Hill – Mankiw 9th Edn Chapter 19 – Earnings and Discrimination
Mankiw, N. G. (2021) Principles of microeconomics (9th ed.)
Principles of economics (9th ed.)
Mason, OH: South-Western Cengage Learning.
Rod Hill
University of New Brunswick, Saint John campus
Saint John, New Brunswick, Canada
email: rhill@unb.ca
Chapter 19 – Earnings and Discrimination
Here are some things to consider when reading this chapter.
- Compensating Wage Differentials
As presented in the textbooks, including Mankiw’s, compensating wage differentials are explained using perfectly competitive labour markets. This implicitly assumes that workers have perfect information about the nature of different kinds of work and workplaces, and that people can move freely between jobs. In Mankiw’s examples, coal miners and night shift workers know the health risks that accompany such jobs, and have alternatives easily available to them. Compensating wage differentials change firms’ costs so that, on average, firms and the buyers of their products pay the true marginal cost of production.
If this was really the case, there would be no need for workplace health and safety regulation. So, for example, the estimated 2.9 million work-related deaths in 2019 (Takaka et al. 2023) would have involved risks which workers knowingly accepted in exchange for appropriately higher wages. Workplace health and safety regulation would not only be redundant, it would potentially prevent workers from willingly accepting high risk jobs in exchange for higher pay. As Kevin Purse remarks: “In the Panglossian world of neoclassical economics, the power relations that are integral to the employment relationship are ignored and conflict over workplace health and safety between employers and workers is abstracted out of existence” (2004, p. 613).
Workplace health and safety regulations are justified by incomplete information and the difficulty that people have in assessing statistical risks from such things as exposure to hazardous materials in the workplace. Of the 2.9 million annual deaths noted above, almost 2.6 million are the result of work-related diseases that develop over a long period of time – cancers and cardiovascular diseases, for example. As Purse comments: “In contrast to traumatic fatalities, the risks associated with industrial disease tend to be very much less visible and, consequently, less likely to be known to workers at the time of their exposure” (2004, p. 614). Those who most underestimate the risks of risky jobs will take them and accept too little compensation compared with the true risks, if they knew them and assessed them correctly.
Even if the risks are known, compensating differentials depend on the ability of workers to take jobs elsewhere. For example, suppose working in a coal mine becomes more hazardous. In the simple neoclassical labour market, the marginal worker is indifferent between working in the coal mine and taking employment elsewhere. More hazardous work would cause such workers to take a different job, reducing the supply of coal miners and increasing their wages and thus the compensating differential.
However, more realistically, worker mobility can be inhibited by relocation costs, loss of social connections, and the cost of getting information about job opportunities elsewhere. In many occupations, workers have also acquired skills whose value can be lost by taking another job. The resulting imperfect mobility of labour weakens the adjustment of wages to changes in job characteristics.
The practical problems with compensating differentials do not end there. In a well-known paper, George Akerlof and William Dickens (1981) showed that even if people have the facts, they may choose to believe something else. Applying the psychological concept of cognitive dissonance, they explained that “workers have preferences over their states of beliefs”. They write that people “who choose to believe the job is safe do not experience the unpleasant feelings of constant fear or unsettling doubts about how wise it was to take such a dangerous job” (p. 308). In the simple model developed by Akerlof and its Dickens, workers who have risky jobs and have chosen their beliefs subsequently have the option of buying safety equipment to reduce their risk of injury. Safety legislation can be beneficial because workers have systematically undervalued the safety equipment which, because of their altered beliefs, they have chosen not to purchase.
Given all this, why would actual wages offer appropriate compensation for those facing workplace hazards?
Finally, as the Commentary for the previous chapter pointed out, the perfectly competitive model omits critical features of labour markets which are important in determining wages. As Kurt Lavetti writes, “Although the concept of a compensating differential is straightforward and intuitive, its simplicity belies how difficult it is to empirically quantify compensating differentials in real-world labor markets” (2023, p. 190). In their attempts to do so, labour economists have ended up employing models that incorporate not only imperfect information, but also wage bargaining and rent sharing.
- Labour Unions
(i) Labour Unions: the missing facts and analysis of the long decline in union membership
Mankiw’s mention of labour unions could hardly be shorter: just a single paragraph in the entire book (p. 385). In contrast, a nonstandard text such as Microeconomics in Context (Goodwin et al. 2023) spends two pages on the role of unions in explaining variations in wages (pp. 291-93).
