What do Unions do?

Abstract: This paper attempts to answer the question of what unions do. Put simply, are they good or bad? Do they harm the economy? What effects do they have, on productivity, financial performance, inflation, productivity growth and investment?

All these questions are answered in detail using a number of studies and applications of various models.

I.  INTRODUCTION

The structure of this paper will be as follows; First it will outline the two faces of trade unions as first created by Freeman and Medoff. The basis of much of the literature surrounding unions has stemmed from this. Secondly a large section of the paper has been devoted to productivity. This is because it is one of the most important aspects of the effect of trade unions from the firm’s point of view and the economy as a whole. Thirdly, there will be a section on productivity growth, which is distinct from productivity. Fourthly the broader concept of financial performance will be assessed. The question of whether unions impact on investment will be answered in the next section and finally a brief section will be devoted to unions and inflation. All this while not a complete analysis of the topic should go someway towards answering the question of what unions do.

II. REVIEW OF THE TOPIC

This paper draws heavily on four main sources. The earliest is that of Freeman and Medoff entitled What Do Unions Do?. It was this study that first created the concept of the two faces of unionism. Following this is Addison and Hirsh’s The Economic Analysis of Unions. The broadest study yet of what unions do comes in Alsion Booth’s book The Economics of the Trade Union. The purpose of this book is to “impose some structure on this body of literature…to provide a selective review of the crucial framework used by economists in modelling trade unions” The goal of this paper is similar. Where it differs is that it draws on international evidence from six countries in total; Canada, U.S.A., Australia, U.K., Japan and Germany. Much of this comes from a paper by David Metcalf entitled Unions and Productivity, Financial Performance and Investment: International Evidence. It provides an interesting insight into countries such as Japan and Germany where union density is much higher than that of the U.S. Although Freeman and Medoff’s study concluded that unions were as a whole good for productivity much of the literature has shown that their results have not held up well. Unions as a whole have had mixed results regarding productivity and inflation, have negatively affected productivity growth, investment and financial performance. Though these results have themselves varied across industries and countries, as will be shown in this thesis.

III. RESULTS AND ANALYSIS

  The Two Faces of Trade Unionism

Trade unions have been defined for the last twenty years by two faces, as outlined by Freeman and Medoff in What Do Unions Do? The first face, put forward primarily by neo-classical economists is that of the Monopoly face. This side of trade unions assumes that most unions have the power to raise wages above competitive levels.  This it is argued has harmful economic effects. Freeman and Medoff break down the definition of these faces into three sections. The first section is on economic efficiency. In this, according to the monopoly face, unions work rules decrease productivity.  The next section is on the distribution of income. Here the monopoly face increases income inequality by raising the wages of highly skilled workers and they create horizontal inequities by creating differentials among comparable workers. Finally on the social nature of the organisation, union’s monopoly face breeds corrupt and non-democratic elements, and discriminate in rationing positions.

The Collective voice/ institutional face on the other hand is quite different. Under this, unions have positive effects on productivity, by reducing quit rates, inducing managers to alter methods of production and adopt more efficient policies and thus improving morale among workers. Also, unions represent a wider range of interests than just the ‘marginal worker’ that is catered for under a non-union system.  This leads the firm to choosing a better mix of compensation and personal policies for employees. For example under this system a firm would cater for senior employees and retirement packages whose voice would otherwise not be heard. In a non-union system the firm would only be inclined to cater for the young and marketable marginal worker, one who might leave because of small changes in the conditions of employment.

Regarding the distribution of income unions standard rate policies reduce inequality among organized workers in a given industry or firm. For example they can raise the wages of blue-collar workers in relation to white-collar workers. Quantitatively, the “inequality-reducing effects of unionism outweigh the inequality-increasing effects...”

The collective voice also defends unions on the political stage. It emphasis that unions are democratically elected organisations that represent the will of its members, especially those who are dis-advantaged and of lower income levels.

Unions and Productivity.

Many studies have been done on this, perhaps the most contentious side of trade unions. It has yielded mixed results, depending on the industry, the models used and other factors such as how competitive that particular industry is. The monopoly view of trade unions and productivity is that they achieve a wage gain at the expense of other parties, be it the consumers who receive higher prices if the cost is passed onto them, or non-union workers who receive lower wages.

