Jobs & Contracts
Meetings & Events
Maps & GIS
This technical summary reports MTC auto ownership level forecasts for all forecast years included in
the ABAG's Projections '98 database (1990, 1995, 2000, 2005, 2010, 2015 and 2020). Also included are
MTC interpolations for years 1996 and 1998. These 1996 and 1998 intermediate year forecasts are for use
in MTC transportation planning studies, including base year updates for the balance of MTC's travel
These auto ownership level forecasts are prepared at MTC's 1099 regional travel analysis zone level.
For sake of comprehension, results are summarized and reported at MTC's 34 superdistrict and nine
county level. Superdistricts are aggregations of MTC's regional travel analysis zones and are used in
the analysis of sub-county level demographic and travel forecasts.
Following this introduction are sections discussing the technical details of the data inputs and
models; and a section discussing the detailed findings.
The MTC model that predicts the share of households by workers in household and by vehicles in
household level is the "Workers in Household, Auto Ownership" choice model, or WHHAO.
In casual conversation we may call this the "MTC auto ownership model." Strictly speaking, this
model is used to predict the distribution of households by vehicles available to the
household, using a decennial census definition of vehicles as "automobiles, vans and trucks of one-ton
capacity or less that are kept at home for use by members of the household."
The MTC WHHAO model was estimated using data from the 1990 MTC household travel survey, and then
calibrated and validated against household data from the 1990 Census (Census Transportation Planning
Package / Urban Element.) Calibration and validation was completed at the MTC 34 superdistrict level.
This means that the results from the base year model validation for 1990 were adjusted to match 1990
Census counts of households by workers in household and vehicles available in household levels.
WHHAO model inputs include:
Two of these input variables - share of multi-family of total households and share of population age
62-or-older - are developed by MTC staff using data from the 1990 Census and from ABAG's county-level
age cohort forecasts. ABAG does not estimate small area population-by-age nor small area single-family
vs. multi-family household data.
The WHHAO model is a nested logit choice model. Logit choice models are a commonly
used model formulation used in travel models with discrete choices (e.g., auto ownership level or mode
choice). The "upper level" of the WHHAO nested logit choice model splits the total number of households
into households by workers in household. The "lower level" nest of this model further splits the
households by auto ownership level. A diagram of this nested logit choice model is shown below:
In the above graphic, NWHH refers to non-working (or zero worker) households; SWHH refers to
one-worker (single worker) households; and MWHH refers to multi-worker households. The term AO refers
to auto ownership (strictly speaking, vehicle availability).
Note that the data inputs to this model are stratified by household income quartile. This means that
the zone-of-residence level output from this model are the number of households stratified by income
quartile (4), by workers in household (3), by vehicles in household (3), for a total of 36 different
types of households by each travel analysis zone.
(Technical readers interested in the detailed inner workings of this nested logit choice model may want to refer to a 10/3/95 technical memorandum included in the MTC report San Francisco Bay Area: 1990 Travel Model Development Project: Compilation of Technical Memorandum: Volume III (March 1996). This technical memo provides detailed examples on how to apply this particular nested choice model at a disaggregate, or household level. Other non-published memos are available that show how this disaggregate choice model is adapted for application at an aggregate, or zone-of-residence level.)
Regional and county tabulations are included in Tables 1 and 2. Results of the workers in household
level forecasts and related statistics are included in Table 1. Results of the vehicles in household
level forecasts and related statistics are included in Table 2. Superdistrict-level tabulations are
included as Tables S.1 through S.16.
Superdistrict, county and regional level data from the 1980 and 1990 decennial censuses is included
in Tables A.1 through A.3. This data is used to convert the number of two-or-more vehicle households
into the total number of vehicles available to households. A map of the MTC 34 superdistricts is
included at the very end of this technical summary.
These data are not discussed in the Summary of Findings but are discussed in
this section. These results are from the "upper nest" of the MTC Workers in Household, Auto Ownership
choice model (WHHAO).
Non-working households include retired households and unemployed households. Analysis of 1990 Census
data for the Bay Area suggests that approximately 75 percent of non-working households are "retired
households" (e.g., head of household is of retirement age.) Other non-working households are a mix of
households with unemployed workers, disabled person households, or households of single
Key in the prediction of households by workers in household level is the number of employed
residents in the travel analysis zone of residence, the share of population age 62-or-over, and the
average household income by income quartile. Certain neighborhoods with very high shares of elder age
population and low numbers of employed residents are likely to be retirement communities with very high
shares of non-working households (e.g., Leisure World in Vacaville and Rossmoor in West Walnut Creek.)
