American Community Survey

The American Community Survey (ACS) provides detailed sociodemographic and socioeconomic information about American households, provided by the US Census Bureau annually. The five-year estimates are available beginning in 2005/09 to present; we use the 2014/18 dataset for our analyses. The data are presented on multiple geography levels including block group, tract, county, and state; our research is interested in tract level information. We used the dataset to characterize the Fairfax County workforce, for example, to locate which tracts had a higher percentage of near retirement population or to create composite indicators identifying where workers are located who may be vulnerable to job loss.

LEHD Origin-Destination Employment Statistics (LODES)

Longitudinal Employer-Household Dynamics (LEHD) Origin-Destination Employment Statistics (LODES) provides data that characterizes workforce dynamics and is comprised of administrative data through the US Census Bureau. For each geography, LODES compiles the number of jobs located in the given area. LODES compiles three different datasets, ranging in time span from 2002 to 2017:

• Origin-Destination (OD) data -- jobs totals associated with home and work Census blocks

• Residence Area Characteristic (RAC) data -- jobs are totaled by home census block

• Workplace Area Characteristic (WAC) data -- jobs are totaled by work census block

LODES data includes additional characteristics of these jobs, such as job sector and worker demographics. Because LODES totals jobs by workplace and by residence areas, one can use these datasets to track differences in workplace dynamics, such as on aggregate whether individuals’ workplaces are close to their homes. We used this dataset to map origin tracts for people traveling from outside of Fairfax County to Fairfax County, to determine areas especially vulnerable to pandemic-related job losses using LODES occupation sector information, and to map job locations relative to Fortune 500 company headquarters locations.

2019 Forbes 1000 List

The 2019 Fortune 1000 contains addresses of the top 1000 largest corporations in the United States, as supplied by Fortune Magazine. This dataset contains the city and address of these companies, which we used to geocode their locations and construct a map of Fairfax sites. Using this information, we overlaid company headquarters locations above a choropleth map of employment data to show how the presence of these company headquarters might influence the local economy and serve as magnets for Fairfax County jobs.

Virginia Employers Data

Virginia Employers data from the Virginia Employment Commission contains a list of over 38,000 employers in the state of Virginia along with business descriptions, average salaries, number of employees and the latitude and longitude for each entry in the dataset. We used this data to map the top hundred largest employers in Fairfax County in our exploration of emerging areas.

Virginia Employment Commission Unemployment Insurance Claims

The Virginia Employment Commission’s Economic Information Services Division supplies a dataset that has provided the number of unemployment insurance (UI) claims by ZIP code since 2003. We used the 2020 dataset to determine the total number of initial unemployment claims by ZIP code within Fairfax County and compared these estimates to the total numbers of workers in occupations vulnerable to covid-related job loss to examine the pandemic’s potential impact on Fairfax County workers.

Exploratory Methods

Network Analysis

We employed network analysis to understand the relationships between where Fairfax County residents work and reside at the census tract level using LODES data. The LODES data is produced by the US Census Bureau in order to better understand where workers reside and work. There are three main datasets of the LODES data; we use the Origin-Destination (OD) 2010-2017 files for our networks. The OD datasets indicate where individuals work and reside by census block group. We retrieved the main and auxiliary OD files and aggregated by tract. The main files include jobs for which the residence and workplace are both in-state; the auxiliary files include jobs for which the workplace is in-state while the residence is out-of-state. We combined both the main and auxiliary files, and aggregated the data to census tract level.

While the LODES data provides OD flows for all workers that live or work in Fairfax County, we decided to reduce the network’s data for the DSPG project to only include those that live in Fairfax County. Before constructing networks, we explored the origins of workers who come to work in Fairfax County. We found that of all workers in the county, 41% are also county residents, and 59% arrive from outside the county. This analysis also revealed that some workers who come to work in Fairfax County live in states well outside of driving distance. This created a challenge, as it is difficult to determine whether workers work for businesses with headquarters in Fairfax but work physically located elsewhere, or whether they do work in Fairfax but perhaps only part of the time. For this reason, we decided to only focus our current analysis on flows from within Fairfax County.

Networks are comprised of nodes and edges that represent entities and the relationships between them. In our case, nodes correspond to census tracts, and edges represent job flows or the links between them, created by workers’ residence and workplace connections. Each edge or connection also has a weight, which corresponds to the number of jobs flowing between two tracts, and a direction, which signifies the direction that workers flow in, from their residence tract to their workplace tract. Network centrality measures help explain how relations between tracts change based on the relative importance of those tracts within the overall county network. Using this network analysis framework, we were able to construct three distinct networks and map the networked flows of all jobs, federal jobs, private jobs, and jobs by industry within Fairfax County.

We first constructed an edgelist to examine how workers living within Fairfax County travel from tracts where they live to tracts where they work. We produced network-level descriptive statistics to understand the basic structural components of the network graph as a whole. We ran an extensive set of descriptive statistic measures using the igraph package for R. The key descriptive statistics produced include the number of nodes, edges, diameter, distance, density, transitivity, and the average degree and betweenness centrality. While these global descriptives help us better understand changes at the County level, we also examined node-level centrality measures to identify the most influential areas within Fairfax County. We used leaflet maps to visualize the four measures of centrality defined below: