GDP, business cycles, jobs and indicators
Gross domestic product, the business cycle, the unemployment rate and leading indicators are the four numbers most often quoted about an economy, and each is easy to misread. This lesson explains how real GDP is built and quoted, how the NBER dates a recession, how the unemployment rate is measured, and which indicators move first. A calculator built on the September 2026 US jobs data shows how the rate can fall while jobs are lost. It then follows Caterpillar and Procter & Gamble through the 2007 to 2009 recession.
- Level
- Beginner
- Reading time
- About 27 minutes
- Know this first
- Inflation
By the end you will be able to:
- explain how GDP is measured and the difference between real and nominal GDP
- convert a quarterly growth rate to an annual rate
- describe the phases of the business cycle and how a recession is dated
- calculate the unemployment, participation and employment rates and explain why the first can mislead
- tell leading, coincident and lagging indicators apart and say why a company's results depend on the cycle
The economy in four numbers
The last two lessons looked at prices: how one price is set, and how the general price level moves. This one looks at quantities. How much is the economy producing? Is it speeding up or shrinking? How many people have work? And what can be read from the data before the news arrives? Four ideas answer these: gross domestic product for output, the business cycle for the ups and downs, unemployment for jobs, and indicators for early signals.
They matter to finance because a company's sales are not independent of them. When output falls and jobs are lost, some businesses lose far more than others, and the difference is a large part of what investors and lenders try to judge. The lesson ends with two companies that went through the same recession and came out very differently.
GDP: what an economy produces
The US Bureau of Economic Analysis (BEA) describes gross domestic product (GDP) as the total value of goods and services produced within the country. One way to measure it is to add up what is spent. The textbook formula, which BEA gives as C + I + G + X − M, has four parts: consumer spending (C), business investment (I), government spending on goods and services (G), and exports (X), minus imports (M). Imports are subtracted because consumer, investment and government spending can include imported goods, and GDP is meant to count only what is produced domestically.
Two refinements matter. First, GDP can be measured in current prices or in constant prices. Nominal GDP rises when prices rise, so it can grow without the economy making any more. Real GDP removes the effect of price changes, which is why the inflation lesson comes first, and growth is quoted in real terms. The BEA series used below is real GDP in billions of chained 2017 dollars. Second, quarterly figures are quoted at annual rates. BEA says it does this to make quarters comparable with annual and historical data. In the second quarter of 2026 real GDP was 24,408.0 billion, up 0.55% on the first quarter, which is an annual rate of 2.2%.
The figure is also an estimate that gets revised. BEA publishes an advance estimate near the end of the month after a quarter ends, then a second and a third as more complete data arrive. It reports that between 1996 and 2024 the revision from the advance to the second estimate averaged 0.5 percentage points of growth, ignoring sign. A growth rate of 2.0% in the first headline could later read 1.5% or 2.5%.
The business cycle
Output does not grow smoothly. It rises in an expansion, reaches a peak, falls in a contraction, and bottoms out at a trough. In the United States the National Bureau of Economic Research (NBER) has a committee that dates these turning points. It defines a recession as a significant decline in economic activity that is spread across the economy and lasts more than a few months, and it weighs the depth, spread and duration of the fall together. Its recessions start in the month after a peak and end in the month of the trough.
A common rule of thumb says a recession is two consecutive quarters of falling real GDP. The NBER does not use it. It says most of its recessions include two or more such quarters but not all, and gives the 2001 recession as one with no such pair. The committee looks at monthly employment, real personal income less transfers, real consumer spending, industrial production and price-adjusted sales, and uses no fixed weights for them. Two recent recessions show how different they can be.
| 2007-09 recession | 2020 recession | |
|---|---|---|
| NBER peak month (quarter) | December 2007 (2007Q4) | February 2020 (2019Q4) |
| NBER trough month (quarter) | June 2009 (2009Q2) | April 2020 (2020Q2) |
| Months of contraction | 18 | 2 |
| Real GDP, peak quarter to trough quarter | −3.82% | −9.09% |
| Most quarters of falling real GDP in a row | 4 | 2 |
| Unemployment rate in the peak month | 5.0% | 3.5% |
| Highest unemployment rate | 10.0% (October 2009) | 14.8% (April 2020) |
| Unemployment rate 18 months after the trough | 9.3% (December 2010) | 4.5% (October 2021) |
| Quarters after the trough until real GDP regained its peak-quarter level | 6 | 3 |
The first was long and slow. It lasted 18 months, real GDP fell 3.82% from the peak quarter to the trough quarter, GDP fell in four quarters in a row, and the unemployment rate peaked at 10.0% in October 2009, 22 months after the economy's peak month. Eighteen months after the trough, the rate was still 9.3%. The second was short and sharp. It lasted two months, real GDP fell 9.09% in a single quarter, and the unemployment rate jumped from 3.5% to 14.8% in two months, the highest in the series since 2000. Yet eighteen months after its trough the rate was 4.5%, and real GDP regained its pre-recession level in three quarters. A recession is a label for a decline in activity. It does not say how deep the decline was, how long it lasted or how long the recovery took.
