nep-for New Economics Papers
on Forecasting
Issue of 2015‒10‒04
fourteen papers chosen by
Rob J Hyndman
Monash University

  1. Variable Selection for Inflation : A Pseudo Out-of-sample Approach By Selen Baser Andic; Fethi Ogunc
  2. Point and Density Forecasts Using an Unrestricted Mixed-Frequency VAR Model By Fady Barsoum
  3. Forecasting Output Growth using a DSGE-Based Decomposition of the South African Yield Curve By Rangan Gupta; Hylton Hollander; Rudi Steinbach
  4. The Predictive Content of Business Survey Indicators: evidence from SIGE By Tiziana Cesaroni; Stefano Iezzi
  5. An accuracy assessment of the global employment trends' unemployment rate forecasts By Bourmpoula, Evangelia; Wieser, Christina
  6. Rough electricity: a new fractal multi-factor model of electricity spot prices By Mikkel Bennedsen
  7. Forecasting a large set of disaggregates with common trends and outliers By Guillermo Carlomagno; Antoni Espasa
  8. Discovering common trends in a large set of disaggregates: statistical procedures and their properties By Guillermo Carlomagno; Antoni Espasa
  9. Household Debt and Business Cycles Worldwide By Atif R. Mian; Amir Sufi; Emil Verner
  10. The Explanatory Power and the Forecast Performance of Consumer Confidence Indices for Private Consumption Growth in Turkey By Hatice Gokce Karasoy; Caglar Yunculer
  11. Expected Business Conditions and Bond Risk Premia By Jonas Nygaard Eriksen
  12. An SVAR Approach to Evaluation of Monetary Policy in India: Solution to the Exchange Rate Puzzles in an Open Economy By William Barnett; Soumya Suvra Bhadury; Taniya Ghosh
  13. Multivariate dynamic intensity peaks-over-threshold models By Hautsch, Nikolaus; Herrera, Rodrigo
  14. Interest Rate Surprises and Transmission Mechanism in Turkey: Evidence from Impulse Response Analysis By K. Azim Ozdemir

  1. By: Selen Baser Andic; Fethi Ogunc
    Abstract: In this paper, we analyze the forecasting properties of a wide variety of variables for Turkish inflation, and thereby pin down the ones producing robust forecasts periodically. Defining the lag structure of a variable in two different ways, we determine the non-leading forecasters and leading indicators of inflation. We employ a pseudo out-of-sample approach and compare the forecasting performance of each variable ex-post with the benchmark model. We measure forecast errors over forecast horizons instead of over time for each horizon. Results suggest that no single variable gives the best forecasts at all times, hence inflation is best forecast by different variables each period. This finding promotes the use of forecast combination strategies and/or multivariate model settings.
    Keywords: Inflation, Variable selection, Leading indicator, Turkey
    JEL: C50 C53 E31 E37
    Date: 2015
  2. By: Fady Barsoum (Department of Economics, University of Konstanz, Germany)
    Abstract: This paper compares the forecasting performance of the unrestricted mixed-frequency VAR (MF-VAR) model to the more commonly used VAR (LF-VAR) model sampled a common low-frequency. The literature so far has successfully documented the forecast gains that can be obtained from using high-frequency variables in forecasting a lower frequency variable in a univariate mixed-frequency setting. These forecast gains are usually attributed to the ability of the mixed-frequency models to nowcast. More recently, Ghysels (2014) provides an approach that allows the usage of mixed-frequency variables in a VAR framework. In this paper we assess the forecasting and nowcasting performance of the MF-VAR of Ghysels (2014), however, we do not impose any restrictions on the parameters of the models. Although the unrestricted version is more flexible, it suffers from parameter proliferation and is therefore only suitable when the difference between the low- and high-frequency variables is small (i.e. quarterly and monthly frequencies). Unlike previous work, our interest is not only limited to evaluating the out-of-sample performance in terms of point forecasts but also density forecasts. Thus, we suggest a parametric bootstrap approach as well as a Bayesian approach to compute density forecasts. Moreover, we show how the nowcasts can be obtained using both direct and iterative forecasting methods. We use both Monte Carlo simulation experiments and an empirical study for the US to compare the forecasting performance of both the MF-VAR model and the LF-VAR model. The results highlight the point and density forecasts gains that can be achieved by the MF-VAR model.
