High-Income Residents: Are They the Driving Force Behind DC’s Premium Apartments?

In a recent post, we concluded that the premium apartment rental market is the more popular and ascendant segment of the city’s housing market in the context of the current trend in net population growth. To further elaborate on this topic, we profile the tenants in the city’s Class A and Class B apartment buildings built after 2000 based on income tax data characteristics. The full research paper can be found here.

Economic Profile of Tenants

Table 1 tells us that in 2015 half of the residents who were income tax filers in the 88 Class A and Class B large and mid-sized apartment buildings that were built after 2000 had annual reported income of less than $57,428 and were under the age of 31.5. And, the vast majority of these tenants were single tax filers (unmarried and no dependents) and were relatively new[1] to the city.


[1] We classify a new resident as someone who existed in the city’s income tax data in either 2013, 2014, and/or 2015, but did not exist in 2012 or prior.

Who is more likely to live in new apartment units?

Our data shows that there was a tripling in the number of premium apartment units delivered in 2013 compared to 2012. To better evaluate the data, we divided the buildings into two groups. The first cohort is comprised of all 2015 tax filers found to be residents in multifamily buildings that delivered between January 2000 and December 2012 (relatively older premium multifamily buildings). The second cohort is comprised of all 2015 tax filers found to be residents in multifamily buildings that delivered between January 2013 and December 2015 (newer premium multifamily buildings).  We then fit a statistical model to the data to determine the characteristics of new buildings versus older buildings.

Using T-tests, we find that the newer buildings tended to have units that were an average of 88.3 square feet (10.5 percent) smaller and cost 17.5 percent more per square foot (Table 2). We also found that individual tenants in newer buildings tended to have income that was on average of $9,884 (12.3 percent) less and 1.3 years younger than renters in older buildings.


Using a statistical model to differentiate the characteristics of tax filers living in a newer building in 2015 versus older buildings, we calculate the probability that certain factors affect the choice of residing in newer apartment buildings instead of older buildings.

While the tenants in new and older apartment buildings are generally very similar, we were able to again tease apart a few distinctions in the two populations as well as a few contributing factors for their housing choices.


We find that income has almost no influence on whether a resident chooses to live in a newer or older apartment building (for every $100,000 increase in income, the probability to choose a newer building increases only about 4 percent). Age is also an important factor in determining how likely a resident will choose newer or older apartment units. Younger residents are more likely to reside in newer apartment buildings. For each additional year in age, existing residents are 0.8 percent less likely to reside in newer buildings, while this percentage is 0.2 percent for new residents. We also find that tenants commonly supplement their traditional wage/salary income with additional business income from entrepreneurial or other self-employment endeavors.[2]

Given that 83 percent of all tenants in these buildings are single filers (as shown in Table 1), we find that long time city residents who are head of household tax filers (unmarried income earning adults with dependent children) are 23 percent more likely to live in newer buildings compared to married residents. This is possibly due to the city’s affordable housing efforts to place low-income households in these new buildings via affordable housing programs.  And finally, single residents are more likely to reside in newer buildings compared to married filers, especially when they are relatively long-time residents.

[2] On government tax forms, adjusted gross income is comprised of wages and salaries, business income, investment gains or losses and other income.

Several Ways DC is Changing

In sum, we find the following results. First, 64 percent of the tenants in all the apartment buildings in this study tended to be new to the city. Second, the newest apartment units are smaller and more expensive, and their residents tended to be slightly younger and have less income than residents in the relatively older buildings. Third, residents in the newest units are more likely to have business income as part of their total reported income, which suggests there is an increased tendency for these residents to supplement their traditional wage and salary income with additional income from entrepreneurial or other self-employment endeavors. Lastly and surprisingly, the analysis shows a relatively strong increase in probability for residents in the newer buildings to be head of household filers. This is possibly due to the city’s affordable housing efforts to place low-income households in these new buildings via inclusionary zoning and various housing subsidy programs.

Conventional wisdom assumes that these newer buildings are attracting primarily high-income residents; however, we find that compared to older buildings, the city’s newest and pricier apartment buildings built during the recent residential construction surge (2013 and after) tend to attract a higher percentage of new residents to the city, and also attract a higher percentage of single, young residents with income below the city average. It appears that both the city’s demographics and apartment rental market are continuing to evolve and change in significant ways. And, it is very likely these changes will have considerable implications on the residential and economic patterns of the city in the years to come.


The Data

Using data from CoStar, we identified 88 Class A and Class B large and mid-sized apartment buildings (containing 21,203 total residential units) from across the city that were built after 2000. The list can be found here. This study also uses 2015 individual income tax data for all DC tax filers who listed their home address as being in one of the 88 apartment buildings mentioned above.



