Anonymous wrote:

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:Statistical agency employee here again. I did a little more digging to see what other pieces of information we could get to understand what's actually going on. It seems like a number of people on here are more interested in rehashing their pet complaints than in understanding what's happening, but I'll share for anyone who's interested. If you think my post is too long and just want to complain, save your keystrokes; I don't care.

The Census Bureau also produces something called the CPS Housing Vacancy Supplement (HVS), which reports vacancy rates for owner-occupied and rental housing units for each quarter and for each state and metro. These are pretty noisy estimates, because the CPS only samples something like 72k households a month nationwide, which means each quarterly estimate for DC is based on only ~450 observations split between rentals and owned properties. The vacancy rate is also only an indirect measure of population change, since households can grow or shrink without becoming vacant, and since construction of new units changes the denominator over time. But, this survey provides much more timely data on what happened with population movements. It doesn't have the same issues of laggy data sources or of problematic proportionality assumptions as the intercensal population estimates (see my prior post). It does still have some caveats re: surveys conducted in a pandemic, similar to the concerns about representativeness in the decennial Census that forms the baseline for the intercensal population estimates.

What you can see for DC is that the DC vacancy rate for owner-occupied properties (houses and condos not intended for rental) is quite low and hasn't really changed much from 2018 to Q3 of 2021 (no Q4 data yet). There numbers bounce around a bit quarter to quarter, but if you average over the quarters of the year you get an average between 1.2% and 1.4% for each of the last four years. This is notably well below what the vacancy rates were back before 2015 and especially back before 2010 (often over 3%). So if homeowners with kids are moving to the suburbs, it's clear that new homeowners have continued to replace them.

Looking at rental vacancy rates is another story. Rental vacancy rate estimates are much noisier than homeowner vacancy rates (probably because renters and their landlords are harder to get survey responses from). But, the estimates for 2018 and 2019 suggest a baseline of between 6 and 8 percent rental vacancy, which is on the low end for DC by historical standards going back to 2005. The first two quarters of 2020 followed this same pattern (7.4% and 6.8%), but there was a big increase in rental vacancy starting in Q3 2020. It went to 8.7% and then 9.9% in 2020, and then to 10.5% and 11.0% in the first two quarters of 2021, before falling back considerably to 9.9% in Q3 2021. These are mostly within the margin of error of one another (which again, is very big because of the small sample), but taking them at face value suggests a big jump up that has started to retreat but still has quite a ways to fall to get back to where it was before the pandemic.

It's also worth mentioning that these trends don't show up if you look at DC metro level data or national data, only if you look at DC itself. This suggests that city renters left the city specifically, but renters didn't leave rentals more generally. However, since DC is the only city-state we have, it's really hard to know how DC compares to other cities in that regard. It would probably be possible for someone to recreate these statistics using the public CPS microdata for a few of the larger cities in the US, but it would be a project.

Overall, these data say a few things to me. First, as mentioned by myself and others, it looks like the vast majority of the movement has been among renters. The low vacancy rates and strong price appreciation for owned properties in DC doesn't provide any evidence to suggest a decline in demand for this type of living. Second, the big jump in 2020 Q3 and the big decline in 2021 Q3 suggest to me that a good chunk of this is in the form of lack of the usual young transplant pipeline. In other words, recent college grads moved home instead of moving here in summer 2020; in summer 2021 that was perhaps a little less true but still an issue. Going forward, we might expect any big shifts in population and vacancy to coincide with summer turnover and incoming new residents. Third, the issues I raised with lag in the intercensal population projections may not be as bad as I thought. Barring major sampling issues in 2020, it seems pretty clear that the peak of rental vacancy, and presumably of peak population loss, was in early 2021. Fourth, we still only have one quarter of data to evaluate the recovery, but it does seem to be substantially underway.

If you're actually interested in this topic, I'd suggest keeping an eye on the next few quarterly releases of these HVS data. By this time next year, I imagine that we will have a pretty good sense of how durable the rise in rental vacancies is, and we'll have much more price data as well. Q3 2022 will be a particularly interesting data point to see. I would still wait a couple of years before really leaning on the intercensal population estimates, given the issues that I've highlighted before.

Sources:
https://www.census.gov/housing/hvs/data/rates.html
https://www.census.gov/housing/hvs/methodology/index.html

This is a wonderful analysis. Thank you!

Just thought I should follow up on this. The new HVS for the fourth quarter of 2021 came out. DC saw another large decline in its rental vacancy rate, down to 7.6% which is basically the average fourth quarter vacancy rate for 2005-2019.

So, it looks likely that whatever population decline occurred in the District in 2020 and early 2021 has mostly reversed itself by now. Caveats as above that these are still noisy statistics based on fairly small samples, etc.

But don't census numbers show that DC's population decline actually began around 2014-15 and was only exacerbated by the pandemic? How do we explain the earlier decline?

This is not correct. The Census Bureau population estimates for DC went up in every year from 2010 to 2019. The rate of increase did decline at the end of the decade, but there was always a reported increase.

