Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Risk-Aversion and Trade Union Membership Laszlo Goerke and Markus Pannenberg CESifo GmbH Poschingerstr. 5 81679 Munich Germany Phone: Fax: E-mail: Web: +49 (0) 89 9224-1410 +49 (0) 89 9224-1409 [email protected] www.cesifo.de Risk-Aversion and Trade Union Membership* Laszlo Goerke Eberhard Karls Universität Tübingen Department of Economics Melanchthonstraße 30 D – 72074 Tübingen, Germany E-Mail: Laszlo.Goerke(at)uni-tuebingen.de IZA, Bonn and CESifo, Munich Markus Pannenberg DIW Berlin SOEP Königin-Luise-Straße 5 D – 14195 Berlin, Germany E-Mail: mpannenberg(at)diw.de University of Applied Sciences Bielefeld and IZA, Bonn preliminary version: 04.5.2007 Abstract In an open-shop model of trade union membership with heterogeneity with respect to risk attitudes, a worker's relative risk-aversion can affect the decision to join a trade union and the collective bargaining outcome. Using German panel data (GSOEP) and three novel direct measures of individual risk-aversion, we find evidence for a significantly positive relationship between individual risk-aversion and the likelihood of being a union member. Moreover, our findings suggest that risk-aversion is correlated negatively with wages and, thereby, positively with employment. JEL Classification: J 51 Keywords: employment, membership, risk-aversion, trade union * We are grateful to Carsten Hefeker for helpful comments. 1. Introduction The changes in wellbeing owing to a variation in consumption possibilities are determined by the slope and curvature of an individual's utility function. Accordingly, an assessment of the gains and costs of trade union membership is likely to vary with risk attitudes. However, the nature of such a linkage is largely unexplored. Understanding the impact of risk aversion on the net gain from being a trade union member can consequently contribute to a more profound appreciation of the determinants of trade union density. In addition, the number of members can affect a trade union's bargaining power. Accordingly, being aware of the exact relationship between risk attitudes and union membership can also help to predict how bargaining power and collective negotiations are influenced by changes in the distribution of individual risk-preferences. Risk attitudes and collective bargaining can be related if trade unions attempt to insure their members against income variations by reducing wage and employment uncertainty (e.g. Agell and Lommerud 1992, Burda 1995). The strength of this desire will depend on the risk preferences of union members. However, this literature assumes a given union membership. This is the case because a reduction in income variability represents a public good and, thus, provides no incentives for an individual to join a trade union, unless there are closed-shops. To establish a relation between risk attitudes and an individual's net gain from trade union membership, a union must supply an excludable good. Therefore, in this paper we initially investigate an open-shop union model in the tradition of Booth (1985) and Booth and Chatterji (1995). We assume that workers differ in their aversion to risk and, hence, evaluate the income loss resulting from the union membership fee differently. This establishes an – ambiguous – relationship between the Arrow-Pratt measure of relative risk-aversion and the net gains from union membership. The relation allows us to analyse the effects of risk-aversion on wages and employment which are, again, theoretically indeterminate. The linkage between an individual's extent of risk-aversion and union membership status and the implications for collective bargaining outcomes are, therefore, largely empirical issues. In the respective part of the paper, we use data from the German Socio-Economic Panel (GSOEP), a nationally representative longitudinal survey, and, in particular, exploit a novel set of direct measures of individual risk attitudes. We find that a worker is more inclined to be a trade union member, the more risk-averse he is. An increase in risk-aversion of the decisive union member, ceteris paribus, reduces the preferred wage. However, the rise in membership increases the union's bargaining power and, thereby, contributes to higher wages. Our empirical findings suggest that the first effect may dominate, so that more risk-averse workers obtain lower wages, although their unionisation rate is higher. Accordingly, risk-aversion may contribute to wage moderation and have positive employment consequences. 1 In the literature on open-shop models of collective bargaining - surveyed, for example, by Schnabel (2003) - only few papers have explicitly incorporated notions of risk-aversion and predict either no or a negative impact on union membership. In contrast, we find that a positive relationship can exist. In addition, none of the empirical contributions estimating the determinants of trade union membership known to us has explicitly evaluated the effects of an individual's risk attitude. Finally, open-shop models of union membership have assumed workers to be heterogeneous with respect to the valuation of a union-provided good or a social norm. Variations in individual risk attitudes have not been allowed for. Our contribution is, therefore, twofold: First, it generates predictions on the impact of a worker's risk attitude on the propensity to unionise and the ensuing consequences for collective bargaining outcomes. Second, our paper contains an explicit test of the membership predictions. The paper unfolds as follows. Section 2 sets out the model of endogenous union membership and collective wage bargaining, and relates the findings to the literature. Section 3 contains the description of the data and our empirical specification. In Section 4 we present the empirical results, while Section 5 concludes. 2. Risk-aversion in a Model of Wage Bargaining with Endogenous Union Membership The theoretical model presented below allows determining union membership and wages. The analysis focuses on one open-shop trade union and one firm (or employer association). Wages are not differentiated according to a worker's union status. This assumption is in line with the legal situation in Germany, not allowing for closed shops. Additionally, it is consistent with empirical evidence – elaborated upon in Section 4. To provide an incentive for joining a trade union, despite the existence of a membership fee and the absence of a wage premium, the union supplies a private good solely to its members.1 The incentives to join a union vary across workers, although all of them appraise the private union good identically. This is because the utility loss due to the membership fee varies owing to distinct risk preferences. Membership Decision Let total utility be additively separable in the utility from income and the utility C from consuming the private good.2 Additionally, the utility of worker i from a wage w is assumed to be given by w(1 – σi)/(1 – σi), 0 < σi ≠ 1, implying a constant individual Arrow-Pratt measure σi of relative 1 See Olson (1965) for the notion of selective incentives. A private good may consist of legal advice, the provision of insurances or pensions at lower prices than available elsewhere, or the supply of job-related information. 