What is the link between unionisation and inequality?

I've looked at graphs like this many times over the past few years:

One of the things that I notice in that chart is that countries that are reputed to have strong union movements have market inequality outcomes that don't appear to be all that much better than those with weaker union movements.

So this post offers some cross-country evidence about the empirical link between unionisation and inequality. All data are taken from the OECD. The inequality measure is the Gini coefficient, which means that this post has nothing to say about 'top end' income concentration. (See here for why I think the distinction is important.)

Here are union densities (percentage of workers who are members of a union) plotted against Gini coefficients:

You can see a negative relationship between union membership and inequality, but there are a couple of reasons why you'd be skeptical about drawing a line through that scatter plot. The first is that Iceland would play a larger role that one would think it should. Secondly, what is going on with France and a union density that is less than the US? You hear a lot about the weakening power of unions in the US, but not really so much in France.

It turns out that while union density is low in France, the share of workers covered by collective agreements is high. It seems to me that union coverage - the share of workers who are covered by collective agreements - is a better measure of union influence than union density:

I don't see much of a link between union coverage and inequality there.

Those two graphs are in levels, but the story of the last 20 years or so has been about changes in inequality. If higher unionisation doesn't necessarily mean lower levels of inequality, may it leads to smaller increases in inequality. Unfortunately, the OECD only have long time series data for Gini coefficients for only a handful of coutries.

(The 1990 numbers for Finland and Norway are the averages for 1986 and 1995.) With a couple of exceptions - the Netherlands and New Zealand - the patterns of changes are roughly similar. Here are the *changes* in the Gini coeffiencts since 1990 plotted against *levels* of unionisation in 1990:

Looking at those two charts, it's hard to tell a story in which a strong union presence significantly attenuated increases in inequality.

At this point, it's worth asking why we would expect a stronger union presence to be associated with more equal market outcomes in the first place. I don't think there's much doubt that unions would be a force for equality *within* a given occupation. But it's less obvious that increased unionisation would reduce the earnings gaps between different occupations.

There's a 2004 NBER paper by David Card, Thomas Lemieux and Craig Riddell that looks at the relationship between changes in unionisation and changes in inequality in Canada, the US and the UK. They find that the 'within-group' effect dominates the 'between-group' effect for men, but not for women. This is broadly consistent with the OECD data on changes in unionisation and changes in equality in those three countries: union densities and coverage fell and Gini coefficients increased. 

Union densities fell in the countries in this graph, and with exceptions of the Netherlands and New Zealand (in which union density fell by some 40 percentage points), inequality increased:

But things are less clear when you plot changes in union coverage against changes in the Gini coefficient:

Increases in inequality were accompanied by declines in union coverage in Canada, the US, the UK and some other countries. But similar increases in inequality were accompanied by increased union coverage in Sweden, Norway and Finland.

I don't really see much of a link - either in theory or in the data - between unionisation rates and increases in inequality across occupations. (Inequality within an occupation is a different story.) 

 

 

 

Comments

[Nick Rowe]

Classic Stephen post: "Well, let's have a look at the data!" Yep.

One minor ambiguity in the post: in the first picture you show Ginis both for market income and for disposable income. In all your subsequent pictures, you don't say which of the two Ginis it is. I think it's probably market income, right?

Which leads to one (slightly muddled) thought: the theory of tax incidence says that an exogenous change in tax/transfer policy that reduced inequality of disposable income might lead to an increase in inequality of market income (especially with migration). Union power might be positively correlated with tax/transfer policies that reduce inequality (lefty countries will tend to have both high union power and egalitarian tax/transfer policies). So this would tend to bias the graphs towards making unions appear to create more inequality of market incomes.

I'm not quite sure how to test this. Maybe if high union power were not correlated with low inequality of disposable income, the theory I have sketched here would fail?

[Determinant]

But similar increases in inequality were accompanied by increased union coverage in Sweden, Norway and Finland.

There is an apple-and-oranges comparison in your post, and it's right there in the section I put in italics.

You are assuming that unions perform the same function in all countries, and that's not true. Your experience, Stephen, is obviously based on your Canadian experience in which unions are bargaining agents for workers over terms of work from employers.

But unions in Finland, Sweden and Norway provide a different function. These countries use the Ghent System for the provision of social welfare. The Welfare State in these countries is two-tiered. Tier 1, the basic, actually not generous at all level is provided directly by the State. Tier 2, the very generous level is provided by unions which you have to join voluntarily, though everyone does, which is why unionization rates in these countries are so high. The unions get government subsides for their welfare activities so they are quasi-state agents.

It is possible that unions simply haven't emphasized their bargaining agent function in the Nordic countries; instead concentrating on their welfare state function. Employers could still embrace unions for the welfare provision in such a case.

It's very much a product of the local rules of the game.

[Stephen Gordon]

Okay, but why did those countries with high union densities see similar increases in inequality?

[Bob Smith]

Great post.

