U.S. BUREAU OF LABOR STATISTICS FEBRUARY 2014 • VOLUME 3/ NUMBER 2 R E G I O N A L E C O N O M I E S Persistence of a high unemployment rate in New York City during the recent recovery Author: Martin Kohli Between June 2009, the start of the national recovery according to the National Bureau of Economic Related articlesResearch, and the end of 2012, the national Additional articles or information on unemployment rate dropped by 1.7 percentage points, to 7.8 unemployment in particular U.S. regions are percent, while the rate in New York City declined by only 0.6 available online at the following links: percentage point, to 8.8 percent.1 This relatively small drop y “Recent trends in the characteristics of in the New York City rate occurred during a period of robust unemployment insurance,” Monthly Labor Review, http://www.bls.gov/opub/mlr/ growth in payroll jobs in the city (up 6.2 percent). Nationally, 2012/07/art3full.pdf the unemployment rate has remained high, or even increased, on several occasions during periods of recovery.2 Still, a y “Pay premiums among major industry groups in New York City,” Monthly Labor Review, number of observers of the New York City economy voiced the http://www.bls.gov/opub/mlr/2011/10/ expectation that, given the size of the growth of nonfarm jobs, art3full.pdf the city’s unemployment rate should have dropped further during 2012.3 U.S. BUREAU OF LABOR STATISTICS | FEBRUARY 2014 1 www.bls.gov BEYOND THE NUMBERS R E G I O N A L E C O N O M I E S This article examines the behavior of New York City’s Job growth and unemployment: the historical context unemployment rate from several perspectives. The Over the past 22 years, New York City has experienced several analysis begins by placing the unemployment rate and large swings in the pace of job growth. Chart 1 presents the growth of nonfarm jobs in their historical context. over-the-year changes in nonfarm employment, beginning The aim is to identify other periods in which strong job with August 1991. The shaded areas indicate periods of growth coexisted with persistent high unemployment. over-the-year job growth greater than or equal to 1.5 Then the labor force participation rate and the percent, combined with an unemployment rate at or above employment–population ratio are examined to see if 8.0 percent. As the chart shows, the city experienced three they shed light on the inertia in the unemployment periods of sustained job loss, each of which was followed by rate. The unemployment rate and the count of nonfarm years-long periods of job gains.4 The most recent upturn in jobs are based on two different surveys, of households payroll employment began in April 2010, although the initial and business establishments, respectively. The two pace of job growth was relatively modest, remaining below surveys have different concepts of employment, 1.5 percent until October of that year. From October 2010 making it difficult to infer that increases in nonfarm until August 2013, the over-the-year change in employment payroll jobs will translate into proportional increases ranged from a high of 2.9 percent to a low of 1.5 percent. This in the number of employed city residents. However, an pattern of recoveries beginning with modest growth and analysis of data from another household survey—the then strengthening is also seen in the upturns that began in American Community Survey (ACS)—makes possible June 1993 and May 2004. an examination of whether that survey’s patterns are similar to those in the Bureau of Labor Statistics For the purpose of examining the connection between the (BLS) household survey. The article concludes with growth in payroll jobs and changes in the unemployment a comparison of movements in data from the two rate, it is useful to separate periods of modest job growth household surveys. from periods of strong job growth with a distinguishing Chart 1 Over-the-year change in employment, New York City, seasonally adjusted, August 1991–August 2013 Percent 4 2 0 -2 -4 -6 -8 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 Note: Shaded areas indicate periods of over-the-year job growth greater than or equal to 1.5 percent, combined with an unemployment Note:r Sahtea daet do ra raebaosv ien d 8ic.0a tpee precerinotd. sD oaft oa vfeorr- tAhueg-yuesta r 2jo0b1 3g raorwe tphr eglrimeaitnearr tyh.an or equal to 1.5 percent, combined with an unemployment rate at or above 8.0 percent. Data for August 2013 are preliminary. Source: U.S. Bureau of Labor Statistics, Current Employment Statistics. Source: U.S. Bureau of Labor Statistics, Current Employment Statistics. U.S. BUREAU OF LABOR STATISTICS | FEBRUARY 2014 2 www.bls.gov BEYOND THE NUMBERS R E G I O N A L E C O N O M I E S value or threshold. This article uses a threshold of 2010 through August 2013 meets these criteria, with the 1.5-percent per year.5 Labor force growth in New York City unemployment rate averaging 9.0 percent. It has not has generally been less than 1.0 percent, so it is plausible been widely recognized, however, that the period