2 This simplifying hypothesis suffices to illustrate the main features of the linkage between risk-aversion and trade union membership. 2 risk-aversion. The measure of relative risk-aversion is distributed on the interval [σmin + ε; σmax + ε], ε ≥ 0. Increasing ε from ε = 0 mimics a general rise in risk aversion. The effects of such a rise will be looked at later on. The membership fee is an increasing and weakly convex function G of the wage w, 0 < G(w) < w, ~ := w – 1 > G' > 0, G'' ≥ 0. Accordingly, the income of an employed union member equals w G(w). An unemployed worker receives unemployment benefits which are normalised to zero without loss of generality, derives no benefits from membership, and does not pay a membership fee. Alternatively, it can be assumed that unemployed workers lose their union status. An employed non-member also receives the wage w. The probability that a worker is employed equals N(w), assuming a random draw from the population of all workers, the size of which is normalized to unity. Hence, 0 < N(w) ≤ 1 also describes the employment level. Worker i decides about union membership, anticipating the (equilibrium) levels of union density and wages. The worker will join the union if the gain Z(σi) from doing so is positive. Since the evaluation of incomes differs across workers, there is a wage w making a particular worker just indifferent between leaving the trade union and remaining in it. This worker represents the marginal member. The marginal member's measure of relative risk-aversion is denoted by σ. The gain from membership for this member is zero and then defined by: Z(σ,...) := ~ 1− σ w w1− σ =0 +C− 1− σ 1− σ (1) The derivatives of Z(σ,…) are given by: ∂Z(σ,...) ~ − σ (1 − G ' ) − w − σ := Z w = w ∂w (2) ~ 1− σ ln w ~ − w1− σ ln w ∂Z(σ,...) w := Z σ = − ∂σ 1− σ (3) If the marginal union member is characterized by a measure of relative risk-aversion less than unity (σ < 1) the gain from membership will rise with risk-aversion, implying Zσ > 0. Accordingly, all workers characterised by σi > σ derive a positive gain from membership so that the gains rise with the measure of risk-aversion. Therefore, workers characterized by the highest measure of relative risk-aversion are the first to join the union. However, for a measure of relative risk-aversion greater than unity (σ > 1) the sign of Zσ is theoretically indeterminate. The sign of Zw is also a priori uncertain. Empirically, a worker's wage has not been found to alter the probability of union membership (see the estimates presented below and also Wagner 1990, 1991, 3 Fitzenberger et al. 1999, and Goerke and Pannenberg 2004). In addition, in our empirical analysis we obtain evidence that the marginal trade union member's measure of relative risk-aversion is less than unity. This implies Zw = 0 and Zσ > 0.3 The further theoretical analysis will make use of these findings to simplify the exposition. The impact of risk-aversion on the incentive to become a union member has previously been investigated in a number of theoretical contributions. In Booth's (1984) seminal paper the membership decision is independent of the measure of risk-aversion, for a given wage. This is because the marginal trade union member is defined in such a manner that the utility resulting from the wage less membership fee equals the utility from unemployment. This situation will be attained if the arguments of the utility function coincide. Hence, changes in the slope or curvature of the utility function do not alter the identity of the marginal member (Booth 1984, eq. (13)). In Moreton's (1998, 1999) set-up, a variation in risk-aversion is modelled as a change in the second derivative of the utility function, holding constant the first derivative. Since the membership curve only depends on utility levels, it is unaffected by a change in risk-aversion. Finally, Oswald (1982) ~ >w presumes that the income of a union member is higher than that of a non-member, i.e. that w holds. In this case, the benefits from union membership decline with the measure of relative riskaversion. According to our knowledge the present contribution is, therefore, the first allowing for a positive impact of individual risk-aversion on the incentives to join a trade union. Wage Bargaining The outcome of the wage bargain between the firm and the union is determined by the (symmetric) Nash-solution. The union's preferences are given by those of the median member, whose ArrowPratt measure of relative risk-aversion is denoted by µ. The firm uses labour as only input, while the output price is normalized to unity. The firm's payoff is given by its profits Π. Profit maximization leads to ∂Π/∂N = 0, ∂Π/∂w = -N(w) by the envelope theorem, and ∂N/∂w := N'(w) < 0. The fallback payoff of the firm is normalised to unity, while the median member's utility in the case of no agreement is determined by unemployment benefits, set equal to zero. The wage, ~ 1− µ /(1 − µ) + C > 0 , is consequently determined by: making use of Ω := w V := ~ − µ (1 − G ' ) N( w ) w − K(w ) = 0 , Ω where K ( w ) := ( N( w )) 2 − N' ( w ) > 0 Π Moreover, we have: 3 Using Zw = 0 in equation (3) does not allow to derive further restrictions for the sign of Zσ. 4 (4) Vw = 2 ~ − µ ~ −1 ~ −µ ~ −µ w (N' (w )(1 − G ' ) − N(w )G ' ') − N(w ) (1 − G ' ) w (µw Ω + w ) − K w Ω Ω2 (5) Subsequently, an interior solution to the wage bargaining problem is assumed (Vw < 0), which yields a wage above the full employment level, implying N(w) < 1. For Vw < 0, the wage effect of an increase in the measure of relative risk-aversion µ of the median member is determined by: Vµ = − ~ −µ ⎡ ~ 1− µ ⎤ N( w )(1 − G ' ) w ~− w ⎥ ⎢C ln w ⎢⎣ (1 − µ) 2 ⎥⎦ Ω2 (6) The sign of Vµ is determined by two effects: A larger measure of relative risk-aversion lowers the gain from a higher wage, while its costs in terms of the fall in employment remain unaffected. Moreover, a rise in µ generally alters the gain from bargaining for the median member and will unambiguously raise the incentives to increase wages for σ < 1. The overall impact is, thus, uncertain. A negative wage and positive employment impact of higher risk-aversion in (closedshop) collective bargaining models have been derived, for example, by Sampson (1983), McDonald (1991), Oswald (1982), and Blair and Crawford (1984). The employment result extends to a setting in which the alternative income is