    Keywords: Mixed-frequency, Bayesian estimation, Bootstrapping, Density forecasts, Nowcasting
    JEL: C32 C53 E37
    Date: 2015–09–25
  3. By: Rangan Gupta (Department of Economics, University of Pretoria); Hylton Hollander (African Institute of Financial Markets & Risk Management, Faculty of Commerce, University of Cape Town, South Africa); Rudi Steinbach (Economic Research and Statistics Department, South African Reserve Bank)
    Abstract: Evidence in favor of the ability of the term spread to forecast economic growth of the South African economy is non-existent. Presuming that this could be due to the term spread aggregating, and hence loosing out on important, information contained in the expected spread and the term premium, we: (i) Develop an estimable Small Open Economy New Keynesian Dynamic Stochastic General Equilibrium (SOENKDSGE) model of the in ation targeting South African economy; (ii) Use the SOENKDSGE model, estimated using Bayesian methods, to decompose the term spread into an expected spread and the term premium over the quarterly period of 2000:01-2014:04, and; (iii) Use a linear predictive regression framework to analyze the out-of-sample forecasting ability of the aggregate term spread, as well as the expected spread and term premium. Our forecasting results fail to detect forecasting gains from the aggregate term spread and also the term premium, but the expected spread is found to contain important information in forecasting the output growth over short- to medium-run horizons, over the out-of-sample period of 2004:01-2014:04. In other words, we confirm our presumption, and in the process highlight the importance of the forward looking component of the term spread, i.e., the expected spread, in forecasting output growth of South Africa.
    Keywords: Structural decomposition, Term spread, DSGE, Predictive regression framework, Forecasting output growth, South Africa
    JEL: C22 C53 E32 E43 E47
    Date: 2015–09
  4. By: Tiziana Cesaroni; Stefano Iezzi
    Abstract: Business surveys indicators represent an important tool in economic analysis and forecasting practices. While there is wide consensus on the coincident properties of such data, there is mixed evidence on their ability to predict macroeconomic developments in the short term. In this study we extend the previous research on business surveys predictive content by examining the leading properties of the main business survey indicators coming from the Italian Survey on Inflation and Growth Expectations (SIGE). To this end we provide a complete characterization of the business cycle properties of survey data (volatility, stationarity, turning points etc.) and we compare them with National Accounts reference series. We further analyze the forecast ability of the SIGE indicators to detect turning points using both discrete and continuous dynamic single equation models against their benchmark (B)ARIMA models. Overall the results indicate that SIGE business indicators are able to early detect turning points of their corresponding national account reference series. These findings are very important from a policy making point of view.
    Keywords: Business cycle, Business survey data, Turning points, cyclical analysis, Forecast accuracy, Macroeconomic forecasts
    JEL: C32 E32
    Date: 2015
  5. By: Bourmpoula, Evangelia; Wieser, Christina
    Abstract: This study provides a quantitative assessment of the bias, accuracy, and efficiency of the Global Employment Trends (GET) global and regional unemployment rate forecasts made in three recent annual GET reports. After conducting a series of statistical tests, the results suggest that, on average across all countries with data availability, the GET unemployment rate forecasts are slightly biased; we over-predict one and two years ahead and under-predict three and four years ahead. However, this bias is not significant for one to three years ahead. Moreover, our tests for accuracy show that the shorter the prediction period, the more accurate our forecasts indicated by smaller forecast errors for shorter prediction periods and larger forecast errors for longer periods.
    Keywords: unemployment, unemployed, forecast, evaluation, chômage, chômeurs, prévision, évaluation, desempleo, desempleados, predicción, evaluación
    Date: 2014
  6. By: Mikkel Bennedsen (Aarhus University and CREATES)
    Abstract: We introduce a new mathematical model of electricity spot prices which accounts for the most important stylized facts of these time series: seasonality, spikes, stochastic volatility and mean reversion. Empirical studies have found a possible fifth stylized fact, fractality, and our approach explicitly incorporates this into the model of the prices. Our setup generalizes the popular Ornstein Uhlenbeck-based multi-factor framework of Benth et al. (2007) and allows us to perform statistical tests to distinguish between an Ornstein Uhlenbeck-based model and a fractal model. Further, through the multi-factor approach we account for seasonality and spikes before estimating - and making inference on - the degree of fractality. This is novel in the literature and we present simulation evidence showing that these precautions are crucial to accurate estimation. Lastly, we estimate our model on recent data from six European energy exchanges and we find statistical evidence of fractality in five out of six markets. As an application of our model, we show how, in these five markets, a fractal component improves short term forecasting of the prices.
    Keywords: Energy markets, electricity prices, roughness, fractals, mean reversion, multi-factor modelling, forecasting.
    JEL: C22 C51 C52 C53 Q41
    Date: 2015–09–18
  7. By: Guillermo Carlomagno; Antoni Espasa
    Abstract: This paper deals with macro variables which have a large number of components and our aim is to model and forecasts all of them. We adopt a basic statistical procedure for discovering common trends among a large set of series and propose some extensions to take into account data irregularities and small samples issues. The forecasting strategy consists on estimating single-equation models for all the components, including the restrictions derived from the existence of common trends. An application to the disaggregated US CPI shows the usefulness of the procedure in real data problems.
