New Evidence and Research Direc1ons for Macro and Labor Economists

Size: px

Start display at page:

Download "New Evidence and Research Direc1ons for Macro and Labor Economists"

Kristin Wood
5 years ago
Views:

1 2 nd Annual CUHK- Fudan- Tsinghua Conference on the Chinese Economy New Evidence and Research Direc1ons for Macro and Labor Economists Remarks by Steven J. Davis faculty.chicagobooth.edu/steven.davis/ Tsinghua University, 19 December 2011

2 Two Topics 1. Economic Policy Uncertainty Work with ScoG Baker and Nick Bloom 2. Recessions and the Costs of Job Loss Work with Till von Wachter

3 Topic 1: Outline Policy News and Stock Market Vola1lity A New Index of Economic Policy Uncertainty News- Based Component of the Index Preliminary News- Based Index of Economic Policy Uncertainty for China Evidence that Higher Policy Uncertainty Foreshadows Lower Output, Employment, and Investment Based on Measuring Economic Policy Uncertainty with ScoG Baker and Nick Bloom 3

4 Time Period What Drives Big Stock Market Moves? Event: Daily move in S&P 500 Index Bigger Than +/- 2.5% # of Events Policy News Interest Rates Macro News Earnings News War/ Terror Other % 9% 31% 12% 11% 22% % 3% 35% 12% 0% 10% Other U.S. Recessions % 10% 50% 0% 0% 20% % 9% 9% 9% 73% 0% % 14% 36% 21% 14% 14% Based on the leading cause for the stock market move, as reported in the next day s New York Times.

5 Time Period What Drives Big Stock Market Moves? Fewer than 1 Policy- Driven Event Per Year # of Events Policy News Interest Rates Macro News Earnings News War/ Terror Other % 9% 31% 12% 11% 22% % 3% 35% 12% 0% 10% Other U.S. Recessions % 10% 50% 0% 0% 20% % 9% 9% 9% 73% 0% % 14% 36% 21% 14% 14%

6 Time Period What Drives Big Stock Market Moves? About 12 Policy- Driven Events Per Year # of Events Policy News Interest Rates Macro News Earnings News War/ Terror Other % 9% 31% 12% 11% 22% % 3% 35% 12% 0% 10% Other U.S. Recessions % 10% 50% 0% 0% 20% % 9% 9% 9% 73% 0% % 14% 36% 21% 14% 14%

7 Number of Daily Moves in S&P 500 Bigger than +/- 2.5% by Year & Number Due to Policy News

8 New Index of Policy- Related Economic Uncertainty Our index is built from four components: News- based index (1/2 weight) Forecaster disagreement about infla1on (1/6) Forecaster disagreement about federal government purchases (1/6) Scheduled tax code expira1ons (1/6) 8

9 U.S. Index of Economic Policy Uncertainty Banking Crisis, Obama Election Eurozone Crisis, 2010 Midterms Debt Ceiling; Eurozone Crisis Policy Uncertainty Index Balanced Budget Act 1 st Gulf War Clinton Election Bush Election 9/11 2 nd Gulf War Lehman and TARP Large interest rate cuts January 1985 to October

10 News-based U.S. policy uncertainty index Policy Uncertainty News Index 1 st Gulf War Clinton Election Russian Crisis/LTCM Bush Election 2 nd Gulf War 9/11 Lehman and TARP Stimulus Debate Obama Election, Banking Crisis Debt Ceiling Dispute; Eurozone Crisis Euro Crisis, 2010 Midterm 10

11 Financial Uncertainty Index A validation check of the news-based approach Comparing the VIX to a news- based index of equity market uncertainty Black Monday Russian Crisis/LTCM Asian Crisis 9/11 2 nd Gulf War 1 st Gulf War Lehman Bankruptcy Notes: Frequency of the triple of economy/economic, uncertain/uncertainty and one of a collection of financial market terms (e.g. stock 11 price, equity price, stock market etc ) in US Google news normalized by the smoothed frequency of the word today. Both series scaled to same mean.

