Full section mapJump to a section
A report released this morning may describe last month—or last quarter. Start with the date and the measure. The market reaction can wait.
The free morning news desk links to official releases. This guide helps you read them. It does not predict a stock’s next move or tell you what to buy.
First, identify the report
| Report | What to check | Original source |
|---|---|---|
| CPI | Consumer-price changes; the comparison period and categories included | Bureau of Labor Statistics |
| PCE inflation | Consumer prices in the Personal Income and Outlays report; monthly or yearly change | Bureau of Economic Analysis |
| GDP | Economic output; real or current-dollar figures, growth period and estimate revision | Bureau of Economic Analysis |
| Employment Situation | Payroll employment and the unemployment rate; these come from different surveys | Bureau of Labor Statistics |
| Jobless claims | Initial or continuing unemployment-insurance claims; week covered and revisions | Department of Labor |
| Retail sales | Reported sales, categories and revisions; dollar changes can reflect prices as well as activity | Census Bureau |
CPI and PCE: two inflation measures
CPI tracks price changes for a basket of consumer goods and services. PCE also measures consumer prices, with different coverage and methods. They are related measures, not interchangeable readings.
The PCE price index appears in the BEA’s Personal Income and Outlays release. You may see both overall inflation and a measure excluding food and energy. Record which measure the headline uses; excluding categories does not mean those expenses disappear from a household’s budget.
GDP: read the units before the percentage
GDP measures the value of final goods and services produced in the United States. “Real” GDP adjusts for price changes; current-dollar GDP does not make that adjustment. Quarterly growth is often reported at an annual rate. That is different from growth over the past twelve months.
Check whether you are reading an advance, second or third estimate. A revised estimate can change the story about an earlier quarter. Keep the estimate date alongside the quarter in your notebook.
Jobs reports and jobless claims answer different questions
A payroll count is not the same as the unemployment rate. The Employment Situation includes results from employer and household surveys. Weekly unemployment-insurance claims are another source: initial claims concern new applications, while continuing claims concern ongoing benefit claims.
Check the week or month covered, seasonal adjustment and any revised earlier figures. A weekly claims headline is not a new reading of the monthly unemployment rate.
A fictional headline exercise
Imagine a headline says, “Consumer prices rose 0.2% in June and 3.1% over the year.” Those invented figures describe two comparisons. The first compares June with May; the second compares June with the same month a year earlier. Neither means every item rose by the same amount.
A lower yearly inflation rate usually describes prices rising more slowly. It does not, by itself, mean the overall price level fell. To check that, read the change in the actual price index and its comparison period.
The five-minute reading routine
- Open the original agency release, not just a screenshot.
- Write the report name, release date and period measured.
- Label the comparison: monthly, yearly or annualized; adjusted or unadjusted.
- Check revisions and the parts of the report that drove the change.
- Separate the reported fact from your interpretation and the questions still open.
Use the printable company research checklist when connecting the wider economy to a business. Revenue, debt, customers and costs matter too.
A useful report is not a trading signal
Economic releases can help you understand the environment a business operates in. A headline alone cannot establish whether a particular stock is attractively priced or how it will trade next. Check the source, keep your assumptions visible and leave room for uncertainty.
Your economic release-note worksheet
Use this six-part record beside a jobs, inflation or GDP release. It gives you a finished note with a source, a reproducible comparison and a next question. Download the blank text template, fill it in your own notes app or print it. No account is needed.
Download the blank release-note template (.txt) · See a completed fictional jobs note
- Identify the document. Record the agency, report title, original URL, publication date and time zone, period measured and date you checked it. The report date and the period covered are different fields.
- Name the measure. Copy the exact series or table-row label. Is it a level or a change? Record units, comparison window and seasonal-adjustment status. For employment, distinguish jobs, people, openings and benefit claims.
- Copy the reported results. Keep the newest period and its comparison value from the same release vintage together. Add the table or paragraph so you can find them again. Keep a forecast in a separate line with its own dated source.
- Record revisions separately. For each earlier period, write the previous estimate, the revised estimate and their difference. A revision changes the earlier period; it is not automatically part of the newest month’s change.
- Show your calculation. Write the formula, substitute the numbers with their signs and units, and state your rounding. Then write one sentence about what the result does not establish.
- Finish with one next check. Separate one reported fact from your interpretation. Name the unresolved question and the next original document that could answer it. A usable note can stop before reaching an investment conclusion.
