Unit 6 · The Business Cycle and Fiscal Policy

Think of the economy’s total output, real GDP, as a line on a chart that climbs over the decades. It does not climb smoothly. It runs up for a few years, stumbles, falls back for a while, then climbs again. That wobble is the business cycle: the repeating swing of economic activity above and below its long-run path. It is called a cycle because the pattern repeats, but the lengths and depths differ every time.

The four phases have names. An expansion is the stretch when real GDP is growing, jobs are being added and businesses are opening. The peak is the high point, the month when growth stops. A contraction is the stretch when real GDP is falling, layoffs rise and stores close. The trough is the low point, the month when the fall stops and growth begins again. Then a new expansion starts. One full cycle runs from one peak to the next.

Put numbers on it with a made-up town. Suppose Riverton’s output was $100 million in year 1, $104 million in year 2 and $107 million in year 3. That is an expansion: output rose every year. In year 4 it was still $107 million, the peak. In years 5 and 6 output fell to $103 million and then $101 million, a contraction. Year 6 was the trough. In year 7 output rose to $105 million, and a new expansion was under way. Notice that the year-7 economy was still smaller than at the peak. Recovery takes time.

A common mistake is to think the economy is in a contraction whenever growth slows. Slower growth is still growth. If Riverton went from $104 million to $105 million, output grew, just less than before. That is a weaker expansion, not a contraction. A contraction means output actually shrinks. Since the mid-1940s, expansions in the United States have usually lasted years while contractions have usually lasted about a year or less, so the line spends most of its time climbing.

People throw the word around, but recession has a working definition. A recession is a significant decline in economic activity that spreads across the whole economy and lasts more than a few months. Three parts matter: it must be deep enough to count, broad enough to hit most industries and regions, and long enough not to be a one-month blip. A bad quarter for one industry, like a strike at a car plant, is not a recession.

You will often hear a simpler rule of thumb: two quarters in a row of falling real GDP. It is a handy shortcut, and it usually points the right way. But it is not the official definition in the United States. The official dates come from the Business Cycle Dating Committee of the National Bureau of Economic Research, a private research group of economists. The committee looks at monthly data on jobs, income, sales and production together, not just GDP. It picks the peak month and the trough month.

Because the committee waits for revised data, it usually announces a recession many months after it began, and sometimes after it has ended. The 2020 recession began in February 2020 and ended in April 2020, only two months, the shortest on record. The committee announced the start in June 2020 and the end in July 2021. That lag frustrates people, but it is the price of being sure. Quick guesses are often wrong.

A common error is to say the recession is over when GDP starts growing again, so things must be fine. The trough marks the end of the decline, not a return to normal. In the recession that ran from December 2007 to June 2009, output stopped falling in mid-2009, but the unemployment rate did not peak until about 10% in late 2009 and stayed high for years. Recovery, meaning getting back to the old peak and beyond, is a separate and slower stage.

Growth is reported as a percent change. If real GDP was $20.0 trillion last year and $20.5 trillion this year, the change is $0.5 trillion. Divide by the starting value: 0.5 ÷ 20.0 = 0.025, or 2.5% growth. If instead GDP fell from $20.0 trillion to $19.4 trillion, the change is minus $0.6 trillion, and 0.6 ÷ 20.0 = 0.03, so growth was minus 3%. A negative growth rate is the number that marks a contraction.

Quarterly numbers in the news are usually annualized. That means the quarter’s change is stretched out to show what a full year at that pace would look like. If output falls 1% in one quarter, the annualized rate is roughly minus 4%. The rule matters when you read headlines. In the spring of 2020 you saw reports of output falling at an annualized rate of about 30%. The actual drop in that single quarter was roughly 9%, which was still by far the sharpest quarterly fall on record.

Jobs tell the same story from a different angle. The unemployment rate is the number of unemployed people divided by the labor force, times 100. Suppose a town’s labor force is 40,000 people. In the expansion, 1,600 are looking for work: 1,600 ÷ 40,000 = 0.04, so 4% unemployment. In the contraction, layoffs push the number to 3,600: 3,600 ÷ 40,000 = 0.09, so 9%. The rate more than doubled even though 90% of the labor force still had jobs. That is why a few percentage points feel like a lot.

