CEO pay hasn't risen much lately. By Matthew Yglesias

CEO pay hasn't risen much lately. By Matthew Yglesias

America's top executives made less in 2021 than in 2000

Monday evening, the Working Families Party posted this tweet arguing on behalf of a minimum wage increase:

1

If the minimum wage had continued to rise with workers’ productivity (like it did from the 1940s until the 1970s), today it would be over $22/hr. Instead, it's stuck at $7.25/hr and hasn't budged for 14 years. Instead, CEO pay has grown 1,460%.

The numbers are from a 2022 report by Josh Bivens and Jori Kandra for the Economic Policy Institute titled “CEO pay has skyrocketed 1,460% since 1978,” which offers two different data series for assessing CEO pay.

The 1,460% figure cited in the tweet is from the one that seems methodologically unsound to me. The other (more reasonable, in my opinion) approach finds that CEO pay has grown “only” 1,050% since 1978. This is also a very large number, so I wondered why a think tank would use a finding from a more dubious methodology — 1,050% is already a shockingly steep increase in CEO pay.

I think the answer is this: While the latter method of measuring CEO pay does show a huge increase since 1978, it shows no increase at all since 2000.

Or, rather, it shows a huge crash in the 2000–2009 period from which CEO pay has only partially recovered over the past decade. It’s another indication, in other words, that economic inequality has been falling recently. There’s a sentiment in some quarters that Joe Biden and the Democrats need to try harder to sell a positive narrative about the Biden economy, with its ultra-low unemployment and now-falling inflation. And certainly part of any presidency is touting your successes. But the EPI report and WFP tweet illustrate one reason I think Democratic presidents struggle with this: progressives are sometimes oddly committed to downer narratives, as if convincing people that everything is awful is an obvious precondition for getting them to support policy change. I think this is a pretty questionable proposition.

Two ways of calculating CEO pay

So how did EPI arrive at two different numbers for the CEO pay increase? Essentially, they count CEO pay in two different ways. Or more specifically, they count the component of CEO pay that is made up of stock options in two different ways.

A stock option means that you have the right to buy X shares of stock at price Y. If the value of the stock goes up to a price that’s higher than Y, you can exercise the options and make money. But if the value of the stock declines to less than Y, you don’t actually lose any money; you simply decline to exercise the option. And if you have options to buy a stock at Y, then the value of the options themselves (Z) would be a function both of Y and of market sentiment about the value of the underlying stock.

2

One of the ways that EPI calculates CEO compensation involves calculating the “realized” value of options (the profits that CEOs who exercised options made on their exercises) and the other involves calculating the “granted” value of the options — the imputed Z based on their fair value when granted.

The realized measure of compensation includes the value of stock options as realized (i.e., exercised), capturing the change from when the options were granted to when the CEO invokes the options, usually after the stock price has risen and the options values have increased. The realized compensation measure also values stock awards at their value when vested (usually three years after being granted), capturing any change in the stock price as well as additional stock awards provided as part of a performance award.

The granted measure of compensation values stock options and restricted stock awards by their “fair value” when granted. (Compustat estimates of the fair value of options and stock awards as granted are determined using the Black-Scholes model.) For details on the construction of these measures and benchmarking to other studies, see Sabadish and Mishel 2013.

These two methods generate different answers to the “since 1978” question, but I don’t think this is rhetorically relevant. Either way, there was a huge run-up in CEO pay in the 1980s and 1990s associated with the shareholder value revolution.

3

But since 2000, these measures give very different answers. On the basis of compensation granted, CEO pay has fallen even as wages and stock prices have risen.

I get that “CEO pay has risen a lot since 2009 but is still below its 2000 peak” isn’t a very snappy talking point, so it’s nice to come up with another metric.

But if you’re trying to measure compensation, then measuring the value of grants seems obviously correct. The grants are what the compensation is. I used to have stock options in Vox Media. I never exercised them (it turns out new media startups were a bad investment) and thus never realized any money out of it. But getting the options in the first place was a thing of value, at the time. There was bargaining over it. To not count the grants as part of the compensation Vox Media was offering would incorrectly portray the whole situation. The other problem, which EPI concedes, is that counting based on when options are exercised leaves you hostage to weird outlier flukes.

For example, Elon Musk had a bunch of Tesla options set to expire in 2022, and 2021 happened to be both an excellent year for the overall stock market and also a year that saw a big speculative run-up in Tesla. So he exercised so many options that it broke all the averages and EPI just threw him out of the data.

In 2021, Elon Musk (CEO of Tesla Motors) exercised $23.5 billion worth of stock options that would have expired in 2022. Under our “realized” methodology, this would have made his pay almost 1,000 times that of the average large-company CEO. Including him in our sample would have resulted in an increase of CEO pay in 2021 relative to 2020 of over 300% (the “average” for the sample would have been just under $100 million).

