Friday, February 23, 2018

Lawler: "Census: Population Growth Slowed Slightly in 2017: Immigration Uncertainty Clouds Outlook"

by Calculated Risk on 2/23/2018 03:58:00 PM

From housing economist Tom Lawler: Census: Population Growth Slowed Slightly in 2017: Immigration Uncertainty Clouds Outlook

At the end of last year the Census Bureau released its estimate of the US population for July 1, 2017 and its updated estimates for previous years (the so-called “Vintage 2017” estimates). As of July 1, 2017 Census estimated that the US resident population totaled 325,719,178, up 2,313,062 (or 0.71%) from the upwardly-revised estimate for July 1, 2016. The estimated population increase for 2017 was slightly lower than the 2,366,096 estimated gain in 2016, reflecting slightly lower estimated births, slightly higher estimated deaths, and slightly lower net international migration. Revised population estimates for previous years mainly reflected revised estimates of net international migration, which were driven by an updated methodology for estimating foreign-born emigration and net native-born migration. Below is a table comparing Vintage 2017 population estimates compared to the Vintage 2015 and the Vintage 2016 estimates. The Vintage “estimates” for the subsequent year reflect near-term projections for each Vintage.

Estimated US Resident Population
Vintage 2015Vintage 2016Vintage 2017

While Census had originally planned to release updated long-term population projections in December (the last long-term population projections report was released at the end of 2014 with “Vintage 2013” as a starting point), “senior officials” delayed the updated population release until sometime in March. Many analysts use these population projections to project such things as labor force growth, household growth, etc. Unfortunately, the Census population projections from 2014 are woefully out of date. Amazingly, however, some analysts still use these old projections. E.g., in October of last year the Bureau of Labor Statistics produced updated labor force projections over the next decade based on the outdated Census population projections. Since, as discussed later, the Census 2014 population projections significantly overstated not just actual population growth but likely population growth over the next few years, projection of key economic variables based on these outdated population forecasts are of little use to anyone.

Below is a table showing the US resident population projections from the Census 2014 projections compared to the Vintage 2017 estimates.

Vintage 2017Census 2014 ProjectionsDifference

As the table indicates, the Census 2014 population projection for 2017 were 906,613 higher than the latest 2017 estimate, and the projection for 2018 was 1,223,225 above the latest short-term forecast.

For 2017, here is a breakdown of the 2014 projections’ “component of change” assumptions for 2014 through 2017 compared to the Vintage 2017’s estimates.

Cumulative Components of Change from 7/1/2013 to 7/1/2017
Vintage 2017Census 2014 ProjectionsDifference
Net International Migration4,375,4164,979,119-603,703
Components of Change9,484,67310,496,952-1,012,279
7/1/2013 Starting Point316,234,505316,128,839105,666
7/1/2017 Population325,719,178326,625,791-906,613

As the table indicates, over the 2014 through 2017 period births were lower, deaths were higher, and net international migration was significantly lower (according to the latest estimates) than the assumptions made in the Census 2014 population projections.

While Census has not yet released the Vintage 2017 population estimates by detailed age (these estimates probably won’t be available until June), the 7/1/2017 estimate of the US resident “adult” (18+) population is 252,063,800, or 778,438 below the projection for that date from the Census 2014 population projection report.

For the 2014 to 2017 period (using “actual” labor force participation rates), US labor force growth was about 0.1% per year lower than would have been the case if the Census 2014 population projections had been accurate. Over than same period my “best guess” is that US household growth over than period was about 118,000 lower than would have been the case if the Census 2014 population projections had been accurate. I use the term “best guess” because there are no reliable estimates of the number of US households for that period. (The so-called “household estimate conundrum.”)

The gap between Census 2014 population projections and likely “actual” population counts will almost certainly widen over the next few years, partly because of higher death rates but mainly because the Census 2014 assumptions on net international migration over the next few years seem way too high given the current political climate. As the chart below shows, the Census 2014 population projections assumed that net international immigration would be substantially higher over the 2018-2020 period than was the case over the past several years. Given the current political climate, such an assumption seems very unrealistic.

Net MigrationClick on graph for larger image.

On the immigration assumption front, my “gut” tells me that a major reason Census officials delayed the release of new long-term population projections is that it is far from clear how to assess what immigration numbers are likely to look like under an Administration that has expressed its support for legislation that would sharply reduce immigration. It also is a sensitive political topic, and it seems doubtful that Census officials would want to put out “official” government projections that showed rising net immigration numbers. I’m guessing that when Census finally releases its updated population projections, it will show different immigration scenarios, as they did in the 2012 population projection release, but I also believe that its “base’ or “middle net international migration scenario will show considerable lower numbers than those in the 2014 projections.

For analysts wishing to make economic projections that are dependent on population projections, there a few choices: (1) forego making any projections until Census releases its updated population forecasts, which is tentatively scheduled for release sometime next month; (2) make their own population projections based on their own assumptions about births, deaths, and (especially) net international migration; or (3) (NOT RECOMMENDED) continue the use the outdated Census 2014 population projections because, well, er, they are the last officially released projections!

When the Census’ new long-term population projections are released (again, probably next month), there are a few things analysts should take into account before using them as inputs to other economic/demographic variables. First, of course, analysts should look closely at the underlying assumptions in the projections, and make an assessment of whether they are “reasonable” (with special attention to the assumptions on net international migration). Second, it is highly likely that the “starting point” for Census’ updated population projections will be the “Vintage 2016” population estimate for July 1, 2016. As noted in the table on page one, the population estimate for this date was revised upward at the end of last year by 278,782. As such, rigorous analysts will need to adjust the Census projections to reflect these updated estimates, and may also want to adjust the Census projections to reflect their own views of the key components of change (again, especially with respect to net international migration).

Net, of course, the updated long-term Census population projections released (probably) next month are likely to show SUBSTANTIALLY slower projected population growth than the Census 2014 projections, and folks who only use “official” Census population projections to forecast such things as labor force growth and household formations will show substantially slower growth in these variables from their previous forecasts.