Goodwin and co-authors give historical background, noting that unions were not legally recognized in the United States until 1935. They present data showing the variation over the last 70 years in union membership in the United States as a percentage of the workforce. (See Figure 1.)
The long decline in the unionization rate is partly the result of shifts in employment from manufacturing to service occupations. Compared with manufacturing, these sectors employ a larger proportion of female workers and have grown at the same time that female labour force participation also increased. While there is some evidence that women are more likely than men to want to join a union, this has not offset the greater difficulties of organizing unions in service occupations, such as retail and restaurant work. In addition, women may face additional barriers in unionizing. Schur and Kruse (1992, p. 90) write that “unions may be reluctant to organize female workers because they view women’s higher turnover rates will make them difficult to organize and retain”, while “family commitments and time pressures” restrict their ability to participate in union activities. As well, Schur and Kruse note that the increase in female labour force participation took place at a time when employers mounted greater resistance to unionization.
Goodwin and co-authors note that “economic power is closely related to political power”, writing that “the balance of political power began to shift dramatically in the 1970s and 1980s” in favour of corporate interests (p. 311). The stronger corporate lobbies became, the more they were able to weaken countervailing forces, by creating an increasingly anti-union regulatory environment, for example (p. 292). No ideas like this appear in Mankiw’s text.
Decisions about what to include in a textbook – and what not to include – inevitably involve authors’ views about what is important and what isn’t. Such value judgements necessarily reflect the authors’ ideological outlooks. Goodwin and her co-authors seem to view the gradual weakening of unions in the United States as having undesirable consequences; apparently Mankiw does not. Readers of his text may or may not share that judgement, but they should be aware that it is being made. It’s easy to see what’s included in the text, but it requires some thought to notice what isn’t.
(ii) Labour unions, wages, and employment: alternative views
Mankiw’s description of the effect of unions on wages and employment assumes that unions can set a wage above the competitive equilibrium wage that would otherwise prevail. Employers choose how much labour to employ at that wage. The result is a reduction in employment similar to that of a minimum wage, again in comparison with the competitive equilibrium. (See Figure 5, p. 116). Unions benefit their membership at the expense of other workers, creating inefficiency. Clearly, students are invited to conclude that unions are, on balance, undesirable.
Mankiw’s analysis takes it for granted that the standard of comparison is a perfectly competitive equilibrium. But as the Commentary on the previous chapter pointed out, employers typically have some monopsonistic power. When the labour market is monopsonistic, workers will have an incentive to form unions to counteract employers’ market power. As well, the assumption that product markets are perfectly competitive means that firms have no economic profits in the long run. In reality, unions form in the private sector where markets are oligopolistic and employers have economic profits. Workers, through unions, can bargain for a share of those profits.
A standard assumption in labour economics is that unions care about both wages and employment. In the same way that a minimum wage could potentially be used to raise both wages and employment compared to the pure monopsonistic situation, a labour union could do the same.
Figure 2 shows the range of possible outcomes when a labour union bargains with an employer who has monopsonistic power in the labour market. This is sometimes called ‘bilateral monopoly’ and the resulting negotiated wage depends on whatever factors determine the relative bargaining strength of each side.
If the union negotiates a wage between the monopsony wage and the hypothetical competitive wage shown by point C, the labour supply curve determines the number of workers who will be employed. In that case, both wages and employment increase compared with the monopsony wage and employment. If the union is strong enough to negotiate a wage above C, the value of workers’ marginal product determines how many workers the firm will employ, so higher wages then involve a trade-off in terms of employment. As long as the negotiated wage is less than $80/day, employment would be higher with a union then without one.
What about the empirical evidence? Alan Manning writes that the evidence is sparse and the results mixed, but that many studies support the conclusion of the monopsony model (2003, p. 351).
Finally, this simple static analysis of Figure 2 does not consider how unionization could affect workers’ future productivity, and possibly firms’ demand for labour and their profitability. This can be easily modified by considering the effects of the greater job security the unionized workers may enjoy. This could increase their willingness to invest in additional training and skills development, increasing their productivity.
