Social preferences, monopsony and government intervention

Goerke, Laszlo; Neugart, Michael Article — Published Version Social preferences, monopsony and government intervention Canadian Journal of Economics/Revue canadienne d'économique Provided in Cooperation with: John Wiley & Sons Suggested Citation: Goerke, Laszlo; Neugart, Michael (2021) : Social preferences, monopsony and government intervention, Canadian Journal of Economics/Revue canadienne d'économique, ISSN 1540-5982, Wiley, Hoboken, NJ, Vol. 54, Iss. 2, pp. 864-891, https://doi.org/10.1111/caje.12515 This Version is available at: https://hdl.handle.net/10419/284749 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/4.0/ Social preferences, monopsony and government intervention Laszlo Goerke Universit¨at Trier; Institut f¨ ur Arbeitsrecht und Arbeitsbeziehungen in der Europ¨aischen Union; IZA Bonn; CESifo M¨ unchen Michael Neugart Technische Universit¨at Darmstadt Abstract. Monopsony power by firms and social preferences by consumers are well established. We analyze how wages and employment change in a monopsony if workers compare their income with that of a reference group. We show that the undistorted, competitive outcome may no longer constitute the benchmark for welfare comparisons and derive a condition that guarantees that the monopsony distortion is exactly balanced by the impact of social comparisons. We also demonstrate how wage restrictions and subsidies or taxes can be used to ensure this condition, both for a welfarist and a paternalistic welfare objective. R´ esum´ e. Pr´ ef´ erences sociales, monopsone et intervention de l’ ´ Etat. Le pouvoir de monopsone des entreprises et les pr´ ef´ erences sociales des consommateurs sont des concepts bien ´ etablis. Nous analysons la façon dont les salaires et l’emploi ´ evoluent au sein d’un monopsone lorsque les salari´ es comparent leurs revenus ` aceuxd’un groupe de r´ ef´ erence. Nous montrons que l’effet concurrentiel non fauss´ e ne repr´ esente plus la r´ ef´ erence en mati` ere de comparaison du bien-ˆ etre, et qu’il en d´ erive une condition permettant de garantir que l’effet de distorsion du monopsone soit exactement contrebalanc´ eparl’impact des comparaisons sociales. Nous montrons ´ egalement la façon dont les restrictions salariales, les subventions ou les impˆ ots peuvent ˆ etre utilis´ es pour garantir cette condition, ` a la fois dans une approche favorisant le bien-ˆ etre ou dans une optique paternaliste. JEL classification: D10, H21, J30, J42 Corresponding author: Laszlo Goerke, [email protected] We are grateful for helpful comments by Ronald Bachmann, Leif Danziger, Marco de Pinto, Fabian Kindermann, two anonymous referees and the managing editor, Katherine Cuff, as well as participants of the the 33rd Annual Conference of the Italian Association of Labour Economics (AIEL) in Ancona, the annual meeting of the Standing Field Committee for Population Economics of the Verein f¨ ur Socialpolitik (VfS) in W¨ urzburg, the 2019 International Conference on Public Economic Theory (PET) in Strasbourg, the Asian and Australasian Society of Labour Economics (AASLE) meeting in Singapore and members of the ZiF (Center of Interdisciplinary Research, Bielefeld) research group “In Search of the Global Labour Market: Actors, Structures, and Policies,”of which Michael Neugart was an associate member from October 2017 until July 2018. Canadian Journal of Economics /Revue canadienne d’´ economique 2021 54(2) May 2021. Printed in Canada /Mai 2021. Imprim´ e au Canada ISSN: 0008-4085 /20 /pp. 864–891 /DOI: 10.1111/caje.12515 ©2021 The Authors. Canadian Journal of Economics/Revuecanadienne d'´ economique published by Wiley Periodicals LLC on behalf of Canadian Economic Association. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. 864 L. Goerke and M. Neugart 1. Introduction SOCIAL PREFERENCES,OR more specifically social comparisons, play an important role in life. Individual decisions are substantially influenced by relative or positional considerations. Very often these positional considerations unfold in the labour market. For example, Neumark and Postlewaite (1998) and Park (2010) show that relative income concerns spur the entry of women into the labour market. Bracha et al. (2015) report experimental findings that relative pay enhances labour supply. Additionally, Bowles and Park (2005) present evidence suggesting that greater income inequality is associated with longer work hours. Clark and Oswald (1996) find that workers’satisfaction levels vary negatively with the wages of peers—an outcome that also arises when neighbours with higher earnings are chosen as the reference group (see, e.g., Luttmer 2005). Finally, the increased use of social media, such as Facebook or Instagram, has enhanced the scope for social comparisons (Appel et al. 2016, Krause et al. 2019). Concurrently, market power of firms has risen and labour markets have become more concentrated. The Council of Economic Advisors (2016) points out that increased firm concentration is accompanied by a downward trend in geographic mobility in the United States. Furthermore, forces counteracting monopsony power, such as labour unions, are becoming weaker. In line with these observations, there is evidence that more than a half of regional labour markets in the United States are highly concentrated, comprising 17% of employment (Azar et al. 2020). We also observe the emergence of crowd-working platforms that exhibit particularly high levels of monopsony power (Dube et al. 2020). Moreover, non-compete and non-poaching agreements are no longer restricted to highskilled employees but are also requested from low-skilled staff, such as those at fast food franchises (Krueger and Posner 2018). These various indications of firms’market power suggest that the narrative of a competitive labour market, according to which workers move on to another firm if their current employer lowers the wage by an even very small amount, is misplaced. Whether firms actually have market power can, inter