Regionally the number of non-working households is projected to decline between 1990 and 2010, from
479 thousand non-working households (1990) to 449 thousand non-working households (2010). The aging of
the baby boomer population is expected to reverse this trend in non-working (retired) households, with
an increasing number of non-working households to the year 2020 - 525 thousand non-working households.
The regional share of non-working households is projected to decrease from a high of 21.3 percent of
total households in 1990 to a low of 16.9 percent of total households by the year 2010, then increasing
back to 18.5 percent non-working households by 2020.
At a county level, Napa County has the highest share of non-working households. In 1990, 27.7
percent of Napa County households were non-working. By the year 2020 this share is projected to
decrease to 23.4 percent of all Napa households.
Santa Clara County has the lowest share of non-working households in the nine-county Bay Area. In
1990, 16.0 percent of Santa Clara County households were non-working. By the year 2020 the Santa Clara
County share is expected to decrease to 15.5 percent of all households. The year 2010 forecast for
Santa Clara County shows a low of 13.2 percent non-working of total households.
Regionally, the number of multi-worker (two-or-more) households is projected to increase from 0.94
million households in 1990 to 1.29 million households by the year 2020. Multi-worker household
forecasts for 2015 are identical at the regional level to 2020. Between 2015 and 2020 the number of
multi-worker households is projected to decrease in the central Bay Area counties, and is offset by
increases in multi-worker households in north bay counties. The regional share of multi-worker of total
households is projected to steadily increase from 41.9 percent in 1990 to 48.2 percent by the year
2010. This share is then expected to drop to 45.5 percent by the year 2020.
Santa Clara County has the highest share of multi-worker counties in the Bay Area. Santa Clara's
share of multi-worker households is projected to increase from 47.8 percent of households in 1990 to
52.5 percent of households by 2010. This multi-worker household share is then projected to slide back
to 49.2 percent by the year 2020.
Other counties with very high multi-worker county shares for 2020 include Solano (47.9 percent), San
Mateo (47.5 percent) and Sonoma (46.6 percent).
San Francisco County shows the largest increase in multi-worker household share, increasing from
just 33.9 percent of San Francisco households in 1990 to 44.5 percent of households by the year 2020.
This is due to increasing workers per household as predicted by ABAG in Projections '98 - from 1.28
workers per household in 1990 to 1.40 workers per household in 2020.
The third market - single worker (one-worker) households is expected to steadily increase from 827
thousand households in 1990 to 1.02 million households by the year 2020. The share of single worker of
total households is expected to drop from 36.8 percent in 1990 to 34.7 percent by 2005, then to
increase to 36.0 percent by 2020.
County-level shares of one-worker households tend to hover between 33 and 40 percent. San Francisco
and Marin have the highest shares of one-worker households in the Bay Area, both historically as well
Again, the Summary of Findings section of this report provides a re-cap of
the highlights of these forecasts. These findings are not repeated in this section.
The regional number of zero-vehicle households is projected to decline between 1990 and 2010, then
to increase between 2010 and 2020. This parallels the change in non-working households discussed in the
previous section. Regionally this works out to 237 thousand zero-vehicle households in 1990, 204
thousand zero-vehicle households in 2010, and 207 thousand zero-vehicle households in 2020. In terms of
share of zero-vehicle of total households, these are expected to decline from a high of 10.6 percent of
households in 1990 to 7.3 percent of all Bay Area households by the year 2020.
The regional number of one-vehicle households is also expected to decline, dropping from 725
thousand households in 1990 to 700 thousand households by the year 2000. Between 2000 and 2020, the
number of one-vehicle households in the Bay Area is projected to increase from 700 thousand to 760
thousand. The share of one-vehicle households is projected to decrease from 32.3 percent of households
in 1990 to 26.8 percent of all households by 2020.
The regional trend in multi-vehicle households shows a constant increase, rising from 1.28 million
households in 1990 to 1.87 million households by the year 2020. The regional share is also expected to
steadily increase from 57.2 percent of Bay Area households in 1990 to 65.9 percent of households by the
All of the relevant county-level vehicle ownership results are discussed in the Summary of Findings.
