Labour Economics - Trade Unions

Estimating union effects on wage inequality Firpo, Fortin and Lemieux (2009):

1) looks to create the equivalent of a Blinder-Oaxaca decomposition but for any distribution statistic: 2) Estimates all compositional changes together (note counterfactual here answers question: "What would the wage distribution look like in the past if unionisation rates were the same as today?" 3) allows us to explain each point in the distribution of income as a function of characteristics of the labour market. Once we can do that, we can look at how the characteristics change (compositional effects) and ho this feeds through into changes in the distribution of pay. 3) Can also identify effects - that is, the way the distribution of wages associated with a characteristic (like union membership) has changed 4) Results: Unions operated around middle of labour market in US and thus their shrinking membership reduced wages around middle of the distribution. Decline in union membership explains a proportion of growth in earnings differential between the middle and the top of the earnings distribution. Also helps to explain why wage growth recently has been at top and bottom of income distribution but not middle.

US Union wage Gap over time

All from Borjas (2015) although he references other papers) During the 1930s, union members earned about 39% more than nonunion members. Since 1970s wage gap has hovered in the 15 to 20 % range. In 2012 the union wage gap stood at 15.1 %.

***Methods to net out impact of skill differentials when estimating union wage gap*** ESTIMATING CAUSAL EFFECTS OF UNIONS ON WAGES

Approach 1: Applies sophisticated econometric techniques to estimate "selectivity corrected" estimates of the union wage gain. This method allows us to predict what a union worker would earn if he were to work in a nonunion job and vice versa. Authors: Duncan and Leigh (1980), Robinson and Tomes (1984) Results: Extremely mixed - some studies suggesting union wage gain is improbably high (greater than 50%) or ridiculously low (sometimes even suggesting that unions decrease wages. Approach 2: Use longitudinal data. Data track workers over time so that particular workers workers can be observed either entering or leaving union jobs. Union wage premium is then average change when entering or leaving a unionised job Authors: Freeman (1984), Jakubson (1991) Results: Typically report that the union wage gain is smaller than the union wage gap (10% vs 15) thus showing selection bias has important effect on the calculation of the union wage effect. Issues: The longitudinal studies view the worker's move between the union and non-union sectors as if it were a natural experiment with person being assigned randomly to various hobs - but this isn't the case!

Featherbedding

As long as contract curve is upward sloping, the unionised firm hires too many workers; that is it hires more workers than the competitive level E*. The firm and the union will then have to negotiate "make-work" or featherbedding practices to share the available tasks among the many workers. The firm may maintain a particular capital/labour ratio regardless of changes in the underlying technology. For instance, even though airlines need only two pilots to fly a particular. For instance many communities in Massachusetts require private compares to hire police officers to guide tragic around construction sites when utilities are installing a gas pipe or fixing an electric line

Monopoly Unions (From Borjas 2015)

Assume union's utility depends on wages and employment and unions want more of both. Thus unions indifference curves represented by U. NB Issue with the above assumption - if all workers had the same preferences over wages and employment, and if leadership elected democratically, the unions preferences would be identical to that of typical worker. Doubtful that all workers have same preferences! Thus how to derive unions utility function!? Union wants to max utility but constraint by firm's behaviour. Assume union dealing with profit maximising competitive firm so cannot influence price of output. In a sense, firms labour demand curve can be viewed as constraint on union behaviour. If firms cannot be induced to move off demand curve, maximisation of union utility occurs at a point like M (tangency). NB that union chooses wage and firm moves along demand curve to set the profit-maximising level of employment. NB that the more inelastic the labour demand curve (D'), the higher the wage unions can demand and the smaller the employment loss.

Other studies

Boal and Pencavel 1994: lower employment in union firms compared to non-union firms Blanchflower, Millward and Oswald (1991): estimated employment growth was 3% slower in unionised UK establishments.

Kim Clark (1984) "Unionisation and Firm Performance: The Impact on Profits, Growth and Productivity"

Data and Method: 1) Examines the impact of unionization on profitability, growth and productivity using time series data on over 900 product line businesses in the North American manufacturing sector. Results: 1) In general, it is not possible to infer changes in the rate of return on capital from information on union wage and productivity effects 2) The analysis provides clear evidence that, on average, unionised firms earn substantially lower returns than non-union firms operating in comparable technological and competitive envionments. 3) It is also evident that other dimensions of performance, particularly growth and capital-labour substitution are little affected by unionisation in this data 4) The evidence is thus consistent with a bargaining model of union-firm interaction, in which the union affects the distribution of profits but has little effect on output, or factors of production. 5) In this data, at least, efficiency effects would seem to derive form differences between firms with and without market power, and from differences between industries 6) The favorable impact of unions on productivity (if there is one), however, is not sufficiently large to compensate the firm for its larger payroll costs. 7) As a result, unionized firms have lower profits. A careful study of profits in union and nonunion firms indicates that unions reduce the rate of return to the firm's capital by 19 percent.

