Little evidence of the gig economy in the District

Much has been written about those who work in the “gig economy” (see here, and here), and those of us who try to count them. The term itself, however, is hard to define. Some think of the gig economy as work contingent on demand. Others include an element of technology that connects workers with potential sources of income.

A recent Government Accountability Office study offers various definitions, the narrowest of which produces an estimate of 7.9 percent of the workforce in the gig sector in 2010. But this group includes temps, on-call workers, and contractors: jobs that have been around forever.  Senator Mark Warner, in his recent op-ed at the Washington Post, cites that study (although the Post inadvertently links to a different report from 2000) to conclude that one-third of the U.S. labor force could be in the gig sector, and that these gig-workers “now find themselves piecing together two, three or more on-demand work opportunities to make a living” [emphasis added]. While it is true that the broadest of the GAO definitions produces an estimate of 40 percent of the workforce gigging, there is nothing new about these work arrangements GAO includes in the broad definition: independent contractors, self-employed individuals, and even part-time workers.  These work arrangements existed long before Uber opened for business.

There are good reasons to try to get a better handle on the gig economy. In the gig sector, the types of risks we typically think of as business risk—e.g., lack of customers because of bad weather, sick workers—become the worker’s problem. To be sure, even before the gig revolution, some sectors of the economy worked just like that. Cab drivers, for instance, never had much in the way of benefits such as healthcare, a pension, paid holidays or even sick days. It is no surprise that much of the gig work is beginning in sectors where workers, such as drivers, handymen, and baby-sitters, already took large risks.

The evidence that piecemeal work is replacing traditional employment in the United States is scant.  So we wondered: how about the District?  We ran into the same definitional problems about the gig economy when looking at the District’s data, but we decided to focus on the self-employed, specifically, those who characterize themselves as “self-employed in an unincorporated business they own.” For laymen, those are the people who pick up contract work, get a 1099 from the IRS at the end of the year, and pay self-employment taxes. The Bureau of Labor Statistics and the U.S. Census differentiate between the 1099’ers and self-employed who actually own a business that receives the monies for the services rendered, and in return pays a salary to the business owner, with proper deductions for social security and Medicaid. (This Pew piece on the characteristics of the self-employed provides a much more detailed explanation of the term).

We first look at the number of District taxpayers who have paid self-employment taxes. The data show that the total number of people who pay self-employment taxes has increased in the District from 35,000 in 2006 to nearly 49,000 in 2014. This is a very steep increase (36 percent overall and nearly 4.5 percent annualized) even when compared to the relatively rapid increase in the District’s population and tax filers (tax filers grew at about 2 percent per year during the same period).  But data show that the rapid increase in the number of filers who paid self-employment taxes occurred before 2010. In fact, since 2010, the share of tax filers who pay self-employment taxes has been stable at about 14 percent.


So why did the District see such a rapid expansion in reported self-employment? This, we suspect, has less to do with changes in the underlying economy and more to do with changes in tax policy. Beginning in 2002, the District started offering Earned Income Tax Credits, first at 10 percent of the federal credit, and by 2009, at 40 percent of the federal credit (one of the most generous such programs in the nation).  The credit targets low income families and single parents with children, and the key recipients of this benefit are those who file as head of households.

Since the policy changes began, both the number and the share of heads of households who pay self-employment taxes has increased.  In 2006, only 7 percent of filers who paid self-employment taxes were heads of households.  In 2010—one year after the benefits maxed at 40 percent of federal credit—this share doubled to 14 percent, and then reached 17 percent in 2012. During the same period, there were no significant changes in the share of singles or married filers who reported self-employment income.


One might say that tax data is not the best measure of the gig economy because it captures all taxpayers who pay self-employment income. In the District, for instance, a government employee who teaches a class at a college, or a professor who writes a paper for a non-profit, would all receive a 1099 and pay self-employment taxes. So the data are noisy, mixing moonlighters with the gig-workers.

So let’s turn to the American Community Survey, which inquires about the employment status of workers. Here we present data on District residents who characterize themselves as self-employed. And, surprisingly enough, we see a decline, both in levels and in shares.  In 2014, only 13,100 residents—2.4 percent of District residents older than 16—claimed to be mainly self-employed, down from the post-recession peak of nearly 18,000 self-employed residents (or 3.4 percent of those over the age of 16).  Self-employed persons increased slightly in the District during the recession, but since 2012—the time when resident employment really began to increase—self-employment has gone down.


