nep-for New Economics Papers
on Forecasting
Issue of 2015‒10‒17
eleven papers chosen by
Rob J Hyndman
Monash University

  1. Forecasting German Car Sales Using Google Data and Multivariate Models By Fantazzini, Dean; Toktamysova, Zhamal
  2. Functional Coefficient Moving Average Model with Applications to forecasting Chinese CPI By Chen, Song Xi; Lei, Lihua; Tu, Yundong
  3. An Application of a Short Memory Model with Random Level Shifts to the Volatility of Latin American Stock Market Returns By Rodríguez, Gabriel; Tramontana, Roxana
  4. Constructing Indonesia Financial Stability Index By Anhar Fauzan Priyono; Arief Bustaman
  5. Natural Expectations and Home Equity Extraction By Pancrazi, Roberto; Pietrunti, Mario
  6. Asymmetries, Structural Breaks, and Nonlinear Persistence: Evidence and Implications for Uncovering the Energy-Growth Nexus in Selected African Countries By Njindan Iyke, Bernard
  7. Semiparametric Model Averaging of Ultra-High Dimensional Time Series By Jia Chen; Degui Li; Oliver Linton; Zudi Lu
  8. Textual Analysis in Real Estate By Adam Nowak; Patrick Smith
  9. Stochastic levels and duration dependence in US unemployment By de Bruijn, B.; Franses, Ph.H.B.F.
  10. Not Just Another Mixed Frequency Paper By Sergio Afonso Lago Alves; Angelo Marsiglia Fasolo
  11. Dynamic term structure models: the best way to enforce the zero lower bound in the US By Andreasen, Martin M; Meldrum, Andrew

  1. By: Fantazzini, Dean; Toktamysova, Zhamal
    Abstract: Long-term forecasts are of key importance for the car industry due to the lengthy period of time required for the development and production processes. With this in mind, this paper proposes new multivariate models to forecast monthly car sales data using economic variables and Google online search data. An out-of-sample forecasting comparison with forecast horizons up to 2 years ahead was implemented using the monthly sales of ten car brands in Germany for the period from 2001M1 to 2014M6. Models including Google search data statistically outperformed the competing models for most of the car brands and forecast horizons. These results also hold after several robustness checks which consider nonlinear models, different out-of-sample forecasts, directional accuracy, the variability of Google data and additional car brands.
    Keywords: Car Sales, Forecasting, Google, Google Trends, Global Financial Crisis, Great Recession
    JEL: C22 C32 C52 C53 L62
    Date: 2015
  2. By: Chen, Song Xi; Lei, Lihua; Tu, Yundong
    Abstract: This article establishes the functional coefficient moving average model (FMA), which allows the coefficient of the classical moving average model to adapt with a covariate. The functional coefficient is identified as a ratio of two conditional moments. Local linear estimation technique is used for estimation and asymptotic properties of the resulting estimator are investigated. Its convergence rate depends on whether the underlying function reaches its boundary or not, and asymptotic distribution could be nonstandard. A model specification test in the spirit of Hardle-Mammen (1993) is developed to check the stability of the functional coefficient. Intensive simulations have been conducted to study the finite sample performance of our proposed estimator, and the size and the power of the test. The real data example on CPI data from China Mainland shows the efficacy of FMA. It gains more than 20% improvement in terms of relative mean squared prediction error compared to moving average model.
    Keywords: Moving Average model, functional coefficient model, forecasting, Consumer Price Index.
    JEL: C1 C13 C5 C51 C53
    Date: 2014
  3. By: Rodríguez, Gabriel (Pontificia Universidad Católica del Perú); Tramontana, Roxana (Pontificia Universidad Católica del Perú)
    Abstract: Empirical research indicates that the volatility of stock return time series have long memory. However, it has been demonstrated that short memory processes contaminated with random level shifts can often be confused as being long memory. Often this feature is referred to as spurious long memory. This paper represents an empirical study of the random level shift (RLS) model using the approach of Lu and Perron (2010) and Li and Perron (2013) for the volatility of daily stocks returns data for …ve Latin American countries. The RLS model consists of the sum of a short term memory component and a level shift component, where the level shift component is governed by a Bernoulli process with a shift probability . The estimation results suggest that the level shifts in the volatility of daily stocks returns data are infrequent but once they are taken into account, the long memory characteristic and the GARCH e¤ects disappear. An out-of-sample forecasting exercise is also provided.