Completed practice: new jobs and old revisions are different lines
All figures below are invented, in thousands of payroll jobs, seasonally adjusted. “Month A,” “Month B” and “Month C” are fictional periods, not current economic results. This example shows how to preserve a release’s vintage; it does not forecast the next jobs report.
| Period | Previously reported | Current release | Revision |
|---|---|---|---|
| Month A | +160 | +140 | −20 |
| Month B | +180 | +150 | −30 |
| Month C | Not yet reported | +120 | First estimate |
- Reported fact
- Month C’s first estimate is +120,000 payroll jobs. The current release revises A to +140,000 and B to +150,000.
- My revision calculation
- (140 − 160) + (150 − 180) = −50 thousand. That is a combined downward revision of 50,000 to A and B, not a statement that C added only 70,000 jobs.
- My consistent-vintage comparison
- 140 + 150 + 120 = 410 thousand across the three monthly changes. Divide by three for an average of about 136,700 jobs per month, rounded to the nearest hundred. Do not mix the old A/B estimates with the new C estimate.
- What this does not establish
- The payroll changes are not the total number of employed people or the unemployment rate. They do not identify a particular company’s earnings or tell us which direction its share price will move.
- Next check
- Read the industry detail and the separate household-survey figures, then retain this release so a later revision can be identified.
For an actual report, use the BLS Employment Situation release and its tables. BLS’s methods guide distinguishes the payroll and household surveys; its payroll-survey FAQ explains revisions. Method links checked November 1, 2026. The worksheet and fictional arithmetic are original teaching material.
Keep your completed template on your device, or transfer the public facts and next question into the existing research notebook. The downloaded blank template contains no current report values; fill it from the original release you are reading.
Work through the numbers
Core checkpoint: explain what you found
Name the producing agency, distinguish a monthly move from an annual or annualized one, and check whether a figure was revised. State what the release says and what it cannot establish about the next market move.
You can stop with that record and choose a next step below, or continue with the optional extensions.
Choose your next practical stepOptional deeper practice
Three checks that change the meaning of a headline
A report becomes more useful when you can reproduce its comparison. These original exercises use invented numbers to isolate annualization, the unemployment denominator and revisions. They are separate from the historical CPI release below.
A quarterly gain is not a twelve-month gain
Suppose a seasonally adjusted real-output index rises from 100 in one quarter to 101 in the next. The actual quarter-to-quarter gain is 1%: (101 ÷ 100 − 1) × 100. Expressing that pace at an annual rate gives about 4.06%: [(101 ÷ 100)4 − 1] × 100.
The second calculation asks what the growth would be if the same quarterly pace continued for four quarters. It is neither a forecast nor the observed change from the same quarter a year earlier. Dividing an annualized percentage by four is only an approximation; reversing the compounding requires a fourth root.
Source: BEA’s quarterly and annual-rate formulas. For a real release, keep the quarter, real/current-dollar label and estimate date together.
The unemployment rate has a changing denominator
In a simplified fictional population, 960 people are employed and 40 meet the definition of unemployed. The labor force is 1,000, so unemployment is 40 ÷ 1,000 = 4%. If ten of those unemployed people stop actively looking and no longer meet the unemployment definition, employment stays at 960. There are now 30 unemployed people and a labor force of 990: the rate is approximately 3.03%.
The lower rate in this example does not mean ten people found jobs. The categories changed. In another scenario, people entering the labor force can raise the unemployment count even while employment increases. Read the employment and labor-force figures alongside the rate.
The payroll survey counts nonfarm wage-and-salary jobs; the household survey counts people. A person with two covered payroll jobs can count twice in one and once in the other. Do not subtract payroll employment from the population to calculate unemployment. Source: BLS’s guide to the two surveys and their definitions. The simplified example assumes the stated classification changes and holds everything else constant.
Keep a record of revisions instead of quietly replacing the old note
Imagine a fictional payroll release initially reports a monthly increase of 180,000 jobs. The next release revises it to 150,000 and reports 120,000 for the new month. On the new vintage, the two monthly changes total 270,000, not 300,000. The revision is −30,000 to an earlier estimate; it is not a separate new-month job loss to subtract again.
BLS revises initial payroll estimates as additional information arrives and makes benchmark revisions separately. Save both the period measured and the release date so that comparisons use a consistent vintage. Source: BLS payroll-survey FAQs, revisions and benchmarking.
