State Employment 1995: Slowing to a Recession?
by Mark E. Schweitzer and Kristin M. Roberts
March 15, 1996

Introduction

Employment in 1995 began with a bang and ended with a whimper, inspiring some to prognosticate a recession. A February 1996 spurt in employment growth (with 631,000 new jobs added, the highest monthly gain since 1983) has silenced many of the bears. The sudden boom came largely from new job starts that were delayed by bad weather or, possibly, missed when the government shutdown forced survey dates to be rescheduled. More subtly, the rebenchmarked state employment numbers have cast a rosier light on 1995.

Recessions and state-level employment declines are, in the simplest terms, almost tautologically related: National recessions necessarily affect many states. However, both the severity and the timing of recessions vary substantially at the state level. In fact, our review showed that a number of states were already exhibiting signs of economic weakness before the official start of every recent recession, as dated by the National Bureau of Economic Research (NBER).1

In this Economic Commentary, we review the final 1995 state employment figures, with an eye toward appraising the health of the national economy. We do not investigate the interesting question of the effectiveness of state employment changes as turning-point predictors, which would require a detailed statistical analysis. Instead, we treat the states as indicators of the extent of slowing employment gains. Why look at state figures if our interests are national? First, the annual rebenchmarking, which can dramatically change the history of both national and state employment figures, is completed at the state level months before the revised national numbers are released. Second, the timing and breadth of employment weakness at the state level help analysts evaluate the national economy’s health. Finally, without regard to a national recession, there can be substantial regional differences in the strength of the local economy, which probably influences public opinion on the strength of the national economy.

Return to Index


Figure 1: U.S. Net Employment Growth

[U.S. Net Employment Growth]

a. Not seasonally adjusted.
NOTE: Shaded areas indicate recessions.
SOURCE: Authors’ calculations based on data from the Bureau of Labor Statistics.

Rebenchmarked State Data and U.S. Employment

After the first quarter of 1995, civilian nonfarm employment gains slowed dramatically (see figure 1). On a year-over-year basis, net jobs growth dropped from more than 3.1 percent in February 1995 to just 1.5 percent in December.2 The size and rate of this decline are considerable, and parallel the declines in 1985 and 1989, prior to the "bumpy landing" of the mid-eighties and the recession that began in 1990. However, the national numbers reveal little about changes in the underlying employment patterns that might help us interpret last year’s slowdown in employment growth.

Before we can compare state employment patterns to national changes, we have to deal with one technical issue—the Bureau of Labor Statistics’ (BLS) annual rebenchmarking process. Its purpose is to compensate for the failure of the BLS survey, which covers approximately 400,000 establishments, to include all employers and therefore all workers.3 The key missing component is jobs created by newly formed companies or facilities, which are not initially included in the survey, but rather are estimated by state-specific adjustment factors. In recent years, the rebenchmarking process has yielded changes ranging from small reductions to additions of over half a million workers (about 0.5 percent of total employment). A particularly large addition might alter the series enough to change perceptions about the severity of the recent slowdown.

While we have only a preliminary estimate of the rebenchmarked national figure, the states’ rebenchmarked numbers, released in March, altered some employment levels substantially. For example, Ohio’s 1995 percentage employment gain nearly doubled, adding 61,300 jobs, or 1.2 percent of the state’s December employment. Overall, the updated figures are positive, with the median state gaining 0.3 percent of its employment through December 1995. According to the preliminary BLS estimate, the national benchmark revisions will add 542,000 jobs. For comparison with state data, we provide a national employment trend that applies the sum of the state rebenchmarks to the national data (indicated in figure 1 by the blue line).4 It appears that the upcoming national rebenchmarking will raise the original employment growth rates without affecting the 1995 slowdown in job gains.

Return to Index

Identifying Weakness at the State Level

One way to identify states with weak employment gains is to compare their year-over-year changes with their average growth over the last 15 years. The 15-year average accounts for differences in state growth rates that are associated with long-run trends or state-specific features—for example, population movement toward the Sunbelt states. Our working definition of "slow" employment growth is year-over-year percentage gains that are less than half of a state’s 15-year average growth rate.5 Figure 2 shows the states that experienced weak employment growth in 1995 and the months when these states were weak. Nine states had slow (or negative) employment growth sometime in 1995, compared to 17 states prior to rebenchmarking. All of these states, except Alaska and Hawaii, fell below this cutoff only in the second half of the year, when national employment changes had slowed dramatically.

