Labor Market Update December 3, 1 Daniel Aaronson Vice President and Director of Microeconomic Research Research Department daaronson@frbchi.org 1 Where we are today Chicago Fed National Activity Index (standard deviation from trend, 3-month moving average, shading corresponds with NBER recession periods) 3 standar rd deviation from trend growth. 1-1 - -3 - Trend growth Unemployment rate falls Unemployment rate rises - 197 197 1977 19 197 199 1997 7 th Annual Economic Outlook Symposium December 3, 1 1
Unemployment rate is historically high, especially accounting for changes in skill levels.1 Unemployment rate, education constant (19 levels).1.1 Calculated UR, 19 Education.. actual.. 197 197 19 19 19 19 19 199 199 199 199 199 1 3 Large Resource Gaps: Payroll Employment version Nonfarm Payroll Employment (change, thousands) Since peak (1/7): Empl = -7. million. But gap includes 1,/month to keep up with population growth. So really down 7. million+3 mos*1, = 1.9 million - Monthly change 3 month average -9 ' '7 ' '9 1 To make up that gap plus keep up with future population growth, would take growth rates over these time horizons: k/month over 3 years OR 37k/month over years OR 1/k month over years Historical context 1 emp growth = 7k/mo - emp growth = 13/k mo 1993-99 emp growth = 1/k mo th Annual Economic Outlook Symposium December 3, 1
Recovery is forecasted to be weak relative to past cycles with deep recessions. Real GDP (Quarterly, 1 = Recession Trough) 1 Unemployment Rate (Quarterly, 1 = Recession Trough) 11 1.1% 11 11 1 1 73 7 Current 9 7 Current 73 7 1 1 9 - -3 - -1 1 3 7 9 1 Quarters relative to end of recession - -3 - -1 1 3 7 9 1 Quarters relative to end of recession Note: Quarter this cycle is 9:Q (UR = 9.3%). Q1=11:Q Dashed lines are the Blue Chip forecast (7/1/1). How much of high unemployment is a cyclical versus structural problem? Critical question. Structural (impediments to matching workers with firms) e.g. skills mismatch industry mismatch (e.g. turning construction workers into nurses) geographic mismatch (e.g. house lock) incentives arising from policy (e.g. UI extensions) Solution: Fix impediments. Education/training, information/search assistance, mobility/housing strategies. Key implication: The long-term unemployment rate (NAIRU) is higher. More accommodative monetary policy will lead to inflation problems and not help to clear mismatch/frictions. i ti Cyclical (demand deficiencies) Solution: need more economic activity. Key implication: NAIRU need not have risen. th Annual Economic Outlook Symposium December 3, 1 3
Outline from here Some evidence for increase in structural unemployment 1. Okun s law is broken. Job openings up, hiring not (shift in Beveridge Curve ) 3. Increase in long-run unemployment. UI extensions and work disincentives Some evidence for increase in cyclical unemployment 1. Okun s law is not broken. The behavior of the Beveridge Curve is not that uncommon 3. Workers are not in the wrong sectors. Workers are not in the wrong geographic areas. Skill mismatch is not a significant barrier Summary 7 1. Structural: Okun s law is broken The unemployment rate is higher than expected given GDP path 1 9 7 Unemployment Rate versus Predicted Unemployment Rate based on Okun s law actual Predicted ~1½ percentage points 7q3 7q q1 q q3 q 9q1 9q 9q3 9q 1q1 1q 1q3 * The predicted unemployment rate is based on an Okun s coefficient of. and the actual path of GDP growth. th Annual Economic Outlook Symposium December 3, 1
. Structural: Vacancies up but hiring barely following. Both extremely low. JOLTS: Hires and Vacancies (SA, millions) Hires Vacancies Unemployment to Vacancies (ratio) 3 X? 1 '3 ' '7 '9 197 '7 ' ' '9 '9 ' 1 9 9. Structural: Hiring out of unemployment (or out of labor force) historically very low Probability of moving from unemployment to employment. 3.3.3...1 Thru Sept 1.1 197 197 19 19 19 19 19 199 199 199 199 199 1 Calculations from matched CPS files. 1 th Annual Economic Outlook Symposium December 3, 1
. Structural: Vacancies are high relative to state of labor market.. 3. Beveridge Curve :Q 1:Q3-7 -11 Vac cancy Rate (percent) 3... ~ 3/ percentage points on unemployment rate 1. 3.... 7.. 9. 1. Unemployment Rate (percent) 11 3. Structural: Unemployment spells lasting longer than usual Average Length of Unemployment spell (in weeks) and Percent of Unemployed with Spells lasting > weeks 3 3 % unemployed > weeks Average length of on-going unemployment 1 1 1 th Annual Economic Outlook Symposium December 3, 1