endogenised (see Nickell 1990). The findings for the open-shop model suggest that these predictions are strongly influenced by the assumption of a given number of union members. Comparative Statics The present analysis aims at determining the membership and the ensuing wage effects of a change in risk-aversion in an open-shop setting. To do so, first, the impact of a rise in the measure of relative risk-aversion of all workers on the incentives to join a trade union is looked at. Subsequently, the consequences for wages are investigated. Since a change in the wage does not alter the probability of membership (that is, since Zw = 0), the membership curve is vertical in the wage - risk-aversion space. This implies that the marginal trade union member - and, hence, union density - is uniquely defined by Z(σ,…) = 0. In particular, if the interval [σmin + ε; σmax + ε], ε ≥ 0, from which the measure of risk-aversion can stem, shifts to the right because ε rises from zero to ε > 0, as depicted in Figure 1, the new marginal member will be characterised by the same value of relative risk-aversion than the original one (∂σ/∂ε = 0). However, the level of risk-aversion of the marginal member will be closer to the lower bound of the interval. For Zσ > 0, this effect is equivalent to an increase in the number of union members. If everyone becomes more risk-averse, the worker who just refrained from being a union 5 member, prior to the general rise in risk-aversion, will experience a rise in utility from joining (for Zσ > 0) and become a unionist. The result is independent of the wage and can be summarised in: Proposition 1: For Zw = 0, a general rise in the Arrow-Pratt measure of relative risk-aversion raises trade union density for Zσ > 0. Figure 1: Equilibrium Wage and Marginal Member in Wage - Risk-aversion Space w Z V0 V1 σ min σ min + ε σ σ max σ max + ε σi Given Vw < 0 and an indeterminate sign of Vµ, that is the risk preferences of the median member have an ambiguous wage effect, the slope of the wage-bargaining curve in the wage – riskaversion space cannot be ascertained. For Zw = 0 and ∂σ/∂ε = 0, the wage effect of a shift in the distribution of the measure of risk-aversion is determined by: dw =− dε Vµ ∂µ ∂σ + Vσ ∂ε = − Vµ ∂µ ∂ε Vw Vw ∂ε (7) The measure µ of risk-aversion of the median member is likely to increase with a general shift of the distribution to the right (implying ∂µ/∂ε > 0). However, theoretically a fall of µ cannot be ruled 6 out either (that is, ∂µ/∂ε < 0). The intuition for the ambiguity is the following: The original median member, prior to the general increase in risk-aversion, will obviously be characterised by a greater level of risk-aversion subsequent to the shift of the distribution. Since more workers become union members, the identity of the median will change. If a relatively large mass of the distribution lay to the left of the original marginal member, but were to the right of σ subsequent to the shift of the distribution, the median member's measure of risk-aversion may decline. This possibility can be ruled out if the distribution of the measure of risk-aversion is fairly symmetric around the positions of the original and new marginal member. Assuming, therefore, a uniform distribution the measure of relative risk-aversion of the median member is given by µ = [σ + (σmax + ε)]/2. Accordingly, a general increase in risk-aversion also raises the median member's risk-aversion µ (∂µ/∂ε = 0.5). In addition, Vµ = 2Vσ = 2Vε holds. From equation (7) the wage effect is then given by dw/dε = Vµ/(2Vw) and can be summarised in: Proposition 2: Assume Vµ < 0, that is, a negative relationship between the measure of risk-aversion of the median member and the bargained wage. If, in addition, Zw = 0 holds and the Arrow-Pratt measure of relative risk-aversion is distributed uniformly, a general rise in the measure of relative risk-aversion will lower the bargained wage. To illustrate Proposition 2, suppose that the bargained wage falls with the measure of risk-aversion of the median member (implying Vµ, Vε < 0). A general rise of risk-aversion has two effects: For a given density level, the higher measure of risk-aversion induces the median member to accept a lower wage in the bargaining process. However, a higher union density will result in workers joining the trade union which are characterised by a lower measure of risk-aversion, relative to the maximum. This effect, ceteris paribus, lowers the median member's relative risk-aversion. Given a uniform distribution of risk-aversion, the general rise in risk-aversion dominates the impact of a rise in union density. In this case, the wage-bargaining curve is downward-sloping in the wage risk-aversion space, as depicted in Figure 1. Moreover, a general rise in risk-aversion shifts the wage-bargaining curve downward from V0 to V1. The employment consequences of an increase in risk-aversion are directly determined by the wage effects and summarised in: Corollary 3: For Zw = 0, Vµ < 0, and a uniformly distributed Arrow-Pratt measure of relative risk-aversion, a general rise in the measure of relative risk-aversion raises employment. 7 The net consequences of a change in risk-aversion of all workers on the bargained wage and employment are, therefore, determined by the impact on the median worker's preferred wage, i.e. the sign of Vµ, and the effect of a change in the marginal member on the median's preferences. Since the first effect is theoretically indeterminate and because the second hinges on the distribution of the measure of risk-aversion, the relation between an encompassing rise in riskaversion and the bargained wage is ultimately an empirical issue. 3. Data and Empirical Specifications Data In our theoretical model wages result from negotiations between a trade union and employer. To capture this setting, data on firm characteristics and the bargaining relationship would be desirable. Moreover, information on individual risk preferences of workers is required. Ideally, we would, hence, use matched employer-employee data with detailed information on firms and workers. However, the available linked employer-employee data sets for Germany provide no information on individual risk attitudes, pivotal for putting our theoretical model to an empirical test. Therefore, we utilise data from the German Socio-Economic Panel (GSOEP), a nationally representative longitudinal survey of the resident German population (Wagner et al. 1993, SOEP Group 2001) which provides three novel direct measures of individual risk attitudes. We use data for the years 2003 and 2004, because the 2003 survey includes information on union membership, while the survey in 2004 contains the risk indicators. Our sample consists of full-time workers with valid information on the relevant measures of risk attitudes and union membership