alia, be derived from estimates of the labour supply elasticity to a single firm. After the pioneering work by Nelson (1973) and Sullivan (1989), these labour supply elasticities have been estimated in many contexts. Positive and finite values suggest that monopsony power exists in many countries, sectors and occupations (Sokolova and Sorensen 2018). These highly pervasive facts, the prevalence of social comparisons and of monopsonistic labour markets, constitute the empirical basis for our theoretical analysis. In particular, we investigate what happens if the two distortions interact. In addition, we derive the normative consequences of the coexistence of market power due to monopsony and income or consumption externalities owing to social preferences. We show that policy implications Social preferences, monopsony and government intervention 865 essentially depend on the size of two empirically observable parameters—the labour supply elasticity to the firm and the strength of social comparisons. Individuals who compare their income or consumption with that of others and exhibit jealousy have an incentive to expand labour supply in order to raise income and thereby improve their relative position (Frank 1984). This results in a downward movement of the labour supply curve in the wage– employment space. In monopsony, employment is determined by the marginal cost curve that exceeds the wage (Robinson 1933). Combining both aspects, our positive analysis indicates that predicting the wage and employment impact of social comparisons in monopsony requires restrictions relating to the labour supply elasticity to the firm. Adopting a normative perspective, we characterize efficiency both for a welfarist and a paternalistic objective. Interestingly, a welfarist social planner will not always prefer an employment level that equals the one that occurs on a competitive market without social comparisons. The planner internalizes the externalities resulting from social comparisons, but internalization is not generally equivalent to the absence of social preferences as the latter affects the marginal utility from consumption. Using these insights, we next consider the optimal regulation of wages and use of fiscal incentives. We show that the magnitude of the labour supply elasticity to the firm and an indicator of the strength of social comparisons can be used to determine whether minimum wages or wage caps (respectively taxes or subsidies) are required to achieve efficiency. The empirical evidence suggests that even substantial monopsony power may be balanced by the effects of jealousy, therefore limiting the efficiency-enhancing role of a minimum wage or wage subsidies. While there is extensive work on monopsonistic labour markets (Manning 2003) and widespread interest in the effects of social comparisons on market outcomes (see, among others, Persson 1995; Ireland 2001; Corneo 2002; Liu and Turnovsky 2005; Aronsson and Johansson-Stenman 2008, 2014, 2015, 2018; Wendner and Goulder 2008; Mujcic and Frijters 2015), our paper contributes to a less developed literature that looks into market outcomes when the two distortions, market power and income or consumption externalities, meet. Desiraju and Sappington (2007) and von Siemens (2010, 2012) study the impact of social comparisons in a monopsony. Contrary to our contribution, they are interested in workers’sorting behaviour into particular jobs, and firms’profits when workers have private information relating to their ability or social preferences. Goerke and Neugart (2017) analyze social comparisons in oligopsony in which heterogeneous firms have limited market power and compete for the same pool of labour. Using a framework based on Salop (1979), they show that a stronger prevalence of comparisons decreases wage inequality, shifts the functional income distribution in favour of workers and increases welfare. In contrast to Goerke and Neugart (2017), we scrutinize the suitability of various policy instruments to remedy the welfare losses resulting from the interaction of social comparisons and market power by employers for alternative objectives. Finally, Sandmo (1994) studies a two-part wage 866 L. Goerke and M. Neugart schedule. He shows that the monopsonist will equalize the effort-related wage component and a worker’s marginal productivity and use the fixed income component to raise profits at the expense of wage income. These benefits of second-degree price discrimination extend to a setting in which individuals undertake social comparisons. Consequently, there is no interaction between the two distortions we consider. In the next section, we describe our analytical apparatus. In section 3, we show how social comparisons affect the market outcome in a monopsony. We then characterize optimal employment from a welfarist perspective and investigate the use of wage regulations and taxes and subsidies in section 4. We assume a paternalistic social planner in section 5 as an important robustness check and conclude in section 6. 2. The model 2.1. General set-up We consider a world in which a monopsonist employs a large number of homogeneous individuals. All workers are employed by the monopsonist and variations in employment take place at the intensive margin. Workers derive utility from consumption and exhibit social preferences because they compare their own consumption with that of a reference group. From the perspective of an individual worker, reference consumption is exogenous. This kind of Nash behaviour implies that each individual creates an externality when deciding about consumption—and thereby labour supply—and, hence, neglects the impact on other individuals. We follow earlier contributions (e.g., Persson 1995, Dupor and Liu 2003) and assume that all workers are also identical ex post. This simplifies the exposition considerably and helps us to focus on efficiency aspects of the interaction of monopsonistic market power and social comparisons. In addition, we can then define welfare in a straightforward manner because we do not have to compare payoffs across individuals. In contrast to workers, the monopsonist takes into account that a wage change will alter both the consumption of each employee and the reference level and, hence, correctly anticipates the labour supply effects of altering the wage (Sandmo 1994). Therefore, the monopsonist partially internalizes the externalities due to social comparisons. 