Figure 6: Overall & Policy-Related Economic Uncertainty Normalized Number of News Articles Recession Fears Recession Fears Gulf War I 1987 Stock Market Crash Dissolution of USSR Clinton Election

12 Figure 6: Overall & Policy-Related Economic Uncertainty Normalized Number of News Articles Recession Fears Recession Fears Gulf War I 1987 Stock Market Crash Dissolution of USSR Clinton Election Asian Financial Crisis Russian Financial Crisis/LTCM Bush Election Notes: Overall News-Based Economic Uncertainty Index composed of monthly number of news articles containing uncertain or uncertainty as well as economic or economy (scaled by the smoothed number containing today ). Policy Index set such that monthly average value is 100. Index covers January 1985-July Axis shown as a log scale. Query run on August 11, 2011.

13 The Intensity and Composi]on of Policy- Related Economic Uncertainty by Time Period 13

14 The Intensity and Composi]on of Policy- Related Economic Uncertainty by Time Period 14

15 The Intensity and Composi]on of Policy- Related Economic Uncertainty by Time Period 15

16 The Intensity and Composi]on of Policy- Related Economic Uncertainty by Time Period 16

17 The Intensity and Composi]on of Policy- Related Economic Uncertainty by Time Period 17

18 The Intensity and Composi]on of Policy- Related Economic Uncertainty by Time Period 18

19 The Intensity and Composi]on of Policy- Related Economic Uncertainty by Time Period 19

20 News-Based Economic Policy Uncertainty Index for China, Jan to Dec. 2010, Highly Preliminary January 2008, concerns about infla1on and monetary policy 1ghtening October 2008, rural land reform June 10, pressure for yuan apprecia1on February bank reserve ra1o hike and concerns about housing prices Nov infla1on concerns Sep- 05 Mar- 06 Oct- 06 Apr- 07 Nov- 07 Jun- 08 Dec- 08 Jul- 09 Jan- 10 Aug- 10 Feb- 11 Based on counts of Google News ar1cles that include the following sets of terms: (China OR Chinese)AND(Economy OR Economic)AND(Uncertain OR Uncertainty)AND (policy OR poli1cs OR poli1cal OR tax Or regula1on OR monetary OR exchange rate OR capital controls) and scaled by the number of ar1cles that include today.

21 Estimated Industrial Production and Employment Changes after a Policy Uncertainty Shock (Increase) Industrial Production Impact (% deviation) Employment Impact (millions) The shock is scaled to match the size of the increase in the policy uncertainty index from 2006 to Months Months Notes: This shows the impulse response function for Industrial Production and employment to an 124 unit increase in the policy-related uncertainty index, the increase from 2006 (the year before the current crisis) until the first 8 months of The central (black) solid line is the mean estimate while the dashed (red) outer lines are the onestandard-error bands. Estimated using a monthly Cholesky Vector Auto Regression (VAR) of the uncertainty index, log (S&P 500 index), federal reserve funds rate, log employment, log industrial production and time trend. Data from 1985 to

22 Figure 9: Robustness of Estimates to Different VAR Specifications Industrial Production Impact (% deviation) Three months of lags Reverse order Baseline Uncertainty index has equal weight on measures Nine months of lags Bivariate (uncertainty and log industrial production) Adding VIX first as a control for economic uncertainty Months after the policy uncertainty shock Notes: This shows the impulse response function for GDP and employment to an 124 unit increase in the policy-related uncertainty index. Estimated using a monthly Cholesky Vector Auto Regression (VAR) of the uncertainty index, log(s&p 500 index), federal reserve funds rate, log employment, log industrial production and time trend unless otherwise specified. Data from 1985 to 2011.