3. Structural: even conditional on the state of the labor market Unemployment Rate vs Long-term Share of Unemployment, 19-1 Fraction of Labor Force Unemployed 7 wee eks + 3 3 1 1 Jan 197 to Nov 7 Dec 7 to Oct 1 Early 19s 1 1 Unemployment rate 13 3. Structural: and this will last a long time. Unemployment rate vs share long-term unemployed, selected cycles Fraction of Labo or Force Unemployed 7 we eeks + 1/7-1/1 3 7/1-1/3 3 1/73-11/7 1 1 1 1 Unemployment rate 1 th Annual Economic Outlook Symposium December 3, 1 7
3. Structural: A spell that is initially due to bad economic conditions can turn structural (lost skills, job networks). Probability of finding a job next month, by unemployment duration this month.3.3...1.1. Weeks Note: Calculations based on matched CPS microdata from 197 to 9 1 3. Structural: Involuntary layoffs lead to long and deep earnings losses. Percent of re-employed displaced workers Mean earnings change: -3% % of workers report real earnings within 1% of previous earnings 1 1 % of workers report real earnings losses of 1% or more Why losses? Worse job match, deteriorating human capital/job networks, employer bias against LT unemployed % of workers report real earnings gains of 1% or more Why gains? Better match, hiring employers more productive than firing employers -1-1 Approximate change in real weekly earnings as a percent of pre-displacement weekly earnings* *Approximate change = change in log of workers real earnings*1; extreme values are included in the statistics, but not shown in the histogram. 1 Source: BLS Displaced Worker Survey (covering 7-9), author s calculations. th Annual Economic Outlook Symposium December 3, 1
. Structural: UI has been extended to 99 weeks in many states. 1 Maximum Potential UI Benefit Duration and Fraction of Unemployed Insured, National Average. 9 7.7 Weeks 7 Thru 9....3 Fraction Insured 3 Avg. Potential UI duration. 1 Fraction Insured.1 9 17. Structural: Extensions probably lead to an increase in unemployment duration and rate Our best guess is that a 1 month increase in benefits leads to a.1 month increase in the length of unemployment spells. Implies ~ ¾ to 1 percentage point on UR today. Example: Schmieder, Von Wachter, and Bender (1). German UI system has sharp eligibility extensions at specific ages. NOTE: Combination of work disincentive and easing of liquidity constraints (Chetty 1). Latter is likely very beneficial to 1 economy. th Annual Economic Outlook Symposium December 3, 1 9
Outline Some evidence for increase in structural unemployment Next, evidence for increase in cyclical unemployment 1. Okun s law is not broken. The behavior of the Beveridge Curve is not that uncommon 3. Workers are not in the wrong sectors. Workers are not in the wrong geographic areas. Skill mismatch is not a significant barrier 19 1. Cyclical: Okun s Law revisited 1 9 Unemployment Rate vs. Predicted Unemployment Rate Based on Okun s Law actual 7 Predicted 7q3 7q q1 q q3 q 9q1 9q 9q3 9q 1q1 1q 1q3 * The predicted unemployment rate is based on an Okun s coefficient of. and the actual path of GDP growth. th Annual Economic Outlook Symposium December 3, 1 1
1. Cyclical: The rise in the unemployment rate is not that surprising after all, under other assumptions. 1 9 Unemployment Rate vs. Predicted UR Based on Cyclical Okun s Law actual 7 Predicted cyclical Okun Predicted. Okun 7q3 7q q1 q q3 q 9q1 9q 9q3 9q 1q1 1q 1q3 * The predicted unemployment rate based on a cyclical Okun s coefficient is estimated separately for recession and expansion quarters and applied to the actual path of GDP growth. 1. Cyclical: What about the shift in the Beveridge Curve? Unemployment vs. Vacancy Rate (percent) 1. Based on Help Wanted Ads Based on JOLTS 1. Vacancy Rate 3 1.. Q3 1. 1 1 1. 1. 1. 1..... Unemployment Rate th Annual Economic Outlook Symposium December 3, 1 11
. Cyclical: Beveridge Curve loops are common and consistent with economic theory Unemployment vs. Vacancy Rate (percent) 1. Based on Help Wanted Ads Based on JOLTS 1. Vacancy Rate 3 1. Recessions: Q3 1. 7 7 1 9-9. 7 9 1 1 1. 1. 1. 1..... Unemployment Rate 3 3 Beveridge Curve during the Great Depression also did not sit on a linear curve relative to expansion. th Annual Economic Outlook Symposium December 3, 1 1
. Cyclical: Moreover, what a vacancy is may vary across the cycle. Efforts to fill a vacancy are lower. Recruiting effort down about 1%. Explains about ¼ of hire shortfall S. Davis, J. Faberman, and J. Haltiwanger, The Establishment-level Behavior of Vacancies and Hires, WP Philadelphia Fed. 3. Cyclical. Is the problem related to people being in the wrong sectors? 3 Construction employment growth 1-1 - -3 th Annual Economic Outlook Symposium December 3, 1 13