from to expect job growth rates of 1.5 percent or higher to be December 1996 through May 1998 also combined strong associated with falling unemployment. Observers such as job growth with high unemployment. Over that 18-month the Fiscal Policy Institute and Julie Anna Golebiewski, an period, the unemployment rate was 9.1 percent—almost economist with the New York City Independent Budget the same as that in the most recent period. Office, voiced the expectation that the pace of over-the- If 9.0 percent were used as a threshold for high year job growth in 2012 (which ranged from 2.5 percent to unemployment, the two shaded areas would be smaller 1.5 percent) could have resulted in a falling unemployment but still important. The first period would start with rate.6 December 1996 and last 10 months, and the second would There have been occasions when strong over-the-year start with June 2011 and last 17 months. job growth was associated with substantial reductions in the unemployment rate. For example, from May 1998 Labor force participation rates and the employment– through February 2001 the rate of job growth ranged population ratio from 3.6 percent to 1.6 percent. As chart 2 illustrates, Chart 3 presents seasonally adjusted data from the BLS the unemployment rate declined from 8.0 percent to 5.3 Local Area Unemployment Statistics (LAUS) program (the percent over that same period.7 program that produces state and area unemployment rates), with the same periods shaded as in chart 2.8 One of Chart 2 also shows that strong job growth with persistent the unusual aspects of the labor market in New York City is high unemployment is not unprecedented. The shaded that the labor force participation rate has trended upward areas in the chart show periods with job growth of 1.5 over the past 22 years. As the chart indicates, the rate rose percent or higher, combined with an unemployment by 3.4 percentage points. Nationally, the rate declined by rate at or above 8.0 percent. The period from October 2.8 percentage points over the same period. Chart 2 Unemployment rate, New York City, seasonally adjusted, August 1991–August 2013 Percent 12 10 8 6 4 2 0 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 Note: ShNaodteed: S ahraedase din adriecaaste i npdeirciaotdes p oefr oiovdesr -othf eo-vyeera-trh joe-by egaror wjotbh ggrroewattehr gtrheaant eorr tehqauna ol tro e q1.u5a pl teorc 1e.n5t p, ceorcmebnitn, ecdom wbitihn eadn wunitehm anpluonyemmepnlto ryamtee antt or above 8r.a0t pe eartc oenr ta. bDoavtea f8o.r0 A puegrcuesnt t2. 0D1a3t aa rfeo rp r Aeluimguisnta r2y0. 13 are preliminary. Source:S Uo.uSr. cBeu:r Uea.Su. oBfu Lreaabuo ro Sf tLaatbisotirc Ss,t aLtoisctailc As,r Leoac Uanl Aemrepa lUoynmemenptl oSytmateisntitc Ss.t atistics. U.S. BUREAU OF LABOR STATISTICS | FEBRUARY 2014 3 www.bls.gov BEYOND THE NUMBERS R E G I O N A L E C O N O M I E S Chart 3 Labor force participation rate and employment–population ratio, New York City, seasonally adjusted, August 1991–August 2013 Percent 62 60 Labor force participation rate 58 56 54 Employment–population ratio 52 50 48 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 NoteN: oShtea:dSehda daereda as rienadsi cinadteic paeteri opedrsi oodf so ovef ro-vtheer--tyheea-ry ejoabr gjorobw gtrho wgtreha gtreera tthearn t hoarn e oqru eaql tuoa l1 t.5o p1e.5r cpeenrtc,e cnotm, cboimnebdin wedit hw aitnh uanneumnepmlopylmoyemnte rnattera atte oart abovoer 8a.b0o pveer c8e.n0 tp. eDractean fto. rD Aautag fuosrt A2u0g1u3s at r2e0 p1r3e laimrei npareryli.m inary. Source: U.S. Bureau of Labor Statistics, Local Area Unemployment Statistics. Source: U.S. Bureau of Labor Statistics, Local Area Unemployment Statistics. The chart also shows that the labor force participation rate percentage increases in jobs were not resulting in large and the employment–population ratio in New York City percentage increases in the count of employed city both increased from December 1996 to April 1998. During residents.9 that period, the unemployment rate remained high, in part At the national level, although the BLS household and because of the expansion of the labor force. establishment surveys track well over the long run, The city’s employment–population ratio also trended occasionally they also have reported large differences in upward—at least until March 2008, after which it dropped changes in employment. One article by BLS economists 2.7 percentage points over the next 20 months. Following reviewed efforts to understand why household a low of 53.9 percent reached in November 2009, the ratio employment (from the Current Population Survey) rebounded in the spring of 2010. However, since August increased nationally by 2.3 million from 2001 to 2004 2010, the employment–population ratio has fluctuated while payroll employment (from the Current Employment in a narrow range around 54.3 percent, slightly above its Statistics survey of business establishments) shrank by recessionary low. Over this same period, the labor force 0.3 million.10 An earlier study, using a