Thursday, March 08, 2018

Lawler: More on Deaths ... And Some Thoughts About Next Week’s Release of Census Long-Term Population Projections

by Bill McBride on 3/08/2018 04:41:00 PM

From housing economist Tom Lawler: More on Deaths ...

In discussing available data on deaths earlier this week, I noted that data are deaths are only available with a considerable lag. While that is the case for detailed data on deaths, the National Center for Health Statistics does release provisional estimates for selected indicators of mortality – including both crude and age-adjusted death rates – that are relatively timely. E.g., aggregate crude and age-adjusted death rates are available through the third quarter of 2017, and these data suggested that both crude and age-adjusted death rates increased from 2016 to 2017. For those who track death assumptions in the Census population estimates (which reflect deaths over the 12 month period ending in June), the NCHS provisional death rate estimates imply that US deaths totaled about 2.785 million over the 12-months ended in June 2017, about 41,000 higher than the “Vintage 2017” assumption, and 104,000 above the projection from Census’s 2014 long-term population projection.

If the higher-than-projected death rates remained constant over the next three years (i.e., though 2020), then US deaths over the four-year period ending in 2020 would be over 600,000 higher than the number of deaths assumed in the Census 2014 population projection, with the most of this increase in deaths coming in the 15-44 year old age groups.

US Deaths Per 100,000, 12-month period ending
at end of quarter
CrudeAge Adjusted
Source: NCHS

... And Some Thoughts About Next Week’s Release of Census Long-Term Population Projections

I got confirmation today that Census is on track to release its updated long-term population projections sometime next week. Here are a few things analysts should consider in evaluating/using these projections.

1. The starting point for the projections is “Vintage 2016,” and the latest population estimates (Vintage 2017) are different (though not massively so).

2. The latest data on deaths used for this projection were those for 2015, and I’m guessing the projection does not allow for rising death rates among teenagers and young/middle aged adults that we observed in 2016 and (apparently) 2017. I am guessing that the updated projections may significantly understate deaths for these age groups, and competent analysts should check this.

3. The updated assumptions on net international migration will be “model based,” and will not incorporate any probable or even possible policy changes on immigration. In terms of the immigration component of net international migration, these models look at recent inflows by country relative to their populations, and assume that future immigration will rise proportionately to projections of these countries’ populations. These models will almost certainly over-predict “likely” net international migration over the next few years, and analysts should check the NIM assumptions.

4. My understanding is that Census only plans to release its “baseline” projections next week, but that some alternative scenarios may be released sometime later (though I’m not positive about this).

What I and any competent analyst will do when these new population projections are released is to dig deeply into the key “component of change” assumptions, compare these assumptions either (1) to updated data (e.g., deaths), or (2) to what I believe are “reasonable” assumptions about net international migration, and produce a different set of population projections if the Census assumptions seem “off.” I and competent analysts will also probably produce alternative population projection using different but plausible assumptions about deaths and net international migration.