The voice/institutional response shows that unions fact raise productivity on average. However as even Medoff notes, “the relationship is far from immutable and has notable exceptions.” (1984a) High productivity “appears to run hand in hand with good industrial relations and to be spurred by competition in the product market”, (1984a) while lower productivity under unionism exist under the opposite circumstances. Although unions may cause wages to increase, this is offset by improved labour productivity and thus profitability or employment need not be affected.

Alison Booth outlines two organisational theories of this view. The first is that with unionisation, and its higher labour costs, the management is ‘shocked’ into operating the firm more efficiently. However it must be remembered that this theory assumes that the firm was not efficient prior to unionisation. The second theory is that unionisation improves morale and co-operation and comes with positive changes in procedural arrangements. It also makes an assumption; that the firm operates in a world of imperfect information or uncertainty, and that employment relationships are frequently long-term. It points out that negotiation costs are more often lower with unions than with individuals. Also, individual arrangements may not be ‘incentive-compatible’, in that unions can hold for example, the threat of withdrawing all labour.

What are the sources of these improvements in productivity that the proponents of the collective voice outline? Addison and Hirsch give a considerable section to this question in their book. They conclude that there is little empirical evidence to support the collective voice explanation of productivity change, with the exception of grievances and quits. Regarding the former they highlight studies by Katz, Kochan and Gobeille (1983) and Ichniowoski (1984). They used data from ten paper mills and union plants at General motors to show that the number of grievances is inversely related to productivity.

Source: Automobiles graph drawn with data found in Harry Katz, Thomas Kochan, and Kenneth Gobeille, in Freeman and Medoff pp. 177

Source: Paper mills graph drawn with data found in Bernard Ichniowski, “How do Labour Relations Matter?”

What remains unanswered in these comparisons Hirsch says, is the effectiveness of union voice in reducing and arbitrating grievances relative to the unobserved grievances in non-union plants.

In relation to quits, Hirsch points to a cross-section study conducted by Mincer (1983) which shows that both unionisation and the union wage premium lead to a reduction in quits. Freeman (1980) in his analyses however suggests that it is only unionisation and not higher wages which reduces quits. However, other studies have shown the complete opposite of this. In an industry study of six cement plants, Kim Clark reports either no change or even higher quits following unionisation in half of the plants over time.

Interestingly, a lot of the studies by these authors point to a lower job satisfaction among unionised workers. Hirsch explains this as being a reliance on the voice structure where workers become more conscious of job problems and thus more willing to complain.

On the other side the monopoly face argues that unionisation reduces morale and motivation and obstructs the efficient organisation of capital and labour, since it constrains the choice set of management. It does this by enforcing restrictive practices, such as over manning rules. Also it is argued, unions adopt an adversarial rather than a co-operative approach to industrial relations therefore lowering morale and thus productivity.

 “But dammit, unions don’t let me run my plant the way I want to. Union work rules reduce my flexibility - An archetypical plant manager of an organized plant.” Freeman and Medoff (1984b) analyse this in terms of the rate at which union and non-union managements substitute nonproduction labour and capital for production labour when the relative costs of the latter change. They find that that a 10% increase in the wage of production workers relative to non-production workers is responsible for a 1.9% substitution of nonproduction for production labour in a union setting, compared with a 2.8% substitution in a non-union setting. They conclude; “all told, reductions in flexibility, while irritating to management, have only modest effects on productivity”

The evidence

Alison Booth (1995a) endeavours to assess the body of evidence on productivity for the U.S.A. Her evidence comes from Freeman and Medoff (1984), Hirsch and Addison (1986). The model, first used by Brown and Medoff in 1978, is as follows;

ln(q/n)_ij = Z_i’b₁ + X_j’b₂ + b₃ln(k/n) + b₄D_ij + b₅[ln(k/n).D_ij + m_ij  ]

where X and Z are vectors of industry and regional dummy variables respectively, the b coefficients to be estimated, and output q is proxied by value added. Subscript i represents industries and j denotes regions. In this particular study they find 22% differential between union and nonunion productivity. This, as Booth points out is remarkably high. She remarks that this is possibly because of the fact that the dependent variable is value added rather than a measure of physical output, and that the hypothesis that production function parameters differ between the union and non-union sectors is not tested. Also, a variable bias is likely to be omitted. In his later study with Freeman, Medoff yields mixed results and shows that, as a whole productivity is likely to be raised by the presence of a union.

