Arguably, the most persistent demographic trend through the centuries in Virginia and the U.S. has been the difference in socioeconomic status between Black Americans and non-Black Americans. By many measures the socioeconomic gap between Black Americans and non-Black Americans has not changed considerably in half a century, adding fuel to assertions that many of our social institutions are structurally racist and disproportionately exclude Black Americans.

While an income gap exists between the incomes of Black Virginians and other Virginians, for some Black Virginians, the income gap is much smaller or almost non-existent. Exploring why this difference in income exists among Black Virginians can help us to understand at least one dimension of why Black Virginians typically earn less than other Virginians.

Since 1970, the income gap between Black and non-Black Virginians has not grown considerably smaller or larger. Though it has moved down in recessions and up during economic expansions, the median family income for Black Virginians has hovered around 70 percent of Virginia’s total median family income for the last 50 years. The fact that the income gap for Black Virginians has not changed considerably since 1970 is particularly notable because the intention of the Civil Rights Era reforms and the Great Society programs that have existed since the late 1960s are in large part to help close the income gap.

Virginia Median Family Income 2

Source: Census Public Use Micro Data, in 2018 dollars. Data for Asian and Hispanic families were not published in every decade.
Read Full Article →


With the novel coronavirus causing a worldwide paradigm shift, I was genuinely tempted to jump into the data and analyze all possible permutations of what the potential impacts may be. Should I first plot the number of cases in Virginia with corresponding population densities on a map, or should I find causal connections between the age-distribution and COVID deaths in each of our 133 cities and counties? Will the impact on births be short or long term? How can we capture the symbiotic relationship between immigration rates and the labor market, amidst fears of a second wave? After the initial adrenaline rush of consuming all the data and attempting to produce a coherent narrative, I paused to consider the pitfalls of analysis paralysis. With all the chaotic commentary floating around us, I had to ask if my attempt would be adding value to the conversation, or just more noise. Read Full Article →

The census, which is conducted every ten years, provides a snapshot of the general characteristics of the U.S. population, including place of residence on April 1. This year, in the midst of the COVID-19 pandemic, obtaining an accurate census count will be especially challenging for a variety of reasons: Some people may not have been living where they typically would on April 1; nursing homes in a community may have been experiencing higher mortality rates than usual; prison populations may have been reduced; and enumerators (Census workers who go door-to-door to gather data for those who have not yet responded) may not be able to reach hard-to-count populations due to health safety concerns. As a result, certain communities will likely experience an undercount or miscount of their population in the 2020 census and may receive less federal funding for critical resources, services, and programs they need.

The information obtained from the census is vital to our governing processes and provides essential information for businesses, governments, and researchers.  It is the foundation for all data collections.  On the government side, census data informs how congressional districts are redrawn and determines where government funds go. On the private sector side, businesses use the data to decide where they should set up shop. Institutions big and small use census-derived data to decide how to set goals, distribute resources, and evaluate outcomes. An accurate census is critical for all.

Image 1

Source: WBUR

Read Full Article →

As a result of COVID-19, higher education institutions across the U.S. had to quickly transition from in-person to online courses for the Spring 2020 semester. In Virginia, the number of students completing coursework entirely online jumped from 120,000 to over half a million in a matter of weeks. Colleges and universities are now preparing scenarios that will shape not only the next few months, but also the next several years. Their decisions and preparations—previous and future—will likely determine their survival in a setting of already decreasing enrollments.

Some institutions have a head start in online instruction. In Virginia, just over half of post-secondary institutions offered at least one online education course and nearly 40 percent of students enrolled in at least one online course, which is higher than the national figure of 35 percent. Online learning in Virginia has been on a steady rise over the past 15 years, but future trends will depend on how institutions adapt to a more virtual environment.

Distance learning trends for Virginia

The number of higher education institutions in Virginia that offer online education opportunities has gradually increased since the data was first reported in 2004. However, the data shows that in recent years that number has stalled at around 90 out of 160 Virginia institutions[1]. Nearly all growth has come from private institutions as public institutions have consistently offered online courses since 2014.