The obvious answer is that in a competitive market where goods and capital are mobile, you wouldn't expect unions to be able to increase wages of their members, therefore decrease inequality (at least to the extent their members on in the bottom half of the income distribution - I'll come back to this). The old story was that unions existed in industries where employers had the ability to collect some sort of economic rents (owing to trade barriers, economies of scale, etc. which kept outsiders out - think the auto industry, steel, heavy industry and, of course, the public sector). In that context, union were able to extract a portion of those rents on behalf of their members. But, lower trade barriers and introduce competition from other countries and destroy those economic rents (well, other than in the public sector), and you'd expect that the best a union could negotiate would be competitive wages (i.e., w=mpl). Unless you can tell some story about unionization increasing productivity, there would be no reason to expect a shift in unionization rates to impact wages and therefore inequality in the longer run.

The other side of the inequality story is, who is unionized? If one believes that unions can improve the wages of their members, they're only going to reduce inequality if you believe that unions represent the poor oppressed proletariat, one-step removed from a Dickensian sweatshop. But, that's hardly an accurate representation of the modern labour movement (despite the self-serving labour day rhetoric of its leaders). One can argue whether, say, Ontarians should be paying teachers, police officers, power workers and sociology professors 6-figure pay packets (particularly as there seem to be no shortage of people willing to do some of those jobs - but I digress), but since people making $100K a year are (almost) in the top 5% of income earners, if union's increase the income of their members, in that context, they might INCREASE inequality, rather than decrease it (moreover, not only do they potentially increase market inequality, but to the extent that scarce public funds are required to pay their wages - rather than, say, provide benefits to the poor - they potentially increase after-tax/transfer inequality). No doubt the reason that unions have jumped on the 1% bandwagon is that if they were targeting the top 10% of income earners, a fair number of their members would tell them to get stuffed.

[genauer]

I think you have 3 significant problems here:

a) what counts for people is their disposable income
b) looking at 2010 is for most European countries, you mostly plot here, is BEFORE the necessary corrections
c) little outliers are overemphazised (how about a bubble plot)

Plotting ca 1995 - exactly 2012, I estimate, would give you a picture, I see as "reality"

[Stephen Gordon]

a) I agree, but unions bargain over market incomes.
b) What necessary corrections?
c) If I were to actually draw a regression line, I would weight for population. But there are too few countries to justify that sort of exercise.

[Bob Smith]

a) "I agree, but unions bargain over market incomes."

I suppose you might be able to tell a story about stronger unions exercising political influence in a way to reduce after-tax inequality. In that story, unions affect after-tax/transfer income, not market wages. That powerful labour unions can have political influence is certainly plausible (look no further than the success of the Ontario Teacher's.. err... Liberal... Party). The notion that they use that influence to affect after-tax/transfer income of society at large, rather than focusing on the market wages of their members, however, s questionable. In any event, I think there'd be a correlation issue with that story - do strong unions result in redistributive tax/social policy, or are governments/societies that support redistributive tax/social policy more receptive to labour unions?

[genauer]

Stephen,

this is mainly about the classical divide between mainland europe and the anglophone world. No wonder that my comments look erratic to you.

To put it in other words, we dont have a gerneral minimum wage here, and strongly believe that is better this way.

To the points:

a) intelligent unions haggle about a lot more than just a top line number
and the metal worker union here is 5 million feet strong and rising
b) dependent, where you take the starting point in time, and will take the end point in time in a few years, those plots are looking drastically different, and those are NOT "necessary corrections" you nhave to make in your analysis
c) I fully agree that any "regression" however weighted, is pretty arbitrary

I am actually reading up on Wilhelm Röpke.

[Guest]

the first 2 graphs (coverage & density v. inequality) use different scales

is the Gini coefficient the correct comparator?
seems sloppy
http://en.wikipedia.org/wiki/Gini_coefficient

[genauer]

All the social issues we are discussing here (Germany, and central Europe), do not show up in this Gini coefficient, or if, mostly the other way.

Changing tax code to tax the upper higher makes "income" showing up that wasn't there before, looking more unequal.

Requiring lower "labor" income folks to file for tax now, redefining "household", adds to the population showing up in statistics at the lower end, looking more unequal, but actually taxing capital income higher.

Canada is in some unique one-neighbor situation. Most central european states have multiple (like 10) borders, some to higher income, some to lower income. What actually goes on with wage levels is not properly reflected in national statistics.

I am sorry, I am tired, and I can not do any better tonight, but just with the sketchy arguments above, to say, nobody here discuses things in terms like Gini.

"Union power" comes from sustained low unemployment numbers,

and we will see a lot more of that now here in Germany, and the hair cutters do it already, raising wages by > 30%, despite nearly non-existing official union membership.

[Determinant]

It is possible that unions simply haven't emphasized their bargaining agent function in the Nordic countries; instead concentrating on their welfare state function. Employers could still embrace unions for the welfare provision in such a case.

is my answer to this:

Okay, but why did those countries with high union densities see similar increases in inequality?

If the decrease in market bargaining power/position of wages is greater than the welfare role of Nordic unions, inequality will increase.

If you are working and healthy in a Ghent System country your are collecting few benefits from the Ghent System; it may be possible your union has chosen to concentrate on its social welfare provision role and has let your market position deteriorate. This would show up as an increase in inequality.