Jobs: how unemployment is measured
The unemployment rate has a precise meaning that is easy to get wrong. In the US monthly household survey, run by the Bureau of Labor Statistics (BLS) on about 60,000 households, a person is employed if they did any paid work in the survey week, and unemployed if they have no job, have actively looked for work in the previous four weeks and are available to take one. A person who is neither is not in the labour force, which covers many students, retirees and people with family responsibilities. The labour force is the employed plus the unemployed.
The rate is a fraction, and a fraction can move because the top changes or because the bottom does. If the unemployed stop looking for work, they leave the labour force, and the rate falls even though nobody found a job. If people who were not looking begin to look, the rate can rise even while employment grows. That is why economists read the unemployment rate together with the participation rate and the employment ratio, which does not have this problem because its bottom is the whole population.
The calculator uses the BLS household data for September 2026, which were 163,439 thousand employed, 6,859 thousand unemployed and 105,256 thousand not in the labour force, out of a population of 275,554 thousand. These are not seasonally adjusted figures. The seasonally adjusted headline rate for the same month was 4.2% in the FRED series, against 4.03% unadjusted here. Seasonal adjustment removes regular patterns such as holiday hiring.
When the rate and the jobs disagree
Set two things: the change in the number of people employed, and the change in the number outside the labour force. The population is fixed, so a person who leaves the labour force is taken from the unemployed, and the calculator works out the rest. Everything starts from the September 2026 figures, so with both sliders at zero the unemployment rate is 4.03% (BLS rounds to 4.0%), the participation rate is 61.80% and the employment ratio is 59.31%. Counts are in thousands.
If employment does not change and no one enters or leaves the labour force, the unemployment rate is 4.03%, the same as the starting 4.03%. The participation rate is 61.80% and the employment ratio is 59.31%.
Four experiments. First, add 500 thousand jobs and leave the labour force unchanged. The unemployment rate falls to 3.73%, 0.29 points lower, and the employment ratio rises. Nothing is misleading here, because every new job comes out of the unemployed. Second, set employment to −400 thousand and move the second slider to 600 thousand leaving. The rate falls to 3.92%, lower than the starting figure, in spite of 400 thousand fewer jobs. The people who left are no longer counted as unemployed. The employment ratio and the participation rate both fall, and they give the more honest picture. Third, set employment to +300 thousand and move the second slider to 800 thousand joining. The rate rises to 4.30%, 0.27 points higher, even though 300 thousand more people have jobs. More people began to look than there were jobs created. Fourth, push both sliders to −1,000 thousand. Employment falls by a million and a million people join the labour force, and the rate rises to 5.17%. Reading the rate by itself would be a mistake in each of these.
Indicators: what moves first
GDP is published after the quarter has ended and is then revised, so it cannot tell anyone what is happening today. Economists therefore watch a wider set of series, usually sorted by timing. Leading indicators tend to turn before the economy does, coincident indicators move with it, and lagging indicators turn afterwards. The Conference Board, a research group, publishes composite indexes of each type for the United States, designed to signal peaks and troughs in the business cycle. These are the components in its December 2025 technical notes.
| Type | Components |
|---|---|
| Leading (10) | Average weekly hours, manufacturing; Average weekly initial claims for unemployment insurance; Manufacturers' new orders, consumer goods and materials; ISM New Orders Index; Manufacturers' new orders, nondefense capital goods excluding aircraft; Building permits, new private housing units; S&P 500 stock index; Leading Credit Index; Interest rate spread, 10-year Treasury bonds less federal funds; Average consumer expectations for business conditions |
| Coincident (4) | Employees on nonagricultural payrolls; Personal income less transfer payments; Industrial production; Manufacturing and trade sales |
| Lagging (7) | Inventories to sales ratio, manufacturing and trade; Average duration of unemployment; Consumer installment credit outstanding to personal income ratio; Commercial and industrial loans; Average prime rate; Labor cost per unit of output, manufacturing; Consumer price index for services |
The pattern makes sense. The logic is plausible. Firms tend to change hours worked and the orders they place before they change their payrolls, so those can lead output. Employment, income, production and sales are what the economy is doing now. Duration of unemployment and the prime rate are slower to respond, so they lag. Two caveats apply. No indicator is reliable on its own, and the Conference Board describes its indexes as an analytic system of series, not a single signal. And a leading indicator that fell does not guarantee a recession. Another leading series in the list, the gap between long-term and short-term interest rates, is the subject of a later lesson on the yield curve.
