Session 17: The Macroeconomics of Uncertainty and Volatility

We analyze how tariff uncertainty affects exchange rates, motivated by the US dollar’s depreciation after the 2025 tariff announcements. Standard macrotrade models predict that unilateral tariffs appreciate the implementing country’s currency, but we show this result can be overturned by policy uncertainty. We build a two-country general equilibrium model with risk-averse agents and segmented financial markets, where tariff volatility enters uncovered interest parity through a risk-premium wedge. Higher tariff uncertainty increases precautionary savings and risk premia, leading to immediate currency depreciation even as tariffs rise. Quantitatively, the model replicates the size and timing of the observed dollar depreciation episode dynamics.

As the risks of government bonds change over time, does this change the compensation that investors require for holding them? While realized excess returns show little relation to bond risk, we find that subjective expected excess returns constructed from professional forecasts of future long-term yields are tightly linked to bonds’ stock mar-ket betas, consistent with a CAPM-style relation. In a U.S. month-by-maturity panel from 1988–2024, the correlation of subjective excess returns with rolling bond–stock betas is 66%. The estimated market price of risk is comparable to the equity premium and stable when controlling for time and maturity fixed effects. Realized excess returns are predicted by subjective excess returns, but this predictability is driven by higher-frequency variation and not by betas. Similar results hold in an international panel of developed countries from 1989–2024. The change in betas from positive to negative accounts for half of the decline in long-term U.S. Treasury yields from the 1980s to the 2010s and implies a negative term premium as early as 2001. During quantitative easing episodes, the price of bond risk declines, suggesting an increased investor willingness to bear risk.

The source of uncertainty determines its effect on prices: demand uncertainty is disinflationary, supply uncertainty is inflationary, a distinction existing measures conflate. Using firm-level survey data from Italy (INVIND) and the United States (SBU), we measure demand uncertainty as the dispersion of expected sales growth and supply uncertainty as the dispersion of expected price growth, and instrument both using a shift-share design based on industry exposure to economic policy uncertainty and oil price volatility. A one-standard-deviation increase in demand uncertainty reduces firm price growth by 0.3 percentage points and contracts real sales; the same increase in supply uncertainty raises price growth by 0.6 percentage points while also contracting real sales. Monetary policy transmission inherits this state-dependence: under supply uncertainty, price pass-through to an accommodative shock rises by 25 percent and its effect on real output falls by 40 percent; under demand uncertainty, the pattern reverses. Both results replicate in the United States.

Using minute-level wearable-device data on 51,191 adults in the National Institutes of Health’s (NIH) All of Us Research Program, covering over 19 million person-nights, we show that sleep quality is a high-frequency biomarker of macroeconomic and financial uncertainty. Higher uncertainty predicts lower deep sleep and sleep efficiency for several nights, a “wake-and-see” effect that complements the classic “wait-and-see” channel. The effects are robust to environmental, physiological, and behavioral controls, are stronger for uncertainty than for first-moment market shocks, and are larger in higher-income areas. We also document a reverse channel from physiology to markets: nights of unusually poor sleep predict weaker opening-hour equity returns, lower next-day market liquidity, and higher realized volatility, with stronger effects in retail-dominated stocks. Finally, poor sleep positively relates to the conditional market price of aggregate risk, consistent with lower aggregate risk-bearing capacity. Overall, population sleep provides a high-frequency physiological state variable linking uncertainty, human capital, and market functioning in real time.

We construct a new data-driven measure of U.S. trade flow uncertainty (TFU), based on the unforecastability of trade flows, and contrast it with the widely used news-based measure of trade policy uncertainty (TPU). Using threshold VARs and smooth transition local projections, we uncover a striking asymmetry: TPU shocks decrease investment, production, and employment only when TFU is already elevated, but these shocks have little real effect when underlying trade flows are stable. Firm-level evidence confirms this state dependence and uncovers an amplifying role for financial constraints. To interpret these results, we develop and calibrate a model in which high TFU amplifies the real effects of TPU through two channels: a financial channel where high TFU makes borrowing constraints more likely to bind, magnifying firms’ investment responses, and a belief channel where firms place more weight on noisy signals about future trade policy when trade flows are volatile.

We investigate whether the conduct of monetary policy can contribute to secular trends in macro-finance since the 1980s: the decline in long-term real interest rates, the rise in private debt, and the fall in capital investment. We develop a tractable general equilibrium model in which heterogeneous agents face aggregate and idiosyncratic risk and make portfolio decisions between nominal bonds and real capital. By committing to aggressive inflation stabilization, central banks compress the risk premia on nominal bonds through two distinct channels: a reduction in inflation risk and a reduction in interest rate risk. Savers reallocate toward nominal bonds: private debt expands and capital investment is crowded out, yet the return to capital does not fall. The model provides a unified mechanism linking these trends to credible inflation targeting.