[Mark A. Sadowski]

Unions, Norms, and the Rise in U.S. Wage Inequality
Bruce Westerna and Jake Rosenfeld
March 2011

Abstract:
"From 1973 to 2007, private sector union membership in the United States declined from 34 to 8 percent for men and from 16 to 6 percent for women. During this period, inequality in hourly wages increased by over 40 percent. We report a decomposition, relating rising inequality to the union wage distribution’s shrinking weight. We argue that unions helped institutionalize norms of equity, reducing the dispersion of nonunion wages in highly unionized regions and industries. Accounting for unions’ effect on union and nonunion wages suggests that the decline of organized labor explains a fifth to a third of the growth in inequality—an effect comparable to the growing stratification of wages by education."

http://www.wjh.harvard.edu/soc/faculty/western/pdfs/Unions_Norms_and_Wage_Inequality.pdf

What's interesting about it is it provides a decomposition of the relative importance of unions and education in determining the change in inequality of compensation between 1973 and 2011 on Table 2. As you can see the decline in unions are responsible for 33.9% of the increase among US men over this period and 20.4% among women. Education is resonsible for 41.3% of the increase in men and 43.9% of the increase among women.

Since women are 47.0% of the employed according to the household survey a simple weighted average yields 27.6% of the increase in inequality of compensation is due to the decline in unions and 42.5% due to education.

[The Iron Robot]

Determinant: I've read that the Ghent system (historically used in Sweden, Denmark and Finland but not Norway) has been weakened in recent years, as union membership is no longer a condition for membership in a union-administered unemployment fund in some of the places that use it. Since 1992, Finland has also had a non-union-affiliated unemployment fund that people can join instead. So my expectation would be that unions would devote more attention to bargaining than they have in the past, because welfare provision is no longer a means of attracting or retaining members.

I think the increasing decentralization of bargaining (which used to be highly centralized and - AIUI - conducted with the goal of eventually equalizing wages across industrial sectors) over the last few decades is the reason why market income inequality increased even in the Nordic countries in which collective agreement coverage increased.

[Mark A. Sadowski]

On the other hand I've looked into the near universal decline in the labor share of income in the advanced world during the age of disinflation and have concluded that union coverage rates cannot explain it.

Peak Core CPI Rate*, Peak and Recent Labor Share of Income (Total Economy) (*Except Portugal)

Nation------CPI-Year-- Peak-Year-Recent-Year-Change
US----------12.4-1980-69.6---1980-63.7---2010--5.9
Japan-------20.2-1974-72.5---1977-56.6---2009-14.9
Germany------6.8-1974-76.3---1974-68.5---2011--7.8
UK----------22.1-1975-75.6---1975-71.3---2010--4.3
France------12.7-1980-79.2---1981-68.4---2010-10.8
Italy-------22.3-1980-83.4---1971-68.1---2010-15.3
Spain-------26.4-1977-76.4---1977-59.9---2011-16.5
Canada------11.1-1980-68.1---1971-59.8---2008--8.3
Australia---12.8-1977-75.5---1975-61.3---2006-14.2
Neth.-------10.5-1975-77.1---1975-68.5---2010--8.6
Sweden------12.5-1980-77.9---1978-63.3---2011-14.6
Switzerland--8.9-1974-67.5---2002-65.9---2010--1.6
Austria-----11.1-1981-98.5---1978-66.3---2011-32.2
Norway------12.2-1981-73.7---1977-55.4---2011-18.3
Portugal*---33.1-1977-83.9---1975-66.4---2010-17.5
Denmark-----10.6-1978-73.7---1980-69.5---2011--4.2
Finland-----17.5-1975-76.9---1991-66.1---2011-10.8
Ireland-----21.2-1981-79.3---1980-60.9---2010-18.4
New Zeal.—--17.2-1982-60.7---1975-49.0---2006-11.7

The international labor share data comes from the OECD and is not consistent with BEA data:

http://stats.oecd.org/Index.aspx?queryname=345&querytype=view

Select *Total Economy*.

The biggest problem with the theory that the decline of unions are responsible for the decline in the labor share of national income is that the decline in labor share of national income was essentially universal throughout the advanced world, and yet union coverage rates have not declined everywhere. See Figure 10 on page 14:

http://www.cepr.net/documents/publications/unions-oecd-2011-11.pdf

Note that union coverage rates increased from 1980 to 2007 in Austria, Finland, France, Norway, Spain and Sweden and if you look at the table I posted above the labor share of income declined by 10.8 to 18.3 points in those countries. In fact I've regressed the change in labor income share for almost all of these same OECD members (all but Austria) against the change in union coverage rates, and the R-squared value is 0.0038 meaning that only 0.38% of the change in labor share of national income can be explained by the change in union coverage rates.

The history of U.S. unionization rates doesn't match up with the history of changes in labor share of national income very well either.

For example the peak year for unionization in the US was 1954 when 34.8% of all wage and salary workers were in a union (see Table A-1):

http://digitalcommons.ilr.cornell.edu/cgi/viewcontent.cgi?article=1176&context=key_workplace

At that time the labor share of National Income was only 61.8%:

http://research.stlouisfed.org/fred2/graph/?graph_id=98737&category_id=0

By the time labor share of National Income peaked at 67.7% in 1980 union membership had already declined to 22.3%. By 2003 union membership had dropped to 12.4% and yet the labor share of National Income last year was 62.1%, or higher than when union membership was at its peak.

But labor share of income is probably more related to income concentration (what you call "top end").

