Macroeconomics: GDP, inflation, unemployment, the business cycle

Level 1 Plain words

For every reader: the idea in plain words, no formulas.

The size of the economy

Gross domestic product, GDP, adds up the value of everything produced in a country in a year. When GDP grows, more is being made and earned; when it shrinks for a while, we call it a recession. Most of US output goes to households buying goods and services, so the spending habits of ordinary people move the whole economy.

Rising prices

Inflation is the rise in the general level of prices. A little inflation is normal and most central banks aim for about 2 % a year. High inflation hurts anyone holding cash or bonds that pay a fixed amount, because each payment buys less. Deflation, falling prices, sounds pleasant but tends to come with weak demand, lost jobs and debts that get harder to repay.

Jobs

The unemployment rate counts people who want a job, have looked for one recently and do not have one, as a share of everyone working or looking. It rises in recessions and falls in expansions, with a lag. For households it is the measure that matters most, since a lost job hurts far more than a dip in a portfolio.

Booms and busts

Economies do not grow in a straight line. They expand for several years, reach a peak, contract in a recession and then recover. That rhythm is the business cycle. The phases have no fixed length: the shortest US recession on record lasted two months, the longest more than three years.

Level 2 How it works

For readers who want the mechanism, with the diagrams.

Beginner path stops here

GDP and its parts

Measured by spending, GDP is the sum of four parts: consumption by households ($C$), investment by businesses in equipment, buildings and inventories plus new housing ($I$), government purchases ($G$) and net exports, which are exports minus imports ($NX$). In 2025 US GDP averaged $30.9 trillion at an annual rate. Consumption made up 67.6 %, investment 18.0 %, government 17.4 %, and net exports −3.0 %, negative because the US imports more than it exports (Bureau of Economic Analysis, via FRED).

Investment is the smallest of the big three and the most volatile. Businesses cut spending on equipment and buildings fast when demand weakens, and housebuilding falls hard when mortgage rates rise. Much of the swing in GDP through a cycle comes from investment, even though consumption is far larger.

Real and nominal

Nominal GDP counts output at current prices. Real GDP takes out price changes using a price index called the GDP deflator. If nominal GDP rises 5.5 % in a year and prices rise 2.5 %, real output grew a little under 3 %. Growth figures in the news are almost always real and annualised.

Measuring inflation

The consumer price index (CPI), from the Bureau of Labor Statistics, tracks the price of a fixed basket of goods and services bought by urban households. The PCE price index, from the Bureau of Economic Analysis, covers a wider range of spending, including what employers and the government pay for medical care on households' behalf, and it updates the basket as people switch between goods. PCE inflation usually runs a few tenths of a point below CPI. The Federal Reserve's 2 % target uses PCE. Core measures leave out food and energy, whose prices jump around; they give a better reading of the trend, though households still pay for food and fuel.

Unemployment and participation

The headline rate, called U-3, counts people with no job who have looked in the past four weeks. U-6 adds people working part time who want full-time work and people who want a job but have stopped looking; it runs several points higher. The participation rate is the share of adults working or looking. When participation falls, the unemployment rate can drop even with no new jobs, because discouraged people stop counting as unemployed.

Okun's law

Arthur Okun noticed in 1962 that output and unemployment move together in a steady ratio: when unemployment rises above its normal level, output falls below its potential by a multiple of that rise. Modern estimates put the multiple near two. The output gap, the percentage difference between actual and potential GDP, is the measure the Taylor rule in Chapter 7 uses.

The Phillips curve and the 1970s

A. W. Phillips showed in 1958 that British wage inflation tended to be high when unemployment was low. Economists read that as a menu: accept a bit more inflation and buy lower unemployment. The 1970s broke the menu. Inflation and unemployment rose together, a mix called stagflation. In May 1975 US unemployment reached 9.0 % while consumer prices were 9.5 % higher than a year before. Milton Friedman and Edmund Phelps had argued that once people expect inflation, it stops buying jobs, and the decade bore them out. The curve still exists in a weaker form: tight labour markets push wages up, all else equal.

The business cycle

The National Bureau of Economic Research, through its Business Cycle Dating Committee, names the month of each peak and trough. It defines a recession as a significant decline in activity spread across the economy and lasting more than a few months, judged by income, jobs, spending and output together. The popular rule of two quarters of falling GDP is a rough guide, and the NBER does not use it. The committee announces its dates months or even a year after the fact.

Leading, coincident and lagging indicators

Indicators fall into three groups by timing. Leading indicators tend to turn before the economy does: building permits, new orders, initial claims for unemployment benefits, the stock market and the slope of the yield curve. Coincident indicators move with it: payroll employment, industrial production, real income and sales. Lagging indicators turn after: the unemployment rate at its peak, the length of unemployment spells, business loans outstanding. Leading indicators give false alarms, so analysts look for several turning together.

Level 3 The math

For Series 65 and CFA readers: formulas and worked examples.

Series 65 path stops here

Real growth from nominal growth and the deflator

In words, one plus real growth equals one plus nominal growth divided by one plus the rise in the price deflator.

With nominal growth of 5.5 % and a deflator up 2.5 %, real growth is 2.93 %; the shortcut of subtracting gives 3.00 %. The gap matters little at low inflation and a lot at high inflation.

Okun's coefficient

Write $u$ for unemployment, $u^*$ for its normal or natural level and $y$ for the output gap in percent.

