I had an interesting
conversation with Prof.
Bill Henderson,
who recently authored an
empirical
study of single-tier versus two-tier partnerships in
the AmLaw 200.   I summarized his presentation in an
earlier
post.  Essentially, Bill’s paper found that:

• two-tier firms experienced lower profits per
  partner than single-tier
  firms, adjusting for all pertinent variables, in all market segments;
• two-tier firms nevertheless had higher leverage; and
• two-tier firms were less "prestigious" (based on Vault surveys)
  than single-tier.

Bill is willing (indeed, eager, to hear him tell it) to have the
thrust of his paper scrutinized by AmLaw 200 lawyers.  Yet,
Bill reports that many of the skeptical comments he’s received
from lawyers are at odds with what the data actually shows. 

This is where it gets interesting:  “To
what extent,” Bill
asked me, “Is my interpretation of this data a law firm analogue
to Moneyball?”

For those of you who don’t follow
baseball (which I don’t, really, until September), or who don’t
follow the writings of Michael Lewis (which I do, passionately),
Moneyball is the title of a renowned book by Michael Lewis
that chronicled how Billy Beane, the general manager of the Oakland
Athletics, used detailed statistical analysis to identify inefficiencies
in the market for baseball talent.   Specifically, Moneyball
elaborates on how Beane decided that certain factors major league teams
typically paid very very good money for, based on scouting reports and
other traditional information sources, were simply not cost-justified based
on how players with those attributes performed.

In other words, Beane identified a disconnect between the conventional
wisdom and what the statistics on player performance actually
showed.  As a result, he was able to assemble consistently
winning Oakland A’s teams for years on a relative shoestring budget.

Beane’s reward?  Scouts, other team managers, the baseball
press, and other baseball insiders sneered at Beane’s
numbers-driven approach even after the A’s fielded a championship
caliber team on one-third the budget of their large market rivals.  (Lewis
discusses the Billy Beane story in this excellent
NPR interview.)  The scouts et al. couldn’t contradict
Beane’s data (baseball, as we know, is as data-intensive a sport as there
is); they could just denigrate his approach, without offering an alternative
approach of their own consistent with the same data.

The advantage of statistical modeling
(a technique that is utterly mainstream in finance and biomedicine)
is that we can go beyond well-reasoned theories—the lawyer’s
greatest strength—into the realm of falsifiable hypothesis.

Here is a simple example.  Bill
asks, “What are the determinants of a firm having one versus
two or more partnership tracks?”  Using multivariate
regression to predict the tier structure, Bill includes four variables
(i.e., possible determinants) in his model:

1. Percentage of lawyers in New York .  Single-tier
   status may be influenced by cultural factors that are more common
   in New York. After all, New York still has a disproportionately
   large concentration of single-tier firms.
2. Firm size.  As a firm gets bigger, a two-tier
   structure can improve the monitoring of nonequity partners and
   reduce admission mistakes into the equity tier.
3. Profitability.  Lower PPP presumably puts pressure on the
   firm to limit the number of equity partners, thus necessitating
   a non-equity track.
4. Prestige.  Firms with lower indices of prestige have
   a harder time (a) attracting clients based on firm reputation,
   (b) and recruiting capable associates and laterals. A nonequity
   tier can thus reduce harmful attrition and consolidate the
   power of rainmakers—who might otherwise leave the firm.

Remarkably, prestige, as measure by
the Vault
rankings, was the only variable that emerged as a statistically
significant predictor of tier structure (and it is highly statistically
significant—less than a 1 in 20,000 chance that the pattern
occurred by random chance).  The other three theories had
NO empirical support.

So are most lawyers like the disbelieving
scouts in Moneyball?  In the Adam Smith poll on switching to
two-tiers, the most
common reason for switching to two-tiers is “To retain valuable
associates we would otherwise have had to lose.” 

But what’s driving this perception?  Prestigious
single-tier firms, like Cravath or Sullivan & Cromwell or Covington & Burling,
are—to judge by their behavior—unconcerned
about this cost.  To the contrary!  Another of Bill’s
findings is that, at a very high level of statistical significance,
every rating of "associate satisfaction" (likelihood of staying
two years, "family friendliness" of the firm, transparency of firm
finances, communications with partners, straight talk about career
prospects) is strongly negatively correlated with profits
per partner. 

Or, as one of my correspondents succinctly put
it:  "The more I’d like to be partner at firm X, the less
I could stand being an associate there."  Precisely.

So why does any associate put up with this?  In hopes of winning
the partnership "tournament" in a very prestigious firm.  In
contrast, a less prestigious firm may need a non-equity tier to
mitigate harmful attrition.  The non-equity tier provides more
of a "lifestyle" choice to associates who would wash out of single-tier
firms on quality or productivity grounds, or who simply don’t have
the client-development skills needed in any AmLaw 200 firm.  Why
don’t firms solve the attrition problem simply by promoting all
excellent technicians to equity partner?  Obviously, because
that would upset the firm’s
financial ratios and impair the loyalty of its rainmakers. 

Citing
these dynamics, Bill claims that non-prestigious single-tier firms
are “inherently
unstable,” and thus Bill believes that his
theory explains the massive migration to the two-tier format
over the last two decades.  (In 1985, essentially the entire
AmLaw 100 was single-tier; today, 80% of the [expanded] AmLaw 200
is two-tier.)

So we return to the Moneyball question:  If you agree with
Bill’s theory, let me know.  If you disagree, also let me know—but
tell me what your alternative theory is for the two-tier migration,
and, most importantly, make it comport with the existing data. Professor
Henderson is more than willing to test any alternatives.