In 2014, the self-employed in the District made up about 5 percent of total resident employment.  This figure has been relatively stable, except for 2012. Self-employment income has likewise been rather stable at 3 percent of personal income. District residents who are self-employed routinely generate about 70 percent of their income from their self-employment work.


Incidentally, the District’s self-employed residents—just like its employed residents—are better educated than those elsewhere in the United States. Nearly 60 percent of District’s self-employed have a graduate or a professional degree (compared to only 13 percent across the United States), and fewer than one in five completed schooling only up to high school (compared to 36 percent in the country as a whole).


Is it possible that the data are not capturing the gig economy? We can think of two reasons—one relatively unique to the District, and the other more general.

  1. It is possible that the District’s gig workers—the Uber drivers, the Amazon flex folks, the Taskrabbits—are not District residents, just like the many District workers who receive minimum wage do not live in the District.
  2. It is possible that some workers do not fully report their income because they do not realize that they must report earnings from Etsy, Sittercity, or airbnb.

It matters to us to measure the gig economy correctly because we need to be able to track the changes in the District’s economy and understand how work activities connect to incomes. We plan to dig a bit deeper, looking at who might be the gig workers in the District and what types of jobs they hold.

What exactly is this data? 

Data on the number of people paying self-employment taxes in the District by tax filer type is from the IRS (2013 data set is now public). Data on class of worker are from various years of ACS. DC data on self-employment has error terms of +/- 0.5 percent to +/- 0.7 percent depending on the year (or about 2,000 workers).

DC’s Residential Property Market –$2 Million may be the new $1 Million for the Luxury Market

Many reports have focused on the rapid pace of price appreciation and sales volumes of luxury homes valued more than $1 million. Year to date, sales of homes with prices exceeding  $1 million account for more than 18 percent of all single family home sales in DC.  With sales of $1 million becoming more and more “commonplace”, it may be time to set the bar higher in terms of what constitutes luxury.  This redefinition would also better align the DC market with other expensive housing markets in the nation where $1 million no longer carries the same cachet and exclusivity that it did ten years ago.

Here’s data that shows the share of overall sales of single family homes valued more than $1 million. These are broken out in three ranges: 1) $1-1.5 Million 2) $1.5-$2 Million and 3) $2 Million and greater

Share of Overall Sales of Single Family Homes Valued more than $1 Million, 2001-2015



  • The share of overall sales of single family homes valued more than $1 million has risen from only 3 percent of all home sales in 2001 to over 18 percent in 2015.
  • Sales of homes valued more than $1 million broke the ten percent barrier of overall sales in 2006.
  • Sales of homes valued more than $2 million now account for more than 4 percent of all sales.
  • Given the upward trend in prices, it may not be too long before 5 percent of all homes sold in the District top $2.0 million.

Shown below are the main neighborhoods where the majority of transactions valued more than $2 million has occurred.

Number of Sales of Homes Valued More than $2 Million by Major Neighborhoods: 2001-2015


  • In most of these neighborhoods, and in particular the bellwether neighborhood of Georgetown, sales of homes valued more than $2 million jumped beginning in 2014.
  • Georgetown accounts for almost one-third of all single family homes sold valued more than $2 million.
  • Altogether, the neighborhoods shown above account for almost 75 percent of all sales valued $2 million and over.

What exactly is the data?

Data on sales is from the DC Recorder of Deeds and the DC Office of Tax and Revenue. Building codes 11-13 were used to identify single family homes.  Only arm’s length transactions were considered.

Betty Alleyne contributed to this post





BEA’s sharp revisions to commuter income earned in the District reveal the challenges of tracking the city’s changing income flows

In its September comprehensive revisions to all states’ Personal Income, the US Bureau of Economic Analysis (BEA) reduced its estimate of District’s 2014 Personal Income by $4.2 billion to $46.0 billion, down 8.5 percent. The reductions go back ten years, telling us that the District residents have been earning less than we thought they did for almost a decade.

BEA’s revised estimates do not change the basic story that income earned by DC residents has grown over the past decade as population and jobs have grown.  But the new numbers dial back considerably the proportion of the growing income earned in DC that stayed in the city.

The Personal Income calculation

To calculate Personal Income, BEA starts with income earned in the District by everybody working in the city.  Earned income is wages and salaries, benefits, and proprietor’s income. BEA then makes what is called a “net resident adjustment” to obtain income earned by DC residents.  The net resident adjustment involves (1) subtracting income earned in DC by non-residents (mostly commuters), and (2) adding income earned outside of DC by DC residents.

BEA 2014 revision increased by $1.7 billion the amount earned in DC, but also took $5.8 billion more away for the net resident adjustment using a new analysis of Census Bureau journey-to-work data and IRS tax filings.  Revisions to the non-earnings portion of Personal Income were relatively minor.


