Chapter 4 — How to Read Economic Data

Numbers, Charts, and the Stories They Tell (and Hide)

This is the chapter Mankiw doesn't have. Most introductory economics textbooks treat data as something you eventually arrive at after understanding the theory — first you learn what supply and demand are, then you learn how to look up actual prices and quantities. This book inverts that. We're going to learn how to read economic data first, before any of the technical machinery starts. The reason is practical: by Chapter 5, you will be reading about real markets and real prices, and the rest of the book will assume you can look up the numbers yourself without panicking. By Chapter 22, when we get to GDP, we will be talking about the actual GDP report you can pull off the BEA's website rather than an abstraction. By Chapter 27, when we get to monetary policy, we will assume you can read what the Federal Reserve actually publishes about its decisions. The skill of looking at a real economic dataset and knowing what it means is the foundation of every other skill this textbook is trying to teach.

It is also a deeply useful skill outside of economics. Public discourse is awash in confident claims about the economy ("inflation is the highest it has ever been," "unemployment is at a 50-year low," "wages are stagnant," "the Fed should cut rates"). Some of these claims are true, some are misleading, and some are flatly wrong. Telling them apart requires knowing where the numbers come from, what they actually measure, and how they can be presented in ways that flatter one story or another. By the end of this chapter, you should be able to fact-check most economic claims you see in the news using free, public, government data. That is a real skill, and economics graduates who have it are unusually employable. If you have it without the rest of an economics degree, you are still ahead of most of the people who comment on economic news for a living.

The chapter has six sections. Section 1 walks through the institutions that produce the data — the BLS, the BEA, the Census, the Federal Reserve, FRED. Section 2 walks through the actual layout of a BLS jobs report and shows you the three numbers journalists usually highlight and the three numbers you should also look at. Section 3 does the same for a CPI release. Section 4 introduces FRED, the free interface that gives you instant access to most of the data the U.S. government publishes. Section 5 walks through three common ways economic charts mislead. Section 6 tells you what to do when you encounter an economic claim in the news. By the end, you will have done at least one real data lookup yourself, which is the only way to actually learn this material.

4.1 Where economic data comes from

In the United States, most economic data is produced by four federal agencies and a couple of private organizations. Knowing which agency produces which data tells you a lot about what you're looking at and how reliable it is.

The Bureau of Labor Statistics (BLS) is part of the U.S. Department of Labor. It produces the monthly Employment Situation report ("the jobs report"), the Consumer Price Index, the Producer Price Index, the Job Openings and Labor Turnover Survey (JOLTS), and a great deal of other data on wages, hours, employment, and prices. The BLS is generally regarded as one of the most professional and methodologically careful statistical agencies in the world. Its data is released on a strict schedule, often months in advance, and political interference is almost unheard of. (When political interference has been alleged — for instance, during the 2020 election cycle — the response inside the agency was unanimous and forceful in defense of the data's integrity.)

The Bureau of Economic Analysis (BEA) is part of the U.S. Department of Commerce. It produces the quarterly Gross Domestic Product report, the personal income and consumer spending statistics, the trade statistics, and the balance of payments. Like the BLS, the BEA is highly professional. Its data revisions — common in macroeconomic statistics, where preliminary numbers are updated as more information comes in — sometimes get blamed for confusion in real time, but the revisions themselves are an honest reflection of how complicated it is to measure a $25 trillion economy in real time.

The U.S. Census Bureau is also part of the Department of Commerce. It produces the decennial census (every ten years), the American Community Survey (annual, more detailed), and a host of demographic and economic surveys including data on poverty, household income, business formation, and retail sales. The Census Bureau is the source for most data on who people are and what they earn — the demographic and distributional information that shows up in every chapter of this book that touches on inequality or housing or labor markets.

The Federal Reserve System publishes its own data on monetary policy, the financial system, banking, household finances, and many other topics. The Federal Reserve Bank of St. Louis maintains FRED — the Federal Reserve Economic Data system — which is the single most useful free economic data resource in the world. We'll spend §4.4 on it.