Put the pieces together and you can read a downturn like Maya’s downtown. Output falls for two or three quarters. Employers cut hours first, then jobs, so the unemployment rate climbs. Consumers spend less, so sales drop, so more employers cut. The numbers move together because they measure the same thing from different sides. The trick in reading them is to ask two questions of every figure: is it a level or a rate of change, and is it for a month, a quarter or a year?

Maya’s aunt noticed the thin lunch crowd before the news said anything. Economists look for the same kind of early sign in data. A leading indicator is a measure that usually changes direction before the overall economy does. No single one is reliable, so forecasters watch a bundle of them. When most turn down together for several months, a contraction often follows.

Some of the standard leading indicators make sense once you think about them. Building permits: a builder pulls a permit months before the house is finished, so a drop in permits means less construction work ahead. New claims for unemployment insurance: when layoffs start, the first filing happens the same week, long before the monthly unemployment rate moves. Stock prices: investors bet on future profits, so a broad slide can signal trouble ahead. Consumer expectations: when survey respondents say they expect harder times, they often cut spending soon after.

Others are less obvious. New orders for manufactured goods fall before factories cut production. Average weekly hours in manufacturing fall before jobs do, because employers trim overtime first. The gap between long-term and short-term interest rates has also turned negative before most recent U.S. recessions. A private research group, The Conference Board, combines about ten of these into a single Leading Economic Index that is reported each month.

The catch is false alarms. Stock prices in particular have fallen sharply many times when no recession followed. Leading indicators shift the odds; they do not read the future. A careful reader asks how many indicators agree, for how long, and how big the moves are. Three months of small dips in one series is noise. Six months of falling permits, rising claims and gloomy consumers together is a warning worth taking seriously.

If leading indicators are the forecast, coincident indicators are the weather report. They move at about the same time as the overall economy, so they tell you where you are right now. The main ones are the number of workers on payrolls, personal income after government payments are subtracted, industrial production and total sales by manufacturers, wholesalers and retailers. When these fall together for several months, a contraction is happening, whether or not anyone has declared it.

The recession dating committee leans heavily on coincident indicators, which is one reason its dates are trusted. Payroll employment, for example, is measured every month from a survey of about 120,000 businesses and government agencies, so it is a broad and timely picture of the job market. When jobs and income and production all peak in the same month, that month is very likely the peak of the cycle.

A lagging indicator turns after the economy does, often by several months. The unemployment rate is the famous example. Employers wait to be sure a slowdown is real before laying people off, and they wait to be sure a recovery is real before hiring back. So unemployment keeps rising for a while after the trough. Other lagging indicators include the average length of unemployment, business borrowing, the ratio of inventories to sales and the prime interest rate that banks charge their best customers.

Lagging indicators sound useless, but they do two jobs. First, they confirm. If unemployment finally starts to fall, the recovery that leading indicators promised is real. Second, they warn about imbalances. When inventories pile up faster than sales, businesses will soon cut orders. The mistake to avoid is treating a lagging number as news about today. A rising unemployment rate in the month after the trough describes the past few months, not the next few.

In earlier chapters you drew supply and demand for one good, like lemonade. Macroeconomics does the same thing for everything at once. Aggregate demand is the total amount of a country’s output that households, businesses, government and foreign buyers want to purchase at each overall price level. It is the four GDP pieces added up: consumption, investment, government purchases and net exports. When the price level rises, total spending tends to fall a little, so aggregate demand slopes down like an ordinary demand curve.

Aggregate supply is the total amount businesses are willing to produce at each price level. In the short run it slopes up: when prices rise faster than wages and other costs, producing more is profitable, so firms expand output. In the long run, wages and costs catch up, and output settles at the level the economy’s workers, machines and technology can sustain. That long-run level is called potential output, and the business cycle is the swing of actual output around it.

Where the two meet is the economy’s short-run equilibrium: an overall price level and a level of real GDP. Now shift a curve. Suppose households get nervous and cut spending, as in Maya’s downtown when people ordered water instead of soda. Aggregate demand shifts left. Output falls below potential and the price level rises more slowly or even falls. That is a demand-side recession. If instead a war or a bad harvest raises the cost of oil or food, short-run aggregate supply shifts left. Output falls and prices rise at the same time, a painful mix.

The model is a way of organizing questions, not a machine that spits out answers. It cannot tell you exactly how far output will fall. It does tell you which direction to expect and which kind of shock you are facing. A demand shock lowers output and inflation together. A supply shock lowers output but raises inflation. Knowing which one is at work matters later, because the government’s tools work well against one and poorly against the other.