slightly different participation rate has remained within 0.5 percentage point methodology, had looked at the period from 1994 to 2000, of 59.6. The persistence of high unemployment during that when the job count increased by 17.5 million whereas timespan reflects the persistence of the near-recessionary the number of employed people grew by 12.1 million.11 level of the employment–population ratio combined with a Both of these articles noted that the household survey relatively high labor force participation rate. includes not only wage and salary workers, but also the self-employed, unpaid family workers, people employed The question thus arises as to why the employment– by private households, and workers temporarily absent population ratio in New York City remained little changed from work without pay. The household and establishment despite the increase in nonfarm jobs. A number of surveys also differ in their reference periods and observers raised the related question of why large geographies (place of work versus place of residence), U.S. BUREAU OF LABOR STATISTICS | FEBRUARY 2014 4 www.bls.gov BEYOND THE NUMBERS R E G I O N A L E C O N O M I E S their treatment of multiple jobholding, and the age of Chart 4 presents annual average unemployment rates workers, among other things.12 Both articles concluded from the ACS, along with rates from the LAUS models and that, although various adjustments for industry, class the CPS. (The inclusion of both the LAUS and the CPS rates of worker, and multiple jobholding explain some of the allows us to see the effects of the models on the survey’s discrepancies, it is not possible to completely reconcile annual averages.14) The chart begins with 2008 data movements in the two national series for the periods in because the ACS changed its questionnaire in that year to question. ask unemployed people if they actively looked for work—a question that is similar to that in the CPS; the chart ends In New York City, the sample sizes are smaller than those with 2012 because that year’s data are the most current on the national level, resulting in larger standard errors. ACS data on unemployment.15 The LAUS unemployment This difference makes it difficult to analyze over-the-year rates differ from the CPS rates by tenths of a percentage changes for those categories, such as the self-employed, point, indicating that the models had relatively small used in attempts at reconciliation. In short, the conceptual effects on the annual averages. In contrast, during the differences between the two series are at least partially 2008–2012 period, the ACS measure exceeded the official responsible for some of their substantive differences, estimates produced by the LAUS program by an average but complete reconciliation is not always feasible at the of 1.6 percentage points, although the difference was as national level, and it is even more difficult with New York much as 2.2 percentage points in 2011. The fact that the City data. numbers differ is not surprising, given the difference in questions, collection methods, and samples between the ACS and LAUS series. The fact that the ACS numbers are The American Community Survey: another higher than the estimates produced by the LAUS program perspective on city residents and the CPS is also not surprising: generally, comparisons The Census Bureau introduced the American Community of ACS unemployment rates with CPS rates for the nation Survey (ACS) to replace the long-form questionnaire and for the states have found that the ACS rates were used in the decennial census. Like the long form, higher.16 the ACS is designed to be self-administered and collected by mail, although other means of collection For the purposes of this article, chart 4 is valuable because are used to follow up with those who do not respond it shows that the ACS measure has followed the same by mail. By contrast, the Current Population Survey general path as the LAUS and CPS measures. Specifically, (CPS) is conducted by trained interviewers. Like the the ACS indicates that a high unemployment rate in New CPS, the ACS asks questions about the employment York City persisted in 2011 despite the growth in nonfarm and unemployment of household members. These jobs. The ACS also indicates that the unemployment rate questions differ, however, from those posed in the CPS. may have dropped slightly in 2012. However, the broader In particular, respondents to the ACS are not asked about pattern of persistent high unemployment is still evident: in specific job search activities. The two surveys also have none of the three series is the unemployment rate for 2012 different approaches to defining their reference period significantly lower than it was in 2009. and to accounting for the worker’s place of residence.13 Nonetheless, the fact that the ACS counts people (rather The LAUS measure indicates that the employment– than nonfarm jobs) and that it provides estimates by population ratio for New York City residents dropped place of residence (as well as place of