A Summary of “Union Density, Productivity and Wages”

Welcome back! Today, I’ll be summarizing this interesting paper by Erling Barth, Alex Bryson, and Harald Dale-Olsen. I chose it because it examines Norway’s union system, where unions are organized by sector and coordinate their bargaining, as compared to the US’ uncoordinated and establishment-level unions. The former, I believe, has many positive effects over the latter, and is less examined. Remember, my comments are in brackets [like this]!

1. Introduction

This study exploits the change in the price of union membership in Norway from 2001–2012, aiming to find a relationship between productivity and union membership rate. It’s not immediately clear from theory which direction this relationship could go. On the one hand, increased union membership could lead to lower investments as investors move to more profitable areas. This means less capital and therefore lower productivity. Plus, union membership may cause workers to shirk their duties due to the extra protection unions provide. On the other hand, unions enhance the voice of workers, potentially communicating productivity-enhancing information that only they know. Workers are also more likely to stay in a union job (since they bring better pay and benefits), and this makes employers more likely to invest in workers’ education and skills.

The relationship here is generally tough to figure out, because it’s hard to find something exogenous to measure the effect of union membership on productivity directly. After all, if high productivity firms are more likely to unionize in this first place, this could lead to a spurious correlation. Or, perhaps unionized firms pick more productive employees to hire. Luckily, Norway’s tax system basically provides a subsidy for union membership. This subsidy varies over time. Since you’re more likely to join a union if it costs less, this allows us to measure the effect of union membership on productivity by comparing firms whose difference is just what the price of union membership is at the time.

2. Theory and Previous Empirical Literature

Past literature on unions often treats unions as “cartels”, who collaborate and “exploit” their employers. Adam Smith even noted this, while also noting employers can similarly form cartels, suggesting that where wages end up depends on the relative power of each force. It’s safe to assume, given the interests of workers, that being in a union increases your wage. These more expensive jobs may lead to employers automating some workers, which increases productivity (remember that’s measured in output per hour, so automation lowers labor hours but leaves output unchanged). [Interestingly, sectoral unions often run training programs, which can help retrain those displaced by automation into better jobs. A classic quote from this NYT piece: “In Sweden, if you ask a union leader, ‘Are you afraid of new technology?’ they will answer, ‘No, I’m afraid of old technology… The jobs disappear, and then we train people for new jobs. We won’t protect jobs. But we will protect workers.”] On the other hand, unions may strike, which decreases productivity.

An important channel through which unions act is compressing the wage distribution, increasing the wages of workers with low earnings, and decreasing the wages of workers with high earnings. In many European countries, wages are set by sectoral unions bargaining at the national level. This can achieve a high level of coverage. For example, 95% of workers in France are covered by a bargaining agreement. These high coverage rates lessen the potential negative effects of unionization, since it’s harder for investors to shift investment towards non-union sectors in the economy.

Norway specifically has strong sectoral unions, with 50% of private workers being union members [and 70% being covered by a collective bargaining agreement]. There may, of course, be local bargaining that raises wages above the sectorally-negotiated level, and 79% of those covered by sectoral agreements have such local agreements.

3. A Simple Model of Union Membership

This section is heavy on the statistics, so I’ll just list a couple important points about their model. They assume unions set higher wages, and provide some insurance against bad employer actions, in addition to legal services. They model the cost of joining a union as the union fee minus the tax subsidy given by the state.

4. The Norwegian Tax Legislation and the Union Membership Fee

Due to shifts in political power, the subsidy for union fees (given as a tax deduction) has generally increased from 2001–2012, see below.

Their data contains the union membership fees paid by all individuals as reported to the tax authorities in Norway for those years. They take as a instrumental variable the subsidy/(union fee-subsidy), and note that lower wage earners generally pay lower membership fees.

5. Empirical Approach and Data

The authors measure the union fee when firms first enter the data, which hopefully screens out endogenous effects of folks wanting to unionize due to productivity changes in a firm over time rather than price changes to joining the union. They assume that the elasticity of union membership demand to cost is constant. Norway’s administrative data contains all the information needed about the size of the subsidy and union fees [you can read some about how thorough Nordic administrative data is here].

6. Results

The elasticity of union membership to the subsidy is 0.108, meaning “a 10 percentage point increase in the subsidy rate yields a 1.08 percentage point increase in the probability of union membership”. A subsidy increase of 100 NOK leads to a 4–7% increased likelihood of union membership for someone paying average fees. They also find that a 1% increase in union membership leads to a 1.7–1.8% increase in firm productivity. Also, a 1% increase in union density also leads to a 1–1.5% increase in wages. They find as well that higher productivity firms have a higher wage premium from union membership, lending credence to idea that unions increase wages through rent sharing.

7. Conclusion

We shouldn’t expect the above numbers to remain the same in countries with different institutional setups. For example, enterprise-level union membership may have very different effects than the sectoral union membership examined in Norway. The graph below compares the actual union membership rate from 2001–2012 to the counterfactual where the union subsidy did not increase from its value in 2001. The increased subsidy increased union membership by 6% overall. [This to me is a shockingly high number for a fairly small subsidy. The subsidy increased 800 NOK from 2001 to 2012, or $143 at 2011 exchange rates. Imagine if it was instead something like $1,000 a year, which is reasonably affordable.]