Can states help us understand the national employment picture?

Employment gains for May 2016 were disappointing. Only 38,000 additional jobs were added to U.S. payrolls that month, compared to a monthly average of 222,000 during the past 12 months. Clearly, the May figure of 38,000 additional jobs is the lowest monthly increase during this record-setting period of 75 months of consecutive monthly employment gains for total private sector jobs and 68 straight months for total nonfarm employment (seasonally adjusted). Only twice in the 68-month period were monthly gains in the May range: 42,000 in January 2011 and 45,000 in December 2013. Even 13 months ago, in March 2015, employment for that month was only 85,000, but employment the next month jumped to 251,000 payroll jobs. Maybe a similar rebound will occur in June but what is concerning analysts this time is the slow decline in employment gains during 2016.  Monthly job gains have slowed in 2016 from 233,000 in February to 186,000 in March to 123,000 in April, and now 38,000 in May. 

Analysts offer a few explanations for the slow May employment gains that relate to one-off events. One is the strike by 37,000 Verizon employees in May, which undoubtedly affected the overall employment number for the month. As of June 1, the strike is over and employees are returning to work.  Another is the weakness in a few key industries, such as construction, utilities, and wholesale trade. 

Instead of focusing on specific events or industries and viewing them in isolation, I would like to switch our view to regional differences in employment growth, specifically differences across states. It is well understood that sectors within regional economies are closely related.  What happens in manufacturing or energy extraction within a local economy, for example, spills over to retail and personal services as workers from export-based sectors purchase goods and services from other local sectors. Consequently, since sectors within regional economies are intertwined, it may be more enlightening to see what’s happening at the state level, and to see if we can detect any trends from that perspective.

To start, let’s look at the current business cycle with respect to the number of states that during any one month experienced an increase in jobs, compared to those that experienced a decline in jobs. Figure 1 shows the number of states with job gains (line) superimposed against national employment gains (bars).  Instead of looking at monthly changes in employment, we consider year-over-year changes, in order to eliminate the volatility inherent in monthly data (even when seasonally adjusted). The figure begins with the peak of the previous business cycle, which occurred December 2007.  Actually, total nonfarm employment didn’t peak until the next month, January 2008.  At that time, as one can see from the graph, 45 states experienced employment growth. Those 45 states accounted for 118 million of the 138 million payroll jobs in the 50 states, and they generated 1.2 million jobs (year over year) at the time. The five states that experienced employment losses at that time accounted for 19 million jobs and had lost 204,000 since the year before. From that time on, the employment picture quickly deteriorated.  Within 12 months, all but one state (Alabama) experienced year-over-year employment losses, and the nation was well within the grips of the Great Recession. However, it was another year before national employment levels hit bottom.

As seen in Figure 1, the number of states experiencing year-over-year employment declines tracks year-over-year national employment changes quite well. The correlation is 96 percent. The same strong relationship is found when monthly employment changes are used to define states with job losses and to account for national employment changes, as shown in Figure 2. The greater volatility in monthly changes is obvious in Figure 2 compared with Figure 1, but the correIation between the number of states with job gains and employment change is still high: 88 percent. 

Does this regional view of the national economy lend any insights into future trends, particularly whether employment gains are seriously slowing and reasons for the slowdown, if it has occurred? What is apparent from Figure 1 is that the number of states with employment gains has begun to decline since early 2015. For five consecutive months prior to March 2015, all 50 states enjoyed employment gains.  Then, beginning in March 2015, one state began to experience job losses followed by a few others.  By April 2016, seven states were experiencing employment losses. All seven states are heavily reliant on energy extraction. West Virginia, heavily dependent on coal mining, was the first state to experience job loss.  It was followed the next month by Wyoming, which is dependent on both coal and oil. The following month, North Dakota, well known for its oil boom and now bust, also began to lose jobs. Three months later Louisiana joined the group, and three months after that Oklahoma and Alaska came on board—all three states with heavy dependence on oil. Kansas, the 10th largest oil-producing state, began to lose jobs in February of this year. Clearly, the well-publicized oil slump has taken its toll in these states.  And at least one other—New Mexico—may follow before that sector turns around.  If New Mexico passes over to the negative side of the employment ledger, all but three of the top 10 oil-producing states would be included among the group of job-loss states. The three not included—Texas, California, and Colorado—are large economies and do not appear to be affected by only one sector, such as oil extraction. For the past year or so, the year-over-year employment growth for Texas has ranked around the middle range of the states, whereas the employment changes for California and Colorado have been above the 75th percentile, averaging more than 3 percent each year (on a year-over-year basis).

Does the fact that seven states have now slid into negative territory say anything about future changes in national employment? It is tempting to look at the sheer numbers and note that when the nation was standing on the precipice of the Great Recession, five states were already shedding jobs. Returning to Figure 1, there is a noticeable increase in recent months in the number of states losing jobs. However, the seven states in employment decline account for only 5 percent of employment in the 50 states, whereas the five states that led the nation into the Great Recession accounted for 14 percent. And so far, employment change on a year-over-year basis is still above the 2 million level, and it hasn’t trended down in any serious way, except for May. Unless states with larger populations, such as Texas and California or even some of the industrial-belt states, begin to slide into negative territory, the current trend may be only a blip.

It’s also interesting to note that the first seven states to lose jobs in the past year are not the same states that first entered the recession in the latter part of 2007 and into 2008.  On the contrary, those seven states were among the last states to enter the recession.  The first states to enter the Great Recession were Rhode Island, Michigan, Indiana, Wisconsin, Nevada, and Florida. The latter two states in this group were closely related to the housing bubble, whereas the others were tied more closely to manufacturing.  During the past year, these states have performed fairly well. Employment gains on a percentage basis have placed most of them in the upper half of all states, with only Wisconsin falling significantly below the mid-point.  Indiana is closely tracking Texas’s employment growth rate, and Michigan’s employment growth rate is consistently in the upper half of the states.  However, Rhode Island has been bouncing around in the lowest 25th percentile range. For the past twelve months, the average employment growth rate for all states is 1.5 percent (not weighted by state population), and for the six states to first enter the recession, the rate is 2.0 percent.  (The average employment growth rate for all 50 states combined, which is in essence weighted by state population, is 1.7 percent.)

Has a state-by-state perspective offered any new insights into the current national employment situation? Clearly, the cumulative fate of state economies colors the national employment picture. The dramatic fall in oil prices and other commodity prices has taken a toll on local economies that depend heavily on these sectors, which is evident from looking at state data. With oil prices rebounding to some extent in the past several months, we may see these states moving back into positive territory with respect to employment changes. If the oil slump is the major reason for the slower employment growth in recent months, it may mean that employment gains will return to around 200,000 a month as oil prices continue to rise. However, a more complete view of the employment situation may require looking for signs of strength or weakness that permeate throughout the broader economy and not simply a few sectors isolated in a few states. 

Randall W. Eberts, President, W.E. Upjohn Institute for Employment Research