Supply of union jobs

Depends on 1) Costs of organising workforce 2) legal environment that permits certain types of union activities and prohibits others 3) resistance of management to collective bargaining 4) Whether firms are making excess rents

Productivity effects - Unions and The Voice Mechanism

Freeman and Medoff - 1984 - The two faces of unionism: - View 1: Unions reduce efficiency through wage demands and obstruction - View 2: Unions raise productivity through the provision of a 'collective voice' The Voice mechanism: 1) Workers encounter undesirable conditions in their employment 2) Classic market solution is entry and exit - workers leave jobs and move (only option to register dissatisfaction is to up and leave 3) Voice refers to attempts to bring actual and desired contains closer together Why collective voice: 1) Improving situations individual creates a public-good hence, unions help internalise this externality and mitigate the free-rider problem 2) Collection action offers protection from the high personal costs potentially involved in individual activism Benefits to productivity: 1) Reduce quit rates (even after controlling for effects of higher compensation packages (Freeman 1980))*** 2) Alter systems of production 3) Raise morale.

Unions, fringe benefits and total premiums

Fringe benefits from unionsised include: 1) health and life insurance, vacation, sick days, pensions 2) The value of fringe benefits is 20% in unionised firms and only 15% in nonunion firms and, because union wages are higher than nonunion wages, the package of fringe benefits is thus greater. 3) Thus total compensation eexceeds union wage effect 4) Evidence suggests the union compensation gap (the percent difference in total compensation between workers in the union and nonunion jobs, may be about 2 to 3 percentage points higher than the union wage gap

Jakubson (1991) Estimation and Testing of the Union Wage Effect Using Panel Data

Idea: 1) A Considerable body of empirical work, based on cross-section data, suggests that the union wage premium is on the order of 20% 2) This large rent for union workers may seem implausibly large, and several economists have questioned whether cross-sections studies estimate a pure union effect. 3) NB that a rational employer might respond to higher union wages by raising hiring standards leading to a positive covariance between the unmeasured productivity characteristics and the union status variable. This covariance will impact an upqard bias to cross section estiamtes of the union wage effect. Method: 1) We present estimates of the union wage effect controlling for unmeasured individual effects 2) If productivity differences are constant over time once can control for them by estaing a fixed-effects model. Results: 1) For PSID men the union wage effect is 5-9% after controlling for person effects, as opposed to 20% in cross-section. 2) There is evidence that the union effect varies among different types of workers, even after controlling for unobserved individual -specific effects. Younger workers, less educated workers, and non-whites gain more from union coverage. Issues: 1) Potential measurement error 2) Implicit assumption that union status changes are random a. Changes in union status generally require job changes, since certification and de-certification elections are infrequent events. If job changes depend on the wage, then the change in union status is endogenous and the estimates here are subject to the usual simultaneous-equations bias.

Estimating Unions, Investment & Productivity Machin and Wadhwani (1991)

Idea: 1) Aim is to test whether firms were more or less likely to invest depending on whether the firm was unionised or not. Data and Method: 1) Cross sectional survey sample of 630 UK firms about introduction of new capital (looking at 3 year time period) 2) 27% made some conventional investment and 48% made some advanced tech investment over the proceeding three years 3) Data from 1984 Workplace Industrial Relations Survey 4) Conventional technical change is the introduction of new plant, machinery or equipment not incorporating micro-electronics (excluding routine replacement 5) Advanced technical change dehined as the introduction if new plant machinery or equipment that includes the new micro-electronics technology Result: 1) There is no significant association between unionism and investment. The coefficient on UNION is positive, so there does not appear, at first sight, to be any evidence for the conventional view that unions deter investment. 2) 2) We also noted that, in terms of raw correlations, unionism is positively associated with investment. This appears to be true because unionism is associated with higher wages, greater organisational change, and formalised consultative procedures. Issues: 1) NB require that unionisation is not positively correlated with unobservable characteristics of the firm that are themselves positively correlated with investment 2) Measures of innovation 3) selection bias - unions could choose to organise at firms in growing industries where there is likely to be investment

Estimating union effects on wage inequality Fortin and Lemieux (1997) "Institutional changes and rising wage inequality: Is there a linkage?"

Idea: 1) Explore impact of decline in real value of minimum wage, decline in unionization rate and movement of economic deregulation on rising wage inequality in the united states. Results: 1) We find that deunionization had a significant effect on the rise in inequality for men but no effect for women. Not surprising due to low levels - 15% in 1979 - of unionisation 2) The only institutional factor that really matters for women is the minimum wage 3) our view, the diagrams presented in the previous section make a prima facie case that a falling real minimum wage, deunionization and deregulation have con- tributed to greater wage inequality. Issues: 1) Institutional changes are the result rather than the cause of rising wage inequality - they are endogenous

Farber (1978) "Individual Preferences and Union Wage Determination: The Case of the United Mine Workers"

Idea: 1) In order to investigate the preferences of union members and the formation of union bargaining goals, a model of union behaviour applicable to the United Mine Workers and based on maximization of the expected utility of the median-aded member of the union is developed, 2) Results shed some light on such aspects of preferences of union memebrs as 1) risk preferences, 2) relative valuation of direct wage payments and fringe benefts, 3) rate of time preference Results: 1) Preferences of UMW which are also suggestive of the preferences of a broader class of union members a. Union members found to be quite risk averse b. Discount rate found to be quite low c. Estimated that union members vlue a dollar spent on fixed fringe benefits almost 40% more than a dollar spend on discretionary income - lending support to the argument that the income tax makes non-taxable fringe benefits a more attractive method of compensation than taxable income Issues: 1) Implicit in the analysis is that there are no direct costs to the union of negotiating or imposing a price for labour

Allen (1984) "Unionized Construction Workers Are More Productive"

Idea: 1) Looks at effects of unions on productivity in construction industry Data and Method: 1) Census of Construction Industries (CCI) data , combined with estimates of percent union and labour quality variables from the Current Population Survey (CPS) was used for 27 states and regions 2) Looking at construction of single family housing 3) Controlled for capital-labour ratio, capital recentness, measurable labour quality, scale of production, industry, region, and interstate price differences Results: 1) Estimated union producitivity effect is 20% (or is it 9? Textbook says 9...)

