Firms’ investment is a key driver of economic growth, yet corporate investment in Europe has remained subdued since the Covid-19 pandemic (Revoltella et al. 2024). One possible explanation lies in how firms perceive the probability of severe downturns in the future. A growing empirical literature shows that the first moment of expectations (the mean of the outlook) and the second moment (uncertainty around it) both matter for firms’ investment decisions (Gennaioli et al. 2016, Kumar et al. 2023). At the same time, macroeconomic tail risk plays a central role in many quantitative asset pricing and business cycle models (Barro 2006, Gourio 2012, Kozlowski et al. 2020). What has been missing is direct firm-level evidence on whether – and how – beliefs about rare disasters influence real corporate decisions. These macro tail risk beliefs might help explain why corporate investment dynamics in the euro area remains weak.

In a new paper (Menkhoff 2026), I fill this gap using novel survey data from the ifo Business Survey, a representative monthly panel of about 6,000 German firms typically answered by the CEO or owner. I combine observational panel variation, an information-provision randomised control trial (RCT), and hypothetical scenarios to pin down the role of macro tail risk beliefs in firms’ investment.

Firms think tail events are fairly likely and disagree a lot

Starting in October 2022, I asked firms about the probability that four concrete macro tail events occur within the next five years: a drop in annual GDP of at least 5%, a financial crisis of 2008/2009 magnitude, a COVID-sized pandemic, and an escalation of geopolitical conflicts. I repeated the questions every October through 2025.

Three facts stand out. First, firms assign non-negligible probabilities to these events: in October 2022 the median firm expected a large GDP drop with 50% probability over the following five years. While the probability decreases, it remains elevated in the following years, as shown in Figure 1, Panel a. Second, beliefs are widely dispersed: in October 2025 the interquartile range spans 10% to 40% for the GDP drop scenario. Third, beliefs are strongly revised within firms over time: p10-to-p90 revisions regularly range from −40 to +30 percentage points year-to-year as visible in Panel b.

Higher macro tail risk beliefs correlate with more pessimistic business outlooks (first moments) and higher subjective uncertainty (second moments). But, as I show below, they also matter beyond shifts in first and second moments for firms’ actions.

Figure 1 Distribution of firms’ macroeconomic tail risk expectations over time

a) Distribution of subjective probabilities

b) Distribution of year-to-year revisions of subjective probabilities

Sample: ifo Business Survey, October waves 2022–2025.

A second layer: ambiguity about firm-specific exposure

Macro tail risk for a firm has a second layer: conditional on an event, how badly would the firm itself be hit? I asked firms about their expected change in real sales in a hypothetical 5% GDP drop, eliciting best-case, average-case, and worst-case scenarios. The responses reveal wide heterogeneity and sizable uncertainty: the median firm sees a 10% sales decline in the average case, with a 20-percentage-point gap between best and worst cases.

An open-ended follow-up reveals why: 93% of firms who comment on assessability say predicting exposure is “difficult” or “(almost) impossible”; when they do provide numbers, they more often quote ranges or max/min scenarios than point estimates; and they repeatedly anchor on a small set of past experiences (the Great Recession, COVID-19).

This combination, firms have scenarios in mind but cannot confidently assign probabilities to them, is the textbook description of Knightian uncertainty or ambiguity (Ilut and Schneider 2022). Under ambiguity aversion, firms act as if the worst-case exposure will be realized. This yields a testable prediction: an increase in the tail event probability should lower investment beyond first- and second-moment channels, concentrated among firms with high expected worst-case exposure.

Higher macro tail risk beliefs strongly reduce investment

I link firms’ October tail-risk beliefs to their investment plans for the following year, elicited in November. These plans map tightly into balance-sheet data: a planned cut corresponds to an average 4–5% decline in the capital stock the next year.

Figure 2 Firms’ macro tail risk expectations and investment plans

Notes: Binned scatterplots. Time fixed effects absorbed. Panel (b) additionally controls for business expectations and uncertainty. Sample: ifo Business Survey, October/November waves 2022–2025.

A one standard deviation (≈30 percentage points) increase in the subjective probability of a large GDP drop predicts a 6.3 percentage point higher probability of cutting investment, which is a 20% increase relative to the unconditional mean as shown in the binned scatter plot in Figure 2, Panel a. Conditioning on business expectations and uncertainty flattens the slope, but only by less than half: macro tail risk beliefs retain substantial predictive power beyond first- and second-moment channels (see Panel b).