Two companies through one downturn
The 2007–09 recession is the clearest recent test of how the cycle reaches a company. Caterpillar makes construction and mining machinery, goods that customers buy when they are expanding and defer when they are not. Procter & Gamble sells household products that people buy in good times and bad. Their results for the year that included the trough are below. The years are not identical: Caterpillar reports for the calendar year 2009, and Procter & Gamble for the year ended 30 June 2009, which sits inside the recession.
| Caterpillar, calendar 2009 | Procter & Gamble, year to 30 June 2009 | |
|---|---|---|
| Sales | $32.396 billion, down 37% from $51.324 billion | $79.0 billion, down 3% |
| Profit | $895 million, down 75% from $3,557 million | Net earnings $13.4 billion, up 11% |
| Diluted earnings per share | $1.43, down from $5.66 | $4.26, up 17% |
| Other measures given | Earnings per share excluding redundancy costs: $2.18 | Organic sales growth +2%; unit volume −3% |
Caterpillar's sales fell 37% and its profit fell 75%, about 2.0 times as fast as sales. Its chief executive called the year "the worst our company has experienced since the Great Depression," and the release said the recession caused construction spending to decline in many countries and mining companies to reduce output. Procter & Gamble's net sales fell 3%, its organic sales rose 2%, and its profit rose. Its release described one of the most difficult macroeconomic environments in decades, and said unit volume fell 3%. The same recession produced a 37% fall in sales at one company and a 3% fall at the other. Caterpillar's 2009 profit also includes redundancy costs, and excluding them its earnings per share were $2.18 rather than $1.43, so the profit fall is partly a restructuring effect. The size of the drop in Caterpillar's profit compared with its sales is the leverage idea in the operating leverage lesson: when costs do not fall as fast as sales, profit falls faster than revenue. Investors call the first kind of business cyclical and the second defensive, and the difference is a reason to read a company's results against the cycle before judging them.
What this lesson leaves out
- One country's data. The figures are US series. Other countries define and publish GDP, unemployment and recession dates in their own ways, though the ideas carry over.
- The calculator is arithmetic. It shows how the rates are built, using real September 2026 levels. It does not forecast jobs, and the real survey has sampling error, which BLS reports.
- Unemployment is not the whole labour market. The rate misses people who work part-time but want full-time work, and those who want a job but have stopped looking.
- Two recessions are not a theory. They show how different episodes can be. They do not explain what caused either.
- Indicator lists are a snapshot. The Conference Board has changed the components of its indexes over time, and the list used here is the one in its December 2025 notes.
- Companies were chosen to show a contrast. They are not a sample, and one year is not a record.
Common mistakes
Quoting nominal growth as real growth
Nominal GDP can rise on prices alone. Growth is normally a real figure.
Taking a quarterly rate for an annual one
BEA quotes quarterly growth at an annual rate. A 0.55% quarter is not 0.55% growth for the year.
Treating "two quarters of decline" as the definition
It is a rule of thumb. The NBER weighs depth, spread and duration, and uses monthly data.
Reading the unemployment rate on its own
It can fall because people stop looking, and rise because they start. Read it with the participation rate and the employment ratio.
Expecting a leading indicator to be right every time
Each is a signal that has often turned before the cycle. None is a promise.
Try it yourself
1. In September 2026 there were 6,859 thousand unemployed and 170,298 thousand in the labour force. What was the unemployment rate?
Show the working
6,859 ÷ 170,298 = 4.03%. The labour force is the employed plus the unemployed, 163,439 + 6,859 = 170,298.
2. The population aged 16 and over was 275,554 thousand. What was the participation rate, and what was the employment ratio?
Show the working
Participation: 170,298 ÷ 275,554 = 61.80%. Employment ratio: 163,439 ÷ 275,554 = 59.31%.
3. Real GDP was 24,274.383 billion in the first quarter of 2026 and 24,408.011 billion in the second. What was the growth, and what is it at an annual rate?
Show the working
24,408.011 ÷ 24,274.383 − 1 = 0.55% for the quarter. At an annual rate, 1.00554 − 1 = 2.2%.
4. Real GDP was 16,915.191 billion in the fourth quarter of 2007 and 16,269.145 billion in the second quarter of 2009. By what percentage did it fall?
Show the working
16,269.145 ÷ 16,915.191 − 1 = −3.82%. That is the fall from the peak quarter to the trough quarter.
5. How can the unemployment rate fall in a month when the number of people with jobs falls?
Show the working
If enough unemployed people stop looking for work, they leave the labour force. The unemployed count and the labour force both shrink, so the rate falls. In the calculator, employment −400 thousand and 600 thousand leaving gives 3.92%, below the starting 4.03%.