Wessels (1994) 'Do Unionized firms hire better workers?"

Idea: 1) The standard story for unionised firms: A union raises wages and the firm foils the union by hiring better-quality workers. 2) Problem with this myopic story is that it ignores what happens in the future. It assumes that unions will sit idly by and let union firms whittle away any union wage differential by hiring better quality workers 3) I would argue that it is more likely that unions would respond to better quality workers by raising wages even more. Employer, anticipating this, may respond by hiring lower-quality workers Evidence: 1) A major body of evidence for the lower-quality result comes from comparing the union wage gains estimated by simply OLS regression with the union wage gains estimated by regression that adjust for selectivity 2) In cross-sectional studies with many observations, the estimates controlling for the endogeneity of union status most often found larger unionw age differentials than the unadjusted OLS estiamtes. 3) If these results do correctly control for selectivity then unionized employers do hire workers with lower-quality. Results: 1) The evidence suggests unionization on average does result in lower-quality workers or in no change in the quality of workers hired.

Clark (1990) "Efficient Bargains and the McDonald Solow Conjecture

Idea: 1) This article explores the McDonald and Solow ( 1981) conjecture that the outcome from a bargain over wages and employment can be ap- proximately achieved by a bargain over wages and the labor-capital ratio (the "crew size"), where the latter can be argued to be more representative of the real world. 2) many bargains in the real world are over wages and work rules. I assume that these rules influence the firm's labor-capital ratio Data: Method: 1) I derive a crew size contract curve and show that this conjecture is not generally true Results: 1) The results suggest that, although the crew size contract curve lies in- between the labor demand curve and the McDonald-Solow contract curve, a bargainover wages and the labor-capital ratio will not reap much of the efficiency gain over points on the labor demand 2) o t is then unlikely that the two contract curves are coincident. Bargaining over crew size leads to an efficient out- come in terms of the variables being bargained over, but the efficiency implied by satisfying marginal conditions in w, Z space is not the same as the efficiency discussed by McDonald and Solow.

Hrisch 2004 What Do Unions Do for Economic Performance

Idea: 1) This essay focuses on perhaps the most contentious topic in What Do Unions Do? - union effects on productivity, growth, profits and investment 2) Little literature existed at the time What Do Unions Do? Was written and, even now, the empirical evidence in this area is limited and cannot establish (or reject) causal union effects or their magnitudes nearly so conclusively as one might like Result: 1) Freeman and medoff are clearly correct that variation in union productivity effects vary substantially across workplaces 2) But their conclusion that union effects are generally positive, which I interpret to mean a nontrivial positive average effect, cannot be sustained, subseqnent evidence suggesting an average union productivity effect near zero and at most modestly positive (say, 2-5 percent). 3) Subsequent literature has continued to fin union associated with lower profits. Unlike Freemand and Medoff, several of the studies conclude that unions tax not just monopoly profits, but also the quasi-rents or long-run normal returns emanating from long-lived physical capital and R&D. 4) Lower profits and the union tax on tangible and intangible capital has led to slower employment and productivity growth among union companies, reinforcing the downward trend in priate sector unionism

Eberts and Stone (1986)

Idea: 1) This paper derives a multidimensional (hedonic) contract-curve model in which employment-security provisions are used to maintain efficient bar- gains outside the employer's demand curve and distinguishes em- pirically between the contract-curve and demand-constraint models using data for public school teachers in New York State. 2) The feature that distinguishes the contract-curve and demand-constraint models is the role played by the employment-secutirty provision. Results: 1) Estimates clearly support the contract curve model of the demand-constraint model by linking the gap between compensation and the value of the marginal product to the strength of employment-security provisions. 2) We find strong support for the contract curve model 3) The employment-related contract provisions enter the regression with signs that support only the predictions of the contract-curve model Issues: 1) Maybe weak external validity. Public sector teacher unions are about as powerful as unions get?

Layard and Nickell (1990) "Is Unemployment Lower if Unions bargain over Employment?"

Idea: 1) o We consider an economy in which all firms are unionized and bargain with their own union 2) The conventional are obtained from a partial-equilibrium analysis, in which the environment outside the representative firm is assumed to be the same whatever bargaining scheme is in operation 3) Once a general equilibrium framework is adopted, the propositions collapse, and the view that featherbedding can maintain jobs is called into question. Weakness of textbook analysis: 1) v Method: Results: 1) THIS PAPER ARGUES THAT EMPLOYMENT DOES NOT NECESSAIRLY INCREASE WHEN UNIONS BARGAIN OVER EMPLOYMENT! 2) If unions bargain over employment as well as wages, unemployment will be the same as if they bargain over wages only, provided that the production function is Cobb- Douglas. It will be lower if the elasticity of substitution between labor and capital is smaller than unity. 3) . If we start from a fully competitive labor market and then move to one of efficient bargaining (over wages and employment), unemployment rises. This is so even if the marginal utility of income is constant, so that bargaining is "strongly efficient." Explanation: 1) To some it may appear counterintuitive that bargaining over employment could be bad for employment, even though unions like employment. The reason is that, if unions can bargain over employment (and not only over wages), this gives them more power. They may thus secure higher wages 2) The effects of extra power may outweigh the employment gains from giving more expression to the unions' concerns over employment. Issues: 1) They assume a fully unionised economy! 2) Different results if economy not fully unionised?

