Incentives at UBS and in General - 8/22/08

Incentives at UBS and in General

Posted: August 22nd, 2008, by Andy  E-mail this post

http://speroconsulting.com/?p=645

As we mentioned in the previous post, The Wall Street Journal’s breakingviews.com column today discusses newly proposed performance measures at UBS: aptly titled “UBS Seeks New Incentives.”  (It is at breakingviews.com, not wsj.com.)

We certainly disdain UBS’s current approach of rewarding performance
with shares, but rather than restate our criticism of the (general) use
of universal performance measures, we point the reader to our essay One Performance Measure to Rule Them All. 
Of course, those looking for a criticism of the converse, i.e., a
multiplicity or overuse of performance measures will find that, too. 
That one is called If One is Bad, Then 400 Must be Good.

Today’s column of interest discusses something called “phantom
equity.”  We’re not sure how many jokes could be written to define that
phrase, especially during the continuing financial crisis, but we have
no desire to offend shareholders at most of the larger firms; so, we
will skip it.

Anyway, phantom shares seem to be the product of some measurement of
divisional earnings, with all the attendant accounting assumptions and
allocations, multiplied by few other arbitrarily chosen numbers,
including some type of earnings multiple that comes from who knows
where.  (We do like the attempt to equate three made-up divisional
values to the overall market value.  To us, it sounds like solving one
equation with three unknowns.  We can vaguely hear someone say, “I
remember that Mrs. Pfeiffer said that we can’t solve one equation with
two unknowns, but she never said anything about three unknowns; so,
let’s keep trying. I don’t care if the answer keeps changing.”)

Under such an earnings-based scheme, it woud seem that once the
parties—corporate and divisional management—agreed to those
multipliers, divisional employees would be rewarded based upon
divisional performance.  Unless, each employee is performing an
identical job so that his or her individual performance is nearly
perfectly measured by his or her share of divisional income, the new
scheme is essentially no different than the old, share-based one;
so, we once again refer interested readers to our essay One Performance Measure to Rule Them All, which discusses both cases.

By disaggregating the divisions and switching from equity to earnings, the firm’s managers may possibly
reduce the risk imposed upon certain employees—we can’t be sure of that
unless we know the relationships (think correlations) between and among
the different measures.  In the process, however, they trade the
possible reduction in risk for the increased capacity to behave
subjectively: they, themselves, not their employees. Thus, while
decreased risk may permit lower risk premia and
thus reduce expected bonuses and increase expected profits, the
increased subjectivity usually increases compensation costs and
demotivate emplyees who become wary of senior management.

This subjectivity may be obvious or not.  It maybe in the form of
the opportunistic use of cost allocation (for centralized and shares
services and resources) to reduce a particular division’s income. 
Without the use of effective commitment mechanisms by management, that
arbitrariness usually increases the level of distrust within the firm. 
(We recall mentioning something like that in a slightly different
context in If One is Bad, Then 400 Must be Good.) 
Note that such schemes may also either directly or indirectly introduce
a level of competition within the firm, but that is not always a bad
thing.  They will also likely make the firm more political, and that is
rarely a good thing.

Thus, the use of phantom equity may leave the the firm in similar
situation as the current scheme but with possible additional problems,
too.  A very rough analogy: think of a single global performance
measure as a banquet food, say, chicken with some unknown white sauce
on it. It probably doesn’t map to anyone’s taste buds.  Adding a few
more items may satisfy a few, but it can lead to more problems and
higher coordination costs and possibly illness if that food is prepared
incorrectly or sits too long.  By comparison, a restaurant permits much
closer mappings to tastes than banquets.  (That is why few banquet
halls operate as restaurants.)  Restaurants may be more expensive, but
customers usually think they are worth it, ergo utility is maximized. 
Think of us as a food critic.

So, what is the solution for UBS?  The same as with any other firm. 
We follow an algorithm by asking: how does the person’s actions and
decisions affect wealth creation?  What signals are available of those
actions and decisions? What are the characteristics of those
signals? And, how should the signals be weighed to effectively evaluate
performance?  Remember that is done to motivate effort and not for its
own sake.

At its core, our approach is statistical but considers qualitative
factors, too.  Saying any more would betray firm secrets, and
needlessly destroy our human capital.  However, given this brief
description, we must add: what are the chances that shares or phantom
equity would be the optimal choice for each semi-autonomous employee in
each of the three divisions?   Seems that it would be rather
remarkable, doesn’t it?

Finally, let us note that this post skips over a whole host of
related issues like asymmetric information and its fraternal twins
moral hazard and adverse selection.  These problems create the need for
performance measures in the first place.  Double finally, we can’t
resist mentioning that we think incentive pay is quite overused and
thus very costly to corporate America and its shareholders.