Why does the whole economy turn at once? Start with the spreading. Suppose Riverton’s largest employer, a parts factory, loses a big contract and lays off 500 workers. Those workers cut spending by, say, $3,000 each over the next year, which is $1.5 million less spent at Riverton’s stores. The stores cut hours, and their workers spend less too. One layoff becomes several. Economists call this chain the multiplier effect, and it is why a shock to one industry can pull a whole region down.

Now the shocks themselves. Demand shocks come from a sudden change in spending. Households may turn cautious after a stock market slide or a jump in debt. Businesses may cut investment when they expect weak sales. A trading partner may fall into recession and stop buying exports. Supply shocks come from costs: a spike in oil prices, a drought, a war that cuts off imports, or in 2020 a virus that closed workplaces. Financial shocks come from the banking system: when lenders fail or freeze, even healthy businesses cannot borrow, and spending collapses.

Policy can push either way. A central bank that raises interest rates sharply to fight inflation will slow borrowing and can tip the economy into contraction, as happened in the early 1980s. A government that cuts taxes or boosts spending can add to demand and lengthen an expansion. The point is not that policy is always wrong or always right. It is that policy is one of the forces moving aggregate demand, and it can be timed badly.

Booms have causes too. A new technology, such as the spread of the internet in the late 1990s, can lift investment for years. Easy credit can inflate a housing boom, as it did in the mid-2000s. Rising confidence feeds on itself: more spending means more income means more spending. The trouble is that booms built on borrowing and optimism tend to end when the borrowing stops. Many busts are the unwinding of the boom that came before, which is why economists watch expansions as closely as contractions.

The worst contraction in American history began in the summer of 1929, before the famous stock market crash of October. The crash made it far worse. Stock prices lost most of their value over the next three years, wiping out savings and confidence. Households cut spending. Businesses stopped investing. Between 1929 and 1933, real output fell by roughly a quarter, and the price level fell by about a quarter. Falling prices sound nice until you have a loan: the dollars you owe are fixed while the dollars you earn shrink.

The banking system turned a bad recession into a catastrophe. Banks then had no deposit insurance. When rumors spread that a bank was weak, depositors rushed to pull their money out, and even a healthy bank could not pay everyone at once. Thousands of banks failed in waves between 1930 and 1933. Each failure destroyed depositors’ savings and cut off loans to farms and businesses. The money supply shrank sharply. In March 1933 the new president closed every bank in the country for several days to stop the panic.

By 1933 about one worker in four was unemployed. There was no unemployment insurance and little public relief. Families lost farms and homes. In the following years the government created the Federal Deposit Insurance Corporation to guarantee bank deposits, the Social Security system, and federal unemployment insurance. These programs still exist, and they are part of why later recessions, though painful, have not repeated the 1930s.

Economists still debate the causes, but most agree on two lessons. First, a collapse in aggregate demand can feed on itself for years if nothing stops it. Second, the banking system is the economy’s plumbing, and when it fails, everything downstream fails with it. The Great Depression did not fully end until massive government spending during the Second World War pushed demand far above where it had been. That experience shaped how governments and central banks have responded to every downturn since.

In the early and mid-2000s, U.S. house prices rose year after year, and lenders grew careless. Mortgages went to borrowers with little income or savings, often at low starting rates that would jump later. The loans were bundled into securities and sold to banks and investors around the world, who believed house prices would keep rising. Home building boomed. Construction employed millions. Families borrowed against their rising home values to spend.

House prices peaked in 2006 and began to fall. Borrowers who owed more than their homes were worth stopped paying. The securities built on those mortgages lost value, and no one knew exactly which banks held the losses. Lenders stopped trusting each other. In September 2008 a major investment bank, Lehman Brothers, failed, and credit markets froze. Businesses that had nothing to do with housing suddenly could not borrow to meet payroll. The financial crisis had become a recession for everyone.

The recession officially ran from December 2007 to June 2009, eighteen months, the longest since the Great Depression. Real GDP fell about 4% from peak to trough. About 8.7 million jobs were lost. The unemployment rate climbed from under 5% to about 10% in late 2009, and it did not return to 5% until 2015. Millions of families lost homes to foreclosure. Illinois was hit hard: its unemployment rate rose above the national rate and stayed there for years.