Evidence on 'the voice mechanism' Freeman and Bryson (2006)

Idea: Comparison of UK and US demand for union voice suing workplace employee survey data 2) Demand for unions depends on workplace needs 3) Conditional on needs, unions are in greater demand when a) management has positive view of unions, b) managements has strongly negative view of unions 4) Demand for union services unmet in the US, whereas workers in UK tend to free-ride whilst enjoying voice substitutes (e.g. HR practices like open-door management systems or worker committees).

Estimating Unions, Investment & Productivity DiNardo and Lee (2004)

Issue with both previous studies - selection bias. Unions could choose to organize at highly profitable and productive firms that are more likely to grow and pay higher wages. This study: 1) Try to overcome this by exploiting rules that exist in US on how unions are allowed to organise in an establishment. 2) Rules are: to have the right to negotatite with employers at an establishment, unions in US must win majority in a vote of all the workers currently employed in the establishment. 3) Assume that establishments where the vote was only just short of a majority have very similar characteristics to those where there was only just a majority. REGRESSION DISCONTINUITY APPROACH REsults: 1) No obvious difference in productivity across the firms, nor is there any difference around to 50% cut-off 2) also find no appreciable difference in wages between whether the union just won the ecltion or just lost. These unions do not seem to be at all successful. Maybe THIS explains the no change in productivity 3) Study finds no significant effect on employment. Issues: 1) does it really capture the 'normal' effect of a union in a firm. In firms that only just got a union, the union is arguably in a very weak bargaining position relative to a situation in which they won comfortably!

McDonald and Solow (1981) "Wage Bargaining and Employment" Need to reread******

Key Points: 1) If it is impractical to specifiy the level of employment in the contract, an efficient outcome may be approxicamtely achievable if the contract specifies the number of workers per machine or some similar rule 2) Since the objective is to increase employment beyond the level given by the labour demand schedule, manning agreements or 'featherbedding; are likely to be adopted. 3) NOTE THAT THE BARGAINS ALONG THE CONTRACT CURVE ARE EFFICIENT ONLY FROM THE POINT OF VIEW OF THE EMPLOYER AND THE FIXED MEMBERSHIP OF THE UNION, UNLESS L IS AT THE COMPETITIVE LEVEL Issues: 1) Our methods, and therefore our concluions, are entirely partial equilibrium 2) Key assumption is that product-market condirions are more sensitive to the businesses cycle than the reservation wage is.

Productivity effects - Unions and Investment

Key determinant is whether unions raise or lower investment by firms. Higher wages through union bargaining directly influences the desired capital stock and hence the rate of investment. However the effect is ambiguous: 1) Higher wages raise costs, lower outputting hence reduce capital requirements 2) Higher wages encourage firms to invest more heavily in order to reduce reliance on expensive labour Hold up effect (Grout 1984): 1) If a firm invests in a new project, unionised workers may capture some of the returns on the project in the form of higher wages (since there is union bargaining) 2) this hold up effect reduces the incentives to invest in what would otherwise by productive projects 3) result hinges on ability to renegotiate wages after investment has taken place Hold up model: 1) Consider firm with profit function π = R(K,L) - wL - rK 2) Wage is determined by Nash bargain between union and firm. If union can credibly commit to no wage renegotiation, the firm faces no risk of making an investment that could be immobilised by strikes, leading to losses. 3) the firm's part of Nas bargain is just profits, as in standard RTM model above. The optimal level of capital KI is then obtained by maximising profit at a given wage: Rk(K*,L) = 4) suppose instead wage can be renegotiated after employer has chosen capital stock. 5) the bargaining wage now depends on level of capital w(K), which gives FoC for K**: Rk(K**, L) = w'(K**)L + r 6) As w'(K) is positive, underinvemstment occurs when wages can be renegotiated with unions.

Bargaining over Work Rules rather than employment

MacDonald and Solow "If it is impractical to specify the level of employment in the contract, an efficient outcome may be approximately achievable if the contract specifies the numberof workers per machine, or some other similar rule, and leaves the overall aggregate to the discretion of the employer" A wide range of bargains have occurred over working conditions, health and safety, job functions, manning requirements,and other aspects of work organization. Crew-size (Labour capital ratio) 1) A bargain over crew size allows flexibility in output in response to shocks 2) o If capital is fixed, then bargaining over w and N/K is the same as bargaining over w and N by definition. 3) o The Crew size bargain, represents an efficiency gain over the right-to-manage bar- gain, but it is generally Pareto inferior to the McDonald-Solow bargain Examples of Work Rules: 1) Restriction on the amount of materials that can be handled by a worker, or number of machines to be tended 2) Demarcation limits, where workers of one type cannot do the job of some other type of employee 3) Apparently needless repetition 4) Unnecessary extra work, such as that three coast of plaster 5) Requirements that crews be unreasonable large 6) Restrictions on adoption of new technology and modern tools 7) Rules governing a worker's maximum speed of work and use of excessive relief time

Efficient Contract Empirics

Many of the studies in this literature estimate regressions that relate the employment in union firms to the union wage and to the competitive wage in the industry. If union sbehave like monopoly unions, level of employment in firms would depend only on union wage and not on competitive wage. In contrst if union conctracts were strongly efficient, level of employment in union firm should be unrelated to union wage, but depends on competitive wage Empirics: 1) There is not strong evidence that unions and firms negotiate over employment. This conclusion is reinforcedby case studies of the docks, printing, railroad and other industries in the United States and more recent work in the United Kingdom, which have revealed that unions rarelybargainover the number of men at the place of work directly. 2) One possible reason why we don't often observe bargains over wages and employment is that firms and unions face uncertainty about future demand - firms have an incentive to report demand is lower than expected (firms private information) Despite this, borjas argues that available studies seem to indicate that wage-employment outcomes in unionised firms do not lie on the labour demand curve.