I study the dynamic transmission across a network of shocks with uncertain dynamic properties using tools from robust control theory to measure system performance. I characterize the response of networks to (1) specified stochastic shocks, where performance is measured by output volatility, or (2) uncertain but norm-bounded inputs, where robust performance is worst-case shock amplification. I apply the framework to production networks, showing how dynamic shock amplification depends on network topology, aggregation, and the directionality of input-output linkages. I first develop some analytic examples, then apply the methods to data on the US production network. I find that over time the US economy has become more fragile and subject to volatility coming through the production network. This reflects both the growing concentration of activity in services, real estate, and finance, and the increased importance of these sectors as inputs.

The CBOE Volatility Index (VIX) is a widely used measure of market uncertainty that has become a foundational component of dynamic macroeconomic models. This paper shows that a simpler, more transparent metric—the normalized price of an at-the-money forward put option or Downside Protection Premium (DPP)—tracks the VIX almost perfectly (0.99 monthly correlation) and permits a direct decomposition of uncertainty into the probability of a market shortfall and the conditional severity of that shortfall. Using an entropic tilting procedure to recover the risk-neutral density, I document that nearly all time-series variation in downside risk during crises is driven by changes in the severity of shortfall, and not by changes in the probability of shortfall. Vector autoregressions show that standard macro “uncertainty shocks” are empirically indistinguishable from shocks to the DPP, suggesting that the relevant variation is best interpreted as tail-severity repricing. I then embed this distinction into a standard RBC model. A novel downside-severity shock process generates a much larger, more persistent, and more dispersed contractionary response to bad shocks vis-a-vis standard time-varying uncertainty.

Why do stocks and bonds sometimes hedge each other and sometimes move together? We show that an unexplored explanation lies in the source of uncertainty about the future path of monetary policy. Using daily revisions in policy rate expectations, we decompose a novel market-based measure of policy path uncertainty into macroeconomic uncertainty and uncertainty about the Federal Reserve’s reaction function. These two components generate sharply different macroeconomic dynamics. Macroeconomic uncertainty lowers output and inflation, prompting monetary policy easing, while reaction function uncertainty produces stagflationary pressures and monetary policy tightening. These contrasting dynamics carry through to financial markets. Both uncertainty sources reduce equity valuations, but macroeconomic uncertainty lowers Trea-sury yields whereas policy response uncertainty raises yields through higher real term premia. Consequently, stock-bond correlations turn negative when macroeconomic uncertainty dominates and positive when reaction function uncertainty prevails. Our decomposition thus provides a structural explanation for why the stock-bond correlation switches sign over time.

We show that government policies interact in ways that shape how capital markets price risk. We introduce the concept of fiscal insurance: a mechanism through which certain fiscal tools are perceived by investors to mitigate the risks created by other government policies. Exploiting the 2018–19 U.S. tariff shocks and concurrent federal procurement spending, we find that firms facing higher tariff exposure earn higher risk premia, but this effect is substantially attenuated for firms receiving greater procurement. Procurement itself increases more for politically connected and economically vulnerable firms, revealing both political and economic channels of fiscal insurance. Overall, our evidence documents the existence of fiscal insurance, a cross-policy mechanism, with implications for firm valuation and real activity.

Over 19% of workers in the U.S. are foreign-born, raising the possibility that foreign geopolitical risk affects U.S. firm productivity through immigrant workers. Using restricted Census employee–employer matched microdata, we demonstrate this channel by constructing each firm’s labor exposure to foreign geopolitical risk (FGPR) based on geopolitical shocks to the origin countries of its immigrant employees. We find that FGPR significantly increases firm labor productivity, profitability, and investment. This positive causal effect is robust and distinct from the trade channel, which typically disrupts U.S. firms. Further worker-level evidence shows that immigrants increase working hours when geopolitical risk rises in their origin countries, driven by reduced return-migration prospects and stronger remittance motives. This response positively spills over to domestic coworkers by increasing their labor supply. Overall, our findings identify a novel channel through which foreign geopolitical risk pressures immigrant workers while benefiting their employing firms.

We develop a model of product switching over the firm lifecycle in which firms learn about product-specific expertise and adjust their product portfolios accordingly. Firms collect and process information to extract signals, and their information- processing capacity improves over time. However, limited data processing ability early in the lifecycle generates noisier beliefs and distinctive product-switching behavior among young firms: young firms are more likely to retain poorly matched products and drop better-matched ones, slowing the reallocation of resources toward high-performing products. This mechanism introduces a novel within-firm margin of resource allocation through product switching that complements the standard firm selection channel and provides a framework for evaluating policies such as product search and entry subsidies. Using firm–product-level administrative data and the Management and Organizational Practices Survey (MOPS) from the U.S. manufacturing sector, we find product switching patterns across the firm lifecycle and managerial ability that are consistent with the model’s predictions.