The response was large. The Federal Reserve cut its interest rate target almost to zero and lent heavily to banks. Congress passed a $700 billion program to shore up the financial system in late 2008 and a stimulus package of roughly $800 billion in early 2009. Economists still argue over what worked best, but there is broad agreement on one comparison: the 1930s show what happens when a financial collapse is allowed to run, and 2008 shows a collapse that was, at great cost, contained.

Every recession in this chapter so far took years to build. The 2020 recession took weeks. In March 2020 the spread of COVID-19 led governments, businesses and households to shut down travel, restaurants, schools, theaters and many workplaces almost at once. It was a supply shock, because workers could not safely produce, and a demand shock, because customers stopped buying, hitting in the same month.

The numbers were unlike anything in the modern record. More than 20 million jobs disappeared in April 2020 alone. The unemployment rate jumped from about 3.5% in February to nearly 15% in April, the highest since the 1930s. Real output fell by roughly 9% in the spring quarter, which is why headlines quoted an annualized drop of around 30%. Restaurants and hotels were hit hardest; many office workers kept their jobs and worked from home.

The peak came in February 2020 and the trough in April 2020. Two months made it the shortest recession in the recorded history of U.S. business cycles, which goes back to the 1850s. The turn came fast because the shock was not a slow unwinding of debt but a sudden stop that could partly be reversed as businesses reopened. It also came fast because the response was enormous. Congress passed relief measured in trillions of dollars, including direct payments to households and expanded unemployment benefits, and the Federal Reserve cut rates to near zero and lent across the economy.

The recovery was uneven and it brought its own problem. By 2022 the unemployment rate was back below 4%, but the combination of strong demand, snarled supply chains and higher energy prices pushed inflation to its highest level in about forty years. That gave students a live lesson from the last section: a supply shock plus a demand boost produces rising prices. The 2020 episode is now the standard example of how different two recessions can be, and how a recovery can carry the seeds of the next problem.

Step back from the cycle. If you drew U.S. real GDP per person over the last century, the recessions would look like small dents in a line that rises steadily. Real output per person today is several times what it was in the 1920s. That climb is economic growth, and it is what turns a country from poor to rich. The cycle decides whether this year is good or bad. Growth decides whether your grandchildren live better than you.

The engine of growth is productivity: how much output a worker produces in an hour. Suppose a bakery with five workers makes 500 loaves in an eight-hour day. That is 500 ÷ 40 worker-hours, or 12.5 loaves per hour. The owner buys a second oven and a dough mixer. Now the same five workers make 800 loaves a day: 800 ÷ 40 = 20 loaves per hour. Productivity rose 60%. The bakery can pay higher wages, cut prices or earn more profit, and the country as a whole has more bread from the same labor.

The oven and the mixer are physical capital: tools, machines, buildings and infrastructure that help people produce. Adding capital is the most direct way to raise productivity. But there is a limit. A third oven helps less than the second, and a tenth might sit idle. Economists call this diminishing returns. A country cannot grow forever just by piling up more of the same machines. Something has to make the machines themselves better, which is the next lesson.

Small differences in growth rates become enormous over time. A useful shortcut is the rule of 70: divide 70 by the yearly growth rate to get the number of years it takes output to double. At 1% growth, output doubles in about 70 years. At 2%, in about 35 years. At 3.5%, in about 20 years. Over a working lifetime of 40 years, a country growing at 2% ends up about 2.2 times richer, while one growing at 1% ends up about 1.5 times richer. That gap is the difference between a struggling economy and a thriving one.

The bakery bought a second oven and productivity rose. Now suppose the owner sends a worker to a course on bread chemistry, and she comes back knowing how to cut waste and speed up proofing. Output rises again, with no new machine. Her knowledge and skill are human capital: what workers know and can do. Education, training and experience build it. Countries with more schooling per worker tend to have much higher output per worker, and the gap has grown as work has become more skilled.

Even skilled workers with good tools hit a ceiling eventually. What lifts the ceiling is technology: new knowledge about how to produce. Technology is not just gadgets. The assembly line, the shipping container, hybrid corn on Illinois farms, the barcode and the spreadsheet were all technologies that let the same workers and machines produce far more. Illinois corn yields today are several times what they were in the 1930s, on the same land, mostly because of better seeds, machinery and farming methods.

Put the sources together and you have the standard recipe for long-run growth: more physical capital, more human capital and better technology, working inside institutions that protect property, enforce contracts and let people keep the rewards of their effort. A country that invests in schools and research, saves and builds, and keeps its rules stable will grow. One that neglects any of these tends to stall, no matter how many machines it imports.

