Economic reporting problems: how to read releases

Reading an economic release starts with the label attached to its number, not with the headline built around it. A percentage point can be mistaken for a percent, slower inflation for lower prices, and an early estimate for a final count. These slips are common, but their corrections are often brief.

Why are percent and percentage points easy to mix up?

A percent change compares a change with its starting value; a percentage-point change is the difference between rates. Moving from 4 percent to 5 percent is a 1 percentage-point rise, or a 25 percent increase against the starting rate, calculated by dividing the change by the base.

What does the release say this number measures?

Check labels such as annualized, real, average, and advance or preliminary before using a figure. Annualized converts a rate to a twelve-month basis, real adjusts for price changes, average divides an aggregate across its covered population, and advance or preliminary signals a first estimate that may be revised.

The Bureau of Labor Statistics publishes the monthly Employment Situation, also called the jobs report, plus the Consumer Price Index, Producer Price Index, and Job Openings and Labor Turnover Survey. The Bureau of Economic Analysis publishes gross domestic product, including its advance estimate, and the PCE price index.

The Census Bureau publishes retail sales and other business surveys, the Federal Reserve publishes industrial production, and the Labor Department provides weekly unemployment insurance claims. Real GDP and its components use chain-weighted indexes, so components do not add exactly to the total as fixed-weight figures would; presenting their contributions as a sum claims arithmetic the data do not support.

Does slower inflation mean prices are lower?

A slowing inflation rate means prices are rising more slowly, not necessarily falling. The Consumer Price Index can rise more slowly, hold flat, or fall; report the rate and the index direction separately because each describes a different movement in the release.

The same distinction matters when explaining household costs: slower inflation does not mean an individual bill has fallen. Use household context to explain how a national measure relates to household experience, not to recast the aggregate as everyone's bill.

Does average earnings tell me what one worker got?

No. An average increase is not a raise paid to every worker or household. State which employee population is covered, whether earnings are hourly or weekly, how composition effects matter, and which price deflator applies; otherwise the aggregate may be mistaken for an individual payroll record.

The BLS real earnings technical note says its constant-dollar series combine Current Employment Statistics earnings with CPI deflators. It uses different CPI series for specified employee groups, making the result a defined aggregate rather than a payroll record for each worker.

Can one release explain what made a number move?

No: movement in one CPI or jobs-report category does not by itself show that policy, weather, or sentiment caused it. Keep description separate from explanation, and reserve causal claims for evidence with additional data, timing, a mechanism, and research; see policy attribution problems.

Which reporting periods can I compare fairly?

Compare rates only when their periods and methods match, or make the conversion explicit and valid. A monthly rate, a quarterly annualized rate, and a year-over-year rate are not directly comparable as if they measured the same interval without explanation.

Annualizing a monthly rate means compounding it across twelve months, not multiplying the monthly rate by twelve. If conversion is justified, name the method and show the calculation; otherwise keep the periods apart in separate chart panels or a table that labels each interval clearly.

Does dollar growth prove shoppers bought more?

No. Higher dollar sales do not establish that more goods were purchased until price changes are adjusted for. Identify the release's adjustment and the category definition as well, since a sales total can change in value without showing the number of units sold.

The Census Bureau's Advance Monthly Sales report labels estimates as adjusted for seasonal, holiday, and trading-day effects, but not price changes. Figures on that page change monthly; the lasting point is to distinguish those adjustments from a price adjustment before describing sales as more goods purchased.

Read the category definition, too. Retail-establishment data group sales by the establishment's primary business, which may differ from the product a reader imagines. A store category can include many types of goods, while one product may pass through several categories; do not treat the classification as a product count without supporting data.

What should I do when a release revises its numbers?

Treat revisions as part of the result, not as a footnote when they change the recent pattern. Jobs reports and GDP estimates begin with an early figure, then change as more complete data arrive; put a large revision in the opening if it changes the pattern.

Keep an older vintage only when showing how estimates evolved. Compare the current release with earlier release vintages, since an advance estimate can be replaced by a larger sample and a benchmark can rewrite a longer history. A chart that joins unlike vintages without a label can make a false change.

Health desks use status labels to distinguish stages of evidence; health alert labels offer a useful comparison. For economic series, identify the estimate's vintage and say what the revision changes about the recent pattern, rather than silently carrying an old number forward as though it were current.

Is a forecast miss enough to call a report good or bad?

Not by itself. Calling a jobs report or inflation reading good or bad only because it beat or missed consensus leaves out the measure's level, trend, revision history, uncertainty, and relevance to households or businesses; show the forecast as one comparison among several.

How do I keep a chart tied to the release it describes?

Record where the data came from and preserve the numbers behind the conclusion. A chart that refreshes itself while its text describes an older release combines two versions of the story in one file, so keep an archive of the data used when the conclusion was written.

The St. Louis Fed's ALFRED preserves vintages for many series, and the Bureau of Labor Statistics and Census Bureau offer their own retrieval tools. Record the release the figures came from, not only the retrieval date; include the series ID, units, adjustment, and archive location so later readers can identify the exact data.

A hypothetical budget placed beside government data can look official unless it is clearly marked as an example. Label it, publish the inputs and arithmetic, and state plainly that it is not an estimate of a typical household. The figures should not be presented as though they came from the release.

Can seasonally adjusted and unadjusted figures share a story?

Yes, if you give a reason and label each series clearly. Do not subtract one from the other or compare them as though they were one series: adjusted and unadjusted figures answer different questions about the same month and must remain distinct in the copy.

Does a revision automatically mean the agency erred?

No. Many statistical programs publish an early estimate and later add more complete data, so revision is part of the design and is not, by itself, evidence of an error. Report the revision's size and direction and explain how it changes the recent pattern.

What should I do first?

Begin by checking the release's label, comparison period, adjustment, and vintage before writing the takeaway. Then decide whether the sentence describes a movement, compares it with a forecast, or explains a cause; the claims require different evidence and should not be collapsed.