Articles Posted in Benchmarks

Power-law distributions rarely crop up in lunchrooms and hallways. Lawyers often feel uncomfortable with mathematics, let alone something as esoteric as a function with an exponent. But reluctance and distaste does not change the fact that a number of occurrences that happen in legal departments have the frequency structure described by the omnipresent power-law function.

So I wrote an article about power laws and the National Law Journal courageously published it. I invite you all to liberate your mathematician and consider what it explains.

Total legal spending as a percentage of revenue (TLS/Rev) leads the metrics pack in terms of importance (See my post of Dec. 5, 2007: stability of the ratio over a decade; and Jan. 12, 2009: benchmarks over time with 8 references.).

https://www.lawdepartmentmanagementblog.com/median-benchmarks-for-comparables-stay-quite-stable-over-periods-of-five-or-more-years/

Longitudinal data from a law department benchmark survey appeared in a study recently of a law department I assisted. The data, for the five years from 2004 through 2008, arrayed total legal spending as a percentage of total revenue for the law department’s industry group, revenue supported ($3-6 billion), and number of attorneys (11-25).

A recent post describes a multi-column chart that color codes a law department’s metrics against key benchmarks (See my post of Sept. 22, 2009: three comparisons on nine benchmarks.). It reminded me that I have long felt the most meaningful benchmark comparisons to be with companies in the same industry. The data described in that post gave me an opportunity to see how well benchmark metrics correlate to each other across industry, revenue, and number of in-house attorneys.

As to total legal spending as a percentage of revenues, this company was median in its industry, second quartile in its attorney size group, and first quartile among companies of similar revenue. A similar dispersion across three columns occurred with inside legal spending as a percentage of revenues and outside legal spending as a percentage of revenue. Hence, for three basic spending metrics, the three groups showed very different benchmark figures. The correlation was low.

At the other extreme, the three groups – industry, revenue, and headcount – must have had very similar figures for inside and outside spending per lawyer because this company was in the same quartile (column) on all three. Note that each of those benchmarks is normalized by attorney count.

Connoisseur of benchmarks that I am, I’m drawn to what we don’t know (See my post of Dec. 7, 2008: four missing metrics and my article; and May 20, 2009: time spent by in-house lawyers that is non chargeable is an unknown metric.). In the category of data missing about external spend, here are 10 figures on my benchmark-bucket wish list. If you know some of these, I would appreciate hearing from you.

1. Blended rates by size of law firm (See my post of Dec. 5, 2005: blended billing rates with 7 references, fixed fees.).

2. Effective hourly rates of law firms that account for 75 percent or more of fees paid (See my post of March 9, 2009: effective billing rates with 9 references.).

During a recent consultation, I reviewed a chart that covered metrics for nine basic benchmarks. The chart had five columns, the first being the metric, such as outside legal spending per lawyer. The next column was 1st quartile, then 2nd quartile, 3rd quartile and 4th quartile.

The legal department had put a colored circle in the middle of a column to show on that particular benchmark where it stood by quartile. For example, on inside spending per lawyer this department was in the fourth quartile so the colored circle was in the middle of the rightmost column; on total legal spending as a percentage of revenues it was at the median so its colored circle was on the line down the middle (between 2nd and 3rd quartile).

The chart went further. Each metric had color-coded a circle in a column to show the legal department’s match-up to the median of its industry group, group of companies of roughly comparable revenues, and group of companies with roughly the same number of attorneys. All in all, an impressive and effective display of quantitative information.

Serengeti has found consistently that the law departments in its survey have spent relatively more of their total legal spend on the law department rather than on law firms. The data in ACC Docket, Vol. 27, July/Aug. 2009 at 18, states the trend in terms of the ratio of median spending on outside counsel to median spending on law departments. The decline goes from 2.0 in 2004, to 1.8 in 2005, to 1.57 in 2006, and to 1.29 in 2007.

I converted those ratios to percentages of total legal spend going to outside counsel. The decline conveyed that way seems less dramatic than the decline in the ratio of medians. Percentages of total spending spent outside declined from 66.7 percent in 2004, to 64.3 percent in 2005, to 61.1 percent in 2006, and to 56.3 percent in 2007.

My only point is that how data is conveyed makes a difference. Here, the medians state a 36 percent drop over the four years, whereas the percentages state a more modest 16 percent drop. Whichever, the trend is unmistakable and it is good that Serengeti points it out.

The scattered distribution of posts here on statistics has reached the point where a hyperpost is in order. The accumulation so far consists of seven metaposts with a total of 84 back references (See my post of April 5, 2009: Bayesian statistics with 6 references; Feb.13, 2008: correlations with 16 references; Dec. 31, 2008: regression statistics with 6 references; Sept. 9, 2009: rolling averages with 8 references; Aug. 10, 2009: standard deviation with 9 references; Jan. 20, 2007: statistics with 28 references; and Sept. 13, 2009: statistics since Dec. 2007 with 11 posts.).

Here are 10 totally free courses on statistics you can take on your own time to help you learn, improve and hone your stats knowledge (See my post of Jan. 20, 2007: statistics with 28 references.). Whether you just want to brush up on your statistics knowledge or learn something new about it altogether, you don’t have to spend lots of money and time taking a class at a nearby college. Instead, take some free courses from some of the most prestigious institutions in the U.S. instead.

1. Applied Statistics: This course from MIT will teach you the basics of statistics and data analysis, but you’ll need to have a little familiarity with calculus and linear algebra beforehand. (MIT)
2. Introduction to Probability and Statistics: If you don’t remember much from your stats class, consider this course as a refresher. It’ll cover the basics from probability models to linear regression. (MIT)

The technique of averaging figures for set periods of time, such as for the previous three months, produces what is called a rolling average. Several posts on this blog have applied that methodology (See my post of Jan. 6, 2006: rolling averages help detect longer-term trends; March 9, 2007: rolling forecasts for budgets; Oct. 29, 2007: base discount level on rolling average; Jan. 13, 2008: sizeable matters can use rolling forecasts; Jan. 21, 2009: quarterly updates of budget at JDS Uniphase; Nov. 21, 2008: performance against budget, on rolling basis, determines assignments of work; May 3, 2009: burn rates expressed in rolling terms; and Aug. 11, 2009: just-in-time budget method.).

The Intellectual Property Owners Association (in a report issued in May 2009) asked its members about the number of applications for new U.S. patents they foresaw in 2009. Are you surprised that 29 percent thought they would increase, 41 percent thought they would decline, and 30 percent thought the number of filings would remain unchanged? This finding is in the ABA J., Vol. 95, Sept. 2009 at 15, and it caused me to remember a similar distribution of answers when lawyers are asked to predict the future.

The three-hump pattern showed up in the survey on Legal OnRamp last year where Aric Press and I asked about predictions five years out. On several questions, roughly one-third of the respondents saw an increase – whatever the question, one-third say a decrease, and one-third saw status quo.

Unsurprisingly, when survey respondents have only shallow knowledge, the future is a toss-up. For those kinds of questions, a balance between up, down, and sideways is quite plausible. Try it on the stock market.