Union membership and the union wage Premium in Ireland Frank Walsh School of Economics University College Dublin Frank.walsh@ucd.ie Conference on Irish Economic Policy Are Trade Unions good or bad for employment and efficiency Theory suggests it depends on: • The objective of unions • The level and degree of co-ordination in bargaining • The legal/institutional framework and how competitive labour and product markets are Empirical literature • Impact of unions on firm performance and productivity is mixed • Across countries evidence that unions increase unemployment unless there is co-ordination across firms and sectors Conference on Irish Economic Policy • Arguably if unions are un-representative or if membership is unbalanced co-ordination across sectors is more difficult • Walsh 2009 looked at trends in Trade Union membership 2001 2006 • No matter how you looked at the data there seemed to be a decline in the percentage of employees who were members • Changes in composition of job and worker type could not explain this decline • The exception is Public Administration • This is consistent with international evidence of decline in membership across a wide range of countries (concentrated in the private sector) Conference on Irish Economic Policy Quarterly National Household Survey quarter two 2003/11 0.40 0.38 0.37 0.36 0.35 0.34 0.34 0.34 0.33 0.33 0.32 0.32 0.32 0.32 0.30 % Employees who are union members 0.29 0.28 0.28 % Labour force who are union members 0.27 0.26 0.26 0.25 0.25 0.24 0.24 0.24 • Density recovers during recession as a share of employees but continues to decline as a share of labour force • Indicates decline in union number of employees less than in non-union 0.22 2011 2010 2009 2008 2007 2006 2005 2004 2003 0.20 This does not include • Non-employed members • Non-members covered by union contracts but not members • Other agreements: ( JLC’s, REA’s) Conference on Irish Economic Policy Quarterly National Household Survey quarter two 2003/11 • Boom and bust cycle much greater for total number employees than for union employment 135 130 125 120 118 115 118 Index of number of employees 115 Index of number of union members 110 108 Index of number of non-union employees 107 105 103 102 100 100 99 97 98 100 102 101 97 95 93 90 91 2003 2004 2005 2006 2007 2008 2009 2010 2011 Conference on Irish Economic Policy Percentage members in Public vs. Private sector • International literature shows a decline in membership across most countries over time • Very of often the decline greater for private sector workers • This trend is pronounced in the Irish case Conference on Irish Economic Policy Quarterly National Household Survey quarter two 2003/10 0.7 0.6 0.62 0.61 0.62 0.59 0.59 0.6 • Steady decline in private sector density continues throughout 0.61 0.61 0.5 0.4 0.3 0.2 % Public sector who are union members 0.29 0.27 • The density is fairly stable for public sector employees % Private sector who are union members 0.25 0.23 0.23 0.22 0.22 0.21 0.1 0 2003 2004 2005 2006 2007 2008 2009 2010 Proxy for Public Sector is Public Administration, Education and Defence. Private sector is all other employees Conference on Irish Economic Policy Changes in the composition of jobs/workers • Is the pattern in membership reflecting a change in the composition of the workforce/jobs or is there an underlying trend? • We can look at the probability that an employee is a union member in each year and control for worker and job characteristics • Controls are for: age, gender, education, nationality, region, urban status, industry, occupation and firm size • Create an index starting at 100 and compare the raw density with the change in probability of membership when control for worker and job characteristics Conference on Irish Economic Policy Quarterly National Household Survey quarter two 2003/10 101.0 100.0 When we control for changes in composition there is a steady decline in union density over the period 100.0 99.0 98.0 98.1 97.9 97.0 96.6 96.0 96.3 96.1 96.1 95.6 95.5 95.6 95.0 94.0 95.0 Index of membership (No Controls) 94.8 94.3 Index of membership (With Controls) 94.4 93.0 92.0 91.0 2003 2004 2005 2006 2007 2008 2009 2010 Conference on Irish Economic Policy • What if follow the same workers over time and observe those who switch union status. Does their wage rise or fall? Ability is fixed • A problem