Some of the more interesting results are as follows:

Source : Freeman and Medoff, “What Do Unions Do” pp. 166

Freeman notes “ unless the prices charged by union and nonunion firms are the same, any finding of higher value added (shipments) per worker in the organized establishments could reflect not the higher physical output per worker but rather a higher price per unit of output.”(1984c) For industries in which the market is not truly competitive one must take care in estimating a productivity effect, as there can be confusion of price with quantity. The physical measure of output can alleviate this problem but it is narrowed to only a few goods.

Most interesting here is the fact that the productivity effect can go from positive to negative over time, shown in the Bituminous coal mines. Productivity rose though in most plants, such as construction (no.5). These increases in productivity are explained by a number of factors. Clark study of six cement factories breaks them down into the following; (Freeman 1984d)

Worker responses: Turnover, Absenteeism, Discipline problems, Morale. Management responses: Plant manager (all replaced) Supervisors (all replaced)

Management Practices: Before Union, After Union.

Regarding turnover the study finds that in half of the plants turnover is down. Half the plants report and improvement in morale, and only one plant experiences increased discipline problems. Under management practices three plants before unionisation half of plants rated their management as “Authoritarian” but after unionisation half the plants reported major improvements in methods of management such as productivity targeting, periodic meetings with workers and new reporting and accounting systems. This study is all the more interesting as it uses disaggregated data rather than aggregate data.

However, the latter yield quite different results. Warren (1985), (pp196 Hirsch) using time series measurements of ‘real gross private domestic product’ of the private domestic economy finds that productivity is lowered by 0.81. More recent studies have shown the same; Clark (1984) Hirsch (1990) and Bemmels (1987) all conclude that unionisation lowers productivity, although it is worth noting that in certain industries it is associated with a positive productivity differential, usually in the private sector where there is a degree of competition.

A time series study of Canadian productivity from 1926-78 by Maki concludes that the growth in union density in the 70’s reduced the annual growth in total factor productivity by 1.7% per year and the strike activity had an effect of 1.0% per year.

David Metcalf (2002) surveys this and studies of five other countries as well; Japan, U.S.A., Australia, Germany and the U.K. His results are most interesting. For Germany, which has a high union density the results are mixed. The system of industrial relations has a dual structure. At industry level, collective agreements are negotiated between unions and employers while work councils watch the implementation and co-ordination of such agreements at workplace level. Addison (2000) surveyed 1025 manufacturing plants in Lower Saxony. Using value added per employee and the work council acting as a dummy variable, he estimated that there was a “significant positive link between work councils and labour productivity” (Metcalf 2002a) for large establishments of 101-1000 employees. The link for smaller establishments was positive but “statistically insignificant”.(Metcalf 2002a) However regarding trade unions, the results were quite different. Schnabel (1991) and Addison’s study finds a negative but statistically insignificant link and commenting on his later survey of five studies on the matter, Schnabel says, “trade union density seems to exert a negative, but quantitively small, influence on labour productivity”.

Japan’s results were mixed. Initial studies on the issue were contradictory. Brunello (1992- pp18) concludes that unions have a negative impact while Muramatsu (1984 – Metcalf 2002) and Morishima (1991- Metcalf 2002) suggest the opposite. Recent studies have produced more reliable results. They show that it is through indirect channels such as longer tenures and the presence of full-time union officials in the plant that unions raise productivity. Tachibanaki and Noda (2000)measure productivity as value added per employee and include firm size, capital/labour ratio and average job tenure as controls. They find that although union recognition is negatively associated with productivity, the union/tenure interaction has a positive effect. In unions with tenure above 15 years there is a positive effect. “such co-operative behaviour in unionised firms raises employees’ work incentives, skill formation and possibly solidarity, and their loyalty to the firm as well as mutual trust”

In 1994, Benson’s study of the Kansai region notes an important role for full-time union officials, though finding in general no significant link between recognition and productivity.  Firms with a full-time union official have slightly higher levels of productivity than nonunion firms. He says, “ Full-time local union officials are more likely to enforce agreements and contracts relating to working conditions… Their intimate knowledge of the enterprise often means that management relies heavily on them to solve disputes, to contribute to the smooth running of the organisation and to assist in the effective management of human resources.”