This is why economists draw a sharp line between the cycle and the trend. A stimulus program can shorten a recession, but it does not make a country permanently richer. A new vaccine, a better school or a faster computer chip does. Over a single year the cycle dominates the news. Over fifty years the trend dominates everything: it determines the wages people earn, the goods they can afford and the choices open to their children. Both matter. Do not confuse them.

When a city raises its bus fare or a state builds a new highway, that is a budget decision. When the national government changes taxes or spending in order to steer the whole economy, economists call it fiscal policy. The word fiscal comes from the Latin word for a money basket or treasury. Fiscal policy has two tools. Government spending includes purchases like roads and payments like unemployment benefits. A tax takes income out of people’s and businesses’ hands.

In the United States, fiscal policy is made by Congress and the President together, because both taxes and spending require a law. The Constitution gives Congress the power to tax and says no money can leave the Treasury unless a law allows it. Tax bills must start in the House of Representatives. The President proposes a budget each year and can sign or veto the bills Congress passes. The Federal Reserve does not make fiscal policy. It makes monetary policy, which works through interest rates and the money supply.

Follow one decision. Suppose Congress passes a law to spend $50 billion repairing bridges over two years, and the President signs it. The Treasury pays construction companies. Those companies hire workers and buy steel and concrete. That spending adds directly to the G in the GDP formula, C + I + G + NX. If Congress instead cut income taxes by $50 billion, nothing would be added to G. The effect would come only if households used their larger paychecks to spend more, raising C.

A common mix-up is to call every government money decision fiscal policy. A school board’s budget matters for its district, but economists use the term mainly for national choices about total taxes and spending. Another mix-up is to say the Fed spends money on programs. It does not. When you hear taxes, spending, deficits or stimulus checks, think Congress and the President. When you hear the interest rate target, think the Fed.

Go back to the aggregate demand model from the last chapter. In a recession, total spending falls short of what the economy can produce, and output sits below potential. Expansionary fiscal policy tries to fill that gap by adding demand: the government spends more, cuts taxes, or both. More government purchases add to G directly. Lower taxes leave households more income to spend, raising C, and can encourage businesses to invest, raising I. Either way, aggregate demand shifts right.

Put numbers on it with a made-up economy. Suppose potential output is $1,000 billion but actual output has fallen to $960 billion. The $40 billion shortfall is called a recessionary gap. Congress does not need to spend $40 billion to close it, because of the multiplier you will meet in the next section. If each dollar of new spending raised output by about $2 in this economy, roughly $20 billion of new spending could close the gap. In the real world no one knows the exact number, which makes sizing a stimulus hard.

The opposite problem is an economy running too hot: output above potential, workers scarce, prices rising fast. Contractionary fiscal policy cuts government spending or raises taxes to slow demand and ease inflation. It shifts aggregate demand left. It is used far less often than expansionary policy, for a plain reason: voters rarely reward politicians for raising taxes or cutting programs. In practice, fighting inflation is usually left to the Federal Reserve.

Watch for the mistake of judging a policy by its label. A tax cut during a boom is expansionary even if it is sold as tax reform, and it can add to inflation. A spending cut during a recession is contractionary even if it is sold as responsible budgeting, and it can deepen the slump. The question to ask is always the same: does this change add to total spending or take away from it, and is that what the economy needs right now?

Some fiscal policy happens with no vote at all. When a factory lays off workers, many of them file for unemployment insurance, a program run by the states with federal backing that replaces part of a lost paycheck for a limited time. Suppose a worker earned $1,000 a week and her benefit is $450 a week (an example; each state sets its own formula). Her spending falls, but by much less than if her income had dropped to zero. Across thousands of laid-off workers, total spending falls less.

Taxes do the same job from the other side. Under a progressive tax, the share of income paid in tax rises as income rises. Take a made-up tax: 10% on the first $20,000 of income and 20% on everything above. A worker earning $60,000 pays $2,000 plus $8,000, or $10,000, and keeps $50,000. If a recession cuts her pay to $40,000, she pays $2,000 plus $4,000, or $6,000, and keeps $34,000. Her pay fell $20,000, but her take-home income fell only $16,000. The tax system absorbed $4,000 of the blow.

These are called automatic stabilizers because they switch on by themselves as the economy changes. In a recession, tax collections fall and benefit payments rise, so the government’s deficit grows. In a boom, more people pay more taxes and fewer claim benefits, so the deficit shrinks. Food assistance through SNAP works the same way. None of this requires Congress to notice the recession, agree on a plan or pass a bill.