Figure 2: Timing of Slow and Negative Employment
Growth in the United States, 1995

[Timing of Slow and Negative Employment Growth in the United States, 1995]

SOURCE: Authors’ calculations based on data from the Bureau of Labor Statistics.

While this cutoff for what defines slow growth is somewhat arbitrary, it has typically been reached before states experienced a period of employment losses in the 1980s and 90s or by states that continue to grow at a diminished rate during a recession.6 Figures 3 and 4 show the same cutoff during two recent periods of slow or negative national jobs growth: the mid-eighties’ "bumpy landing" and the 1990-91 recession. These figures illustrate the periods in which states (not individually identified) fell below the growth cutoffs on a year-over-year basis, with the NBER recession dates indicated by vertical lines. Both figures support the idea that growth of less than half of a state’s 15-year average is a reasonable indicator of continuing problems at the state level, rather than of randomly occurring weak periods. Furthermore, the "first-in-last-out" pattern of employment weakness applies to both of these periods: The first states to weaken tend to be the last to recover.7


Figure 3: Timing of Slow and Negative Employment
Growth in the United States, 1990-1992

[Timing of Slow and Negative Employment Growth in the United States, 1990-1992]

SOURCE: Authors’ calculations based on data from the Bureau of Labor Statistics.
During the 1990-91 downturn, a substantial number of states were already affected by the time the national recession began in July 1990 (see figure 3), as was typical of previous recessions. It is not surprising that a collection of states would be more sensitive to an employment shock of national scope (for example, a rapidly weakening industry) or would have experienced negative shocks during the period of feeble employment growth preceding a national recession. In 1990-91, the first states to slow were concentrated in the Northeast and were closely tied to defense-industry rollbacks and general weakness in financial services. Another look at figure 2 reveals that the current employment slowdown has neither involved as many states nor affected them for protracted periods. Since 1951, the only recession when so few states showed early weakness was that of 1973-75, but the typical pattern involves softening in a number of states before the business cycle peak.


Figure 4: Timing of Slow and Negative Employment
Growth in the United States, 1985-1987

[Timing of Slow and Negative Employment Growth in the United States, 1985-1987]

SOURCE: Authors’ calculations based on data from the Bureau of Labor Statistics.
It is also clear that several states can experience adverse employment situations without the nation in aggregate falling into a recession. As figure 4 shows, in the mid-eighties up to 16 states simultaneously had slow employment growth, without the nation experiencing a recession. Until now, the current employment slowdown has affected far fewer states.

Return to Index

Regional Patterns of Strength and Weakness

Although few states currently show weak employment growth, regional patterns are emerging. Figure 5 maps employment changes, again relative to each state’s 15-year history. The categories were chosen according to average employment growth relative to the 15-year average of all states (125 percent), with the outer categories representing substantially superior or weaker performance in 1995. Each category includes 9 to 14 states, with the weakest category being the smallest. While some states stand out from the regional patterns (for example, Washington, Wisconsin, Hawaii, and Massachusetts), we want to emphasize the broad regional trends.


Figure 5: State Employment Growth, December 1995a

[State Employment Growth, December 1995]

a. Percent of 15-year average employment growth.
SOURCE: Authors’ calculations based on data from the Bureau of Labor Statistics.

The most conspicuously weak region is the Atlantic Coast from Virginia to Connecticut. For the majority of these states, jobs growth fell below the 15-year average only late in 1995. The unusually harsh winter was not responsible, as the December survey data were tallied before the severest storms hit. Indeed, employment gains in these states had been sliding lower for months. Some Atlantic Coast states were among the last to recover from the 1990-91 recession: New York, New Jersey, Connecticut, and Maryland were weak well into 1993, while Pennsylvania was sluggish until late 1992. This suggests that some sources of weakness that led to long recessions for these states (hobbled defense-related and financial services industries) may still be hindering their economies.

