8 ms·

Equilibrium Effects of Pay Transparency

63 points by jandrewrogers 5y ago

versale 5y ago
> Our model predicts that transparency reduces the individual bargaining power of workers, leading to lower average wages. So, it's in employers' interest to make the salaries transparent, isn't it? I bet the model is incomplete.
- ElViajero 5y ago
  That is a very good point. If employers have the information, and they could already share it to lower salaries. Why are they not doing it? Or something is missing, or the employers are going to realize their mistake and implement transparency.
  - sokoloff 5y ago
    I’m in favor of a light version of salary transparency. (“Here’s the anonymized spread of compensation and you can find where you land in the spread.”) I’m not in favor of publishing the individual salary information of people who joined my company without that publication being part of the bargain. (I view it as a serious privacy violation whether I publish “here’s what XYZ makes” or whether I publish enough data that lets a data scientist figure out which point in the dataset is XYZ with greater than 80% certainty.)
    - callmeal 5y ago
      >I’m in favor of a light version of salary transparency. In countries with salary transparency, (the ones I've worked in anyway), there is usually a published grid of roles, and salary ranges. Similar to how the government does things[0]. Companies then make use of "benefits" to reward better performing employees, since those benefits are not part of salary. [0]: https://www.federaljobs.net/salarybase.htm#SALARY_TABLE_2015-GS https://www.federaljobs.net/salarybase.htm#SALARY_TABLE_2015...
- akdor1154 5y ago
  > So, it's in employers' interest to make the salaries transparent, isn't it? I bet the model is incomplete. Not necessarily - without transparency, an employer might decide to hire Dr Super Engineer, and might be happy to pay her 2× the mean wage of their existing engineering team. With transparency, they might choose not to hire her, because they are only happy to pay her 2× if the existing engineering team are not aware, and she might refuse to take the job if they won't give her 2×.
  - callmeal 5y ago
    >With transparency, they might choose not to hire her, because they are only happy to pay her 2× if the existing engineering team are not aware, and she might refuse to take the job if they won't give her 2×. What's stopping them from creating a new role/position/title that pays 2x?
    - lsiebert 5y ago
      Nothing, but there's no need for that, because there's nothing stopping them from paying her 2x for the same position, based on the actual skills and experience someone has. You can totally pay people different wages under these laws for the same position, if they are better at their job, you just have to show that that's why.
      - dolni 5y ago
        Employers are already intrinsically motivated to pay as little as they can get away with. Isn't the fact that they are willing to pay more the best kind of proof?
        lsiebert 5y ago
        That would only work as proof if you believe that salary is a perfect metric for job performance, and that nobody is paid less than they could get elsewhere (IE that the labor market operates with perfect efficiency). I imagine we both know neither of those is the case. Since those are not the case, you can ask yourself, are bills requiring equal pay for equivalent work and pay transparency each likely to improve salary as a metric for job performance, and labor market efficiency? I suspect they do both, but if you believe otherwise, feel free to explain.
- Mauricebranagh 5y ago
  Doesn't seem to work for CEO pay :-)
cinntaile 5y ago
You don't really need to make up a theoretical model for this. Just analyze Norway or Sweden where pay transparancy exists to discover how it impacts pay. I remember reading that it reduces overall pay as the model here predicts, but I don't remember the source.
- occz 5y ago
  Pay transparency exists in Sweden? You mean like taxable income being part of public records?
  - cinntaile 5y ago
    Yes. That should be pretty accurate since most people only work at one job and don't switch each year.
  - karatinversion 5y ago
    I’m not familiar with Sweden, but this is how it works in Finland. The data are not published online, but can be queried by phone or in person from the tax office. National newspapers publish lists of the top earners each year when the data are published; here is one example (in Finnish), including everyone with income over €100,000: https://www.iltalehti.fi/verokone https://www.iltalehti.fi/verokone
  - mellavora 5y ago
    Yes, taxable income is part of public records in Sweden.
- pydry 5y ago
  A non theoretical model's results wouldnt be able to eliminate confounding variables in Finland/Norway. Did pay go down because of pay transparency or because of 17 other variables that were changing at the same time? No way to tell. A theoretical model is likely trash - you can pretty much reverse the outcome by tweaking a couple of assumptions. The only way I can think of that might work would be to measure the effect on pay for similar jobs across a state border - where one side implemented pay transparency and the other didn't (similar to how the Dube, Lester, Reich study was constructed).
anothernewdude 5y ago
All the good workers end up being contractors anyway.
let_me_ask_this 5y ago
The model is complete nonsense. If you are an employer in a jurisdiction with pay transparency, and you need to get a superstar employee, you just create a titled position for them. Then you can justify paying them 350% the average wage. The paper claims that there will be a wage equilibrium. Yes, for average employees there will be a wage equilibrium, and average employees will see their wages increase because of this. The paper tries to claim that this wage equilibrium will negatively affect superstar employees. This is complete nonsense. It bases this on the assumption that employers cannot justify discriminating between normal employees and superstar employees. But this is an unwarranted assumption. Literally just give the guy a title. Secondly it assumes that even for these superstar employees their wage would reach an equilibrium. This is another misguided assumption, because these superstar employees are not a fungible resource like the rest of the employees in the same occupation. They very likely possess unique experience and skillsets. So the pay of other superstar employees should not have any bearing on how much they themselves get paid. Basically the model seems to think that somehow all the employees are equal and entirely fungible, but at the same time some employees somehow manage to negotiate much higher salaries. Now of course I am not an economist so it is ridiculous to believe that I have found the fatal flaw in a published paper, but because of the political nature of such articles I find it hard to trust the "hard numbers". I remain skeptical.
- mulvya 5y ago
  Right, what is the spread of salaries among star and journeymen players in professional sports leagues? The Indian cricket league IPL has a public auction every year. Salaries range from 2 million at the top to 20/30K at the bottom. And most players are towards the bottom.
