Understanding the Decline in Frequency

Transcription

1 2014 Annual Issues Symposium Understanding the Decline in Frequency Harry Shuford, PhD Practice Leader and Chief Economist NCCI Holdings, Inc. May 9, 2014 Orlando, Florida

2 The Mystery of the Disappearing Frequency Is It Real? Can It Continue? 2

3 Old Lessons: What Drives Frequency? Old Lessons and New The downtrend goes back 80 years The cycles track the economy New Observations: The downtrend is a global phenomenon See for example: Narrative to Accompany Barriers to Reporting, Lenore S. Azaroff, in Use of Workers Compensation Data for Occupational Injury & Illness Prevention, April 2010; How Much Work-Related Injury and Illness Is Missed by the Current National Surveillance System?, KD Rosenman, et al., Journal of Occupational and Environmental Medicine, April 2006; Has the Burden of Disabilities for Workers Changed?, Peter Rousmaniere, workcompcentral.com, April 7,

4 Old Lessons 4

5 Looking Primarily at the BLS Data 1. Using data from the US Bureau of Labor Statistics (BLS) 2. Linked to Occupational Safety and Health Administration (OSHA) reporting 3. Incidence rates for workplace injuries All reported cases Just those resulting in lost work time 4. This is similar to, but more comprehensive than, workers compensation frequency 5

6 Searching for the Factors Driving the Change in Frequency Looking at 50 Years of Experience Frequency changes reflect both: A long-term tendency to decline A cyclical pattern around that trend 6

7 Searching for the Factors Driving the Change in Frequency Looking at 50 Years of Experience Swings in the business cycle are associated with similar swings in frequency: Downward pressure in recession Upward pressure during periods of robust growth 7

8 Frequency: The Long-Term View 8

9 Frequency: A Long-Term Drift Downward Manufacturing Total Recordable Cases Rate of Injury and Illness Cases per 100 Full-Time Workers 25 WW II Depression Young Boomers OSHA 10 5 OSHA fines 0 '26 '28 '30 '32 '34 '36 '39 '41 '43 '45 '47 '49 '52 '54 '56 '58 '60 '62 '65 '67 '69 '71 '73 '75 '78 '80 '82 '84 '86 '88 '91 '93 '95 '97 '99 '01 '04 '06 '08 '10 '12 Recessions indicated by gray bars Source: US Bureau of Labor Statistics (BLS) 9

10 Frequency Since 1990 Is This Period Different? 10

11 Incidence Rates Are Dropping in Manufacturing and Total Private Sector : Frequency of Disabling Injuries per Million Man-Hours Worked; : Injury and Illnesses Cases per 100 Full-Time Equivalent Workers, Total Recordable Cases; no data point available for 1971; effective 2003, based on NAICS Frequency of observation: annual; latest available data point: 2012 Sources: Historical Statistics of the United States: Colonial Times to 1970, Census Bureau, 1975; Statistical Abstract of the United States, Census Bureau, various years; Timothy Webster, 1999, US Bureau of Labor Statistics (BLS) 11

12 The Change in Frequency in the 1990s Key Findings The decline in the 1990s was broad-based: Across virtually all states Across industries and occupations Across virtually all injury demographics including: Gender Event Source Body part injured 12

13 The Change in Frequency in the 1990s The declines are not due to shifts from higher frequency to lower frequency: Industries Occupations Workers 13

14 Looking at Industries and Occupations Key Findings In spite of the dramatic declines, no change in the relative position of industries and occupations In general: The most risky remain the most risky The safest are still the safest 14

15 Frequency and Economic Structure The Frequency Decline Is Primarily Due to Workplaces Getting Safer by Design If no change in industrial mix Observed Injury and Illnesses Cases per 100 Full-Time Equivalent Workers, Total Recordable Cases, All Private Industry Frequency of observation: annual; latest available data point: 2009; no data points are available for 2001 and 2002 due to changes in industry classification; tick marks indicate beginning of year; data points are mid-year Sources: Bureau of Economic Analysis (BEA), US Bureau of Labor Statistics (BLS), 15

16 The Change in Frequency in the 1990s The downtrend in frequency in the 1990s: It s Remarkable and It s Global 16

17 Incidence Rates for Nonfatal Injuries Have Been Declining in Major Industrialized Nations Incidence Rates Indexed so That Rate in 1995 = Index (1995 = 100) Canada France Germany United Kingdom United States Sources: Industrial Labor Organization (ILO); US Bureau of Labor Statistics (BLS) 17

18 Why Do Injury Rates Tend to Decline Over Time? 18

19 The Long-Term Decline in Frequency What can explain this tendency for frequency to decline over extended periods of time? According to the Federal Reserve Bank of Dallas: Competitive labor markets require continuing improvement in working conditions and productivity. Source: Have a Nice Day, Annual Report 2000, Federal Reserve Bank of Dallas 19

20 The Change in Frequency in the 1990s The downtrend in frequency in the 1990s: Is it more than technology and competitive markets? Workplace safety likely becomes more valuable to workers as worker income increases If so, this should be embedded in international data on workplace safety 20

