Facebook Business Activity Trends during Crisis

Step 1: Count public posts. We analyze the posting activity by administrators of Facebook business pages before and after a crisis event. All data shared with partners is de-identified and aggregated at administrative levels.

Step 2: Compare activity to pre-crisis levels. We compare post-crisis posting activity to a baseline distribution for that page in the year prior to the crisis.

Step 3: Produce outputs. We aggregate these comparisons daily for each US county or global equivalent, grouping business pages by economic sector. The final output is an average across pages, similar to a percent change.

Full details here: https://dataforgood.facebook.com/dfg/resources/business-activity-trends-methodology-paper

Business verticals: We derived the business verticals by aggregating categories as defined by the admins on the business pages.

All: Refers to all businesses in the polygon. This includes but is not limited to all the following categories. This includes all of the following categories except public good, because the activity of public good pages tends to differ from other businesses during crises.

Grocery and convenience stores: Retailers that sell everyday consumable goods including food (typically unprepared foods and ingredients) and a limited range of household goods (like toilet paper). These can include grocery stores, convenience stores, pharmacies and general stores.

Retail: Retail other than grocery and convenience stores such as auto dealers, home goods stores, personal goods stores and general merchandise/big-box stores like Walmart.

Restaurants: Businesses that sell prepared food and beverages for on-premise or off-premise dining.

Local events: Events, activities and businesses that sell real-life experiences, such as amusement parks, bowling alleys, concert venues and social clubs.

Professional services: Services driven by demand from an individual event such as a legal need or health issue that require high customization. Providers usually have an advanced degree or certification and are considered experts and “knowledge workers.” Examples include CPAs, lawyers, medical professionals, architects.

Business and utility services: Business services offering business-to-business services like construction, office cleaning, advertising and marketing companies and business software solutions. Utility services offer commodity services like electric, phone, internet, water and energy.

Home services: Services driven by demand from an individual event at home such as plumbing or electrical work. Examples include home repairs, photographers, cleaning, mechanics, plumbers, electricians, landscapers, interior decorators.

Lifestyle services: Specific to beauty, care and fitness services. These businesses offer standardized services that are part of a customer's regular routines. Examples include gyms, salons, barbers, and nonmedical and noneducational supervision, like childcare nurseries and pet care.

Travel: Businesses that provide or sell transportation or accommodation services, such as airlines, hotels, car rentals and tour operators.

Manufacturing: Businesses that manufacture durable goods (like furniture and cars) or consumable goods (like food and personal goods) and have no or limited business-to-customer sales.

Public good: Includes government agencies, nonprofits and religious organizations.

Codebook: Polygon ID (polygon_id): Unique identifier from theDatabase of Global Administrative Areas (GADM) for the polygon representing the region or administrative area.

Polygon name (polygon_name): Name of the polygon based on the Database of Global Administrative Areas (GADM).

Business vertical (business_vertical): The business category of the aggregation. Business verticals are defined internally within Facebook by categories selected by the page admins. We use business verticals as a proxy for local economic sectors. Included as a business vertical is the category ‘All’, which includes all other business verticals except ‘Public Good’.

Activity quantile (activity_quantile): The level of activity as a quantile relative to the baseline period. This is equivalent to the 7-day average of what University of Bristol researchers call the aggregated probability integral transform metric (see this article in Nature Communications). It’s calculated by first computing the approximate quantiles (the midquantiles in the article) of each page’s daily activity relative to their baseline activity. The quantiles are summed and the sum is then shifted, rescaled and variance-adjusted to follow a standard normal distribution. The adjusted sum is then probability transformed through a standard normal cumulative distribution function to get a value between 0 and 1. We then average this value over the last 7 days to smooth out daily fluctuations. We give this metric a quantile interpretation since it compares the daily activity to the distribution of daily activity within the baseline period, where a value around 0.5 is considered normal activity. This is a one-vote-per-page metric that gives equal weight to all businesses and is not heavily influenced by businesses that post a lot.

Date (ds): The date of the activity provided in YYYY-MM-DD format defined in Pacific Time (PT).

Country (country): 3-character (ISO alpha-3) country code.

Business Activity Trends during crisis measures how local businesses are affected by and recover from crisis events using data about posting activity on Facebook. Given the broad presence of small businesses on the Facebook platform, this dataset aims to provide timely estimates of global business activity without the common limitations of traditional data collection methods, such as scale, speed and nonstandardization. This method for understanding local economic activity, first described by the University of Bristol team and published in Nature Communications (https://www.nature.com/articles/s41467-020-15405-7), is intended to provide humanitarian organizations, researchers and policymakers with a broad view of subnational economic activity after disasters.

For rows with a polygon_id from USA, here is a guide for the states: 'USA.1' (Alabama); 'USA.10' (Florida); 'USA.11' (Georgia); 'USA.18' (Kentucky); 'USA.19' (Louisiana); 'USA.25' (Mississippi); 'USA.34' (North Carolina); 'USA.41' (South Carolina); 'USA.43' (Tennessee); ‘USA.44’ (Texas); 'USA.47' (Virginia); 'USA.49' (West Virginia)