Working Paper Series no. 1036. Post-pandemic inflation highlighted tensions between price stability and growth objectives. We evaluate this risk using probabilistic responses from professional forecasters’ surveys. Our dynamic factor model with time-varying uncertainty and asymmetry captures the joint dynamics of inflation and growth and decomposes them into demand and supply components. We find that tail risk is prominent in US data in the 1980s and during the Great Recession for inflation, and during the 1980s and in the period following COVID-19 for GDP growth. The post-pandemic inflation is driven by temporary adverse supply and persistent positive demand. The model-implied correlation between inflation and growth is time-varying, negatively related to nominal term premiums and on average positive, suggesting that professional forecasters do not have stagflationary beliefs. In 2022, stagflation risks increased after three decades of near-zero probabilities.
Figure 1. Time-varying probability of stagflation
Non-Technical Summary
Identifying whether macroeconomic fluctuations are mainly attributable to demand or supply factors is essential to setting the appropriate policy mix. A shock like COVID-19 has drawn renewed interest in distinguishing between these determinants because this period is characterized by several large concurrent demand and supply shocks. In this paper, we address this question for the United States (US).
We jointly assess the risks on prices and production by exploiting forward-looking survey information to capture macroeconomic risk perceptions. We identify demand and supply factors for the US by imposing basic theoretical assumptions on a dynamic factor model. Following economic theory, demand factors are those that drive inflation and real activity to move in the same direction. In contrast, supply factors are captured by movements in inflation and economic activity in the opposite direction. The dynamic factor model features time-varying volatility (uncertainty) and asymmetry to study the relationship between inflation and the real economy through the lens of the probabilistic responses of professional forecasters. The use of surveys allows us to study the perception of the joint relationship between inflation and GDP growth of informed market participants that regularly nourish monetary policy discussions. Importantly, such data reveal information beyond past and current realized inflation and GDP growth because they are forward-looking and inform about tail-risk expectations without the need for such fears to materialize, making them a rich and relevant source of information for our analysis.
Beyond standard realized inflation and output growth series, we leverage probabilistic responses of the Survey of Professional Forecasters (SPF) to exploit changes in the entire distribution of future expected inflation and GDP growth. In our setup, we enrich the identification of drivers by allowing changes in the shape of the distribution of expected inflation and GDP growth to inform about the nature (demand/supply) of the shock hitting the economy. Hence, our model features time-varying uncertainty and asymmetry (tail risk) to jointly estimate the dynamics of realized and expected inflation and GDP growth rates and fit the conditional first-, second- and third-order survey-moments at different horizons. Moreover, our model allows for a trend and cycle decomposition, which enables us to study the drivers of prices and real activity at business-cycle and lower frequencies.
We estimate our model on US data spanning the period 1981-2024. Our model can fit realized inflation and growth as well as all three first conditional moments of their forecast distributions. Importantly, the sample includes both calm periods and very large shocks (e.g., 1980s, Great Recession, COVID-19), therefore, our model performs well under any economic condition. Our demand and supply factors capture key events that relate to demand and supply shocks, respectively. Moreover, our volatility factor marks its sharpest increase during the COVID-19 crisis and features peaks around every NBER recession in our sample.
Our findings suggest that the output gap is determined by a mix of demand and supply factors before the Great Recession and in the last quarters of our sample, while it is mainly driven by demand between 2008 and 2020. Moreover, the price gap exhibits a sharp rise in the last quarters of our sample, in line with the idea that high inflation realizations are partly explained by transitory components. Interestingly, the post-Covid increase in inflation is attributed to temporary negative supply factors and more persistent positive demand factors after the COVID crisis. These findings are consistent with the increase in bottlenecks and supply chain disruptions since COVID-19, as well as the large pandemic economic stimulus and relief packages that were implemented in the United States. Finally, we compute the time-varying correlation between inflation and economic activity, revealing the evolving importance professional forecasters assign to aggregate supply and demand. We observe that this correlation varies over time and is negatively related to nominal term premiums, in line with structural modeling of the term structure of interest. On average, the relationship between inflation and economic activity is positive, suggesting that professional forecasters do not have stagflationary beliefs. Computing probabilities of stagflation risk in future horizons, we also show that in 2022, these risks reemerged after three decades of near-zero probabilities.
Keywords: Dynamic Factor Model with Stochastic Volatility, Uncertainty, Asymmetry, Tail Risk, Inflation, Output Growth, Demand, Supply, Trend, Cycle
Codes JEL : C32, E31, E32, E44
Download the full publication
Updated on the 25th of February 2026