagile Lesame (Department of Economics, University of Pretoria, Pretoria, 0002, South Africa); Mark E. Wohar (Department of Economics, University of NE-Omaha, USA and School of Business and Economics, Loughborough University, UK) |
| Abstract: | We propose a logistic smooth transition autoregressive fractionally integrated [STARFI(p,d)] process for modeling and forecasting US housing price volatility. We discuss the statistical properties of the model and investigate its forecasting performance by assuming various specifications for the dynamics underlying the variance process in the model. Using a unique database of daily data on price indices from ten major US cities, and the corresponding daily Composite 10 Housing Price Index, and also a housing futures price index, we find that using the Markov-switching multifractal (MSM) and FIGARCH frameworks for modeling the variance process helps improving the gains in forecast accuracy. |
| Keywords: | US housing prices, GARCH processes, MSM processes, Model confidence set |
| JEL: | C22 C53 C58 |
| Date: | 2019–10 |
| URL: | http://d.repec.org/n?u=RePEc:pre:wpaper:201977&r=all |
| By: | Tae-Hwy Lee (Department of Economics, University of California Riverside); Bai Huang (CUFE); Aman Ullah (UCR) |
| Abstract: | When some of the regressors in a panel data model are correlated with the random individual effects, the random effect (RE) estimator becomes inconsistent while the fixed effect (FE) estimator is consistent. Depending on the various degree of such correlation, we can combine the RE estimator and FE estimator to form a combined estimator which can be better than each of the FE and RE estimators. In this paper, we are interested in whether the combined estimator may be used to form a combined forecast to improve upon the RE forecast (forecast made using the RE estimator) and the FE forecast (forecast using the FE estimator) in out-of-sample forecasting. Our simulation experiment shows that the combined forecast does dominate the FE forecast for all degrees of endogeneity in terms of mean squared forecast errors (MSFE), demonstrating that the theoretical results of the risk dominance for the in-sample estimation carry over to the out-of-sample forecasting. It also shows that the combined forecast can reduce MSFE relative to the RE forecast for moderate to large degrees of endogeneity and for large degrees of heterogeneity in individual effects. |
| Keywords: | Endogeneity, Panel Data, Fixed Effect, Random Effect, Hausman test, Combined Estimator, Combined Forecast. |
| JEL: | C13 C33 C52 |
| Date: | 2018–12 |
| URL: | http://d.repec.org/n?u=RePEc:ucr:wpaper:201906&r=all |
| By: | Kai Carstensen (University of Kiel, Ifo Institute, CESifo); Markus Heinrich (University of Kiel); Magnus Reif (University of Kiel, Ifo Institute); Maik H. Wolters (University of Jena, Kiel Institute for the World Economy, IMFS at Goethe University Frankfurt) |
| Abstract: | We estimate a Markow-switching dynamic factor model with three states based on six leading business cycle indicators for Germany preselected from a broader set using the Elastic Net soft-thresholding rule. The three states represent expansions, normal recessions and severe recessions. We show that a two-state model is not sensitive enough to reliably detect relatively mild recessions when the Great Recession of 2008/2009 is included in the sample. Adding a third state helps to clearly distinguish normal and severe recessions, so that the model identifies reliably all business cycle turning points in our sample. In a real-time exercise the model detects recessions timely. Combining the estimated factor and the recession probabilities with a simple GDP forecasting model yields an accurate nowcast for the steepest decline in GDP in 2009Q1 and a correct prediction of the timing of the Great Recession and its recovery one quarter in advance. |
| Keywords: | Markov-Switching Dynamic Factor Model, Great Recession, Turning Points, GDP Nowcasting, GDP Forecasting |
| JEL: | C53 E32 E37 |
| Date: | 2019–09–17 |
| URL: | http://d.repec.org/n?u=RePEc:jrp:jrpwrp:2019-006&r=all |
| By: | Ruiz Ortega, Esther; Luo, Yun; González-Rivera, Gloria |
| Abstract: | We approximate probabilistic forecasts for interval-valued time series by offering alternative approaches. After fitting a possibly non-Gaussian bivariate VAR model to the center/log-range system, we transform prediction regions (analytical and bootstrap) for this system into regions for center/range and upper/lower bounds systems. Monte Carlo simulations show that bootstrap methods are preferred according to several new metrics. For daily S&P500 low/high returns, we build joint conditional prediction regions of the return level and volatility. We illustrate the usefulness of obtaining bootstrap forecasts regions for low/high returns by developing a trading strategy and showing its profitability when compared to using point forecasts. |