Residents move into the city for jobs, move out for housing

The District added about 90,000 net new residents between 2000 and 2014, but the population churn has been great. Current Population Survey data show that more than half a million people report moving to the District from some other state or jurisdiction during that period—this is on average 8 percent of the city’s resident population every year. Residents also move within the city frequently: In 2014, for example, nearly 60,000 residents moved houses within the city—this is approximately 9 percent of District’s resident population.


Jobs—or the prospect of one—is the top reason why the District receives new residents from other states. Between 2000 and 2014, the District received nearly 165,000 new residents because either they got a job in the District or their job was relocated in the District. Another 55,000 moved here to attend college, or they had just completed college, and found the District to be an attractive job market. Not all of these newcomers stayed, but it is interesting to see that over 42 percent of District residents who had moved to the city sometime in the previous 12-month period did so for their careers. Convenience of living in the city follows jobs, with another 10 percent of District residents suggesting that they moved into the city for an easier commute.


Why do people move out? It is housing. The top two reasons people report moving out the District in the last fifteen years have to do with wanting better housing, seeking cheaper housing, or wanting to own a house, for example, and these reasons account for 36 percent of the moves out of the District whereas they account for only 12 percent of the moves into the District. Jobs however, account for 12 percent of the people moving out compared to 32 percent moving in.


Incidentally, suburbs do a better job in attracting District residents than the other way around. Between 2000 and 2011, the District sent 391,000 of its residents to Maryland or Virginia (42 percent of those who left the city), but received only 191,000 new residents from these two states (30 percent of the residents who moved into the city). Looking at the reasons why people move to the suburbs, housing still plays a role, but the top reason is to establish a household. It appears from the data that those who share housing in the District with roommates are most likely to move out to the suburbs when they want a place of their own.


What exactly is this data?  Data on move rates  are extracts from the Current Population Survey data maintained by Miriam King, Steven Ruggles, J. Trent Alexander, Sarah Flood, Katie Genadek, Matthew B. Schroeder, Brandon Trampe, and Rebecca Vick. Integrated Public Use Microdata Series, Current Population Survey: Version 3.0. [Machine-readable database]. Minneapolis: University of Minnesota, 2010. The sample for the District is small–therefore the post looks at a combined 15 year period.

Note: an earlier version of this post was published before the draft was completed.

Homeownership in the District

In 2014, according to data compiled by the Federal Housing Finance Agency on home purchase prices, homes in the District sold for over three times the prices they commanded in 2001.  During that time, home prices in the U.S. also increased, but not nearly as fast—2014 prices were 45 percent greater than 2001 prices.  The great recession did dampen prices in the District (shaded in the graphs below), but not enough to undo the rapid gains in early 2000s and since the end of the recession, rapid price increases once again became the norm.image003

So how did all this affect homeownership?  In 2014, 44 percent of District residents lived in homes they owned—that is down 4 percentage points from 2001 and down five percentage points from 2007 (right before the great recession) when ownership rates reached 49 percent. As a relative decline, this is about 10 percent (5 out of 49).  Homeownership rates declined in the US too, but not as rapidly.  Ownership rates declined by 4 percentage points since the beginning of the great recession from 71 percent to 67 percent, but given that ownership rates in the nation were much higher to begin with, this is a relative decline of 6 percent.


We have written many times on this blog about the changing demographics and gentrification in the District (see here, here and here).  Homeownership lies at the heart of these issues.  So we checked: how did home ownership change among different income groups?  We divide the District’s resident population into three groups: low-income, which includes all households with incomes in the bottom 25 percent of the income distribution in 2014, high-income, which include households in the top 25 percent of the distribution, and middle-income, which is all the households in between. We look at these groups since 2001, adjusting income thresholds for inflation.  This way, we are comparing similar groups based on today’s demographics.

In 2014, 19 percent of households who fall in the bottom 25 percent of the income distribution owned their homes.  If we looked at the same income group in 2001, we would have seen that 31 percent of them owned their homes.  That is a relative decline of 40 percent.  Homeownership among the middle-income groups increased through the 2000s, only to go back to their 2001 levels in 2014, at 43 percent.  Homeownership among high income residents also lost ground, but only slightly, going down from 77 percent to 72 percent.image007

There are many issues at play here: increasing prices, transient population, limited growth in housing stock, demographic change (read: more singles who are less likely to own across all income groups and overall growth in population).  But what is clear is that the income composition of homeownership is changing, with ownership of housing shifting towards middle and high income residents.

What exactly is this data?  The home price data is the quarterly index of home prices based on estimated purchase price.  Homeownership and income data are extracts from the Current Population Survey data maintained by Miriam King, Steven Ruggles, J. Trent Alexander, Sarah Flood, Katie Genadek, Matthew B. Schroeder, Brandon Trampe, and Rebecca Vick. Integrated Public Use Microdata Series, Current Population Survey: Version 3.0. [Machine-readable database]. Minneapolis: University of Minnesota, 2010.

Should we all just pack up and move to (insert city here)?

