nep-for New Economics Papers
on Forecasting
Issue of 2020‒10‒05
eleven papers chosen by
Rob J Hyndman
Monash University

  1. Proper scoring rules for evaluating asymmetry in density forecasting By Matteo Iacopini; Francesco Ravazzolo; Luca Rossini
  2. The merge of two worlds: Integrating artificial neural networks into agent-based electricity market simulation By Fraunholz, Christoph; Kraft, Emil; Keles, Dogan; Fichtner, Wolf
  3. Forecasting the Covid-19 recession and recovery: lessons from the financial crisis By Foroni, Claudia; Marcellino, Massimiliano; Stevanović, Dalibor
  4. The Predictive Power of NZX Dairy Futures By Scott, Ayesha; Schoen, Tilman; Fernandez-Perez, Adrian
  5. The Corona Virus, the Stock MarketÕs Response, and Growth Expectations By Niels J. Gormsen; Ralph S.J. Koijen
  6. Volatility Forecasting with 1-dimensional CNNs via transfer learning By Bernadett Aradi; G\'abor Petneh\'azi; J\'ozsef G\'all
  7. FRED-SD: A Real-Time Database for State-Level Data with Forecasting Applications By Kathryn Bokun; Laura E. Jackson; Kevin L. Kliesen; Michael T. Owyang
  8. Forecasting the Leading Indicator of a Recession: The 10-Year minus 3-Month Treasury Yield Spread By Sudiksha Joshi
  9. Does the Phillips curve help to forecast euro area inflation? By Bańbura, Marta; Bobeica, Elena
  10. Supervised learning for the prediction of firm dynamics By Falco J. Bargagli-Stoffi; Jan Niederreiter; Massimo Riccaboni
  11. Scenario Forecast of Cross-border Electric Interconnection towards Renewables in South America By Wenhao Wang; Jing Meng; Duan Chen; Wei Cong

  1. By: Matteo Iacopini; Francesco Ravazzolo; Luca Rossini
    Abstract: This paper proposes a novel asymmetric continuous probabilistic score (ACPS) for evaluating and comparing density forecasts. It extends the proposed score and defines a weighted version, which emphasizes regions of interest, such as the tails or the center of a variable’s range. A test is also introduced to statistically compare the predictive ability of different forecasts. The ACPS is of general use in any situation where the decision maker has asymmetric preferences in the evaluation of the forecasts. In an artificial experiment, the implications of varying the level of asymmetry in the ACPS are illustrated. Then, the proposed score and test are applied to assess and compare density forecasts of macroeconomic relevant datasets (US employment growth) and of commodity prices (oil and electricity prices) with particular focus on the recent COVID-19 crisis period.
    Keywords: asymmetric continuous probablistic score, asymmetric loss, proper score, density forecast, predictive distribution, weighted score, probabilistic forecast
    Date: 2020–09
  2. By: Fraunholz, Christoph; Kraft, Emil; Keles, Dogan; Fichtner, Wolf
    Abstract: Machine learning and agent-based modeling are two popular tools in energy research. In this article, we propose an innovative methodology that combines these methods. For this purpose, we develop an electricity price forecasting technique using artificial neural networks and integrate the novel approach into the established agent-based electricity market simulation model PowerACE. In a case study covering ten interconnected European countries and a time horizon from 2020 until 2050 at hourly resolution, we benchmark the new forecasting approach against a simpler linear regression model as well as a naive forecast. Contrary to most of the related literature, we also evaluate the statistical significance of the superiority of one approach over another by conducting Diebold-Mariano hypothesis tests. Our major results can be summarized as follows. Firstly, in contrast to real-world electricity price forecasts, we find the naive approach to perform very poorly when deployed model-endogenously. Secondly, although the linear regression performs reasonably well, it is outperformed by the neural network approach. Thirdly, the use of an additional classifier for outlier handling substantially improves the forecasting accuracy, particularly for the linear regression approach. Finally, the choice of the model-endogenous forecasting method has a clear impact on simulated electricity prices. This latter finding is particularly crucial since these prices are a major results of electricity market models.