Estimating Unions, Investment & Productivity Denny and Nickell (1992) "Unions and Investment in British Industry"

Method: 1) Use longtitudnal data for sample of UK manufacturing firms between 1973 and 1985 2) Use fixed effects approach - eliminates unobservable fixed effects Data: 1) In this study we utilise two data sets. The first is a group of 72, 3 digit industries into which we merge information from the I980 and I984 Workplace Industrial Relations Surveys (WIRS) 2) The second set consists of 54, 3 digit industries (or industry groups) which have been carefully matched across the changes in the standard industrial classification in I980 and stretches from I973 to I985. Results: 1) So unions have an overall negative impact on investment which is larger in the competitive sector of the economy 2) when a firm is installing new capital, it seems that if there is a union, it is helpful if as many workersas possible are members. 3) REsults: find significant negative effects of unions on investment Issues: 1) The only wrong sign is that on the expected growth rate of technical progress, which should be positive. 2) selection bias - unions could choose to organise at firms in growing industries where there is likely to be investment

Monopoly Unionism, efficiency and Resource Allocation

Monopoly Unionism is inefficient because unions reduce total value of labour's contribution to the national income. If employers move along the demand curve as result of union-mandated wage increases, unions reduce employment in union firms and increase employment in nonunion firms. Thus wage differs between two sectors and there is allocative inefficiency. The last worker hired by the non union firm would have greater productivity if he or she had been hired in union sector

Right to Manage (RTM model) - 2 - Solution

NB - FOC is mark-up over reservation wage Interpretting Beta: 1) How quickly either side needs to resolve negotiation Comp statics: 1) Wages increase as power of union rises 2) Wages decrease with the absolute value of the two wage elasticities. More elastic labour demand (e.g. from competitive pressure) lowers markup NB: 1) as union power (beta) tends to zero, the wage tends to the competitive outcome 2) as union power (beta) tends to 1, we get the so-called monopoly union model ( see below slide

Unions, Productivity and Profits - the Evidence

NB, we shouldn't be surprised if we find unions increase productivity, after all, union wage reduces employment causes a move up labour demand curve where marginal product of labour is higher. Allen (1984) found that, in a study of the concrete industry, productivity of unionised workers is 9% higher than non unionised workers. However, the positive impact of unions on productivity is not enough to offset the larger payroll costs and thus unions have lower profits. Clark (1984) found that profits of union and nonunion firms indicate unions reduce rate of return to capital by 19%

Abowd (1989) "The Effect of Wage Bargains on the Stock Market Value of the Firm"

Points from textbook 1) Strongest evidence of vertical contract curve is given by Abowd 1989. 2) Looks at the relationship between the timing of union contracts and the value of the firm in the stock market. 3) Indicates that a $1 unexpected increase in share of rents going to union workers reduces value of firm by exactly $1. 4) This is what we expect with a vertical contract curve! Idea: 1) I develop a model for estimating the expected change in union labour costs around the time of a wage contract renegotiation, conditional on information available to both the union and the employer three months prior to the eventual settlement Results: 1) Find dollar for dollar traded 2) Also finds support for the hypothesis that collective bargains maximise the sum of shareholders' and union members' wealth. The empirical relation between the unexpected change in labour costs and the unexpected change change in common stock values is consistent with a strongly effcieitn contracting environment if the shareholds' wealth change is equal and opposite to the union members' weath change 3) The findings are inconsistent with any continuing productivity enhancing activity by the union

Costs of Strikes

Private costs: 1) Borne by the union and workers Social costs: 1) Include foregone output in economy and adverse spillover effects on other industries and reduction in annul income

Empirical Determinants of Strike Activity

Problems with estimating empirical determinants of strike activity as variables seldom observed (such as initial wage demand and bottom line wage). Empirical implications from model: - Higher initial demands increase length of strike - Unions will not make excessive wage demands in periods of high unemployment - Costs of strike in terms of forgone output and revenues are deterrent Several studies (McConnel 1989 and Card 199) have estimated union resistance curve by relating the union wage settlement to length of strike. Evidence suggests the settlement wage falls by about 2 percent after a 50-day strike and 4 % after a 100-day strike. Recent studies have shown that strikes are more likely to occur when unions are uncertain about firm's financial condition. Tracy (1987) shows likelihood of strike increases if firm has volatile stock value. A strike reduces the value of shareholders wealth by around three percent (DiNardo and Hillock 2002)

Effects of Unions on Investment - Summary

Pros: 1) Voice effects - workers in unionised workplaces become more receptive to changes in work organisation 2) Union wage mark-up induces firm to invest in more labour-saving equipment Cons: 1) Unions may inhibit change in work practices required to use investment effectively. Thus adding to cost of installation. Eg Fleet street employers in 1970s - Employers, in their attempts to overcome union opposition to new technology, committed themselves to improved pensions and redundancy 2) Appropriation of returns on durable capital 3) Inflexible work rules may be imposed by unions which reduces marginal product of capital and deters investment.