1 It can sell its output at a fixed price normalized to unity. Therefore, income changes do not alter product or labour demand. All workers are paid a wage, w, as there is no wage discrimination. They supply an amount of labour, L, resulting in labour income, wL. In addition, profits are redirected to workers. Accordingly, the functional income distribution is without impact, and we concentrate on the efficiency 1 This is not feasible in oligopsony, because wages of competitors and, hence, reference income, are given from the perspective of each oligopsonistic firm (Goerke and Neugart 2017). Social preferences, monopsony and government intervention 867 consequences of social comparisons in the normative part of our analysis. Workers are price-takers. They view profit income, π, as given and, thus, as unaffected by labour supply decisions. This assumption and the differential ability of the monopsonist and an individual worker to affect reference consumption reflect the idea that the firm has market power, while each individual’s actions have negligible effects on market outcomes. 2.2. Preferences A worker’s utility, U, increases in own consumption, c, at a decreasing rate and decreases in working time, L, at a weakly increasing rate, such that U¼Uðc,cr,γ,LÞ(1) and U cc ,U L <0<U c and U LL ≤0 hold, where subscripts denote partial derivatives. Moreover, utility varies with consumption of a reference group, cr. In previous empirical contributions, various kinds of such groups have been looked at, such as neighbours, parents, people who are comparable with respect to age, education, etc., and individuals who have the same occupation or colleagues (Luttmer 2005, Senik 2009, Goerke and Pannenberg 2015, Clark et al. 2017). Given our setting, we focus on colleagues. Moreover, we consider the case of jealousy, as defined by Dupor and Liu (2003), such that Ucr<0 holds. 2 Accordingly, in our model the employment-reducing impact of monopsony power could be counteracted by the employment-enhancing effect of social comparisons. Finally, the parameter γ≥0 indicates the intensity with which employees compare their consumption with that of the reference group. If, for example, social preferences are of the additive, or subtractive, type (Clark and Oswald 1998), we could specify utility as U¼Uðcγcr,LÞ. This specification is often distinguished from a multiplicative formulation (Carroll 2000), in which the ratio of own to reference consumption determines utility, such that Uðc=ðcrÞγ,LÞholds. For both specifications, the signs of U γ and Ucr, as well of U cγ and Uccr, coincide. Nonetheless, the distinction is analytically helpful, because it allows us to vary the intensity, γ, of social comparisons exogenously, whereas reference consumption, cr, is determined endogenously. As in Persson (1995), Corneo (2002) or Goerke and Hillesheim (2013), among others, we assume that utility is separable in consumption and labour supply (UcL ¼UcrL¼0). In our context, this enables us to unambiguously determine the impact of wages and social comparisons on labour supply. Moreover, the constraint implies that jealousy is equivalent to “keepingup-with-the-Joneses”(KUJ) preferences, which are empirically validated in a series of studies, see, e.g., Clark and Oswald (1996), Blanchflower and Oswald (2004), Ferrer-i-Carbonell (2005), Luttmer (2005) or Senik (2009). 2 In Goerke and Neugart (2020), we extend our analysis to the case of admiration, i.e., if utility is an increasing function of reference consumption or income. 868 L. Goerke and M. Neugart With additively separable preferences, KUJ implies Uccr>0 (which we subsequently assume to be the case). That is, an increase in reference consumption enhances the marginal utility from own consumption. Finally, we assume that the direct positive impact of a general increase in consumption dominates the indirect one via reference consumption. This holds both for the utility level, U, (Dupor and Liu 2003) and the marginal utility from consumption, U c (Liu and Turnovsky 2005), implying that UcþUcr>0>Ucc þUccrfor dc ¼dcr>0. These restrictions ensure that the aggregate labour supply curve is upward sloping. 2.3. Labour supply Individual labour supply: To reduce notational burden, we set the number of workers equal to one. The representative worker chooses working hours or labour supply to maximize utility subject to the budget constraint, c= wL +π. Because each worker regards profits πas fixed, the first-order condition for a utility maximum is dUðc,cr,γ,LÞ dL ¼Ucðc,cr,γÞwþULðLÞ¼0:(2) Given the separability assumption, marginal utility from consumption does not depend on working time directly, Uc¼Ucðc,cr,γÞ, and the marginal disutility from working is independent of consumption levels, U L =U L (L). Accordingly, individual labour supply is increasing in the wage, w, if the direct substitution effect dominates the income effect. Aggregate labour supply: Next, we consider the consequences of a higher wage paid by the monopsonist. To determine the impact of an encompassing wage increase, we have to incorporate not only the effect on own consumption, ∂c/∂w=L, but also the repercussion on the reference level, ∂cr=∂w, which will be positive if it is also financed by labour income. Moreover, the reference group will also adjust labour supply. Holding constant profits, the change in aggregate labour supply can be derived from Ucðcðw,LÞ,crðw,LÞ,γÞwþULðLÞ¼0:(3) Totally differentiating the above expression for c=wL +πand c¼crðw,LÞ yields the slope of the aggregate labour supply curve: dL dw ¼ dðUcðc,cr,γÞwþULðLÞÞ dw dðUcðc,cr,γÞwþULðLÞÞ dL ¼ UcþwðUccLþUccr∂cr ∂wÞ wðUccwþUccr∂cr ∂LÞþULL :(4) Because workers are homogeneous, reference consumption equals own consumption (c¼cr). As there are no costs other than wages, and with the production function denoted by f(L), profits can, hence, be written as π= f(L) −wL. It follows that consumption equals c¼cr¼wL þπ¼fðLÞ. In addition, we have ∂cr=∂w¼∂c=∂w¼0 and ∂cr=∂L¼∂c=∂L¼f0ðLÞ,so that the slope of the aggregate labour supply curve in equation (4) becomes Social preferences, monopsony and government intervention 869 Lw¼ Uc wðUcc þUccrÞf0ðLÞþULL >0:(5) Hence, the aggregate labour supply curve reflects the substitution effect of a wage increase, but no income effect anymore. Moreover, a greater importance of reference consumption raises aggregate labour supply, L=L(w,γ), as U cγ is positive. 