In Australia the presence of more than one union in a firm has proven to be detrimental to productivity. A study by Crockett (1992) shows that “the negative union effect is strongest when unionism is measured by the number of unions…. Causes inter-union competition and communication problems and may be associated with possible conflicts between different union voices” As a whole the study shows that unions have a negative effect around the region of 10%.

This has also been the case in the U.K. Metcalf concludes that most of the evidence points to a negative relationship between unions and productivity in general but a positive one in the 1980’s. In 1980 41% of establishments employing 2000 people or more had 3-5 unions present and a further 38% had 6 or more present. (2000b) Metcalf infers therefore that this is a major factor in the negative relationship with productivity as “change is likely to occur at a lower pace than in systems not suffering from these handicaps”

Union effects on productivity growth

“All right, productivity is generally higher under unionism, but that isn’t what really matters. What’s important is how unions affect the growth of productivity. Don’t unions reduce growth by opposing technological change? – An archetypical critic of unionism” (Freeman and Medoff 1984e)

Freeman and Medoff, as well as Booth, touch briefly on this important aspect of union effects. The former rightly point out that because unions that succeed in blocking technological change go out of business the critic of unions above is no more than a myth. They point out that in the 1980’s some unions even pressed management to modernise their plants with new investments. Their study shows that the negative relationship with productivity is too weak to support the claim that unions reduce dynamic efficiency. The model used is as follows;

DLn(q/n)= Dln A_it + aDln(k_it/n_it) + (1-a)(g-1)DD_it + Dv_it

Up to now we have examined the union impact on levels of labour productivity. The few studies on productivity growth tend to point to a negative or insignificant correlation, Booth points out. But she concludes that it is “rather premature to conclude that there is a conflict between the results of the productivity level and growth studies.”

Financial performance

As Freeman and Medoff rightly point out, profitability is one of the most difficult economic variables to measure. They present a set of results by Clark. (1980) The difference in profitability is calculated by the price cost margin and the quasi rents divided by capital. Here it is shown that there is a clear negative impact on profitability, even more so than that of wage and productivity effects. This is explained by Medoff as being due to the fact that profits are a relatively small component of an industry’s income flow. The possible exceptions to the rule would occur in two types of situations. The first is where cost increases brought on by unions lead the industry to charge monopoly level prices, and the second is when cost decreases. He concludes “ beneficial to organized workers, almost always; beneficial to the economy, in many ways, but harmful to the bottom line of company balance sheets: this is the paradox of American trade unionism” (1984f)

Metcalf criticizes theses measurements of profitability by saying how it is notoriously difficult to measure the value of capital, and “profits should be, conceptually measured in present rather than on an annual basis” (2002c) More recent studies have used a measure asking the following question: “ I now want to ask you how your workplace is is currently performing compared with other establishments in the same industry. How would you assess your workplace’s financial performance?” (2002c) Responses went along a 5 point scale from a ‘a lot better than average’ to ‘a lot worse than average’

Recent studies also include other factors such as percent part-time, workplace characteristics such as size and location, management characteristics like the extent of human resource management, and the nature of the product market such as the number of competitors. Bronars and Addison and Hirsch’s review of 16 studies found the same results as Freeman and Medoff. They state that there is “fairly strong and significant evidence that the total effect of higher union coverage is to reduce profitability.”

In Japan the most comprehensive study was that performed by Benson (1994) He states that union presence “reduced the probability of managers reporting a rate of return of 6% or more [and] increased the probability that these enterprises would have lower profits” They suggest that the reason for this is higher wage costs. In the U.K. nearly all the studies point to a negative association. Metcalf (1993) reports that “ Eight U.K. studies use workplaces, firms or industries to analyse the link between profitability and unionisation: all but one show a negative association” This study was conducted in the 1980’s but a more recent study by Wilkinson (2000) shows that by the end of the 1990’s there was no overall association between union presence and financial performance. This could be explained as weaker unions effects on pay balancing out the stronger ones. Also the decrease in multi-unionism is a major factor. Pencavel (2001) however, shows that multi-unionism still has adverse effects on financial performance. Also he shows that when there are 5 or few competitors, unions are associated with significantly worse financial performance but union recognition has no such impact when the market is more competitive.