In addition, several other federal agencies publish important data: the Internal Revenue Service (income and tax data), the Department of Agriculture (farming and food prices), the Energy Information Administration (energy production and prices), the Centers for Disease Control (vital statistics, life expectancy), and others. International organizations — the World Bank, the International Monetary Fund, the OECD, the United Nations — produce comparable data for other countries and for cross-country analysis.

A few private organizations also produce widely cited economic data. The Conference Board publishes the Consumer Confidence Index. ADP (the payroll company) publishes a private monthly employment report based on its actual payroll records. Moody's Analytics and other private forecasters publish their own GDP estimates. The Conference Board also produces the Leading Economic Indicators series. None of these has the authority of government data, but each can be useful as a cross-check.

For this chapter, we'll focus on the four federal agencies and especially on FRED, which gives you a unified interface to most of what they produce.

4.2 How to read a BLS jobs report

The Employment Situation Report — informally, "the jobs report" — is released by the BLS on the first Friday of every month at 8:30 a.m. Eastern time. It is the most-watched single piece of economic data in the U.S. economy. Reporters cover it. Markets move on it. The Federal Reserve uses it. Politicians cite it. It has its own ritualized rhythm: the numbers get released, headlines appear within minutes, market reactions happen within seconds, and the analytical takes follow over the next few hours.

The report is also widely misreported. The headlines almost always emphasize one or two numbers, and journalists who don't know better treat those numbers as if they were the whole story. They are not.

Let's walk through what's actually in the report.

The two surveys

The jobs report is built from two surveys conducted by the BLS each month:

The Establishment Survey (also called the Current Employment Statistics, or CES) is a survey of about 122,000 businesses and government agencies, covering roughly 666,000 worksites. It asks employers how many people they had on payroll during the pay period that includes the 12th of the month. The Establishment Survey produces the headline payroll employment number — "nonfarm payrolls" — that you see in the news.

The Household Survey (also called the Current Population Survey, or CPS) is a survey of about 60,000 households conducted by the Census Bureau on behalf of the BLS. It asks individuals about their employment status, hours worked, and other labor force characteristics. The Household Survey produces the unemployment rate.

The two surveys ask different questions of different respondents and produce different numbers. They usually point in the same direction, but they often disagree on the details. The reasons for the disagreement are interesting and we'll come back to them.

What journalists usually report

When the jobs report comes out, the typical news headline says something like: "U.S. Adds 250,000 Jobs in March; Unemployment Rate Holds at 3.8%."

That headline contains two numbers, both real: - Nonfarm payroll employment change (250,000) — from the Establishment Survey - Unemployment rate (3.8%) — from the Household Survey

These are the two most-cited numbers. They are also genuinely important. But they leave out a lot.

What you should also look at

If you read past the first paragraph of the BLS press release, you'll find:

1. The labor force participation rate. This is the share of the working-age population (16+) that is either working or actively looking for work. The unemployment rate only counts people who are actively looking. If discouraged workers stop looking, they leave the labor force entirely, and the unemployment rate falls — even though no one got a job. The labor force participation rate tells you whether the falling unemployment rate reflects more hiring or more discouragement. Both are real, but they're very different.

For context: U.S. labor force participation peaked at about 67.3% in early 2000 and has been declining for two decades, partly because of demographic aging (older workers are less likely to participate) and partly because of structural changes in the labor market. It dropped sharply during COVID and has only partially recovered. As of early 2026, it's around 62.5%, which is several percentage points below the late-1990s peak.

2. The U-6 unemployment rate. The headline unemployment rate is U-3 — the BLS's narrowest definition. U-6 includes: - The U-3 unemployed (people who lost a job and are looking) - "Marginally attached" workers (people who want work and have looked recently but aren't actively looking this month) - "Discouraged workers" (people who want work but have given up looking because they think no jobs are available) - People working part-time who would prefer full-time

U-6 is typically about twice as high as U-3. In a tight labor market it might be 7%; in a recession it can be 18%. The gap between U-3 and U-6 tells you about the quality of the employment situation, not just the quantity.