Figure 1: Number of Virginia Higher Education Institutions Offering Online Learning Opportunities, 2004-2018

Online_Figure 1

Source: U.S. Department of Education’s National Center for Education Statistics (NCES), Integrated Postsecondary Education Data System (IPEDS) 

Read Full Article →

Lower mortgage rates and a rise in concentrated wealth nationwide has resulted in more people buying second homes. With a multitude of natural, historical, and cultural amenities, Virginia is an attractive location to spend leisure time and a prime location to purchase a vacation home. Within the Commonwealth, the percentage of homes in an area that are vacation units, or the vacation share of housing, varies by the region. However, prices in areas with a high percentage of vacation homes are increasing faster than the median price of homes in Virginia and the U.S. overall.  A high or increasing vacation share of housing can indicate the area is an attractive destination, which can boost the local economy, but it also may bring unintended consequences, such as an increase in overall housing costs.

Home Vacation rental

Vacation Home Definition
The majority of home buyers purchase a property to use as a primary residence, but some buyers also purchase a vacation home for family use, for equity gain, for rental income or for use as a primary residence during retirement. Housing units occupied by persons with usual residence elsewhere are classified by the Census Bureau as “vacant for seasonal, recreational, or occasional use.” Thus, we define these seasonally vacant homes as vacation homes.
These are vacant units used or intended for use only in certain seasons or for weekends or other occasional use throughout the year. Seasonal units include those used for summer or winter sports or recreation, such as beach cottages and hunting cabins. Interval ownership units, sometimes called shared-ownership or timesharing condominiums, also are included here.

Read Full Article →

The 2020 Census is currently underway, and its accuracy relies on two equally important processes: complete data collection, and accurate data reporting. Complete data collection (in the words of the Census Bureau, “counting everyone once, only once, and in the right place”) is essential to an effective decennial census, which is estimated to cost $15.6 billion and which guides the allocation of more than $675 billion to states, communities, programs, and organizations.

Please note “guides the allocation” in the second half of the previous sentence, as the complete data collection is only part one of an accurate Census. Part two, accurate data reporting, drives the allocation of those funds (often awarding a certain amount of funds per person counted in each place), and also informs planning and service provision to citizens across the country. It tells local leaders how many school children to expect for the next year; rescue squads how many ambulances to have on hand; businesses how many working-age people might be available to fill their jobs. 

Alarmingly, even if data collection in this Census is complete and perfect, the data released will be far from accurate due to the implementation of a new approach to data privacy named by the Census Bureau as “Differential Privacy Disclosure Avoidance System” (DP).


Differential Privacy is a new mathematical procedure in which all data below the state level (anything pertaining to counties, cities, or towns) will be infused with “noise” in pursuit of the goal of greater privacy protection. Sounds good, until it becomes clear that privacy protection comes at the great cost of data accuracy and utility.

On October 30, 2019, the Census Bureau posted 2010 data altered with the proposed DP procedures for 2020.  This was designed to help data users better understand the 2020 Census disclosure avoidance system and to evaluate its impact on data quality. Analyzing the differences between the 2010 count and the 2010 noise-infused data (referred to as DP onwards) for the case of Virginia highlights several issues.

[Download Handout here: How Differential Privacy Harms Census Data in Virginia]

Read Full Article →

This morning the Weldon Cooper Center at the University of Virginia released its 2019 population estimates for Virginia’s counties and cities. The estimates show Virginia’s population inched up to 8,535,519 in 2019, after passing 8.5 million in 2018. Though Virginia has added over half a million new residents since the last census in 2010, population growth has slowed significantly across Virginia in recent years, falling below U.S. growth levels to reach the lowest population growth rate since the 1920s.Average Annual Population Growth chart Virginia

