nep-for New Economics Papers
on Forecasting
Issue of 2020‒12‒21
eight papers chosen by
Rob J Hyndman
Monash University

  1. Density Forecasts in Panel Data Models: A Semiparametric Bayesian Perspective By Laura Liu
  2. Forecasting Realized Stock-Market Volatility: Do Industry Returns have Predictive Value? By Riza Demirer; Rangan Gupta; Christian Pierdzioch
  3. Forecasting Stock Market Recessions in the US: Predictive Modeling using Different Identification Approaches By Felix Haase; Matthias Neuenkirch
  4. Determinants and Forecasting of Female Labour Force Participation Rate in India: Testing of Feminization U hypothesis By Gaurang Rami
  5. Wheat Futures Trading Volume Forecasting and the Value of Extended Trading Hours By Joseph Janzen; Nicolas Legrand
  6. Capturing GDP nowcast uncertainty in real time By Paul Labonne
  7. Forecasting Financial Crashes: A Dynamic Risk Management Approach By J-C Gerlach; Dongshuai Zhao, CFA; Didier Sornette
  8. Public Mobility Data Enables COVID-19 Forecasting and Management at Local and Global Scales By Cornelia Ilin; Sébastien E. Annan-Phan; Xiao Hui Tai; Shikhar Mehra; Solomon M. Hsiang; Joshua E. Blumenstock

  1. By: Laura Liu (Indiana University, Bloomington, Indiana)
    Abstract: This paper constructs individual-specific density forecasts for a panel of firms or households using a dynamic linear model with common and heterogeneous coefficients and cross-sectional heteroskedasticity. The panel considered in this paper features a large cross-sectional dimension N but short time series T. Due to the short T, traditional methods have difficulty in disentangling the heterogeneous parameters from the shocks, which contaminates the estimates of the heterogeneous parameters. To tackle this problem, I assume that there is an underlying distribution of heterogeneous parameters, model this distribution nonparametrically allowing for correlation between heterogeneous parameters and initial conditions as well as individual-specific regressors, and then estimate this distribution by pooling the information from the whole cross-section together. Theoretically, I prove that both the estimated common parameters and the estimated distribution of the heterogeneous parameters achieve posterior consistency, and that the density forecasts asymptotically converge to the oracle forecast. Methodologically, I develop a simulation-based posterior sampling algorithm specifically addressing the nonparametric density estimation of unobserved heterogeneous parameters. Monte Carlo simulations and an empirical application to young firm dynamics demonstrate improvements in density forecasts relative to alternative approaches.
    Keywords: Bayesian, Semiparametric Methods, Panel Data, Density Forecasts, Posterior Consistency, Young Firm Dynamics
    JEL: C11 C14 C23 C53 L25
    Date: 2020–03
  2. By: Riza Demirer (Department of Economics & Finance, Southern Illinois University Edwardsville, Edwardsville, IL 62026-1102); Rangan Gupta (Department of Economics, University of Pretoria, Pretoria, 0002, South Africa); Christian Pierdzioch (Department of Economics, Helmut Schmidt University, Holstenhofweg 85, P.O.B. 700822, 22008 Hamburg, Germany)
    Abstract: Yes, they do. Utilizing a machine-learning technique known as random forests to compute forecasts of realized (good and bad) stock market volatility, we show that incorporating the information in lagged industry returns can help improve out-of sample forecasts of aggregate stock market volatility. While the predictive contribution of industry level returns is not constant over time, industrials and materials play a dominant predictive role during the aftermath of the 2008 global financial crisis, highlighting the informational value of real economic activity on stock market volatility dynamics. Finally, we show that incorporating lagged industry returns in aggregate level volatility forecasts benefits forecasters who are particularly concerned about under-predicting market volatility, yielding greater economic benefits for forecasters as the degree of risk aversion increases.