    Date: 2015–07
  4. By: Anhar Fauzan Priyono (Department of Economics, Padjadjaran University); Arief Bustaman (Department of Economics, Padjadjaran University)
    Abstract: Financial system stability is required to ensure a sustainable economic development. Indonesia's financial system has experienced some major shocks in its history (e.g., Asian Financial Crisis in 1997, and most recently 2008 sub-prime mortgage crisis). These financial shocks to certain extent affect the stability of Indonesia financial system, later, transmitted to contraction of Indonesia's economic growth. By this point, one might consider that financial stability and economic performance are highly correlated. This study has the following objectives. First, it offers an alternative concept and methodology for constructing Indonesian Financial System Stability Index (IFSI). Second, perform a forecast of IFSI. This study employed two methods IFSI namely Aggregation with Variance Equal Weight, and Principal Component Analysis (PCA). We found that Indonesia financial stability index constructed by the two approaches have the same trend pattern. By reducing dimension of variables in PCA technique loosing information on the movement of the index variation, we found that banking performance has major contribution to IFSI. Moreover, in the term of forecasting performance, we found that Adaptive Holt Winters Exponential Smoothing is superior to Autoregressive Moving Average (ARMA) technique.
    Keywords: Indonesia financial market, Financial Stability Index
    JEL: G0
    Date: 2015–10
  5. By: Pancrazi, Roberto (Department of Economics University of Warwick); Pietrunti, Mario (Banca d’Italia and Toulouse School of Economics)
    Abstract: In this paper we propose a novel explanation for the increase in households' leverage during the recent boom in U.S. housing prices. We use the U.S. housing market's boombust episode that led to the Great Recession as a case study, and we show that biased long-run expectations of both households and, especially, nancial intermediaries about future housing prices had a large impact on households' indebtedness. Specically, first we show that it is likely that financial intermediaries used forecasting models that ignored the long-run mean reversion of housing prices after a short-run momentum, thus leading to an overestimation of future households' housing wealth. We frame this finding in the theory of natural expectations, proposed by Fuster et al. (2010), to the housing market. Then, using a tractable model of collateralized credit market populated by households and banks, we find that: (1) mild variations in long-run forecasts of housing prices result in quantitatively considerable dierences in the amount of home equity extracted during a housing price boom; (2) the equilibrium levels of debt and interest rate are particularly sensitive to nancial intermediaries' naturalness; (3) home equity extraction data are better matched by models in which agents are fairly natural.
    Keywords: Natural expectations ; Home equity extraction ; Consumption/saving decision ; Housing pricecreation-date: 2015
    JEL: E21 E32 E44 D84
  6. By: Njindan Iyke, Bernard
    Abstract: The paper utilizes the nonparametric Triple test, the Bai-Perron test, and the KSS test to examine whether the paths of energy consumption and economic growth for 19 African countries are characterized by asymmetries, structural breaks, and nonlinear persistence over the period 1971-2011. We find evidence of deepness and steepness asymmetry, structural breaks, and nonlinear persistence in energy consumption and economic growth for these countries. The implications of these findings are that: (i) the findings of studies which examine the energy-growth nexus for these countries in linear settings may be doubtful; (ii) forecasts of energy consumption and economic growth which rely on linear models may contain sizeable forecasting errors. We recommend that future research on the energy-growth nexus should attempt to account for these nonlinearities in order to report more efficient estimates.
    Keywords: Asymmetries, Persistence, Energy Consumption, Economic Growth
    JEL: Q43 Q47
    Date: 2015–09–01
  7. By: Jia Chen; Degui Li; Oliver Linton; Zudi Lu
    Abstract: In this paper, we consider semiparametric model averaging of the nonlinear dynamic time series system where the number of exogenous regressors is ultra large and the number of autoregressors is moderately large. In order to accurately forecast the response variable, we propose two semiparametric approaches of dimension reduction among the exogenous regressors and auto-regressors (lags of the response variable). In the first approach, we introduce a Kernel Sure Independence Screening (KSIS) technique for the nonlinear time series setting which screens out the regressors whose marginal regression (or auto-regression) functions do not make significant contribution to estimating the joint multivariate regression function and thus reduces the dimension of the regressors from a possible exponential rate to a certain polynomial rate, typically smaller than the sample size; then we consider a semiparametric method of Model Averaging MArginal Regression (MAMAR) for the regressors and auto-regressors that survive the screening procedure, and propose a penalised MAMAR method to further select the regressors which have significant effects on estimating the multivariate regression function and predicting the future values of the response variable. In the second approach, we impose an approximate factor modelling structure on the ultra-high dimensional exogenous regressors and use a well-known principal component analysis to estimate the latent common factors, and then apply the penalised MAMAR method to select the estimated common factors and lags of the response variable which are significant. Through either of the two approaches, we can finally determine the optimal combination of the signicant marginal regression and auto-regression functions. Under some regularity conditions, we derive the asymptotic properties for the two semiparametric dimension-reduction approaches. Some numerical studies including simulation and an empirical application are provided to illustrate the proposed methodology.