Union coverage rates may still be an important factor in explaining changes in the *distribution of labor income* (which is related to what you call "first order").

[Camille Roy]

"why did those countries with high union densities see similar increases in inequality?"

Your mistake is your focus on union density as the factor influencing inequality. Other factors to consider: financialization, globalization. This paper looks at all these and attempts to judge their importance :
http://www.ipe-berlin.org/fileadmin/downloads/working_paper/ipe_working_paper_17.pdf

Financialization and globalization reduce returns to labor despite the degree of unionization.

[Bob Smith]

"Your mistake is your focus on union density as the factor influencing inequality. Other factors to consider: financialization, globalization."

But Stephen isn't the one (I think) making the claim that unions reduce inequality. That's a claim made by others (see, for example, www.policyalternatives.ca/publications/commentary/why-unions-matter). I think he's questioning (rightly, in my view) the merits of that claim.

[Mandos]

This is a very strange discussion fueled by an odd exclusion, which is based in Stephen's distinction between "first order" and "top end" inequality. Most of the complaints (and social and political problems) that are attributed to high inequality can be traced to the growth and behaviour of the top end, and the merit of unionization is mostly concentrated on clawing back the gains of the rentier class/1%. So it wouldn't surprise me if unionization wasn't that correlated with smaller gap between the lower end of the working class and the higher end of the working class, that now calls itself the "middle class" although it is actually not much less disposable.

So what? I mean, that point of the CCPA article that Bob Smith cites is exactly those issues that accrue to the conflicts between the "ownership class" and those who are engaged in wage labour for basic sustenance and/or some luxuries.

[Mandos]

In fact the whole thing really exemplifies the relationship between economism and the left wing---strawmanning. An argument from the economism side is made that recasts a left-wing argument into a the sort of intellectually constricted framework that economists wish to use. Then the argument is demolished with the finding of non-correlation between things that aren't actually part of the current left wing political claim.

[genauer]

Nice, that Mark A Sadowski has shown up! He knows and understands a lot more than most tenured folks in the US, despite us having some disagreements in the past : - )

I have wondered a little bit, how I put my view in the shortest possible, but clear way:

1. In former times in those Gini coefficient plots, Germany showed up on the one end, as part of the more equal social democratic, north/central European block, and the US to the other end. Now suddenly they are next to each other in the center, in the first plot in this thread. Germany has not become more unequal, and the US certainly not more equal. These numbers are obviously very dependent on who calculates them how.

2. If you look at net wages / income, from the median/ center perspective, you get a surprisingly simple picture, the 10th percentile has half the median, the 90th percentile twice, and that is very similar in the 2 countries I also know details from living there, the US and DE (Germany).
For the US look at epi.org, for DE for the SOEP data (DIW) I try to get specific links later.

3. this factor 2 picture (10/50/90 percentile) is actually extremely stable and common. Germany 20, 50 , 100 years ago, US now / 50 / 100 years ago, and even ancient Rome 2000 years ago (only 5% tax , aaaawh) , AS GOOD AS WE KNOW it.

4. For Germany at least, this is not because this is some natural law, but because we make it so, with taxes and social benefits.

Some more, after I listen to a Bank of England, Regulatory Prudent Valuation presentation : - )

[Bob Smith]

"Most of the complaints (and social and political problems) that are attributed to high inequality can be traced to the growth and behaviour of the top end, and the merit of unionization is mostly concentrated on clawing back the gains of the rentier class/1%."

Which would be a nice story, if there were any evidence that unions achieve that goal. Stephen's point is that it isn't obvious that they do. My point is that it isn't obvious why we would expect them to achieve that goal in a competitive market (since unions can only extract economic rents from the "rentier class" where there are economic rents to be extracted). Is there any evidence that rates of union coverage are correlated with lower incomes of the 1% on a country by country basis?

More to the point, as I noted earlier, the story about conflict between the "ownership" class and those "engaged in wage labour for basic sustenance", while fun to tell, isn't consistent with what is actually happening. Generally, unions don't represent (and never have represented) people "engaged in wage labour for basic sustenance". Indeed, they have often pursue policies that are directly contrary to the interests of such people (think of the opposition to labour reform in France, which makes the young and the poor all but unemployable, or the AFL's historical opposition to immigrants to the US, which played a key role in restricting immigration to the US in the inter-war years, or more recently the efforts of Swedish labour unions to restrict immigration). And as I think Stephen (or Mike or Kevin Milligan, I don't recall), in contrast to 50 years ago, at least in Canada, the incomes of the 1% is largely employment (or self-employment) income, rather than income from capital (since the 1% largely consists of management, small business owners, and professionals, such as doctors and dentists).

[Mandos]

More to the point, as I noted earlier, the story about conflict between the "ownership" class and those "engaged in wage labour for basic sustenance", while fun to tell, isn't consistent with what is actually happening. Generally, unions don't represent (and never have represented) people "engaged in wage labour for basic sustenance". Indeed, they have often pursue policies that are directly contrary to the interests of such people (think of the opposition to labour reform in France, which makes the young and the poor all but unemployable, or the AFL's historical opposition to immigrants to the US, which played a key role in restricting immigration to the US in the inter-war years, or more recently the efforts of Swedish labour unions to restrict immigration). And as I think Stephen (or Mike or Kevin Milligan, I don't recall), in contrast to 50 years ago, at least in Canada, the incomes of the 1% is largely employment (or self-employment) income, rather than income from capital (since the 1% largely consists of management, small business owners, and professionals, such as doctors and dentists).

