Fifth District State Business Cycles

Intro

To model national and state business cycles, this post first explores fluctuations of payroll employment around its long-run trend, comparing the correlation of the state-level employment fluctuations to the national series. Second, it explores how a popular rule of thumb recession indicator, the Sahm Recession Rule, performs when applied to state-level data compared to national data during the past four economic downturns. Third, this analysis not only provides us with insight into states' historical relationships with the U.S. business cycle, but it also reveals how they might fare in future economic downturns.

Background on Regional Business Cycles

The business cycle refers to the fluctuation of national economic activity over time, broadly measured through variables such as gross domestic product or employment. An economic variable's oscillations around a long-term trend can be parsed into phases, where a trough-to-peak period indicates an expansion, and a peak-to-trough period indicates a recession. While studying the theory and policy implications of the national business cycle has been a major focus of macroeconomists since the Great Depression, regional economists have long studied the fluctuations in the economies of subnational geographies such as census regions, states, and metropolitan areas.

Research into regional business cycles is primarily concerned with how and why cycles in these smaller geographies differ from the national business cycle. The most popular explanation are differences in industrial structure and trade relations. This body of research suggests that areas with high manufacturing activity, particularly in the production of durable goods, and more export-oriented output, tend to be the most sensitive to the business cycle. A 2019 article from the Philadelphia Fed explores how industry mix might explain why manufacturing-heavy metro areas' unemployment rates (e.g., Hickory, North Carolina, Rockford, Illinois, Elkhart, Indiana) are much more sensitive to the business cycle than their college town peers (e.g., Lawrence, Kansas, Madison, Wisconsin, State College, Pennsylvania).

Beyond industry mix and trade relations, other factors have also contributed to the degree of synchronization of state-level business cycles. Michael Owyang, David Rapach, and Howard Wall find that average establishment size, agglomeration (the degree of geographic concentration of firms), and the characteristics of a state's neighbors are important factors in explaining a state economy's correlation with the national business cycle. When looking at the individual phases of the business cycle through state coincident indexes, Owyang, Wall, and Jeremy Piger find that recession growth rates are driven by industry mix, while expansion growth rates are better explained by demographic factors such as age and education.

Findings from regional business cycle studies often have important policy implications. Gerald Carlino and Robert Defina find that, based on highly persistent spillovers in personal income growth across census regions, macroeconomic policies could better achieve aggregate growth objectives by targeting the areas with the largest potential for spillovers. In a later article, the authors find that regional differences in the mix of interest rate-sensitive industries explain differences in responsiveness to monetary policy, and that areas with higher shares of manufacturing activity are more sensitive to monetary policy shocks.

Cycles of Payroll Employment

To estimate business cycles for each of the Fifth District areas (District of Columbia, Maryland, North Carolina, South Carolina, Virginia, West Virginia), I apply the bandpass filter developed by Lawrence Christiano and Terry Fitzgerald to state-level quarterly annualized payroll employment growth series from 1985 through 2022. The filter extracts the cyclical component from the seasonally adjusted payroll employment series by removing frequencies shorter than six quarters and longer than 32 quarters from its estimated long-term trend.

Several patterns are apparent in the estimated state business cycles above. North and South Carolina appear to be highly sensitive to the business cycle: They tend to perform worse than the United States during recessions but better during expansions (particularly through the Great Recession). In contrast, the District of Columbia and West Virginia appear to be less sensitive to the business cycle: They have undergone shallower recessions and weaker expansions than the nation in three of the most recent four recessions. However, during the short-lived COVID-19 recession in 2020 Maryland and the District of Columbia performed much worse than their peer district states and the nation.

To better understand the co-movement between the state and national business cycles, a simple correlation analysis occurs between each of the series and the nation from 1985 and 2022. A correlation value of 0 indicates that there is no relationship between a state business cycle and the national cycle, while a value of 1 indicates perfect co-movement between the two series. The analysis reveals that business cycles in the District of Columbia (.73) and West Virginia (.89) are the least synchronized with the United States, while business cycles in Maryland (.96) are the most synchronized.