Similar to the 1920s, Virginia’s population is growing more slowly because for the past six years, the number of Virginians moving out of the commonwealth has been greater than the number of people moving to Virginia from other states. During the 1920s, the increasing mechanization of agriculture caused thousands of Virginians to move out of the countryside, often to Northeastern cities to find work. Today, migration patterns have reversed, with Virginia attracting thousands of new residents every year from the Northeast but losing an even larger number of Virginians who have moved south to fast growing metro areas and popular retirement communities. In 2018, the most recent year IRS migration data is available, nearly 15,000 more Virginians moved to either Florida, the Carolinas or Texas than residents of those states moved to Virginia. Read Full Article →

There is an axiom in the technology sector, known as Amara’s law, which states that the short term impact of a new technology will typically be overestimated while the long term impact is usually underestimated. This is particularly true when considering the impact the internet was expected to have on where people choose to live. During the 1990s dot-com boom, increasing access to the internet was expected to make geography matter less—with goods, services and even work becoming accessible through the internet, the differences between regions, countries, cities and rural areas were expected to shrink. Thomas Friedman’s bestseller The World is Flat popularized this idea even further. But the acceleration in job and population growth in many major U.S. cities over the past two decades has caused a number of demographers and economists to question the initial assumption that the internet would be a “great leveller”. In fact, in a digital economy, the growth in jobs and population has appeared to be more concentrated than ever in a few large metro areas.Job Growth BLS

Source: Oregon Office of Economic Analysis, BLS data

However, Census Bureau data released in December shows some signs that the internet may be entering the second stage of Amara’s law where the places Americans chose to live are becoming less connected to where their employer is based. In the last three years that Census data is available (2015-2018), the number of Americans who primarily telecommute (working from home) rose by over 1.6 million, after increasing by less than 1.5 million between 2000 and 2010. If Virginians who primarily worked at home were grouped together as an industry, it would easily be Virginia’s fastest growing industry, increasing by 43 percent since 2010. As of 2018, about 6 percent, or nearly a quarter of a million Virginians, worked from home, a little less than the share of Virginians who worked in manufacturing (7 percent). Read Full Article →

It is important that kids from economically-disadvantaged families have access to good schools so that they will have a better outlook for employment, income, and overall wellbeing than their parents’ generation. My first post on this topic provided an introduction to the interrelation between an area’s school proficiency and its cost of housing. By giving an overview of both school quality and housing cost across Virginia, the post provided context in relation to the quality of schools children in lower-income housing are attending. One way that governments provide opportunity for low-income households is by offering them rental subsidies to live in areas where job centers,  amenities, and higher-quality schools are located. This second post will focus on (1) where subsidized housing is located in Virginia and (2) how this relates to school quality.

 Affordable housing is an ambiguous term that means different things to different people. In this study, it focuses on federally subsidized or assisted rental housing that provides lower-income households the financial ability to have affordable homes. Federally subsidized or assisted housing comprises the majority of subsidized units in service today, including those under the Low-Income Housing Tax Credit (LIHTC), Section 8, Public Housing, HOME, HUD Insured, USDA Rural Rental Housing, and Section 202 programs. Data for this analysis are drawn from the National Housing Preservation Database (NHPD).

Figure 1: Federally subsidized housing locations in Virginia, 2019

Sub Housing Figure1

Source:The National Housing Preservation Database (NHPD)
U.S. Department of Housing and Urban Development, Data Source: Great Schools (proficiency data, 2013-14); Common Core of Data (4th grade school addresses and enrollment, 2013-14); Maponics (attendance boundaries, 2016)

Read Full Article →


Political punditry and polls often form the basis for predicting election results, but rarely are they consistent or comprehensive. Given the importance of state-level voting, and the central role played by the Electoral College in the democratic process, projecting the number of eligible voters in each state may add value to understanding the November 2020 election cycle. For the current analysis, eligible voters are defined as U.S. citizens ages 18 and over, including those born in the U.S. as well as those born in other countries who have immigrated and become naturalized citizens.1

The sheer population size of the individual states is a good starting indicator of how large a pool of potential voters may reside in them. If the elections were held today, California would have over 25.8 million voters, while Wyoming would have fewer than 500,000 people eligible to vote. Texas, Florida, and New York would each be home to over 10 million eligible voters, closely followed by Pennsylvania and Illinois. Read Full Article →