    Keywords: Kernel smoother, penalised MAMAR, principal component analysis, semiparametric approximation, sure independence screening, ultra-high dimensional time series.
    JEL: C14 C22 C52
    Date: 2015–10
  8. By: Adam Nowak (West Virginia University, Department of Economics); Patrick Smith (Georgia State University, J. Mack Robinson College of Business)
    Abstract: This paper incorporates text data from MLS listings from Atlanta, GA into a hedonic pricing model. Text is found to decrease pricing error by more than 25%. Information from text is incorporated into a linear model using a tokenization approach. By doing so, the implicit prices for various words and phrases are estimated. The estimation focuses on simultaneous variable selection and estimation for linear models in the presence of a large number of variables. The LASSO procedure and variants are shown to outperform least-squares in out-of-sample testing.
    Keywords: textual analysis, big data, real estate valuation
    JEL: C01 C18 C51 C52 C65 R30
    Date: 2015–08
  9. By: de Bruijn, B.; Franses, Ph.H.B.F.
    Abstract: We introduce a new time series model that can capture the properties of data as is typically exemplified by monthly US unemployment data. These data show the familiar nonlinear features, with steeper increases in unem- ployment during economic downswings than the decreases during economic prosperity. At the same time, the levels of unemployment in each of the two states do not seem fixed, nor are the transition periods abrupt. Finally, our model should generate out-of-sample forecasts that mimic the in-sample properties. We demonstrate that our new and flexible model covers all those features, and our illustration to monthly US unemployment data shows its merits, both in and out of sample.
    Keywords: Markov switching, duration dependence, Gibbs sampling, unemployment, stochastic levels
    JEL: C11 C22 C24 C53 E24
    Date: 2015–09–23
  10. By: Sergio Afonso Lago Alves; Angelo Marsiglia Fasolo
    Abstract: This paper presents a new algorithm, based on a two-part Gibbs sampler with FFBS method, to recover the joint distribution of missing observations in a mixed-frequency dataset. The new algorithm relaxes most of the constraints usually presented in the literature, namely: (i) it does not require at least one time series to be observed every period; (ii) it provides an easy way to add linear restrictions based on the state space representation of the VAR; (iii) it does not require regularly-spaced time series at lower frequencies; and (iv) it avoids degeneration problems arising when states, or linear combination of states, are actually observed. In addition, the algorithm is well suited for embedding high-frequency real-time information for improving nowcasts and forecasts of lower frequency time series. We evaluate the properties of the algorithm using simulated data. Moreover, as empirical applications, we simulate monthly Brazilian GDP, comparing our results to the Brazilian IBC-BR, and recover what would historical PNAD-C unemployment rates look like prior to 2012
    Date: 2015–09
  11. By: Andreasen, Martin M; Meldrum, Andrew (Bank of England)
    Abstract: This paper studies whether dynamic term structure models for US nominal bond yields should enforce the zero lower bound by a quadratic policy rate or a shadow rate specification. We address the question by estimating quadratic term structure models (QTSMs) and shadow rate models (SRMs) with at most four pricing factors. Our findings suggest that QTSMs give a better in-sample fit than SRMs with two and three factors, whereas the SRM marginally dominates with four factors. Loadings from Campbell-Shiller regressions are generally better matched by the SRMs, which also outperform the QTSMs when forecasting bond yields, particularly with four pricing factors.
    Keywords: Bias-adjustment; forecasting study; quadratic term styructure models; sequential regression approach; shadow rate models
    JEL: C10 C50 G12
    Date: 2015–09–29

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.