from West and East Germany.4 Observations from self-employed persons, apprentices and civil servants ('Beamte') are excluded. Moreover, in the regression analyses all respondents with missing information on relevant variables are dropped. In our theoretical framework the individual decision to join the union depends on wages, the utility of the private good provided by the trade union and the membership fee. Hence, information on how individual risk attitudes affect the evaluation of monetary payments is required. To meet this requirement we, first, utilize a survey question requiring respondents to indicate their willingness to take risks in financial matters. This willingness is recorded on an eleven-point scale. A value of zero (ten) indicates a total unwillingness (readiness) to take risks. We, second, employ a survey question corresponding more closely to a standard lottery measure in experiments. In particular, information is collected on the individual's investment choices, based on a hypothetical lottery 4 Schnabel and Wagner (2003) conclude that the same set of variables can be used to analyse the determinants of union membership in West and East Germany. To control for regional differences in the level of union density we include a full set of state dummies in our regression analysis. 8 prize of 100.000 €. The questionnaire allows for 6 opportunities to invest this prize. More precisely, the household can purchase a risky asset for 100.000 €, 80.000 €, 60.000 €, 40.000 €, or 20.000 € or can, finally, refrain from such an acquisition, with equal chances of doubling the amount invested or losing half of it. Since the second measure is based on explicit stakes and probabilities, it holds risk perceptions constant across individuals. Moreover, within the expected utility framework, the hypothetical lottery allows us to calculate the Arrow-Pratt measure of relative risk-aversion for each respondent (see for example Caliendo et al. 2006 or Guiso and Paiella 2005).5 As a check of robustness, we additionally employ a survey question on the willingness to take risks in general. The GSOEP risk measures have been validated in a field experiment with representative subject pools. Dohmen et al. (2005) find that questionnaire responses to the general risk question are reliable predictors of actual risk-taking behaviour in the field experiment. Moreover, answers to the general risk question are strongly correlated with those to the other two questions. Finally, Dohmen et al. (2005) demonstrate that the best predictor of a specific outcome is the risk measure most closely related to the relevant context. Therefore, we are confident that the GSOEP risk measures we use are high-quality proxies for actual risk preferences in our specific context. Due to data availability we link the information on individual risk attitudes collected in 2004 with data on union membership and its determinants measured in 2003. Hence, the crucial assumption of our empirical work is that individual union membership in 2003 does not alter peoples' risk attitudes until they are collected in 2004. This requirement is consistent with evidence provided by Andersen et al. (2005) and Barsky et al. (1997) that risk preferences elicited from hypothetical lotteries in surveys are stable over time. Empirical Specifications Since we are interested in assessing the effect of individual risk attitudes on trade union membership we start with the following specification of a standard probit model: P (U i ,03 = 1| Ri ,04 , X i ,03 ) = Φ (α Ri ,04 + β ' X i ,03 ) , (8) where U i ,03 =1 if the individual is a union member (in 2003), Ri ,04 is the relevant measure of individual risk attitudes (in 2004), X i ,03 is a vector of control variables (also measured in 2003), α , β ' are (vectors) of unknown parameters and Φ () is the cdf of the standard normal distribution. Estimated marginal effects and standard errors robust with respect to clustering at the household level are documented. 5 Given the construction of the lottery, the Arrow-Pratt measure of relative risk-aversion is bounded by 0.4 (4) from below (above). See also Table A1 in the Appendix. 9 To address the potential endogeneity of our plain risk measures, we also use an instrumental variable probit estimator (Wooldridge 2002, 472-477). Our according specification builds on a structural form latent variable specification of latent union membership ( U i*,03 ), where U i ,03 = 1 [ U i*,03 > 0 ] and on a reduced-form specification of the particular risk measure Ri ,04 : U i*,03 = α Ri ,04 + β ' X i ,03 + ε1,i Ri ,04 = ∂Heighti ,04 + ∂ ' X i ,03 + ε 2,i (9) (10) We assume that the errors ( ε1,i , ε 2,i ) follow a zero mean, bivariate normal distribution and are independent of the instrument Heighti ,04 . Individual height is used as an instrument, since height (i) is plausibly exogenous to the indicators of individual risk attitudes and has a significant impact on individual risk preferences, (ii) is not correlated with the error term in the union membership equation and (iii) has no direct impact on the likelihood of being a union member. The parameters are estimated using maximum likelihood as the estimation criterion. Standard errors are robust with respect to clustering at the household level. Standard tests for the validity of the instruments and for endogeneity of our risk measures as well as marginal effects of the estimated parameters and their standard errors are documented. Both probit estimators require a distributional assumption to be made, i.e. a particular normality assumption. However, there is no prior knowledge on the validity of this assumption. As a check of robustness, we therefore employ a semi-nonparametric estimator originally suggested by Gallant and Nychka (1987) and used in applied work, for example, by Gabler et al. (1993), Gerfin (1996) and Stewart (2005). Essentially, this estimator approximates the true distribution of the error terms by a Hermite series. The approximation can be expressed as the product of the normal density and a squared polynomial and thus nests the standard probit model of equation (8). It can be estimated by maximizing a pseudo-likelihood function and usual test procedures can be applied. We adopt the framework proposed by Stewart (2004) to test the validity of the distributional assumption of the probit model in our application and to estimate the parameters of interest.6 The vector of control variables consists of the usual covariates which have been found to affect the probability of union membership in Germany in previous studies: Age, age squared, tenure, tenure squared and dummy variables for being a foreigner, being male, different firm size categories, having completed an apprenticeship, having a university degree, being member of a works council, having preferences for the social democratic party (SPD) or the Christian-democratic parties (CDU/CSU), the father being self-employed when the respondent was 15 years old, being a blue 6 See Stewart (2004) for a detailed description of the econometric specification and his corresponding stata-tool. 10 collar worker, the industry (NACE 1-digit) in which the respondent works and the state of residence ('Bundesland').7 Since we link data from the years 2003 and 2004 in our sample, longitudinal sample weights are calculated and used to take into account the sampling design of the different subsamples of the GSOEP as well as panel attrition (cf. Pannenberg et al. 2004). 