Source: https://www.census.gov/data/tables/time-series/demo/popest/2010s-state-total.html

The other thing to remember is that the annual population estimates always use the prior Decennial and the prior year as their baseline, and then adjust up or down based on tax data, etc. Any issues with the methodology likely compound over the course of the decade and then sort of get recalibrated by the next Decennial. So, the big drop reported in the 2020 Decennial was probably a mix of recalibrating if some of the 2010-2019 estimates overshot the mark, whatever population moves happened between July 2019 and April 1, 2020, and possibly some undercounting in the Decennial because of the pandemic. It's more likely, in my opinion, that DC's population never quite hit 705k then it is that DC's population declined by 15k when the pandemic had only affected the city for a couple weeks. I would also be cautious about comparing pre-2020 figures to post-2020 figures for the same reason.

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:Statistical agency employee here again. I did a little more digging to see what other pieces of information we could get to understand what's actually going on. It seems like a number of people on here are more interested in rehashing their pet complaints than in understanding what's happening, but I'll share for anyone who's interested. If you think my post is too long and just want to complain, save your keystrokes; I don't care.

The Census Bureau also produces something called the CPS Housing Vacancy Supplement (HVS), which reports vacancy rates for owner-occupied and rental housing units for each quarter and for each state and metro. These are pretty noisy estimates, because the CPS only samples something like 72k households a month nationwide, which means each quarterly estimate for DC is based on only ~450 observations split between rentals and owned properties. The vacancy rate is also only an indirect measure of population change, since households can grow or shrink without becoming vacant, and since construction of new units changes the denominator over time. But, this survey provides much more timely data on what happened with population movements. It doesn't have the same issues of laggy data sources or of problematic proportionality assumptions as the intercensal population estimates (see my prior post). It does still have some caveats re: surveys conducted in a pandemic, similar to the concerns about representativeness in the decennial Census that forms the baseline for the intercensal population estimates.

What you can see for DC is that the DC vacancy rate for owner-occupied properties (houses and condos not intended for rental) is quite low and hasn't really changed much from 2018 to Q3 of 2021 (no Q4 data yet). There numbers bounce around a bit quarter to quarter, but if you average over the quarters of the year you get an average between 1.2% and 1.4% for each of the last four years. This is notably well below what the vacancy rates were back before 2015 and especially back before 2010 (often over 3%). So if homeowners with kids are moving to the suburbs, it's clear that new homeowners have continued to replace them.

Looking at rental vacancy rates is another story. Rental vacancy rate estimates are much noisier than homeowner vacancy rates (probably because renters and their landlords are harder to get survey responses from). But, the estimates for 2018 and 2019 suggest a baseline of between 6 and 8 percent rental vacancy, which is on the low end for DC by historical standards going back to 2005. The first two quarters of 2020 followed this same pattern (7.4% and 6.8%), but there was a big increase in rental vacancy starting in Q3 2020. It went to 8.7% and then 9.9% in 2020, and then to 10.5% and 11.0% in the first two quarters of 2021, before falling back considerably to 9.9% in Q3 2021. These are mostly within the margin of error of one another (which again, is very big because of the small sample), but taking them at face value suggests a big jump up that has started to retreat but still has quite a ways to fall to get back to where it was before the pandemic.

It's also worth mentioning that these trends don't show up if you look at DC metro level data or national data, only if you look at DC itself. This suggests that city renters left the city specifically, but renters didn't leave rentals more generally. However, since DC is the only city-state we have, it's really hard to know how DC compares to other cities in that regard. It would probably be possible for someone to recreate these statistics using the public CPS microdata for a few of the larger cities in the US, but it would be a project.

Overall, these data say a few things to me. First, as mentioned by myself and others, it looks like the vast majority of the movement has been among renters. The low vacancy rates and strong price appreciation for owned properties in DC doesn't provide any evidence to suggest a decline in demand for this type of living. Second, the big jump in 2020 Q3 and the big decline in 2021 Q3 suggest to me that a good chunk of this is in the form of lack of the usual young transplant pipeline. In other words, recent college grads moved home instead of moving here in summer 2020; in summer 2021 that was perhaps a little less true but still an issue. Going forward, we might expect any big shifts in population and vacancy to coincide with summer turnover and incoming new residents. Third, the issues I raised with lag in the intercensal population projections may not be as bad as I thought. Barring major sampling issues in 2020, it seems pretty clear that the peak of rental vacancy, and presumably of peak population loss, was in early 2021. Fourth, we still only have one quarter of data to evaluate the recovery, but it does seem to be substantially underway.

If you're actually interested in this topic, I'd suggest keeping an eye on the next few quarterly releases of these HVS data. By this time next year, I imagine that we will have a pretty good sense of how durable the rise in rental vacancies is, and we'll have much more price data as well. Q3 2022 will be a particularly interesting data point to see. I would still wait a couple of years before really leaning on the intercensal population estimates, given the issues that I've highlighted before.

Sources:
https://www.census.gov/housing/hvs/data/rates.html
https://www.census.gov/housing/hvs/methodology/index.html

This is a wonderful analysis. Thank you!

Just thought I should follow up on this. The new HVS for the fourth quarter of 2021 came out. DC saw another large decline in its rental vacancy rate, down to 7.6% which is basically the average fourth quarter vacancy rate for 2005-2019.