3. Average hourly earnings (and average weekly hours). Wages are also in the jobs report, but they don't usually make the headline. If wages are growing faster than inflation, real wages are rising; if wages are growing slower than inflation, real wages are falling. The headline jobs number can look great (lots of jobs added) while real wages are stagnant or declining. Both pictures are part of the labor market story.

4. The "diffusion index." This is a measure of how broadly employment growth is spread across industries. If employment is rising in 80% of industries, that's a sign of broad-based growth. If employment is rising overall but only in a few industries (say, healthcare and government), that's a sign of narrower, more vulnerable growth. The diffusion index doesn't get reported in news headlines, but it's in the BLS press release and it tells you something the headline doesn't.

5. Revisions to previous months. The BLS publishes revisions to the previous two months' data in each new report. These revisions can be substantial — sometimes 50,000 jobs or more in either direction. If you look at this month's report and ignore the revisions, you might miss that last month's "great" report has been revised down to "okay," or that the previous "okay" report has been revised up to "great." Always look at the revisions.

6. The household-establishment gap. As noted above, the two surveys often disagree. When the establishment survey says employment rose by 250,000 and the household survey says it rose by 50,000 (or fell), something is happening that requires interpretation. The BLS itself usually has a paragraph in the press release explaining the gap. Read it.

A worked example

Let's walk through a hypothetical jobs report and show what the careful reading looks like.

Hypothetical report: "March 2026 — U.S. economy added 280,000 nonfarm payroll jobs. Unemployment rate fell to 3.9%."

The headline tells you that employment rose and unemployment fell. Good news, surely?

Now look at the rest of the report: - Labor force participation: 62.4% (down 0.1% from February). The unemployment rate fell partly because some people stopped looking for work, not entirely because they got jobs. - U-6: 7.3% (down 0.1% from February). Roughly half the U-3 decline showed up in U-6, which means some of it reflects real labor market improvement and some reflects discouragement. - Average hourly earnings: Up 4.1% year-over-year. Inflation over the same period was 3.2%. Real wages are growing at about 0.9% — modestly positive. - Revisions: February was revised down by 35,000 (from 230,000 to 195,000). January was revised up by 12,000. - Household survey: Reported employment rose by 165,000 — substantially less than the establishment survey's 280,000. The gap is unusual but not unprecedented.

What does this report actually say? It says: the labor market is still growing, but more slowly than the headline suggests, with some discouraged workers leaving the labor force, real wage growth that's positive but modest, and a household survey that's notably weaker than the establishment survey. A more accurate headline would be: "U.S. Job Growth Continues at Steady Pace, but Underlying Picture Mixed."

Most news organizations will not write that headline, because it doesn't fit in 70 characters and doesn't drive clicks. But that's the honest summary of what the BLS just released. If you can read the report carefully and write the honest summary in your own head, you are doing better than 90% of the people who report on economic data.

4.3 How to read a CPI release

The Consumer Price Index is released by the BLS in the second week of each month, usually around the 10th–13th. It is the second most-watched economic data release after the jobs report. It is also chronically misunderstood, partly because the methodology is genuinely complicated and partly because the political stakes are large.

What the CPI measures

The CPI is an index that tracks the average price of a fixed basket of goods and services purchased by urban consumers in the United States. The basket is updated every two years based on the Consumer Expenditure Survey, which tells the BLS what households actually spend their money on.

The basket has roughly 80,000 items in it, organized into eight major categories: 1. Food and beverages 2. Housing (rent, owners' equivalent rent, fuels and utilities, household furnishings) 3. Apparel 4. Transportation (cars, gasoline, public transportation) 5. Medical care 6. Recreation 7. Education and communication 8. Other goods and services

Every month, BLS price collectors visit thousands of retail outlets, restaurants, gas stations, doctors' offices, and other establishments across 75 urban areas. They record the prices of the items in the basket. Those prices get aggregated, weighted by how much households spend on each category, and combined into the overall index.