This is the fundamental ideological difference that is exactly what I am talking about. The efforts of Swedish labour unions to restrict immigration are to maintain the status of those who are engaged in wage labour for basic sustenance within their own polities. The (deliberate and openly admitted) creation of unemployment in Southern Europe is to create a pressure for "labour reform" so that in the future, the rentier class---who are probably a lot less than 1%---can pay workers less. I don't consider bare subsistence to be "basic sustenance"---I consider relatively equal access to what we consider the good life, income security, health care, etc to be "basic sustenance". Which was the point of the CCPA article.

[Mandos]

Yet again, it all connects to the free trade ideology. It's the missing link. Two thing have happened that are the major components of the left-wing claim:

1. internationalization of trade and capital that allows *external pressure* to be placed upon workers in high-union countries. Hence the rising Gini in high labour political power countries since 1990.

2. (related) large-scale attacks on labour's power through "labour market reforms".

Especially in the absence of a longer time series (ideally going back to 1950 or longer), it easily explains the changes in Gini---which in any case doesn't show top-end inequality, or the increased stress on workers, reduced access to "the good life" and so on that come from the long-term trend of de-unionization. But it's not surprising since a correlation between Gini and some function of union membership does not a very robust model of reality make.

[Bob Smith]

"1. internationalization of trade and capital that allows *external pressure* to be placed upon workers in high-union countries. Hence the rising Gini in high labour political power countries since 1990.

2. (related) large-scale attacks on labour's power through "labour market reforms". "

I don't disagree that free trade has undermined the ability of business to extract rents from consumers (and therefore of unions to split those rents with their members) - I just don't think that's a bad thing. But that just reinforces Stephen's thesis that labour unions don't have any impact on income inequality. Moreover, your second point is inconsistent with Stephen's (and Mark's) observation that increases in income inequality (or labour share of income) seem to have little or no relation to the strength (or lack thereof) of national labour movements. As Stephen noted, that's a story that labour unions tell in the US, but it's a hard story to tell about France, which has worse levels of income inequality (as least measured by gini coefficient).

[Mandos]

I don't disagree that free trade has undermined the ability of business to extract rents from consumers (and therefore of unions to split those rents with their members)

Which is of course not at all what I said or meant.

Moreover, your second point is inconsistent with Stephen's (and Mark's) observation that increases in income inequality (or labour share of income) seem to have little or no relation to the strength (or lack thereof) of national labour movements. As Stephen noted, that's a story that labour unions tell in the US, but it's a hard story to tell about France, which has worse levels of income inequality (as least measured by gini coefficient).

Which, as Stephen himself devoted a whole prior post to, is not a good way of measuring "top-end" inequality---which was the entire point of my other comments!

[Determinant]

Which would be a nice story, if there were any evidence that unions achieve that goal. Stephen's point is that it isn't obvious that they do. My point is that it isn't obvious why we would expect them to achieve that goal in a competitive market (since unions can only extract economic rents from the "rentier class" where there are economic rents to be extracted). Is there any evidence that rates of union coverage are correlated with lower incomes of the 1% on a country by country basis?

Your depiction of unions glossed over the United Food & Commercial Workers, which is the largest union in Canada (Unifor is second) represents supermarket workers, yes, even the party-time staff, and the construction unions, among others. Both of these are very competitive industries. Your premise here, Bob, is flawed.

[Bob Smith]

"Your depiction of unions glossed over the United Food & Commercial Workers, which is the largest union in Canada (Unifor is second) represents supermarket workers, yes, even the party-time staff, and the construction unions, among others. Both of these are very competitive industries. Your premise here, Bob, is flawed."

Not really. Unlike, say, steel workers or auto-workers, those industries have not been subject to international competition. Moreover, even if there is significant competition between their employers, if the union itself has market power in the supply of workers in those industries (i.e., 1 union, multiple employers) so that it can still extract economic rents (which employers can pass on to consumers through higher prices, because all their competitors are subject to higher costs). It is telling, for example, that in Ontario, employers in the construction industry in Ontario are bound by province-wide collective bargaining (see this memo prepared by the City of Toronto with respect to it's obligations: http://www.toronto.ca/fairwage/pdf/labour_trades_contractual_obligations.pdf). Gee, imagine that, when you have a monopoly on the provision of labour services in the Province, you can extract economic rents.

Gotta say, that's probably not the best argument I've ever heard for labour unions (i.e., everyone likes the story when they're shaking down big bad capitalists, that story sounds a lot worse if they're screwing over consumers).

Indeed, this rather supports my thesis, that unions can exist (and can extract higher wages) only in those industries where they extract economic rents (whether by splitting rents realized by employers, or by having market power of their own). Moreover, it illustrates the ambiguous impact of unions on inequality. If the cost of higher wages are borne by consumers, rather than owners of capital (because, hey, someone has to bear them), even if they can extract higher wages, it may have no impact, or aggravate, income inequality (for example, if union members are in the upper-end of the income distribution) and worsen consumption inequality (for example, if the cost of higher wages for construction or food service workers are born through more expensive housing or basic groceries - which are likely to make up a larger chunk of the budget of the poor).

