How to check an economic claim against the original source

Turn rhetoric into a testable sentence

Claims such as inflation exploded, debt doubled, or the economy shrank leave essential choices unstated. Before searching, rewrite the statement with a named measure, geography or population, starting period, ending period, units, and comparison. If the speaker did not provide those details, mark them unknown instead of silently filling them in.

This step separates factual verification from interpretation. A table may confirm that a defined index changed by a certain percentage while leaving words such as good, unaffordable, sustainable, or unprecedented unresolved. Those judgments require an explicit benchmark and often additional evidence. The narrow factual claim should be checked first.

• Replace “inflation” with the exact index if known: CPI-U, PCE price index, or another measure.
• Replace “debt” with gross debt, debt held by the public, or the source’s precise label.
• Replace “now” with a release period and publication date.

Sources: [bls-faq] [treasury-debt]

Find the producer, not just a republisher

Search for the agency that produces the statistic. For consumer prices, that may be BLS; for PCE and GDP, BEA; for recorded Treasury debt, Treasury Fiscal Data. A news article or social post can point you toward a release, but it should not be the final evidence when the public primary source is available.

On the agency page, distinguish a learning article from the data release and a release from a downloadable table. Learning pages explain definitions; releases provide narrative and selected tables; datasets provide the records needed for reproduction. Use the layer that supports the claim, and link definitions as well when a term is likely to be misunderstood.

• Confirm the domain and agency name rather than trusting a screenshot’s visual style.
• Prefer the release, table, API documentation, or dataset maintained by the producing agency.
• Keep a stable URL and access date with your notes.

Sources: [bls-faq] [bea-quality] [treasury-debt]

Run the label checklist

Trace the number to a row and column. Record the full measure, units, seasonal-adjustment status, population or scope, geography, and period. Check whether the value is a level, percent change, percentage-point change, annual rate, or share of another measure. Similar-looking values can answer very different questions.

Then read nearby footnotes and methodology. BLS explains that CPI includes many indexes and both adjusted and unadjusted presentations. BEA explains that its estimates are updated as more complete source data arrive. Treasury distinguishes debt categories and publishes datasets at specific frequencies. These details are not escape clauses; they define what the data mean.

• Measure and scope: exactly what people, transactions, or obligations are included?
• Time and treatment: monthly, quarterly, fiscal year, twelve-month, seasonally adjusted, or annualized?
• Units and status: dollars or percent, nominal or inflation-adjusted, observed or projected, preliminary or revised?

Sources: [bls-faq] [bea-quality] [treasury-debt]

Check the vintage and reproduce the comparison

Economic statistics can be revised. BEA states that it routinely updates estimates as more complete source data become available and labels successive GDP estimates. A later table may therefore differ from the value available when an older article was written. Preserve the release date or vintage before accusing either source of a mistake.

Recalculate only after matching definitions. Use the agency’s prescribed formula and unrounded data when available. If your result differs, check date alignment, units, seasonal adjustment, annualization, and revisions before concluding the source is wrong. Document the calculation so another reader can repeat it without guessing.

• Save the publication date and note whether the table says preliminary, revised, advance, second, or third estimate.
• Compare like with like: same series, same period convention, same units, and same vintage.
• Report small rounding differences as rounding differences, not evidence of manipulation.

Sources: [bea-quality] [bls-faq]

Worked check: did output grow, or did prices rise?

Suppose a headline says the economy grew 10 percent. Before accepting that as an output claim, check whether the underlying series is nominal or real GDP. BEA defines GDP as the value of final goods and services produced in the United States. Nominal, or current-dollar, GDP uses market prices in the measured period. Real GDP removes price-change effects to help compare output over time. A rise in nominal value alone does not establish how much real production grew.

Consider an illustrative economy producing one final product. It makes 100 units at $10 each in year one, for $1,000. In year two it makes the same 100 units at $11 each, for $1,100. Nominal output rose 10 percent, while quantity was unchanged. Valuing both years at the original $10 price gives $1,000 each year and zero real growth in this simplified example. The verdict is not that the nominal number is false: it is that the number does not support a claim of 10 percent more output.

These invented teaching numbers are not U.S. observations or a current inflation estimate. BEA uses chain-type measures for a many-product economy; holding one price fixed here only illustrates the distinction. For an actual national-output claim, use BEA's published real measure rather than treating this example as a method for rebuilding the national accounts.

Sources: [bea-gdp] [bea-real]

Reproduce the arithmetic without mixing measures

Now change year two in the same illustrative economy to 105 units at $11 each. Nominal output is $1,155, a 15.5 percent increase from $1,000. At the original $10 price, output is $1,050, a 5 percent increase. Prices rose 10 percent and quantity rose 5 percent, but nominal growth is 15.5 percent rather than exactly 15 percent: the higher price also applies to the additional units.

The multiplication is 1.10 × 1.05 = 1.155. Reversing this simplified example gives (1.155 ÷ 1.10) − 1 = 0.05, or 5 percent real growth. Subtracting rounded growth rates can provide a rough intuition, not an exact calculation. Do not divide nominal GDP by a consumer-price index and assume the result reproduces BEA's real GDP series. The source's own real estimates are the appropriate starting point for checking a claim about output growth.

• Illustrative nominal growth: (($1,155 ÷ $1,000) − 1) × 100 = 15.5 percent.
• Illustrative quantity growth: ((105 ÷ 100) − 1) × 100 = 5 percent.
• Label every example as illustrative and preserve the exact series definition for real-world calculations.

Sources: [bea-gdp] [bea-real]

Finish the GDP check with period and vintage

Choosing real GDP does not finish the verification. BEA explains that quarterly GDP data are reported at annual rates unless otherwise specified, and that seasonal adjustment removes recurring calendar patterns. A quarterly dollar level expressed at an annual rate is not simply the dollars produced during that quarter. Likewise, an annualized quarterly growth rate is not the same comparison as a change from the same quarter a year earlier.

BEA's advance quarterly estimate uses the best information available at that time; the second and third estimates incorporate additional source data. Record which release the headline used before comparing it with a newer table. Your final testable sentence should name real or nominal GDP, level or growth rate, the exact comparison periods, seasonal adjustment, annualization where applicable, and the release vintage. If the headline omits these labels, report that gap instead of supplying whichever interpretation makes the claim look strongest.

Sources: [bea-gdp]

Write a verdict with calibrated confidence

A useful verdict is more specific than true or false. The source may support the number but not the causal explanation; support the direction but not the magnitude; or support the statement only under a particular measure. Quote as little as necessary, link the exact source, and explain the gap in plain language.

If evidence is missing, say what would resolve the question. A chart without a series identifier may be unverifiable even if its shape looks plausible. A projected value should not be narrated as an observed result. Honest uncertainty is an outcome of verification, not a failure of it.

• Supported: the exact primary-source measure matches the claim.
• Supported with qualification: the number matches, but scope, period, or interpretation needs correction.
• Not established: the available source does not identify or support the claimed measure.

Sources: [bea-quality] [treasury-debt]