Average is Over, installment #1437

Over the past few decades, we find that about 80% of the widening residual wage inequality to be within jobs.

Furthermore, performance-pay incidence is the single largest factor behind that, accounting for 42% of those changes.  Of course this brings us back to the least popular explanation for growing income inequality, namely that we measure productivity better than before, and reward it accordingly.

That is all from a new NBER working paper by Rongsheng Tang, Yang Tang, and Ping Wang.

Comments

Is there an implicit assumption here that performance pay accurately measures productivity and is correctly and fairly allocated? What would be interesting to explore is the role that bias plays in the awarding of performance pay. In my experience cliques and insider groups are very good at gaming performance management systems inside larger firms to award each other fat bonuses. Pay goes to the best connected, not necessarily to the hardest working. So even if performance pay within jobs is a driver of inequality, this may not necessarily imply that the explanation for increasing inequality is anodyne.

That is not an assumption, it is a possible explanation for the results.

Respond

Add Comment

"fairly"

Indeed, it's hard to have adult discussions with children who worry about what's "fair".

The quality of your comment obviously marks you out as an adult.

Respond

Add Comment

Calling your workers "children" is a neat way to stifle wage negotiations. It usually doesn't work.

Respond

Add Comment

hardest working Not the same as most productive. Participation trophy anyone?

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

'namely that we measure productivity better than before, and reward it accordingly'

You mean the most productive cashiers at Walmart are paid according to their productivity?

And honestly, we do a horrible job of measuring CEO productivity, and an equally horrible job paying CEOs for their actual contribution to anything but their own self-interest in ensuring the largest compensation package for themselves.

I'm in Europe but most places here have robot self checkouts. The best cashiers are probably only 2 or 3 times as fast as the average cashier. The kinds of jobs where worker-productivity distributions are uniform or normal now make up a smaller part of total output than jobs where that distribution is pareto or exponential.

"The best cashiers are probably only 2 or 3 times as fast as the average cashier" Only? 3x as fast as average is a huge difference in this context. Huge.

Right, I mean I erred on that side because my overall point is about the difference between cashiers and knowledge workers. Of course, they deserve to be paid according to their productivity multiple.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Who cares how productive a cashier is? If you don’t shop online that’s your problem and there are still these things called markets that make sure that work that that can be done by a 7 year old get done by people deserving commiserate pay.

Respond

Add Comment

Prior gets a D- for reading comprehension.

“Over the past few decades, we find that about 80% of the widening residual wage inequality to be within jobs.”

“Of course this brings us back to the least popular explanation for growing income inequality, namely that we measure productivity better than before, and reward it accordingly.”

You responded with “but muh cashiers.”

+1

Respond

Add Comment

+1

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

If only this were true.
I think to the super productive executives at Hertz who loaded the company with debt to support the stock price.
I think to the productive executives at Boeing who RIFed people who knew how to build planes safely.

The Great Stagnation In CEO Competence

Respond

Add Comment

This is what happens when you bail out the financial sector.

Respond

Add Comment

Respond

Add Comment

This is a self fulfilling prophecy. In that the people in the positions to be the most productive can be the most productive, and then justify the highest wages.... I feel like this must be the perspective you get when you are in academia and do not see how offices actually function.

It stinks of James Buchanan's legacy...

It is an elaborate self justification for the status quo when it is obviously failing.

Yes, my mood affiliation proves that the status quo is obviously failing.

Respond

Add Comment

Respond

Add Comment

No progress of knowledge here. It's just that it became socially more acceptable to measure individual productivity and adjust reward accordingly. This is the flip side of the decline in collective bargaining. It is not established that this trend is conductive to higher average productivity nor, of course, to higher social welfare.

Respond

Add Comment

It is an elaborate self justification for the rich getting richer. As long as that is occurring, the status quo is meaningless.

Respond

Add Comment

"Counterfactual analysis shows the contributions of performance-pay incidence and job relatedness are about 42% and 26%, respectively, both higher than that of job-specific productivity."

Give me a break. My counterfactual analysis is that these results can't be replicated in any meaningful sense. But nice try.

>My counterfactual analysis is that these results can't be replicated in any meaningful sense. But nice try.

This is a very compelling argument! Thank you for the comment sir/madam.

Respond

Add Comment

Respond

Add Comment

Rather than just measuring productivity better than before, isn't it the case that modern knowledge-work allows greater potential for large actual productivity gaps? You've heard of the 10x programmer; I can imagine a 10x social media manager, or game designer, but there are probably no 10x truck drivers or assembly line workers.