[Bob Smith]

Moreover, it's also worth questioning the assumption that the surviving unions are successful at extracting wage premiums from their employers. They might survive by providing their employees with "soft" benefits (for example, protection against harassment by management) that employees value, but which cost employers nothing. It's interesting to note that Andrew Jackson (who knows a thing or two about labour unions -being the former policy wonk for the Canadian Labour Congress) recently observed that: "Controlling for differences such as occupation, education and age, the union wage premium—the difference in wages between union and non-union workers—has been falling, from about 20 percent in the 1980s to a modest 8 percent in 2002, and likely even less today(Fang and Verma 2002)". Unions can survive in a competitive industry, if their members are willing to accept competitive wages.

As an aside, it was interesting to read Andrew Jackson's comments about the declining wage premium for union members in light of the Canadian Labour Congress's latest PR campaign advertising the massive wage premium for unionized workers relative to non-unionized workers (http://www.canadianlabour.ca/about-clc/canada). The CLC isn't lying, but since it's not accounting for differences in occupation, age, education, etc. of unionized vs non-unionized workers, it's certainly misleading the public. I'm not sure if they're shameless or just desperate.

[Determinant]

Not really. Unlike, say, steel workers or auto-workers, those industries have not been subject to international competition. .

You are seriously claiming the grocery industry is uncompetitive? Riiight... See Walmart, Target, all the other entrants.

Gee, imagine that, when you have a monopoly on the provision of labour services in the Province, you can extract economic rents.

And the law profession, and....

Second, nice fallacy about market power.

Somebody has to have market power. Pure equality of market power of participants is a theoretical construct and not borne out in reality.

It is ever the question, who has and should have market power.

[Mandos]

It's interesting to note that Andrew Jackson (who knows a thing or two about labour unions -being the former policy wonk for the Canadian Labour Congress) recently observed that: "Controlling for differences such as occupation, education and age, the union wage premium—the difference in wages between union and non-union workers—has been falling, from about 20 percent in the 1980s to a modest 8 percent in 2002, and likely even less today(Fang and Verma 2002)". Unions can survive in a competitive industry, if their members are willing to accept competitive wages.

So, if unions forego the entire point of being unions---skewing the market for labour against employers---they can survive in an environment where they are competing against workers who are not being protected by a union. Wonderful.

Again, the subtext of all of this is the neoliberal free trade ideology---force classes of workers that have "made it" to compete with those who have not, with the faint hope waved in front of everyone's faces for a little bit of naive redistribution. Workers who do well by limiting the market for their labour are blamed, in Bob Smith's world, for the workers who don't do well. The answer is obviously to implement those damned transfers from the fabulously wealthy, and yet, mysteriously, the fabulously wealthy get fabulously wealthier...while the workers who have "made it" are cast down again.

A planet of slums is our future.

[Mandos]

You know, this all reminds me of the "logic" that was deployed shortly after the Bangladesh textile building collapse disaster. Anyone who dared notice that this was just the logical result of labour arbitrage against first world workers, and that, you know, we might at least consider making such things less rewarding for the importers was shouted down for wanting to send Bangladeshis back to their miserable farms...no matter the protests even in Bangladesh. It was as though Wal-Mart was some kind of charity operation, and its managers were forced, forced to skim off most of the surplus...

In the same way, when workers organize, it is automatically assumed to be the case that they are doing so against consumers. Capital? Apparently, it has no agency whatsoever. We are ruled by robots who happen to own yachts. And the robots are mocking us.

[Determinant]

Oh yes, and Bob dragged up the Auto Workers fallacy again. It was not unions or union contracts that forced GM and Chrysler into bankruptcy. Ford has the same unions and survived. It was bad management producing cars that people didn't want, or didn't want enough, that did in those two. And GM in Canada didn't go bankrupt, it was only the corporate parent.

The failures at those two companies were failures of management to develop products that the market wanted. That has nothing to do with labour.

[Bob Smith]

"You are seriously claiming the grocery industry is uncompetitive? Riiight... See Walmart, Target, all the other entrants."

Yes, and note Andrew Jackson's comments about the declining union wage premium. Coincidence?

"And the law profession, and...."

Nice try, but there's a world of different between regulatory barriers to enter a profession (and, more to the point, since the material bar to entering the legal profession is getting hired as an articling student, it's a bar that actually reflects demand for legal services, more than a true regulatory bar) and mandatory province-wide bargaining on the wages of members of that profession. Fees in the legal profession are remarkably cut-throat, since clients can negotiate freely with dozens of more or less identical providers - you can't do that in the trades.

"Somebody has to have market power"

Says who? That's just a silly statement. If it's "borne out by reality" you should have no problem providing empirical evidence for that proposition. In any event, I'm not sure why you think that disproves my point. Even if I were to concede that employers "always" have market power vis-à-vis their customers (which, facially, is a silly proposition), I think you'd be forced to concede that the degree of market power maintained in most industries has been undermined by competition. Moreover, it's plainly silly to suggest that companies always have market power vis-à-vis their employees (i.e., in the labour market) since they have to compete not only with companies in their own industry, but with companies in other industries and other regions (this might have been a better story in the era of the company town, but given the high degree of modern labour mobility, it doesn't withstand critical scrutiny).