Your completed reading note
For the fictional output example: “Measure: seasonally adjusted real-output index. Period: quarter 2 versus quarter 1. Values: 101 and 100. My calculation: +1% quarterly, equivalent to approximately +4.06% at an annual rate. Unknown: actual year-over-year growth, because the year-earlier value is absent. Next check: obtain the same series and vintage for that quarter.”
Try one more calculation
A fictional index rises from 200 to 201 in one quarter. The quarterly increase is 0.5%; annualizing gives (1.0054 − 1) × 100, or approximately 2.02%. A headline saying “output rose 2.02% during the quarter” would omit the annual-rate convention.
Method sources checked October 24, 2026. Calculations and scenarios are original teaching examples, not current economic readings. Use the research notebook to retain a source, its date and the question still unanswered.
A historical CPI release has more than one correct number
The June 2024 CPI exercise later in this guide is a useful place to practice reading a whole release. Open the BLS release published July 11, 2024 and keep its vintage visible. The agency reported a seasonally adjusted monthly all-items decline of 0.1%, while the unadjusted twelve-month increase was 3.0%. Food increased 0.2% over the month and gasoline fell 3.8%. These are historical readings, not the latest inflation data.
A broad index can decline while some categories rise. It is not a vote in which each category gets one equal share. Nor does a fall in gasoline imply that rent, insurance or groceries became cheaper. A useful note identifies the measure behind each number before explaining the result. Avoid mixing a category’s monthly movement with the whole index’s annual movement.
The release also carries a reissue notice explaining that certain inpatient-hospital-service figures were removed because they did not meet publication criteria. Keep that notice with the archived release. It is a concrete reason to preserve the source version rather than relying on a screenshot. This correction notice does not, by itself, mean every published CPI figure was withdrawn.
Our reading exercise asks you to reproduce the historical publication, not to overwrite it with today’s revised series. The BLS CPI questions and answers explains that seasonal factors are re-estimated and previously published seasonally adjusted indexes can be revised. If a current database differs from an archived release, first check series, adjustment and vintage before assuming either was copied incorrectly.
Lower inflation can leave a much higher price level
Use this fictional fixed-basket index to separate the level from its rate of change. Set the starting index to 100. One year later it is 110, a 10% increase. Another year later it is 112.2, a 2% increase from 110. Inflation slowed from 10% to 2%, yet the basket is 12.2% more expensive than at the starting point. The second year did not undo the first year’s increase.
Diagram scrolls horizontally on small screens.
To return from 112.2 to 100, the required decline would be (100 ÷ 112.2 − 1) × 100, approximately −10.87%. A 12.2% decline would overshoot because it uses the higher level as the denominator. Percentage increases and decreases of the same size do not generally cancel.
Try another path: an invented index goes from 200 to 210 to 212.1. The annual increases are 5% and 1%; the total increase is 6.05%. Adding 5% and 1% gives only an approximation because the second increase applies to 210. This is the same compounding principle used in the compound-interest lesson, applied to a price index rather than an investment balance.
A changing annual comparison also has a changing base: each new twelve-month window drops an old month and adds a new one. An annual rate can therefore change even if the latest monthly movement is modest. Read the monthly and annual comparisons together, with their adjustment conventions, instead of treating one as a full explanation of the other.
A category’s size matters as well as its price change
Imagine a fictional household buys exactly the same quantities each year. Its starting basket costs $100: $80 for category A and $20 for category B. Category A becomes 10% more expensive; category B becomes 10% cheaper. The new basket costs $88 + $18 = $106, a 6% increase. Averaging +10% and −10% to get zero would assign equal importance to unequal spending amounts.
| Category | Initial cost | Price change | New cost |
|---|---|---|---|
| A | $80 | +10% | $88 |
| B | $20 | −10% | $18 |
| Total | $100 | +6% | $106 |
Now reverse the initial spending: $20 in A and $80 in B. The same category price movements produce $22 + $72 = $94, a 6% decline. Nothing about the price changes changed; the basket weights did. This is why an aggregate measure and a particular household’s experience can differ without either being fabricated.
BLS explains that CPI represents a defined population’s average experience, not every household’s personal budget. The actual index also uses methods more complex than this two-item fixed basket. Do not rebuild official CPI by casually averaging headline categories, which can overlap or use different weights and adjustments. Use the exercise to ask a better question: how much does this category contribute to the relevant measure?