(iii) Effects on nonunion wages
Mankiw’s brief discussion deals only with a labour market in which all workers are unionized. He cites the estimated 10 to 20 percent union-nonunion wage differential but does not explain how this could be interpreted. It is not the amount by which the wage has been increased in the unionized firm compared to what it otherwise would have been.
In his labour economics text, George Borjas explains that collective agreements make it difficult for unionized firms to fire or to lay off workers. As a result, “the unionized firm will want to screen job applicants very carefully” and a higher union wage will give them plenty of applicants to choose from. “Over time, the firm’s workforce will be comprised of workers who are relatively more productive than workers in nonunion firms” (2019, p. 365). But the union-nonunion wage gap should be estimated by comparing equivalent workers in union and nonunion jobs. Because the higher skill level in unionized firms can only be imperfectly observed, the effect of unions on wages is over estimated.
Wages in nonunionized firms could be affected in several ways. For example, if the creation of a union reduces employment in unionized firms, as in Mankiw’s analysis, it increases the supply of labour available to nonunion firms in the local labour market, lowering wages in such firms at least in the short term. (In the long run, all else equal, such lower wages could induce some workers to leave the local labour market for greener pastures.)
On the other hand, if the monopsony model sketched out in the previous section is relevant, the formation of unions will tend to increase wages for workers in nonunionized firms. If unionized firms increase employment, employers in nonunionized firms would face increased competition for workers in the local labour market. As well, the increased prevalence of unions in a local labour market is also consistent with the so-called ‘threat effect’: nonunionized firms increase wages to reduce their workers’ incentive to unionize.
The above discussion implicitly assumes the unionized and nonunionized firms in a sector are relatively similar. As noted earlier, large firms with market power and profits are attractive targets for unionization, even though management in large firms has ways of discouraging unionization. If nonunionized firms are smaller, less profitable and their workers less productive, this helps to explain their nonunion status.
(iv) Effects of unions on general working conditions
Goodwin and co-authors quote Lawrence Mishel (2012, p. 9), who gives examples of how union activity has benefited workers as a whole. Unions
affect non-union pay and practices [by instituting] norms and practices that have become more widespread throughout the economy, thereby improving pay and working conditions for the entire workforce… Many fringe benefits, such as pensions and health insurance, were first provided in the union sector and then became more commonplace. Union grievance procedures, which provide due process in the workplace, have been adopted to many nonunion workplaces. Union wage setting… has frequently established standards for what workers expect from their employers.… [Unions] remain a source of innovation in work practices (e.g., training and worker participation) and in benefits (e.g., childcare, work-time flexibility, and sick leave).
(v) Summing up Mankiw on unions
Mankiw’s brief portrayal of unions is superficial, omitting historical context and the decline of unionization rates that has been partly due to the growing lobbying power of business. It analyses the effects of unions in a perfectly competitive market, ignoring the possibility that unions exist to respond to employer power in imperfectly competitive labour markets, while seeking to share economic profits that firms have in imperfectly competitive product markets.
- Labour Markets Are Not Like The Market For Bananas
Yet, as Yanis Varoufakis writes, the basic neoclassical textbook model of the labour market “assumes that labour can be purchased in a manner similar to the way bananas change hands in some fruit market” (1998, p. 170). Hence, the application of the supply and demand model to the labour market that we see in Mankiw.
However, labour markets involve the purchase of the time and effort of people, not inanimate objects. As a result, the “fact that workers bring with them not only their labour-power but also their past history and norms of justice in the workplace is considerably more important in determining both their relative and average level of wages than purely market forces of supply and demand” (Seccareccia 1991, p. 45).
In this post-Keynesian view, these social and equity norms are strong enough to marginalize market forces, so much so that the labour market really doesn’t exist. “Wages have little to do with an individual’s productivity”, leaving “a wide range of indeterminacy in the establishment of wages by nonmarket forces” (Seccareccia 1991, p. 48). In any case, an individual’s marginal productivity is often fundamentally unobservable, as described the first part of the Commentary on Chapter 19.
Mankiw does mention one of the special features of labour markets. In Chapter 22, he acknowledges the importance of fairness: “in Chapters 18 and 19, we discussed how wages were determined by labor supply and labor demand. Some economists have suggested that the perceived fairness of what the firm pays its workers should also enter the picture.” His example: workers might expect a share of the firm’s economic profits (or rents) and firms “might well decide to give workers more than the equilibrium wage for fear that the workers might otherwise try to punish the firm with reduced work effort, strikes, or even vandalism” (p. 460).
