Monday, August 15, 2016

“Glass Half Full” author concedes problems with estimates of solo practitioner incomes and headcounts (updated 8/18)

Professor Benjamin H. Barton recently responded to critiques of his estimates of solo practitioner incomes. Barton does not answer the specific questions that I posed about his use of IRS data, but he generally concedes that the IRS data is problematic. 

  1. Barton wrote:

“Is it possible that the IRS data undersells the earnings of solo practitioners?  Yes, for the reasons I state above and for some of the reasons that you and Professor Diamond point out.”  

I applaud Professor Barton’s honesty.  I encourage him to acknowledge the problems with the IRS data in future editions of “Glass Half Full” and to correct his CNN and Business Insider posts.

  1. Barton wrote:

“Do I think that the IRS data are off by a factor of 3.5 or even 2?  No.”  

I encourage Professor Barton to present a revised estimate that he thinks is more accurate. Several studies that he cites for support suggest that his solo income estimates are off by a factor of approximately 2 to 3 (see below for details).

  1. Barton defends his use of IRS data on three grounds, each of which is problematic:

a. “The IRS data on lawyer earnings is the longest running data I could find and thus the best dataset for a discussion of long term trends.”

Professor Barton overlooked the U.S. Census Bureau’s Decennial Census, which has data on Lawyer’s incomes since 1950 (which reports 1949 incomes).[i]  The IRS data presented by Barton starts 18 years later, in 1967.

When considering long term trends in occupational incomes, it’s important to consider changes in the race and sex of members of the occupation.  Across occupations, women and minorities generally earn less than white men.  Race and sex variables are available in Census Household data, but not public-use IRS data.

b. The IRS data “separates lawyer earnings into solo practitioners and law firm partners”

Professor Barton acknowledges that his data misses incorporated self-employed lawyers, and that this group likely has higher incomes than those that he captures.[ii]

This means that Professor Barton’s IRS data is much less useful for identifying small and solo practitioners in 2013 than it was in 1970.  This is because the proportion of solo and small attorneys who incorporated has likely increased dramatically.  In 1970, 5 percent of full-time self-employed lawyers were incorporated.  By 2014, the share increased to more than 50 percent.[iii].  Barton is missing many solo and small time practitioners.  If trends toward incorporation continue, his data will become less useful every passing year.  The IRS data has different biases at different points in time, making trends potentially unreliable. 

Professor Barton’s schedule C data likely contains many filings from part time side businesses.  Using IRS Schedule C data in the way Barton uses it suggests that the overwhelming majority of unincorporated U.S. small business owners are living below the poverty line. 

Even this clearly distorted data shows that in relative terms, legal services on average is more profitable than 97% of sole proprietorships.  In other words, Barton’s data suggests that a part time side business is usually less lucrative than a similar full-time job.  But legal services are apparently more lucrative than most side businesses. 

c. IRS “data also jibe with several other sources, including After the JD [and the Economic Census]”

After the J.D. is not consistent with Barton’s solo practitioner data. 

After the JD suggests far fewer solo practitioners than Professor Barton’s IRS data.  After the JD III (page 28) suggests that about 10 percent of law graduates work as solo practitioners within 12 years of graduation.  Barton’s data suggests that there are around 350,000 or more solo practitioners—perhaps around 25 to 40 percent of lawyers.  Barton also previously claimed that solos constitute “The Middle Class of American Lawyers” and that changes in solo incomes were indicative of the fate of the profession.

Professor Barton now concedes that because of data limitations he “does not know exactly how many working lawyers there are [or] what they earn.”  I encourage Professor Barton to post a correction to his CNN and Business Insider posts, and to revise the next edition of “Glass Half Full, noting that several data sources suggest that solo practitioners constitute a small fraction of full-time lawyers.  I encourage him to refrain from over-stating the size and importance of this group of lawyers to legal practice.  After the JD also suggests higher solo practitioner incomes than Barton’s IRS data.[iv]  

UPDATE: August 18, 2016.  Mean solo practitioner incomes from After the JD III's national sample (about 12 years after graduation) were approximately $114,000 in 2013 overall, and $131,000 including only those working full-time, compared to Professor Barton's IRS schedule C estimate of $49,000.  The means are not published in the AJD report, only the medians.  Thanks to Bryant Garth, Ronit Dinovitzer, and Ioana Sendroiu for supplying this information.

Identifying Solo Practitioners in the Economic Census by Number of Employees Implies Average Solo Incomes Close to or Exceeding $100,000 per year

Professor Hadfield looked at the U.S. Economic Census (ECN)—a survey of establishments that have Taxpayer Identification Numbers.  ECN is useful for estimating establishment revenue, but is not well suited to estimating individual occupational incomes.  Nevertheless, with a few modifications to Professor Hadfield’s approach, ECN data can be reconciled with estimates from ACS.