- naturalauction 5y ago
  >Now of course I am not an economist so it is ridiculous to believe that I have found the fatal flaw in a published paper, but because of the political nature of such articles I find it hard to trust the "hard numbers". I remain skeptical. I admittedly haven't read the paper properly, but did study economics at undergrad and read many similar papers. This is how a lot of economics works, you end up with published papers claiming completely different results because of small differences model assumptions. Multiple models can be consistent with observed data while providing vastly different policy implications. If you want to look at a well publicized example (though I think it might have to do more with different econometric assumptions) look at the research published on the Mariel boatlift and its impact on labor markets. See here - https://www.bruegel.org/2017/06/the-mariel-boatlift-controversy/ https://www.bruegel.org/2017/06/the-mariel-boatlift-controve...
saurik 5y ago
> Our model predicts that transparency reduces the individual bargaining power of workers, leading to lower average wages. But, are those of us advocating for pay transparency trying to increase the average wage, though, or trying to reduce pay disparity?
- sokoloff 5y ago
  Why ask others to guess what your motivations are if you can instead tell us? Reducing disparity within a company is surely achievable and almost inevitable. I don’t personally believe it’s desirable in endeavors where value contributions can be heavily unevenly distributed, because it likely leads to the company being unable to employ the very top value contributors as another company can [and likely will choose to] pay them more, which is consistent with the main claims in the article.
  - lsiebert 5y ago
    yeah, as best I can tell you absolutely can pay someone more if they are better at their job under equal pay laws. In California for example, the equal pay act requires 'equal pay for employees who perform “substantially similar work,” when viewed as a composite of skill, effort, and responsibility.' https://www.dir.ca.gov/dlse/California_Equal_Pay_Act.htm https://www.dir.ca.gov/dlse/California_Equal_Pay_Act.htm Someone who contributes more inherently must have greater skills and/or put in greater effort, and likely greater responsibility, given they are accomplishing more tasks. You can pay differently based on a seniority system, a merit system, a system that measures earnings by quantity or quality of production, or another bonafide factor such as education, training or experience. Here's a copy of that law, it's pretty short. https://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?lawCode=LAB&sectionNum=1197.5 https://leginfo.legislature.ca.gov/faces/codes_displaySectio....
    - sokoloff 5y ago
      MA law is fairly similar. My concerns around transparency are mostly individual privacy concerns rather than believing that employees (or even regulators) can’t understand generally why two people could justly be paid different quantities of the local currency. No place is (yet) mandating perfect sameness of the latter and I don’t expect that to be policy in my lifetime or while we have the concept of money as a society.
OhNoMyqueen 5y ago
Bargaining power is mainly dependant on offer and demand; it can be low, or it can be high, depending on how valuable you are to the company. Pay transparency will simply increase the number of employees who will try to bargain. I guess that if you're not valuable enough that could backfire for everyone. If you do are valuable the company will just need to pay.
- stuaxo 5y ago
  The company is made up of people, and people have biases in who they value hence these inequalities.
rkangel 5y ago
The model skips an important part of the reasoning - some people might deserve to be paid more because they're more valuable (due to being more productive or whatever the equivalent is in a given industry). Pay transparency means that if you want to pay someone more they have to be able to justify it. In a well managed company you need to have some form of employee assessment system to judge how well they're doing. This is already true so that good people get rewarded, but becomes even more important with transparent pay because you have to be able to point to it and say this is why that person gets 50% more than you. You have to be able to have real conversations like "they brought on 2 new clients and delivered all their projects on time, whereas your last project was on budget but two weeks late". Being good and useful (and comparing salaries at other companies) provides the employee upward leverage on price (the inverse version of my role play above). That said - I don't disagree with the conclusions of the paper (showing correlation between pay transparency and reduced pay), but I don't think it's an important conclusion. Pay is a frequency distribution. Talking about the mean/median of that distribution is less important than making there there are good reasons for the people at either ends of it (e.g. "I'm at the lower end because I'm female). To put it another way - I'm more worried about inequality than overall pay level because the people in the middle are fine, it's people at the bottom who need help.
- BeFlatXIII 5y ago
  The problem with specific justifications is that many roles do not have objective KPIs like this. How do you compare the worst teammate on a successful project with a superstar on a project that was doomed from the start? For that matter, there are also the office intangibles that don’t show up nicely on an evaluation but do cause problems when nobody does them.
  - rkangel 5y ago
    These are all problems that exist already and are just highlighted by pay transparency. I agree that they're difficult problems, but they're important ones. You definitely can't do a proper employee evaluation by a completely objective set of numerical KPIs (even in places it seems possible like sales numbers, it results in gaming the system and knock on negative effects). On the other hand, you can definitely have a consistent system for looking at the different areas of people's performance and judging how much they've added. It takes some effort to get right, and there will always be an element of subjectivity but it is possible to do a decent job at it.
jl2718 5y ago
I think this is conflating two separate effects. One is pay transparency, and the other is legislated pay equity. Legislated pay equity will always result in lower wages simply because the employer is legally unable to raise their bid for talent. No surprises there.
hammock 5y ago
This seems like it might not be a holistic model. Pay transparency might result in lower wages at the transparent company, as underpaid employees jump ship to another company - leaving only the lower-value companies being perhaps more fairly paid. That doesn't mean lower wages overall for all workers involved, though. The ones that jumped ship may be getting much more somewhere else, raising the average in a way that wasn't looked at in this model.

