I promised a more substantive gloss on the AmLaw 100 results, and, while I have not run this by any Ph.D.'s in statistics, I think it tells a fascinating and possibly powerful story—which, more importantly, might even accord with some of the hypotheses and intuitions floated hereabouts on the trend of consolidation among the AmLaw 200.

First, eyeball this:

What, you may rightly ask, is going on here?

Let's back up.  One of my hypotheses, articulated when I first posted this year's results, is that "the rich are getting richer."  By this I meant to suggest that the leaders of the AmLaw 100 pack are pulling away from the laggards at an accelerating rate.  So far, most of my basis for surmising this has been based on anecdote, isolated "data points," and caffeinated or lubricated conversations with friends and colleagues about Big Trends for the Next Ten Years.

Then it occurred to me that if there is something to this hypothesis—if the AmLaw 50, say, are gaining ground on the AmLaw 51-100—then the year over year revenue growth might reveal something.  So on this graph I calculated the year-on-year revenue growth for the AmLaw 100 for the past three years (yielding two one-year-over-one-year series and a two-year series).

So what do we see?  I would argue:  Striking visual evidence that the hypothesis is correct.   For each of the three series, the rate of increase in revenue is higher for larger firms than for smaller.  

I performed one other test.  Using Excel's "forecast" function, which extrapolates a series out to a future data point, I asked what revenue growth would look like for AmLaw firm #200, and these are the almost shocking results:

• Based on an extrapolation of the 2004 vs. 2003 series, firm #200 would have revenue "growth" of -7.18%.  (Yes, that's a minus sign.)
• Equally damning, based on an extrapolation of the two-year series (2004 vs. 2002), that same firm would scarcely enjoy any net revenue growth at all for the two years:  Specifically, a charitable-to-call-it-anemic 0.12%.

I anticipate copious emails from the incredulous, the triumphant, and the Ph.D. statisticians.

---

Methodology Update: Posted by Bruce at 8:45 am 30 June 2005:

You may be wondering whether I tracked the revenue-growth performance of each individual firm in the AmLaw 100 during this period or whether I analyzed the overall composition of the group.  Answer B. 

In other words, my interest was in "what it takes" to stay in the AmLaw 100 at a macro level, not whether the strategies of individual firms were, relatively speaking, successes or failures.  So, for example, firm #10 in 2004 was Sullivan & Cromwell; in 2003 it was Kirkland & Ellis; and in 2002 it was McDermott, Will & Emery.  My data point for "firm #10" on these time series, therefore, actually represents the performance of three entirely different firms. 

But since my goal, as noted, was to analyze what it takes to be (in this case) "Firm #10," this methodology seemed the correct one.

Other questions?