E37 Forecasting and Simulation
Refine
Document Type
- Working Paper (2)
Language
- English (2)
Has Fulltext
- yes (2)
Is part of the Bibliography
- no (2)
Keywords
- Animal spirits (1)
- Bruttoinlandsprodukt (1)
- Business cycle (1)
- European Central Bank (1)
- European Union (1)
- Europäische Union (1)
- Europäische Zentralbank (1)
- Eurozone (1)
- Forecasting (1)
- Geschäftszyklus (1)
Institute
This paper examines the role of uncertainty in the context of the business cycle in the Eurozone. To gain a more granular perspective on uncertainty, the paper decomposes uncertainty along two dimensions: First, we construct the four different moments of uncertainty, including the point estimate, the standard deviation, the skewness and the kurtosis. The second dimension of uncertainty spans along three distinct groups of economic agents, including consumers, corporates and financial markets. Based on this taxonomy, we construct uncertainty indices and assess the impact on real GDP via impulse response functions and further investigate their informational value in rolling out-of-sample GDP forecasts. The analysis lends evidence to the hypothesis that higher uncertainty expressed through the point estimate, a larger standard deviation among confidence estimates, positive skewness and a higher kurtosis are all negatively correlated with the business cycle. The impulse response functions reveal that in particular the first and the second moment of uncertainty cause a permanent effect on GDP with an initial decline and a subsequent overshoot. We find uncertainty in the corporate sector to be the main driver behind this observation, followed by financial markets’ uncertainty whose initial effect on GDP is comparable but receding much faster. While the first two moments of uncertainty improve GDP forecasts significantly, both the skewness and the kurtosis do not augment the forecast quality any further.
This paper examines the recession probabilities for the Eurozone along four different dimensions: First, we identify the best performing indicators for a recession within the next 12 months based on 43 underlying single variables and their different transformations in a benchmark model. We find that a modified version of the yield curve incorporating the shadow interest rate removes the downward rigidity of the front-leg and restores part of the informational content of the term spread at the zero lower bound. However, the best performing single indicator of the benchmark model is Real M1 followed by the Purchasing Managers Index (PMI), the investment grade corporate bond spread and the Terms of Trade. Second, the paper establishes three submodels to increase the lead-time and the stability of recession models: (i) Monetary transmission channels via principal component analysis; (ii) Bivariate regressions to identify paramount combinations; (iii) Unstable surges vis-à-vis the Hodrick-Prescott trend to detect animal spirits and hawkish mistakes. Third, the analysis is extended over various forecasting horizons (6m, 18m and 24m). Fourth, the results are analyzed from the perspective of risk-affine and risk-averse investors.