BEA GDP Methodology: How the US Measures Output

The US Bureau of Economic Analysis (BEA) publishes the official US GDP series in the National Income and Product Accounts (NIPA). Quarterly releases use chain-weighted Fisher indexes for real GDP and a documented revision schedule. Here is exactly how the figures are built.

Source: BEA NIPA Concepts and Methods Manual, BEA Methodology Papers

The NIPA Framework

The National Income and Product Accounts (NIPA) are an integrated set of accounts the BEA maintains to measure the US economy. They are organised around the basic identity that aggregate spending equals aggregate income equals total production. NIPA presents GDP through the expenditure approach (the headline figure), through the income approach (called GDI), and via the production-side industry accounts (the satellite Industry Accounts).

Each NIPA table has a number. Headline GDP appears in Table 1.1.5 (current dollars, nominal GDP) and Table 1.1.6 (chained 2017 dollars, real GDP). The standard growth-rate quote uses Table 1.1.1 (percent change from preceding period in real GDP). NIPA Table 1.10 carries the income-side GDI breakdown.

The full set of NIPA tables exceeds 200 and is published at bea.gov. Methodology is documented in the BEA NIPA Concepts and Methods Manual (most recently revised 2025).

Chain-Weighted Real GDP

Since 1996 the BEA has used chain-weighted Fisher indexes to construct real GDP. The motivation: fixed-weight measures (such as the constant-1996-dollar series the BEA used before 1996) produce growing distortions over time as relative prices change. When personal computers became cheaper, consumers bought more of them; a fixed-1996-weight measure would understate the growth attributable to computing investment because it ties weights to a year when computing was a smaller share of the basket.

The chain-weighted Fisher index avoids this by recalculating weights each year. For each pair of adjacent years, the BEA computes two growth-rate measures, the Laspeyres (using the earlier year's weights) and the Paasche (using the later year's weights). The Fisher index is the geometric mean of the two. These year-over-year growth rates are then chained together back to the base year (currently 2017).

A consequence is that chained-dollar component series do not add up to chained-dollar GDP. The BEA publishes a separate residual to reconcile. This non-additivity confuses first-time users of NIPA data but is a mathematical feature of chain-weighting, not an error.

Key NIPA Tables

Source Data

The BEA does not collect primary data. It synthesises GDP from dozens of source agencies including the Census Bureau (retail sales, manufacturers' shipments, services surveys, international trade), the Bureau of Labor Statistics (employment and earnings data), the Treasury (federal receipts and outlays), the Federal Reserve (industrial production, flow of funds), the Department of Agriculture (crop and livestock production), the IRS (corporate tax returns, used in annual benchmarks), and many specialised industry surveys.

For the advance estimate, much source data is missing and the BEA uses trend extrapolation, judgmental adjustments, and indicator series to fill the gaps. For the third estimate, most source data is in. For annual updates, the BEA incorporates the comprehensive Census Bureau Annual Surveys (Retail Trade, Wholesale Trade, Services, Manufactures), the Treasury monthly statements, and the IRS Statistics of Income corporate data.

Comprehensive revisions every five years incorporate the once-a-decade Economic Census plus updated source data going back further. The most recent comprehensive revision was in September 2024. Each comprehensive revision typically nudges historical estimates of total GDP up or down by a fraction of a percent and can materially change individual components.

Release Schedule and Revisions