When connecting inflation to a business, its input mix matters too. A company buying fuel, imported components and labor does not have the same exposure as a consumer basket. First identify the actual cost categories in company filings; then ask which published measures, if any, are relevant proxies. A national index is context, not a complete company cost model.
Separate a larger dollar total from more activity
Suppose a fictional store sells 1,000 identical units at $10 each in the first period, generating $10,000 of sales. In the second period it sells 980 units at $11, generating $10,780. Dollar sales rose 7.8%, even though units fell 2%. A headline describing the dollar increase as 7.8% more units would be wrong.
The relationship is multiplicative: a 10% price increase combined with a 2% quantity decline gives 1.10 × 0.98 = 1.078. In a simplified example with a suitable price index, real growth can be calculated as (1 + nominal growth) ÷ (1 + price growth) − 1. If nominal growth is 3% and prices rise 2%, the result is 1.03 ÷ 1.02 − 1 = approximately 0.98%. Subtracting 2% from 3% gives a close approximation, not the exact compounded result.
Real economic releases use their own coverage, weighting and methods. Do not divide an unrelated sales series by any convenient inflation index and label it an official real measure. Verify whether the agency already publishes the needed real series and read its definition. For a company, changing product mix, currency and acquisitions can also alter reported dollar growth.
Answer check: if identical-unit volume is unchanged while price increases 4%, nominal revenue rises 4% under those assumptions. That says nothing by itself about profit: costs might rise faster, and payment timing might differ. Follow the revenue-to-profit lesson before treating a larger sales total as a larger shareholder return.
Label percentage points and forecast surprises
A fictional rate moves from 4% to 5%. That is an increase of one percentage point, or 25% relative to the original 4% rate. It is not a five-percentage-point increase. Write both the starting and ending values when a headline uses “percent higher,” because the intended denominator may be unclear.
Separate the actual release from anyone’s forecast. Imagine a private forecast calls for 100,000 additional jobs, while a release estimates 120,000. The reported increase is 120,000; the difference from that forecast is 20,000. Neither number is the total level of employment. If the prior month is revised lower, that revision belongs to the earlier period and needs a separate line.
A “beat” or “miss” therefore needs a named forecast source, collection date, units and comparison definition. Different forecasters can disagree. An agency release generally supplies the estimate, not a universal market expectation or a guaranteed price response. Even a correctly calculated surprise does not tell you how an individual stock must move.
Use three columns in your note: agency result, dated forecast and your comparison. If no reliable forecast source is available, leave the forecast blank. A good factual summary does not need to manufacture a consensus number to sound informed.
Connect the economy to a company through a testable mechanism
Start with one exposure, not the entire economy. A fictional delivery business might buy fuel, employ drivers, borrow for vehicles and serve customers whose demand changes. Lower fuel prices could reduce one cost while weaker demand reduces revenue. Fixed debt might not immediately reprice when policy rates change. Those competing channels are questions to investigate in its actual contracts and filings.
Build a short chain: published measure → possible business exposure → company evidence → limitation. For example: “The agency reports a historical gasoline-price decline. Fuel expense may matter to this delivery business. I need its fuel-cost share, hedging arrangements and pricing terms. The headline alone does not establish next quarter’s margin.” This is an inference with missing evidence, not a reported company result.
A practice release and answer key
Consider this entirely fictional packet: an all-items price index is 212.1, versus 210 a year earlier; a retailer’s revenue rose 3%; a relevant simplified price measure rose 2%; and an analyst expected 2.5% revenue growth. A social post says “Prices returned to normal, shoppers bought 3% more, and the retailer beat inflation by 0.5%.” Identify the denominator or evidence problem in each claim.
Read the worked answer
The index increased 1% over the year: 212.1 ÷ 210 − 1. It did not return to an earlier level merely because the rate is smaller. A 3% revenue increase does not establish 3% volume growth; with the exercise’s suitable 2% price measure, the simplified real increase is approximately 0.98%. The 0.5-percentage-point difference compares reported revenue growth with the analyst’s 2.5% forecast, not with inflation. Keep the forecast comparison separate from the real-growth calculation.
A corrected note records the index level and comparison period, the nominal sales result, the stated price assumption and the named forecast. It concludes that these facts do not establish profit growth, valuation or a future stock return.