Employment Growth: Selected Industries, 19:Q-1:Q1 Construction Durable manufacturing Nondurable manufacturing 3 1 1-1 - 3 1-1 - 1 - -1-1 -3-3 - Transportation and utilities Wholesale Trade Retail Trade 1 1 - -1-1 1 1 - - - - -1 1 1 - - - - - - - - Finance, insurance, and real estate 1 - - - Services 1 - - Government 7 Noncyclical Employment Growth: Selected Industries, 19:Q1-1:Q1 Construction Durable manufacturing 1 1 1 1 - - - -1 - - -1-1 - - -1-1 Nondurable manufacturing Transportation and utilities Wholesale Trade Retail Trade - - - - -1 3 1-1 - -3 - - - - - - - - - Finance, insurance, and real Services Government estate. 1. 1. -. - -1-1. - - -. - -3 - Source: Ellen Rissman, Chicago Fed th Annual Economic Outlook Symposium December 3, 1 1
3. Cyclical. Is it industry reallocation? Not likely Noncyclical measure of employment reallocation 7 I * * * * t sit git gt i 1 1/ 3 1 Perhaps mismatch is within industry. Hard to measure. Source: Ellen Rissman, Chicago Fed 9 3. Cyclical: No evidence in the wage data that there is unusual demand in particular sectors. Dispersion of private industry hourly earnings, year-over-year. Highest industry less average Standard deviation of industries..3..1. 1973 197 197 197 1977 197 1979 19 191 19 193 19 19 19 197 19 199 199 1991 199 1993 199 199 199 1997 199 1999 1 3 7 9 1 Average hourly earnings, by private industry (CES). Excludes mining and logging. Red line = highest industry less average industry. Blue line = standard deviation of industry average hourly earnings. 3 th Annual Economic Outlook Symposium December 3, 1 1
. Cyclical. Is it people being in the wrong place? Homeowner migration rates have barely budged.. State to State Migration over month intervals, SIPP..1 renters.1. owners Jan- Sep- May- Jan- Sep- May-7 Jan- Sep- May-9 Jan-9 Sep-9 May-91 Jan-9 Sep-9 May-93 Jan-9 Sep-9 May-9 Jan-9 Sep-9 May-97 Jan-9 Sep-9 May-99 9 Jan- Sep- May-1 Jan- Sep- May-3 Jan- Sep- May- Jan- Sep- May-7 Jan- Sep- May-9-7 1/-11/9 Difference Renters.3. -.3 Owners..1 -. Difference -. ( 1) Similar result if look at states with largest price declines 31. Cyclical. Is it people having the wrong skills? Unemployment has gone up proportionately for highly skilled.1 Unemployment rate: recent college grads*.1 Unemployment rate: Engineers, IT, & Nurses..1. Everyone < college graduates.1. Everyone Engineers, IT, Nurses...3.3.......11...1. 197 197 19 19 19 19 19 199 199 199 199 199 1 197 1977 1979 191 193 19 19 19 199 1991 1993 199 1997 199 7 9 * Also no UI, no homes, highly mobile 3 th Annual Economic Outlook Symposium December 3, 1 1
. Cyclical. Is it people having the wrong skills? Then why not grab folks from other countries? Regular H1B Visa Cap Reached Dates (Month of previous year) 1 1 x 77% Full through Nov. 7 9 1 11 The regular H1B cap has been, since. There are also, Masters exemption visas. 33. Cyclical. Is it people having the wrong skills? Then why not grab folks from other firms? Quits as a share of total employment (*1) 9. 7. 7. Davis, Faberman, and Haltiwanger (1)*.. JOLTS 199 1991 199 1993 199 199 199 1997 199 1999 1 3 7 9 1 *Davis, S, J Faberman, and J Haltiwanger, 1, "Worker Flows and Job Flows in the Cross- Section and Over time, WP Philly Fed. 3 th Annual Economic Outlook Symposium December 3, 1 17
. Cyclical. Lots of people in the door but still not getting enough hours Part-time for economic reasons as a share of total employment (*1). 7 7.... 3.. 1.. 19 19 197 197 19 19 199 199 1 3 Summary of labor markets There is probably some extra impediments in the ability of firms and workers to match right now. NAIRU (in the medium-term) has risen from to -7 percent. A good chunk of this may be temporary (UI). The high unemployment rate is probably primarily due to deficiencies in the demand for goods and services. Little direct evidence of any kind of mismatch problem (at this point). Recession (and mild recovery) has been broad-based. No inflation (prices or wages) As the modest economic recovery continues, the unemployment rate will come down in-line with past relationships between aggregate economic activity and labor market activity. 3 th Annual Economic Outlook Symposium December 3, 1 1
Policy actions Fed Funds Rate (percent) Federal Reserve Assets (Bils. $) 7 History Market Expectations All Other Assets ($1. bil.) TALF ($.9 bil.) MMIFF 1 CPFF AIG ($9.3 bil.) 3 1 Maiden Lane ($.1 bil.) 1 Oct 1 AMLF TSLF PDCF 1999 '1 '3 ' '7 '9 '11 '13 7 9 1 TAF Fiscal policy might help too. E.g. employment subsidies that lower the marginal cost of additional hiring (possibly through payroll taxes), work/job sharing programs, job search assistance, education and skill development (longer-run). 37 th Annual Economic Outlook Symposium December 3, 1 19