Before the revision, income earned by the District residents (in D.C. or elsewhere) appeared to be 46.1 percent of the income earned in the District in 2014 (by residents and commuters), up from 39 percent in 2004.  The revision puts the 2004 share to 38.2 percent, and 2014 share to 39.9 percent. Here is another way to look at it: before the revision, we thought that the District residents accounted for 62 percent of the increase in all income earned in the District between 2004 and 2014. With this revision, this share is now closer to 44 percent.


With these revisions, the increases in the proportion of income earned in DC staying in the city become more modest than what one might expect just by looking at the jobs numbers.  From 2004 to 2014, resident employment grew 21.0 percent while wage and salary jobs located in DC increased just 11.8 percent.  We have written about this before, considering explanations such as more DC residents working in higher paying jobs outside the District and more DC residents taking jobs vacated by commuters.  New data gives us pause since it shows that the growth in the resident employment seems not to have been not accompanied by a similar measure of growth in incomes. We do not know why.

The difficulty of tracking income flows in DC’s changing economy.

There is nothing unusual about BEA revising its estimates as more information becomes available, but the scale of the September 30, 2015 revisions to DC’s Personal Income is unusual. The percentage revisions to DC’s Personal Income were far greater than for any state data. Over the past three years there were only 5 instances of state changes of more than 3 percent in any one year, the biggest being a 4.8% increase for Alaska in 2012. Why was DC so different? The reason is the net resident adjustment factor. Tracking the in-and out-flows among jurisdictions is one of the more difficult tasks in compiling earnings estimates, and those flows proportionately are much larger in DC than for any state. They also can be influenced heavily by the changes in job or residence location by a relatively small number of high income households.

In its latest revisions, BEA turned to the most recent American Community Survey (ACS) data available for all US counties (the average for the years 2006 through 2010), along with the most recently available IRS tax data. Accurately tracking DC’s Personal Income is likely, however, to continue to be a challenge for BEA because DC’s population and economy keep changing—and the key data sources used to determine Personal Income are available only with a considerable lag.

Revised DC Personal Income compared to the US

The new data still show that DC Personal Income grew faster than the US average from 2004 to 2014 (DC 59%, US 46%). Over the past four years, however, DC increases have been quite close to the US average, and in 2014 the 4.4% growth in the US exceeded DC by more than a percentage point.

DC’s per capita income, now estimated at $69,828 in 2014, is 52% above the US average and is still the highest in the nation (compared to states). With higher population growth, however, DC’s per capita income has also grown more slowly than the national average over the past four years.image012


What exactly is this data?

The Data is from BEA’s Regional Economic Accounts.  The latest data are available here.

Where does DC stand when it comes to nontax revenue: What do the numbers tell?

Last month the Office of the Chief Financial Officer (OCFO) issued an updated report on the District’s general purpose nontax revenue, which is largely comprised of charges for services, fees and fines. Nontax Study Report – 2015. We further compared the revenue composition of the District’s General Fund to that of five neighboring jurisdictions: the City of Alexandria, VA; Arlington County, VA; Fairfax County, VA; Montgomery County, MD; and Prince George’s County, MD. Intergovernmental transfers/revenue, such as payments from the federal and state/local governments are excluded from this analysis since our focus is on District generated revenue. For those who haven’t read about the results of the study as reported in the Washington Business Journal, here is another chance.

The District appears to rely heavily on non-tax revenue sources, such as, fines and fees, when compared with neighboring jurisdictions. Based on our analysis, the significance of nontax revenue seems to be dependent on the level of government under consideration, i.e. whether it is a city, a county or a state. D.C.’s unique character may have contributed to this.Local revenue comparison chart-2014

The percentage share of non-tax revenue in the District has been relatively stable, varying between 13 percent and 16 percent of the total General Fund revenue over the period FY 2005 – FY 2014. In terms of dollars however, it looks like the city’s non-tax revenue has gone up by about 27 percent.General Fund CompositionPercentage Increase NT-FY 2014We also looked at the trend of the top ten nontax revenue sources to see if there was a single source of revenue driving the majority share of the 27 percent increase. We found that collections from Emergency Ambulance Fees, Traffic Fines and Building Permits were the leaders each growing at an average rate of 17%, 8.2%, and 8% respectively.

Traffic related fines however consistently topped the list during the past ten years, FY 2005 – FY 2014. The table below presents the top 10 revenue sources in the order of the share they contributed to the total non-tax revenue base.Top 10 NT revenue - FY 2014We will have a series of posts diving into these top 10 non-tax revenue sources in the upcoming few months.