In Germany the evidence seems to point in the same direction. Schedlitzki (2002) concludes that work councils have a negative impact on profitability, which is unusual, as they are not allowed to strike or bargain on wages. Addison (1996) explains that the “dual system of industrial relations in Germany by no means excludes the possibly that work councils are rent seeking agencies.” Fitzroy and Kraft (Metcalf 2002) offer another explanation; that work councils are themselves linked to various legal provisions which constrain competent managers, which would not be present if there were no councils.

Unions and Investment.

Investment is of crucial importance as capital accumulation is key to long run growth. Studies on the matter tend to analyse in terms of technology and process innovations, research and development, product improvements through design advertising and after sales service and human capital. A direct effect of unions affecting investment could occur through unions delaying the introduction of new machinery. Also inflexible work rules can infer that “investment is not used to its full capacity” (Metcalf. 2002d) Indirect effects in the rent-seeking model are explained by Metcalf. Unions can “tax” the returns on a new project therefore creating a “holdup” reducing the incentive for new investment. This is especially true if the firm is operating in a competitive environment.

The argument against this is that the union-set wage is exogenous in the traditional demand curve and the firm simply adjusts along this curve. Therefore “ a higher union wage stimulates investment because the firm substitutes away from expensive labour” (Metcalf 2002d)

The Evidence.

Research on this matter has yielded consistent results. Hirsch’s study for the U.S. (1990) finds that unions reduce investment by 20%. Half of this, Metcalf explains, was a result of rent-seeking activity and the other half from reduced profitability. Canadian studies show more or less the same outcomes. Odgers and Betts (1997) show that Unionised industries experience an 18-25% reduction in investment. In Canada, the U.S. and the U.K. the largest declines in investment came at low levels of unionisation. One explanation for this is that voice effects of a larger union presence are more effective then those industries with partial coverage. But this does not mean that a 100% density of unions completely countered the negative effect of unions, as evidenced in the U.K. by Denny and Nickel (1991,1992) 

Addison, Kraft and Wenger find that in Germany, work council presence lowers the investment rate by between a fifth and a third of those firms without work councils. (1993)Japan’s evidence is not so conclusive as there is no study on investment rates. Instead Benson (1994 ) uses labour’s share in total cost saying that it “reflects capital investment in the long-run”  He finds that unions generate greater capital intensity, and there is no evidence that they detract form returns to capital or impose restriction on its use.

Unions and Inflation

In their discussion of inflation, Hirsch and Addison point out that it is important to distinguish between direct and indirect causation. “Unions are essentially a passive force, responding to but not themselves generating inflation” (Hirsch 1984a) Excess monetary growth stimulates output in the economy and this creates an excess demand for labour which feeds into wage increases. The price changes also affect the expectations of workers in both union and non-union sectors. This at least is the orthodox view of trade unions and inflation. It also states that unions may “dampen the responsiveness of wage change to excess demand because of the time consuming nature of union negotiation procedures.” (Hirsch 1984a) Finally it accepts that the wage inflation may be held at a constant rate above the market rate.

The “counter-orthodox” view is that unions can and very often do affect the rate of change in wages independently of demand. The wage inflation rate is determined by three main factors which Hirsch discusses in depth; union power, union militancy and union spillover forces. Union power can be based on the ability of the union to extract wage gains from the employer. Hirsch quotes the theoretical model Johnston, and says that “ the rate of change in wages will be a function of the rate of change in the real wage claim, the rate of profit per worker, union ability to endure a strike, and strike costs to the firm by length of strike.” (Hirsch 1984 b) The problem he concludes, with the empirical application of this model is “that of differentiating it from the expectations augmented Philips curve.” Under union militancy he introduces the identity:

W = (1 – T)Wn +TWu

Where Wn is the average non-union wage, Wu is the union wage, and T is union density. Militancy is a more subjective term he says, and can be used to describe the aggressiveness of the trade unions. The density (T) is often employed as a measurement of union militancy, as first put forward by the economist A.G. Hynes. Other interpretations tend to focus on the union-nonunion differential, but the problem here is that this may fluctuate for a number of reasons other than militancy.