"there are probably no 10x truck drivers or assembly line workers."

10x? I don't know. But there are absolutely 2x-5x difference. I work in factory automation. Downtime is measured in minutes per shift. A good worker will easily have half or less downtime per shift. Furthermore, the good workers will tend to be given the more complicated systems that are more error prone.

Workers in automated plants don't spend their time building things with their hands. They are using an HMI (computer interface) to start the sequence, monitor for alarms and ensure that the material flow is steady and well stocked.

In addition, the better workers are proactive about problems and will alert managers and/or engineers of odd behaviors that are often early indicators of imminent failure. This allows the maintenance to happen on a scheduled basis, instead of after a failure in the middle of production.

Very interesting observations, and I agree we could do a much better job at measuring productivity even in these kinds of roles.

Respond

Add Comment

Respond

Add Comment

From purely anecdotal experience I’d put the programmer proficiency differential at a possible 50x. Or maybe even higher.

Disclaimer: I was most definitely NOT among the most proficient. I was in awe of those programmer Gods. I was, however, aware enough to ask for their help when I needed it.

Seconded. There are a few people out there who write effective code rapidly and almost never write mistakes. They get more done in an hour than others do in a week. Once they realize they can demonstrate and market this ability, the sky is the limit for their wages.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

The increasing prevalance of "performance pay" also explains why wages have become less "sticky". It's almost automatic today that when the economy declines, measured average wages also decline due to lower bonuses, option income, and other types of incentive/performance pay. Employers don't have a contractual obligation to pay bonuses.

+1, excellent point

Respond

Add Comment

Respond

Add Comment

So what's the game plan for average people? By definition we all can't be above average. So just give up and die?

The measure of inequality is irrelevant to average people. Their welfare is not dependent on the incomes of others.

Yes. One can live pretty well at the average wage. At the average wage, one can live in a climate controlled dwelling with people they love.

Respond

Add Comment

Average people will need to accept that their betters will control the direction of this country in almost every way that matters: how we trade, who we make war on, how much coal we burn, who we let in, how much college costs, what we are allowed to think, etc. Average people need to accept that it is their station in life to have no influence over such things. This is nothing new.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

The inequality in incomes is much greater than inequality in wages, even if you count stock options for CEOs of large corporations from wage labor not rent seeking

Respond

Add Comment

I'm not that concerned about income inequality, it's wealth inequality that concerns me. Why? Financial and economic instability. The connection between high levels of wealth inequality and financial and economic instability is undeniable, although, as with coronavirus, in some quarters there is plenty of denial. I know, income inequality is the first step to wealth inequality, but today wealth inequality is mostly a product of happenstance: being born in the right family. There's a reason we have come to rely on rising asset (stock) prices for prosperity. There's a reason innovation and productivity have stagnated. There's a reason opportunity for many has been shipped elsewhere.

+1
Unfortunately, those who have designed and maintain the actual system we live in have discovered that they can push the idea of systemic racism and...voila...they're off the hook.

Respond

Add Comment

Respond

Add Comment

So RIDDLE ME THIS: If we've gotten better at measuring and rewarding productivity, and we've lowered marginal income tax rates to boost the productivity of high earners....why is productivity growth slowing?

There may be a chicken and egg scenario here. As society gets richer, a certain number of people are more complacent and rewarding productivity more is need for growth.

Respond

Add Comment

".why is productivity growth slowing?"

Tyler wrote an entire book on the answer to that question.

"The Great Stagnation"

Because we've reached a technological plateau and now innovation is just harder than it was in the past? What a weak thesis. Of course in hindsight the breakthroughs of the past 300 years seem like they were easy! And his justification that we've reached a tech plateau is based on the slowing of productivity (circular reasoning). Bestselling books based ad hoc empirical arguments that cannot be falsified are why Economics has never made it into the rung of more respected hard sciences.

The technological plateau we reached was speed of communication. We have instantaneous what 100 years ago took week, 50 years ago took a day, and 25 years ago took an hour. The only feasible improvement we have is one person thinking something and another person automatically knowing it; of course, I'm not convinced that would be an actual improvement.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Perhaps wages and production would increase if stock options and bonuses were awarded further down the ranks to noted productive employees rather than to revolving door executives.

+1, I think that would be a good idea, however, you can't effectively award a low level employee with stocks and have it be a strong incentive, because their actions don't move the stock very much. Bonuses in that case ought to be based upon some metric that a single employee can significantly affect when possible.