Pencavel and Catherine Hartsog (1984) "A Reconsideration of the Effects of Unionism on Relative Wages and Employment in the United States

Provide some evidence that the union wage gap is slightly countercyclical; it widens in periods of high unemployment and narrows during economic expansions.

Estimating union effects on wage inequality: Holmes and Mayhew (2015)

Same approach as above but looking at UK Results: 1) Decline in union membership has a negative effect for more of distribution and may even be stronger at bottom end. 2) Not got diagram here but effect is weaker for mid 990s- late 2000s 3) Occupation and education far more important than union density decline Issues: 1) Dont control for industry

Efficient Contracts Model - McDonald and Solow (1981)

Shows the RTM model is inefficient because employment is not included in the bargaining process. In RTM model, equilibrium is always on the labour demand curve, and this may rule out wage/employment combinations that are Pareto superior. See Diagram: 1) Labour demand curve is derived from maximum points of iso-profit curves for the firm 2) If union were a monopoly union it would pick point M on the demand curve. 3) Note however the firm could try talk the firm into moving to point Q where the union would be indifferent and firm would be better off. Similarly union could try talk firm into moving to point R.. 4) Any point between Q and R would make both firm better off 5) Suppose that highest wage firm can pay without incurring loss is wz, then the iso-profit line πz gives all the zero profit wage-employment combinations and thus provides upper bound to wage-employment combinations firms offer. 6) PZ gives all the (Pareto optimal) tangency points - contract curve 7) Important to note that contract curve lies to the right of the demand curve thus for any given wage, an efficient contract leads to more employment than would be observed with monopoly unionism. 8) This relies on unions having risk-averse preferences (so indifference curves are not flat). If Beta = 0 then indifference curves are flat NB that wage employment combinations on an upward sloping contact curve are efficient only in the sense that they exhaust all bargaining opportunities between the employer and the union They are not efficient in the allocative sense as they do not yield an optimal allocation of labour within the firm and between the union and nonunion sectors (see below)

Why have American Unionisation rates declined?

Structural (supply) factors can explain at most a third of the fall in the unionisation rates. In addition to structural shifts - demands of workers for unionisation appear to have declined. 1)There has been a significant drop in the proportion of certificiation elections won by unions - In 1955, unions won more than 66% - By early 1990s they won fewer than half

Strikes and the Hicks Paradox

The irrationality of strikes has come to be known as the Hicks paradox. Strikes are costly to both parties. The firm's profits decline; may lose customers; may diminish long-run value of brand . Size of pie to be split between firm and worker shrinks. Strikes are not pareto optimal. When the parties have reasonably good information about the costs and likely outcome of the strike, it is irrational to strike.

Unions and Wages - Empirics

Theory just that unions raise wages. Standard way of thinking about this is to estimate a union wage premium (or union wage effect as termed in discussion below. Run a regression of the form in picture. Most estimates are positive and are of the order of 5-15%. Estimates are unlikely to have a causal interpretation due to non-random selection of unionised workers. If, for example, more able workers choose to become unionised and we cannot control perfectly for ability, we will find a positive link between wages and unionisation even if union wage premium actually zero.

Unions, Investment & Productivity

Theory suggests that unions may reduce investment by firms who are worried about the hold up problem. Unions may also be resistant to change and adoption of new technology and working practices. On the other hand, training tends to be higher in unionised firms and unions may facilitate the adoption of new work practices by ensuring worker support. In general, estimated results are conflicting depending on the available data.

Unions and Wage Inequality

Unions and wage inequality: 1) Implication of Card study - effects are stronger for less-skilled than higher end. Thus unions may reduce overall wage inequality. 2) In addition, unions often try to explicitly reduce wage disparities within a firm 3) On the other hand unions also raise their members wages relative to non-members which widens wage disparities. 4) Decline in unionisation may explain widening of wage inequality we have witnessed in UK and US recently! Why do unions decrease wage inequality? 1) unions are a more homogenous group 2) Unions offer lower payoff for skills (perhaps due to pay equity considerations in bargaining

Key Points

Unions can arise and prosper only when firms earn above-normal profits (rents). In effect, unions provide an institutional mechanism through which employers share rents with workers.

Threat and Spillover Effects and their effects on union wage differentials

Unions have an impact on the wage go non-union workers. As a result calculating the wage differential between union jobs and non-union jobs does not truly measure the union wage gain (even in absence of selection biases)! Threat effects: 1) Profit maxing emplyoers in idstury have incentive to keep union out and might be willing to share some of excess rents in hope workers will not unionised. Unions have a positive impact on non-union wages. Spillover effects 1) As workers lose jobs in union firms, supply of labour increases in non-union sectors. Competitive wage falls. Estimation: Evidence on such effects is typically based on the sign of the correlation between the nonunion wage in a labour market and the rate of unionisation in that market. Sign tells us which effect dominates. Many studies suggest that unions have both effects. Extreme example of how unions affect the wages of nonunion workers is given by the provision in the Davis-Bacon Act of 1931 requiring that workers employer in federally subsidized construction projects beep paid a prevailing wage defined as the union wage

Strongly Efficient Contracts

When the shape of the unions indifference curves generate a vertical contract curve. Thus firm hires same number of workers whether unionised or not. Because employment is same regardless of which deal is struck, on the vertical contract curve, the firms output and revenue are also constant. Thus vertical contract curve essentially describes the many ways in which a fixed-size pie can be divided between union and worker. Efficient in two ways: 1) Exhaust all bargaining opportunities between employer and union 2) hire the 'right' number of workers so union does not distort the allocation of labour, and there is no deadweight loss to national economy.

