nep-for New Economics Papers
on Forecasting
Issue of 2019‒05‒20
eight papers chosen by
Rob J Hyndman
Monash University

  1. Forecasting Causes of Death using Compositional Data Analysis: the Case of Cancer Deaths By Søren Kjærgaard; Yunus Emre Ergemen; Malene Kallestrup-Lamb; Jim Oeppen; Rune Lindahl-Jacobsen
  2. Longevity forecasting by socio-economic groups using compositional data analysis By Søren Kjærgaard; Yunus Emre Ergemen; Marie-Pier Bergeron Boucher; Jim Oeppen; Malene Kallestrup-Lamb
  3. A Stock Selection Method Based on Earning Yield Forecast Using Sequence Prediction Models By Jessie Sun
  4. Predicting China’s Monetary Policy with Forecast Combinations By Pauwels, Laurent
  5. The impact of forecast errors on fiscal planning and debt accumulation By Ademmer, Martin; Boysen-Hogrefe, Jens
  6. Implications of the permanent-transitory confusion for New-Keynesian modeling, inflation forecasts and the post-crisis era By Cukierman, Alex
  7. Heterogeneous component multiplicative error models for forecasting trading volumes By Naimoli, Antonio; Storti, Giuseppe
  8. Spatial modelling of the two-party preferred vote in Australian federal elections: 2001-2016 By Jeremy Forbes; Dianne Cook; Rob J Hyndman