63 points by jandrewrogers 5y ago

> Our model predicts that transparency reduces the individual bargaining power of workers, leading to lower average wages. So, it's in employers' interest to make the salaries transparent, isn't it? I bet the model is incomplete.

ElViajero 5y ago
That is a very good point. If employers have the information, and they could already share it to lower salaries. Why are they not doing it? Or something is missing, or the employers are going to realize their mistake and implement transparency.
- sokoloff 5y ago
  I’m in favor of a light version of salary transparency. (“Here’s the anonymized spread of compensation and you can find where you land in the spread.”) I’m not in favor of publishing the individual salary information of people who joined my company without that publication being part of the bargain. (I view it as a serious privacy violation whether I publish “here’s what XYZ makes” or whether I publish enough data that lets a data scientist figure out which point in the dataset is XYZ with greater than 80% certainty.)
  - callmeal 5y ago
    >I’m in favor of a light version of salary transparency. In countries with salary transparency, (the ones I've worked in anyway), there is usually a published grid of roles, and salary ranges. Similar to how the government does things[0]. Companies then make use of "benefits" to reward better performing employees, since those benefits are not part of salary. [0]: https://www.federaljobs.net/salarybase.htm#SALARY_TABLE_2015-GS https://www.federaljobs.net/salarybase.htm#SALARY_TABLE_2015...
akdor1154 5y ago
> So, it's in employers' interest to make the salaries transparent, isn't it? I bet the model is incomplete. Not necessarily - without transparency, an employer might decide to hire Dr Super Engineer, and might be happy to pay her 2× the mean wage of their existing engineering team. With transparency, they might choose not to hire her, because they are only happy to pay her 2× if the existing engineering team are not aware, and she might refuse to take the job if they won't give her 2×.
- callmeal 5y ago
  >With transparency, they might choose not to hire her, because they are only happy to pay her 2× if the existing engineering team are not aware, and she might refuse to take the job if they won't give her 2×. What's stopping them from creating a new role/position/title that pays 2x?
  - lsiebert 5y ago
    Nothing, but there's no need for that, because there's nothing stopping them from paying her 2x for the same position, based on the actual skills and experience someone has. You can totally pay people different wages under these laws for the same position, if they are better at their job, you just have to show that that's why.
    - dolni 5y ago
      Employers are already intrinsically motivated to pay as little as they can get away with. Isn't the fact that they are willing to pay more the best kind of proof?
      - lsiebert 5y ago
        That would only work as proof if you believe that salary is a perfect metric for job performance, and that nobody is paid less than they could get elsewhere (IE that the labor market operates with perfect efficiency). I imagine we both know neither of those is the case. Since those are not the case, you can ask yourself, are bills requiring equal pay for equivalent work and pay transparency each likely to improve salary as a metric for job performance, and labor market efficiency? I suspect they do both, but if you believe otherwise, feel free to explain.
Mauricebranagh 5y ago
Doesn't seem to work for CEO pay :-)