The index is set to 100 for a base period (currently 1982–1984 = 100). The current value tells you how much more expensive the basket is than in the base period. As of early 2026, the CPI is around 320 — meaning that the average basket of goods costs about 3.2 times what it cost in 1982–84.

Headline vs. core

The headline CPI is the all-items index. Inflation is usually reported as the year-over-year percentage change in the headline index.

Core CPI excludes food and energy. The reason: food and energy prices are highly volatile in the short run because they depend on weather, oil markets, and other transient factors. Excluding them gives you a better measure of underlying inflation trends. Central banks (including the Federal Reserve) generally watch core inflation more closely than headline.

This is one place where ordinary people's intuitions diverge from professional economists' framing. Most consumers think of food and gas as the most important things they buy — and they are absolutely right that those are the things they notice. But for monetary policy purposes, the question is "what are prices doing in the underlying economy, abstracting from temporary food and energy shocks?" — and that question is answered by core inflation, not headline.

When people complain that "the official inflation rate doesn't reflect what I see at the grocery store," part of what they're noticing is the food/core distinction. Food prices in 2022–23 rose much faster than core inflation. The headline inflation rate captured this; the core measure did not. Both rates were correct measurements of different things.

The three biases

The CPI is famously biased in three ways. The biases are not deliberate — they are mathematical features of how a fixed-basket index works. Each tends to cause the CPI to overstate true inflation by some small amount per year.

Bias 1 — Substitution bias. When the price of one good rises, consumers substitute toward cheaper alternatives. If beef gets expensive, people buy more chicken. The fixed basket doesn't capture this substitution; it keeps measuring the cost of the same beef quantity. So the fixed-basket CPI overstates the true cost-of-living increase, because consumers have already adapted to the price changes.

Bias 2 — New goods bias. New goods are introduced over time, often at high prices that fall rapidly. Smartphones in 2007 cost hundreds of dollars; today they're more affordable and dramatically better. The CPI is slow to incorporate new goods, so it misses much of the benefit consumers get from technological progress. (BLS does add new goods over time, but the lag means the CPI tends to miss the early years of price decline.)

Bias 3 — Quality change bias. A 2025 car is not the same product as a 1995 car. It has airbags, ABS, navigation, better fuel economy, better safety, longer expected lifespan. If the price has roughly doubled since 1995, you might say "cars have gotten 100% more expensive." But you'd be wrong, because you're comparing different products. The car has gotten better in ways that should be counted. The BLS tries to adjust for quality changes ("hedonic adjustment"), but the adjustments are imperfect, and quality changes that the BLS misses cause the CPI to overstate true inflation.

The Boskin Commission, in 1996, estimated that the combined effect of these three biases was to overstate true inflation by about 1.1 percentage points per year. The BLS has since reformed several measurement procedures, and modern estimates of the bias are smaller — perhaps 0.5–0.8 percentage points. But the bias is still there, and it accumulates over time. Compounded over 40 years, even a 0.5 percentage point per year overstatement would add up to a roughly 22% cumulative overstatement of inflation.

This matters for several real-world purposes. Social Security cost-of-living adjustments use the CPI. Tax brackets are indexed to the CPI. Many private contracts use CPI escalators. If the CPI overstates inflation, all of these adjustments are slightly too generous (or too punitive, depending on which side of them you're on). This is a politically sensitive issue, which is why the BLS is very careful about how it makes methodological changes.

Why your experience differs from the official rate

A common complaint: "The government says inflation is 3%, but everything I buy is going up much faster than that. They must be lying."