In words, the output gap is about minus Okun's coefficient times the amount by which unemployment exceeds its natural rate.

What inflation does over time

A price level growing at $\pi$ a year for $n$ years multiplies by $(1 + \pi)^n$, so a fixed dollar amount buys $1/(1+\pi)^n$ of what it bought at the start.

In words, after n years a dollar buys one divided by one plus inflation to the power n of what it bought at first.

Recession arithmetic

The NBER table gives each recession's length from peak month to trough month. Of the 15 contractions since 1929, the longest ran 43 months (1929 to 1933) and the shortest 2 months (2020). The 13 recessions since 1945 lasted 10.2 months on average and none more than 18 months, while the expansions between them averaged 64 months. Since the war the US economy has spent far more time growing than shrinking, which is why a long-horizon stock investor has been paid for sitting through the bad stretches.

Level 4 History, analysed

For readers who want past episodes run through the Level 3 math.

Fifteen recessions since 1929

The table lists every NBER contraction with its length and, where S&P Dow Jones Indices publishes the closes, the stock market's peak-to-trough fall around it. The market's fall often began before the recession and ended before the trough.

Patterns across the recessions

Several regularities show up. Recessions have grown shorter and expansions longer since the Second World War, a change economists link to a larger service sector, automatic stabilisers such as unemployment insurance and more active central banks. Unemployment keeps rising after the trough: it peaked at 10.8 % in November 1982, the trough month of that recession, and at 10.0 % in October 2009, four months after the 2009 trough. Stocks, as a leading indicator, usually bottom before the economy does, so waiting for good economic news to buy has meant buying after the rebound. Deep falls in stocks cluster around recessions that came with a financial crisis (1929 and 2007) or an inflation shock (1973).

1929 to 1933: the Great Contraction

A fall of that size needs a gain of +825 % to recover, and the Dow did not close above its 1929 peak again until 1954. This contraction combined a stock crash, thousands of bank failures, falling prices that raised the real burden of debts, and a Federal Reserve that let the money supply shrink. Those lessons shaped the policy responses to 2008 and 2020.

1973 to 1975: oil and stagflation

Adding the unemployment rate and the twelve-month inflation rate gives the old "misery index", which stood at 18.5 % in May 1975. On the Phillips menu, high unemployment should have brought low inflation. It did not, because the shock hit supply rather than demand: it raised costs and cut output at once. Chapter 5 analyses the oil market side of the same event.

1980 to 1982: a recession made to stop inflation

Most recessions start with a shock or a bust. This pair started with policy. With consumer prices up 14.8 % in the year to March 1980, the Federal Reserve under Paul Volcker held short rates far above inflation until inflation broke. Okun's law gives the cost: unemployment about four points above an assumed normal rate near six percent implies output roughly −8.0 % below potential at the worst. Consumer prices rose 3.8 % in the year to December 1982, and the long expansion of the 1980s followed. Chapter 7 runs the Taylor rule over the same years.

2001: a stock bust with a mild recession

This case shows how loosely stocks and the economy can line up. The recession was one of the mildest on record, yet the market's fall was among the deepest, because it started from very high valuations in technology shares. The bear market began a year before the recession and ended almost a year after the trough. Investors who watched only GDP would have seen little of what happened to their portfolios.

2007 to 2009: the financial crisis

A housing bust turned into a banking crisis when losses on mortgage securities spread through banks and funds that had borrowed heavily against them. Credit stopped flowing, investment fell, and job losses fed back into more defaults. Recoveries after financial crises tend to be slow, and this one was: unemployment did not return to its 2007 level for years. Chapter 17 follows the chain from mortgages to Lehman Brothers.

2020: the shortest recession

This recession broke most patterns. It started with public-health orders instead of a financial imbalance, and fiscal and monetary support on a very large scale arrived within weeks. By August 2020 the S&P 500 had regained its February high. Indicators built on past cycles, such as the yield curve, had little to say about its timing.

The yield curve as a recession signal

A yield curve plots Treasury yields from short maturities to long. Normally long yields sit above short ones. When the ten-year yield falls below the two-year yield, the curve has inverted, which tends to happen when markets expect the Federal Reserve to cut rates in a coming slowdown. Arturo Estrella and Frederic Mishkin showed in 1996 that the slope of the curve forecast US recessions better than most other indicators.

We count each episode in which the 2s10s spread (the ten-year yield minus the two-year, FRED series T10Y2Y) closed below zero after at least four months above it, and the months to the next NBER peak.

Of the 8 episodes, 6 came before a recession, with leads from 6 months to 24 months and an average of 15 months. In 1998 the inversion lasted only about two months and was followed by growth; the 2019 inversion lasted three days, and the recession that followed came from a virus that no bond market foresaw. A long and uncertain lead makes the signal hard to trade: stocks often rose for a year or more after an inversion.

Our last row, 2022, remains an open case. The spread dipped below zero on 1 April 2022, went deeply negative from July and stayed below zero most days until September 2024. By the end of our data, 54 months after the first inverted close, FRED's NBER-based recession indicator showed no recession after April 2020. Either the lead this time is far longer than before, or the signal failed. One explanation offered: large pandemic savings and fixed-rate mortgages shielded households from the rate rises that usually cause the slowdown. Chapter 3 shows with Bayes' rule why even a good signal gives many false alarms when recessions are rare.