Tuesday 20 January 2015

CEO pay and inequality

The ODT's Eileen Goodwin asked me for comment on CEO pay in New Zealand; I'm quoted briefly in this story. Here's what I'd sent her; the ODT story wasn't able to include any of it but did quote from our prior phone chat. I wish she had made some use of the data I provided on top-1% earnings for context though.
We should start by looking at whether there actually has been any great increase in earnings in the top 1% relative to others, just so we have the data right.

I created this chart from the World Incomes Database. You can check my numbers at http://topincomes.g-mond.parisschoolofeconomics.eu/#database . The only difference between those figures and mine is that I averaged incomes between 1999 and 2000 for everybody, because the change to the 39% top marginal tax rate had a pile of people bring forward income to 1999 from 2000, making 1999 look super rich and 2000 super bad. Best is then just to average the two. I set everyone’s income in 1953 to be equal to 100 to get an index score, then looked at growth from then onwards relative to the 1953 baseline. That makes it really easy to compare income growth in different cohorts. Note that these are real inflation-adjusted figures. 

As you can see, there is zero evidence of any blow-out in top 1% earnings in New Zealand. Since 1953, real earnings in the top 1% are up by 50%. But real earnings for the bottom 90% doubled.

This is further confirmed by the Ministry of Social Development’s recent report. You can check it at the link; I’ve pasted the relevant table below.
"Overall, there is no evidence of any sustained rise or fall in inequality in the last two decades. The level of household disposable income inequality in New Zealand is a little above the OECD median. The share of total income received by the top 1% of individuals is at the low end of the OECD rankings.  
Income inequality in New Zealand, 1984 to 2013 HES


1984
1994
2004
2009
2012 & 2013 for HES, 2010 & 2011 for tax records
Household disposable income, adjusted for household size … data from sample surveys (HES)
Gini x 100 (trend-line)
26.6
32.5
32.9
32.9
32.9
Share ratio, D10 to D1
6.1
8.2
9.1
8.6
8.3
Share ratio, Q5 to Q1
4.1
5.1
5.5
5.4
5.3
Share ratio, D10 to D1-4 (Palma)
0.92
1.21
1.31
1.29
1.27
Percentile ratio, P90 to P10
3.5
4.1
4.2
4.4
4.2
Percentile ratio, P80 to P20
2.4
2.7
2.9
2.9
2.7
Individual market income … data from tax returns – avg of year noted and the one either side
Top 1% share
5.6
8.9
9.0
7.8
7.8
Top 10% share
28
33
33
30
30
Top 10% - 1% share (ie P90 to P99)
23
24
24
22
22
 Income inequality in New Zealand compared with other OECD countries, c 2011-2012
(%)
NZ
OECD-34 median
DNK
NOR
FIN
FRA
AUS
CAN
UK
US
Gini x 100 (trend-line)
32.9
30.5
25.3
25.0
26.1
30.9
32.4
31.6
34.4
38.9
Share ratio, D10 to D1
8.2
7.6
5.3
6.1
5.5
7.4
8.5
8.5
9.6
16.5
Share ratio, Q5 to Q1
5.2
4.8
3.6
3.7
3.7
4.7
5.4
5.2
5.6
8.2
Share ratio, D10 to D1-4 (Palma)
1.27
1.18
0.87
0.85
0.93
1.18
1.27
1.19
1.40
1.74
Percentile ratio, P90 to P10
4.2
3.8
2.9
2.9
3.2
3.6
4.5
4.1
4.1
6.1
Top 1% share – tax records
8
The latest available from 2009 to 2012
6
8
8
8
9
12
13
19
Top 5% share – tax records
21
17
19
21
21
21
27
28
36
Note:  See the main report for details about the sources for the figures in the above tables.
Eric here again now.

So, top 1% earnings here are hardly high by either a historical comparison within New Zealand, nor by any international standard. And inequality in New Zealand is middling by OECD standards.

The MSD report also points out that while wealth inequality is higher than income inequality (and data on it less reliable), what we do know about it is that the top decile’s wealth share here is about on par with France and Canada, and lower than Norway and the US. Further, when we account for that much of the country’s wealth is held in the form of mortgage-free houses by the elderly who have low incomes, the joint distribution of wealth and income is far more equal than if we look at wealth alone: a lot of people with high wealth have little income.

In my view, inequality has taken on a lot of salience in New Zealand for two main reasons. First, inequality in the US has risen substantially – though not here. New Zealand is way too quick to assume that whatever’s happening everywhere else is also happening here. Please look at the data above and help people to at least base their opinions on what the data actually says about things here. Second, the ridiculous run-up in housing prices, due mostly to Council restrictions preventing new building, has made housing really really unaffordable for a lot of even middle- to higher income New Zealanders. They then blame richer people for bidding up the price of housing – and they’re especially prone to blaming rich foreigners. It’s right to be concerned about housing costs, and especially about that housing costs eat up way too much of household incomes for lower to middle income families. But that’s not a problem with inequality – it’s a “not building enough houses” problem.