This paper investigates the fundamental trade-off between firm specialization and diversification. While management theories emphasize core competency to maximize efficiency, finance favors diversification to mitigate cash flow volatility. We develop a heterogeneous-firm general equilibrium model to quantify these forces. We devise a novel empirical approach, grounded in theory, to measuring them. We first extract granular revenue shares from 10-K filings, using a large language model. Next, we use the co-movement between sectoral aggregates as (an inverse) measure of the economic distance between them. The observed relationship between scope and markups at the firm-level pins down the price of risk. These sufficient statistics also recover productivity at the firm-sector level. The results reveal that risk considerations lower aggregate output significantly (by 10%). We also find that firms are more likely to operate in related sectors, where they tend to have higher productivity on average.

We examine how central banks can effectively communicate forecast uncertainty in a two-part experimental study. Part I tests how different visual media – fan charts, dot plots, box-and-whisker plots, speedometers, and ranges – communicate uncertainty to both the general public and expert audiences. We find that fan charts are well under- stood and perform best at jointly conveying both expectations and uncertainty. Part II implements a novel dynamic information experiment with 1,600 UK participants across four stages, examining the effects of uncertainty communication on expectations and uncertainty perceptions over time. We find that while point forecasts anchor expectations marginally more than fan charts initially, forecast errors significantly de-anchor expectations, particularly for “unlucky” errors that move inflation away from target. Critically, fan charts materially mitigate this de-anchoring, acting as an “insurance policy” that helps protect central bank reputation. We also document that the public consistently underestimates the degree of uncertainty, and that communicating uncertainty via fan charts helps the public learn more realistic uncertainty perceptions. Our findings have important implications for central bank communication strategies.

We analyze firms’ perceptions of their cost of capital using hand-collected data. We show that firms with a higher perceived cost of capital invest less and earn higher returns on invested capital, suggesting that the perceived cost of capital determines the long-run allocation of capital. In inefficient markets, firms aiming to maximize their current market value should set their perceived cost of capital equal to the expected returns on their debt and equity. We strongly reject this market-value maximization benchmark, as little variation in firms’ perceived cost of capital can be explained by variation in market-based expected returns. Alternatively, firms may aim to maximize their fair value, which is the value corrected for financial market inefficiencies. We find evidence in favor of this approach, as a fundamental risk benchmark explains half the variation in the perceived cost of capital. Most variation in the perceived cost of capital can be explained by firms incorporating investors’ biased return expectations, which is inconsistent with firms maximizing their current market value. Using a quantitative model, we find that distortions in the perceived cost of capital can generate substantial capital misallocation and thereby reduce aggregate productivity.

We study the role of macroeconomic disagreement in explaining the term structure of interest rates, accounting for evolving beliefs regarding the structure of the economy. To capture aggregate macroeconomic disagreement, we construct a summary measure that reflects the common component of the dispersion in survey responses on forecasts of multiple economic variables. We show that disagreement has important predictive power for excess bond returns, beyond the information in yields and macroeconomic factors. Introducing disagreement in a shadow-rate macro-finance term structure model estimated with real-time data improves the fit to the cross section of yields, and exhibits better performance in out-of-sample predictions of future yield movements. The model suggests that observable macroeconomic disagreement is an important factor explaining term-premia.

We introduce an improved measure of geopolitical risk that builds on Caldara and Iacoviello (2022) and uses artificial intelligence to evaluate newspaper content. Our approach replaces keyword matching with semantic understanding: instead of searching for specific word combinations, we use one of the flagship ChatGPT models (GPT-4o mini) to read newspaper articles and assess their geopolitical risk intensity. The daily AI-GPR index scores about 5 million articles from the New York Times, Washington Post, and Chicago Tribune from 1960 through 2025. The approach reduces false positives from articles mentioning war or terrorism in non-geopolitical contexts while capturing relevant articles that lack exact or common dictionary terms. The AI-GPR index also assigns gradations of risk intensity rather than simple yes-or-no classifications, providing more nuanced measurement even at high frequencies. We demonstrate the potential of our approach with several applications. The AI-GPR index produces a stronger and more precisely estimated negative effect of geopolitical risk on stock returns. Combined with a second LLM classification layer, it yields a historical time series of geopolitical risk-driven oil supply disruptions by region. A third classification layer maps directed networks of geopolitical actors—initiators, respondents, and spillover countries—across major historical episodes. Finally, the layered architecture decomposes geopolitical risk by event type, by country, and by bilateral country pair.