2.4. Wage choice The production function, f(L), is characterized by standard properties, that is, f(0) =0, f0ð0Þ!∞and f0>0, f 00 <0 for L>0. The monopsonist maximizes profits by setting the wage, taking into account the impact on aggregate labour supply (as described in equation (5)): π¼fðLðw,γÞÞwLðw,γÞ:(6) Using the definition of the (aggregate) wage elasticity of labour supply, ϵ(w,L(w,γ),γ)=L w w/L>0, the first-order condition for a profit-maximizing choice can be expressed as πw¼f0ðLÞLwLwLw ¼Lðw,γÞϵðw,Lðw,γÞ,γÞ wf0ðLÞw1þϵðw,Lðw,γÞ,γÞ ϵðw,Lðw,γÞ,γÞ  ¼0:(7) The monopsonist will set a wage equal to the marginal product of labour, corrected by a factor that depends on the labour supply elasticity. The second-order condition is πww ¼Lðw,γÞϵðw,Lðw,γÞ,γÞ w f00ðLÞLw1þϵðw,Lðw,γÞ,γÞ ϵðw,Lðw,γÞ,γÞþw ðϵðw,Lðw,γÞ,γÞÞ2 dϵðw,Lðw,γÞ,γÞ dw <0: " (8) Given an upward-sloping aggregate labour supply curve, the secondorder derivative will surely be negative if the wage elasticity of labour supply, ϵ(w,L(w,γ),γ), weakly declines with the wage, w, or does not rise too strongly. Once the wage has been determined as shown in equation (7), the employment level can be found by calculating labour supply, implicitly defined by equation (3). 3. Positive analysis In this section, we investigate how wages and employment change with the intensity of social comparisons. We also consider two particular utility functions often used to resolve some of the ambiguities that remain for the general specification of preferences. 870 L. Goerke and M. Neugart 3.1. A general result The wage and employment effects of a change in the strength of social comparisons are summarized in the following proposition. PROPOSITION 1. Sufficient conditions for a greater intensity of KUJ preferences: (1) to increase labour supply and employment are ∂ϵ/∂γ≥0and ∂ϵ/∂w≤0.and (2) to decrease the wage are ∂ϵ/∂γ≤0and ∂ϵ/∂L≤0. Proof. See appendix. ▪ We can explain the proposition graphically and thereby also provide intuition. Figure 1 contains the textbook illustration of a monopsony. The thin, upward-sloping lines (γ=0) refer to the case without social comparisons. As is well known, the marginal cost curve for the monopsonist (thin dashed line) is situated above the labour supply curve it faces. The relative difference between the marginal product of labour and the wage is determined by the inverse of the labour supply elasticity to the firm, i.e., Pigou’s measure of exploitation (Boal and Ransom 1997, p. 88) Incorporating social comparisons (γ>0) has no impact on labour demand, as it is independent of employees’incomes. Jealousy, however, shifts the labour supply curve downwards in the wage-employment space. The bold line illustrates this effect in figure 1. Moreover, social comparisons affect the monopsonist’s marginal costs w(1 +(1/ϵ)) via alterations in the labour supply elasticity. If the labour supply elasticity weakly rises with more intense social comparisons and with a lower wage, originating from the shift in the labour , (=0) (>0) (=0) (>0) FIGURE 1 Wages and employment in monopsony with social comparisons NOTES: Sis labour supply and Dlabour demand. Bold lines for labour supply refer to the case of γ>0 and thin lines refer to the case of γ=0. Dashed lines correspond to the marginal costs (MC) of the monoposonist. The underlying calculations are based on a Cobb–Douglas production function and a utility function as in Ljungqvist and Uhlig (2000). Social preferences, monopsony and government intervention 871 π¼fðLðw,γÞÞð1þtÞwLðw,γÞ:(19) Because considerations of individuals are unaffected, the features of the labour supply curve are the same as outlined in section 2.2. Any tax receipts are returned in a lump-sum manner. Similarly, in case of tbeing a subsidy, a profit tax or another non-distortionary means of raising revenue is assumed to balance the government’s budget. Consequently, the only impact of the tax is the change in the firm’s wage choice. Maximization of profits as defined in equation (19), possibly amended to incorporate profit taxation or lump-sum payments, yields as first-order condition f0ðLM;tÞwð1þtÞ1þϵ ϵ¼0, (20) where L M,t denotes employment in the presence of a payroll tax or subsidy. Combining equation (20) with the outcome of the individual optimization (cf. equation (2)), we obtain f0ðLM;tÞ¼ð1þtÞ1þϵ ϵ ULðLM;tÞ Ucðγ≠0Þ:(21) The socially optimal outcome is defined by the derivative (A4) in the appendix, where the proof of proposition 2 is found. Evaluating this derivative at the market outcome, L M,t , and using our notation of Ucr=Ucðγ≠0Þ¼γ, we obtain dW dL γ≠0;L¼LM;t¼ðUcðγ≠0ÞþUcrÞð1þtÞ1þϵ ϵ ULðLM;tÞ Ucðγ≠0ÞþULðLM;tÞ ¼ULðLM;tÞ1ð1γÞð1þtÞ1þϵ ϵ  : (22) The expression in square brackets will be zero, such that welfare is maximized if topt ¼1 1γ ϵ 1þϵð1γÞ  :(23) The optimal tax or subsidy rate will be zero if the two distortions just balance out and the wage set by the monopsonist induces the optimal employment level. If 1 −γ<ϵ/(1 +ϵ), the impact of consumption externalities dominates the consequences of market power and t opt will be positive. In a competitive labour market (ϵ→∞), the optimal tax equals toptðϵ!