work) means that by 1.8 percentage points from 2008 to 2009 and has that survey offers another tool for examining the effects remained below 55 percent since then. Chart 5 shows of the recovery on New York City residents. Both surveys that the comparable ACS measure, despite always being publish data on New York City as annual averages (and higher than the CPS measure, was little changed in 2011, not on a monthly basis). although it rose slightly in 2012.17 U.S. BUREAU OF LABOR STATISTICS | FEBRUARY 2014 5 www.bls.gov BEYOND THE NUMBERS R E G I O N A L E C O N O M I E S Chart 4 Unemployment rate, New York City, 2008–2012 Percent 12 2008 2009 2010 2011 2012 8 4 0 Local Area Unemployment Statistics Current Population Survey American Community Survey Sources: Current Population Survey and Local Area Unemployment Statistics data: U.S. Bureau of Labor Statistics; American Community Survey data: U.S. Census Bureau. Chart 5 Employment–population ratio, New York City, 2008–2012 Percent 62 60 2008 2009 2010 2011 2012 58 56 54 52 50 48 Local Area Unemployment Statistics Current Population Survey American Community Survey Sources: Current Population Survey and Local Area Unemployment Statistics data: U.S. Bureau of Labor Statistics; American Community Survey data: U.S. Census Bureau. U.S. BUREAU OF LABOR STATISTICS | FEBRUARY 2014 6 www.bls.gov BEYOND THE NUMBERS R E G I O N A L E C O N O M I E S The labor force participation rate, as measured by LAUS, little effect on the employment–population ratio, as was 59.5 percent in 2008 and has remained within 0.6 measured by both the CPS and the ACS, and the continued percentage point since then. Similarly, the ACS measure low level of the ratio is part of the reason that high of the labor force participation rate has changed relatively unemployment has persisted. In 2012, the ACS indicated a little since 2008, remaining within 0.7 percentage point of slight improvement in the employment–population ratio the 2008 estimate of 63.5 percent. and a dip in the unemployment rate, but the pattern of high unemployment remained.  Conclusion This BEYOND THE NUMBERS summary was prepared The recent period of persistent high unemployment by Martin Kohli, economist in the New York Regional Office combined with strong growth in the number of payroll for Economic Analysis and Information, U.S. Bureau of jobs in New York City (a period that began in October 2010 Labor Statistics. Email: kohli.martin@bls.gov. Telephone: and extended through August 2013) is not unprecedented. (646) 264-3620. During an earlier period, from December 1996 to May 1998, job growth was also robust and the arithmetic Information in this article will be made available to average of the unemployment rate was virtually the same individual with sensory impairments upon request. Voice as it has been during the period that started in October phone: (202) 691-5200. Federal Relay Service: 1-800-877- 2010. That earlier period was followed by a sustained 8339. This article is in the public domain and may be period of lower unemployment. reproduced without permission. Other characteristics of the labor market for these periods Suggested citation: were not so similar. During the 1996–1998 period, the Martin Kohli, “Persistence of a high unemployment labor force participation rate and the employment– rate in New York City during the recent recovery,” population ratio both rose. In contrast, during the most Beyond the Numbers: Regional Economies, vol. 3, recent period both of these measures have been little no. 2 (Bureau of Labor Statistics, February 2014), changed. The labor force participation rate, for example, http://www.bls.gov/opub/btm/volume-3/persistence- has remained within 0.5 percentage point of 59.6 percent of-a-high-unemployment-rate-in-New-York-City- during the recent recovery. during-the-recent-recovery. By contrast, nationally the labor force participation rate declined from 64.4 percent in October 2010 to 63.2 Upcoming articles percent in August 2013. Thus, one of the reasons that the y An analysis of occupational concentration by unemployment rate in New York City has remained high industry relative to the national rate is that New York City’s labor y A general overview of STEM groups using OES force participation rate, unlike the nation’s, has not declined. data and the 2010 SOC System Finally, data from the ACS confirm that the unemployment Visit our online archives to access past publications at rate for residents of New York City was little changed in http://www.bls.gov/opub/btn/archive/home.htm. 2011. In that year, the large increase in payroll jobs had U.S. BUREAU OF LABOR STATISTICS | FEBRUARY 2014 7 www.bls.gov BEYOND THE NUMBERS R E G I O N A L E C O N O M I E S Notes 1. The use of a different start date, October 2009, as designated by the coincident economic indicator of the New York Federal Reserve Bank, leads to similar numbers. 