4. Results Descriptive Evidence Figure 2 shows the distribution of individual risk attitudes in financial matters by union membership status. Each bar in the histograms indicates the percentage of respondents choosing a number on the eleven-point scale, indicating their willingness to take risks in financial matters. The degree of reluctance to take risks in financial matters is striking: 86% of all non-members and 91% of all union members choose a value of 5 or less on the eleven-point scale. Moreover, Figure 2 reveals that union members are more risk-averse than non-unionists. This is confirmed by means of a Kolmogoroff/Smirnov test, which rejects the null hypothesis of equality of the distributions of risk attitudes for union members and non-members at the α = 0.002 level. Figure 3 corroborates this impression with respect to the lottery measure of individual risk attitudes: 64% of all union members and 54% of all non-members in Germany do not invest a positive amount of the hypothetical lottery prize in the risky asset. In this case a Kolmogoroff/Smirnov test rejects the null hypothesis of equality of the distributions of the lottery measure for union members and nonmembers at the α = 0.061 level. With respect to general risk attitudes, Figure A1 (in the Appendix) documents that roughly 63% of all workers choose a value of 5 or less on the eleven-point scale. However, for this measure we do not observe significant differences in the distributions of risk attitudes for members and non-unionists. Considering the Arrow-Pratt measure of relative risk-aversion the findings confirm our previous results that non-members are less risk-averse than union members, given that the mean of the Arrow-Pratt measure is higher for union members (see Table A1 in the Appendix). The Kolmogoroff/Smirnov test again rejects the equality of the distributions (α = 0.061). This is consistent with the assumption in the theoretical section of the paper that those workers characterized by the highest measure of relative risk-aversion are the first to join the union. Moreover, the figures indicate that the minimum of the coefficient of relative risk-aversion is less than unity, as it has been assumed in Section 2. 7 Descriptive statistics for all control variables are presented in Table A3 in the Appendix. 11 Regression Results The descriptive evidence points to remarkable degrees of risk-aversion in our sample of full-time workers. Furthermore, risk-aversion seems more prevalent among union members. In a first step, we therefore use the standard weighted probit specification to assess whether individual risk attitudes have an impact on union membership, conditional on a set of control variables. For the sake of a more intuitive interpretation we recoded the eleven-point scale of the two GSOEP measures of risk attitudes for our regression analysis in reverse order, i.e. '0' indicating strong risklove and '10' total reluctance to take risks. Consequently, for all measures of risk a higher value indicates greater risk-aversion.8 As a preliminary exercise, we estimated regressions also including the log of the monthly gross wage as covariate. Irrespective of the measure of individual risk-aversion employed, we could replicate previous findings by Wagner (1990, 1991), Fitzenberger et al. (1999), and Goerke and Pannenberg (2004) that the individual gross wage does not influence the probability of membership.9 Accordingly, the wage is not included into our main specifications. In terms of our theoretical model, the assumption Zw = 0 captures the insignificant wage effect. Table 1 presents the estimated marginal effects and their standard errors for the four measures of individual risk attitudes using the probit specification. Since our focus is on the relationship between risk-aversion and union membership we do not discuss the parameter estimates for the set of control variables which are in line with results from other studies.10 Column 1 reveals that fulltime workers who are more risk-avers in financial affairs exhibit a significantly higher probability of being a union member than their less risk-averse colleagues. The marginal effects imply that someone who switches from extremely risk-loving ('0') to extremely risk-averse ('10') in financial matters exhibits a roughly 7 percentage points higher likelihood of being a union member than before. Given that the unconditional mean of being a union member is around 20 % in our sample, the size of the estimated marginal effect is remarkable. Column 2 presents the estimates for the lottery measure of individual risk-aversion. Again, we observe a similar significant increase in the probability of being a union member if someone decides not to invest a positive fraction of the hypothetical prize in the risky asset. Column 3 contains the findings when using the Arrow-Pratt measure of relative risk aversion. These estimates are in line with the ones for the lottery measure 8 With respect to the lottery measure, we do not have to recode our variable, since “1” indicates an investment of 100.000 € and “6” indicates an investment of 0 €. 9 Table A2 in the Appendix documents the parameter estimates for a specification with the measure of individual risk attitudes in financial matters. 10 See, for example, Fitzenberger et al. (1999), Fitzenberger et al. (2006), Goerke and Pannenberg (2004), and Schnabel and Wagner (2003). Only the significantly positive effect of acting as a works councillor represents a new result. See Goerke and Pannenberg (2007) for evidence on the relationship between union membership and works councillorship. 