In an effort to identify solo and small practitioners, Hadfield focused on the bottom 40 percent of Offices of Lawyers by revenue.  According to Professor Hadfield, this group includes only 57,612 establishments.  This is far lower than most estimates of the number of solo practitioners—especially Barton’s.  Since there are roughly 1 million to 1.2 million lawyers according to ACS and CPS, this implies that this group represent roughly the bottom 5 to 15 percent of employed lawyers (57,612/1,000,000 = 0.06).

The bottom 40% group by revenue does not appear to be a representative sample of solo/small practitioners, but rather a skewed subset of low revenue, unsuccessful offices of lawyers in a particular year (which for many may be an unusually bad year).

Rather than assuming that small practitioners are the bottom X percent of Offices of Lawyers by revenue and then discussing the low revenue for which they were specifically selected, a better approach to identifying small practitioners is to use the Economic Census’s field for number of employees.  One of ECNs greatest advantages is the presence of a measure of firm size tied to number of employees—similar measures are not available in ACS.[v]   

Using revenues for smallest employee categories in ECN and otherwise accepting Professor Hadfield’s assumptions suggests lawyer-proprietor incomes that are close to or above $100,000.[vi]

Professor Barton believes that a handful of other sources support his solo practitioner estimates, but as a general matter, they do not.  Urban Lawyers and Chicago Lawyers suggest that Barton’s estimates are off by a factor of 2 to 3.[vii]  Others, such as NALP full-time starting salary data, are inapposite.[viii]  Still other sources are not nationally representative and may rely on questionable sampling procedures.[ix]

The remainder of Professor Barton’s post focuses on the likelihood that ACS self-employed, unincorporated lawyers includes both law firm partners and solo practitioners.  I alluded to this in my first post, and elaborated in detail in extensive correspondence with Professor Hadfield shortly after she commented on Professor Diamond’s blog.  Professor Hadfield subsequently discussed many of these issues with Professor Barton.  I’m pleased to report that Professor Barton and I generally agree on many of these issues.

To briefly summarize a lengthy and technical conversation:

Solo and small practitioners are likely present in two ACS Class of Worker categories: “Self Employed, Incorporated” and “Self Employed, unincorporated.”  Lawyers who own interests in larger practices are also likely be included in at least one and probably both of these categories.  Ultra-high income lawyers will be top-coded—there are almost no lawyers in sample with incomes exceeding $1.3 million, and few with incomes above $500,000.

Combining the two self-employed categories produces a higher income distribution than just the unincorporated category.  In other words, by rejecting the “unincorporated” category as the baseline, Professor Barton is arguing in favor of a higher income distribution that is farther from his initial estimates of small practitioner incomes.

The data suggest that full-time solo practitioners could be a small segment of the legal profession and low paid—not the broad middle class of the legal profession, but rather toward the low end of the distribution.  Or solo practitioners could be a larger segment of the legal profession and reasonably well paid.  But the data suggests that solos probably cannot be both a very large fraction of full-time lawyers and very low paid.  All self-employed lawyers—including both solos and law firm partners— account for only about a quarter of full-time lawyers.

Self-employed, not incorporated full-time lawyers’ median incomes in recent years were around $100,000.  For the self-employed incorporated category, full-time median incomes were closer to $120,000.  The 75th percentile was around $260,000 to $300,000.  At the 25th percentile, toward the bottom of the distribution, incomes were around $50,000 to $80,000.  Including those working part time, incomes are lower—around $35,000 to $65,000 at the 25th percentile and around $225,000 to $280,000 at the 75th percentile. 

Thus, some self-employed full-time lawyers are making incomes close to Barton’s estimates, but these are self-employed full-time lawyers toward the low end of the distribution.[x]  

I thank Professor Barton for engaging in this discussion, and also thank Professors Hadfield and Diamond for their contributions.


[i] The Decennial Census has some lawyer and judge earnings data starting in 1940 / 1939, but variables become more consistent with those currently in usage starting in 1950/1949.  The Current Population Survey has data every year from 1967, but has a much smaller sample size than the Decennial Census.  From 2000 forward, annual data is available from the American Community Survey (ACS), which surveys around 3.5 million households compared to CPS’s 60,000.  Census IPUMS concepts of income have changed less over the decades than tax concepts of income.

[ii] Individuals who are dedicated to running a full-time legal practice may be more likely to file their taxes as C-corporations or S-corporations than as sole proprietorships, and will not show up in Barton’s data.  Such individuals will show up in ACS because they will report their occupation as “Lawyer.”

[iii] This is according to ACS and Decennial Census data.  ECN also shows a substantial proportion of offices of lawyers that are incorporated.

[iv] After the JD III shows median incomes for full-time solo practitioners around $50,000-$60,000 12 years after graduation.  Mean incomes—the measure Barton uses for IRS data—are much higher than the median at around $114,000, or for only those working full-time, $131,000.  Moreover, attorney incomes typically continue to increase more than 12 years after graduation according to Census data.  Footnote Updated August 18, 2016.