Finally, he discusses wage spill over and imitation. Union settlements, he points out can directly or indirectly spill over into the non-union sector. After discussing various models he concludes that the evidence does not support them. One explanation for this could be the small size of the union sector in the U.S., yet despite this the British evidence points in the same direction even though it has a higher percentage of union workers. As a whole he concludes that the empirical evidence for the contribution of trade unions to wage inflation appears modest.

Conclusion.

It has been the purpose of this paper to assess and quantify the mass body of evidence on the crucial question of what unions do. What must be first noted is that it was not the purpose of Freeman and Medoff’s book to prove that unions were unambiguously good. Rather, it set out to challenge the conventional view of neoclassical economists that they were inherently bad. In this regard their paper has sparked a massive interest and produced a massive amount of literature attempting to explore in depth the issues raised by them.

Regarding productivity, the results have been rather mixed. In the U.S. despite Freeman and Medoff’s claims, productivity tends to be lower with the presence of a union as is the case for Australia. Yet the opposite result was found in Germany.

Regarding productivity growth the evidence seems to overwhelmingly point to a negative relationship with unions.

When profitability is concerned, the evidence again points to a negative relationship, but Wilkinson’s study shows that there is room for further investigation.

Unions for the most part reduce investment, though Japan is is the exception.

Finally, inflation appears to be only mildly affected by trade union activity.

The evidence before us leaves us unable to answer the simple question of whether unions are good or bad. Instead the conclusion could be that they are both bad and good, that there are indeed two faces to unionism but that for the most part, the monopoly face prevails.

References

FREEMAN, R. and MEDOFF, J. ,1984. “What Do Unions Do?”

1. Pp.180
2. Pp.172
3. Pp.167
4. Pp.175
5. Pp.169
6. Pp.190

MINCER, JACOB (1983) “Union Effects: Wages, Turnover, and Job Training.”

FREEMAN, RICHARD. B ,1980. “Job Satisfaction as an Economic Variable.” ()

CLARK, KIM, “ the Impact of unionisation on productivity; a Case Study – industrial and labour relations review,” July 1980, pp. 451-69

HIRSCH, BARRY T. and ADDISON, JOHN T., 1986 “The Economic Analysis of Unions”

1. Pp. 219
2. Pp. 221

BOOTH, ALISON, 1995 “The Economics of the Trade Union”

1. Pp.195
2. Pp.199

WARREN, RONALD 1985 “The Effect of Unionisation on Labour Productivity: Some time Series Evidence.”

BEMMELS, B, 1987, “ How Unions Affect Productivity in Manufacturing Plants”

MAKI, D., 1983 “The effects of unions and strikes on the rate of growth of total factor productivity in Canada”

METCALF, DAVID, 2002 “ Unions and Productivity, Financial Performance and Investment: International evidence.”

1. Pp. 18
2. Pp. 20
3. Pp. 24
4. Pp. 33

ADDISON, J.T. 1996 “Worker Participation and Firm Performance: Evidence from Germany and Britain.”

SCHNABEL, 1991, “Trade Unions and Productivity : The German Evidence”

BRUNELLO, G., 1992, “The Effects of Unions on Firm Performance in Japanese Manufacturing”

TACHIBANKI,T. and NODA, T. , 2000 “The economic Effects of trade unions in Japan.”

BENSON, J. ,1994 “ The Economic Effects of Unionism on Japanese Manufacturing Enterprises”

CROCKETT, G. , 1992 “The Impact of Unions on Workplace Productivity in Australia”

METCALF, D, 1993, “Industrial relations and Economic Performance”

WILKINSON, D, 2000 “Collective Bargaining and Workplace Financial Performance in Britain”

PENCAVEL, J., 2001, “Worker Participation; Lessons from Worker Co-ops of the Pacific Northwest”

SCHEDLITZKI, D., 2002, “German Works Council, Employee Involvement Programs and their Impact on Establisment Profitability”

HIRSCH, B. , 1990, “Innovative Activity, Productivity Growth and Firm Performance”

ODGERS, C, and BETTS, J., 1997, “Do Unions Reduce Investment?”

DENNY, K, and NICKEL, S., 1992, “Unions and Investment in British Industry”

ADDISON,J.T., KRAFT,K. and WAGNER,J. 1996 “German Works Councils and Firm Performance”