| Keywords: | Qml Estimation; Logarithmic Transformation; Coverage Rates; Constrainted Regression; Bootstrap |
| JEL: | C53 C22 C01 |
| Date: | 2019–10–15 |
| URL: | http://d.repec.org/n?u=RePEc:cte:wsrepe:29054&r=all |
| By: | Gloria Gonzalez-Rivera (Department of Economics, University of California Riverside); Yun Luo; Esther Ruiz (Universidad Carlos III de Madrid) |
| Abstract: | We approximate probabilistic forecasts for interval-valued time series by offering alternative approaches. After fi tting a possibly non-Gaussian bivariate VAR model to the center/log-range system, we transform prediction regions (analytical and bootstrap) for this system into regions for center/range and upper/lower bounds systems. Monte Carlo simulations show that bootstrap methods are preferred according to several new metrics. For daily S&P500 low/high returns, we build joint conditional prediction regions of the return level and volatility. We illustrate the usefulness of obtaining bootstrap forecasts regions for low/high returns by developing a trading strategy and showing its pro fitability when compared to using point forecasts. |
| Keywords: | Bootstrap, Constrainted Regression, Coverage Rates, Logarithmic Transformation, QML estimation. |
| JEL: | C01 C22 C53 |
| Date: | 2019–09 |
| URL: | http://d.repec.org/n?u=RePEc:ucr:wpaper:201921&r=all |
| By: | Timo Dimitriadis; Andrew J. Patton; Patrick Schmidt |
| Abstract: | Rational respondents to economic surveys may report as a point forecast any measure of the central tendency of their (possibly latent) predictive distribution, for example the mean, median, mode, or any convex combination thereof. We propose tests of forecast rationality when the measure of central tendency used by the respondent is unknown. These tests require us to overcome an identification problem when the measures of central tendency are equal or in a local neighborhood of each other, as is the case for (exactly or nearly) symmetric and unimodal distributions. As a building block, we also present novel tests for the rationality of mode forecasts. We apply our tests to survey forecasts of individual income, Greenbook forecasts of U.S. GDP, and random walk forecasts for exchange rates. We find that the Greenbook and random walk forecasts are best rationalized as mean, or near-mean forecasts, while the income survey forecasts are best rationalized as mode forecasts. |
| Date: | 2019–10 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:1910.12545&r=all |
| By: | MORIKAWA Masayuki |
| Abstract: | This study presents an ex post evaluation of the accuracy of long-term macroeconomic forecasts made by economics researchers. The results indicate, first, that the economic growth and inflation forecasts for the next ten years are biased upward. Second, there are positive correlations between real gross domestic product (GDP) and total factor productivity (TFP) growth forecasts, and between nominal GDP growth and consumer price index (CPI) inflation forecasts, resulting in the same correlations between forecasting errors for these macroeconomic variables. Third, GDP growth forecasts by academic researchers in economics are less upwardly biased than those by professional forecasters in private institutes. However, the upward bias of academic researchers specializing in macroeconomics and economic growth is larger than those in the other research fields. These results indicate that long-term economic forecasting involves significant uncertainty, even for economists. |
| Date: | 2019–10 |
| URL: | http://d.repec.org/n?u=RePEc:eti:dpaper:19084&r=all |
| By: | MORIKAWA Masayuki |
| Abstract: | This study presents ex post evaluation of the accuracy of long-term macroeconomic forecasts presented by economics researchers. The results indicate, first, that the GDP growth and inflation forecasts for the next ten years are biased upward. Second, there are positive correlations between real GDP and TFP growth forecasts and between nominal GDP growth and CPI inflation forecasts, resulting in the same correlations between forecasting errors for these macroeconomic variables. Third, academic researchers' GDP growth forecasts are less upward biased than those of professional forecasters in private institutions. However, the upward bias of academic researchers specializing in macroeconomics and economic growth is larger than those in the other research areas. These results indicate that long-term economic forecasting involves significant uncertainty even for professional economists. |
| Date: | 2019–10 |
| URL: | http://d.repec.org/n?u=RePEc:eti:rdpsjp:19058&r=all |
