How a True Number Can Mislead, and the Questions That Catch It

Most numbers quoted in an argument about the economy are accurate. What misleads is the frame around them: where the clock starts, which dollars, which measure, what pace. Seven questions catch most of it, and the almanac's own data shows each one cutting both ways.

In one sentence

Most economic figures quoted in an argument are accurate; what misleads is the frame around the figure, and a short list of questions about the frame catches most of it.

New articles and charts arrive in one short email on Monday mornings. Get the weekly digest

The analogy

A cropped photograph. Every pixel in it is real, and the crop decides what it shows: a packed section of the stands and a half-empty stadium can be the same game. An economic number gets cropped the same ways, by where the frame starts, how close the lens is and which caption goes under it. Nobody has to invent a figure to mislead with one.

Each section below names one of those crops, gives the question that catches it, and shows a figure from the almanac's own data that supports opposite claims depending on the frame. The examples are figures as of late October 2026; the charts stay current.

Where does the clock start?

In the last week of October 2026 a gallon of regular cost $4.27. "Gas has more than doubled" was true: in late April 2020, with much of the country shut down, it cost $1.77, and $4.27 is 141% more. "Gas has come down" was also true: in mid-June 2022 it cost $5.01, and $4.27 is 15% less. Same series, same week, two honest headlines. The start date did all the work.

The same thing happens to presidents. Counted from inauguration, jobs fell by 58,000 a month in Trump's first term; counted through February 2020, before the pandemic, they rose by 180,000 a month. Counted from inauguration, Biden's term added 321,000 a month, the most of any term since 1939; shift the whole window a year later, which drops the rebound and adds his successor's first year, and it is 179,000. Indexed to Inauguration lays out every term both ways.

Ask: why that start date, and is it a peak or a trough? What does the line say from a year earlier, or a year later?

Ten years of regular gasoline. The April 2020 low and the June 2022 peak are both on it, and any claim about the change since then has picked one of them. Full chart and context →

Is it a speed or a distance?

In September 2026 inflation was 3.7%, down from 9% in mid-2022, and prices were nearly a third higher than in 2019. "Inflation has come way down" and "everything costs a third more" describe the same economy read off two gauges: the rate is a speedometer, the price level an odometer, and the odometer never runs backward. The deficit and the debt are the same pair. A smaller deficit means the debt is growing more slowly, not shrinking. Rate vs. level goes through the pairs.

Ask: is this how fast something is changing, or where it stands? What does the other gauge say?

A record in which dollars?

In June 2026 a gallon of whole milk cost $4.32, the highest in the Bureau of Labor Statistics series, which began in 1995. Adjusted for inflation, that same June price was cheaper than four months in five since then. A record at the register, and below most of its own history.

It runs the other way too. Average hourly pay for production and nonsupervisory workers, $32.60 in September 2026, set a record in dollars every month from February 2021 on, and in about nine months out of ten since 1964, because dollars shrink. "Record pay" is nearly always true, which is why it says so little: measured against a year earlier, pay at best kept pace with prices from April through August 2026, and in September it fell behind, 3.3% against 3.7%. Anything priced in dollars drifts toward records for the same reason, which is why the almanac's price tiles are read in today's dollars and its government figures as a share of the economy.

Ask: in today's dollars? For a government figure, as a share of the economy?

Whole milk as charged, and the same prices restated in today's dollars. As charged, the peak was June 2026; in today's dollars, it was June 2004. Full chart and context →

Which measure?

Most things have more than one official measure, often from the same agency and sometimes in the same release.

Ask: is there another official measure of the same thing, and what does it say? If the speaker usually quotes the other one, why not this time?

The headline unemployment rate and U-6, from the same monthly survey. They rise and fall together; the gap between them is the people the headline leaves out. Full chart and context →

At what pace?

The U.S. reports quarterly growth as if the quarter's pace kept up for a year. The third quarter's 3.4% means the economy grew 0.84% in those three months. Europe reports the quarter itself. For the second quarter of 2026, the first U.S. estimate was 1.5% and Eurostat's first estimate for the euro area was 0.4%. The U.S. looked nearly four times as fast. In like terms it grew 0.37% to the euro area's 0.4%. (The U.S. figure has since been revised to 2.2%.)

Monthly figures can be stretched the same way. Consumer prices rose 0.65% in September 2026 alone, about 8% at an annual pace, while the rise over the year was 3.7%. "Inflation is running at 8%" and "inflation is 3.7%" both described that month.

Ask: over what period, and is it annualized? Is one month being stretched into a year?

Is this the first print?

The first estimate of most numbers comes from a sample. It is revised as late reports arrive, and again a year later against fuller records. May 2026's jobs number was first reported as 172,000; by October it stood at 63,000. The second quarter of 2022 was first reported as a contraction of 0.9% and now shows growth of 0.6%. A claim built on a first print can be true when made and wrong a month later, in either direction, and first prints have tended to run high as the economy turns down and low as it turns up. Revisions and vintages has the pattern, and the argument over what the 2024–25 run of downward revisions meant.

Ask: has this number been revised since, and is it due to be?

Who was driving?

A number moving during someone's term is not the same as that person moving it. Short-term interest rates are set by the Fed, and its moves take a year or more to reach hiring. Tax and spending laws need Congress and often take a year or more to reach paychecks. The price of oil is set on a world market; what a president can do about it, and how much that moves a gallon, is on the gas page. And the business cycle is mostly inherited. Every president since Reagan who took office with unemployment above 6% left it lower, and every one who took office with it below 6% has seen it rise, the current term included so far.

Ask: when did the policy take effect, and what else changed at the same time?

Where the analogy breaks

A cropped photo has an uncropped original. An economic number doesn't. Every comparison needs a start date, every rate a period, every dollar figure a choice of whose dollars, and none of those choices is neutral. The test isn't whether someone chose. It is whether they say what they chose, and whether they would make the same choice if it pointed the other way. Some crops are the honest version: setting aside the pandemic months, or reading core inflation through an oil shock, can show a trend better than the full picture does. An honest crop is stated, and it stays the same when the numbers turn.

What people argue about

Whether the economy does better under one party. From 1949 to 2013, real GDP grew about 4.3% a year under Democratic presidents and 2.5% under Republican ones, and 41 of the 49 quarters spent in recession came under Republicans. Alan Blinder and Mark Watson tested whether a gap that size could be a statistical fluke across only sixteen terms, and found it very unlikely. One side reads that as a record of stewardship. The other points to the same paper, which finds that fiscal and monetary policy don't explain the gap (if anything, policy was slightly more favorable to growth under Republicans, a difference the authors say they would not make much of) and traces up to 56% of it to milder oil shocks, faster productivity growth, more optimistic consumers and, counting the Korean War, defense spending: things the authors call blends of good policy and good luck. The rest is unexplained, and the figures stop in 2013.

When leaving something out is honest. Comparisons that drop 2020, and inflation measures that drop food and energy, are criticized as convenient: the pandemic happened, and people buy groceries and gas. Defenders answer that a one-time shutdown says nothing about policy, and that core inflation has been a better guide to where headline inflation is going than the headline itself. Each side is right about something. The trouble starts when the exclusion is used only when it helps.

See also

Sources

The weekly digest

One email on Monday mornings: the week's new articles and charts, the numbers that moved most, and what comes out in the week ahead. No ads, and no tracking of opens or clicks.

Past issuesHow addresses are kept