The honest answer is that you and the BLS are measuring different things. Your experience reflects: - What you actually buy. If you eat out a lot, drive a lot, or pay for childcare, you're probably experiencing higher inflation than the typical urban consumer in the BLS basket. If you mostly buy electronics and stay home, you're probably experiencing lower inflation. - The most recent prices. Humans remember recent price increases more vividly than recent price decreases or unchanged prices. Behavioral economists call this "salience bias." We're more likely to notice the carton of eggs that doubled in price than the pair of jeans that's the same price as last year. - Loss aversion. Psychologically, a $1 increase in something we buy hurts more than a $1 decrease helps. So we feel inflation more sharply than we feel deflation or stability. - The impact on what we already own. If your rent went up 8% but your mortgage payment stayed the same, the BLS's housing index will reflect a mix of both. Your personal experience reflects which one applies to you.

The BLS publishes a research series called the "Chained CPI" (C-CPI-U), which tries to account for substitution bias more carefully. It also publishes various subindexes for different demographic groups. None of these is the "true" inflation rate; they're all measurements of slightly different things.

The honest framing: the official CPI is one reasonable measurement of an inherently fuzzy concept. It is not lying. It is not the only legitimate measure. And your personal inflation rate may be quite different from the official one — but the difference is usually about what you buy, not about whether the BLS is being honest.

4.4 FRED — your most useful tool

FRED stands for Federal Reserve Economic Data. It is maintained by the research department of the Federal Reserve Bank of St. Louis, and it is — for free, with no registration, no paywall, no ads — one of the best data resources in the world. As of 2026, FRED hosts over 800,000 data series from more than 100 sources, including all of the U.S. federal agencies described in §4.1, the IMF, the World Bank, the OECD, and various private sources.

You can find it at fred.stlouisfed.org. If you're going to do anything in this textbook, do this: open FRED in a browser tab and keep it open as you read.

The basic interface

FRED's homepage has a search bar. Type any economic concept you can think of — "unemployment," "inflation," "GDP," "median household income," "housing starts," "federal funds rate," "Brazilian coffee prices" — and FRED will give you a list of matching data series. Click on one, and you get a chart, a description of the series, the source, the frequency (monthly, quarterly, annual), and the option to download the data as a CSV.

Each series has a unique ID (a short alphanumeric code like "UNRATE" for the U.S. unemployment rate or "GDPC1" for real GDP). If you know the ID, you can go directly to the chart by typing fred.stlouisfed.org/series/UNRATE in your browser. The IDs are stable — UNRATE today is the same as UNRATE last year — which is convenient for citation.

Some series IDs you should bookmark: - UNRATE — civilian unemployment rate (monthly, U.S.) - CIVPART — labor force participation rate (monthly, U.S.) - CPIAUCSL — Consumer Price Index, all items, urban (monthly, U.S.) - CORESTICKM159SFRBATL — Sticky-price core CPI (Atlanta Fed, less volatile than headline core) - GDPC1 — real Gross Domestic Product (quarterly, U.S.) - A939RX0Q048SBEA — real GDP per capita (quarterly, U.S.) - MEHOINUSA672N — real median household income (annual, U.S.) - DGS10 — 10-year Treasury yield (daily, U.S.) - FEDFUNDS — federal funds rate (monthly, U.S.) - CSUSHPINSA — Case-Shiller national home price index (monthly, U.S.) - PAYEMS — total nonfarm payroll employment (monthly, U.S.)

You should not memorize these. You should know that they exist and that you can find them in seconds.

Things you can do in FRED that journalists can't

Most economic news articles show you a single chart — usually a recent slice of one series. FRED lets you do much more.

1. Compare two series on the same chart. Want to see whether wage growth is keeping up with inflation? Plot average hourly earnings against the CPI on the same axes. FRED makes this easy.

2. Compare across countries. Want to see how U.S. unemployment compares to UK unemployment compares to French unemployment? FRED has the data, and you can plot all three on the same chart.

3. Convert nominal to real. Most economic series come in both nominal (current dollars) and real (inflation-adjusted) versions. FRED lets you toggle between them.

4. Adjust the time period. Most news charts show the most recent few years. FRED lets you adjust the time window to anything from a few months to the full historical record. The story often looks completely different over different time windows.