With that established, let’s look at some of the evidence on CEO compensation.
Edward Glaeser Kevin Murphy finds that executive compensation in the US is strongly related to firm performance. See the first paper in this set.
Adams et al find that Swedes with higher cognitive and non-cognitive abilities get put in charge of larger companies. Better CEOs sort to the bigger firms where their skills are more needed; CEO pay then reflects both their abilities, and the size of the firms they have to manage.
Kaplan and Rauh find that top CEOs have been gettingbigger pay packets largely because the market has gotten a lot bigger and morecomplicated: the CEO job has gotten harder at large firms, and so the gains from having the best people have gone up too.
Again, this is consistent with the view that shareholders generally do well in keeping an eye on their boards, that boards pay what’s necessary to attract top talent, and that the importance of having strongly performing CEOs has increased.
Another way of testing? What happens to a firm’s value when the CEO dies unexpectedly. Here’s Nguyen and Nielsen. In short, there are sharp drops in firm valuation (share price) on the death of a CEO. They argue that for every extra dollar of wealth that a CEO creates for the company *over and above what someone else in the job would have created*, that CEO keeps about $0.75 and shareholders get $0.25. Other papers using different measures find that CEOs keep a much smaller fraction of the wealth they create for shareholders; the high estimate here may reflect some drop in value due just to the uncertainty about who the successor will be.
If there are problems in CEO compensation, they will be specific to individual firms. If a CEO has managed to pack the Board with friendly supporters, that CEO could be getting paid far more than he or she deserves. That makes the case not for any kind of blanket rules on CEO pay but rather for making sure that corporate governance is strong: that shareholders maintain control of the firms they own, that Boards are active in monitoring performance, and that shareholders sack Boards and CEOs if they can get a better deal elsewhere.
I hope this helps.
 It's always interesting to keep track of what bits get used and which bits don't.

Update: for the benefit of those who, like Keith Ng, seem to want to assume I did something dodgy in choosing 1953 as a baseline year, go and check the original source data. The basic reporting unit changed from 1952 to 1953 so the data isn't consistent; they have 'em as different data series. I chose 1953 because it was the earliest consistent year in that dataset; I chose that dataset because it's the longest term data I've yet seen on income distribution in New Zealand. Maybe Keith would be happier with a 1984 base year because then he could be all mad about income increases for the top 1% in the late 80s; maybe somebody else would like a 1999 base year so they could be all mad about income declines for the top 1% since then. The nice thing about choosing the longest possible data series is that anybody can then truncate it to later points if they want to think about alternative base years. I'm hardly hiding the increase in top 1% incomes in the late 80s; I'm just failing to hide the prior long term decline in earnings among the top 1%.

Update2: Keith's only unhappy that I took 1953 as base year, not that I included the full data series. People can use whatever base year they want. In absence of some big reason for picking one over another, I go with the earliest; you can easily run ocular re-basing to whatever base-year's desired.

7 comments:

  1. Headlines which read "Still no change in income or wealth inequality after 30 years" don't sell papers.


    And, for the leaders of some political parties, the very rare skill is convincing people that there has been a significant deterioration in inequality. If you're in the business of benefiting from wealth redistribution, you're very literally not getting paid to believe that income or wealth inequality doesn't exist.

    ReplyDelete
  2. To reverse the title of a seminal paper in the area, "it's not how CEOs are paid, it's how much they're paid that matters" for many critics. In other words, there'd still be howls of outrage about the size of CEO pay packets even if they were all doing a brilliant job. This sort of nonsense should just be ignored--what matters is whether CEOs are paid in an economically efficient manner.

    There is precisely zero evidence that CEO pay in NZ is inefficient or self-serving. Indeed, possibly the opposite--my own work with Helen Roberts found that CEOs who sat on their own compensation committee (not exactly a characteristic of supposed good governance) actually received lower pay increases on average. Of course, that's not to say there aren't some egregious counter-examples (such as the Lyttelton port CEO), but the point is that such cases have been, to this point in NZ, atypical.

    Even if overpaid CEOs were more common, this should only matter to critics in academia and the media if they were able to show it had a significant adverse effect on economic performance. If excessive CEO pay is just a transfer from shareholders to CEOs, that's an agreement between consenting adults that's of no relevance to anybody else.

    ReplyDelete
  3. Have to say I do not think the ODT did a decent job of using the material you provided. Its one thing to quote your 'opinion' - which makes it into a game of "he says, she says" but when the data is easily available to the journalist that should be central to the argument. And it would be interesting to hear the 2 academics from Otago give their view of the data.

    ReplyDelete
  4. She had a story she wanted to tell; I was the "balance" quote, that's all. That's how it works.

    ReplyDelete
  5. Agree entirely. You should provide that paper to the folks running the hate-on-CEOs conference.

    ReplyDelete
  6. Most interesting finding of this wealth research is working Richard have replaced passive investors as the rich

    ReplyDelete
  7. Speaking as someone not well versed in the minutiae of economics - it seems a little off to set everyone's income to the same level and just compare the percentage rise from 1953.


    If a low income worker has an income of say 20K (making numbers up here to illustrate the bit that worries me) and a high income worker has an income of 100K in 1953, then a 90% rise in income for the lower person only takes the income to 38K - and 18K rise. Where as the high income worker is now on 150K - a 50K rise. It might be less in terms of a percentage, but it's still rising faster.


    Which, if you play it out, years down the line would result in low incomes being a tiny, tiny percentage of the high incomes.


    You may well be right about the lack of growth of inequality (you haven't convinced me with this work and I think blaming it on Auckland council for not building houses is just ... weak). I just don't see how using percentage increases in income illustrates the point though. Would it be better to plot the Gini scores over the same period?

    ReplyDelete