"In the same way, when workers organize, it is automatically assumed to be the case that they are doing so against consumers. Capital? Apparently, it has no agency whatsoever. We are ruled by robots who happen to own yachts. And the robots are mocking us."

Quite the contrary, the assumption is that to the extent they can, the owners of capital will pass along costs to consumers. That's an assumption that is both empirically supported and consistent with your world view that the owners of capital are greedy money sucking pigs. Workers may like to think that they're sticking it to the "man", but funny enough, the "man" doesn't like getting stuck. Mind you, to the extent that the owners of capital CAN'T pass along those costs to consumers, of course, they'll shut down the factory and invest their money in a more profitable locale. So, yes, to the extent that unions can extract above market-wages, they're doing so at the expense of consumers, because capital has agency, and is willing to use it.

As an aside, why are you offended by the suggestion that workers are working against consumers? Isn't this obvious? That's not a moral statement about workers, that's a recognition that human beings are self-interested. The old Marxist framework of labour vs. capital is fun and all, but at the end of the day, they're both inputs on the production side of the equation. Both have an incentive, to the extent possible, to fleece consumers - disputes between labour and capital are just over how to divvy up the wool.

[Bob Smith]

"The failures at those two companies were failures of management to develop products that the market wanted. That has nothing to do with labour."

My, you're just full of straw-men today. I didn't argue that unions was the source of the big-3's problems, I said that increasing competition was the source of their problem. It's impossible to argue with that proposition. That they were badly managed (and Ford only less badly than the other two) was indisputable (successful mature companies don't have a "burn" rate), but only a problem as a result of cut-throat competition from the Japanese and the Koreans - they were no better managed back in the good ole' days (when their cars were even crappier), they just didn't have real competition to worry about. Mind you, a large chunk of their bad management was agreeing to labour contracts and long-term commitments that they couldn't afford in the long-run - a proposition that the UAW/CAW appear to have accepted, given the considerable give-backs they offered up in 2009 in the face of Chrysler and GM's bankruptcy (and, I'm sure you're aware that Ford, while not in bankruptcy, got in on that deal too).

In any event, don't be so defensive, the collapse of the Big-3 was such a business disaster that there's lots of blame to go around.

[Mandos]

So, yes, to the extent that unions can extract above market-wages, they're doing so at the expense of consumers, because capital has agency, and is willing to use it.

By agency, I meant moral/ethical/political agency, of course. A programmed automaton also has "agency" in the sense that you use it---and you are proposing that capital is not a political agent, but a kind of programmed automaton. *If* what you say is true, the conclusion is obviously the need for immediate confiscation.

As an aside, why are you offended by the suggestion that workers are working against consumers? Isn't this obvious? That's not a moral statement about workers, that's a recognition that human beings are self-interested. The old Marxist framework of labour vs. capital is fun and all, but at the end of the day, they're both inputs on the production side of the equation. Both have an incentive, to the extent possible, to fleece consumers - disputes between labour and capital are just over how to divvy up the wool.

And within the implicit framework of analysis you are using, the relative purchasing power of the consumer base has eroded in countries where the system is attempting to impose "labour reform", or did you miss the half-decade or so of sluggish demand? The workers are the bulk of consumers, and pitting worker against worker is the MO of of the 0.01%.

So to say that where workers manage to achieve some amount of prosperity is therefore supposedly at the expense of other workers...and economism accuses the left of playing zero-sum games?

[Adrian]

"The inequality measure is the Gini coefficient, which means that this post has nothing to say about 'top end' income concentration."

Really? The Gini coefficient is NOT completely insensitive to 'top end' income concentration. In fact it is arguably more sensitive to income inequalities at the top than at the bottom. This is because at the bottom end the Gini coefficient is limited by the fraction of poor people in the population. For example if 10% of the population earn nothing but the rest all earn the same non-zero amount (say $50k) then the Gini coefficient is 10%.
On the other hand at the top end the Gini coefficient is set by the total income of the high earners as a proportion of the national income and this can be potentially much higher. Thus if the top 1% or 0.1% earn between them 20% of the national income, then the Gini coefficient will be at least 20%. The Gini coefficient is thus sensitive to inequality at both ends of the spectrum.

In the second graph above (Union Density and Inequality in the OECD 201) all the countries with union densities over 40% have Ghent systems of unemployment benefit. The remainder show little overall pattern. But why should they? As you said:
"I don't think there's much doubt that unions would be a force for equality *within* a given occupation. But it's less obvious that increased unionisation would reduce the earnings gaps between different occupations."

Some at the TUC in the UK are campaigning for a stronger unionized workforce and more collective bargaining in the belief that it will reduce inequality. Personally I am sceptical. Even when unions in the UK were strong such as in the 1970s much of the industrial action was driven by the desire of one group of workers to "maintain their differentials" with respect to a group of lower paid workers. Unions tended to compete against each other as much as they did against shareholders.