Because inclusion of this extreme outlier would have made this year’s numbers incomparable with previous years’ numbers, we opted to exclude Tesla and Musk from our sample entirely.

This is not a very principled way to handle your data, and the fact that EPI feels compelled to make this kind of ad hoc adjustment just shows that the realized compensation methodology is bad. The granted compensation method is the correct one, and it shows that CEO pay has not yet recovered to 2000 levels, while wages are up about 17%.

4

The inequality surge happened in the past

To be clear, the huge surge in economic inequality that occurred after the Reagan Revolution is very real. That was an important historical moment in the political economy of the United States and it’s totally fair to talk about.

At the same time, it happened a pretty long time ago. Most of the people who voted for Ronald Reagan in 1980 are dead now. A lot of the people who voted for Bill Clinton in 1992 and 1996 are now Republicans. The economic trends of the past 20–25 years feel like a reasonably long-term historical trajectory for any working-age person alive today. And there’s a big difference between “inequality went up a lot at the end of the 20th century and then stopped” and “inequality has been soaring relentlessly for generations.”

And if you look at the CBO’s report on income distribution, the top 1% share was smaller in 2019 than it was in 2000. Not dramatically smaller. Not small enough to have reversed the prior rise in inequality. But smaller.

And not smaller through some kind of crazy coincidence; smaller because the net impact of the Clinton/Bush/Obama/Trump ping-pong of tax-and-spend policies was to make the overall system more redistributive. Since the pandemic, a tight labor market has meant wage compression: earnings are growing faster at the median than at the 90th percentile, and faster at the 10th percentile than at the median.

The whole CEO pay exercise is kind of fake since they’re only asking about the chiefs of the 350 biggest companies which, while big, don’t make up that large a share of overall employment. But it points in the same direction as these other broader indicators. The push against inequality has delivered results, and I think it would be constructive for Democrats to talk about that more. In politics, people often offer public characterizations of their enemies. Trump-era Democrats tend to do this in terms of battling white supremacy and other forms of “hate” or abstractly in terms of “MAGA Republicans.”

I think it would be useful to acknowledge that there are some people out there who don’t necessarily have any particular problem with women or trans people or racial diversity but are mad that Democrats are helping the working class gain ground on the rich and are therefore highly motivated to cut big checks to Republicans.

It’s good to focus on real problems

Rhetoric aside, the other thing that the wage chart shows is that median wages have been falling in the Biden era. The wage compression itself is good, but wage compression in the context of falling inflation-adjusted median wages is not good.

The main reasons for this, I think, are 1) supply shocks that are now in the rearview mirror and 2) a huge surge in non-wage income that is also in the rearview mirror. People got a lot of stimulus payments during Covid that they didn’t spend because they were being cautious, and they’ve been spending that money down in the post-reopening era in a way that temporarily pushed price growth ahead of wage growth.

It now looks like we are making real progress on inflation, so I’m fairly optimistic. But that still leaves us with the reality that in an era of wage compression, you really do want to be thinking a lot about how to boost productivity and economic growth.

A big part of the point of putting all this inequality stuff on the policy agenda in the first place was to argue that the link between productivity and pay had been broken in a problematic way. And that very much did characterize the CEO pay boom of the ‘80s and ‘90s. Executives received what were, in effect, huge payouts for engaging in financial engineering that boosted shareholder returns far more than it boosted productivity or national economic performance. Precisely because this is an important point, it’s also important to acknowledge that it does not characterize the current economic situation — we have a well-functioning full-employment economy that is doing a good job of serving marginal workers.

That’s great, but it means that right now all this neoliberal-shill, productivity-boosting stuff really does matter, both for earnings at the median and also for making the full-employment economy sustainable.

1

Ideally, I would embed the tweet here, but because Elon Musk is petty, he has made tweet embedding on Substack impossible. So I’ll quote it instead, which I think is worse for me and also worse for Twitter.

2

Here’s a much more technical explanation of how options work.

3

Very generally, it used to be that CEOs were very lightly supervised by shareholders and could make a lot of decisions driven by sentimentality, ego, or whatever else. Then some legal changes and the rise of corporate raiders and leveraged buyout operations started encouraging CEOs to focus much more on delivering returns to shareholders. In exchange for being put on this tighter leash, CEOs got pay packages that were tied to share prices. Giving executives a clear mission of higher stock prices generated higher stock prices, which in turn generated soaring compensation-based pay.

4

Total compensation is, I believe, up even more than that because the value of job-based health insurance has tended to grow faster than wages — in part because compensation received in the form of health insurance isn’t taxed.