Last week we critiqued a Columbus, Ohio marketing campaign that attempts to lure area residents away by promoting the city’s low cost of living. We found that living in Columbus isn’t going to save a typical millennial as much as you might expect. This left us wondering how the cost of living in the District compares to other jurisdictions. In order to compare, we used the same methodology that we used last week. What we found is that District millennials, even after paying for expensive housing, end up with more purchasing power.

8 City Comparison

Millennials living in Baltimore, Charlotte, and New York City have similar purchasing power to those living in the District. Median monthly rent for a one bedroom apartment in Baltimore and Charlotte is roughly half of what residents pay in the District – however salaries in these jurisdictions are considerably less. Conversely, New York City millennials pay some of the highest rent in the country but large salaries offset the Big Apple’s exorbitant housing prices.

Unfortunately for millennials living in Boston and San Francisco, there isn’t much disposable income remaining after paying rent each month. Boston’s high housing costs and low median salary drastically reduce the amount of money millennials have to spend on non-housing goods. As for San Francisco, residents are burdened with the highest rent in the country. The median salary in the Bay Area is also high, but not high enough to offset the city’s expensive rental market. Millennials living in these housing markets might want to consider getting a roommate to help with rent.

As we stated in our post last week, millennials that want to buy a house or have children in the District will experience a larger hit to their wallets than if they lived in a comparable city. The median cost of a three-bedroom, two-bathroom home in D.C. is $788,000. In Charlotte it’s $255,569 and in Baltimore it’s $447,021. Child-care for two children in the District is 39 percent more expensive than in Baltimore and 62 percent more expensive than in Charlotte. Despite the large housing and child-care cost differences in these two cities, District millennials can take solace in knowing that the median cost of a three-bedroom, two-bathroom home in Manhattan is $1,355,865 and that child-care costs are 15 percent higher than in the District.

What exactly is this data? “Millennial” means people between the ages of 18 and 34. We got the median millennial earnings from the Census Bureau and the rent data from zumper.com. To compare non-housing expenses in D.C. and Columbus, we used the family budget calculator from the Economic Policy Institute. We subtracted childcare costs from the family budgets to get a more realistic budget for millennials living in cities. Data on median home prices is from NerdWallet. Child care costs are from the Economic Policy Institute.

Ginger Moored contributed to this article.

Where is the vacant land in the District?

Along with news on population growth and increasing housing prices, we often hear the concern that the small footprint of the city is an impediment to growth.  The discussions on height limitations in the city, work done by the Urban Institute and by our good friends at the Office of Planning made us think: How much vacant land is there in the District that can one day be developed?

We searched the District’s real property tax database for empty, unbuilt land.  Of course, not all of this land is immediately buildable.  Zoning and ownership factors greatly affect allowable development on these lands.  And some of this land might never be buildable (Parks, rail tracks, other bits of land–do we want to see buildings in the Arboretum?).  So think of this as an exercise over 300 years where population growth might make our current preferences less relevant.

Here is what we found:


  •  The District has approximately 300 million square feet of vacant, unimproved land. This accounts for about 19 percent of all land in the District’s real property tax database.
  • Of this vacant land, only 14 percent (42 million square feet) is in the hands of private entities and individuals. These would be the simplest to build on (barring zoning limitations).  Privately owned lots are concentrated in zip codes 20017 through 20020 and in 20002.
  • Non-profit, non-taxable entities such as hospitals, universities, and churches own another 24 million square feet. Of course, we have seen many examples of such land being converted into mixed use.  But, it takes a bit more time to get there.
  • The District owns 34 million square feet, some by the District’s Housing Authority. These lands could be transformed into housing or mixed use, but the development must follow government procedures.  It takes time.
  • The United States government owns 197 million square feet of land (or 66 percent of the vacant and unimproved land). Of this, 133 million square feet are east of the river.
  • In zip code 20032 (Washington Heights and Bellevue neighborhoods) only, the District and the U.S. governments combined hold 53 million square feet of land.

What does this all mean? The Office of Planning estimates, under plausible conditions, the District could add 175,000 new households by 2040.  This would mean an increase in demand for over 200 million square feet of housing to house the new residents (see pages 27 and 28 here).

Twenty five years is not a very long time, so if we limit our new construction to privately owned vacant  land only, to meet the 200 million mark, we would have to see a floor-area-ratio of 4 or greater–that is the usable construction area is four times or grater than the size of the lot.  This is pretty dense.  But of course there are other sources of land including parking lots, older buildings or properties with underutilized capacity.

Here is a map of all vacant land in the District by zip codes by ownership and tax status. We made interactive maps, too, which you can get to using the link here or clicking on the map below.


What exactly is this data? The data captures all lots and squares marked as vacant in the District’s real property database classified as commercial or residential.  We eliminated smaller lots unless they are in the same square and combined made 2500 sq. ft. of land or more. The zip code results might be influenced by where ownership records fall within large properties.