So far in Inequalityfest 2011 we’ve focused largely on moral and ethical issues, as well as on the distinction (if one can be made) between inequality of opportunity and inequality of outcome. These are very important issues, but I’m interested in one that I think is overlooked.
Lets assume an ethical model where, above ensuring a minimum threshold of frugal comfort, we aim to maximise overall production 1 by ensuring that we are somewhere on the production possibility frontier – thus maximising human welfare. Furthermore let us also assume that output is maximised by allocating resources by way of complete markets (not “free”, but lacking imperfections from any source, government or laissez faire) which returns to each input the contribution it makes to output. Here, it is “fair” when the income received by a given person reflects their productivity.
There’s no fundamental reason to adopt this ethical model, but I think this reflects the implicit assumptions of many people, even if they haven’t thought about it directly, and something similar is dubbed the “just deserts“.
When we look at workers we see great disparities in income. Let focus on full time workers. The lowest paid full time workers receive $29634 a year. The average 2 full time worker receives $65488. The average base rate for CEOs is $2.05 million but including bonuses has a median of $5.2 million and can be found as high as $16.16 million for Ralph Norris.
Given the assumptions we have about allocation in complete markets and maximising output we now have two possible hypotheses.
a) Markets are complete, or at least close enough for intents and purposes and this pay reflects worker productivity. Thus the average CEO, by base pay is 69 times more productive than a minimum wage worker, and 31 times higher than the average worker. By full compensation he (almost certainly a he) is 99 times more productive. Ralph Norris contributes over three times as much to the Commonwealth Bank as does the average CEO to their company.
b) Markets are not complete. Salaries do not indicate productivity. Additionally, the lack of complete markets means production is not maximised and therefore neither is human welfare.
How do we distinguish between these? It’s very difficult to measure productivity, especially disentangling the effects of one input (or even defining them in the case of capital). Outside the Cobb Douglas framework there’s also the issue of disentangling productivity in labour from it’s institutional and social settings, as I incoherently discussed here.
But we can guess which seems more probable, and I feel it is far more likely we live in world where b) is true. Whilst I am happy to accept there is variation in human productivity, you’d have a hard time convincing me that Ralph Norris was 50 times as productive as a minimum wage worker, let alone 500 times. In fact, the only reason to even remotely think the latter is if we assume that markets are complete, an assumption that should not be made lightly. The burden of proof is on someone who claims that the average CEO is at least 31 times more productive than the average worker, rather than the person who claims that the observed inequalities result from other factors, including incomplete markets. And if the markets are incomplete, output and welfare is not maximised.
So even people with no concern about equity need be concerned with inequality as a sign that something else is wrong, and aggregate output is suffering. And this is before we consider that inequalities of opportunity may prevent the maximising of productivity through inequal access to education.
I am fully aware that I have chosen the most egregious examples of unproductive pay, that the incompleteness in the market for CEOs is clear due to the principal-agent problem in corporate governance (and thus the victims are most likely shareholders who are themselves comfortable). But the principal is valid at other levels. If we want to maximise welfare through output we should check if the magnitude of the disparity between incomes is plausibly explicable by productivity (examples are easily found), or whether we should look for market incompleteness that is preventing maximum production. The alternative is to assume complete markets to vindicate whatever incomes are being received.
1 Given the difficulties in measuring output, also let it be assumed that this refers to the market price of goods (GNP) rather than unobservable utility.
2 Irritatingly I can’t find out whether this is median or mean.
I think it is a very valid point (if it is your point) that ‘productivity’ and output are maximised through increasing equality of opportunity, and that this has implications for the acceptable range of equality of outcomes.
I would like to see one of this blog’s inequalityfests considering how to square what appears to me to be a circle: that a certain and probably quite high level of inequality of outcomes appears to be a necessary feature of the kind of society that improves wellbeing, as opposed to remaining stagnant.
In particular, Australia can probably afford to screw it up a fair bit either way. What should developing countries whose most pressing problem is aggregate (or perhaps average) well-being do?
