A comprehensive evaluation of macroeconomic forecasting methods

We employ datasets for seven developed economies and consider four classes of multivariate forecasting models in order to extend and enhance the empirical evidence in the macroeconomic forecasting literature. The evaluation considers forecasting horizons of between one quarter and two years ahead. We find that the structural model, a medium-sized DSGE model, provides accurate long-horizon US and UK inflation forecasts. We strike a balance between being comprehensive and producing clear messages by applying meta-analysis regressions to 2,976 relative accuracy comparisons that vary with the forecasting horizon, country, model class and specification, number of predictors, and evaluation period. For point and density forecasting of GDP growth and inflation, we find that models with large numbers of predictors do not outperform models with 13–14 hand-picked predictors. Factor-augmented models and equal-weighted combinations of single-predictor mixed-data sampling regressions are a better choice for dealing with large numbers of predictors than Bayesian VARs.     

Do foreign exchange forecasters apply asymmetric loss functions? Evidence from three major exchange rates

This article investigates which type of loss function is consistent with the hypothesis that major exchange rate forecasts, i.e. the euro, the British pound, and the Japanese yen vis-à-vis the US dollar, are rational. We apply a comprehensive data set, which also allows us to examine different forecast horizons and heterogeneity of forecasters.     

Political forecast cycles

A moral hazard model is used to show why overly optimistic revenue forecasts prior to elections can be optimal: Opportunistic governments can increase spending and appear more competent; ex post deficits emerge in election years, thereby producing political forecast cycles – as also found for US states in the empirical literature. Additionally, we obtain three theoretical results which are tested with panel data for Portuguese municipalities. The extent of manipulations is reduced when (i) the winning margin is expected to widen; (ii) the incumbent is not re-running; and/or (iii) the share of informed voters (proxied by education) goes up.     

Knowing is half the battle: Seasonal forecasts,adaptive cropping systems,and the mediating role of private markets in Zambia

This paper examines how smallholders living in regions where a drought is forecasted adapt their farm practices in response to receiving seasonal forecast information. The article draws on a unique longitudinal dataset in Zambia, which collected information from farm households before and after a significant drought caused by the 2015/2016 El-Niño Southern Oscillation. It finds that farmers residing in areas forecasted to be drought-affected and receiving seasonal forecast information are significantly more likely to integrate drought tolerant crops into their cropping systems compared to similar households not receiving this information. Moreover, the probability that a farmer implements these adaptive farm management strategies in response to seasonal forecast information is found to increase substantially as the number of private grain buyers in the farmers’ village increases. This suggests that climate change adaptation and resilience strategies that integrate the generation and dissemination of weather information with agricultural market development can achieve greater impact on farmers’ adaptive responses than approaches that treat these activities in isolation.     

Forecast-error-based estimation of forecast uncertainty when the horizon is increased

Past forecast errors are employed frequently in the estimation of the unconditional forecast uncertainty, and several institutions have increased their forecast horizons in recent times. This work addresses the question of how forecast-error-based estimation can be performed if there are very few errors available for the new forecast horizons. It extends the results of Knüppel (2014) in order to relax the condition on the data structure that is required for the SUR estimator to be independent of unknown quantities. It turns out that the SUR estimator of the forecast uncertainty, which estimates the forecast uncertainty for all horizons jointly, tends to deliver large efficiency gains relative to the OLS estimator (i.e., the sample mean of the squared forecast errors for each individual horizon) in the case of increased forecast horizons. The SUR estimator is applied to the forecast errors of the Bank of England, the US Survey of Professional Forecasters, and the FOMC.     

A hybrid machine learning model for forecasting a billing period’s peak electric load days

Many models have been studied for forecasting the peak electric load, but studies focusing on forecasting peak electric load days for a billing period are scarce. This focus is highly relevant to consumers, as their electricity costs are determined based not only on total consumption, but also on the peak load required during a period. Forecasting these peak days accurately allows demand response actions to be planned and executed efficiently in order to mitigate these peaks and their associated costs. We propose a hybrid model based on ARIMA, logistic regression and artificial neural networks models. This hybrid model evaluates the individual results of these statistical and machine learning models in order to forecast whether a given day will be a peak load day for the billing period. The proposed model predicted 70% (40/57) of actual peak load days accurately and revealed potential savings of approximately USD $80,000 for an American university during a one-year testing period.     

ALICE: Composite leading indicators for euro area inflation cycles

This paper develops Area-wide Leading Inflation CyclE (ALICE) indicators with the aim to provide early signals for turning points in the euro area headline and core inflation cycles. The headline (core) ALICE leads the reference cycle by three (four) months and the lead times extend to five (nine) months based on longer-leading series. Both ALICE indicators signalled turning points in the inflation cycles ex post and also performed well in a simulated real-time exercise. Moreover, in particular the headline ALICE also appeared to be useful for quantitative forecasting of the direction and level of inflation.     

How efficient is the market for Australian firms’ earnings information?

We construct a measure of the speed with which forecasts issued by sell-side analysts accurately forecast future annual earnings. Following Marshall, we label this measure earnings information flow timeliness (EIFT). This measure avoids the aggregation problem inherent in price-based measures of information efficiency. We document large variation in EIFT across firm-years, and show that EIFT is positively associated with the extent of analyst following, consistent with increased analyst coverage improving the speed with which earnings-related information is recognised. We also find that EIFT is higher for firm-years classified as ‘bad news’ (i.e., where analysts’ forecasts at the start of the financial period exceed the reported outcome). However, when we separately consider instances where analysts appear to forecast non-GAAP (or ‘street’) earnings rather than GAAP earnings, we find that the greater timeliness of bad news is concentrated among observations where analysts forecast non-GAAP earnings, where unusual items are typically excluded. We conclude that the market for accounting information is more efficient for negative operating outcomes than for negative outcomes reflecting unusual items.     

Can incentives mitigate student overconfidence at grade forecasts?

Research shows that college students exhibit bias in their forecasts of exam performance. Most students are overconfident in their forecasts, academically weaker students are the most overconfident, and top-performing students are underconfident. The literature identifies negative repercussions of these biases, including inadequate preparation for exams. A recurring attribute of this literature is the absence of meaningful incentives for students to forecast accurately. We implement an extra credit scheme to incentivize accurate forecasts. Depending on how forecast bias is measured, the scheme mitigates bias by top-performing students and marginally mitigates bias by other students. Our results have several implications. First, we illustrate an extra credit tool instructors can use to incentivize students to make more thoughtful assessments of their future exam performance. Second, we show how the association between incentives and forecast bias differs across student groups. Finally, we show that results in this literature are sensitive to how bias is measured.     

Out-of-sample density forecasts with affine jump diffusion models

We conduct out-of-sample density forecast evaluations of the affine jump diffusion models for the S&P 500 stock index and its options’ contracts. We also examine the time-series consistency between the model-implied spot volatilities using options & returns and only returns. In particular, we focus on the role of the time-varying jump risk premia. Particle filters are used to estimate the model-implied spot volatilities. We also propose the beta transformation approach for recursive parameter updating. Our empirical analysis shows that the inconsistencies between options & returns and only returns are resolved by the introduction of the time-varying jump risk premia. For density forecasts, the time-varying jump risk premia models dominate the other models in terms of likelihood criteria. We also find that for medium-term horizons, the beta transformation can weaken the systematic effect of misspecified AJD models using options & returns.