12 itself. If we use the encompassing measure of individual risk attitudes - instead of the two measures related to financial affairs - we do not find a significant effect (column 4). This is in line with evidence presented by Dohmen et al. (2005) that a context specific measure of individual risk attitudes is the best predictor of actual individual behaviour in that context. Summing up, we observe a significantly positive association between individual risk-aversion in financial affairs and the likelihood of being a union member. Hence, in terms of the theoretical discussion of Section 2 an overall increase in risk-aversion leads to an increase in union density, a finding which is consistent with Proposition 1. The previous empirical specifications have assumed that the measures of individual risk attitudes are not correlated with the errors in the union membership equation. However, one might argue that union membership is endogenous with respect to risk attitudes. We assess the potential endogeneity of individual risk attitudes in the membership equation by means of the IV-probitestimator, using the instrument “individual height” as described above. To test the validity of the instrument “height”, we start with a simple Wald-test of the estimated parameter of height in a reduced-form equation in which the individual risk measures are regressed on height and the set of control variables of the union membership equation. It turns out that height is a valid instrument in terms of significant parameter estimates for the two risk measures based on the willingness to take risks in financial matters and in general, but not for the two measures of individual risk preferences based on the hypothetical lottery (see Table 2, Wald_IV). Moreover, the Wald-tests of exogeneity indicate that we cannot reject the hypothesis that the relevant measures of risk-aversion are exogenous (see Wald_Ex).11 Hence, taking these test statistics at face value, there is no need for the IV-specification. However, the estimated parameter of individual risk-aversion in financial matters on the probability of being a union member is significantly positive and, once again, points to a positive association of risk-aversion and trade union membership. Moreover, the estimated parameter of the overall risk-aversion measure is significantly positive and the marginal effect indicates that a worker revealing the average degree of risk-aversion of 5.2 in the sample ceteris paribus has a roughly 50 percentage points higher probability of being a union member than an extremely risk-loving worker. This again suggests a positive relationship between individual risk-aversion and trade union membership and puts into perspective the insignificant effect for the encompassing measure of risk attitudes in our base regression (cf. Table 1, column 4). 11 This result is confirmed by the test-statistic of the Rivers/Vuong test of exogeneity. See Wooldridge (2002, pp. 472) for a description of both tests. 13 As pointed out above, a priori there is no evidence that the normality assumption underlying the probit models is appropriate. Performing likelihood-ratio tests comparing the log-likelihoods of the probit and the semi-nonparametric specification of the union membership equation described above indeed indicate that the normality assumption of the probit model might not be upheld. Therefore, in Table 3 we additionally present the estimated parameters for the different measures of individual risk attitudes based on the semi-nonparametric specification. The estimated parameters of the two direct measures of individual risk preferences as well the one of the individual measure of relative risk aversion are significantly positive. Moreover, they are similar in size to the estimated parameters of the standard probit specification (not documented). Hence, relaxing the distribution assumptions of our econometric specifications again leads to parameter estimates which confirm the finding of a positive relationship between individual risk-aversion and trade union membership. Summing up, there is substantial evidence that the probability of trade union membership of German full-time workers rises with their aversion to risk. Further Correlation Analysis The theoretical model shows that the degree of risk-aversion of the median trade union member has an impact on the bargained wage and therefore on the level of employment. Our estimates indicate that an overall increase in the degree of risk-aversion raises the individual probability of union membership and, ceteris paribus, leads to a rise in union density. Our preliminary estimations also suggest that Zw = 0 applies (see Table A2 in the Appendix). As a consequence, the bargained wage will decrease and the employment level increase if Vε, Vµ < 0 hold, i.e. if there is a negative relationship between the bargained wage and the measure of risk-aversion of the median member (see Proposition 2 and Corollary 3). To get a crude grasp of the sign of Vµ, we specify standard Mincer-earnings regressions including our measures of individual risk attitudes, an estimate of the industry-specific union density based on the GSOEP data, individual union membership as well as the set of covariates used in the regression analysis described above. Table 4 documents the parameter estimates for the measures of individual risk attitudes, union density and individual union status. We observe significantly negative correlations between (log) gross monthly wages and both measures of the individual willingness to take risks, while the parameter estimates of the two risk measures based on the hypothetical lottery are not significantly different from zero. Our finding of a negative correlation between the individual willingness to take risks and individual wages (see Bonin et al. 2006 for a similar result) is consistent with the assumption of Vε, Vµ < 0. Moreover, 14 the parameter estimates reveal a significantly positive correlation between industry-specific union densities and wages, while the individual union membership status has no impact on wages. Our simple correlation exercise provides supportive evidence for the two opposite effects of an increase in individual risk-aversion in the theoretical model: On the one hand, greater risk-aversion leads to a rise in union density, which tends to increase bargained wages. On the other hand, higher risk-aversion goes along with lower wages. This finding is consistent with the interpretation that an increase in individual risk-aversion leads to a change in the preferences of the median member, which unions take into account when bargaining over wages and which ceteris paribus has a negative impact on wages. 