Also, I will guess that fn2 is mean and that the median is lower.
Finally, and you can delete this sentence, the principle in the second sentence of your last paragraph is not a person.
Richard, great post.
That average earnings figure is the mean. The median for full time employees was $52000 per year in August 2009 according to ABS 6310.0 (Table 6). The median for full time employees was $56160 in August 2008 according to ABS 6306.0 (Table 6). The two surveys differ in a few ways (the first is an employee survey, the second an employer survey, etc). A new median earnings figure will be released in the next few weeks.
While I agree there is a principal-agent problem in CEO pay (ie, the same people would do the same jobs for much less) but it is actually quite plausible that in large companies CEO-level decisions make far more than their salaries in differences to profits.
Of course they do not personally make the good or service that ultimately makes the profit, but they set the context in which goods and services are produced.
To get some insight into one reason some people are increasingly paid much more than their marginal product, have a read of this book by Robert Frank. For a shorter and simpler exposition of the thesis, try this polemic by Krugman.
a certain and probably quite high level of inequality of outcomes appears to be a necessary feature of the kind of society that improves wellbeing, as opposed to remaining stagnant.
That’s not at all what the empiric wellbeing research shows – your society has to be awfully rich to adequately compensate you (in terms of wellbeing) for it being unequal. That suggests that the very high inequality aversion (ie being willing to trade off a lot of national income to avoid gros inequality) implied by progressive tax scales and welfare states is in fact just what is needed to keep the bulk of people happy.
Maybe it depends on what you mean by ‘quite high’?
Andrew Norton: “While I agree there is a principal-agent problem in CEO pay (ie, the same people would do the same jobs for much less) but it is actually quite plausible that in large companies CEO-level decisions make far more than their salaries in differences to profits. ”
Sol Trujillo’s smelly tenure as Telstra CEO occasioned a one-quarter drop in the company’s value- that’s $25 billion. Not bad going for four years work. Was his house and other assets seized? Was he subject to any type of penalty at all? Feck no, he pocketed $30 million and fled the country on short notice, not unlike a gaggle of Latin American kleptocrats during the 1970s. The Trujillo case is not unusual either, in fact it is a pattern so often repeated that we no longer even notice the stench.
[…] than post another lengthy contribution to what Richard Tsukamasa Green has dubbed Inequalityfest 2011, I thought I’d post a few charts on the distribution of […]
Yes, Mel, but I don’t think that AN was saying that only people who happen to be CEOs are capable of making those decisions…just that the decisions of those who happen to be CEOs can make a very big impact.
I doubt you disagree, given your comments on Trujillo…;)
Patrick – Yes, Trujillo shows that CEOs matter a lot. For good or ill, most other employees cannot do as much good or as much harm to a firm (rogue traders in banks are the only obvious exception).
While the golden parachutes are larger than required, there is logic given the CEO job market. Boards are pretty ruthless with under-performing CEOs, and tenure at the top is often short. From the CEO’s perspective, it reduces the risks involved. From the board’s perspective, it makes it easier to get rid of CEOs without too much fuss.
Boards are pretty ruthless with under-performing CEOs …
Err, up to a point Lord Copper.
It depends which way they’re underperforming. Sol’s disastrous tenure lasted four years – yet anyone with a modicum of nous could foretell what must happen to a monopoly wholly dependent on govermnent regulation, if its CEO kept insisting on pushing those regulations to the limit and then rubbing governments’ faces in their own impotence. That’s why I never bought any Telstra shares – I never predicted the NBN, but did predict that sooner or later some minister would find a way to fix Telstra’s wagon real good.
Of course it also depends what you mean by “ruthless”. I wouldn’t mind some of the separation packages those poor CEOS have suffered at the hands of ruthless boards. Which is Mel@7’s point – once you reach the upper heights, whatever happens you come out way ahead.
[…] (blog) Rather than post another lengthy contribution to what Richard Tsukamasa Green has dubbed Inequalityfest 2011, I thought I’d post a few charts on the distribution of […]