cinntaile 5y ago

You don't really need to make up a theoretical model for this. Just analyze Norway or Sweden where pay transparancy exists to discover how it impacts pay. I remember reading that it reduces overall pay as the model here predicts, but I don't remember the source.

occz 5y ago
Pay transparency exists in Sweden? You mean like taxable income being part of public records?
- cinntaile 5y ago
  Yes. That should be pretty accurate since most people only work at one job and don't switch each year.
- karatinversion 5y ago
  I’m not familiar with Sweden, but this is how it works in Finland. The data are not published online, but can be queried by phone or in person from the tax office. National newspapers publish lists of the top earners each year when the data are published; here is one example (in Finnish), including everyone with income over €100,000: https://www.iltalehti.fi/verokone https://www.iltalehti.fi/verokone
- mellavora 5y ago
  Yes, taxable income is part of public records in Sweden.
pydry 5y ago
A non theoretical model's results wouldnt be able to eliminate confounding variables in Finland/Norway. Did pay go down because of pay transparency or because of 17 other variables that were changing at the same time? No way to tell. A theoretical model is likely trash - you can pretty much reverse the outcome by tweaking a couple of assumptions. The only way I can think of that might work would be to measure the effect on pay for similar jobs across a state border - where one side implemented pay transparency and the other didn't (similar to how the Dube, Lester, Reich study was constructed).

anothernewdude 5y ago

All the good workers end up being contractors anyway.

let_me_ask_this 5y ago

The model is complete nonsense. If you are an employer in a jurisdiction with pay transparency, and you need to get a superstar employee, you just create a titled position for them. Then you can justify paying them 350% the average wage. The paper claims that there will be a wage equilibrium. Yes, for average employees there will be a wage equilibrium, and average employees will see their wages increase because of this. The paper tries to claim that this wage equilibrium will negatively affect superstar employees. This is complete nonsense. It bases this on the assumption that employers cannot justify discriminating between normal employees and superstar employees. But this is an unwarranted assumption. Literally just give the guy a title. Secondly it assumes that even for these superstar employees their wage would reach an equilibrium. This is another misguided assumption, because these superstar employees are not a fungible resource like the rest of the employees in the same occupation. They very likely possess unique experience and skillsets. So the pay of other superstar employees should not have any bearing on how much they themselves get paid. Basically the model seems to think that somehow all the employees are equal and entirely fungible, but at the same time some employees somehow manage to negotiate much higher salaries. Now of course I am not an economist so it is ridiculous to believe that I have found the fatal flaw in a published paper, but because of the political nature of such articles I find it hard to trust the "hard numbers". I remain skeptical.

mulvya 5y ago
Right, what is the spread of salaries among star and journeymen players in professional sports leagues? The Indian cricket league IPL has a public auction every year. Salaries range from 2 million at the top to 20/30K at the bottom. And most players are towards the bottom.
naturalauction 5y ago
>Now of course I am not an economist so it is ridiculous to believe that I have found the fatal flaw in a published paper, but because of the political nature of such articles I find it hard to trust the "hard numbers". I remain skeptical. I admittedly haven't read the paper properly, but did study economics at undergrad and read many similar papers. This is how a lot of economics works, you end up with published papers claiming completely different results because of small differences model assumptions. Multiple models can be consistent with observed data while providing vastly different policy implications. If you want to look at a well publicized example (though I think it might have to do more with different econometric assumptions) look at the research published on the Mariel boatlift and its impact on labor markets. See here - https://www.bruegel.org/2017/06/the-mariel-boatlift-controversy/ https://www.bruegel.org/2017/06/the-mariel-boatlift-controve...

saurik 5y ago

> Our model predicts that transparency reduces the individual bargaining power of workers, leading to lower average wages. But, are those of us advocating for pay transparency trying to increase the average wage, though, or trying to reduce pay disparity?

