|
on Forecasting |
| By: | Mitchell, James (University of Warwick); Weale, Martin (King's College London) |
| Abstract: | This paper considers the production and evaluation of density forecasts paying attention to if and how the probabilities of outlying observations are quantified and communicated. Particular focus is given to the ‘censored’ nature of the Bank of England’s fan charts, given that - which is commonly ignored - they describe only the inner 90% (best critical region) of the forecast distribution. A new estimator is proposed that fits a potentially skewed and fat tailed density to the inner observations, acknowledging that the outlying observations may be drawn from a different but unknown distribution. In forecasting applications, motivation for this could reflect the view that outlying forecast errors reflect (realised) unknown unknowns or events not expected to recur that should be censored before quantifying known unknowns. |
| Keywords: | forecasting uncertainty; fan charts; skewed densities; best critical region; density forecasting; censoring; forecasting evaluation; |
| JEL: | C24 C46 C53 E58 |
| Date: | 2019 |
| URL: | http://d.repec.org/n?u=RePEc:wrk:wrkemf:27&r=all |
| By: | Jakob Fiedler; Josef Ruzicka; Thomas Theobald |
| Abstract: | We integrate newly created financial stress indices (FSIs) into an automated real-time recession forecasting procedure for the Euro area and Germany. The FSIs are based on a large number of financial indicators, each of them potentially signaling financial stress. A subset of these indicators is selected in real-time and their stress signal is summarized by principal component analysis (PCA). Besides conventional measures of realized financial stress, such as volatilities, we include variables related to the financial cycle, such as different types of credit growth, for which strong increases may anticipate future financial market stress. Building blocks in our fully automated real-time probit forecasts are then i. the use of a broad set of widely acknowledged macroeconomic and financial variables with predictive power for a real economic downturn, ii. the use of both general-to-specific and specific-to-general approaches for variable and lag selection, and iii. the averaging of different specifications into a composite forecast. As a real-time out-of-sample analysis shows, the inclusion of financial stress leads to an improved recession forecast for the Euro area, while the results for Germany are mixed. Finally, we also evaluate the predictive power of the change in bank lending (credit impulse) and find that it adds little additional information. |
| Date: | 2019 |
| URL: | http://d.repec.org/n?u=RePEc:imk:wpaper:198-2019&r=all |
| By: | Tara M. Sinclair (The George Washington University) |
| Abstract: | Throughout the history of macroeconomic forecasting, several major themes have remained surprisingly consistent. The failure to forecast economic downturns ahead of time is perhaps the most significant of these. Forecasting approaches have changed, but forecasts for recessions have not improved. What can we learn from past evaluations of macroeconomic forecasts? Is it possible to predict major economic shocks or is it a fool’s errand? This chapter discusses how forecasting techniques have evolved over time and yet the record on forecasting recessions remains dismal. There are several competing hypotheses for why forecasters fail to foresee recessions, but little evidence any of them are going to be addressed before the next recession occurs. This suggests planners and policymakers should expect to be surprised by the arrival of downturns and develop ways to be prepared for recessions without having clear warning of their coming. |
| Keywords: | Forecast evaluation, recessions |
| JEL: | E37 C53 |
| Date: | 2019–08 |
| URL: | http://d.repec.org/n?u=RePEc:gwc:wpaper:2019-003&r=all |
| By: | Nassim Nicholas Taleb |
| Abstract: | What do binary (or probabilistic) forecasting abilities have to do with overall performance? We map the difference between (univariate) binary predictions, bets and "beliefs" (expressed as a specific "event" will happen/will not happen) and real-world continuous payoffs (numerical benefits or harm from an event) and show the effect of their conflation and mischaracterization in the decision-science literature. We also examine the differences under thin and fat tails. The effects are: A- Spuriousness of many psychological results particularly those documenting that humans overestimate tail probabilities and rare events, or that they overreact to fears of market crashes, ecological calamities, etc. Many perceived "biases" are just mischaracterizations by psychologists. There is also a misuse of Hayekian arguments in promoting prediction markets. We quantify such conflations with a metric for "pseudo-overestimation". B- Being a "good forecaster" in binary space doesn't lead to having a good actual performance}, and vice versa, especially under nonlinearities. A binary forecasting record is likely to be a reverse indicator under some classes of distributions. Deeper uncertainty or more complicated and realistic probability distribution worsen the conflation . C- Machine Learning: Some nonlinear payoff functions, while not lending themselves to verbalistic expressions and "forecasts", are well captured by ML or expressed in option contracts. D- Fattailedness: The difference is exacerbated in the power law classes of probability distributions. |
| Date: | 2019–07 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:1907.11162&r=all |