Automatic stabilizers soften the cycle, but they do not end it. They are built to cushion a downturn, not reverse it. Their great advantage is speed: they act in the same month income falls. A common mistake is to read a rising deficit in a recession as proof that Congress went on a spending spree. Much of the increase is usually automatic: fewer paychecks to tax, and more people claiming help they qualify for under laws already on the books.

Suppose the government pays a contractor $100 to fix a stretch of sidewalk. The contractor now has $100 more income. Say she spends $80 of it at a hardware store and saves $20. The hardware store owner now has $80 more income and spends 80% of it, or $64, at a diner. The diner owner spends 80% of $64, which is $51.20. The first $100 set off a chain of spending, each round smaller than the last. This is the multiplier at work.

The share of each extra dollar of income that people spend is the marginal propensity to consume, or MPC. In this example it is 0.8. The share they save is the marginal propensity to save, here 0.2. Add up all the rounds, $100 + $80 + $64 + $51.20 and so on, and the total comes to $500. There is a shortcut: the simple multiplier equals 1 ÷ (1 − MPC). With an MPC of 0.8, that is 1 ÷ 0.2 = 5. Each $1 of new spending raises total income by $5 in this simple model.

The size of the multiplier depends on how much people spend. If the MPC were 0.5, the multiplier would be 1 ÷ 0.5 = 2, and $100 of new spending would raise total income by $200. If the MPC were 0.9, the multiplier would be 1 ÷ 0.1 = 10. Higher spending shares mean longer, stronger chains. That is why aid aimed at people likely to spend it right away, such as families that have lost income, tends to have a bigger effect than aid to people who will mostly save it.

The multiplier also runs in reverse. If the government cuts $100 of spending, or a factory closes, the chain runs downward. Remember that the multiplier is a model, not a measured law. Real economies leak spending into savings, taxes and imports at every round, so the real effect is much smaller than 5. Economists’ estimates for real spending programs vary widely, often between about 0.5 and 2, depending on the program and how deep the slump is. The next lesson explains why.

Compare two plans that each cost $100 billion. Plan A: the government buys $100 billion of road work and equipment. The whole $100 billion is spent in the first round. Plan B: the government cuts taxes by $100 billion. With an MPC of 0.8, households spend $80 billion and save $20 billion in the first round. Plan B starts the chain with $80 billion instead of $100 billion. In the simple model, total income rises by $400 billion under Plan B instead of $500 billion under Plan A.

A transfer payment, such as a stimulus check, a Social Security benefit or an unemployment check, works like a tax cut. The government gives money to households, and they decide how much to spend. That is why it matters who receives the money. Studies of the 2008 and 2020 payments found that households with little savings spent more of their checks quickly, while households with plenty of savings spent less. A check to a family behind on rent moves more spending than one to a family that banks it.

There is another leak. When the government spends more than it collects, it borrows by selling a bond, a promise to repay with interest. That borrowing competes with businesses and families for the same pool of savings. If it pushes interest rates up, some businesses skip a new factory and some families skip a car loan. Economists call this crowding out. It is likely to be large when the economy is near full employment, and small in a deep recession, when savings sit idle and rates are already low.

Put the leaks together and you see why economists argue about stimulus. One side says spending in a deep recession has a multiplier well above 1, because idle workers and machines are put back to use. The other side says a large share leaks into savings, imports and higher interest rates, and that households expecting future taxes may save more now. Both sides agree on the direction of these effects. They disagree about their size, which is exactly what the evidence has trouble pinning down.

Suppose a recession starts in March. When does fiscal policy arrive? First comes the recognition lag, the time it takes to see that a recession has started. Data arrive late and are revised, so this can take several months. Next comes the decision lag: Congress must agree on a plan, both chambers must pass it, and the President must sign it. Last comes the implementation lag, the time between signing a law and the money actually being spent.

Each lag depends on the tool. Payments sent through the tax system can reach bank accounts within weeks. The first 2020 payments began arriving in April, less than a month after the law was signed on March 27. A bridge project is slower: engineers need plans, contractors need bids, and work may not start for a year or more. Economists sometimes say good stimulus is timely, targeted and temporary. It arrives while the economy is weak, goes to people likely to spend it, and ends when the slump does.