| By: | Ooft, G.; Bhaghoe, S.; Franses, Ph.H.B.F. |
| Abstract: | For many countries, statistical information on macroeconomic variables is not abundant and hence creating forecasts can be cumbersome. This paper addresses the creation of current year forecasts from a MIDAS regression for annual inflation rates where monthly inflation rates are the explanatory variables, and where the latter are only available for the last one and a half decade. The model can be viewed as a hybrid New-Keynesian Philips curve (NKPC). Specific focus is given to the forecast accuracy concerning the high inflation period in 2016-2017. |
| Keywords: | Inflation, New Keynesian Phillips curve, Rational Expectations, MIDAS Regression, Forecasting |
| JEL: | E12 E17 |
| Date: | 2019–09–01 |
| URL: | http://d.repec.org/n?u=RePEc:ems:eureir:120337&r=all |
| By: | Domenico Delli Gatti; Jakob Grazzini |
| Abstract: | We propose two novel methods to “bring ABMs to the data”. First, we put forward a new Bayesian procedure to estimate the numerical values of ABM parameters that takes into account the time structure of simulated and observed time series. Second, we propose a method to forecast aggregate time series using data obtained from the simulation of an ABM. We apply our methodological contributions to a medium-scale macro agent-based model. We show that the estimated model is capable of reproducing features of observed data and of forecasting one-period ahead output-gap and investment with a remarkable degree of accuracy. |
| Keywords: | agent-based models, estimation, forecasting |
| JEL: | C11 C13 C53 C63 |
| Date: | 2019 |
| URL: | http://d.repec.org/n?u=RePEc:ces:ceswps:_7894&r=all |
| By: | Halbleib, Roxana; Dimitriadis, Timo |
| JEL: | C1 C4 C5 |
| Date: | 2019 |
| URL: | http://d.repec.org/n?u=RePEc:zbw:vfsc19:203669&r=all |
| By: | Hinterlang, Natascha |
| JEL: | C45 C53 E47 |
| Date: | 2019 |
| URL: | http://d.repec.org/n?u=RePEc:zbw:vfsc19:203503&r=all |
| By: | Michael Coelli (Department of Economics, The University of Melbourne); Jeff Borland (Department of Economics, The University of Melbourne) |
| Abstract: | We review a highly influential study that estimated potential job loss from advances in Artificial Intelligence and robotics: Frey and Osborne (FO) (2013, 2017) concluded that 47 per cent of jobs in the United States were at ‘high risk’ of automation in the next 10 to 20 years. First, we investigate FO’s methodology for estimating job loss. Several major problems and limitations are revealed; especially associated with the subjective designation of occupations as fully automatable. Second, we examine whether FO’s predictions can explain occupation-level changes in employment in the United States from 2013 to 2018. Compared to standard approaches which classify jobs based on their intensity in routine tasks, FO’s predictions do not ‘add value’ for forecasting the impact of technology on employment. |
| Keywords: | employment; technology; prediction; job loss; AI and robotics |
| JEL: | J21 O33 |
| Date: | 2019–10 |
| URL: | http://d.repec.org/n?u=RePEc:iae:iaewps:wp2019n10&r=all |
| By: | Tae-Hwy Lee (Department of Economics, University of California Riverside); Yiyao Wang |
| Abstract: | We aim to find a forecast in the Survey of Professional Forecasters (SPF) that is closest to the Greenbook forecast of the Federal Reserve Board. To do it, we look for an SPF cross-sectional percentile that is not encompassed by the Greenbook forecast under the Greenbook's estimated asymmetric quadratic loss function with allowing asymmetry to be time-varying. To evaluate each SPF percentile in terms of the Greenbook's asymmetric quadratic loss function, we introduce the encompassing test for the asymmetric least square regression (Newey and Powell 1987). From the analysis of the U.S. quarterly real output and inflation forecasts over the past four decades, we find that almost all SPF percentiles are encompassed by the Greenbook forecast in full data period. However there is evidence in sub-periods that many SPF percentiles are not encompassed by Greenbook. Among them, the best SPF percentile that is not encompassed by Greenbook and is closest to Greenbook for real output growth forecast is near the median of the SPF percentiles, while the best SPF percentile for inflation forecast is far below the median in the left tail of the SPF cross-sectional distribution. It indicates that the common practice of using the SPF median can be misleading. |
| Keywords: | Asymmetric least squares, Encompassing test, Estimating asymmetric quadratic loss function, Forecast averaging, Model averaging. |
| JEL: | C1 C2 C3 C4 C5 |
| Date: | 2018–08 |
| URL: | http://d.repec.org/n?u=RePEc:ucr:wpaper:201904&r=all |