Projections of Household Vehicle Consumption in the United States

Abstract

Forecasts of household vehicle consumption are important for automobile market analyses and related socioeconomic planning. This chapter employs the ProFamy extended cohort-component method to project household vehicle consumption from 2000 to 2025 across four regions of the United States (the Northeast, Midwest, South, and West). The results show that the total number of household vehicles (The term “household vehicles” refers to vehicles for household use in this book) in 2025 will reach 235 million, representing a 31 % increase over 25 years. About a half of the increase is due to the consumption of cars, while the household consumption of vans will increase at a faster rate than that of cars and trucks. Household vehicle consumption will grow more in white non-Hispanic and Hispanic households in comparison with black non-Hispanic and Asian and other non-Hispanic households. Owners of household vehicles in the United States will be aging quickly. Among households of different sizes, the largest increase in household vehicles will come from two-person households. Across the four regions, the largest increase in household vehicle consumption will be in the South, followed by the West, Midwest, and Northeast.

Appendices

Appendix 1: The Four Regions Defined By the US Census Bureau

Northeast (9 states): Maine, New Hampshire, Vermont, Massachusetts, Rhode Island, Connecticut, New York, New Jersey, Pennsylvania;

Midwest (12 states): Ohio, Indiana, Illinois, Michigan, Wisconsin, Minnesota, Iowa, Missouri, North Dakota, South Dakota, Nebraska, Kansas;

South (16 states and DC): Delaware, Maryland, District of Columbia, Virginia, West Virginia, North Carolina, South Carolina, Georgia, Florida, Kentucky, Tennessee, Alabama, Mississippi, Arkansas, Louisiana, Oklahoma, Texas;

West (13 states): Montana, Idaho, Wyoming, Colorado, New Mexico, Arizona, Utah, Nevada, Washington, Oregon, California, Alaska, Hawaii.

Appendix 2: Consistency Examination Across the Four Regions

• 1. The sum of all domestic immigrants and domestic out-migrants in the four regions in each of the projection years is equal to zero;

• 2. The sum of the net international migration of the four regions in each of the projection years is equal to the whole country’s net international migration in the corresponding year forecasted by the Census Bureau;

• 3. In the years of the forecasting, the estimated region-race-specific summary measures must pass the consistency check. That is, the weighted race-specific average summary measures for the whole country, which are derived from the estimated region-race-specific summary measures and the regional proportions of each race group, are equal to the directly estimated summary measures of the whole country.

Appendix 3: Consistency Examination for Percentile Distribution of Income Categories

The age-race-specific proportions of the high, middle I, middle II, and low incomes for each household type/size category from 2001 to 2025 are assumed to be the same as those obtained from the 2000 census 5 % sample dataset, as has been justified in the existing literature and discussed in the text. But the aggregated race-specific proportions of the four income categories for each household type of all ages combined are not constant over time because they are a weighted average of the proportions across ages, and the age structure of the householder (i.e., the weights of the aggregate proportions) change over time. Similarly, the age-specific proportions of each income category of all races combined for each household category are not constant over time because they are a weighted average of the proportions across races, and the race compositions of households change over time. In sum, the overall proportions of each income category for each race and all ages combined, the age-specific proportions of each income category for all races combined, and the overall all-age-race-combined proportion of income category are dynamic from 2000 to 2015, due to changes in the household distributions and age structure of the householder. At the same time, the census-based or American Community Survey (ACS)-based age-race-household category-specific proportions of each income category, which measure the race-age-sex-region differentials of income distributions, are basically kept constant.

The procedure for the consistency check is as below:

I _k (t), the percent of income category k;

P _k (i, x, t, r, j), proportion of households of kth income category among households of type/size i with householder of age group x, race group r, and region j in year t; one may assume that P _k (i, x, t, r, j) in the projection year is the same as that observed in the most recent year or assume some systematic changes. In any case, ${\displaystyle \sum_k{P}_k\left(i,x,t,r,j\right)=1.0}$

H(i, x, t, r, j), number of projected households of type/size i with householders of age group x, race group r, and region j in year t;

H(i, x, t, r, j) P _k (i, x, t, r, j), the first estimate of the number of households with income category k, household type/size i, and householder of age x, race group r, and region j in year t.

Because of the changes in compositions of households of different types/sizes, and age structure of householders in projection year t, ${\displaystyle \sum_i{\displaystyle \sum_x{\displaystyle \sum_r{\displaystyle \sum_j}}H\left(i,x,t,r,j\right){P}_k\left(i,x,t,r,j\right)}}/{\displaystyle \sum_i{\displaystyle \sum_x{\displaystyle \sum_r{\displaystyle \sum_j}}H\left(i,x,t,r,j\right)}}$ may not be exactly equal to I _k (t) although the discrepancy is usually not large. Thus some adjustments are needed as below.

$$ \frac{C_k(t){\displaystyle \sum_i{\displaystyle \sum_x{\displaystyle \sum_r{\displaystyle \sum_j}}}}H\left(i,x,t,r,j\right){P}_k\left(i,x,t,r,j\right)}{{\displaystyle \sum_i{\displaystyle \sum_x{\displaystyle \sum_r{\displaystyle \sum_j}}}}H\left(i,x,t,r,j\right)}=0.25 $$

$$ {C}_k(t)=\frac{0.25{\displaystyle \sum_i{\displaystyle \sum_x{\displaystyle \sum_r{\displaystyle \sum_j}}}}H\left(i,x,t,r,j\right)}{{\displaystyle \sum_i{\displaystyle \sum_x{\displaystyle \sum_r{\displaystyle \sum_j}}}}H\left(i,x,t,r,j\right){P}_k\left(i,x,t,r,j\right)} $$

(11.1)

$$ P{\hbox{'}}_k\left(i,x,t,r,j\right)={C}_k(t){P}_k\left(i,x,t,r,j\right) $$

(11.2)

$$ P\hbox{'}{\hbox{'}}_k\left(i,x,t,r,j\right)=P{\hbox{'}}_k\left(i,x,t,r,j\right)\frac{1.0}{{\displaystyle \sum_kP{\hbox{'}}_k\left(i,x,t,r,j\right)}} $$

(11.3)

We compute the quartiles of high, middle I, middle II, and low income again if their relative differences from 0.25 are all less than 0.01, say, or another criterion, we accept P’’  _k (i, x, t, r, j). More specifically, if $\left\{\left[\frac{{\displaystyle \sum_i{\displaystyle \sum_x{\displaystyle \sum_r{\displaystyle \sum_j}}}}H\left(i,x,t,r,j\right)P\hbox{'}{\hbox{'}}_k\left(i,x,t,r,j\right)}{{\displaystyle \sum_i{\displaystyle \sum_x{\displaystyle \sum_r{\displaystyle \sum_j}}}}H\left(i,x,t,r,j\right)}-0.25\right]/0.25\right\}<0.01$ for all income categories k (for example, k = 1, 2, 3, 4), we accept P’’  _k (i, x, t, r, j). Otherwise, we repeat the adjustment procedure expressed in formulas (11.1), (11.2), and (11.3) until the criterion is met.