Blog Post < Previous | Next >
Guy
Bay Area Population Projection Too High
The California Department of Finance, Demographic Research Unit (DRU) recently released their projection for California's population out to 2070. In a recent article, I analyzed their population projection for the whole State of California. Now, I will focus on benchmarking their San Fransisco Bay Area forecast (including the separate forecast for each of the 9 counties).
Right of the bat, the DRU forecast appears overestimated. When you compare their forecast for 2024 (first forecast annual period) vs the early actual estimate, their forecast overestimated the Bay Area population by close to 40,000. That's a large error.
Below, I detail the DRU overestimation in 2024 in %. They overestimated the Bay Area population by close to + 0.50% in the very first year of the forecast.
The DRU has a long history of overestimating population growth.
Regarding this Bay Area forecast, based on historical demographic trends, DRU carried their overestimation up to the end of their forecast in 2070.
Reviewing the Bay Area counties DRU natural growth rate
Natural growth rate is:
(# Births - # Deaths)/Population
This is calculated on a yearly basis. It indicates by how much a population grows or shrinks in % absent any migration.
The graphs below disclose the natural growth rate for the 9 counties. The historical data ranges 2009 to 2023 (before and until the vertical red dashed line). The DRU projections start in 2024 and go out to 2070.
When you look at the graphs above, 7 out of the 9 counties are associated with reasonable natural growth rate projections. Notice that often these natural growth rates turn negative by 2030. This reflects population aging. This is especially the case for Marin County.
However, 2 counties have projected natural growth rates that are much too high. They are:
Contra Costa
Sonoma
While the 7 counties experience a consistent decline in natural growth rate over time, Contra Costa's rate does not decline much, and rises in the outer years.
Meanwhile beyond 2030, Sonoma's natural growth rate surges upward.
From a demographic standpoint, it is unclear why these two counties, with a surge in natural growth in the outer years, would diverge from the other 7.
Next, let's look at the birth rate to see what is going on.
Reviewing the Bay Area counties birth rate
As expected the same pattern emerges here. Contra Costa and Sonoma respective birth rates keep on rising. This is unlike the other 7 counties. For both Contra Costa and Sonoma, such rates keep on rising much above the record levels over the 2009 - 2023 historical period.
The overestimated birth rates for Contra Costa and Sonoma are what bends their respective natural growth rates upward. To generate a reasonable forecast, we will have to fix the birth rates for these two counties.
Fixing DRU's Sonoma and Contra Costa respective birth rates
Doing the fixing is easy and reasonable. I just looked for the other county that had the most similar birth rate history to Contra Costa and Sonoma. And, when building my alternative model projection, I will use these proxy counties' birth rates from 2030 to 2070.
For Sonoma (left-hand graph) the closest match was Napa. You can see that from 2009 to 2023, their respective birth rate curves match up closely.
For Contra Costa (right-hand graph) the closest match was San Mateo. Notice that for the majority of the time from 2009 to 2023, San Mateo's birth rate was much higher than for Contra Costa. There is no reason that during the forecast period, Contra Costa's birth rate all of a sudden would chronically be that much higher than San Mateo's.
Next, let's review the migration rate. The latter is a big component in sustaining demographic growth throughout developed countries who all have by now either flat or negative natural growth rates.
Reviewing Bay Area counties migration rate
The DRU migration rate projections for 3 counties look reasonable. They are:
Napa
San Francisco
Santa Clara
For the other 6 counties, the projected migration rates are too high. Some are way too high (Contra Costa, Solano). Often the projected migration rates revisit or beat the records during the historical period and remain at near record level til 2070 (end of DRU forecast).
On the graphs below, I show such migration rates only until 2035, because they remain nearly unchanged until 2070.
This uptick in migration into the 6 counties is unlikely because:
- Several years into this exacerbated nationwide immigration crisis that cost Democrats the Elections, California and the Bay Area have not experienced such a surge. Current Bay Area migration levels are still much below the highs during the historical period, and they are still most often negative.
- The Work From Home (WFH) is and will have a permanent effect on the Bay Area migration flows. You can see the impact of WFH on the San Francisco office vacancy rate since COVID kicked in (after 2020 Q1).
- Trump will pretty much lock down the border and expel migrants. This will further deflate the migration outlook. Immigration is a bipartisan issue (remember originally it was a Democrat only one tied to labor unions). Thus, the US anti-immigration regime could last for a long while.
- Besides the current immigration crisis drivers such as the war in Ukraine, the near collapse of the Venezuelan government and other dire situations in South America and the Middle East, these drivers are temporary.
- Much of the West, and the Americas are experiencing flattening if not negative natural growth rates. Thus migration from Canada, Mexico, Europe, the Americas into the US will slow down.
How to fix the DRU migration rates that are too high
For the counties with excessively high migration, I took the average of the 6th and 7th highest annual migration rate during the history period (2009 - 2023). This is just a little bit higher than if I would have taken the median (average of the 7th and 8th highest out of 15). And, I used these averages of the 6th and 7th over the 2030 - 2070 forecast period.
Constructing a more realistic forecast
Three counties have a reasonable DRU forecast. So, I left those unchanged. They had both reasonable natural growth rate, birth rate, and migration rate forecasts.
Six counties have an alternative forecast, much lower than DRU
For the other 6 counties, I adjusted the migration rate for all 6 of them. And, I adjusted the birth rate for Contra Costa and Sonoma. My resulting forecast (Model, green line) was often very much lower than DRU's. This is especially the case for Contra Costa, Solano, and Sonoma, as shown in the graphs below.
My alternative Model comes up with close to 1 million fewer by 2070 vs the DRU projections. And, I do that by just using the mentioned more realistic migration rates and birth rates.
Below, I am just outlining the respective percentage growth (left table) over the whole period and the annual growth rates (right table).
Below I just graph the Bay Area total population projections for the DRU vs my model.
DRU forecast generates nearly an extra 1,000,000 by 2070.
The level of overestimation at around 1 million is in line with DRU's overestimation level in their 2023 forecast.
When focusing on Marin County, the DRU's overestimation still pales compared to local officials who believe that Marin County's population could grow by 34.6% to 49.4% over the next 20 years. That is what they infer by suggesting that the Larkspur Ferry parking lot should be expanded by that much because of projected growth in Marin County's population.
THE END
Gaetan Lion is an independent researcher with expertise in economics, econometrics, investments, risk management, statistics, and quantitative analysis. His areas of interest include demographics, economics, investments, Climate Change, health care, sports, and politics.