Differences between Union and Non-Union workers

Why may union workers be more productive? 1) Collective bargaining agreements typically make it harder for firms to lay off workers - unionised firms will want to screen workers very carefully 2) 15 % wage premium attracts more workers and thus firm can pick most attractive from pool Wessels (1994) suggests unionised firms hire worse workers

Decision to join a union graphically

Worker initially at nonunion firm offering w*. Potential union promises wage increase to Wu (rotate budget line. Cost of union is cutback in employment. Suppose that the firm's demand curve for labour is downward sloping and elastic, firm responds to union wage by reducing workers hours each week to say h0. Therefore worker is worse off. However if firm's demand curve for labour is inelastic, employment reduction is small and union offers wage employment combination at point P1, and worker better off

Strikes and Asymmetric Information

Workers may not be well informed about firm's financial status and may have unreasonably optimistic expectations about size of pie and how much firm will be willing to give away. Strike 'teachers workers a lesson.' See diagram: 1) Workers, in their uninformed state, make initial demand of w0 - this may be optimal for union because asymmetry encourages firm to lie about financial condition 2) The occurrence and duration of a strike signal to the union that perhaps the firm is not as profitable as the union thought it was and thus moderates demands (and strikes are hard for members) 3) Eventually union demands fall to mwin 4) Firm knows union will moderate demands, and thus compares present value of profits if it gives in to initial demand with present value of profit if strike lasts one period, two periods and so on. 5) The firm then chooses the strike duration that maximizes the present value of profits. 6) Firm moves to lowest possible isoprofit curve and maximises present value of profits by choosing tangency point between isoprofit curve and union resistance curve. Issues with model: 1) Assumes no bargaining over employment? 2) Unions surely know and understand this model

Demand for union services

Workers rationally join a union if the package of benefits that are offered exceed the costs. Depends on 1) Union wage differentials 2) Non-wage benefits (e.g. unemployment insurance in Belgium) 3) Existence of substitutes for union services (e.g. minimum wages) 4) Cost of membership 5) extent of employment loss 6) Macroeconomic conditions The interaction of these factors with 'supply' helps explain differing levels of unionisation

Right to Manage (RTM model) - 1

* Unions and firms bargain over wages * Firms choose the employment level that maximises profits, given the wage that has been bargained over Assumptions: 1) Both parked have fall back option if no agreement made (for firms its zero profits, for union members its their reservation wage) 2) We assume unions are risk neutral. Surplus for firms (see picture) = profit Surplus for union is surplus per worker multiple by number employed. NB unions care about employment even though they only bargain over wages

Efficients Contracts model

* Unions and firms bargain over wages and employment

Important aspects of union bargaining:

1) Centralisation vs Decentralisation 2) Degree of co-ordination 3) Scope of union negotiations

Estimating union effects on wage inequality DiNardo, Fortin and Lemieux (1996):

1) Generate a counterfactual wage distribution that answers the question "What would the wage distribution look like today if unionisation rates were the same as in the past?" 2) Takes the form of reweighing the wage data at the end of the period 3) Result: the study finds around 20% of the widening in male wage inequality in the US during the 1980s could be attributed to fall in union density - but not for women 4) Issue: reweighing approach only looks at one compositional effect at a time

Ways of assessing the power and importance of unions in the labour market

1) Membership (doesn't take into account size of labour market) 2) Density 3) Coverage

Determinants of Union density

1) Political effectiveness of union movements 2) Political parties etc 3) Benefits of unions

Trends in Union density in UK

Discussion of decline from Bryson and Freeman (2006) 1) Under Prime Ministers Thatcher and Major the Conservatives 'set about dismantling the props to joint regulation and restricting the power of trade unions' 2) Legislation limited the condtions under which unions could legally engage in industrial action, and made unions and their officials financially liable for unlawful actions. 3) Limited collective bargaining 4) The change in public policy undermined the perceptions that unions were the ocially desired form of worker representation in dealing with employers 5) In early 1990s, government outlawed the closed shop which required workers to join a TU as a condition of employment. a. Thus union density declined 6) Between 1984 and 1998, the proportion of workpalces with 25 of more employees with recognized unions dropped from 67% to 42%

Right to Manage (RTM model) - 4 - grapiccally

Employment always lies on the labour demand curve.

Trends in Union Density across OECD

Fraction of Irish workers unionised dropped from 53% to 35% between 1970 and 2003 whereas French unionisation dropped from 22 to 8 percent. Less pronounced in Italy where unioniszation rate dropped from 37 to 34 percent in that period In Sweden unionisation rose form 68 to 78 percent between 1970 and 2003

Does having a voice decrease satisfaction?

Hersch and Stone (1990) show that union members report being less satisfied with their jobs than nonunion members. this may be thought of as to contradict the 'exit-voice' hypothesis because unions should remedy worker grievances! But in order for unions to be effective, workers' voices must be heard 'loud and clear'. Therefore a by-product of unionisation might well be the politicisation of the workforce which results in union members expressing less job satisfaction. NB this job satisfaction is not genuine as it does not lead to more quits.