What exactly is this data? The above analysis is based on the District’s Annual Operating Budget for fiscal years 2005-2014 and the FY 2014 Annual Comprehensive Financial Report (CAFR). Non-tax revenue is income earned by a government from sources other than taxes such as fees and fines. The rankings are calculated based on the total amount of actual revenue reported for each non-tax revenue source within the District’s system of Accounting and Reporting (SOAR) which was later rolled over into the CAFR. The dollar amount reported for Traffic Fines above is inclusive of past due collections of outstanding traffic related tickets by the District’s Central Collection Unit (CCU).

The role of demographics in NAEP and PARCC scores

Last week, the District received its results for two school assessment tests: the National Assessment of Educational Progress (NAEP) and the Partnership for Assessment of Readiness for College and Careers (PARCC).  NAEP is a national test of reading and math skills given to 4th and 8th graders.  PARCC is the District’s replacement for the DC CAS, and it is connected to the Common Core curriculum. The results reveal both good and bad news:  District students made important gains in the NAEP scores, but the PARCC results for 10th graders show most high school students are not prepared for college.

What role do demographics play in these test scores? Here are our findings:

  • Demographic change among students taking NAEP in D.C. explains only part of the increase in NAEP scores the city has seen since 2005. When we control for changes in race, test scores still rise, just not as much.
  • PARCC scores tend to be lower for high schools with larger at-risk populations, though some schools defy expectations, especially charters.

Demographic change explains only part of the increase in NAEP scores (by Kevin Lang)

The 2015 NAEP results were mostly positive for the District.  Fourth graders posted statistically significant gains in both math and reading. Eighth grade reading scores were unchanged from last year and math scale scores saw a slight decline. The District’s NAEP scores have, for the most part, been increasing during the 10 year period we analyzed.

Comparing multiple years of data could be problematic, especially in a city like ours where there are rapid demographic changes.  Since 2005, the proportion of test takers in D.C. that identify as Black continuously declined while the proportion of Hispanic and White students steadily increased.

demographic changeGiven that White and Hispanic students in D.C. consistently out-score Black students on NAEP, does the increase in NAEP scores capture an increase in learning or a shift to students who traditionally do better on such tests?

To answer this question we looked at what would have happened to the NAEP scores if the demographics of test-takers in D.C. had stayed the same between 2005 and now. Our calculations show that even if the racial breakdown of students stayed the same, the District would have still seen growth in NAEP scores, but just smaller. That is, demographic changes influenced test results, but they cannot explain all the increases; it seems learning in the city’s schools has also improved.

NAEP DEMO CONSTANTHere, we are collapsing all factors that affect learning into one: race.  It is also possible that students moved up the socioeconomic ladder, regardless of race, or that some other change took place among students that would affect their test scores. However, the percent of test-takers who are eligible for free and reduced lunch has fluctuated from year to year, and NAEP results have increased not just across race, but across all subgroups such as gender, disability status, and ELL status. This we take as further evidence that learning has improved.

NAEP Subgroup Scale Scores

If you want to further explore how the demographics of NAEP test-takers in D.C. have changed, check out this interactive graph of ours. And to see how the percent of students scoring proficient or above on NAEP in D.C. has changed over time, click here.

PARCC scores tend to be lower for high schools with larger at-risk populations, but some schools defy expectations (by Ginger Moored)

Overall, the PARCC scores for D.C. high school students were pretty abysmal, with only 10 percent of students meeting or exceeding expectations in math, and 25 percent of students meeting or exceeding expectations in English.

As the graph below shows, though, test scores for each high school varied a lot. The larger a school’s at-risk population, the more likely it was to have lower scores—though there were plenty of exceptions, especially among charter schools.

For example, at SEED and Thurgood Marshall—charter high schools where about half of the students are at-risk—around 60 percent of students met or exceeded expectations on the English test (which is about twenty percentage-points higher than schools with similar at-risk populations).


If you’d like to sort through the PARCC data yourself, and compare the scores at different schools for different types of students (black, white, male, female, etc.), you can use the interactive graphs below.


indvidual school

What exactly is this data?

NAEP data is from the National Center for Education Statistics NAEP Data Explorer.

PARCC data is from the Office of the State Superintendent of Education (OSSE), DC Public Schools, and the DC Public Charter School Board. At-risk data is from OSSE. An at-risk student is defined as a student who is homeless, in foster care, qualified for food stamps or TANF, or a high school student who has been held back a grade.