∞Þ¼γ=ð1γÞ¼Ucr=ðUcðγ≠0ÞþUcrÞ>0. If the effects of social comparisons are relatively weak, and 1 −γ<ϵ/(1 +ϵ), the monopsonist will be subsidized. In the limiting case of preferences exhibiting no social comparisons, t opt (γ=0) =−1/(1 +ϵ)<0. Alternatively, an income tax, τ, or consumption tax, s, could be imposed on workers, such that their budget constraint, in the absence of any transfer or lump-sum tax, reads wL(1 −τ)+π−c=0orwL +π−c(1 +s)=0. In this case, the labour supply elasticity also depends on the tax (τ,s>0) or subsidy (τ,s<0). Proceeding in the same manner as in the derivation of t opt , the 878 L. Goerke and M. Neugart optimal income tax or subsidy rate, setting the consumption tax rate to zero, is (implicitly) defined by τopt ¼1þϵðτoptÞ ϵðτoptÞ ϵðτoptÞ 1þϵðτoptÞð1γÞ  , (24) while s opt =t opt . The optimal income tax rate, τ opt , will be positive (negative) if 1 −γ<(>)ϵ/(1 +ϵ). In the absence of labour market imperfections, the optimal tax rate equals τopt ðϵ!∞Þ¼γ¼Ucr=Ucðγ≠0Þ>0. 9 Wecansummarizetheconsiderationsofthissectioninthepropositionbelow. PROPOSITION 4. Assume that a social planner can affect welfare by setting tax or subsidy rates. A welfarist social planner will set the tax/subsidy rate on labour costs or on consumption expenditure in accordance with equation (23) and the tax/subsidy rate on wage income in line with the expression in (24). Proof. Follows from the above. ▪ Accordingly, in our simple setting either a minimum wage or a subsidy can raise employment if it is below the optimal level. Alternatively, a tax or a wage cap are both equally suitable as policy instruments if the effects of social comparisons dominate the monopsony distortion and employment needs to be reduced to enhance welfare. 4.4. An empirical assessment Our investigation reveals that the optimal use of wage regulation and fiscal incentives depends crucially on the magnitude of two measures. These are the ratio, γ, of the marginal utility from reference consumption to the marginal utility from own consumption on the one hand and the labour supply elasticity to the firm, ϵ, on the other hand. Estimates of the strength of income comparisons have, inter alia, been obtained from data on consumption choices and subjective well-being. Maurer and Meier (2008) and Alvarez-Cuadrado et al. (2016) estimate values of γ between 0.11 and 0.44 and of around 0.3 for US and Spanish data, respectively. Most findings for the parameter γ, however, result from discrete choice experiments, in which individuals compare two hypothetical situations. In one of them, own income is higher than in the other situation, while the reverse is true with respect to relative income. Such experiments have been conducted for various countries, groups of individuals and sample sizes. Most show a substantial heterogeneity in the prevalence of social comparisons. Estimates of the average magnitude of γrange from about 0.25 (Solnick and Hemenway 9 See, e.g., Persson (1995), Ljungqvist and Uhlig (2000), Dupor and Liu (2003), Aronsson and Johansson-Stenman (2010, 2013, 2018). From the results obtained by Liu and Turnovsky (2005) and Alvarez-Cuadrado (2007), we can derive comparable expressions, taking into account that they incorporate various taxes. Social preferences, monopsony and government intervention 879 1998, 2005 [United States]; Carlsson, Nam, et al. 2007 [Vietnam]; Shigeoka and Yamada 2019 [Japan]) to values of between 0.4 and 0.6 (Alpizar et al. 2005 [Costa Rica]; Johansson-Stenman et al. 2002 [Sweden]; Carlsson, Johansson-Stensman, et al. 2007 [Sweden]; Yamada and Sato 2013 [Japan]; Clark et al. 2017 [Japan]). 10 Similar magnitudes are obtained when comparisons refer to consumption goods, such as cars or housing (Alpizar et al. 2005; Carlsson, Johansson-Stensman, et al. 2007). In sum, the conclusion by Wendner and Goulder (2008, p. 1978) referring to earlier studies, according to which γ∈[0.2,0.4] is “a range for the status parameter that is consistent with the existing survey experimental evidence,”may be rather conservative. Turning to monopsony power, at least for the United States there is a prominent if not dominating view that “[m]onopsony prevails in a large number of ... labor markets”(Marinescu and Posner 2020). This interpretation is often based on indicators of labour market concentration, such as the Herfindahl–Hirschman index (Azar et al. 2019), and estimates of the labour supply elasticity to the firm. The latter vary widely across markets and countries (Manning 2011). The meta-analysis by Sokolova and Sorensen (2018) reports a mean elasticity of 7 and a median of 1.43 for European countries. The respective values for other advanced countries, including the United States and Canada, are somewhat lower. In addition, there is evidence for the United States that the labour supply elasticity has decreased substantially from 1.2 to about 1 on average over the last two decades (Webber 2021). For online labour markets, labour supply elasticities, ϵ, as low as 0.1 have been computed (c.f. Dube et al. 2020). Combining information on the labour supply elasticity to a firm and the strength of social comparisons allows us to determine the nature of optimal wage regulation and welfare-enhancing fiscal incentives. Assume, initially, a low impact of reference group income on the marginal utility from consumption (γ=0.2). In such a case, a minimum wage or subsidies enhance welfare if the labour supply elasticity, ϵ, is less than 4. The extant literature suggests that this is usually the case. For an intermediate value for the strength of social comparisons, namely γ=0.4, employment will be excessive if the labour supply elasticity is greater than 1.5. The meta-analysis by Sokolova and Sorensen (2018) indicates that about 50% of all estimated values exceed this threshold. If, finally, a high comparison intensity is presumed, γ=0.6, employment will be excessive if the labour supply elasticity is greater than 2 3. This is likely to be the case in most labour markets. Assuming ϵ=1.5 and γ=0.5, the optimal tax rate on labour costs as defined in equation (23) would be t opt (ϵ=1.5,γ=0.5) =0.2, setting the other rates to zero (s opt =τ opt =0). The optimal income and consumption tax rates would be τoptðϵ¼1:5, γ¼0:5Þ¼0:16 6 and s opt (ϵ=1.5,γ=0.5) =0.2, 10 Shigeoka and Yamada (2019) estimate a somewhat smaller value for γfor the United Kingdom, but observe no envy or jealousy for respondents from the United States. 