Near Washington, D.C., with its dwindling federal employment, Maryland and Virginia have fallen far short of their typical growth rates since May 1995. While U.S. government jobs represent a large share of these states’ total employment (around 6 percent in Maryland versus about 2 percent nationally), the direct decline of federal employment (-1.6 percent last year for Maryland) does not account for such weakness. Rather, potentially associated losses and slowdowns across most industries explain Maryland’s poor performance.8 This situation is typical of states that enter a recession early—some proximate cause may be identifiable, but the losses are widespread. While the pattern could continue to act as a damper on these states, their specific weakness would be unlikely to harm other states, where federal employment is less concentrated.

On the other hand, many states continue to grow vigorously, albeit at rates lower than at the beginning of 1995. Strong showings in Ohio (238 percent of its 15-year growth rate), Illinois (260 percent), and Michigan (178 percent) point toward continued strength in the Great Lakes states traditionally associated with manufacturing.9 Employment gains are also substantial in several Plains states where jobs are highly concentrated in agriculture, although many of these states—most notably Iowa and Kansas —have above-average manufacturing intensities (18 percent and 16 percent of workers, respectively).

The other strong region is the West. While Utah, Montana, and Oregon are in better shape than their neighbors, most western states are outperforming the economy, even relative to their 15-year averages (which are also higher than most). Several states (for example, Utah, Nevada, and Colorado) have been big gainers for the last several years, as migrants from other states rapidly expanded their populations. Although the West continues to flourish, the growth of most of its states has tapered off. Several states in the oil-oriented Southwest are also doing quite well.

Return to Index

Conclusion

Overall, U.S. employment growth has slackened, reflecting a slower aggregate economy. However, many states have thus far avoided entering a "soft patch." There is no definitive evidence here for a recession; rather, our analysis shows that jobs continue to grow in most states (although at a reduced rate). Only a few states are experiencing declining employment growth that might suggest an increasing likelihood of local recessions —notably on the East Coast. These states represent a substantial portion of the economy, and their performance may tend to color perceptions beyond their boundaries.

Return to Index

Footnotes

1. Randall Eberts reviews recessions from 1950 to 1990 and finds that in most of them, a few states were already experiencing employment losses when the peak was reached nationally. See "Can State Employment Declines Foretell National Business Cycles?" Federal Reserve Bank of Cleveland, Economic Commentary, September 15, 1990.
Back to commentary

2. We use year-over-year employment changes (calculated by subtracting the previous year’s non-seasonally-adjusted monthly figure from this year’s), not the more commonly reported seasonally-adjusted monthly changes.
Back to commentary

3. Rebenchmarking corrects these figures by using far more complete information from states’ unemployment insurance records.
Back to commentary

4. Because of technical differences in the procedures used to calculate civilian nonfarm employment levels, the state numbers do not sum to the national figures.
Back to commentary

5. Note that the 15-year average includes substantial recessionary periods during which most states lost employment.
Back to commentary

6. The cutoff was actually arrived at by studying the frequency with which states fell below their average growth rates. States fell below half of their 15-year average growth only 30 percent of the time. Average employment growth rates varied from West Virginia’s 0.3 percent long-term rate to booming Nevada’s 4.7 percent.
Back to commentary

7. Eberts (footnote 1) identified this pattern in state employment losses from 1950 to 1990.
Back to commentary

8. Industry figures at the state level have not yet been rebenchmarked. They are used primarily to calculate states’ relative industry intensities, which do not change rapidly and are unlikely to be affected by rebenchmarking.
Back to commentary

9. Ohio and Michigan both have manufacturing employment concentrations well beyond national averages—23 and 24 percent, respectively. Illinois is only slightly above average in manufacturing employment.
Back to commentary

Return to Index


Mark E. Schweitzer is an economist and Kristin M. Roberts is a senior research assistant at the Federal Reserve Bank of Cleveland. The authors thank Randall Eberts for insightful comments.

The views stated herein are those of the authors and not necessarily those of the Federal Reserve Bank of Cleveland or of the Board of Governors of the Federal Reserve System.


http://www.clev.frb.org/research/com96/