More generally, unlike the price of bananas, people’s views about what wages should be affects what wages actually are. For example, it was long considered acceptable to pay women less than men because men were considered to be the primary or even the sole earners in a family, while women’s income was seen as supplementary. This view became untenable when women’s participation in the labour force greatly increased from the early 1970s, while at the same time divorce rates increased significantly. The old social norm became socially unacceptable. Political pressure led to ‘equal pay for equal work’ regulations which helped to shrink the gap.
The concept of efficiency wages, given three sentences on page 385, effectively acknowledges that the story told earlier in the text about the production function is incomplete. A firm can hire an additional hour of work, but its marginal product depends on the worker’s choice of work effort. This is something that the employer can’t fully observe and can only try to influence (as Mankiw briefly describes later on p. 448). The efficiency wage hypothesis discards the idea that employers are price takers in a competitive labour market; instead, they can choose to offer a higher wage in the hope that this will elicit greater work effort.
As Robert Prasch observed: “If the marginal product of labor schedule is at least partially determined by the wage level, then the model may no longer be said to feature a unique equilibrium solution” (2004, p. 152). As he also pointed out, the marginal productivity theory says “that a high level of marginal productivity is the fundamental cause of high wages”, but “the direction of causation runs both ways”: high marginal productivity could be caused by high wages (p. 152).
The existence of a well-defined labour demand curve depends on the firm knowing how much additional output is produced by an additional hour of work, where labour is homogeneous in the sense that it doesn’t matter which worker is providing the additional hour of work. If there is no firm link between hours of work and work effort, the simple lines in Mankiw’s Chapter 13, illustrating inputs and outputs (the production function) and outputs and costs (the total and marginal cost functions), don’t exist. Instead, the actual relationships between the variables lie in some region around them, depending on “the social relations between bosses and workers, between workers themselves or about the social environment in which firms operate. It would be like admitting that history, sociology, politics, etc. should have a say in the theory of the firm”, as Yanis Varoufakis writes (1998, p. 172).
REFERENCES
Akerlof, George, and William Dickens (1982) “The economic consequences of cognitive dissonance”, American Economic Review, 72(3): 307–19.
Borjas, George (2019) Labor Economics, International Student Edition, McGraw-Hill.
Doucouliagos, Hristos (Chris), Richard B. Freeman, Patrice Laroche, and T.D. Stanley (2018) “How Credible is Trade Union Research? 40 Years of Evidence on the Monopoly-Voice Trade-Off”, ILR Review, 71(2): 287–305.
Goodwin, Neva, (2023) Microeconomics in Context, Routledge.
Lavetti, Kurt (2023) “Compensating Wage Differentials in Labor Markets: Empirical Challenges and Applications”, Journal of Economic Perspectives, 37(3): 189-212.
Mishel, Lawrence (2012) “Unions, Inequality, and Faltering Middle-Class Wages”, Economic Policy Institute, Issue Brief #342. Washington, DC.
Prasch, Robert (2004), ‘How is labor distinct from broccoli? Some unique characteristics of labor and their importance for economic analysis and policy’, in D.P. Champlin and J.T. Knoedler (eds), The Institutionalist Tradition in Labor Economics, M.E. Sharpe, pp. 146-58.
Purse, Kevin (2003) “Work-related fatality risks and neo-classical compensating wage differentials”, Cambridge Journal of Economics, 28(4): 597–617.
Schur, Lisa and Douglas Kruse (1992) “Gender Differences in Attitudes toward Unions”, ILR Review, 46(1): 89-102.
Seccareccia, Mario (1991) “An alternative to labour-market orthodoxy: the post-Keynesian/institutionalist policy view”, Review of Political Economy, 3(1): 43-61.
Sloane, Peter, Paul Latreille and Nigel O’Leary (2013) Modern Labour Economics, Routledge.
Takala J., P. Hämäläinen , R. Sauni, C.-H. Nygård, D. Gagliardi, and S. Neupane (2023) “Global-, regional- and country-level estimates of the work-related burden of diseases and accidents in 2019”, Scandinavian Journal of Work, Environment & Health, published online November 12. https://doi.org/10.5271/sjweh.4132
Related commentaries
Birks, Stuart: “Efficiency wage”