880 L. Goerke and M. Neugart respectively. Raisingthe measureof the intensity of social comparisons to γ=0.6, increases optimal tax rates to t opt (ϵ=1.5, γ=0.6) =0.5, τoptðϵ¼1:5, γ¼ 0:6Þ¼0:33 3 and s opt (ϵ=1.5, γ=0.6) =0.5. Because real-world fiscal systems consist of a combination of many more tax and contribution rates than we consider, the optimal rates as defined in equations (23) and (24) are more indicativeofthetotalmarginaltaxburdenthancomparabletoactualrates. These illustrative computations clarify that the question whether social comparisons mitigate monopsony power, or even dominate its effects, is not only of academic interest but also of great empirical relevance. Our summary indicates that wage restraints or taxation of income, labour costs and consumption may be a relevant policy option even in many monopsonistic labour markets. 5. Paternalistic social planner Section 4 assumes that the social planner maximizes the utility of the representative worker. However, it has been argued that individual preferences that incorporate jealousy may be inappropriate as starting point of a normative investigation. Accordingly, the analysis of optimal taxation in competitive settings has occasionally been based on the assumption of a paternalistic or nonwelfarist social planner (see, for example, Aronsson and Johansson-Stenman 2018, Dodds 2012, Eckerstorfer and Wendner 2013, Micheletto 2011). In this case, the social planner maximizes the payoff of the representative consumer, ignoring repercussions of her choice via changes in utility due to social comparisons. This is tantamount to maximizing utility as specified in equation (1) for an exogenously given reference consumption. 11 While it is clearly debatable if jealousy harms another individual, if this individual has no such preferences and, thus, if jealousy has to be disregarded, it is nonetheless insightful to investigate in how far propositions 3 and 4 depend on the specification of the normative objective. Therefore, we subsequently assume that the social planner is paternalistic (or non-welfarist). We denote her objective by W p and indicate the modification by expressing utility as function of an exogenous level,  cr, of reference consumption, Wp¼Uðc, cr,LÞ. Workers are homogeneous and obtain the monopsonist’s profits as (exogenous) income. Consequently, we continue to focus on the efficiency properties of the allocation. Maximization of W p yields dWp dL ¼UcðfðLÞ, crÞf0ðLÞþULðLÞ¼0:(25) 11 Hence, we adopt the approach chosen, for example, by Aronsson and Johansson-Stenman (2018) and Aronsson et al. (2019). The former also provide a thorough discussion of the merits of such an approach. If we instead assume that the objective depends on consumption, c, and working time, L (see Dodds 2012, inter alia) only, the social planner completely ignores relative preferences. Social preferences, monopsony and government intervention 881 We denote the ratio of the marginal utility from own consumption when repercussions via social comparisons are ignored, UcðfðLÞ, crÞ, to the marginal utility in their presence, U c (f(L),f(L),γ), by μ(L,γ), 1 >μðL,γÞ¼UcðfðLÞ,  crÞ=UcðfðLÞ,fðLÞ,γÞ>0. This ratio is an (inverse) indicator of the strength of the consumption externality resulting from social comparisons. This is because an increase in the strength of social comparisons and in reference consumption raises the marginal utility from own consumption, U c (f(L),f(L),γ), and reduces μ(L,γ). Employing the ratio μ, we can establish the proposition below. PROPOSITION 5. A paternalistic social planner who can affect welfare solely: (1) by fixing the wage will set it at a lower level than the monopsonist if 1/μ(L,γ)>(1 +ϵ)/ϵ, (2) by taxing or subsidizing labour costs (t)or consumption expenditure (s) will set the tax/subsidy rate equal to t opt,W =s opt,W =1/μ(L,γ)[ϵ(s opt,W )/ (1 +ϵ(s opt,W )) −μ(L,γ)] or (3) by taxing or subsidizing wage income (τ)will set the tax/subsidy rate equal to τ opt,W =1−μ(L,γ)(1 +ϵ(τ opt,W ))/ϵ(τ opt,W ). Proof. See appendix. ▪ The gain from more employment is given by the increase in consumption for the representative worker. The respective gain for the worker is larger than the social planner’s. This is the case because the worker also fares better in terms of social comparisons. Since the utility loss from working more is unaffected by the existence of social preferences, the individual incentives to work more are greater than the socially optimal ones. Consequently, labour supply is, ceteris paribus, excessive not only for a welfarist but also for a paternalistic social planner. However, as the labour market features a monopsonistic employer, employment is, ceteris paribus, too low. Proposition 5, part (a), indicates that if the supply side distortion, as captured by 1/μ(L,γ), exceeds the demand side distortion, (1 +ϵ)/ϵ, wages need to be lowered because employment is excessive. In line with this, labour costs, consumption or income will have to be taxed if μ(L,γ)<ϵ/(1 +ϵ). 12 12 If the social planner completely ignores relative preferences, W p depends on only consumption, c, and working time, L, only, and she sets an employment level that equals the one that results in a competitive market in the absence of social comparisons. Our previous results are not affected qualitatively by whether reference consumption is held constant or omitted, since they depend on the magnitude of the labour supply elasticity, ϵ, relative to an indicator of the consumption externality, given by ~μðL,γÞ¼UcðfðLÞÞ=UcðfðLÞ,fðLÞ,γÞ. Note that if utility were not separable, UcL,UcrL≠0, the ratios μ(L,γ)or~μðL,γÞ would have to be redefined to also include the ratio of the marginal utility from work in the presence of social comparisons and when they are ignored. Given this modification, results would also be unaffected. 