WestJet was an airline that used to be owned via shares by its employees. Because personal In airlines are both highly trained and low in number each individual had a big effect on the airline when it was small.
Once WestJet got big enough they bought back all the employee stock and now they’re just like any other airline.

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Did they account for year of release from slavery?

Respond

Add Comment

Please rate yourself on a scale of 1 to 5, where 1 means you will get fired, and 5 means you will get a raise, while noting carefully that the only allowable answer is 3. Your self-assessment will then be reviewed by your manager, who will have final say on what your self-assessment score is. All non-3 responses will be corrected to a value of 3. Congratulations, your self-assessment score is 3. Please sign below your manager's signature to affirm that you agree with this assessment.

Oh, yes. We're absolutely wonderful at measuring productivity.

+1 This is all too real. Been there.

Respond

Add Comment

Respond

Add Comment

The bimodal distribution of attorney starting salaries would be an example: https://abovethelaw.com/2020/05/the-bimodal-salary-of-starting-lawyers/

Respond

Add Comment

I had an econ professor in 1977 or 78 who said, something along that line about productivity had allowed him to beat the markets in his investments.

Respond

Add Comment

I think this is my favorite list of co-authors of all time.

Respond

Add Comment

An excellent thread.

Respond

Add Comment

Does performance pay explain a large part of the overall wage inequality? I did a little bit of investigation:
- An earlier version of the paper (as I can't access the NBER one) says "within group wage inequality or residual wage inequality counts for around 2/3 of total wage inequality" (footnote 1, https://editorialexpress.com/cgi-bin/conference/download.cgi?db_name=EEAESEM2017&paper_id=414)
- This paper's abstract says they "find that about 80% of the widening residual wage inequality to be within jobs"
- They built a model that "can account around 92% of the changes in within-job inequality among the highly educated from 1990 to 2000"
- "Counterfactual analysis shows the contributions of performance-pay incidence and job relatedness are about 42% and 26%"

My best guess is the above implies that (based on their model and data) performance-pay accounts for a total of 0.42*0.92*0.8*0.67=20% of overall wage inequality. So "the least popular explanation for growing income inequality, namely that we measure productivity better than before, and reward it accordingly" does not seem to explain a big part of the overall puzzle?

I could be wrong as I'm just reading off the abstract and an earlier version of the paper.

(The abstract also says "Counterfactual analysis shows the contributions of performance-pay incidence and job relatedness are about 42% and 26%, respectively, both higher than that of job-specific productivity." Seems like they are looking at the impact of performance-pay on top of job-specific productivity, measured in some way. But I can't understand better what this means without reading the paper further.)

Respond

Add Comment

Could it not be that the increase in wage disparity within an occupation could equally be the increasing success of a certain group of employers in finding employees that will work for less? Thus you might have large institutional employers that pay possibly historically higher wages (maybe still unionized) and disrupter employers who startup new companies or perhaps utilize new modes (employees as contractor) to drive down the wages that they have to pay...

Anecdotally in my own work experience, I never saw an employer rewarding the super workers...only finding ways to reduce the pay of the average or lesser workers. And yes that included at least one case of a 10x (100x?) programmer. In order to get paid significantly more than the 1x people, he had to quit and start his own company.

"Anecdotally in my own work experience, I never saw an employer rewarding the super workers...only finding ways to reduce the pay of the average or lesser workers. And yes that included at least one case of a 10x (100x?) programmer. In order to get paid significantly more than the 1x people, he had to quit and start his own company."

I think that might actually be fundamental to the metrics of a large, older company. Generally speaking larger old school companies don't often have a very significant difference in pay ranges for their employees. And essentially all pay outside of management is capped at a relatively low rate. Nor will the other employees "allow" their 1% co-workers to be compensated at a much higher level. And you aren't going to make 99 programmers unhappy to make the 1 guy happy.

Clearly there's a difference for some of the new companies, Google, Apple, etc, but they are a small piece of the big picture.

About the only way to break out of that model is to become a contractor for a large company. That's a model where an employee who doesn't want to manage (and it's a misuse of his abilities) can turn around and get 2-3x his previous wages if he's got very high end skills.

The other approach is to quit an form your own company as you mentioned. But then you are likely going to have to take up a larger management roll in that case.

>Nor will the other employees "allow" their 1% co-workers to be compensated at a much higher level. And you aren't going to make 99 programmers unhappy to make the 1 guy happy.

IDK. My experience is that reductions in performance-based pay are coming from senior management, not other employees (who tend to be grumpy about that change)

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Yes Tyler lets pick out the paper that confirms my priors and ideology.

Respond

Add Comment

Respond

Add Comment