    Keywords: Stock market; Realized volatility; Industry returns, Market efficiency and information
    JEL: G17 Q02 Q47
    Date: 2020–12
  3. By: Felix Haase; Matthias Neuenkirch
    Abstract: Stock market recessions are often early warning signals for financial or economiccrises. Hence, forecasting bear markets is important for investors, policymak-ers, and economic agents in general. In our two-step procedure, we first iden-tify stock market regimes in the US using three different techniques (Markov-switching models, dating rules, and a na ̈ıve moving average). Second, we predictrecessions in the S&P 500 with the help of several modeling approaches, utilizingthe information of 92 macro-financial variables. Our results suggest that severalvariables are suitable for forecasting recessions in stock markets in-sample andout-of-sample. Our early warning models for the US equity market, in particu-lar those using principal components to aggregate the information in the macro-financial variables, provide a statistical improvement over several benchmarks. Inaddition, these generate economic value by boosting returns, improving the sharpratio and the omega, and substantially reducing drawdowns.
    Keywords: Dating Algorithms; Markov-Switching Models; Predictions; PrincipalComponent Analysis; Specific-to-General Approach; Stock Market Recessions.
    JEL: C53 G11 G17
    Date: 2020
  4. By: Gaurang Rami
    Abstract: Greater involvement of women within the labour force has economic and social impact. The Female Labour Force Participation Rate (FLFPR) for India remains appallingly low at around 27%, while the male labour force participation rate has been 79.9%. In India, during 1990–2016, the FLFPR (% of female population age 15+) showed a declining trend. In this paper, the determinants of FLFPR for India have been estimated using regression analysis for the time period 1990–2016. Data on all the relevant variables have been taken from World Development Indicators, World Bank. It has been found that FLFPR has strong negative linear correlation with LN GDP (constant 2010 US$), LN GDP per capita, PPP (constant 2011 international $), female tertiary school enrollment (% gross), and literacy rate and young female (% of females age 15–24). Results of regression analysis suggest that 87.9% variations in LFPR (% of female population age 15+) in India are explained by independent variables together. To forecast FLFPR, time series analysis with Auto Regressive Integrated Moving Average (ARIMA) has been applied. The ARIMA (1,2,0) model has been found to be the most suitable for forecasting value of FLFPR (for both, in-sample and out-sample). As per out-sample forecast, FLFPR in India will have increasing trend and it will be around 33.55% in 2035. From the results of trend analysis, scatter plot and correlation analysis, it can be concluded that the declining phase of Feminization U hypothesis has clearly been explained in India during 1990–2016. Results of curvilinear regression and out-sample forecast up to 2035, using the ARIMA technique, suggests that FLFPR (% of female population aged 15+) will have increasing trend; this may explain increasing phase of Feminization U hypothesis in India. To conclude, Femi-nization U hypothesis in India has been partially reinforced and it will be fully supported if forecasting of FLFPR (% of female population age 15+), LN GDP (constant 2010 US$), and LN GDP per capita, PPP (constant 2011 international $) are true. The aim is not only to increase participation of females in labour force, but to create an environment, providing opportunities and freedom for women to attain decent and dignified work which will contribute significantly in the economic empowerment and the holistic development of women, thereby ensuring gender equity in the labour market in India.
    Date: 2020–06
  5. By: Joseph Janzen (University of Illinois at Urbana-Champaign [Urbana] - University of Illinois System); Nicolas Legrand (SMART - Structures et Marché Agricoles, Ressources et Territoires - AGROCAMPUS OUEST - Institut Agro - Institut national d'enseignement supérieur pour l'agriculture, l'alimentation et l'environnement - INRAE - Institut National de Recherche pour l’Agriculture, l’Alimentation et l’Environnement)
    Abstract: Electronic trading in modern commodity markets has extended trading hours, lowered barriers to listing new contracts, broadened participation internationally, and encouraged entry of new trader types, particularly algorithmic traders whose order execution is automated. This paper seeks to understand how these forces have shaped the quantity and timing of trading activity, using the world's multiple wheat futures markets as a laboratory. To do so, we extend existing models for forecasting trading volume found in the literature on volume weighted average price (VWAP) order execution (e.g. Bialkowski, et al 2008 and Humphery-Jenner 2011) to applications beyond trading algorithm design. We consider a setting with multiple trading venues for related commodities, specifically the front-month Chicago Mercantile Exchange Soft Red Wheat and Paris Euronext Milling Wheat futures contracts. We compare a series of nested forecasting models to infer whether past trading history, intraday volume dynamics, cross market trading activity, and other information are useful predictors of trading activity. We assess the value of extended trading hours and the existence of alternative trading venues by testing whether trading volume is more predictable at particular times throughout the trading day.