23 How Does Policy Uncertainty Hold Back the Economy? Five Channels 1. More precau1onary savings, more deleveraging 2. When investment and hiring decisions are costly to reverse, greater uncertainty depresses and delays investment and hiring 3. Debt finance becomes more costly 4. Managers act in a more risk- averse manner 5. Agency problems intensify, reducing the value of new and exis1ng employment, business and financial rela1onships 23

24 More on Economic Policy Uncertainty Uncertainty and the Slow Recovery, Gary S. Becker, Steven J. Davis and Kevin M. Murphy, Wall Street Journal, 3 January Policy Uncertainty Is Choking Recovery, ScoG R. Baker, Nicholas Bloom and Steven J. Davis, Bloomberg View, 5 October 2011 Policy Uncertainty and the Stalled Recovery, ScoG R. Baker, Nicholas Bloom and Steven J. Davis, VOX, 22 October Measuring Economic Policy Uncertainty, ScoG R. Baker, Nicholas Bloom and Steven J. Davis, working paper, 10 October

25 Topic 2: Outline of Remarks Cyclical Movements in the Incidence of Job Loss and Job Displacement Magnitude and Cyclicality of Earnings Losses Associated with Job Displacement Effects of Job Loss in Leading Theore1cal Models of Unemployment and Labor Market Dynamics Based on Recessions and the Costs of Job Loss with Till von Wachter, forthcoming in Brookings Papers on Economic AcBvity

26 Fig. 1. Job Loss Indicators, Quarterly, % of Employment Job Destruction (BED) Unemployment Inflows (CPS) Layoffs (JOLTS) Initial Claims for UI Benefits (Right Axis)

Job Loss Indicators, Quarterly, % of Employment 10.0 9.0 Job Destruction (BED) Unemployment Inflows (CPS) Layoffs (JOLTS) ~9 million per quarter 7.0 6.0 8.0 5.0 7.0 4.0 6.0 3.

27 Job Loss Indicators, Quarterly, % of Employment Job Destruction (BED) Unemployment Inflows (CPS) Layoffs (JOLTS) ~9 million per quarter Initial Claims for UI Benefits ~5 million per quarter

28 Defining Job Displacements in SSA Data 1. A worker is displaced in year y if he Separates from his (main) employer in y Has 3+ years of job tenure with the employer, and Employer experiences a mass- layoff event in y 2. A worker separates in year y if he has earnings with the employer in y- 1 but not in y. 3. Employer criteria for a mass- layoff event in y: 50+ employees in y- 2 Employment contracts by 30-99% from y- 2 to y Employment in y- 2 < 130% of employment in y- 3 Employment in y+1 < 90% of employment in y- 2

29 Fig. 2. Annual Job Displacement Rates, Percent Men, 50 and younger, 3+ Years of Prior Job Tenure Displacement Rate Based on a 1% random sample of persons with SSA earnings Job Destruc1on At Firms with 50+ Employees (leq axis)

30 Fig. 2. Annual Job Displacement Rates, Percent About 1/3 of workers in SSA data are at firms with 50+ employees and have 3+ years of job tenure 20% cumula1ve displacement 5 from 1980 to 1985 ~ million persons

31 Estimating the Dynamic Pattern of Annual Earnings Losses for Workers Displaced in Displacement Year y Distributed Lag Model Quartic von in Wachter, Age, etc. Song, Manchester (2011) Individual and Calendar Year Fixed Effects Allowing for differential earnings trends k= years since or to job displacement Event Job displacement in y, y+1, y+2 Control Group Workers not separating from employers in y, y+1, y+2 (same tenure and 50+ requirements) Identification Evolution of control group earnings Assumption is a valid counterfactual for earnings of displaced workers in the absence of job displacement, conditional on controls

32 Fig 5C. Average Annual Earnings Losses Before and A`er Job Displacement Rela]ve to Control Group Earnings, Frac]on of Pre- Displacement Annual Earnings, Men 50 or Younger with At Least 3 Years of Job Tenure