with this is that a small % of workers randomly miscode in all surveys. These will almost always look like they change union status from one period to the next • A high percentage of the fraction of the sample who appear to be moving may be just miscodes • The union premium will be biased downwards since wages will not change for the workers with miscodes • SILC data for Ireland asks workers if paid a union sub on last wage packet. Separate question asks how much it is. For over 60% of sample the payslip is observed • Arguably for this sub-sample measurement error in the union variable will be very low. You would have to incorrectly report paying a sub or not, then report an incorrect amount that is consistent with this and the interviewer would observe your payslip and fail to resolve this inconsistency Conference on Irish Economic Policy Coefficient on Union dummy from log wage regressions SILC data 2006/10 ALL IF PAYSLIP SEEN Worker characteristics: Weekly wage (Private sector only) Cross-section Fixed Effects Cross-section Fixed Effects 0.137 0.058 0.125 0.109 Union (0.000)** (0.087) (0.000)** (0.029)* 0.579 0.970 0.602 0.979 R-squared Worker and job characteristics: Weekly wage (Private sector only) 0.074 0.061 0.071 0.100 Union (0.000)** (0.076) (0.000)** (0.048)* 0.642 0.970 0.672 0.979 R-squared Worker characteristics: Weekly wage (Private and public sector) 0.144 0.039 0.136 0.079 Union (0.000)** (0.149) (0.000)** (0.039)* 0.141 0.013 0.147 0.034 Public sector (0.000)** (0.607) (0.000)** (0.293) 0.027 -0.022 0.019 -0.044 Union*public (0.228) (0.459) (0.479) (0.247) 0.574 0.964 0.583 0.971 R-squared Worker and job characteristics: Weekly wage (Private and public sector) 0.088 0.049 0.087 0.076 Union (0.000)** (0.074) (0.000)** (0.049)* 0.105 0.020 0.128 0.028 Public sector (0.000)** (0.426) (0.000)** (0.382) 0.020 -0.026 0.004 -0.036 Union*public (0.355) (0.381) (0.872) (0.349) 0.638 0.964 0.658 0.971 R-squared We cluster over the individual identifier in the cross-section regressions. Robust p values are in parentheses. * denotes significant at 5% and ** significant at 1%. All regressions also include month and year dummies, eight region dummies, five health dummies, an urban dummy, age, age square weekly hours worked, weekly hours squared, weekly hours cubed, a born in Ireland dummy, a male dummy, thirteen education dummies, a marital status dummy years of employment experience and years of employment experience squared. Additional controls for job characteristics are one digit industry and occupation dummies, five firm size dummies, job tenure and tenure squared. Conference on Irish Economic Policy Wage premium declining over time for all specifications 0.2 0.18 0.172 0.17 0.16 0.157 0.145 0.14 Private sector worker & firm characateristics 0.12 0.1 0.104 0.106 0.094 0.096 0.114 0.107 0.112 Private sector worker characateristics 0.098 0.089 0.084 0.08 Private & publicsector worker & firm characateristics Private & public sector worker characteristics 0.073 0.063 0.06 0.05 0.042 0.04 0.049 0.042 0.02 0 2006 2007 2008 2009 2010 Conference on Irish Economic Policy Discussion/Conclusion • Union density on a downward trend especially in private sector • Union wage premium at around 8-10% from 2006-10 but seems to have fallen substantially during recent recession • A labour market with co-ordinated sectoral bargaining does not seem feasible without representative employer groups • A key difference between countries with low and high rates of representation is the degree to which worker representative bodies are involved in the provision of other services (pensions/social security) Conference on Irish Economic Policy Table 1: Summary statistics Private sector Mean Weekly wage Standard Deviation Weekly wage Mean Union member ship Standard deviation Union membership Observations Number of individuals Switchers per thousand Public and Private sector Mean Weekly wage Standard Deviation Weekly wage Mean Union member ship Standard deviation Union membership Observations Number of individuals Switchers per thousand ALL WORKERS PAYSLIP SEEN PAYSLIP NOT SEEN 739 774 696 626 593 662 0.22 0.24 0.19 0.41 0.43 0.39 7,905 5,867 9.36 4,338 3,329 5.76 3,567 2,538 14.01 802 846 731 620 593 655 0.38 0.44 0.28 0.48 0.50 0.45 11,793 8,265 14.92 7,234 5,223 11.33 4,559 3,042 20.6 Conference on Irish Economic Policy