Identifying State Recession Entry Using the Sahm Rule

The National Bureau of Economic Research's (NBER) Business Cycle Dating Committee defines recessions as "a significant decline in economic activity that is spread across the economy and lasts more than a few months." The committee dates recessions by observing a range of aggregate economic data series such as payroll employment, real personal income, and industrial production. Since the committee relies primarily on backward-looking data, it often announces the occurrence of a recession several months after it has occurred.

Economists have searched for alternative recession indicators to assess recession risk in real time such as the term spread between long- and short-term treasury security yields and economic surveys. Economist Claudia Sahm developed a simple and reliable recession indicator using the U.S. unemployment rate, the Sahm Rule, which "signals the start of a recession when the three-month moving average of the national unemployment rate (U3) rises by 0.50 percentage points or more relative to its low during the previous 12 months." While the NBER and Sahm Rule recession entry dates do not always agree, the two are typically within a few months of one another.

Applying the Sahm Rule to the Fifth District states and the District of Columbia enables an assessment of how quickly the region enters national recessions. The charts below depict the difference between unemployment rates' three-month moving average and their lows for the previous 12 months while the purple shading indicates official NBER recession months.

According to the Sahm Rule, most Fifth District states (Maryland, Virginia, West Virginia, North Carolina) entered a recession in August 1990 — the same month the NBER dated the beginning of the early 1990s recession — but two months earlier than the Sahm Rule's recession flag for the United States (October 1990). However, the unemployment rates for the District of Columbia and South Carolina flagged recessions five months earlier (March 1990) and four months later (December 1990) than the NBER.

Ahead of the early 2000s recession, the Sahm Rule indicates that North and South Carolina had entered a recession several months earlier than the NBER national entry date of April 2001 (North Carolina in September 2000, South Carolina in February 2001), while Virginia (April 2001) and the District of Columbia (May 2001) entered a recession close to the NBER date. However, Maryland (August 2001) and West Virginia (December 2001) entered a recession several months later.

Unlike the two previous recessions, none of the Fifth District states' unemployment rates signaled the Great Recession earlier than the official NBER start in January 2008. While Virginia was flagged a month later (February 2008), most of the states were flagged in June or July (Maryland, District of Columbia, North Carolina, South Carolina). However, West Virginia's unemployment rate did not signal a recession until November 2008. For the two-month COVID-19 recession in spring 2020, all states' unemployment rates signaled a recession in either March (West Virginia) or April (District of Columbia, Maryland, North Carolina, South Carolina, Virginia).

Based on the Sahm Rule, it seems that no Fifth District state is consistently an early bellwether state for national recession risk. However, in three of the most recent four recessions, the unemployment rate for West Virginia signaled a recession several months later than the national unemployment rate.

What do recent data say about whether a recession is within sight? In the figure below, we plot the Sahm Rule measure using state-level unemployment rates, and the national unemployment rate, from 2022-2023. As of this writing, three of the states in the Fifth District have breached the Sahm Rule threshold (District of Columbia, North Carolina, and Virginia) over that period. However, North Carolina's reading has fallen back below its recession threshold, and only the District of Columbia has consistently risen above it in recent months. The national measure does not indicate that there is currently a recession.

Closing Thoughts

Regional data can provide insight into how subnational geographies behave during, and differ from, the national business cycle. An analysis of payroll data since the mid-1980s suggests that, out of Fifth District states, the District of Columbia and West Virginia are the least synchronized with the national business cycle. Using the Sahm Rule methodology to date state recession entry with unemployment rates does not indicate that any state consistently leads the nation as a recession indicator. However, the Sahm Rule suggests that West Virginia lagged the nation in recession entry by several months in three of the most recent four recessions.