5. Apply transformations. FRED lets you compute year-over-year changes, percent changes, log transformations, indexing to a particular date, and more. If you've ever wondered "is this the kind of variable that should be plotted in levels or in growth rates?" — FRED lets you try both.

6. Download the data. Every chart can be downloaded as a CSV (or Excel). If you want to do your own analysis, FRED is the cleanest free source for U.S. economic data.

A walkthrough exercise

Let's do a real exercise. Open FRED. In the search bar, type "civilian unemployment rate" and click on the first result (UNRATE). You should see a chart of U.S. unemployment from January 1948 to the most recent month.

Look at the chart. Notice:

• The big spikes. There's a spike around 1975 (the recession of 1973–75). Another around 1982 (the Volcker recession). Smaller spikes around 1991, 2001, and 2003. A massive spike in 2008–09 (the Great Recession). The biggest spike of all in April 2020 (COVID lockdowns).
• The general trend. After each recession, unemployment falls again as the economy recovers. Sometimes the recovery takes years (after 2008, full employment wasn't restored until about 2017). Sometimes it takes months (after COVID, employment recovered remarkably quickly).
• The 1990s and 2000s. Unemployment hovers in a relatively narrow range (4–7%) for most of this period.
• The 2020s. The COVID spike, then a sharp recovery, then a remarkably tight labor market.

Now use the FRED interface to do the following: - Change the date range to "1948 to present" (you may already be there). - Then change it to "2020 to present." How does the picture look different? - Then "1970 to 1985." What stories does this period tell? - Click "Edit Graph" and add a second series: real GDP (GDPC1). Plot both. Notice that recessions show up as both unemployment spikes and GDP dips — but the timing is slightly different.

Spend ten minutes doing this. You will learn more about the U.S. economy in those ten minutes than you would learn from reading three news articles, because you are interacting with the actual data instead of with someone's interpretation of it.

4.5 Three ways economic charts mislead

Even when the underlying data is good, the way it's presented in a chart can deeply shape (and sometimes deceive) the reader. Here are three of the most common tricks. Once you learn to spot them, you'll see them everywhere.

Trick 1 — Truncated y-axis

A line chart with a y-axis that doesn't start at zero can make small changes look enormous. If unemployment goes from 3.8% to 4.2%, plotting it on a y-axis that runs from 0% to 10% shows a barely noticeable upward tick. Plotting it on a y-axis that runs from 3.5% to 4.5% shows a dramatic spike. The data is the same. The visual story is wildly different.

When you see a line chart, always look at the y-axis. If it doesn't start at zero, ask yourself whether the truncation is justified or whether it's hyping a small change.

There are legitimate reasons to truncate. Many time series fluctuate in a narrow range, and forcing the y-axis to start at zero would compress the data into an uninterpretable squiggle. (The Federal Reserve's interest rate charts are a good example — rates have ranged from 0% to 20% over the historical record, and forcing every chart to span the full range would obscure the meaningful variation in any short window.) But truncation is also one of the easiest ways to make a small change look big, and partisan or attention-seeking publications use it constantly.

Trick 2 — Cherry-picked time period

The same data can tell completely different stories depending on which window you show. The 1990s saw declining unemployment; the late 2000s saw rising unemployment; the 2010s saw declining unemployment again. A chart that shows only one of these periods — without context — tells one story. A chart that shows all of them tells a different story. A chart that shows only the recent six months might not show any meaningful pattern at all.

When you see a chart, always check the time period. If it starts in 2009, ask why. If it ends in 2019, ask why. If the period is unusually short, ask whether a longer view would change the story.

A particular variant of this trick: starting a chart at the depth of a recession or the peak of an expansion. From the bottom of the 2009 Great Recession, unemployment fell almost continuously for over a decade. A chart that starts in 2009 makes this look like the longest unbroken expansion in history. A chart that starts in 2007 (before the recession) makes the same period look like a partial recovery from a very deep hole. Both are accurate. They tell different stories.