Oswald (1992) "Efficient contracts are on the labour demand curve: Theory and facts"

Idea and Theory 1) The purpose of this paper is to develop a seniority model of the trade union which blends elements from both existing theories. 2) The seniority model developed in later sections begins from the realistic assumption (even the briefest examination of union contracts make it clear that lay-offs are almost never decided randomly.In practice, workers are dismissed according to their length of service with firm' of lay-offs by seniority rather than the usual one of lay-offs by random draw. 3) Senior workers know themselves to be insulated from all but the most immense falls in product demand. Under majority voting, therefore, the trade union will act as though it were locally indifferent to the level of employment. Senior workers, who can out-vote any small group of juniors, will wish simply to raise wages. 4) This means that around the equilibrium, the union will be content to let firm set employment unilaterally - efficient contracts will then lie on labour demand curve 5) Key point of model, union indifference curves are horizontal for points past the median voter's seniority position because the median person can always be employed and therefore only cares about rate of pay 6) This appears to offer one solution to the contradic- tion - noted years ago - between the presumed efficiency of labour contracts and the empirical lack of employment negotiation. Empirics: 1) The British Institute of Management publication, Smith (1974), revealed that only 14 percent of firms employing less than one hundred workers used 'last in, first out', but that between 60 percent and 70 percent of companies employing more than one hundred people used the criterion of seniority. Result: 1) The model predicts that (i) efficient contracts will lie on the labour demand curve, (ii) there will be negotiations over pay but not employment, (iii) unions will have no direct interest in the elasticity of labour demand, (iv) in certain circumstances there will be extreme wage rigidity, (v) exceptional recessions will produce concession bargaining, and (vi) there will be no 'shrinking union' problem. 2) Only a starling demand shock could pull the efficient contract off the labour demand curve Issues: 1) Econometric evidence is against the seniority/flat-indifference curve model 2) What if senior workers are at risk of law-off or earnings reduction? Counter: The justification for the seniority model must be that it is closer to the truth than conventional random-draw union theories, not that it is a perfect mirror of reality. 3) Involuntary lay-offs are sometimes made by criteria like age and productivity rather than by seniority. Voluntary redundancies are also common. Counter: Thee key features of the seniority model depend only on the assumption that there is some known lay-off ordering. That lay-off ordering could be on the basis of age, productivity or even hair colour. 4) • Common observation shows that unions worry about employment losses

Farber and Western (2001) Accounting for the Decline of Unions in the Private Sector, 1973 - 1998

Idea: 1) There has been a long decline in private sector unionism over previous 50 years. 2) Present an accounting framework that decomposes the sharp decline in private sectore union membership rate into components due to a. Differential growth rates in employment between union and non-union sectors b. Changes in the union new organisation rate, specifically the success of unions in organizing new members through certification elections Data: 1) data on the 25 years from 1973 using data from the current population survey Result: 1) Find that most of decline in union membership is due to differential employment growth rates and changes in union organizing activity have had relative little effect 2) The decline in the private sector union membership rate was due primarily to changes in the economic environment that made union representation of less value to workers or more costly to employers a. Increased global competitiveness and mobility of capital (meaning that owners of capital unlikely to pay union wage premium likely to be important factors

Card (1996) The Effect of Unions on the structure of Wages: A Longitudinal Analysis

Idea: 1) This paper studies the effects f unions on the structure of wages, using an estimation technique that explicility accounts for misclassificiation errors in reported union status, and potentnial correlations between union status and unobserved productivity Methods: 1) Use longitudinal data 2) Basic approach is to look at the same workers over time and estimate the wage effect from switching between a unionised and non-unionised job (controls for fixed effects) 3) Fixed effects regression 4) Militate against measurement error problem - asks both employees and employers about union status of worker Data: 1) Model estimated for 5 skill groups using a large panel data set formed from US Current Population Survey Results: 1) Results suggests that unions raise wages more for workers with lower levels of observed skills 2) Among workers with lower levels of observed skill, unionized workers are positively selected, whereas union workers are negatively selected from among those with higher levels of observed skill 3) Among workers with lower levels of observable skills, union memebrs are positively selected, leading to a positive bias in OLS union wage gap 4) Among workers with higher levels of observable skills, union members are netagtively selected, leading to a negative bias in the OLS wage gap 5) Even correcting for these selection biases, the union wage effect is bigger for workers with lower levels of observed skill 6) Longitudinal estimated overall premium is 17%. Across skill groups, the differences in premium are much smaller and positive for all groups. This tells us standard OLS has positive section bias at bottom end of income distribution and negative section bias at the top Other interesting conclusions: 1) the 10.8% decline in male unionisation accosts for a third of the increase in 90-50 gap between 1979 and 1988 2) By contrast, it has a negative effect on the 50-10 gap. 3) In other words, the de-unionization story was going the wrong way at the low end during the 1980 4) The results from DiNardo et al. (1996) thus imply that the negative effect of de- unionization on low-end inequality was more than offset by the large effect of the declining minimum wage over this period. Issues: 1) Longtitudinal estimators are highly sensitive to measurement error o Coutner: This paper uses a lontidunial estimator that explicitly accounts for misclassification errors in reported union status

Union wage gain vs Union wage gap

Typically we only observe one of the two wages needed for the union wage gain. As a result, we instead calculate a very different sort of union-nonunion wage differential, the union wage gap. which is just the percent differential between union and nonunion jobs. Why does union wage gap not measure the union wage gain? *Because union workers will be qualitatively different from non union workers