    Keywords: Agent-based simulation,Artificial neural network,Electricity price forecasting,Electricity market
    Date: 2020
  3. By: Foroni, Claudia; Marcellino, Massimiliano; Stevanović, Dalibor
    Abstract: We consider simple methods to improve the growth nowcasts and forecasts obtained by mixed frequency MIDAS and UMIDAS models with a variety of indicators during the Covid-19 crisis and recovery period, such as combining forecasts across various specifications for the same model and/or across different models, extending the model specification by adding MA terms, enhancing the estimation method by taking a similarity approach, and adjusting the forecasts to put them back on track by a specific form of intercept correction. Among all these methods, adjusting the original nowcasts and forecasts by an amount similar to the nowcast and forecast errors made during the financial crisis and following recovery seems to produce the best results for the US, notwithstanding the different source and characteristics of the financial crisis. In particular, the adjusted growth nowcasts for 2020Q1 get closer to the actual value, and the adjusted forecasts based on alternative indicators become much more similar, all unfortunately indicating a much slower recovery than without adjustment and very persistent negative effects on trend growth. Similar findings emerge also for the other G7 countries. JEL Classification: C53, E37
    Keywords: Covid-19, forecasting, GDP, mixed-frequency
    Date: 2020–09
  4. By: Scott, Ayesha; Schoen, Tilman; Fernandez-Perez, Adrian
    Abstract: We assess the market efficiency of four New Zealand Exchange (NZX) dairy futures, which are cash-settled to Global Dairy Trade (GDT) results. Importantly, we identify quantitative drivers of the deviations from futures to spot prices and compare the forecasting accuracy of the four dairy futures. The results of this study are relevant for local and international participants on New Zealand dairy spot and futures markets, from farmers and dairy processors to traders and international food manufacturers. Established in 2008, the GDT is an online platform that runs twice-monthly ascending price clock dairy auction events, providing widely-accepted price references for sellers and buyers of dairy commodities including whole milk powder (WMP), skim milk powder (SMP), butter, and anhydrous milk fat (AMF). WMP futures were launched by NZX in October 2010, SMP and AMF futures followed in February 2011, and NZX butter futures were introduced in December 2014. With these four contracts combined, over 800,000 metric tons of dairy futures were traded to July 2018. As New Zealand is one of the world’s largest dairy exporters, accounting for one-third of internationally traded dairy products in 2017 despite only being the eighth-largest dairy producer, both GDT spot prices and the NZX dairy futures have an immediate and significant impact on international dairy trade. The dataset spans October 2010 to July 2018, comprising of 179 GDT auction events, 179 WMP, 157 SMP futures, 151 AMF and 79 butter futures data and consists of data that is auction- but not commodity-specific as well as data that is both auction- and commodity-specific. For example, the quantity of volume sold at each GDT event is available for WMP, SMP, AMF, and butter separately, while the auction participation of buyers from different regions is only given for the entire auction event. The GDT data set of this study allows the assessment of futures’ forecast accuracy or market efficiency and additionally the identification of quantitative drivers thereof as well as interdependencies of spot and futures markets. We use standard regression methods to assess predictive power of futures and identify drivers of their forecast accuracy. We provide evidence the four examined NZX dairy futures – whole milk powder, skim milk powder, anhydrous milk fat, and butter – perform efficient price discovery, as they provide market participants with accurate and unbiased GDT price forecasts. We also identify several statistically and economically significant factors influencing futures’ forecast accuracy or market efficiency. For example, SMP, AMF, and butter futures are found to forecast each month’s first GDT result significantly more accurately than the second. Moreover, the forecast error is not driven by trade volumes or open interest, confirming existing literature that efficient forecasting is not dependent on high liquidity but informed trading. Also, several futures’ forecast accuracies were positively or negatively affected by the participation of certain buyer groups on GDT, hinting toward varying levels of activity among these buyers on the respective NZX futures market. Lastly, WMP futures are proven to provide significantly more accurate GDT price forecasts than the other three contracts.
    Keywords: Agribusiness
    Date: 2020–09–16
  5. By: Niels J. Gormsen (University of Chicago - Booth School of Business); Ralph S.J. Koijen (University of Chicago Booth School of Business)
    Abstract: We use data from the aggregate equity market and dividend futures to quantify how investorsÕ expectations about economic growth across horizons evolve in response to the corona virus outbreak and subsequent policy responses. Dividend futures, which are claims to dividends on the aggregate stock market in a particular year, can be used to directly compute a lower bound on growth expectations across maturities or to estimate expected growth using a simple forecasting model. We show how the actual forecast and the bound evolve over time. As of March 16, expected growth over the next quarter declined by 2.5% in the US and 3.5% in Europe (both annualized) compared to the beginning of the year. The lower bound on expected GDP growth has been revised down by as much as 10% in the US and 12% in the EU. There are signs of catch-up growth from year 4 to year 10. News about economic relief programs on March13 appear to have increased stock prices by lowering risk aversion and lift long-term growth expectations, but did little to improve expectations about short-term growth.The events on March 16 reflect a deterioration of fundamentals in the US and predicta deepening of the economic downturn. We also show how data on dividend futures can be used to understand why stock markets fell so sharply, well beyond changes ingrowth expectations
    Date: 2020
  6. By: Bernadett Aradi; G\'abor Petneh\'azi; J\'ozsef G\'all
    Abstract: Volatility is a natural risk measure in finance as it quantifies the variation of stock prices. A frequently considered problem in mathematical finance is to forecast different estimates of volatility. What makes it promising to use deep learning methods for the prediction of volatility is the fact, that stock price returns satisfy some common properties, referred to as `stylized facts'. Also, the amount of data used can be high, favoring the application of neural networks. We used 10 years of daily prices for hundreds of frequently traded stocks, and compared different CNN architectures: some networks use only the considered stock, but we tried out a construction which, for training, uses much more series, but not the considered stocks. Essentially, this is an application of transfer learning, and its performance turns out to be much better in terms of prediction error. We also compare our dilated causal CNNs to the classical ARIMA method using an automatic model selection procedure.