Trick 3 — Wrong measure

Some data series measure things that are easy to confuse with related things, and journalists often pick the measure that tells the story they want. Some examples:

• Median vs. mean. Median income is the income of the middle household. Mean income is the average. Mean is pulled up by very high earners, so it's almost always higher than the median. A claim about "average income" without specifying which is being used can be misleading.

• Real vs. nominal. Nominal numbers are in the dollars of the time. Real numbers are adjusted for inflation. "Average household income has risen from $50,000 in 1995 to $75,000 in 2025" sounds like a 50% increase — and it is, in nominal terms. In real terms, accounting for inflation, the increase is much smaller. Always check whether a number is real or nominal.

• Stock vs. flow. Some variables are stocks (the total amount at a point in time, like the national debt) and some are flows (the rate at which something changes, like the federal deficit). The difference is fundamental, but news articles often mix them up. "The deficit is $1.5 trillion" and "the debt is $34 trillion" are both true, refer to different things, and are sometimes presented as if they were the same.

• U-3 vs. U-6 unemployment. As discussed in §4.2, these measure different things. A claim about "unemployment" without specifying which measure can be picking the one that supports the writer's preferred narrative.

• Headline vs. core inflation. Headline inflation includes food and energy; core excludes them. Different stories for different purposes. Make sure you know which you're being shown.

The general principle: whenever someone shows you a number, ask "is this the right measure for the question being asked?" Often it isn't, and the discrepancy is doing rhetorical work.

4.6 What to do when you encounter an economic claim in the news

You now have enough to fact-check almost any economic claim you read in the news. The procedure has six steps.

Step 1 — Identify the claim. What is the writer asserting? Be precise. "Inflation is the highest it has been in 40 years" is different from "year-over-year CPI growth in [month] reached the highest level since [month]" is different from "core CPI is rising fast." Each of these can be true or false, and they imply different things.

Step 2 — Find the source. Where does the writer say the data comes from? If they cite the BLS or BEA or Census or FRED, you can verify it directly. If they cite "experts say" or "studies show," ask which experts and which studies — and search for the original source, not the journalist's paraphrase.

Step 3 — Look up the data. Open FRED. Find the relevant series. Look at the actual number, not the journalist's characterization of it.

Step 4 — Check the time period and the measure. Is the time window suspicious? Is the measure the right one? Is the claim about real or nominal numbers? Headline or core?

Step 5 — Check the context. Even if the headline number is correct, is it the most informative number? What does the broader picture look like?

Step 6 — Form your own judgment. Based on what you actually find, what would you say about the claim? Sometimes the journalist was right. Sometimes the headline is misleading but the underlying claim is roughly correct. Sometimes the claim is just wrong. Knowing which is which is the entire payoff of this chapter.

This procedure takes ten to fifteen minutes the first few times you do it. After you've practiced for a few weeks, you can do most of it in two minutes. The skill is enormously useful in your civic life and surprisingly fun once you have it.

4.7 Where this is going

You now have the foundations. In Chapter 5, we'll meet supply and demand — the most consequential model in microeconomics. From here on, when the textbook references real economic data, you can look it up yourself rather than taking the textbook's word for it. That's the whole point of this chapter. Take it seriously: do at least one real FRED lookup before you move on. You will be glad you did.

Key terms recap: time series — a sequence of measurements over time index — a measurement set to a base period of 100 base year — the period the index is normalized to percent change — the change divided by the original, times 100 year-over-year (YoY) — the change from the same period one year earlier seasonally adjusted — corrected for predictable seasonal patterns FRED — the Federal Reserve Economic Data interface BLS, BEA, Census Bureau — the main federal statistical agencies headline vs. core — total CPI vs. CPI excluding food and energy real vs. nominal — adjusted for inflation vs. current dollars

Themes touched: Data tells stories (foundational), Disagreement (about measurement), Affects daily life (every chapter from here on uses real data).