Accessible Version
Assessing Recession Risks with State-Level Data, Accessible Data
Figure 1. Illustration of L- and U-shaped recessions
Figure 1 presents a schematic illustration contrasting L-shaped and U-shaped recoveries from recessions. The figure uses simple line graphs to illustrate the concepts: both lines begin with an economic expansion and a recession, but diverge in their aftermath. The U-shaped line shows a strong rebound, eventually converging to the pre-recession trend indicated by a black dotted line. In contrast, the L-shaped line never returns to this trend, depicting a persistent shortfall in output—this is the empirical signature of hysteresis. The shaded area denotes the recession period, visually reinforcing the distinction between transitory and lasting economic damage.
Note: The X-axis denotes periods after a business cycle peak. The black dotted line indicates the hypothetical output level if a recession had not occurred; the shaded area represents periods in which the economy is in a recession. The periods prior to the recession correspond to expansionary phases.
Source: Authors’ calculation.
Figure 2. National-Level Recession Probabilities
Figure 2 plots estimated probabilities of experiencing L-shaped and U-shaped recessions at the national level, with Panel A based on payroll employment and Panel B based on GDP growth. The red lines represent L-shaped probabilities, and the blue lines represent U-shaped ones. Overall, the payroll-based estimates correspond well with NBER recession dates but differ from GDP-based ones in key periods—most notably, the Great Recession is identified as L-shaped in employment but U-shaped in GDP. The COVID-19 recession is classified as U-shaped in both, although the second wave is interpreted differently: payroll data suggests a new U-shaped episode, while GDP data sees continued expansion. This divergence illustrates that employment measures may better capture persistent labor market weaknesses.
Note: The figures display the probabilities of L-shaped and U-shaped recessions at the national level, estimated using nonfarm payroll employment growth (Panel A) and real GDP growth (Panel B). For 2025:Q4, nonfarm payroll employment growth is computed as the average of October and November, as December data are unavailable at the time of writing. The blue dashed lines represent the probability of a U-shaped recession, while the red solid lines represent the probability of an L-shaped recession. The y axis indicates the probability, and the x-axis indicates the date. The shaded bars indicate U.S. recessions as dated by the National Bureau of Economic Research: April 1960–February 1961, December 1969–November 1970, November 1973–March 1975, January 1980–July 1980, July 1981–November 1982, July 1990–March 1991, March 2001–November 2001, December 2007–June 2009, and February 2020–April 2020.
Source: Authors’ calculation.
Figure 3. Probabilities of L- and U-shaped Recessions across States over Time (1960:Q1-2025:Q3)
Figure 3 presents two heatmaps (Panels A and B) that display, across all U.S. states, the quarterly probabilities of experiencing L-shaped and U-shaped recessions from 1960 to 2025. Each row is a state, and each column is a period, with darker colors indicating higher probabilities. Panel A (L-shaped) reveals that since the 1990s, L-shaped recessions have become more prevalent, while Panel B (U-shaped) shows a marked decline in such recoveries. Some states display isolated recessionary episodes independent of national cycles—for example, Louisiana in 2005 and North Dakota in 2015. The COVID-19 recession, however, stands out as broadly U-shaped across nearly all states, reflecting a rapid bounce back from an unusually deep shock.
Note: The figures show the probability that states experience L-shaped recessions (Panel A) and U-shaped recessions (Panel B), estimated using state-level nonfarm payroll employment data. As of the time of writing, state-level nonfarm payroll employment growth is available through August; for 2025:Q3, we use the average of July and August. States are shown on the y-axis, and the x-axis is calendar time. Darker colors in each heat map indicate higher recession probabilities.
Source: Authors’ calculation.
Figure 4. Macroeconomic indicators of recessions
Figure 4 presents two figures that display layoffs indicators and financial market indicators. Panel A plots initial unemployment insurance claims, advance layoff notices, and layoffs and discharges, which all peak very high in 2020 and then decline in 2021. Initial unemployment claims go back down to 2019 levels in 2022 and stay there. The other two series dip below 2019 levels in 2021 then also revert to near 2019 levels in 2022. Panel B plots the Treasury spread and the credit spread. The treasury spread is briefly negative in 2019 before rising to 2 percent in 2022. It then sharply declines to almost -2 and stays negative until late 2024 when it begins to hover around 0. The credit spread spikes during the pandemic from 1.5 percent to above 3 percent. It then falls to nearly 1 percent in 2021, rebounds to around 2 percent in 2022, then steadily declines to near 1 percent again through 2025.
Note: In panel A, initial UI claims are reported as a 4-week moving average. The y-axis is truncated at 250 to prevent pandemic-era observations from muting recent variation. In panel B, The Treasury spread is defined as the difference between the 10-year and 3-month Treasury yields. The credit spread is the difference between yields on BBB corporate bonds and 10-year Treasuries. Shaded areas spanning February – April 2020 denote the NBER recession. Units are in percentage points.
Source: Authors’ calculation; ICE Data Indices LLC, used with permission; Federal Reserve Board of Governors; U.S. Bureau of Labor Statistics; U.S. Department of Labor.