The California State Department of Motor Vehicles (DMV) publishes an end-of-year estimate of
"fees-paid" registered vehicles by county-of-registration. Data is published for automobiles,
commercial vehicles, trailers and motorcycles. According to the DMV, the DMV statistics are the:
"estimated number of fee-paid vehicle registrations which occurred in California
counties during the past calendar year. It represents the official DMV count of vehicle registration
transactions. This count should not be used for revenue projections because it is based upon a
percentage distribution formula. It is not an exact count." (http://www.dmv.ca.gov/profile/97fee.htm)
According to the federal Truck Inventory and Use Survey (TIUS, 1997), 68.0 percent of
commercial-plated vehicles are light duty trucks. MTC staff apply this 68.0 percent figure to DMV
commercial registrations, summed with automobile registrations, to obtain a census-comparable "personal
use estimated fee paid vehicles" by county-of-registration. This technique works fairly poorly for 1990
- the Census Bureau (and MTC) estimates 3.95 million household vehicles in the nine-county Bay Area;
the State DMV, 4.32 million household vehicles. On the other hand, the comparison for 1997/98 is fairly
good: MTC estimates 1998 vehicle ownership at 4.37 million vehicles; DMV-based estimates of December
1997 vehicle ownership is also 4.37 million vehicles. The following table summarizes these DMV to MTC
DMV and MTC Estimates of Vehicles Registered - 1997/98
DMV-based estimates of personal use vehicles show just a 1.7 percent increase between
December 1989 and 1997, from 4.32 million to 4.37 million vehicles. This same technique yields 4.52
million personal use vehicles registered in the Bay Area for 1996. This represents a 3.3 percent
decrease in vehicles registered in the Bay Area between December 1996 and December
1997. This apparent reduction in vehicles registered in the Bay Area (and California) is due, in part,
to seasonal fluctuations in license registration revenue received by the DMV over the years. Another
plausible explanation is that increased insurance requirements in California have resulted in an
increased number of unlicensed vehicles.
DMV data may serve as an independent check on MTC vehicle ownership forecasts.
Unfortunately the DMV does not provide estimates of the number of households by vehicle ownership level
(the basic data that MTC is predicting!) The best check on these forecasts will be to compare these
forecasts against data from the Year 2000 Decennial Census. Year 2000 Census data should become
available by around the year 2002.
MTC superdistrict, county and regional level auto ownership and workers in household level forecasts
are included as Tables S.1 through S.16. Data is provided for all forecast years, including the
interpolated demographic forecasts for the years 1996 and 1998.
Total households by MTC superdistrict is (intentionally) repeated in Table S.4 and S.11, as a
summary check on households by workers in household and vehicles in household level.
It is important to note the wide variation in the number of households by superdistrict. In 1990,
this ranges from a low of 13 thousand households in superdistrict #28 (upper Napa County) to a high of
165 thousand households in superdistrict #18 (greater Oakland and Alameda). Most of the comparative
analyses included in this section relate to the share of households by either workers in household or
vehicles in household level.
The distribution of households by workers in household, by superdistrict, is the subject of Tables
S.1 through S.7. In 1990, the non-working household share ranged from a low of 10.6 percent of
superdistrict #13 (South San Jose / Almaden) households to a high of 35.9 percent of superdistrict #1
(northeastern San Francisco) households (see Table S.5). These two superdistricts share the same
distinction in 2020, with 8.3 percent of superdistrict #13 and 29.1 percent of superdistrict #1
households projected to be non-working.
Alameda County has the largest range in non-working household shares, ranging (in 2020) from a low
of 8.7 percent in superdistrict #16 (Fremont / Union City) to a high of 28.1 percent in superdistrict
#18 (Oakland / Alameda).
In Contra Costa County, the fairly high share of non-working households in superdistrict #22 (Walnut
Creek / Lamorinda) (26.0 percent in 1990 increasing to 28.8 percent in 2020) is due in part to the
Rossmoor retirement community in western Walnut Creek.
Other superdistricts with high shares of non-working households in 2020 include: superdistrict #20
(Richmond / El Cerrito), superdistrict #28 (St. Helena / Calistoga), and superdistrict #18 (Oakland /
Alameda), all at 28.1 to 28.2 percent non-working households.
In terms of multi-worker household shares (see Table S.7), superdistrict #1 (northeastern San
Francisco) has the lowest share, at 21.8 percent of 1990 households and 30.4 percent of 2020
households. Superdistrict #12 in Santa Clara County (Milpitas / East San Jose) has the highest
multi-worker household shares, increasing from 57.0 percent of households in 1990 to 62.0 percent of
households by the year 2020.