21 An Analysis of Trends in International Work-Related Fatality Rates 21

22 International Fatality Rates Data Sources Trends Over Time The Impact of Wealth Regional Differences 22

23 Data Sources International Labour Organization (ILO) Frequency rate of fatal occupational injury per 100,000 workers (fatality rate) World Bank Gross domestic product per capita, in current US dollars (wealth) 23

24 International Work-Related Fatality Rates Across Countries and Over Time The Data Fatalities per 100,000 Workers (log)

25 International Work-Related Fatality Rates Across Countries and Over Time The Data Median Middle 50% Fatalities per 100,000 Workers (log) Middle 80%

26 Trends Over Time Fatality Rates (log) Median Fatalities per 100,000 Workers (log)

27 Trends Over Time Generally, fatality rates have declined over time There appear to be two kinks: After 1991 A coincidence? After 2007 The global financial crisis 27

28 Linear Regression of Fatality Rates on Time 6 5 y = x Fatalities per 100,000 Workers (log)

29 Trends Over Time Generally, fatality rates have declined over time Linear regression indicates an annual decrease of approximately 4.3%* over the last 30 years In any given year, what explains the variation among countries? * = e ^ ( 0.044) 1 29

30 International Fatality Rates Is there a wealth effect? 30

31 Linear Regression of Fatality Rates on GDP: Lower Fatality Rates With Higher Wealth y = x Fatalities per 100,000 Workers (log) GDP per Capita (log) 31

32 The Impact of Wealth Generally, fatality rates are lower for wealthier countries Linear regression indicates that over the last 30 years, every 10% increase in GDP per capita was associated with a reduction in the fatality rate of approximately 3.6%* But what about the underlying trends in wealth over time? * = 1.1 ^ ( 0.385) 1 32

33 Trends Over Time in GDP per Capita On Average, Wealth Has Been Trending Upward Does This Account for the Decline in Fatality Rates? GDP per Capita (log) Median Middle 50% Middle 80%

34 The Impact of Wealth Incorporating variables for time and wealth in a single regression model: ln(fatality) ~ * ln(gdp) * Year Both coefficients are significant at the 5% significance level, indicating a significant relationship between fatality rate, wealth, and time In a given year, a 10% increase in a country s GDP per capita was associated with a 3.1%* decrease in the fatality rate For a given level of wealth, fatality rates decreased approximately 2.1%** per year * = 1.1 ^ ( 0.334) 1 ** = e ^ ( 0.021) 1 34

35 Fatality Rates, Wealth, and Time There Is Nothing Systematic in the Residuals ln(fatality) ~ * ln(gdp) * Year Residual Fitted Fatality Rate per 100,000 Workers (log) 35

36 Comparison of Models Wealth Wealth and Time Coefficients (GDP) (GDP) (Year) AIC BIC Log likelihood* The model incorporating both wealth and time shows us that: There is a gradual decline in fatality rates over time Wealthier countries tend to have lower fatality rates *p value = 1 pchisq(2( ), df = 1) = x

37 Regional Differences There is considerable variation across countries and regions in the level and growth rates of fatality rates and GDP per capita Do the relationships between fatality rates, wealth, and time vary by geographic region? 37

38 Regional Patterns in Fatality Rates: Different Levels, Similar Downtrends 3.5 Fatalities per 100,000 Workers (log) East & South Asia Europe & Central Asia Latin America & Caribbean Middle East & North Africa North America Sub-Saharan Africa Median 38

39 Regional Patterns in GDP per Capita: Different Levels, Similar Trends 12 GDP per Capita (log) East & South Asia Europe & Central Asia Latin America & Caribbean Middle East & North Africa North America Sub-Saharan Africa 39

40 Modeling Fatality Rates on Wealth and Time by Region Region Wealth Coefficient Time Coefficient All Regions * * East & South Asia * * Europe & Central Asia * * Latin America & Caribbean * Middle East & North Africa * * North America * Sub Saharan Africa * * Indicates significance at the 99% confidence level 40

41 Analysis of International Fatality Rates Summary of Results Over the last 30 years, occupational fatality rates have declined globally Higher levels of wealth (as measured by GDP per capita) are associated with lower rates of occupational fatality rates A decline in international occupational fatality rates is apparent even after accounting for the wealth effect While different regions have different levels of wealth, the relationships with wealth and the passage of time are common for all regions 41

42 Is the Data on Workplace Safety Credible? Old Lessons: The downtrend goes back 80 years New Observations: The downtrend is a global phenomenon The evidence strongly supports the view that workplace injuries and fatalities are falling in the United States and globally. See for example: Narrative to Accompany Barriers to Reporting, Lenore S. Azaroff, in Use of Workers Compensation Data for Occupational Injury & Illness Prevention, April 2010; How Much Work Related Injury and Illness Is Missed by the Current National Surveillance System?, KD Rosenman, et al., Journal of Occupational and Environmental Medicine, April 2006; Has the Burden of Disabilities for Workers Changed?, Peter Rousmaniere, workcompcentral.com, April 7,

43 2014 Annual Issues Symposium Thank you