So, it looks likely that whatever population decline occurred in the District in 2020 and early 2021 has mostly reversed itself by now. Caveats as above that these are still noisy statistics based on fairly small samples, etc.

But don't census numbers show that DC's population decline actually began around 2014-15 and was only exacerbated by the pandemic? How do we explain the earlier decline?

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:Statistical agency employee here again. I did a little more digging to see what other pieces of information we could get to understand what's actually going on. It seems like a number of people on here are more interested in rehashing their pet complaints than in understanding what's happening, but I'll share for anyone who's interested. If you think my post is too long and just want to complain, save your keystrokes; I don't care.

The Census Bureau also produces something called the CPS Housing Vacancy Supplement (HVS), which reports vacancy rates for owner-occupied and rental housing units for each quarter and for each state and metro. These are pretty noisy estimates, because the CPS only samples something like 72k households a month nationwide, which means each quarterly estimate for DC is based on only ~450 observations split between rentals and owned properties. The vacancy rate is also only an indirect measure of population change, since households can grow or shrink without becoming vacant, and since construction of new units changes the denominator over time. But, this survey provides much more timely data on what happened with population movements. It doesn't have the same issues of laggy data sources or of problematic proportionality assumptions as the intercensal population estimates (see my prior post). It does still have some caveats re: surveys conducted in a pandemic, similar to the concerns about representativeness in the decennial Census that forms the baseline for the intercensal population estimates.

What you can see for DC is that the DC vacancy rate for owner-occupied properties (houses and condos not intended for rental) is quite low and hasn't really changed much from 2018 to Q3 of 2021 (no Q4 data yet). There numbers bounce around a bit quarter to quarter, but if you average over the quarters of the year you get an average between 1.2% and 1.4% for each of the last four years. This is notably well below what the vacancy rates were back before 2015 and especially back before 2010 (often over 3%). So if homeowners with kids are moving to the suburbs, it's clear that new homeowners have continued to replace them.

Looking at rental vacancy rates is another story. Rental vacancy rate estimates are much noisier than homeowner vacancy rates (probably because renters and their landlords are harder to get survey responses from). But, the estimates for 2018 and 2019 suggest a baseline of between 6 and 8 percent rental vacancy, which is on the low end for DC by historical standards going back to 2005. The first two quarters of 2020 followed this same pattern (7.4% and 6.8%), but there was a big increase in rental vacancy starting in Q3 2020. It went to 8.7% and then 9.9% in 2020, and then to 10.5% and 11.0% in the first two quarters of 2021, before falling back considerably to 9.9% in Q3 2021. These are mostly within the margin of error of one another (which again, is very big because of the small sample), but taking them at face value suggests a big jump up that has started to retreat but still has quite a ways to fall to get back to where it was before the pandemic.

It's also worth mentioning that these trends don't show up if you look at DC metro level data or national data, only if you look at DC itself. This suggests that city renters left the city specifically, but renters didn't leave rentals more generally. However, since DC is the only city-state we have, it's really hard to know how DC compares to other cities in that regard. It would probably be possible for someone to recreate these statistics using the public CPS microdata for a few of the larger cities in the US, but it would be a project.

Overall, these data say a few things to me. First, as mentioned by myself and others, it looks like the vast majority of the movement has been among renters. The low vacancy rates and strong price appreciation for owned properties in DC doesn't provide any evidence to suggest a decline in demand for this type of living. Second, the big jump in 2020 Q3 and the big decline in 2021 Q3 suggest to me that a good chunk of this is in the form of lack of the usual young transplant pipeline. In other words, recent college grads moved home instead of moving here in summer 2020; in summer 2021 that was perhaps a little less true but still an issue. Going forward, we might expect any big shifts in population and vacancy to coincide with summer turnover and incoming new residents. Third, the issues I raised with lag in the intercensal population projections may not be as bad as I thought. Barring major sampling issues in 2020, it seems pretty clear that the peak of rental vacancy, and presumably of peak population loss, was in early 2021. Fourth, we still only have one quarter of data to evaluate the recovery, but it does seem to be substantially underway.

If you're actually interested in this topic, I'd suggest keeping an eye on the next few quarterly releases of these HVS data. By this time next year, I imagine that we will have a pretty good sense of how durable the rise in rental vacancies is, and we'll have much more price data as well. Q3 2022 will be a particularly interesting data point to see. I would still wait a couple of years before really leaning on the intercensal population estimates, given the issues that I've highlighted before.

Sources:
https://www.census.gov/housing/hvs/data/rates.html
https://www.census.gov/housing/hvs/methodology/index.html

This is a wonderful analysis. Thank you!

Just thought I should follow up on this. The new HVS for the fourth quarter of 2021 came out. DC saw another large decline in its rental vacancy rate, down to 7.6% which is basically the average fourth quarter vacancy rate for 2005-2019.

So, it looks likely that whatever population decline occurred in the District in 2020 and early 2021 has mostly reversed itself by now. Caveats as above that these are still noisy statistics based on fairly small samples, etc.