    Date: 2020–09
  7. By: Kathryn Bokun; Laura E. Jackson; Kevin L. Kliesen; Michael T. Owyang
    Abstract: We construct a real-time dataset (FRED-SD) with vintage data for the U.S. states that can be used to forecast both state-level and national-level variables. Our dataset includes approximately 28 variables per state, including labor market, production, and housing variables. We conduct two sets of real-time forecasting exercises. The first forecasts state-level labor-market variables using five different models and different levels of industrially-disaggregated data. The second forecasts a national-level variable exploiting the cross-section of state data. The state-forecasting experiments suggest that large models with industrially-disaggregated data tend to have higher predictive ability for industrially-diversified states. For national-level data, we find that forecasting and aggregating state-level data can outperform a random walk but not an autoregression.
    Keywords: space-time autogregression; factor model; VAR; industrial diversity
    JEL: C33 R11
    Date: 2020–08–14
  8. By: Sudiksha Joshi
    Abstract: In this research paper, I have applied various econometric time series and two machine learning models to forecast the daily data on the yield spread. First, I decomposed the yield curve into its principal components, then simulated various paths of the yield spread using the Vasicek model. After constructing univariate ARIMA models, and multivariate models such as ARIMAX, VAR, and Long Short Term Memory, I calibrated the root mean squared error to measure how far the results deviate from the current values. Through impulse response functions, I measured the impact of various shocks on the difference yield spread. The results indicate that the parsimonious univariate ARIMA model outperforms the richly parameterized VAR method, and the complex LSTM with multivariate data performs equally well as the simple ARIMA model.
    Date: 2020–09
  9. By: Bańbura, Marta; Bobeica, Elena
    Abstract: We find that it does, but choosing the right specification is not trivial. We unveil notable model instability, with breaks in the performance of most simple Phillips curves. Euro area inflation was particularly hard to forecast in the run-up to the EMU and after the sovereign debt crisis, when the trend and for the latter period, also the amount of slack, were harder to pin down. Yet, some specifications outperform a univariate benchmark most of the time and are thus a useful element in a forecaster's toolkit. We base these conclusions on an extensive forecast evaluation over 1994 - 2018, an extraordinarily long period by euro area standards. We complement the analysis using real-time data over 2005-2018. As lessons for practitioners, we find that: (i) the key type of time variation to consider is an inflation trend; (ii) a simple filter-based output gap works well overall as a measure of economic slack, but after the Great Recession it is outperformed by endogenously estimated slack or by estimates from international economic institutions; (iii) external variables do not bring forecast gains; (iv) newer generation Phillips curve models with several time-varying features are a promising avenue for forecasting, especially when density forecasts are of interest, and finally, (v) averaging over a wide range of modelling choices offers some hedge against breaks in forecast performance. JEL Classification: C53, E31, E37
    Keywords: C53, E31, E37
    Date: 2020–09
  10. By: Falco J. Bargagli-Stoffi; Jan Niederreiter; Massimo Riccaboni
    Abstract: Thanks to the increasing availability of granular, yet high-dimensional, firm level data, machine learning (ML) algorithms have been successfully applied to address multiple research questions related to firm dynamics. Especially supervised learning (SL), the branch of ML dealing with the prediction of labelled outcomes, has been used to better predict firms' performance. In this contribution, we will illustrate a series of SL approaches to be used for prediction tasks, relevant at different stages of the company life cycle. The stages we will focus on are (i) startup and innovation, (ii) growth and performance of companies, and (iii) firms exit from the market. First, we review SL implementations to predict successful startups and R&D projects. Next, we describe how SL tools can be used to analyze company growth and performance. Finally, we review SL applications to better forecast financial distress and company failure. In the concluding Section, we extend the discussion of SL methods in the light of targeted policies, result interpretability, and causality.
    Date: 2020–09
  11. By: Wenhao Wang; Jing Meng; Duan Chen; Wei Cong
    Abstract: Cross-border Electric Interconnection towards renewables is a promising solution for electric sector under the UN 2030 sustainable development goals which is widely promoted in emerging economies. This paper comprehensively investigates state of art in renewable resources and cross-border electric interconnection in South America. Based on the raw data collected from typical countries, a long-term scenario forecast methodology is applied to estimate key indicators of electric sector in target years, comparing the prospects of active promoting cross-border Interconnections Towards Renewables (ITR) scenario with Business as Usual (BAU) scenario in South America region. Key indicators including peak load, installed capacity, investment, and generation cost are forecasted and comparative analyzed by year 2035 and 2050. The comparative data analysis shows that by promoting cross-border interconnection towards renewables in South America, renewable resources can be highly utilized for energy supply, energy matrix can be optimized balanced, economics can be obviously driven and generation cost can be greatly reduced.
    Date: 2020–09

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.