5. Conclusions Why do workers belong to a trade union? Our theoretical analysis is based on the assumption that a union provides its members with an excludable good. This constitutes the gain from union membership. We, furthermore, assume that the costs of membership, namely the utility losses resulting from the membership fee, vary with the extent of risk-aversion. Our empirical findings are consistent with these presumptions. Using the German Socio-Economic Panel (GSOEP), a nationally representative survey, we find that the probability of union membership increases significantly and by a sizeable amount with indicators of individual aversion to risk. In a collective bargaining set-up, a change in risk-aversion will then have two effects. First, the preferred wage (and employment) outcome of the median member will change with his risk attitude. In addition, the change in the level of union membership will alter the identity of the median member. Accordingly, variations in risk-aversion have wage and employment effects. Our findings suggest that wages may fall and employment may rise with an increase in risk-aversion of the labour force. The impact of individual, direct measures of risk-aversion on the individual decision to become a member of a trade union has not been analysed before. Since our information relates to Germany, it will be interesting to see whether this positive correlation between risk-aversion and unionisation is a robust one also existing in other industrial relations systems. The wider implications of a positive relationship, for example, with respect to the effectiveness of union organising campaigns, the preferred degree of centralisation in bargaining, the interaction of product and labour market imperfections, or the role of trade unions as insurance devices in openshop settings also represent topics for future investigations. 15 Figure 2: 0: non-member; 1: member of a trade union Figure 3: 0: non-member; 1: member of a trade union 16 Table 1: Union Membership and Individual Risk Attitudes in Germany - Probit Estimates Risk_finance Risk_lottery Arrow-Pratt – RRA Risk_general Blue_collar Father_self-employed Prefers SPD Prefers CDU/CSU Member of works council Tenure Tenure (sqrd) Apprenticeship University degree Firm size: 20 ≤ X < 200 workers Firm size: 200 ≤ X < 2000 workers Firm size: X > 2000 workers Male Foreigner Age Age (sqrd) ME / s.e. ME / s.e. ME / s.e. ME / s.e. 0.007* ---(0.003) ----0.013* ---(0.006) ----0.013** ---(0.005) -----0.003 ---(0.003) 0.124** 0.125** 0.124** 0.129** (0.020) (0.019) (0.019) (0.020) -0.051* -0.057* -0.057* -0.054* (0.021) (0.020) (0.020) (0.020) 0.105** 0.103** 0.103** 0.101** (0.022) (0.021) (0.021) (0.021) -0.046** -0.047** -0.046** -0.050** (0.016) (0.016) (0.016) (0.016) 0.364** 0.360** 0.361** 0.359** (0.044) (0.043) (0.043) (0.043) 0.010** 0.010** 0.010** 0.010** (0.002) (0.002) (0.002) (0.002) -0.000* -0.000* -0.000* -0.000* (0.000) (0.000) (0.000) (0.000) 0.017 0.017 0.018 0.016 (0.026) (0.025) (0.025) (0.025) -0.042 -0.041 -0.039 -0.044 (0.027) (0.027) (0.027) (0.027) 0.095** 0.094** 0.094** 0.094** (0.027) (0.027) (0.027) (0.026) 0.189** 0.185** 0.185** 0.186** (0.033) (0.033) (0.033) (0.033) 0.256** 0.253** 0.252** 0.251** (0.035) (0.035) (0.035) (0.035) + 0.029 0.027 0.020 0.028 (0.016) (0.015) (0.015) (0.016) 0.016 0.016 0.016 0.020 (0.028) (0.028) (0.028) (0.028) + + + + 0.011 0.011 0.010 0.011 (0.006) (0.006) (0.006) (0.006) + + + -0.000 -0.000 -0.000 -0.000 (0.000) (0.000) (0.000) (0.000) Industry dummies & yes yes yes yes state dummies Wald_X 547.2** 547.0** 547.3** 551.5** N 5369 5370 5370 5372 Source: SOEP 2003-2004. Longitudinal weights are used. ME/s.e.: marginal effect/standard error. Robust standard errors also allow for clustering at the household level. Wald_X: Wald-Test with H0: no joint significance of all regressors. (df=45) Significance levels: ** (0.01), * (0.05), + (0.10). 17 Table 2: Union Membership and Individual Risk Attitudes in Germany - IV-Probit-Estimates Risk_finance α̂ (s.e.) ME(αˆ ) Wald_Ex Wald_IV Wald_X N 0.336* (0.151) 0.099 1.76 5.47* 1083.5** 5350 Risk_general 0 .328* (0 .148) 0 .101 2.06 3.95* 1229.3** 5353 Source: SOEP 2003-2004. Longitudinal weights are used. α̂ : Parameter estimate of particular risk measure. s.e.: standard error. ME( αˆ ) : marginal effect α̂ . Robust standard errors also allow for clustering at the household level. Significance levels: ** (0.01), * (0.05), + (0.10). Set of covariates is identical to specifications in Table 1. Wald_Ex: Wald-Test with H0: no significant correlation between the error terms in the structural and the reduced-form equation of the endogenous variable “riskaversion”. (df=1) Wald_IV: Wald-Test with H0: no significant effect of the instrumental variable “height” in a regression of the reduced-form equation for the endogenous variable “risk- aversion“. (df=1) Wald_X: Wald-Test with H0: no joint significance of all regressors. (df=45) Table 3: Union Membership and Individual Risk Attitudes in Germany - Semi-Nonparametric Estimates (SNP) Risk_finance α̂ (s.e.) LR_D Wald_X N 0.035* 0.017 17.1** 877.4** 5369 Risk_lottery 0.067* 0.033 15.5** 953.0** 5370 Arrow-Pratt –RRA 0.066* 0.028 16.2** 877.1** 5370 Risk_general -0.008 0.015 14.4** 900.6** 5372 Source: SOEP 2003-2004. Longitudinal weights are used. α̂ : Parameter estimate of particular risk measure. s.e.: standard error. Robust standard errors. Significance levels: ** (0.01), * (0.05), + (0.10). Hermite polynomial is of order 3, i.e. three additional parameters are estimated. Set of covariates is identical to specifications in Table 1. LR_D: Likelihood-Ratio test of Probit-Model against SNP extended model. (df=1) Wald_X: Wald-Test with H0: no joint significance of all regressors. (df=45) 18 Table 4: Gross Wages and Individual Risk Attitudes in Germany Risk_finance Risk_lottery Arrow-Pratt-RRA Risk_general Union density Union membership N R2 ME / s.e. ME / s.e. ME / s.e. ME / s.e. -0.014** ---(0.004) ----0.001 ---(0.007) ----0.002 ---(0.005) -----0.015** ---(0.004) 0.010** 0.010** 0.010** 0.010** (0.003) (0.003) (0.003) (0.003) -0.015 -0.018 -0.018 -0.020 (0.018) (0.018) (0.018) (0.018) 4881 4881 4881 4883 0.51 0.51 0.51 0.51 Source: SOEP 2003-2004. Longitudinal weights are used. OLS-Regression of log gross monthly wages on a measure of individual risk aversion, union density, individual union membership and the set of covariates used in the specifications in Table 1. Robust standard errors allow for clustering at the household level. Significance levels: ** (0.01), * (0.05), + (0.10). 19 References Agell, Jonas and Kjell Erik Lommerud (1992), Union Egalitarianism as Income Insurance, Economica 59 (No. 235), 295-310. Andersen, Steffen et al. (2005), Lost in State Space: Are Preferences Stable?, CEBR DP 2005-13. Barsky, Robert B. et al. (1997), Preference Parameters and Behavioral Heterogeneity: An Experimental Approach in the Health and Retirement Study, The Quarterly Journal of Economics 112, 537-579. Blair, Douglas H. and David L. Crawford (1984), Labor Union Objectives and Collective Bargaining, The Quarterly Journal of Economics 99, 547-566. Bonin, Holger et al. (2006), Cross-Sectional Earnings Risk and Occupational Sorting: The Role of Risk Attitudes, IZA DP 1930. Booth, Alison L. (1984), A Public Choice Model of Trade Union Behaviour and Membership, The Economic Journal 94, 883-898. Booth, Alison L. (1985), The Free Rider Problem and a Social Custom Model of Trade Union Membership, The Quarterly Journal of Economics 100, 253-261. Booth, Alison L. and Monojit Chatterji (1995), Union Membership and