The result is not driven by fixed firm heterogeneity. Within-firm revisions of macro tail risk beliefs deliver the same pattern. Consistent with the ambiguity mechanism, the effect is concentrated among firms with high expected worst-case exposure. Upward belief revisions also predict 7–10% lower realised investment the following year.

Two experiments support a causal interpretation. In an RCT, I inform a random half of firms about the average macro tail risk beliefs of other managers (28%), elicited four months earlier in an unpublished survey wave. Updating is asymmetric: firms with low priors revise upward substantially, while firms with high priors do not revise downward, consistent with conservative updating of ambiguous signals (Epstein and Schneider 2008). Focusing on the firms that move, those with below-median priors raise their subjective probability of a large GDP drop by about 2.8 percentage points (roughly 40% of the prior mean). One month later, these same firms are 8 percentage points more likely to plan an investment cut, a 37% increase relative to the control group mean. The effect is only modestly attenuated by controlling for first and second moments and is concentrated among firms with high worst-case exposure. In hypothetical vignettes that fix the mean and variance of macro outlooks and vary only the skewness, firms prefer the symmetric outlook with low tail risk, and the preference is strongest among those reporting the most negative worst-case exposure, the mechanism predicted by the ambiguity model.

Policy: The case for credible symmetric corporate taxation

I embed ambiguous firm-specific exposure into a standard heterogeneous firm model, disciplining the worst-case exposure with the survey data. A tail-risk news shock (a rise in the probability of a disaster, without the disaster itself) generates a decline in investment: the median firm cuts investment by about 10% on average over the first four quarters, with a fat left tail of firms cutting by 25% or more.

I then compare two stabilisation policies: a state-contingent interest rate cut, and a state-contingent switch from asymmetric to symmetric corporate taxation (similar to generous loss carrybacks). Both cushion the investment decline, but with a striking difference. The interest rate cut works about equally well in worlds with and without ambiguous exposure. The fiscal policy, by contrast, mitigates the capital decline by about 20% under ambiguity and is largely ineffective without it. The reason: under ambiguity, firms evaluate investment under their worst-case scenario and symmetric taxation provides exactly the insurance that scales with worst-case losses.

An important qualifier: these policies must be credibly announced as rule-based, state-contingent responses. Discretionary subsidies promised after the event would not shift firms’ worst-case scenario ex ante, which is where the investment decision is anchored.

Takeaways

Macroeconomic tail risk is a distinct dimension of firm expectations, not captured by standard sentiment measures, with first-order implications for investment. Policymaker and central bank communication that anchors firms’ beliefs about the likelihood and severity of disasters is a cheap lever in turbulent times. And for stabilization, automatic stabilizers beat ad hoc interventions: a rule-based switch to symmetric corporate taxation in tail events is particularly effective precisely because it shifts the worst-case scenario that ambiguity-averse firms act on. Tail events may be rare, but firms think about them and those thoughts show up in the capital stock long before any disaster arrives.

References

Barro, R J (2006), “Rare disasters and asset markets in the twentieth century”, Quarterly Journal of Economics 121(3): 823–866.

Epstein, L G, and M Schneider (2008), “Ambiguity, information quality, and asset pricing”, Journal of Finance 63(1): 197–228.

Gennaioli, N, Y Ma, and A Shleifer (2016), “Expectations and investment”, NBER Macroeconomics Annual 30: 379–431.

Gourio, F (2012), “Disaster risk and business cycles”, American Economic Review 102(6): 2734–2766.

Ilut, C, and M Schneider (2022), “Modeling uncertainty as ambiguity: A review”, in R Bachmann, G Topa, and W van der Klaauw (eds), Handbook of Economic Expectations, Elsevier.

Kozlowski, J, L Veldkamp, and V Venkateswaran (2020), “The tail that wags the economy: Beliefs and persistent stagnation”, Journal of Political Economy 128(8): 2839–2879.

Kumar, S, Y Gorodnichenko, and O Coibion (2023), “The effect of macroeconomic uncertainty on firm decisions”, Econometrica 91(4): 1297–1332.

Menkhoff, M (2026), “The devil is in the tail: Macroeconomic tail risk expectations of firms”, CESifo Working Paper 11848.

Revoltella, D, P Harasztosi, J Delanote,  T Bending , and M C André (2024), “Corporate investment in Europe: A snapshot from the 2024 EIB Investment Survey”, VoxEU.org, 19 November