Where this is used
- A company builds a budget by assuming a path for the economy, and asks how its own sales behave when growth slows.
- A lender or analyst judges whether a business is cyclical or defensive before deciding how much debt it can carry.
- A central bank reads growth, jobs and inflation together when it sets interest rates, the subject of the next lessons.
- An investor watches leading indicators for early signs of a turn, knowing none is certain.
- A government or a forecaster compares GDP, jobs and inflation across countries, taking care that definitions differ.
Key terms
- Gross domestic product
- The total value of goods and services produced within a country in a period.
- Real and nominal GDP
- GDP adjusted for price changes, and GDP in current prices.
- Business cycle
- The repeating pattern of expansion, peak, contraction and trough in economic activity.
- Recession
- A significant decline in economic activity, spread across the economy, lasting more than a few months, in the NBER's definition.
- Labour force
- The employed plus the unemployed.
- Unemployment rate
- The unemployed as a share of the labour force.
- Participation rate
- The labour force as a share of the population aged 16 and over.
- Employment ratio
- The employed as a share of the population aged 16 and over.
- Leading, coincident and lagging indicators
- Series that turn before, with and after the economy.
- Cyclical and defensive
- A business whose results swing with the economy, and one whose results hold up better.
Questions people ask
Is a country in recession when GDP falls for two quarters?
Not necessarily by the NBER's definition, though it is a common shorthand. The committee looks at depth, spread and duration, using several monthly series.
Why are GDP figures revised?
The first estimate uses incomplete data. BEA publishes later estimates as more complete data arrive, and updates the history each year.
Does a lower unemployment rate always mean a stronger job market?
No. It can fall because people leave the labour force. The participation rate and the employment ratio show which has happened.
Why do the headline unemployment rate and the table here differ?
The headline rate is seasonally adjusted. The calculator starts from the unadjusted counts, which give 4.03% against the adjusted 4.2%.
Sources and review
- US Bureau of Economic Analysis, The expenditures approach to measuring GDP (The BEA Wire, 3 June 2025), for the definition of GDP, the C + I + G + X − M formula and why imports are subtracted.
- US Bureau of Economic Analysis, Gross domestic product release: additional information (last modified 27 March 2026), for the advance, second and third estimates, the size of revisions from 1996 to 2024 and the use of annual rates for quarterly data.
- US Bureau of Economic Analysis, Real gross domestic product (GDPC1), billions of chained 2017 dollars, seasonally adjusted annual rate, as published by the Federal Reserve Bank of St. Louis (last updated 30 September 2026), for the quarterly values from 2007 to the second quarter of 2026.
- National Bureau of Economic Research, US business cycle expansions and contractions (data last updated 14 March 2023), for the peak and trough months of the December 2007 and February 2020 turning points and the length of each contraction.
- National Bureau of Economic Research, Business cycle dating procedure: frequently asked questions, for the definition of a recession, why the committee does not use the two-quarter rule, and the monthly indicators it examines.
- US Bureau of Labor Statistics, Unemployment rate (UNRATE), percent, seasonally adjusted, as published by the Federal Reserve Bank of St. Louis (last updated 2 October 2026), for the monthly unemployment rates used in the recession table and the seasonally adjusted rate for September 2026.
- US Bureau of Labor Statistics, How the government measures unemployment, for the definitions of employed, unemployed, not in the labour force, the labour force, the unemployment rate and the participation rate, and the sample of about 60,000 households.
- US Bureau of Labor Statistics, The Employment Situation, Table A-1, household data (modified 2 October 2026), for the September 2026 population, labour force, employed, unemployed and not-in-labour-force counts, not seasonally adjusted.
- The Conference Board, US leading economic index technical notes (9 December 2025), for the components of the leading, coincident and lagging indexes and their stated purpose.
- Caterpillar Inc., Fourth-quarter and full-year 2009 results release (Form 8-K, Exhibit 99.1, 27 January 2010), for sales and revenues, profit, profit per share and the company's description of the recession's effects.
- The Procter & Gamble Company, Fiscal year 2009 results release (Form 8-K, 5 August 2009), for net sales, organic sales growth, net earnings, earnings per share, unit volume and its description of the economic environment.
How the numbers were checked: every growth rate, the annual-rate conversion, the unemployment, participation and employment rates and the gaps between the quarters were recomputed in plain code from the figures in the sources above, and the calculator, which runs separately, was compared with that code for every combination of settings it allows. The September 2026 household data are the not seasonally adjusted counts in the BLS table, and the unemployment series is the seasonally adjusted rate. GDP is the series as it stood on its last update on 30 September 2026, and it can be revised.