For your next actual release, save six items in the research notebook: measure, period, release vintage, units/adjustment, reproducible comparison and unresolved question. Use the glossary when a term interrupts the reading, then return to the source. Keep historical practice notes dated; use the news desk’s original agency links when checking a new release.
Additional BLS sources checked October 25, 2026. The June 2024 CPI case is historical. All other new index, basket, retailer, forecast and delivery-business examples are fictional educational exercises. These calculations are not current economic readings or financial advice, and no independent professional review is implied.
Explain GDP growth with contributions, not a list of growth rates
“Consumer spending grew” and “consumer spending contributed to GDP growth” answer different questions. The first describes a component’s own change. The second describes how much that component added to the overall change. A small category can grow quickly and still contribute less than a large category that grows slowly.
Fictional contribution table: imagine a quarterly real-GDP release with the following contributions, expressed in percentage points to the annualized growth rate. These are invented teaching values, not an actual BEA release. Each row is a separate, nonoverlapping contribution; the total is the headline growth rate.
| Component | Contribution, percentage points |
|---|---|
| Consumer spending | +1.4 |
| Private fixed investment | +0.5 |
| Private inventory investment | +0.8 |
| Government spending and investment | +0.2 |
| Exports | +0.3 |
| Imports | −1.1 |
| Real-GDP growth rate | 2.1% |
Check the sum: 1.4 + 0.5 + 0.8 + 0.2 + 0.3 − 1.1 = 2.1. You may say consumer spending contributed 1.4 percentage points to the 2.1% annualized increase. You cannot say consumer spending itself grew 1.4% from this table. Its own growth rate belongs in a different row or table.
Similarly, consumer spending and private fixed investment together contributed 1.9 percentage points. That is not the growth rate of their combined aggregate. BEA calls that aggregate final sales to private domestic purchasers; consult its separately published growth rate when asking about those purchases. This distinction prevents a convincing-looking arithmetic answer to the wrong economic question.
Use the agency’s contribution estimates instead of rebuilding them by summing inflation-adjusted dollar amounts. BEA explains that chained-dollar component values generally are not additive outside the reference year and should not be used to calculate component contributions. In a real contribution table, displayed rounding can also keep your sum from matching the headline perfectly. Check units, rounding and whether you accidentally included both a parent category and its subcategories.
Read imports and inventories as accounting relationships
The negative import contribution is easy to turn into an oversimplified story. BEA’s expenditures explanation describes why imports are subtracted: consumption and other expenditure categories include imported goods and services, while GDP measures domestic production. Subtraction removes foreign production already included in those expenditures.
Second fictional case: a household buys an imported item for $100. Assume its imported value is $80 and the remaining $20 consists entirely of domestic retail services, with no other inputs or accounting complications. Consumption records $100; subtracting the $80 import leaves $20 of domestic value added. Counting the purchase as $100 of domestic production would include the foreign-made portion. Looking only at the minus $80 and calling it an $80 loss to the economy also misses the rest of the transaction.
Inventory investment deserves similar care. The contribution depends on changes in inventory investment, not simply on whether warehouses contain many goods. A positive contribution is not proof that final customers bought more, that managers intended the buildup, or that companies will earn higher profits. BEA also notes that inventory figures can include imported goods, so the inventory and import contributions should be read together before assigning a cause.
Try this source-reading sequence on a release: record its date and estimate vintage; locate the headline’s period and annualization; open the contribution table; identify the largest positive and negative entries; then read the explanation of their underlying source data. BEA’s additional-information page links its growth and contribution tables separately. Save the exact table title with your note so another reader can reproduce it.
Practice: repair a GDP summary
A draft says, “Consumer spending grew 1.4%, imports destroyed 1.1% of the economy, and inventories prove demand is strong.” From the fictional table, replace it with: “Consumer spending contributed 1.4 percentage points and imports contributed −1.1 percentage points to annualized GDP growth. Inventory investment contributed 0.8 percentage point; the table alone does not show why inventories changed.” The GDP rate is 2.1%, while the spending growth rate and business explanation still require separate evidence.
Expansion sources reviewed November 1, 2026: BEA’s expenditures and additional-information pages linked above, plus its chained-dollar glossary. All table values and the imported-item transaction are original fictional examples. Neither case is a current economic estimate or a policy recommendation.
