Alternative Data and Monetary Policy

Source details

Nontraditional data provide timelier signals than traditional government statistics, often available in a few weeks or even days versus weeks or months for official releases.
Example: In late March 2020, internal weekly Fed employment estimates based on ADP payroll processing showed large declines more than a month before the BLS monthly employment report.
Weekly ADP estimates can offer early insight when monthly BLS jobs estimates shift sharply, as occurred in 2025.
Nontraditional data are useful during interruptions to official data, for instance, government shutdowns from congressional impasses.

High-frequency, granular private-sector data enable more direct testing of policy-relevant hypotheses (for example, tariff pass-through to prices).
Alberto Cavallo and collaborators constructed daily price indices from online data of five major US retailers covering 350,000 goods, including country of origin, tariff rates, and selling price.
- Their findings: prices of imported consumer goods rose more quickly than domestic goods relative to pre-tariff trends; effects more pronounced for domestic goods competing with tariffed imports; overall effects relatively modest.
Granular mobility and administrative health data were deployed to monitor shifts in consumer and business behavior during the COVID-19 pandemic.
Real-time measures of supply-chain stress included purchasing manager surveys, shipping price indices, and shipping container movement data.

Traditional government statistics (BLS, Department of Commerce) are survey-based and considered the gold standard but face rising costs and declining participation, reducing precision.
Private data can maintain or improve quality and cost-effectiveness of traditional statistics; the BLS now uses private company data for components of the consumer price index including prices of used cars, airline tickets, and wireless telephone contracts.
Benchmarking and reweighting are required to make private transaction data representative of the broader economy (example: reweighting First Data transactions using the five-year Economic Census).
Careful testing by statistical agencies is necessary to ensure that nontraditional sources improve precision rather than introduce new noise.

Alternative data can improve accuracy of initial real-time estimates and reduce errors from model adjustments:
- The payroll employment estimate relies on a “birth-death model” to account for business entry and exit; pandemic-related shifts in net business formation caused significant errors and revisions.
- Weekly tax filings for employer identification numbers provide a reliable forecast of business formation in subsequent quarters and could improve the birth-death adjustment.
Combining multiple independent estimates can improve understanding of official series; the Federal Reserve Bank of Chicago is blending official and alternative labor market data (Indeed, Lightcast, Google unemployment searches) to estimate the current month’s unemployment rate, though the project is in early stages.

Nontraditional data enable research on distributional effects of monetary policy using household-level credit records, bank accounts, and administrative records.
- CoreLogic property tax forms and deed records showed Black, Hispanic, and lower-income borrowers were less likely to refinance during falling interest rates than Asian, White, and higher-income borrowers; systematic differences in refinancing costs contributed.
- IRS individual income tax records analysis found unexpected tightening increased income inequality mainly by worsening outcomes for low earners, while unexpected easing decreased inequality.
Alternative data can help evaluate policy transmission in near real time after monetary adjustments.

Key challenges of nontraditional data sources:
- Acquisition cost and reliability: private providers may stop sharing data or raise prices, threatening continuity of government statistics.
- Proprietary access limits transparency and external verification; the Fed’s reliance on nonpublic data can reduce accountability and prevent outsiders from replicating analysis—only market participants who pay have access.
- Short time series complicate seasonal adjustment and anomaly troubleshooting.
- Data may embed business-specific growth (e.g., client acquisition by a payment processor) that must be separated from true macroeconomic changes.
Solutions and mitigations:
- Benchmarking private data to official censuses or surveys (example: use of the five-year Economic Census to reweight card transactions).
- Careful agency testing before incorporation into official statistics.
- Strong ties among government statistical agencies, private data providers, government officials, and academics to improve data quality and usability.

Nontraditional data are a valuable complement to traditional statistics, offering timeliness and granularity that help detect turning points and analyze policy transmission.
They should not be viewed as substitutes for official government statistics; their usefulness often depends on and benefits from integration with traditional data.
Improving the quality and policy utility of data requires:
- Continued development of methods to benchmark and validate private data against official sources.
- Institutional arrangements to manage data access, continuity, and transparency.
- Ongoing collaboration among statistical agencies, private providers, policymakers, and researchers.

Source: Alternative Data and Monetary Policy, Claudia Sahm, F&D Magazine, December 2025.