The Equality Impact of the Employment Crisis Elish Kelly, Gillian Kingston, Helen Russell, Fran McGinnity

The Equality Impact of the Employment Crisis Elish Kelly, Gillian Kingston, Helen Russell, Fran McGinnity The Economic and Social Research Institute Irish Economy Conference: Learning from the Crisis 25 February 2015

Background - Recession in Ireland: Labour Market Collapse and Austerity Boom period 1994-2007: Numbers employed almost doubled (from 1.2 million to 2.1 million) Key characteristic of the period: increase in female employment; also immigration, very low unemployment, property/construction boom. Recession hit in 2008: Banking and fiscal crisis (Troika bailout); unemployment soared - massive job losses in the private sector, especially in construction; emigration rose rapidly. Programme of austerity: Public sector changes pension levy, recruitment freeze, pay cuts, increase in hours (O Connell, 2013). But budgetary changes under austerity affected all employed, including the private sector - universal social charge, property tax, social welfare cuts (Callan et al. 2012)

Impact of the Recession Considerable debate on where the costs of the recession have fallen across society in terms of income groups, social class, age groups, household types, gender, nationality (e.g., Barry & Conroy 2013; Callan et al, 2012; Gerlach-Kristen, 2013; Keane et al. 2014, Whelan, et al., 2015) Increase in male unemployment and decrease in participation narrowed the gaps in male and female employment and labour market participation (LMP) (Russell et al., 2014) Young people have been severely affected, particularly in terms of unemployment (Kelly et al., 2014) Certain immigrant and national-ethnic groups have also suffered more than natives (Barrett et al., 2012; Kingston et al., 2012)

Research Questions Paper focuses on the equality impact of the labour market crisis : which groups covered by the Equality Acts have experienced the greatest changes in their labour market fortunes? Full report assesses labour market (employment, unemployment and LFP), income and poverty situation of the groups covered by the Equality Acts for which labour market data exists (5 of the 9 equality grounds) Today s presentation focuses on 3 grounds : i) gender, ii) age, and iii) nationality

Labour Market Outcomes Examined Focus on the impact of the recession on the unemployment risk (ILO measure) of the various equality grounds studied Also, employment in construction sector examined

Theoretical Perspectives: Equality Groups and the Labour Market in Recession Reserve army of labour (or buffer): idea is that women are carers first, workers second and act as a labour reserve (Holst 2000) ; migrants will go home thus acts as a shock absorber (Borjas, 2001) Insider-Outsider those outside the labour market will find it hard to get back in (e.g. young people) Primary and secondary jobs (Doeringer and Piore 1971; Edwards et al. 1975): to the extent that women, non-irish and young people have more flexible, insecure jobs more easily adjusted in the downturn Segregation perspective (Bettio and Verashchagina 2013): concentration of women, young people and non-irish in sectors (or occupations) job losses among groups will be strongly affected by sectoral employment losses

Sectoral Changes in Employment Between 2007 and 2012 Agriculture Manufacturing Construction Retail/wholesale Transport Accomodation & Food Information& commun Financial services Profess, scientific Admin & Support Services Public Admin & Defence Education Health Arts & other servces -200,000-150,000-100,000-50,000 0 50,000

Data Quarterly National Household Survey (QNHS) Measure the situation of groups before and after recession: Q4 2007 (Pre Recession) and Q4 2012 (Post Recession) Cross-sectional data - unable to take account of migration with the QNHS data Investigate differences in i) unemployment and ii) construction sector employment between equality groups in 2007 and 2012, and whether group differences changed over time

Modelling Strategy Probit models using probit and dprobit commands in Stata Dependent variable 1=unemployed 0=employed for unemployment model; and 1 = employed in construction sector 0 = rest of working age population (15-64) Self-employed excluded Models control for age, gender, nationality, marital status, family status, education and geographic region Second specifications estimated in which we included 2012 year interaction terms to test for differences over time

Unemployment Rates of Selected Equality Groups: 2007 and 2012 0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 Male Female Age 15-19 Age 20-24 Age 25-34 Age 35-44 Age 45-54 2007 2012 Age 55-64 Irish UK EU-13 NMS Africa

Probit Model of Unemployment: 2007 and 2012 (Marginal Effects) 2007 2012 Significant Change Gender (Ref Male): Female -0.020*** -0.082*** Yes (0.002) (0.011) Age (Ref Age 35-44): Age 15-19 0.030*** 0.077*** No (0.004) (0.018) Age 20-24 0.021*** 0.072*** Yes (0.003) (0.014) Age 25-34 0.007*** 0.007 Yes (0.003) (0.005) Age 45-54 -0.003** -0.006 No (0.002) (0.008) Age 55-64 -0.017*** -0.001 Yes (0.003) (0.004) Nationality (Ref Irish): UK 0.047*** 0.098*** No (0.015) (0.003) EU-13-0.000 0.007 No (0.013) (0.014) New Member States (NMS) 0.001 0.028*** Yes (0.003) (0.007) Africa 0.082*** 0.225*** Yes (0.012) (0.011) Asia 0.004 0.009 No (0.023) (0.012) North America, Australia and Oceania 0.057-0.007 Yes (0.050) (0.054) Rest of Europe/World 0.035** 0.111 No (0.015) (0.077)

Probit Model of Unemployment: 2007 and 2012 (Marginal Effects) 2007 2012 Significant Change Educational Attainment (Ref Primary or Less) Lower Secondary -0.010*** -0.005 Yes (0.002) (0.009) Upper Secondary -0.031*** -0.087*** No (0.002) (0.008) Post Secondary -0.030*** -0.036*** Yes (0.001) (0.008) Ordinary Degree -0.036*** -0.102*** Yes (0.001) (0.004) Higher Degree -0.044*** -0.139*** Yes (0.005) (0.011) Post-Graduate Degree -0.038*** -0.126*** No (0.002) (0.009) Observations 30,853 21,913 Pseudo R-squared 0.0676 0.108 Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1

