STATEMENT OF WORK revised.pdf

STATEMENT OF WORK

BUREAU OF ECONOMIC ANALYSIS

Purchase of Daily U.S. Credit Card Transaction Data

I. INTRODUCTION

The Bureau of Economic Analysis (BEA) is one of the world's leading statistical agencies.

BEA produces economic statistics that enable the government, economists, statisticians, and the public to understand the performance of the Nation's economy. BEA collects source data, conducts research and analysis, develops and implements methodologies, and disseminates statistics to the public.

In response to the pandemic BEA has used card transaction data from Palantir/Fiserv to track large changes in the economy happening in days and weeks, rather than in months and quarters. This information has been useful in a number of ways: (1) the data has been useful to help inform estimates internally within BEA to provide more accurate estimates to the public and reduce revisions; (2) over the past year, BEA has been using this data to provide timely estimates of consumer spending to the public on a weekly basis. Alternative card transaction data sources all have different properties and BEA staff have gained substantial familiarity with these data, as reflected in multiple analytical papers using these data over the past few years, including Dunn et al. (2020), Dunn et al. (2021), and Chen et al. (2019).

Our weekly spending estimates based on card transaction data are analyzed by a wide variety of data users, including Haver Analytics, Citadel, UBS, Bloomberg, the Council of Economic Advisors, and many others. In fact, Haver Analytics currently disseminates this data through their data portal tool for their users to access. It is important to continue to provide highly valuable timely data that is viewed and used widely by BEA customers.

II. OBJECTIVE

The objective of this data acquisition is to procure card transaction data that provide timely information about spending at establishments across the U.S. aggregated to different levels of geographic detail including national and at the state-level and including expenditure across various retail and service industries. The series should be compatible and consistent with the existing series that is currently used to produce timely estimates for the public. It is also important to maintain consistency for internal analysis of the data that may be compared with previous time periods. The estimates should be constructed in such a way that they better reflect actual economic activity and not the sales fluctuations caused by changes in the business of the data sellers (e.g., addition or removal of a new platform or other factors related to the business practices of the data seller that cause substantial changes in sample size). One way to attain estimates that reflect actual economic activity is to use only those firms or consumers that stay in the data set throughout the period of the estimate, such as the method applied in Aladangady et al. (2019).

Specifically, BEA seeks expenditure data that:

• Provide information on geography at the state-level or finer level of geographic detail (CBSA, MSA, or County)

• Provide transactions or expenditure data with detailed industry (at least 3-digit NAICS code) classification

• Offer frequency at a daily level

• Weight and benchmark estimates relative to Economic Census

• Provide coverage statistics, relative to the Economic Census

III. DELIVERABLE/OUTCOME

The Contractor will:

• Provide access to data through direct weekly or daily transfer

• Update transaction data daily or weekly with a limited data lag (1-5 days)

• Provide data time coverage back until 2014 or earlier

• Provide estimates in dollars

• Provide data with the following specifications:

o Reports: Levels of sales to a fixed panel of firms at a daily frequency across many expenditure categories at the NAICS three-digit level, including retail and services, and across the nation for all states and Washington DC.

o Data Update Frequency: Data should track sales expenditures at a daily frequency, and data should be updated on a daily or weekly basis.

o Geography: National, State (if available, may provide CBSA, City, or County).

o Metrics: Sales/Spending (required), Transactions (optional), Average/Median

Transaction Size (optional), consumer demographics (e.g., gender and age;

optional).

o Data Delivery: via daily accessible file in comma separated file (.csv), Stata, or other accessible data format.

The data merchant is not required to reveal source data but should provide a general description of the type of transaction data (e.g., regional bank cards, credit cards with reward points, credit and debit cards, gift cards). Additional information about changes in sample should be made available when the associated changes may affect estimates (e.g., addition or removal of a new platform or other factors causing substantial changes in sample size).

XIII. PERIOD OF PERFORMANCE

The Government anticipates a base year plus four (4) optional years as follows:

Base Period [January 1, 2022 – December 31, 2022] Option Year 1 [January 1, 2023 – December 31, 2023] Option Year 2 [January 1, 2024 – December 31, 2024] Option Year 3 [January 1, 2025 – December 31, 2025] Option Year 4 [January 1, 2026 – December 31, 2026]

The Contracting Officer Representative for this call order will be ***.