28
The Color of Wealth in Boston
Endnotes
1
As defined by the U.S. Census Bureau, race categories reflect a social definition of race recognized in the United States.
Categories of race are based on respondents’ self-identification and include the following: White, Black or African
American, American Indian and Alaska Native, Asian, and Native Hawaiian and Other Pacific Islander. The concept of
race is separate from the concept of Hispanic origin or ethnicity. In addition to race and ethnicity, the NASCC survey
asked about ancestry and country of origin.
2
The Boston MSA includes the following counties: Essex, Middlesex, Norfolk, Plymouth, and Suffolk in Massachusetts;
and Rockingham and Strafford New Hampshire.
3
All population figures come from the 2012 American Community Survey 1-year estimates. The share of the non-His-
panic white population declined from 81 percent in 2000 to 74 percent in 2012.
4
As of 2012, there were 3,435,332 white residents; 329,500 black residents; 318,181 Asians and Pacific Islanders; and
444,517 Hispanics in the Boston MSA. These categories do not include mixed-race individuals with the exception of
Hispanics/Latinos who may be of any race. Most Hispanics self-identify as “other race” in the U.S. Census.
5
U.S. Census projections at the national level estimate that by 2030 non-Hispanic whites will account for 55 percent of
the nation’s population. Hispanics and non-Hispanic blacks will represent 22% and 13%, respectively. Unfortunately,
population projections at the state level by race and ethnicity are not available.
6
In the United States in 2012, Puerto Ricans and Dominicans accounted for 9.4 percent and 3.1 percent of the Hispanic
population, respectively.
7
Gateway cities in Massachusetts are economically struggling mid-size urban centers. In this report, we use the Federal
Reserve Bank of Boston’s definition of “working cities,” that is, cities in Massachusetts with a population above 35,000
(excluding Boston) that have below median family income and above median poverty rates.
8
U.S. Census Bureau, 2012 American Community Survey, 1-year estimates.
9
The “three-decker” is a unique housing type characteristic of New England cities in the early 20th century. Generally
defined, it is a freestanding, three-story wood frame structure on a narrow lot. Triple-deckers (as they are also called)
are designed as multifamily housing with one family living on each floor, including the owner who typically pays the
mortgage by renting the other two units. They are the dominant housing stock in Dorchester where nearly 5,000 such
structures exist (see Krim 1977).
10
The Center for Survey Research (CSR) at the University of Virginia was the subcontractor that administered the survey.
Tom M. Guterbock, director for the CSR, directed the survey administration. The surveys were translated into Spanish
and Portuguese for the Boston study sample. To complete the survey took an average of 39 minutes.
11
The sample also includes a smaller sample of Asians (14) that we don’t analyze in this report because the sample is not
large enough and because we are not able to differentiate among different subgroups within the Asian category.
12
For the NASCC project in general, about 70,000 personalized advanced letters were sent, 87,000 telephone numbers
dialed 448,000 times, and 12,113 interviewer hours were spent across three shops to conduct 2,746 completed surveys.
13
Among NASCC households, a higher percentage of heads of household have completed college as compared with
households represented in the U.S. Census Bureau’s American Community Survey (ACS) data. For example, for black
households the percentage was 21 percent (ACS data) as compared with 42 percent (NASCC data). Among Hispanics,
NASCC data on educational attainment is similar to the ACS data.
14
In general, the median age of the head of household and the percentage of married households was higher in the NA-
SCC dataset than in the ACS dataset.
15
The median family income for blacks and whites was 10% lower among NASCC households than among ACS house-
holds.
16
We report significance at the 90%, 95% and 99% levels. However, given our small sample sizes it may be difficult to
detect significance at those levels even if differences exist. This is particularly true when estimating asset, debt, and net
worth values. The p-values will be conservative, increasing the likelihood of not detecting significance when, in fact,
there may be significance. Thus, differences in medians can be treated as meaningful in some cases even when statistical
significance is not found at traditional levels.
17
Because of some very high values, using the mean, skews upward estimates of what a typical family owns when measur-
ing wealth. This is especially relevant when comparing groups with small sample sizes, where arithmetic means will be
even more sensitive to outlier values.
18
Cash is not included in these calculations.
19
Tippett et al. (2014) found that 80 percent of whites, 55 percent of blacks, and 60% of Hispanics held checking ac-
counts.
20
It is worth noting that the percentage of Caribbean black households owning savings accounts and the percentage of
Cape Verdean households owning savings accounts did not differ in a statistically significant way.