There are reasons everyone talks about executive compensation, but nobody does anything about it.

Regarding the latest Obama administration initiative regarding compensation on Wall Street generally, Matthew Yglesias agrees with those who accuse the administration of not doing enough and approvingly cites a Brad DeLong post unfavorably contrasting Wall Street compensation schema to those of Silicon Valley ventures:

The engineers of Silicon Valley startups are significantly smarter and work a lot harder than do the traders of Wall Street. Some of the engineers of Silicon Valley make fortunes: they are compensated with relatively low salaries and large restricted equity stakes in the startup businesses they work for, and so if the businesses do well they do very well indeed--in the long run, in the five to ten years it takes to assess whether the business is in fact going to be a viable and profitable going concern. And the engineers of Silicon Valley have every incentive to use all their brains and all their hours to make their firm viable and successful: they get their cash only at the end of the process. They don't get big retention bonuses if they stick around until the end of a calendar year. They don't get big payouts if they report huge profits on a mark-to-market basis.

The traders of Wall Street, by contrast, get their money largely up front. If the mark-to-market position is good, they get paid--even though it is almost surely the case that nobody has tried to actually sell the entire position to somebody else. If the strategy produces short-run profits, they get paid--even though not nearly enough time has passed for anybody to be able to assess what the risks involved in the strategy truly are. They get "traders' options"--we claim that we have made you a lot of money, we claim that the positions and strategies we have left you, the stockholders, with are sound, we claim that we have correctly managed our risks--but we are not interested in putting our own personal money where our mouths are but instead we insist on getting our fortunes up front.

The failure of the major institutions of Wall Street to adopt Silicon Valley compensation schemes in the 1980s and 1990s was always a great worry to regulators and policymakers...

On this topic, attention must be paid to former investment banker The Epicurean Dealmaker, who has been writing about this quite often over the past 18 months.  He has made two main points which haven't been reflected sufficiently in coverage of Wall Street paydays:

1) The major banks already were giving out a very large share of compensation in restricted stock and other instruments dependent on the value of the firm - i.e., the "trader's option" asymmetry was less of a problem than popularly assumed.

2) More importantly, the heavy weightings of Wall Street compensation in restricted instruments still hasn't had, and won't have, the desired effects for the following reasons:

Now, given the differing motivations and incentives of pure traders and pure investors, there are really only two proven ways for the investor to control his trader's assumption of risk. The first is close supervision, monitoring, and control: the investor limits what securities and positions the trader can assume, he monitors daily trading activity and marks positions to market daily, and he intervenes when things go off the rails. This is the simplest model, and it is the one that used to obtain back in the dark ages before investment banks became large, externally funded, global trading houses. Yves Smith points out that this is the model the old Goldman Sachs partnership used to use, before it went public. There really is nothing better to keep some young Turk under control than some grizzled, grouchy old bastard seated next door who used to trade the very same markets you do and whose personal partnership stake you are trading for a living.

This model, as we have seen over the past 18 months, begins to break down when the span of control gets too broad and the chain of supervision becomes too attenuated, like it did in today's huge global banks. Complicated Value at Risk models and professional risk managers are no match for crafty and devious traders, particularly when the money they are trading belongs to some absent, passive institutional investors whom no-one gives a damn about. Markets are too fast today, and securities are too recondite, to make supervision at a distance very successful.

The second way for investors to control their traders' assumption of risk is to make them investors, too. Make a trader eat his own cooking, so to speak, and you will see a marked change in how he handles and assumes risk. The trader will supervise himself. After all, it's his money too. Many hedge funds do this, by paying their important traders in shares of their own trading book, or the overall book of the firm. Investment and commercial banks have been doing this for some time, too, by paying traders--along with everyone else--substantial portions of their annual compensation in long-vesting restricted stock of the firm.

The problem with this method is twofold. First of all, you need to make sure that enough of the trader's compensation and total net worth is tied up in this way; otherwise, he will just view unvested compensation as "house money" to play with, and he will have little incentive to care. The temptation to swing for the fences, or assume dangerous risks, will overwhelm any proprietary instincts for preservation of personal capital. Second, even if the trader has a substantial portion of his wealth tied to the overall results of his firm, the firm cannot be too big in relation to his stake, or he will feel that nothing he does will matter anyway. The rubber band tying his personal trading performance to the price or value of his employer's equity will be too elastic and contingent on the actions of others to act as a real incentive. This is the problem faced by large investment banks, where a trader holding even $50 million in unvested stock feels that nothing he can do--good or bad--will make a difference to the price of Citigroup stock.

(Emphasis addded.)

Those points are some of the best arguments I've seen as to why the extensive employee equity-ownership of the major investment banks - and the most notable examples were none other than Lehman Brothers and Bear Stearns - did not provide sufficient alignment of interests to prevent franchise-destroying risks.  And they also point to the most important term in the Brad DeLong excerpt above: not "Wall Street" or "Silicon Valley," but rather "startups."  The compensation system for a startup with few employees simply does not scale up in the same fashion to an institution the size of a Wall Street firm.  On his sadly-discontinued blog, famed tech entrepreneur Marc Andreessen - who knows the differences between startups and big companies - points this out:

This is why stock options work so well in startups -- and the fewer people in a startup, the better stock options work, since when there are only a few people in a company, it's usually crystal clear to each person how her work will impact the value of the company.

...As a company grows, stock options and other forms of equity-based motivation become less and less useful as an incentive tool, since it becomes harder for many employees in a large company to see how their individual behavior would have any effect on the stock price of the overall corporation. So, more tactical incentives kick in, such as cash bonuses.

(Emphasis added.)

In short, there isn't an easy solution for how to structure compensation at Wall Street firms - and if there is, they'd surely like to hear about it.  But as far as general principles go, I think Andreessen has the right idea in the post linked above:

The design of tactical incentives -- e.g. bonuses -- is a whole topic in and of itself, and is critically important as your company grows. The most significant thing to keep in mind is that how the goals are designed really matters -- as Mr. Munger says, people tend to game any system you put in place, and then they tend to rationalize that gaming until they believe they really are doing the right thing.

I think it was Andy Grove who said that for every goal you put in front of someone, you should also put in place a counter-goal to restrict gaming of the first goal.

So, for example, if you are incenting your recruiters on the number of new employees recruited and hired, you need to also give them a counter-goal (and tie it to their compensation) that measures the quality of the new hires three months in. Otherwise the recruiters are guaranteed to give you what you don't want: a lot of mediocre new hires.

One of the great unwritten Silicon Valley skewed incentive stories was a major datacenter vendor in the late 90's that incented its salespeople based on bookings of long-term datacenter leases, without sufficient counter-goals tied to revenue collection or the customer's ability to pay. Sure enough, soon the company's reported bookings were heading straight up, revenue was flat, and cash headed straight down, resulting in a truly spectacular bankruptcy. The salespeople got paid, though, so they were happy.

More recently, skewed incentives in the mortgage industry -- mortage issuers getting paid based on quantity of mortgages issued, versus ability to pay -- caused many of the current catastrophic Wall Street financial meltdowns you get to read about every day.

Even engineers need counter-goals: incent engineers based purely on a ship date, and you'll get a shipping product with lots of bugs. Incent based on number of bugs fixed, and you'll never get any new features. And so on.

We want to hear what you think about this article. Submit a letter to the editor or write to