  1. By: Søren Kjærgaard (University of Southern Denmark); Yunus Emre Ergemen (University of Aarhus and CREATES); Malene Kallestrup-Lamb (University of Aarhus and CREATES); Jim Oeppen (University of Southern Denmark); Rune Lindahl-Jacobsen (University of Southern Denmark)
    Abstract: Cause-specific mortality forecasting is often based on predicting cause-specific death rates independently. Only a few methods have been suggested that incorporate dependence among causes. An attractive alternative is to model and forecast cause-specific death distributions, rather than mortality rates, as dependence among the causes can be incorporated directly. We follow this idea and propose two new models which extend the current research on mortality forecasting using death distributions. We find that adding age, time, and cause-specific weights and decomposing both joint and individual variation among different causes of death increased the forecast accuracy of cancer deaths using data for French and Dutch populations
    Keywords: Cause-specific mortality, Cancer forecast, Forecasting methods, Compositional Data Analysis, Population health
    JEL: C22 C23 C53 I12
    Date: 2019–05–09
  2. By: Søren Kjærgaard (University of Southern Denmark); Yunus Emre Ergemen (University of Aarhus and CREATES); Marie-Pier Bergeron Boucher (University of Southern Denmark); Jim Oeppen (University of Southern Denmark); Malene Kallestrup-Lamb (University of Aarhus and CREATES)
    Abstract: Several OECD countries have recently implemented an automatic link between the statutory retirement age and life expectancy for the total population to insure sustainability in their pension systems when life expectancy is increasing. Significant mortality differentials are observed across socio-economic groups and future changes in these differentials will determine whether some socio-economic groups drive increases in the retirement age leaving other groups with fewer years in receipt of pensions. We forecast life expectancy by socio-economic groups and compare the forecast performance of competing models using Danish mortality data and find that the most accurate model assumes a common mortality trend. Life expectancy forecasts are used to analyse the consequences of a pension system where the statutory retirement age is increased when total life expectancy is increasing
    Keywords: Compositional data, forecasting, longevity, pension, socioeconomic groups
    JEL: C22 C23 C53 I12
    Date: 2019–05–09
  3. By: Jessie Sun
    Abstract: Long-term investors, different from short-term traders, focus on examining the underlying forces that affect the well-being of a company. They rely on fundamental analysis which attempts to measure the intrinsic value an equity. Quantitative investment researchers have identified some value factors to determine the cost of investment for a stock and compare different stocks. This paper proposes using sequence prediction models to forecast a value factor-the earning yield (EBIT/EV) of a company for stock selection. Two advanced sequence prediction models-Long Short-term Memory (LSTM) and Gated Recurrent Unit (GRU) networks are studied. These two models can overcome the inherent problems of a standard Recurrent Neural Network, i.e., vanishing and exploding gradients. This paper firstly introduces the theories of the networks. And then elaborates the workflow of stock pool creation, feature selection, data structuring, model setup and model evaluation. The LSTM and GRU models demonstrate superior performance of forecast accuracy over a traditional Feedforward Neural Network model. The GRU model slightly outperformed the LSTM model.
    Date: 2019–05
  4. By: Pauwels, Laurent
    Abstract: China’s monetary policy is unconventional and constantly evolving as a result of its rapid economic development. This paper proposes to use forecast combinations to predict the People’s Bank of China’s monetary policy stance with a large set of 73 macroeconomic and financial predictors covering various aspects of China’s economy. The multiple instruments utilised by the People’s Bank of China are aggregated into a Monetary Policy Index (MPI). The intention is to capture the overall monetary policy stance of the People’s Bank of China into a single variable that can be forecasted. Forecast combination assign weights to predictors according to their forecasting performance to produce a consensus forecast. The out-of-sample forecast results demonstrate that optimal forecast combinations are superior in predicting the MPI over other models such as the Taylor rule and simple autoregressive models. The corporate goods price index and the US nominal effective exchange rate are the most important predictors.
    Keywords: Monetary policy indicators; China; forecast combination; optimal weights
    Date: 2019–05–14
  5. By: Ademmer, Martin; Boysen-Hogrefe, Jens
    Abstract: We investigate the impact of errors in medium run tax revenue forecasts on the final budget balance. Our analysis is based on fiscal data for the entirety of German states and takes advantage of revenue forecasts and respective errors that can be considered as exogenously given in the budgeting process. We find that forecast errors at various forecast horizons translate considerably into the final budget balance, indicating that expenditure plans get only marginally adjusted when revenue forecasts get revised. Consequently, the accuracy of medium run forecasts considerably affects the sustainability of public finances. Our calculations suggest that a significant share of total debt of German states results from revenue forecasts that were too optimistic.
    Keywords: fiscal policy,fiscal planning,medium run forecasting,budget balance,public debt
    JEL: E62 H61 H68
    Date: 2019
  6. By: Cukierman, Alex
    Abstract: Decisions about consumption, work, leisure, pricing, investment and other private and public policy decisions rely on forecasts of the future. The permanent-transitory confusion (PTC) refers to the fact that even when they know all past and current information individuals are uncertain about the persistence of the current state. This all pervasive informational limitation makes it optimal, in general, to use all past information when forecasting the future even under rational expectations. The objective of this paper is to remind the profession of this basic fact and point out some of its implications by showing at both the theoretical and the empirical levels that forecasts of the future are generally adaptive in the sense that they depend on available past information even when information is utilized efficiently. This is done along the following dimensions. First by briefly surveying the literature on rational-adaptive expectations from Muth (1960) to Coibion-Gorodnichenko (2015). Second, by showing that the PTC injects the past even into purely forward looking New-Keynesian such as that of Clarida, Gali & Gertler (1999). Third, by showing empirically that inflationary expectations in the US Survey of Professional Forecasters rely on past inflation. The paper concludes with reflections on the persistence of economic and policy changes induced by the global financial crisis.
    Keywords: inflationary expectations; Permanent-transitory confusion; rationally adaptive expectations â?? implications for New-Keynesian framework
    JEL: D84 E12 E31
    Date: 2019–05
  7. By: Naimoli, Antonio; Storti, Giuseppe
    Abstract: We propose a novel approach to modelling and forecasting high frequency trading volumes. The new model extends the Component Multiplicative Error Model of Brownlees et al. (2011) by introducing a more flexible specification of the long-run component. This uses an additive cascade of MIDAS polynomial filters, moving at different frequencies, in order to reproduce the changing long-run level and the persistent autocorrelation structure of high frequency trading volumes. After investigating its statistical properties, the merits of the proposed approach are illustrated by means of an application to six stocks traded on the XETRA market in the German Stock Exchange.
    Keywords: Intra-daily trading volume, dynamic component models, long-range dependence, forecasting.
    JEL: C22 C53 C58
    Date: 2019–05–09
  8. By: Jeremy Forbes; Dianne Cook; Rob J Hyndman
    Abstract: We examine the relationships between electoral socio-demographic characteristics and two-party preference in the six Australian federal elections held between 2001 to 2016. Socio-demographic information is derived from the Australian Census, which occurs every five years. Since a Census is not directly available for each election, spatio-temporal imputation is employed to estimate Census data for the electorates at the time of each election. This accounts for both spatial and temporal changes in electoral characteristics between Censuses. To capture any spatial heterogeneity, a spatial error model is estimated for each election, which incorporates a spatially structured random effect vector that can be thought of as the unobserved political climate in each electorate. Over time, the impact of most socio-demographic characteristics that affect electoral two-party preference do not vary, with industry of work, incomes, household mobility and de facto relationships having strong effects in each of the six elections. Education and unemployment are amongst those that have varying effects. It is also found that between 2004 and 2013, the spatial effect was insignificant, meaning that electorates voted effectively independently. All data featured in this study has been contributed to the eechidna R package (available on CRAN).
    Keywords: Federal election, census, Australia, spatial modelling, imputation, data science, socio-demographics, electorates, R, eechidna.
    JEL: C31 C33 D72
    Date: 2019

This nep-for issue is ©2019 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.