2. See James M. Borbely, “Sizing up the 2007–09 recession: comparing two key labor market indicators with earlier downturns,” Issues in Labor Statistics, Summary 10-11 (U.S. Bureau of Labor Statistics, December 2010), http://www.bls.gov/opub/btn/ archive/sizing-up-the-200709-recession-comparing-two-key-labor-market-indicators-with-earlier-downturns-pdf.pdf. The National Bureau of Economic Research (NBER) also has noted that the unemployment rate often continues to rise after economic activity reaches a trough. See The NBER’s business cycle dating procedure: frequently asked questions (Cambridge, MA: National Bureau of Economic Research, updated daily), http://www.nber.org/cycles/recessions_faq.html. 3. See State of Working New York 2012: Disappointingly Weak Recovery (Albany, NY and New York, NY: Fiscal Policy Institute, September 2, 2012), http://fiscalpolicy.org/category/topics/jobs-wages-income/sub-topics-jobs-wages- income/fpis-state-of-working-new-york; and Julie Anna M. Golebiewski, “Unraveling the discrepancy between city job growth & a high unemployment rate” (New York City Independent Budget Office, February 2013), http://www.ibo.nyc.ny.us/iboreports/febacsemployment2013.html. Both of these reports were issued before the employment and unemployment data for 2012 were benchmarked. The benchmarking produced a revised, lower level of the unemployment rate in New York City during the middle of 2012, but the pattern of strong job growth and a persistently high unemployment rate remained. 4. The periods of job loss and job gain in the city do not coincide neatly with the business cycle dates of the NBER. Because observers of the city’s economy focus on counts of jobs, this article will use periods of local job loss and job gain (rather than the NBER’s dates) as a framework for the analysis presented. 5. BLS does not have a general criterion for distinguishing periods of strong job growth from weak. 6. See State of Working New York and Golebiewski, “Unraveling.” 7. Labor force estimates for New York City are produced by the BLS Local Area Unemployment Statistics (LAUS) program. The models used to estimate employment and unemployment are designed to identify and remove sampling error in monthly data from the Current Population Survey (CPS). For more information on the LAUS estimation methodology, see “Estimation methodology,” Local Area Unemployment Statistics (U. S. Bureau of Labor Statistics, September 11, 2009), http://www.bls.gov/lau/laumthd.htm. 8. Monthly estimates of the labor force participation rate and the employment–population ratio for New York City (and a number of other subnational areas) are given in “Civilian noninstitutional population and associated rate and ratio measures for model-based areas,” Local Area Unemployment Statistics (U. S. Bureau of Labor Statistics, April 19, 2013), http://www.bls.gov/lau/rdscnp16.htm. 9. Golebiewski, “Unraveling.” See also Jason Bram and James Orr, “Good news or bad on New York City jobs?” Liberty Street Economics (New York: Federal Reserve Bank of New York, August 2012), http://libertystreeteconomics.newyorkfed.org/2012/08/good-news-or-bad-on-new-york-city-jobs.html. 10. See Mary Bowler and Teresa L. Morisi, “Understanding the employment measures from the CPS and the CES survey,” Monthly Labor Review, February 2006, pp. 23–38, http://www.bls.gov/opub/mlr/2006/02/art2full.pdf. 11. See Thomas Nardone, Mary Bowler, Jurgen Kropf, Katie Kirkland, and Signe Wetrogan, Examining the discrepancy in employment growth between the CPS and the CES, report to the Federal Economic Statistics Advisory Committee (U.S. Bureau of Labor Statistics and U.S. Census Bureau, October 17, 2003), http://www.bls.gov/bls/fesacp2101703.pdf. 12. For a detailed discussion of the differences between the Bureau’s household and establishment surveys, see Bowler and Morisi, “Understanding the employment measures.” 13. For detailed discussions of the conceptual differences between the surveys, see Shail Butani, Charles Alexander, and James Esposito, Using the American Community Survey to enhance the Current Population Survey: opportunities and issues (U.S. Bureau of Labor Statistics, 1999), http://www.bls.gov/osmr/pdf/st990280.pdf; and Braedyn K. Kromer and David J. Howard, Comparison of ACS and CPS data on Employment Status (U.S. Census Bureau), http://www.census.gov/hhes/www/laborfor/ACS-CPS_Comparison_Report.pdf. See also “American Community Survey (ACS) questions and answers,” Local Area Unemployment Statistics (U.S. Bureau of Labor Statistics, May 4, 2011), http://www.bls.gov/lau/acsqa.htm. 14. As earlier noted, the LAUS models of employment and unemployment are intended to remove sample error from the monthly CPS estimates. 15. See Kromer and Howard, Comparison of ACS and CPS data. Even with this change, the ACS question was considerably more general than questions in the CPS, which asked about specific types of job searches. 16. Ibid. 17. The figures for 2010, 2011, and 2012 were 55.8, 56.0, and 56.8 percent, respectively. In all 3 years, the 90-percent confidence interval was plus or minus 0.3 percentage point around the estimate. U.S. BUREAU OF LABOR STATISTICS | FEBRUARY 2014 8 www.bls.gov