[v] It might also be useful to look at firms with no employees, but unfortunately, data for such firms appears to only be available for “legal services”, not “offices of lawyers.” The non-employer data are reasonably similar to the IRS data on which Barton relied, and probably share similar limitations, such as picking up supplemental revenue from side businesses and some non-legal practice legal services, rather than total individual incomes for full-time legal practitioners.

[vi] The smallest categories in the ECN are (1) “establishments operated for the full year with less than 5 employees” (1-4 employees); and (2) “establishments not operated for the full year.” For Offices of Lawyers in 2012, the 1-4 employee category had an average of 1.95 employees (perhaps a lawyer and a non-lawyer employee or two lawyers with no non-lawyer employees) and the less-than-full-year category hand an average of 0.79 employees (so perhaps one-part time lawyer).  This raises questions about whether and when a lawyer who owns a firm and pays himself a salary would count as an employee for ECN purposes.

The 1-4 employee category had an average revenue per establishment of $331,888 and includes 105,000 establishments.  The part time category had an average revenue of $229,000 per year and includes 24,000 establishments.  (The number of these establishments suggests around 130,000 solo practitioners—still a bit lower than estimates from other sources).

Both the 1-4 employee and part-time categories have much higher average revenue than the $134,000 number suggested by the bottom 40% by revenue estimate.

Offices of Lawyers from ECN 2012 ECN_2012_US_54SSSZ2

Accepting Professor Hadfield’s back-of-the-envelope estimate that lawyer-proprietor income equals revenue, less payroll, less $38,000 in fixed expenses would imply that on average small practitioner incomes exceed $100,000 per year ($331,888 – $90,239 – $38,000 = $204,000).  If this were a firm owned jointly by two-lawyers, it would suggest just over $100,000 per year each.  Unfortunately, we only know the number of employees in ECN, not the number of proprietors. Even lawyers who owned part-time offices would cross the $100,000 mark.  ($228,863 – $71,234 -- $38,000 = $120,000), assuming one proprietor.

[vii] For example, Urban Lawyers estimated mean 1995 solo practitioner incomes in Chicago of $126,600 (inflation adjusted to 2016 dollars). That same year, Barton’s IRS data estimated national solo practitioner incomes of $60,400 (in 2016 inflation adjusted dollars).  Barton’s estimate is around $65,000 lower, or off by a factor of 2.1.  Similarly, Chicago Lawyer estimated mean Chicago solo practitioner incomes of $183,000 in 1975.  That year, Barton’s IRS data estimated national mean solo practitioner incomes of $65,900—lower by $117,000 or a factor of 2.8 (all dollar figures inflation adjusted to 2016).  The estimates that appear in the text of Urban lawyers are in 1995 dollars.  I have inflation adjusted them to 2016 dollars for purposes of this post.  Professor Barton has generously shared a spreadsheet showing his IRS data, both in nominal and 2012 inflation adjusted dollars.

There are two likely explanations.  Either lawyer incomes in Chicago were much higher than the national average in 1975 and 1995, or the Chicago Lawyers and Urban Lawyers studies both have much higher estimates than Barton’s IRS schedule C data.  Lawyer incomes in Chicago were reasonably close to, and perhaps slightly lower than the national average according to combined data from the Decennial Census from 1980, 1990, and 2000.  (The 1970 Census was not used because city codes were not available).  In other words, if Barton has as much faith in Chicago Lawyers and Urban Lawyers as his last post suggests, then he should conclude that his IRS schedule C estimates are likely much too low.

Urban Lawyers also estimated that solo practitioners were a much smaller percent of lawyers than Barton--around 14 percent as of 1995, and trending down. The authors suggested that this might be a slight underestimate relative to national averages because of the urban focus.

[viii] NALP’s director of research, James Leipold, stated by email that NALP does not have data on solo practitioner incomes.  In addition, Census data shows that law graduate starting salaries are typically much lower than subsequent incomes.  According to NALP, its data 9 months after graduation shows around 2% of law graduates working as solo practitioners.

[ix] Bar associations grappling with a wide range of responsibilities and limited resources will sometimes use inexpensive and quick sampling methods that can produce lower quality data.  Large scale, nationally representative empirical surveys are the raison d'ê·tre of the U.S. Census Bureau.  The Census Bureau is supported by thousands of highly specialized experts, over a billion-dollar budget, and over a century of institutional experience.  The American Community Survey alone has an annual budget exceeding $230 million and collects data from over 3 million households every year, while the Decennial Census has a multibillion dollar budget and last collected long-form data from around 18 million households.

[x] At most points in the distribution, self-employed lawyers are not the lowest paid group.  That distinction goes to state and local government workers and lawyers working for non-profit organizations.

Guest Blogger: Michael Simkovic, Legal Profession, Of Academic Interest, Science, Weblogs | Permalink