    Keywords: Cointegration , Pairwise testing , Disaggregation , Forecast model selection , Outliers treatment , Inflation
    Date: 2015–09
  8. By: Guillermo Carlomagno; Antoni Espasa
    Abstract: The objective of this paper is to model all the N components of a macro or business variable. Our contribution concerns cases with a large number (hundreds) of components, for which multivariate approaches are not feasible. We extend in several directions the pairwise approach originally proposed by Espasa and Mayo-Burgos (2013) and study its statistical properties. The pairwise approach consists on performing common features tests between the N(N-1)/2 pairs of series that exist in the aggregate. Once this is done, groups of series that share common features can be formed. Next, all the components are forecast using single equation models that include the restrictions derived by the common features. In this paper we focus on discovering groups of components that share single common trends. We study analytically the asymptotic properties of the procedure. We also carry out a comparison with a DFM alternative; results indicate that the pairwise approach dominates in many empirically relevant situations. A clear advantage of the pairwise approach is that it does not need common features to be pervasive.
    Keywords: Cointegration , Factor Models , Disaggregation , Pairwise tests
    Date: 2015–09
  9. By: Atif R. Mian; Amir Sufi; Emil Verner
    Abstract: A rise in the household debt to GDP ratio predicts lower output growth and higher unemployment over the medium-run, contrary to standard macroeconomic models. GDP forecasts by the IMF and OECD underestimate the importance of a rise in household debt to GDP, giving the change in household debt to GDP ratio of a country the ability to predict growth forecasting errors. We use lower credit spreads and increases in risky debt issuance as instruments for the rise in household debt to GDP to argue that our results are supportive of recent models where debt growth is driven by changes in credit supply, borrowing constraints, or risk premia. We also show that a rise in household debt to GDP is associated contemporaneously with a rising consumption share of output, a worsening of the current account balance, and a rise in the share of consumption goods within imports. This is followed by strong external adjustment when the economy slows as the current account reverses and net exports increase due to a sharp fall in imports. Finally, an increase in global household debt to GDP also predicts lower global output growth. The pre-2000 predicted relationship between global household debt changes and subsequent global growth matches closely the actual decline in global growth after 2007 given the large increase in household debt during the early to mid-2000s.
    JEL: E17 E2 E21 E32 E44 G01 G21
    Date: 2015–09
  10. By: Hatice Gokce Karasoy; Caglar Yunculer
    Abstract: In this study, we assess empirically the relevance of consumer confidence indices (CCI) to future private consumption dynamics for Turkey in a sample period of 2002Q1 to 2014Q4. To this end, we first estimate models for total private consumption, durable and nondurable consumption growth with and without CCI and evaluate in-sample forecast powers. Next, we evaluate one-step-ahead out-of-sample forecast performances of these models from recursive OLS estimates. Finally, we test whether permanent income and precautionary savings hypotheses are capable of explaining our results on the link between consumer sentiment and future consumption expenditures. In our analyses we employ 4 different CCI series. These are overall index of CNBC-e Survey, overall index of TURKSTAT-CBRT Survey, Consumer Expectations Index (CEI) and Propensity to Consume Index (PCI) from CNBC-e Survey. Our results show that CCI have explanatory power on the future growth of both total consumption and its subcomponents. However, when other relevant variables such as real labour income, real stock price index and real interest rate are augmented to the models, CNBC-e and CEI for durable consumption, CEI and PCI for nondurable consumption are able to preserve their explanatory power on future consumption growth. On the other hand, CCI measures improve out-of-sample forecast performance for nondurable consumption growth. Finally, we find no evidence for either precautionary savings motive or permanent income hypothesis on the link between consumer sentiment and future private consumption changes.
    Keywords: Consumer confidence, Private consumption, Forecasting
    JEL: C52 C53 D12 E21 E27
    Date: 2015
  11. By: Jonas Nygaard Eriksen (Aarhus University and CREATES)
    Abstract: This paper studies the predictability of bond risk premia by means of expectations to future business conditions using survey forecasts from the Survey of Professional Forecasters. We show that expected business conditions consistently affect excess bond returns and that the inclusion of expected business conditions in standard predictive regressions improve forecast performance relative to models using information derived from the current term structure or macroeconomic variables. The results are confirmed in a real-time out-of-sample exercise, where the predictive accuracy of the models is evaluated both statistically and from the perspective of a mean-variance investor that trades in the bond market.
