Identifying Turns in Connecticut's Economy - January 2002
Preparatory activities and analyses have begun in the Department of Labor’s Office of Research in preparation for the annual benchmark revision of Connecticut nonfarm employment estimates. Revisions to the currently published nonfarm employment series will replace the estimates from April 2000 through March 2001 and will be released in March of 2002. The benchmark revision process represents a once-a-year re-anchoring of sample-based employment estimates to full population counts which become available later through unemployment insurance (UI) tax reports filed by nearly all employers with the Department of Labor.
Preliminary calculations indicate that Connecticut nonfarm employment (unadjusted) will be revised downward by approximately 7,000 to 13,000 jobs, or 0.4 percent to 0.7 percent. The total number of State jobs is close to 1.7 million. This is an early and rough benchmark revision assessment that could change as more UI tax records are received and processed. Ongoing administrative editing of the UI tax data remains as the most significant determinant of the extent of the final employment revisions. There are almost 110,000 worksites in Connecticut, with 3,500-4,000 included in the monthly payroll survey at estimation, so revisions are inevitable. Nevertheless, the monthly sample-based estimates are key to identifying important economic trends on a timely basis.
Connecticut’s Employment Turns
The currently published employment series shows an abrupt flattening of the rate of employment growth starting July 2000. Actually, seasonally adjusted nonfarm employment for the State declined 1,900 positions from July 2000 (1,699,400 jobs) to December 2000 (1,697,500 jobs). Business services employment, which includes temporary help, computer services, and advertising, peaked in the fourth quarter. The first quarter of 2001 showed further job growth deceleration and second quarter employment was no better as a major healthcare strike confounded May data.
Job loss trends began to accelerate and become more apparent at the end of the second quarter of 2001. Rapid and preemptive interest rate reductions may have served to prop up economic activity levels through the second quarter, as slightly negative linear trends developed in advance of the increased job cuts that began after June 2001. Connecticut has posted seasonally adjusted job losses every month since June 2001 for a total decline of 18,100 jobs between June and November 2001. Some of the worst job losses do appear to be over at this point, however. The warmer than usual weather is helping.
The September 11th events, which may be viewed as acts of economic warfare, have in some ways added a clearer view to the State’s employment backdrop. Much of the discussion regarding the State’s economy has been focused on the 9-11 events and subsequent economic uncertainty, or the third quarter 2001 employment drops, to identify slackening trends for the State. What has been undetectable to date is that the current nonfarm employment series already points to the Connecticut employment cycle peaking in the fourth quarter of 2000. We used peak-trough analysis tools created by the U.S. Bureau of Labor Statistics to identify employment cycles and phases according to criteria similar to National Bureau of Economic Research (NBER) methods. Using these peak-trough analysis tools, the current nonfarm job estimates show an employment peak late in the fourth quarter of 2000 if not earlier, and this is expected to be confirmed following the annual benchmark revisions. In addition, production worker weekly hours, also from the payroll survey, peaked in the fourth quarter of 2000.
Under this peak-trough criterion, a full cycle (peak to trough or trough to peak) must be fifteen months long and phases must be an expansion or contraction of at least five months. These definitions (15 and 5 months) plainly allow for the establishment of trends to prevent calling a one-month drop in employment a change in direction. The almost linear or slightly downward trend in employment in Connecticut after July 2000 qualifies, with a full contracting phase (ending the trough to peak cycle) being established five or so months later in the fourth quarter of 2000. The events of 9-11 may have more firmly established these downward trends, as some of the earlier data may have been inconclusive to establish a turning point in the cycle. The recent action by the NBER to set the onset of the national recession, based predominantly on U.S. nonfarm employment trends, as March 2001, would still preserve Connecticut’s distinction of leading the nation into downturns.
It is interesting to note that using these criteria to look back to 1992’s employment trough, one can see the employment low point was reached sometime immediately after Hurricane Andrew hit in August 1992. Connecticut’s employment recovery was slow after Hurricane Andrew, but it appears to have been a turning or bottoming point when predictions at the time continued to be negative. Hurricane Andrew was the largest insurance claim event prior to September 11, 2001.
Connecticut’s High Employment Correlation with the Nation
It is evident that Connecticut has diversified its employment base in the last decade and is less exposed to specific industry downturns. In fact, statistics show Connecticut has a very strong correlation coefficient with the nation in terms of nonfarm employment (there is a correlation above +0.95 over 120 observations - ten years, where +1.00 is a perfect correlation). A strong positive correlation is more a measure of association (strength) of relationship between variables, and not causation. In other words, if the nation were going into a recession, then most likely Connecticut would also go into recession as an outcome of the strength of this association.
It would also be unrealistic to think that other states did not also restructure their economies over the last decade, which provides ongoing competition. However, it is good information to know that Connecticut has already been in an employment downturn for potentially more than a year at this point. Further, knowing that average periods of diminishing economic activity usually range from 8 to 16 months may indicate a potential for recovery in the foreseeable future, baring an unusually severe winter, a prolonged strike, or another terrorist shock.
Conventional wisdom has been ingrained into economic analysis while virtually unpredictable conditions have existed over the last few years. So, fully buying into conventional analysis can at times do more detriment than good. Assorted economic activity indexes have contributed to the lack of clarity in this downturn, because separate indicators like housing and income have held up so well, masking turning points. Nonfarm employment data, even with its revisions, provides a broad, timely and relatively reliable coincident assessment of Connecticut economic trends and turning points in State economic cycles.