If you want to reduce income inequality at both the top and the bottom then perhaps you should think about a different approach. The paper in the link below sets out one such alternative. Tax companies in proportion to their Gini coefficients.

http://www.cnlfabiansociety.org.uk/Publications.html

[Determinant]

Moreover, it's plainly silly to suggest that companies always have market power vis-à-vis their employees (i.e., in the labour market) since they have to compete not only with companies in their own industry, but with companies in other industries and other regions (this might have been a better story in the era of the company town, but given the high degree of modern labour mobility, it doesn't withstand critical scrutiny).

In the environment of today's unemployment rates, competitive labour conditions as you describe are few and far between. Market power in the labour field is on the employer's side.

[Patrick]

As far as the effect on inequality goes, the whole labour vs. ownership thing seems to me to be a bit irrelevant. Doesn't the data show that the top 1% or even .1% mostly get their income from wages? And FWIW anecdotal evidence is that the it's the masters of universe and the tight clique of professional non-owner denizens of the C-suite, and not some latter day railway barons who are calling the shots - extracting rents, buying legislators etc.

The so called ownership class has become powerless pensioners who own shares through their pension funds. They mostly don't (or can't, being one step removed) even vote their shares. I would argue that they have much more in common with labour than with your typical C-suite tart. Both require firm who prosper *in the long run*. The typical C-suit is not interested in the long run (see under Trojan Horse, Elop, Nokia).

So I actually think that corporate governance reform driven by organized labour's self-interest (properly defined), and the ownership class' self-interest (properly defined) are about the only hope of bringing some sanity to bear. Put crudely, the C-suit is stealing cash from shareholders and wages from workers. Sounds like they have a common problem.

[Bob Smith]

"Market power in the labour field is on the employer's side."

What do you mean by "market power"? Let's define our terms here. I'm using it in the proper sense as being a situation where suppliers can charge price > MC, or as an employer who can pay wages < MPL. How are using it. The mere fact that labour conditions are competitive, or that there is unemployment, doesn't mean that employers have market power.

[Bob Smith]

"Put crudely, the C-suit is stealing cash from shareholders and wages from workers. Sounds like they have a common problem."

Ding, ding, ding! We have a winner.

[Determinant]

Market power in the labour field is on the employer's side."

What do you mean by "market power"? Let's define our terms here. I'm using it in the proper sense as being a situation where suppliers can charge price > MC, or as an employer who can pay wages < MPL. How are using it. The mere fact that labour conditions are competitive, or that there is unemployment, doesn't mean that employers have market power.

I'm using in the sense that by and large, employers can and do say "you will get paid x" and if you say you want to be paid x+y, they will say no and you can walk. Which means that can be and often is x < MPL, or that the MPL is forced to fall.

[Declan Trott]

My comment seems to be stuck in moderation, but the problem with the ICTWSS database was a temporary website glitch - it downloads as an Excel file now.

On the broader topic, I would suggest that there is a strong negative relationship between unionisation and wage inequality, but that, strangely, wage shares have fallen least in those (developed) countries where wage inequality has risen most (particularly the US), so the relationship with overall income inequality is ambiguous.

I have written a paper on this but unfortunately the Australian Conference of Economists 2011 website has disappeared!

[Declan Trott]

Funny, that seemed to post alright. Anyway I was originally just asking where the union coverage data was from, since I couldn't find any at the OECD, and I had trouble downloading it from the ICTWSS (which is linked to from the OECD union density data).

[Stephen Gordon]

Comments aren't held for moderation, but they sometimes fall victim to our capricious spam filter and sometimes they just disappear. I just checked the spam filter and it's not there, so maybe that second thing happened.

I had a hard time tracking down coverage/density data; the numbers used here are from Figure 3.12 of the OECD Employment Outlook 2012. I don't recall offhand what their source was. It might have been the ICTWSS (I too tried looking there and came up empty).

[Min]

In a globalized economy, won't parochial unions be relatively ineffective? Whatever happened to "Workers of the world, unite!"

[Mandos]

In a globalized economy, won't parochial unions be relatively ineffective? Whatever happened to "Workers of the world, unite!"

To echo Bob, ding, ding, ding! We have a winner!

Our Great Leaders have worked hard to implement globalization as a form of worldwide labour arbitrage-enablement, and it shows.

[genauer]

Trade Union Density

[Daniel I. Harris]

Stephen, I love your posts. The emphasis on data is fantastic and convincing, and you always choose subjects that are fascinating.

[hix]

"I don't think there's much doubt that unions would be a force for equality *within* a given occupation. But it's less obvious that increased unionisation would reduce the earnings gaps between different occupations."

Unless the union represents different occupations, as is the norm in countries with strong unions. The Canada/US union model is the expectation, not the rule.

Look at Austria:
-Employer union membership is mandatory
-On the employee side, the ultimate decission authorty for tariffs is with the topc echelon of the Ögb, which represents all sub unions for specific sectors (not trades, thats just the anglo-saxon crap union model).

"It turns out that while union density is low in France, the share of workers covered by collective agreements is high. It seems to me that union coverage - the share of workers who are covered by collective agreements "

This is a big red hearing. High coverage rates can also be a sign of weakness. And why should French unions be strong? If the idea is that they are strong because they do noisy strikes, that has it exactly backwards. They do noisy strikes because they are weak. Strong unions get their demands fulfilled by making phone calls. Only the weak have to resort to strikes.

[Declan]

Thanks Stephen. (It was the ICTWSS BTW.)