    Keywords: Bond risk premia, expected business conditions, predictability, economic value, expectations hypothesis, time-varying risk premia
    JEL: E43 E44 E47 G11 G12
    Date: 2015–09–24
  12. By: William Barnett (Department of Economics, The University of Kansas; Center for Financial Stability, New York City; IC2 Institute, University of Texas at Austin); Soumya Suvra Bhadury (Department of Economics, The University of Kansas;); Taniya Ghosh (Indira Gandhi Institute of Development Research (IGIDR), Reserve Bank of India, Mumbai, India)
    Abstract: Following the exchange-rate paper by Kim and Roubini (2000), we revisit the questions on monetary policy, exchange rate delayed overshooting, the inflationary puzzle, and the weak monetary transmission mechanism; but we do so for the open Indian economy. We further incorporate a superior monetary measure, the aggregation-theoretic Divisia monetary aggregate. Our paper confirms the efficacy of the Kim and Roubini (2000) contemporaneous restriction, customized for the Indian economy, especially when compared with recursive structure, which is damaged by the price puzzle and the exchange rate puzzle. The importance of incorporating correctly measured money into the exchange rate model is illustrated, when we compare models with no-money, simple-sum monetary measures, and Divisia monetary measures. Our results are confirmed in terms of impulse response, variance decomposition analysis, and out-of-sample forecasting. In addition, we do a flip-flop variance decomposition analysis, finding two important phenomena in the Indian economy: (i) the existence of a weak link between the nominal-policy variable and real-economic activity, and (ii) the use of inflation-targeting as a primary goal of the Indian monetary authority. These two main results are robust, holding across different time period, dissimilar monetary aggregates, and diverse exogenous model designs.
    Keywords: Monetary Policy; Monetary Aggregates; Divisia monetary aggregates; Structural VAR; Exchange Rate Overshooting; Liquidity Puzzle; Price Puzzle; Exchange Rate Puzzle; Forward Discount Bias Puzzle.
    JEL: C32 E51 E52 F31 F41
    Date: 2015–08
  13. By: Hautsch, Nikolaus; Herrera, Rodrigo
    Abstract: We propose a multivariate dynamic intensity peaks-over-threshold model to capture extreme events in a multivariate time series of returns. The random occurrence of extreme events exceeding a threshold is modeled by means of a multivariate dynamic intensity model allowing for feedback effects between the individual processes. We propose alternative specifications of the multivariate intensity process using autoregressive conditional intensity and Hawkes-type specifications. Likewise, temporal clustering of the size of exceedances is captured by an autoregressive multiplicative error model based on a generalized Pareto distribution. We allow for spillovers between both the intensity processes and the process of marks. The model is applied to jointly model extreme returns in the daily returns of three major stock indexes. We find strong empirical support for a temporal clustering of both the occurrence of extremes and the size of exceedances. Moreover, significant feedback effects between both types of processes are observed. Backtesting Value-at-Risk (VaR) and Expected Shortfall (ES) forecasts show that the proposed model does not only produce a good in-sample fit but also reliable out-of-sample predictions. We show that the inclusion of temporal clustering of the size of exceedances and feedback with the intensity thereof results in better forecasts of VaR and ES.
    Keywords: Extreme value theory,Value-at-Risk,Expected shortfall,Self-exciting point process,Conditional intensity
    Date: 2015
  14. By: K. Azim Ozdemir
    Abstract: This paper investigates the importance of interest rate shocks in explaining macroeconomic dynamics during the relatively low-inflation period in Turkey after mid-2000s. For this purpose, we compute impulse response functions using not only VAR models but also multi-step ahead forecast regressions, which are referred as Local Projections. Estimations are carried out on two different monthly data sets, a set of conventional series and a newly constructed set of series for measuring real GDP, the price level and the exchange market pressure in Turkey. Impulse responses obtained from newly constructed series exhibit more plausible dynamics than the conventional series after an interest rate shock. Moreover, results from Local Projections show remarkably similar dynamic responses to those obtained from the VAR models. This finding can be interpreted as an evidence that the identified VAR models successfully capture the true relationships among the variables.
    Keywords: Monetary Policy, Identification, VAR, Local Projections, Interpolation
    JEL: C32 E52 C82
    Date: 2015

This nep-for issue is ©2015 by Rob J Hyndman. It is provided as is without any express or implied warranty. It may be freely redistributed in whole or in part for any purpose. If distributed in part, please include this notice.
General information on the NEP project can be found at For comments please write to the director of NEP, Marco Novarese at <>. Put “NEP” in the subject, otherwise your mail may be rejected.
NEP’s infrastructure is sponsored by the School of Economics and Finance of Massey University in New Zealand.