882 L. Goerke and M. Neugart Turning to the specific utility functions in equations (9) and (10), the marginal utility from own consumption, holding constant the reference level or, alternatively, setting γ=0, equals UcðfðLÞ, crÞ¼cβfor c=f(L). If the repercussions of a marginal variation in consumption via reference consumption are taken into account, marginal utility for specification (9) is given by UcðfðLÞ,fðLÞ,γÞ¼ðcβÞ=ð1γÞ. Consequently, we obtain μðL,γÞ¼UcðfðLÞ, crÞ=UcðfðLÞ,fðLÞ,γÞ¼1γ. Comparing proposition 5 with propositions 3 and 4, we can observe that for this specification of preferences, the normative implications of the joint existence of monopsony power and social comparisons are independent of social planner’s preferences. This is the case because the extent of the consumption externality, which is relevant for the welfarist objective, is the same as the extent of the deviation from the undistorted outcome, which determines a paternalist’s behaviour. Although the objectives of a welfarist and a paternalist social planner therefore differ, their choice of wages or taxes/subsidies will be the same. This assertion holds not only for the preferences defined in equation (9) but also more generally for all difference specifications of utility, U¼Uðcγcr,LÞ. 13 For the formulation of preferences in equation (10) proposed by Gali (1994), we have UcðfðLÞ,fðLÞ,γÞÞ ¼cβþγðβ1Þ, implying that μðL,γÞ¼ cγðβ1Þ¼fðLÞγðβ1Þ. Therefore, the social planner’s choice of wages or taxes does not depend only on the strength of social comparisons. This is because the extent of the consumption distortion resulting from social comparisons, which the paternalist social planner needs to internalize, varies with the consumption level. Consequently, for the multiplicative specification of preferences defined in equation (10) the behaviour of a paternalist social planner will deviate from that of her welfarist counterpart. We can conclude that the social planner’s basic trade-off is independent of her objective. The optimal response in terms of wage regulation or taxes/ subsidies depends on the strength of the monoposony distortion relative to the consumption externality. 6. Conclusions We derive fairly general conditions on the labour supply elasticity to the firm that allow us to sign the wage and employment effects of social comparisons in monopsony. Two specific utility functions exemplify the more general conditions. Assuming that workers compare their consumption in absolute terms and using a utility function suggested by Ljungqvist and Uhlig (2000), we find that employment increases in the prevalence of social comparisons, while 13 A similar result is obtained by Aronsson and Johansson-Stenman (2018), who look at optimal taxation and show (cf. proposition 1 and corollary 1) that the marginal tax rate chosen by a welfarist and a paternalist social planner are the same if individuals are homogeneous and preferences are additive. Social preferences, monopsony and government intervention 883 wages decline. Using a utility function as in Gali (1994), we derive for a case of relative comparisons that the employment and the wage effects of more intensive social comparisons will both be positive if the production function is not too concave. Interestingly, a welfarist social planner will not necessarily choose an employment level equal to the one in a competitive market without social comparisons. She will do so only for rather special properties on the marginal utility of a worker’s own and reference consumption. Our findings bear novel and important policy implications. A social planner who tries to achieve optimal employment by setting wages would not always employ a minimum wage. If the labour supply elasticity to the firm is sufficiently large, she would rather cap wages. Such a wage restriction will prevent the monopsonist from choosing employment in excess of the optimal level. This will be the case if the externality due to social comparisons is strong enough. Analogously, we find conditions for an optimal use of either subsidies, or alternatively, taxes in a monopsony with social comparisons. Given the evidence that the labour supply elasticity to a monopsonist varies with the business cycle (Hirsch et al. 2018), this implies that optimal policy may alternate between minimum and maximum wages or positive and negative tax rates, respectively. A qualitatively similar conclusion emerges if supply elasticities vary across labour markets, providing an additional argument for differentiations of wage regulations. Such challenges to determining optimal policies would be augmented if also the intensity of social comparisons varied with the economic situation or regionally. Importantly, these conclusions are qualitatively independent of the exact specification of the welfare objective. This clarifies that policy conclusions are rather robust with respect to society’s preferred outcome because the nature of the distortions is unaffected. In our set-up, focusing on the efficiency properties of the interaction of market power and consumption externalities, one instrument is sufficient to achieve the social planner’s objective. Therefore, in section 4, we consider wages and taxes separately. If the social planner pursued a distributional objective in addition, for example, because individuals were heterogeneous ex post or firms were not owned by workers, she would require more than one instrument to achieve her objective. In particular, non-linear taxes could then help to realize the distributional aims. In addition, we assume that the monopsonist pays uniform wages. If, however, the monopsonist could undertake price discrimination, the monopsonistic employment inefficiency would be mitigated. In case of perfect wage discrimination, only the distortion resulting from consumption externalities would remain. Given incomplete wage discrimination, it can be conjectured that the monopsony distortion becomes less relevant, relative to the consequences of social comparisons, and that taxes and wage caps become more important as policy instruments. The investigation of policy implications in such more comprehensive set-ups, in which also distributional questions become an issue, is clearly beyond the scope of the present paper. It represents a promising topic for future inquiry. 