33 Fig. 6. Annual Earnings Losses in the Third Year of Job Displacement, Men 50 or Younger with at Least 3 Years of Job Tenure Prior to Displacement Earnings Loss Relative to Pre- Displacement Annual Earnings Unemployment Rate in Displacement Year (Year 1)

Constructing Estimates of Present Value Earnings Losses The estimated average annualized log earnings loss between years 6-10 and 11-15, estimated using data for all displacement years Average annual

34 Constructing Estimates of Present Value Earnings Losses The estimated average annualized log earnings loss between years 6-10 and 11-15, estimated using data for all displacement years Average annual decay rate of losses in years 11 to 20 after displacement. Estimated earnings loss in year s from displacement, using average of estimated loss in all or selected displacement years. 5% annual discount rate Estimated earnings loss in year 10 from displacement

35 Table 1. Magnitude of PV Earnings Loss at Job Displacement in Mass Layoffs from : Men, 50 or Younger with 3+ Years of Prior Job Tenure

36 Fig. 7. PV Earnings Losses over 20 Years from Displacement, Men 50 or Younger with 3+ Years of Prior Job Tenure PV Earnings Loss As a Multiple of Pre- Displacement Annual Earnings Unemployment Rate in Displacement Year

37 Table 4. PV Income and Earnings Losses Due to Job Loss in the Basic MP Model

38 ! MP Model of Burgess and Turon (2010) Start with Mortensen and Pissarides (1994) Add costly search on the job and other changes The Model Yields: Worker flows apart from job flows Heterogeneity in productivity and wages A job ladder Job loss spikes due to negative aggregate shocks!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!"#$%&'!()!*'#!+(,!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!-#./$&#0#)*!12%2)3!!!!!!!!!!45!!!!!!!!!!!!!!-6!7!!!!!!!!!!!!!!!!!!!!!!86!7!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!"6!7!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!5!!!!!!!! "#$%&'()*+,$)!-*%#.),$/$,'!01*2&1%3#&!4%*!5%6! 72&,*.),$%(8!92:3+)2;2(,!<$*$(=!+(#!>2+*)1!%(!,12!5%6!!

able 5. PV Earnings Losses Due to Job Loss in Model of Burgess and Turon (2010), MP(1994) with Search on the Job Note: We calibrate the model to match U.S. job-finding rates.

40 able 5. PV Earnings Losses Due to Job Loss in Model of Burgess and Turon (2010), MP(1994) with Search on the Job Note: We calibrate the model to match U.S. job-finding rates. Burgess and Turon calibrate their model to match features of the British economy from 1964 to Using their calibration, which entails much lower job-finding rates for unemployed workers, the Mean PV earnings losses range from 4.4% to 5%.

41 Table 5, Con1nued

42 Summary 1. Job displacement brings large PV earnings losses Unemployment Rate in Displacement Year Less than 6% More than 8% Full Sample, Mean PV Losses as Mul1ple of Pre- Displacement Earnings (Men) 1.4 Years 2.8 Years 1.7 Years 2. Incidence of job loss & displacement rises in recessions and slumps, especially severe ones 3. Job displacement leads to many other nega1ve consequences for workers and their families 4. Job loss is a rather inconsequen1al event in leading models of unemployment fluctua1ons MP models do not account for magnitude of PV losses, or the sensi1vity to condi1ons at the 1me of displacement They miss why workers and policymakers are so concerned about job loss, recessions and unemployment

43 Direc1ons for Research Put specific human capital into DMP models Learning about match quality (Jovanovic, 1979) Learning by doing on the job Investments in job- specific training (Becker, 1962) Account for (other sources of) worker rents Pay equity and fairness norms (Akerlof and Yellen, 1982) High pay to deter shirking (Bulow and Summers, 1986) Appropria1on of quasi- rents on capital (Grout, 1986) Worker sharing of product market rents Downward s1ckiness due to contrac1ng and one- sided commitment by firms (Beaudry and DiNardo, 1992)

44 Direc1ons for Research, 2 Gradua1ng in a Recession Workers who enter the labor market when condi1ons are slack suffer persistent nega1ve effects on future earnings (Kahn, 2010) Las1ng declines in employer quality and las1ng effects of low star1ng wages on wage growth within firms (Oreopolous, von Wachter and Heisz, 2010) These results suggest that weak condi1ons at the 1me of labor market entry (or displacement) slow the accumula1on of rents and specific human capital for many years thereaqer.