Wage Bargaining When Membership is not Compulsory, The Economic Journal 105, 345-360. Bryson, Alex, Cappellari, Lorenzo and Claudio Lucifora (2005), Why so Unhappy? The Effects of Unionisation on Job Satisfaction, IZA DP 1498. Burda, Michael (1995), Unions and Wage Insurance, CEPR DP No. 1232. Caliendo, Marco, Fossen, Frank M. and Alexander S. Kritikos (2006), Risk Attitudes of Nascent Entrepreneurs: New Evidence from an Experimentally-Validated Survey, IZA DP 2168. Dohmen, Thomas et al. (2005), Individual Risk Attitudes: Evidence from a Large, Representative, Experimentally-Validated Survey, IZA DP 1730. Fitzenberger, Bernd, Ernst, Michaela and Isabell Haggeney (1999), Wer ist noch Mitglied in Gewerkschaften?, Zeitschrift für Wirtschafts- und Sozialwissenschaften 119, 223-263. Fitzenberger, Bernd, Kohn, Karsten and Qingwei Wang (2006), The Erosion of Union Membership in Germany: Determinants, Densities, and Decompositions, IZA DP 2193. Gabler, Siegfrid, Laisney, Francois and Michael Lechner (1993), Seminonparametric Estimation of Binary Choice-Models with an Application to Labor-Force Participation, Journal of Business and Economic Statistics, 11, 61-80. Gerfin, Michael (1996), Parametric and Semi-Parametric Estimation of the Binary Response Model of Labour Market Participation, Journal of Applied Econometrics 11, 321-339. Goerke, Laszlo and Markus Pannenberg (2004), Norm-Based Trade Union Membership: Evidence for Germany, German Economic Review 5, 481-504. Goerke, Laszlo and Markus Pannenberg (2007), Trade Union Membership and Works Councils in West Germany, Industrielle Beziehungen, forthcoming. Guiso, Luigi and Monica Paiella (2005), The Role of Risk Aversion in Predicting Individual Behaviour, Banca D’Italia DP 546. McDonald, Ian M. (1991), Insiders and Trade Union Wage Bargaining, The Manchester School of Economics and Social Studies 59, 395-407. Moreton, David R. (1998), An Open Trade Union Model of Wages, Effort and Membership, European Journal of Political Economy 14, 511-527. Moreton, David R. (1999), A Model of Labour Productivity and Union Density in British Private Sector Unionised Establishments, Oxford Economic Papers 51, 322-344. Nickell, Stephen (1990), Unemployment: A Survey, The Economic Journal 100, 391-439. Olson, Mancur (1965), The Logic of Collective Action, Harvard University Press: Cambridge, MA. 20 Oswald, Andrew J. (1982), The Microeconomic Theory of Trade Unions, The Economic Journal 92, 576-595. Pannenberg, Markus et al. (2004), Sampling and Weighting, in: Haisken-DeNew, John and Bernd Frick (eds), Desktop Companion to the GSOEP, Berlin. Sampson, Anthony A. (1983), Employment Policy with a Rational Trade Union, The Economic Journal 93, 297-311. Schnabel, Claus (2003), Determinants of Trade Union Membership, 13-43, in: Addison, John T. and Claus Schnabel (eds), International Handbook of Trade Unions, Edward Elgar: Cheltenham, UK and Northampton, MA, USA. Schnabel, Claus and Joachim Wagner (2003), Trade Union Membership in Eastern and Western Germany: Convergence or Divergence, Applied Economics Quarterly 49, 213-232. SOEP Group (2001). The German Socio-Economic Panel After More Than 15 Years – Overview, Vierteljahrshefte zur Wirtschaftsforschung 70, 7-14. Stewart, Mark B. (2004), Semi-nonparametric Estimation of Extended Ordered Probit Models, Stata Journal 4, 27-39. Stewart, Mark B. (2005), A Comparison of Semiparametric Estimators for the Ordered Response Model, Computational Statistics and Data Analysis 49, 555-573. Wagner, Gerd, Burkhauser, Richard and Friederike Behringer (1993). The English Language Public Use File of the German Socio-Economic Panel, The Journal of Human Resources 28, 429-433. Wagner, Joachim (1990), Gewerkschaftsmitgliedschaft und Arbeitseinkommen in der Bundesrepublik Deutschland, DP No. 155, University of Hannover. Wagner, Joachim (1991), Gewerkschaftsmitgliedschaft und Arbeitseinkommen in der Bundesrepublik Deutschland, Ifo-Studien 37, 109-140. Wooldridge, Jeffrey M. (2002), Econometric Analysis of Cross Section and Panel Data, MIT Press. 21 Appendix Figure A1: 0: non-member; 1: member of a trade union 22 Table A1: Arrow-Pratt Measure of Relative Risk-Aversion for Hypothetical Lottery Union Member Non-Member Minimum 0.40 0.40 Mean 3.04 2.81 Maximum 4 4 Source: SOEP 2003-2004. Longitudinal weights are used. Table A2: Union Membership and Individual Risk Attitudes in Germany - Probit Estimates ME / s.e. 0.007* (0.003) Log(gross wage) -0.028 (0.025) Male 0.043* (0.017) Foreigner 0.002 (0.028) Age 0.011 (0.006) Age (sqrd) -0.000 (0.000) Tenure 0.012** (0.003) Tenure (sqrd) -0.000* (0.000) Apprenticeship -0.008 (0.027) University degree -0.050 (0.028) Prefers SPD 0.106** (0.022) Prefers CDU/CSU -0.048** (0.017) Blue_collar 0.110** (0.022) Father_self-employed -0.030 (0.022) Firm size: 0.099** (0.028) 20 ≤ X < 200 workers Firm size: 0.198** (0.035) 200 ≤ X < 2000 workers Firm size: 0.255** X > 2000 workers (0.038) Member of works council 0.355** (0.047) Industry dummies & state dummies yes N 4881 Wald_X 500.2** Source: SOEP 2003-2004. Longitudinal weights are used. ME/s.e.: marginal effect/standard error. Robust standard errors also allow for clustering at the household level. Wald_X: Wald-Test with H0: no joint significance of all regressors. (df=45) Significance levels: ** (0.01), * (0.05), + (0.1) Risk_finance 23 Tab A3: Descriptive Statistics of Covariates Variable Mean Std. Dev. Male 0.631 Foreigner 0.078 Age (in years) 41.089 Tenure (in years) 10.512 Apprenticeship 0.658 University degree 0.212 Prefers Social Democrats (SPD) 0.191 Prefers Christian Parties (CDU/CSU) 0.185 Blue collar worker 0.367 Father was self-employed 0.101 0.305 Firm size: 20 ≤ X < 200 workers 0.247 Firm size: 200 ≤ X < 2000 workers Firm size: X ≥ 2000 workers 0.235 Member of works council 0.040 Schleswig-Holstein 0.033 Hamburg 0.025 Lower Saxony 0.085 Bremen 0.007 North Rhine-Westphalia 0.219 Hessen 0.065 Rhineland-Palatinate, Saarland 0.054 Baden-Wuerttemberg 0.134 Bavaria 0.160 Berlin (East) 0.015 Mecklenburg/West-Pomerania 0.019 Brandenburg 0.030 Saxony-Anhalt 0.029 Thuringia 0.034 Saxony 0.055 Mining/ quarrying 0.011 Manufacturing 0.004 Electricity/ gas/ water supply 0.320 Construction 0.014 Wholesale and retail trade/ repair 0.067 Hotels/ restaurants 0.120 Transport, storage/ communication 0.022 Financial intermediation 0.059 Real estate/ renting/ business 0.054 Public administration/ defence 0.082 Education 0.056 Health/ social work 0.038 Other services/ Private households 0.106 Individual height (cm) 174.297 Source: SOEP 2003-2004. Longitudinal weights are used. N=5908. 24 0.482 0.267 10.75 9.666 0.474 0.409 0.393 0.388 0.482 0.302 0.460 0.431 0.424 0.197 0.181 0.156 0.279 0.087 0.414 0.247 0.226 0.341 0.366 0.122 0.139 0.173 0.169 0.182 0.229 0.108 0.067 0.466 0.119 0.250 0.325 0.146 0.236 0.227 0.275 0.231 0.193 0.308 9.116