Other superdistricts with high shares of multi-worker households in 2020 include: superdistrict #13
(South San Jose / Almaden) at 59.7 percent; superdistrict #16 (Fremont / Union City) at 58.6 percent;
and superdistrict #15 (Livermore / Pleasanton) at 56.3 percent.
The share of households with zero vehicles (see Table S.12), at superdistrict level, ranges from a
low of 1.0 percent of superdistrict #23 (Danville / San Ramon) households to a high of 62.1 percent of
superdistrict #1 (northeastern San Francisco) households (data for 1990.) For the year 2020, the zero
vehicle household share is projected to decrease to 0.7 percent of superdistrict #23 households and
52.9 percent of superdistrict #1 households.
Other superdistricts with high projected shares of zero vehicle households include superdistrict #2
(Richmond District of San Francisco) at 19.3 percent; superdistrict #18 (Oakland / Alameda) at 16.8
percent; superdistrict #3 (Mission District of San Francisco) at 15.1 percent; and superdistrict #19
(Berkeley / Albany) at 13.6 percent zero vehicle households.
Other superdistricts with low projected shares of zero vehicle households include superdistrict #15
(Livermore / Pleasanton) at 1.6 percent; superdistrict #31 (Healdsburg / Cloverdale) at 1.7 percent;
and superdistrict #15 (Fremont / Union City) at 2.1 percent. Outside of San Francisco, Alameda County
has the widest range of zero-vehicle household shares, ranging from a projected low of 1.6 percent of
East Alameda County households to a high of 16.8 percent of Oakland/Alameda City households.
For 1998, the share of households with zero vehicles ranges from a low of 0.9 percent of
Danville/San Ramon Valley area households to a high of 60.1 percent of greater downtown San Francisco
The share of households with two-or-more vehicles (see Table S.14), at superdistrict level, is
projected to range from a low of 13.2 percent of households in superdistrict #1 to a high of 85.3
percent of households in superdistrict #23 (Danville / San Ramon).
Other superdistricts with high projected shares of multi-vehicle households include: superdistrict
#15 (Livermore / Pleasanton) at 83.0 percent; superdistrict #13 (South San Jose / Almaden) at 81.6
percent; superdistrict #14 (Gilroy / Morgan Hill) at 81.1 percent; and superdistrict #16 (Fremont /
Union City) at 80.1 percent share of multi-vehicle households for the year 2020.
The other San Francisco superdistricts, Berkeley and Oakland are projected to have the lowest share
of multi-vehicle households in the Bay Area by year 2020.
Average total vehicles available per household, by MTC superdistrict, is shown in Table S.16.
Greater downtown San Francisco (superdistrict #1) has the lowest ownership rate, at 0.49 vehicles per
household in 1990 climbing to 0.63 vehicles per household by the year 2020.
Four superdistricts are projected to have year 2020 average vehicles per household in excess of 2.3
vehicles per household: superdistrict #14 (Gilroy / Morgan Hill) at 2.373 vehicles/household;
superdistrict #23 (Danville / San Ramon) at 2.358 vehicles/household; superdistrict #12 (Milpitas /
East San Jose) at 2.339 vehicles/household; and superdistrict #13 (South San Jose / Almaden) at 2.303
vehicles per household.
For 1998, average vehicles per household ranges from a low of 0.527 vehicles per household in greater downtown San Francisco to a high of 2.332 vehicles per household in the Danville / San Ramon Valley area of Contra Costa County.
Travel Demand Models for the San Francisco Bay Area: Technical Summary (June 1997).
This report provides technical specifications for the entire MTC modeling system, including the workers
auto ownership choice model. /datamart/forecast/baycast1.htm
Auto Ownership in the San Francisco Bay Area: 1930-2010 (July 1997). This report is a compilation of statistics from the decennial census, the state DMV, and older vintage MTC auto ownership forecasts. Auto Ownership Paper
Superdistrict and County Summaries of ABAG's Projections '98: 1990-2020: Technical Summary (July 1998). This report summarizes ABAG's Projections '98 databases at MTC's 34 superdistrict and nine-county level.
Metropolitan Transportation Commission • 101 Eighth Street, Oakland, California 94607
This page was last modified Friday November 09, 2012
© 2013 MTC