45 Direc1ons for Research, 3 Who recovers from displacement events and who doesn t? Standard measures of skills, educa1on and cogni1ve ability Social network (size and character) Do personality traits mager and, if so, how much? Resilience, op1mism, perseverance, sociability, organiza1onal skills and prac1ces, propensity to plan, attudes, etc.

46 Additional Slides Not for Prepared Remarks

Policy uncertainty also seems to make stock markets more risky (as stocks move together more) Reproduced from Political Uncertainty and Risk Premia, by Lubos

47 Policy uncertainty also seems to make stock markets more risky (as stocks move together more) Reproduced from Political Uncertainty and Risk Premia, by Lubos Pastor and Pietro Veronesi, University of Chicago, 20 November Political Uncertainty Index is the Economic Policy Uncertainty Index from Baker, Bloom and Davis.

48 Figure 7: Relationship of News-Based Index of Overall Economic Uncertainty to News-Based Index of Policy-Related Economic Uncertainty log(economic Uncertainty) R-Squared: 0.68 Slope: 0.79 (0.05) R-Squared: 0.53 Slope: 1.50 (0.19) R-Squared: 0.88 Slope: 0.98 (0.03) to August 2001 September 2001 Onwards log(policy Uncertainty)

49 Index of Future Tax Code Expirations Tax Legislation Expiration Index Notes: Utilizes List of Tax Expirations from the Joint Committee on Taxation. Each year s forecast is a 10-year horizon 49 of expiring tax laws. Future expirations weighted by 0.5^((T+1)/12) where T is the number of months until the tax expires

50 Our dollar (rather than equal) weighted tax expira1on index shows an even steeper rise Indexes of Scheduled Tax Code Expirations Based on CBO Data, Preliminary 50

51 CPI Forecaster Disagreement index CPI Forecasters IQ Range Index Balanced Budget Act Budget Battle 1 st Gulf War Clinton Election 2 nd Gulf War/ Fed Drops Interest Rates Obama Election, Banking Crisis Notes: From the Philadelphia Federal Reserve Survey of Professional Forecasters. Takes the interquartile (IQ) range of the 1-year ahead forecasts 51 (made every quarter) of total federal government expenditures relative to the mean forecast. Normalized to a mean 100 index prior to Aug 2011.

52 Federal Government Purchases Disagreement Index Federal Expenditures Forecasters IQ Range Index Balanced Budget Act Budget Battle Clinton Election 9/11 Obama Election, Banking Crisis Notes: From the Philadelphia Federal Reserve Survey of Professional Forecasters. Takes the interquartile (IQ) range of the 1-year ahead forecasts 52 (made every quarter) of total federal government expenditures relative to the mean forecast. Normalized to a mean 100 index prior to Aug 2011.

53 Figure 10: Quarterly VAR estimates for GDP and investment Months after the policy uncertainty shock GDP Impact (% deviation) Investment Impact (% deviation) Notes: Shows the impulse response function to an 124 unit increase in the policy-related uncertainty index, the increase from 2006 (the year before the current crisis) until the first 8 months of The central (black) solid line is the mean estimate while the dashed (red) outer lines are the onestandard-error bands. Estimated using a quarterly Cholesky VAR: the uncertainty index, log (S&P 500 index), federal reserve funds rate, log employment, log investment, log consumption and log GDP. Data from 1985 to 2011.