884 L. Goerke and M. Neugart Appendix Appendix A1: Proof of proposition 1 The derivative of the aggregate labour supply curve L(w,γ) with respect to γ, taking into account wage repercussions, is dLðw,γÞ dγ¼LγþLw dw dγ¼LγLw πwγ πww :(A1) Substituting for the wage effect, we obtain dLðw,γÞ dγ¼LγLw f00ðLÞLγþw ϵ2 ∂ϵ ∂γþ∂ϵ ∂LLγ  f00ðLÞLw1þϵ ϵþw ϵ2 ∂ϵ ∂wþ∂ϵ ∂LLw  ¼ Lγ 1þϵ ϵþw ϵ2Lγ ∂ϵ ∂wLw ∂ϵ ∂γ  πww : (A2) Because the denominator is negative according to the second-order condition (8), the employment effect is unambiguously positive for ∂ϵ/∂w≤0 and ∂ϵ/∂γ≥0. This proves part (a). The derivative of the first-order condition of the firm (7) with respect to γ is πwγ¼Lðw,γÞϵ wf00ðLÞLγþw ϵ2 ∂ϵ ∂γþ∂ϵ ∂LLγ  :(A3) Because labour supply rises with the intensity of social comparisons (L γ >0), the term in square brackets will surely be negative if the wage elasticity of labour supply rises neither with the strength of social comparisons nor with employment. This proves part (b). Appendix A2: Proof of proposition 2 Maximizing W=U(f(L),f(L),γ,L) with respect to Lyields as first-order condition in the presence of social comparisons (γ≠0) dW dL γ≠0¼ðUcðγ≠0ÞþUcrÞf0ðLÞþULðLÞ¼0:(A4) Denote the resulting employment level by L opt,γ≠0 . The second-order condition holds, because f00ðLÞ,Ucc,Ucrcr<0, and U LL ≤0. Since there are no distributional effects of the market outcome on welfare in our setting, employment resulting in a competitive market without market power and externalities is equivalent to the social planner’s choice for γ=0. This choice, L opt,γ=0 , is determined by dW dL γ¼0¼Ucðγ¼0Þf0ðLÞþULðLÞ¼0:(A5) Social preferences, monopsony and government intervention 885 Because U L (L) and f0(L) are the same for a given employment level, the social planner’s choice in the presence of social comparisons and the outcome in a competitive market in their absence will coincide (L opt,γ≠0 =L opt,γ=0 ), if Ucðγ≠0ÞþUcr¼Ucðγ¼0Þ.IfUcðγ≠0ÞþUcr>Ucðγ¼0Þholds, L opt,γ≠0 will exceed L opt,γ=0 because Wis strictly concave in L. Appendix 3: Proof of proposition 3 Employment in a monopsony, denoted by L M , is implicitly defined by equation (7). Moreover, labour supply is given by equation (2). Combining both equations yields f0ðLMÞϵ 1þϵ¼ULðLMÞ Ucðγ≠0Þ:(A6) Evaluating the social planner’s choice as defined in (A4) at L=L M yields dW dL γ≠0;L¼LM¼ðUcðγ≠0ÞþUcrÞf0ðLMÞþULðLMÞ ¼ULðLMÞ Ucðγ≠0ÞUcðγ≠0ÞþUcr ½ 1þϵ ϵþULðLMÞ ¼ULðLMÞ 1þϵ ϵ1þUcr Ucðγ≠0Þ  þ1  : (A7) Using Ucr=Ucðγ≠0Þ¼γ, the social planner’s objective will, hence, be maximized by the market outcome if 1 −γ=ϵ/(1 +ϵ) and she will want to increase (reduce) employment above (below) L M if (1 −γ)(1 +ϵ)/ϵ>(<)1 holds, given U L <0. Employment can be increased (decreased) by marginally raising (lowering) the wage above (below) the level set by the monopsonist. Appendix 4: Proof of proposition 5 Evaluating the first-order condition shown in equation (25) at the market outcome, as defined by the equality in (A6), and taking into account that marginal utility depends only on Lyields dWp dL γ≠0;L¼LM¼UcðfðLMÞ, crÞULðLMÞ UcðfðLMÞ,fðLMÞ,γÞ 1þϵ ϵþULðLMÞ ¼ULðLMÞ1μðLM,γÞ1þϵ ϵ  : (A8) The social planner’s objective will, hence, be maximized by the market outcome if μ(L,γ)=ϵ/(1 +ϵ) and she will increase (reduce) employment above (below) L M if μ(L,γ)(1 +ϵ)/ϵ>(<) 1 holds, given U L <0. This proves part (a) of the proposition. If the firm pays a tax on labour costs, t, the market outcome can be described by equation (21). Substituting in equation (25), we obtain 886 L. Goerke and M. Neugart dWp dL γ≠0;L¼LM;t¼UcðfðLM;tÞ, crÞULðLM;tÞ UcðfðLM;tÞ,fðLM;tÞ,γÞð1þtÞ1þϵ ϵþULðLM;tÞ ¼ULðLM;tÞ1μðLM;t,γÞð1þtÞ1þϵ ϵ  : (A9) Solving this expression yields t opt,W as defined in proposition 5, part (b). If the tax is levied on consumption expenditure, such that the worker’s budget constraint equals wL +π−c(1 +s)=0, the market equilibrium can be characterized by f0ðLM;sÞ¼ð1þsÞ1þϵ ϵ ULðLM;sÞ UcðfðLM;sÞ,fðLM;sÞ,γÞ:(A10) Substituting in the first-order condition (25), and evaluating it at the market outcome, yields s opt,W =t opt,W , where ϵ=ϵ(s opt,W ). This completes the proof of part (b) of the proposition. Finally, if the tax is levied on labour income, τ, the market outcome is given by f0ðLM;τÞ¼ 1þϵ ϵð1τÞ ULðLM;τÞ UcðfðLM;τÞ,fðLM;τÞ,γÞ:(A11) Proceeding in the same manner as above, part (c) can be established. References Alpizar, F., F. Carlsson, and O. Johansson-Stenman (2005) “How much do we care about absolute versus relative income and consumption?,”Journal of Economic Behavior and Organization 56(3), 405–21 Alvarez-Cuadrado, F. (2007) “Envy, leisure, and restrictions on working hours,” Canadian Journal of Economics 40(4), 1286–310 - (2016) “Envy and habits: Panel data estimates of interdependent preferences,”Oxford Bulletin of Economics and Statistics 78(4), 443–69 Appel, H., A. L. Gerlach, and J. Crusius (2016) “The interplay between Facebook use, social comparison, envy, and depression,”Current Opinion in Psychology 9, 44–49 Aronsson, T., and O. Johansson-Stenman (2008) “When the Joneses’consumption hurts: Optimal public good provision and nonlinear income taxation,”Journal of Public Economics 92(5), 986–97 - (2010) “Positional concerns in an OLG model: Optimal labor and capital income taxation,”International Economic Review 51(4), 1071–95 - (2013) “Conspicuous leisure: Optimal income taxation when both relative consumption and relative leisure matter,”Scandinavian Journal of Economics 115(1), 155–75 - (2014) “Positional preferences in time and space: Optimal income taxation with dynamic social comparisons,”Journal of Economic Behavior and Organization 101, 1–23 Social preferences, monopsony and government intervention 887