Official fiscal forecasts in EU member states under the European Semester and Fiscal Compact – An empirical assessment

The efficacy of official forecasts in the EU has been under the spotlight since the introduction of the euro, with biases widely reported prior to the 2008–12 financial and sovereign bond market crisis. Changes to the EU fiscal rules and procedures, in the form of the European Semester and Fiscal Compact, in the early 2010s were adopted to improve forecasting, including through providing a role for independent fiscal institutions. Using data for 22 countries between 2013 and 2019, this paper shows that, despite these changes, biases, of a pessimistic form, remain in forecasts of budget balance and output variables in Stability and Convergence Programmes and the European Commission's Spring Forecasts. Econometric analysis indicates forecast errors in both the headline budget balance and the structural budget balance being explained by forecast errors in output variables and by EU fiscal rule requirements. Member states under an excessive deficit procedure provide optimistic headline budget balance forecasts compared to non-EDP countries, while those that have not met their medium-term objective report smaller forecast errors for the structural budget balance. Independent fiscal institutions are linked to a smaller bias to forecasts of the structural budget balance but have no effect on the forecast errors of the headline budget balance.     

News Shocks,Business Cycles,and the Disinflation Puzzle

We argue that key findings of the empirical literature on the effects of news about future technology—including their tendency to generate negative comovement of macro-economic aggregates, and their puzzling disinflationary nature—are due to measurement errors in total factor productivity (TFP). In this paper, we estimate the macro-economic effects of news shocks in the United States using an agnostic identification approach that is robust to measurement errors. We find no evidence of negative comovement conditional on a news shock, and the disinflation puzzle essentially vanishes under our identification strategy. Our results also indicate that news shocks have become an important driver of business-cycle fluctuations in recent years.     

The association between quarter length,forecast errors,and firms’ voluntary disclosures

Approximately 60 percent of adjacent fiscal quarters contain a different number of calendar days. In preliminary analyses, we find the change in quarter length is significantly associated with the changes in sales and earnings and that analysts condition on the prior quarter's results when making their forecasts. These results indicate that it is important for analysts to adjust for changes in quarter length when making forecasts. However, we find the quarterly change in days is positively associated with analysts’ sales and earnings forecasts errors, where forecast error equals the actual earnings minus the forecasted earnings. These results indicate that analysts systematically underestimate (overestimate) performance when quarter length increases (decreases). We find evidence indicating investors make similar errors as returns around earnings announcements are positively associated with the change in quarter length, but only when changes in firm performance is more sensitive to changes in quarter length. Corroborating these findings, managers are more (less) likely to discuss quarter length during conference calls when quarter length decreases (increases). These results are consistent with managers’ strategic disclosure incentives. In summary, our evidence suggests analysts and investors fail to fully take account of the quasi-mechanical effect that quarter length has on firm performance and managers strategically alter their voluntary disclosures to take advantage of these failures.     

Time-variation,multiple testing,and the factor zoo

Bayesians circumvent the need for significance threshold correction when multiple testing and we recommend controlling the Type-S (sign), rather than the Type-1, error rate because it yields more reliable frequency properties for inferences. Our unified Bayesian framework, with theory-informed priors, identifies two breaks (2001 and 2008) in our 1980–2018 sample period. After each break the set of characteristics changes, and only market beta is selected in all regimes. In a portfolio application, the method generates significantly larger Sharpe ratios after transaction costs than a range of benchmark methods, including the same model that uses a Type-1 (not Type-S) error framework.