While many recent papers have demonstrated the effects of volatility on firms, there has been less attention on the effects of volatility on workers, and on which workers bear this form of risk. Using linked employee-employer data from the United States, we document causal evidence that increases in an employer’s stock price volatility negatively affect workers’ earnings growth. Low-earning workers experience the largest losses, especially workers who separate from their employers. To study the macroeconomic effects of an increase in firm dispersion, we use a dynamic contracting model with worker moral hazard, fixed per-worker operating costs, and human capital scarring. Our calibrated model qualitatively and quantitatively matches our empirical findings. We feed the path of idiosyncratic firm volatility observed around the financial crisis into our model and find that this alone accounts for 1.06 percentage points (19 percent) of the rise in unemployment between 2007-2009. Taken together, these findings lend support to the connection between increases in cross-sectional firm dispersion in times of crisis and adverse labor market outcomes.

We introduce a class of uncertainty measures, Entropic Uncertainty, constructed by applying the factor-augmented stochastic-volatility estimator of Jurado et al. (2015) (JLN) to a novel input: the Shannon entropy of the directed weighted graph that double-entry bookkeeping induces on a firm’s financial statements each quarter. Recast as a graph, the financial statements that aggregate uncertainty measures usually overlook become a rigorous, structurally disciplined record of the firm’s internal resource allocation. By applying graph theory to firms’ audited 10-Q and 10-K disclosures, the resulting bookkeeping graph captures, at the level of the firm’s primary economic functions, how a firm allocates its internal resources, and the graph entropy summarizes the distribution of such within-firm resource flows. Running firm-level entropy through the JLN factor-augmented forecasting regression and Bayesian stochastic-volatility filter yields a firm-quarter measure of the conditional volatility of the unforecastable component of within-firm structural composition, which we aggregate into an aggregate Entropic Uncertainty Index (EUI). EUI carries incremental predictive power for firm investment, horse-racing favorably with existing uncertainty measures in both panel and time-series specifications. A structural decomposition of the bookkeeping graph entropy delivers channel-specific uncertainty readings whose differential predictive performance is itself informative. Collectively, the results establish financial-statement structure as an internally grounded, theoretically disciplined, and empirically rich source of uncertainty measurement.

Did forecasters believe the 2021 surge in inflation was transitory or persistent? For many, the answer was neither: they did not know. A forecaster who is certain about persistence reports forecasts that decay geometrically toward their long-run view, while a forecaster who is uncertain about persistence reports forecasts that decay more slowly. The size of that discrepancy identifies their persistence uncertainty. Using point and density forecasts from the US Survey of Professional Forecasters, I find that 43 percent of forecasters are uncertain about persistence, while the rest report forecasts indistinguishable from certainty. Only this uncertain minority revises their beliefs as a Bayesian would, and the size of their revisions matches their measured uncertainty. The estimated beliefs account quantitatively for canonical survey anomalies, including predictable forecast errors and overreaction that grows with the forecasting horizon. Since 2021, long-run inflation expectations have stabilized, but the updating process that generates them has not re-anchored. Long-run expectations are more sensitive to inflation surprises than at any time since the 1990s.

The anchoring of inflation expectations is a major objective of modern monetary policy. Under very weak conditions, allowing for a wide range of deviations from rationality and from the usual linear-Gaussian structure, the local sensitivity of an agent’s expectations to news is equal to their subjective uncertainty multiplied by the precision of the observed signal. Instrumental variables estimates show that uncertainty is a significant determinant of expectational anchoring for survey participants, and the data is consistent the model’s implication of proportionality – a hypothetical agent with zero uncertainty would ignore all signals. At the aggregate level, average uncertainty predicts the sensitivity of average expectations to realized inflation surprises, and with a magnitude consistent with the micro IV-based estimates. The results imply that as of 2025, household inflation expectations were 2–3 times more sensitive to news than prior to 2020.

This paper provides a novel empirical framework to examine how fiscal policy simultaneously affects both the level of economic activity and the uncertainty surrounding the economic outlook. Using quarterly historical US data, we show that an increase in government spending not only raises output but also reduces macroeconomic uncertainty, resulting in a stronger impact on the left tail of the output distribution—that is, on downside economic risks. Additionally, we demonstrate that the framework is useful to assess how the effects of fiscal policy vary depending on the state of the economy and we find that fiscal multipliers tend to be much larger—sometimes exceeding one— when economic conditions are weak and risks are elevated, compared to periods of economic expansion.

Uncertainty appears to have both a short-run and a long-run component, which we measure using implied volatility data from options of 30 days to two years duration. We summarize this term structure by its level (the 30-day implied volatility) and its slope (the 2-year less the 30-day), where the slope measures the persistence of uncertainty. We find that more persistent uncertainty depresses investment more than hiring. Within capital, the effects are largest for durable, irreversible, and structures-intensive assets. To capture the temporal aspect of uncertainty, we develop a two-factor dynamic investment model with a four-state Markov chain governing the level and persistence of volatility, in which asset durability & irreversibility drive sensitivity to long-run uncertainty.