54 Figure 11: Estimates after including controls for consumer confidence Months after the economics policy uncertainty shock Industrial Production Impact (% deviation) Consumer confidence included second in the VAR Consumer confidence included first in the VAR Notes: This shows the impulse response function for Industrial Production and employment to an 124 unit increase in the policy-related uncertainty index, the increase from 2006 (the year before the current crisis) until the first 8 months of The central (black) solid line is the mean estimate while the dashed (red) outer lines are the onestandard-error bands. Estimated using a monthly Cholesky Vector Auto Regression (VAR) of the uncertainty index, log (S&P 500 index), federal reserve funds rate, log employment, log industrial production and time trend. Data from 1985 to Top panel includes the Michigan Consumer confidence index included as the second variable after our uncertainty index, and the bottom panel includes the Michigan Consumer Confidence index included as the first variable.

55 Figure 13: News-Based China and Japan Competition Indexes China and Japan Competition Index Notes: News-Based China and Japan Competition Index composed of monthly number of news articles containing competition and economy and Japan or China (scaled by the smoothed number of articles containing today ). Query run August 26, Index covers Jan 1985-Aug 2011.

56 Appendix Figure A1: News-Based Energy Uncertainty Index Arab Spring Oil Spike Energy Uncertainty Index 1 st Gulf War 2 nd Gulf War Oil Spike Notes: Energy Uncertainty Index composed of monthly number of news articles containing uncertain or uncertainty as well as the term energy (scaled by the smoothed number of articles containing today ). Google query run June 15, Index covers January 1985-May 2011

57 Appendix Figure A3: News-Based Middle East Uncertainty Index Arab Spring Middle East Uncertainty Index 1 st Gulf War 9/11 2 nd Gulf War Notes: News-Based Middle East Uncertainty Index composed of monthly number of news articles containing uncertain or uncertainty as well as the term Middle East (scaled by the smoothed number of articles containing today ). Google query run June 15, Index covers January May 2011

Google economic policy uncertainty citation index is highly correlated with a narrower 5-paper index Correlation=0.

58 Google economic policy uncertainty citation index is highly correlated with a narrower 5-paper index Correlation= paper index is the frequency of uncertain, economic/y and our key policy terms in the USA Today, Miami Herald, LA Times, Wash Post, and Chicago Tribune, normalized by today

59 Fig. 3. Annual Displacement Rates by Tenure, Men 50 or Younger at Firms w/50+ Employees

60 Fig. 4. Annual Displacement Rates by Age, Men w/3+ Years Tenure at Firms w/50+ Employees

61 Table 2. Magnitude of PV Earnings Loss at Job Displacement in Mass Layoffs from : Men, 50 or Younger with 6+ Years of Prior Job Tenure and Women, 50 or Younger with 3+ Years of Tenure

62 Worker Anxie1es about Job Loss and other Labor Market Prospects Percep1ons about job loss and the difficulty of job finding track actual unemployment rates Worker anxie1es about job loss, hours cuts, wage and benefit cuts rose sharply aqer the 2008 financial crises and remain very high These high anxiety levels likely produce important stresses and psychological costs for a large segment of the popula1on.

63 Fig. 8. Perceived Likelihood of Job Loss in Next 12 Months, All Available Years in the General Social Survey, 1977 to 2010

64 Fig. 9. Perceived Likelihood of Job Finding, All Available Years in the General Social Survey, 1977 to 2010

65 Figure 10. Perceived Ability to Find a Quality Job, March 2002 to June 2011, Gallup Polling

66 Table 3. Worker Anxiety Rose Sharply in the Wake of the 2008 Financial Crisis and Have Remained High

Measuring Economic Policy Uncertainty

Measuring Economic Policy Uncertainty Scott R. Baker (Stanford) Nick Bloom (Stanford & NBER) Steve Davis (Chicago Booth & NBER) Princeton, February 2012 Policy uncertainty has recently been argued to be