    Keywords: Trading hours,High-frequency data,Volume predictions
    Date: 2019–05–15
  6. By: Paul Labonne
    Abstract: Nowcasting methods rely on timely series related to economic growth for producing and updating estimates of GDP growth before publication of official figures. But the statistical uncertainty attached to these forecasts, which is critical to their interpretation, is only improved marginally when new data on related series become available. That is particularly problematic in times of high economic uncertainty. As a solution this paper proposes to model common factors in scale and shape parameters alongside the mixed-frequency dynamic factor model typically used for location parameters in nowcasting frameworks. Scale and shape parameters control the time-varying dispersion and asymmetry round point forecasts which are necessary to capture the increase in variance and negative skewness found in times of recessions. It is shown how cross-sectional dependencies in scale and shape parameters may be modelled in mixed-frequency settings, with a particularly convenient approximation for scale parameters in Gaussian models. The benefit of this methodology is explored using vintages of U.S. economic growth data with a focus on the economic depression resulting from the coronavirus pandemic. The results show that modelling common factors in scale and shape parameters improves nowcasting performance towards the end of the nowcasting window in recessionary episodes.
    Date: 2020–12
  7. By: J-C Gerlach (ETH Zürich - Department of Management, Technology, and Economics (D-MTEC)); Dongshuai Zhao, CFA (ETH Zürich - Department of Management, Technology, and Economics (D-MTEC)); Didier Sornette (ETH Zürich - Department of Management, Technology, and Economics (D-MTEC); S wiss Finance Institute; Southern University of Science and Technology; Tokyo Institute of Technology)
    Abstract: Since 2009, stock markets have resided in a long bull market regime. Passive investment strategies have succeeded during this low-volatility growth period. From 2018 on, however, there was a transition into a more volatile market environment interspersed by corrections increasing in amplitude and frequency. This calls for more adaptive dynamic risk management strategies, as opposed to static buy-and-hold strategies. To hedge against market drawdowns, the greatest source of risk that should accurately be estimated is crash risk. This article applies the Log-Periodic Power Law Singularity (LPPLS) model of endogenous asset price bubbles to monitor crash risk. The model is calibrated to 15 years market history for five relevant equity country indices. Particular emphasis is put on the US S&P 500 Composite Index and the recent market history of the "Corona" year 2020. The results show that relevant historical bubble events, including the Corona crash, could be detected with the model and derived indicators. Many of these events were predicted in advance in monthly reports by the Financial Crisis Observatory (FCO) at ETH Zurich. The Corona crash, as the most recent event of interest, is discussed in further detail. Our conclusion is that unsustainable price dynamics leading to an unstable bubble, fuelled by quantitative easing and other policies, already existed well before the pandemic started. Thus, the bubble bursting in February 2020 as a reaction to the Corona pandemic was of endogenous nature and burst in response to the exogenous Corona crisis, which was predictable to some degree based on the endogenous price dynamics. Following the crash, a fast recovery of the price to pre-crisis levels ensued in the following months. This lets us conclude that, as long as the underlying origins and the macroeconomic environment that created this bubble do not change, the bubble will continue to grow and potentially spread to other sectors. This may cause even more hectic market behaviour, overreaction and volatile corrections in the future.
    Keywords: Financial Bubbles, Crashes, Forecasting, LPPLS Model, Dynamic Risk Management, Confidence Indicator
    JEL: C01 C53 C58 G01 G32
    Date: 2020–12
  8. By: Cornelia Ilin; Sébastien E. Annan-Phan; Xiao Hui Tai; Shikhar Mehra; Solomon M. Hsiang; Joshua E. Blumenstock
    Abstract: Policymakers everywhere are working to determine the set of restrictions that will effectively contain the spread of COVID-19 without excessively stifling economic activity. We show that publicly available data on human mobility — collected by Google, Facebook, and other providers — can be used to evaluate the effectiveness of non-pharmaceutical interventions and forecast the spread of COVID-19. This approach relies on simple and transparent statistical models, and involves minimal assumptions about disease dynamics. We demonstrate the effectiveness of this approach using local and regional data from China, France, Italy, South Korea, and the United States, as well as national data from 80 countries around the world.
    JEL: C1 C8 H12 H70 I18 O2 R40
    Date: 2020–11

This nep-for issue is ©2020 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.