Probit Model of Unemployment: Gender*Age Interactions (Marginal Effects) 2007 2012 Change Reference: Males Aged 35-44 Males: Age 15-19 0.028*** 0.056** No (0.007) (0.022) Age 20-24 0.019*** 0.070*** Yes (0.005) (0.006) Age 25-34 0.012*** 0.001 Yes (0.002) (0.012) Age 45-54 0.005 0.007 No (0.003) (0.009) Age 55-64 -0.015*** 0.021* Yes (0.003) (0.012) Females: Age 15-19 0.002 0.035 Yes (0.008) (0.026) Age 20-24 0.003 0.002 No (0.008) (0.025) Age 25-34 -0.010*** 0.014 Yes (0.003) (0.018) Age 35-44 -0.014*** -0.076*** Yes (0.004) (0.012) Age 45-54 -0.017*** -0.031*** Yes (0.003) (0.004) Age 55-64 -0.008* -0.049*** Yes (0.004) (0.017) Observations 30,853 21,913 Pseudo R-squared 0.0685 0.109

Probit Model of Employment in the Construction Sector (Marginal Effects) 2007 2012 Significant Change Gender (Ref Female): Male 0.110*** 0.031*** Yes (0.006) (0.001) Age (Ref Age 35-44): Age 15-19 -0.019*** -0.010*** Yes (0.002) (0.001) Age 20-24 0.022*** -0.005*** Yes (0.006) (0.001) Age 25-34 0.016*** 0.001 Yes (0.002) (0.002) Age 45-54 -0.011*** -0.001* Yes (0.001) (0.001) Age 55-64 -0.018*** -0.005*** No (0.002) (0.001) Nationality (Ref Irish): UK/EU-13-0.010*** -0.002** Yes (0.001) (0.001) New Member States (NMS) 0.026*** -0.003* Yes (0.001) (0.002) Other -0.023*** -0.005*** Yes (0.002) (0.001)

Probit Model of Employment in the Construction Sector (Marginal Effects) 2007 2012 Significant Change Educational Attainment (Ref Primary or Less) Upper Secondary -0.002-0.001 No (0.005) (0.001) Post Secondary 0.021*** 0.012*** Yes (0.005) (0.003) Ordinary Degree -0.015*** -0.003*** Yes (0.002) (0.001) Higher / Post-Graduate Degree -0.032*** -0.010*** Yes (0.002) (0.001) Location (Ref Dublin): Border 0.007** 0.001 No (0.004) (0.003) Midlands 0.016*** 0.001 Yes (0.003) (0.001) West 0.009*** 0.003 No (0.002) (0.002) Mid-East 0.010*** 0.005* No (0.003) (0.003) Mid-West 0.009*** 0.005* No (0.001) (0.003) South-East 0.015*** 0.003** Yes (0.003) (0.001) South-West 0.005 0.004* No (0.004) (0.003) Observations 46,337 33,637 Pseudo R-squared 0.214 0.161

Summary of Findings I Gender: women were not buffers in the Irish recession, this is also true elsewhere in Europe (Bettio and Verashchagina 2013) Appears females were shielded from unemployment by segregation and low concentration in construction; younger women (somewhat) by higher educational qualifications. Age: young people face particularly high unemployment: difficulties entering the labour market in Ireland; Concentrated in cyclically sensitive sectors during 2007 (e.g., construction and retail); Low-educated particularly affected - penalty for low education has risen (see also Bell and Blanchflower, 2011).

Summary of Findings II Non-Irish: higher unemployment in 2007, though change over time same as for Irish African unemployment rose sharply; So did NMS nationals, but the rise in their unemployment may have been more severe if it wasn t for emigration (Slide). Research does not examine wider recession effects e.g., cuts in public services. Further research could investigate equality between groups in recovery (need to wait for data).

Emigration flows, Ireland 2006-2014 100 90 80 70 60 50 40 30 20 Non-EU EU 10/12 EU 13 UK Irish 10 0 2006 2007 2008 2009 2010 2011 2012 2013 2014 Large outflow of NMS during recession has evened off; non-eu emigration remained fairly stable throughout

Summary of Findings II Non-Irish: higher unemployment in 2007, though change over time same as for Irish African unemployment rose sharply; So did NMS nationals, but the rise in their unemployment may have been more severe if it wasn t for emigration (Slide). Research does not examine wider recession effects e.g., cuts in public services. Further research could investigate equality between groups in recovery (need to wait for data).

Conclusions Overall, there are no clear winners : unemployment rose for all groups Though young adults and men have seen bigger changes than 35-44 age group and women, and also NMS and African nationals compared to natives. Unemployment findings highlight the importance of implementing the Youth Guarantee. Also, importance of higher levels of education as protection against unemployment. Need for activation policy to focus on providing retraining/education for the unemployed in those sectors where future growth, and thus jobs, is forecast.

Thank you.

What happened to labour market participation patterns? Model Results Greater decrease in LMP among men than women in recession: levelling down. Women not acting as buffers Youngest age groups (under 25) have lowest LMP and experience sharpest fall between 07 and 12 For 15-19 group this has was due to greater education participation; for 20-24yrs NEET rate doubled NMS nationals have higher LMP rates than Irish nationals: all other groups have lower LMP rates Over recession greater decline in LMP for NMS nationals and Asian group compared to Irish nationals (some buffer effect)

Changes in Labour Market Participation: 15-64 years