sokoloff 5y ago
Why ask others to guess what your motivations are if you can instead tell us? Reducing disparity within a company is surely achievable and almost inevitable. I don’t personally believe it’s desirable in endeavors where value contributions can be heavily unevenly distributed, because it likely leads to the company being unable to employ the very top value contributors as another company can [and likely will choose to] pay them more, which is consistent with the main claims in the article.
- lsiebert 5y ago
  yeah, as best I can tell you absolutely can pay someone more if they are better at their job under equal pay laws. In California for example, the equal pay act requires 'equal pay for employees who perform “substantially similar work,” when viewed as a composite of skill, effort, and responsibility.' https://www.dir.ca.gov/dlse/California_Equal_Pay_Act.htm https://www.dir.ca.gov/dlse/California_Equal_Pay_Act.htm Someone who contributes more inherently must have greater skills and/or put in greater effort, and likely greater responsibility, given they are accomplishing more tasks. You can pay differently based on a seniority system, a merit system, a system that measures earnings by quantity or quality of production, or another bonafide factor such as education, training or experience. Here's a copy of that law, it's pretty short. https://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?lawCode=LAB&sectionNum=1197.5 https://leginfo.legislature.ca.gov/faces/codes_displaySectio....
  - sokoloff 5y ago
    MA law is fairly similar. My concerns around transparency are mostly individual privacy concerns rather than believing that employees (or even regulators) can’t understand generally why two people could justly be paid different quantities of the local currency. No place is (yet) mandating perfect sameness of the latter and I don’t expect that to be policy in my lifetime or while we have the concept of money as a society.

OhNoMyqueen 5y ago

Bargaining power is mainly dependant on offer and demand; it can be low, or it can be high, depending on how valuable you are to the company. Pay transparency will simply increase the number of employees who will try to bargain. I guess that if you're not valuable enough that could backfire for everyone. If you do are valuable the company will just need to pay.

stuaxo 5y ago
The company is made up of people, and people have biases in who they value hence these inequalities.

rkangel 5y ago

The model skips an important part of the reasoning - some people might deserve to be paid more because they're more valuable (due to being more productive or whatever the equivalent is in a given industry). Pay transparency means that if you want to pay someone more they have to be able to justify it. In a well managed company you need to have some form of employee assessment system to judge how well they're doing. This is already true so that good people get rewarded, but becomes even more important with transparent pay because you have to be able to point to it and say this is why that person gets 50% more than you. You have to be able to have real conversations like "they brought on 2 new clients and delivered all their projects on time, whereas your last project was on budget but two weeks late". Being good and useful (and comparing salaries at other companies) provides the employee upward leverage on price (the inverse version of my role play above). That said - I don't disagree with the conclusions of the paper (showing correlation between pay transparency and reduced pay), but I don't think it's an important conclusion. Pay is a frequency distribution. Talking about the mean/median of that distribution is less important than making there there are good reasons for the people at either ends of it (e.g. "I'm at the lower end because I'm female). To put it another way - I'm more worried about inequality than overall pay level because the people in the middle are fine, it's people at the bottom who need help.

BeFlatXIII 5y ago
The problem with specific justifications is that many roles do not have objective KPIs like this. How do you compare the worst teammate on a successful project with a superstar on a project that was doomed from the start? For that matter, there are also the office intangibles that don’t show up nicely on an evaluation but do cause problems when nobody does them.
- rkangel 5y ago
  These are all problems that exist already and are just highlighted by pay transparency. I agree that they're difficult problems, but they're important ones. You definitely can't do a proper employee evaluation by a completely objective set of numerical KPIs (even in places it seems possible like sales numbers, it results in gaming the system and knock on negative effects). On the other hand, you can definitely have a consistent system for looking at the different areas of people's performance and judging how much they've added. It takes some effort to get right, and there will always be an element of subjectivity but it is possible to do a decent job at it.

jl2718 5y ago

I think this is conflating two separate effects. One is pay transparency, and the other is legislated pay equity. Legislated pay equity will always result in lower wages simply because the employer is legally unable to raise their bid for talent. No surprises there.

hammock 5y ago

This seems like it might not be a holistic model. Pay transparency might result in lower wages at the transparent company, as underpaid employees jump ship to another company - leaving only the lower-value companies being perhaps more fairly paid. That doesn't mean lower wages overall for all workers involved, though. The ones that jumped ship may be getting much more somewhere else, raising the average in a way that wasn't looked at in this model.