Lags create a real danger. If a stimulus meant for a recession arrives after the recovery is under way, it adds to demand just as the economy is heating up, which can push up inflation. That is one reason many economists like automatic stabilizers, which skip the recognition and decision lags. It is also why some prefer to leave most short-run steering to the Federal Reserve, which can change its interest rate target at a single meeting.

Politics shapes every step. Tax cuts and new programs are popular; tax increases and program cuts are not. That creates a tilt: stimulus is easier to pass in a slump than restraint is in a boom. Programs meant to be temporary often become permanent. Lawmakers also bargain over where the money goes, so a package can include items that have more to do with home districts than with the recession. None of this makes fiscal policy useless. It means the textbook version and the real version rarely match.

Look at a pay stub. Suppose a worker earns $800 in a week. Some is withheld for federal income tax, some for Social Security and Medicare, and perhaps some for state income tax. Those lines are the government’s main sources of money. The individual income tax, allowed by the Sixteenth Amendment in 1913, is the largest single source of federal revenue, about half of the total in recent years. It is progressive: rates rise in steps as income rises.

The second big source is the payroll tax, which pays for Social Security and Medicare. The worker pays 6.2% of wages for Social Security and 1.45% for Medicare, and the employer pays the same amounts again. On that $800 week, the worker’s share is $49.60 for Social Security and $11.60 for Medicare, $61.20 in all, and the employer pays another $61.20. Payroll taxes bring in roughly a third of federal revenue. The Social Security part stops at a yearly earnings cap, so it takes a smaller share of very high incomes.

The corporate income tax, a tax on business profits, brings in around a tenth of federal revenue. The rest comes from several smaller sources. An excise tax is a tax on one specific good, such as gasoline, alcohol or tobacco. Customs duties on imports, estate taxes and fees add more. The federal government has no national sales tax and no national property tax. Those are state and local tools.

A frequent mistake is to think corporations pay most federal taxes, or that income tax is the only tax most workers pay. Many workers owe little or no federal income tax after credits and deductions, but almost every worker pays payroll tax from the first dollar earned. Revenue also swings with the business cycle. In a recession, incomes and profits fall, so tax collections fall even if no tax rate changes. That is the automatic stabilizer from earlier in this chapter.

Picture the federal budget as a household’s list of bills. The biggest single item is Social Security, which sends monthly checks to retired and disabled workers and their families. It takes about a fifth of all federal spending. Next come the major health programs, Medicare for people 65 and older and Medicaid for people with low incomes, which with other health programs take roughly a quarter. National defense takes a bit over a tenth. Interest on the national debt now takes more than a tenth as well.

Economists sort spending into two kinds. Mandatory spending is set by laws already on the books. Anyone who qualifies for Social Security, Medicare or food assistance gets it, and the amount spent depends on how many people qualify, not on a yearly vote. Discretionary spending is decided each year when Congress passes appropriations bills. It covers defense, highways, national parks, scientific research, education grants and the running costs of federal agencies. Discretionary spending is only about a quarter of the total.

Here is what that means for a real choice. Suppose total federal spending is $6.8 trillion, and a lawmaker wants to cut it by $340 billion, which is 5%. If she cannot touch mandatory programs or interest, she has to find all $340 billion in the discretionary quarter, which is about $1.7 trillion. That is a 20% cut to everything from the military to the weather service. This arithmetic is why budget debates keep returning to Social Security and Medicare, even though changing them is politically hard.

A common belief is that foreign aid is a large share of the budget. In fact it is around 1%. Another is that most spending goes to federal workers’ salaries. Most actually goes out as payments to or for people, such as Social Security checks and Medicare bills. The biggest long-run trend is aging. As more Americans retire, spending on Social Security and Medicare grows faster than the economy under current law, which is one reason forecasters expect large deficits to continue.

Suppose a student earns $400 a month and spends $450. The $50 gap, covered with a credit card, is her monthly deficit. After a year of this, she owes $600, and that total is her debt. The federal government works the same way. A budget deficit is the amount by which spending exceeds revenue in one year. A budget surplus is the reverse. The national debt is the total the government owes: the sum of all past deficits minus all past surpluses.

Now use rounded real numbers. In fiscal year 2024, which ran from October 2023 through September 2024, the federal government spent about $6.75 trillion and collected about $4.9 trillion. The deficit was about $1.8 trillion. The government covered the gap the way it always does: the Treasury sold Treasury securities, bonds and bills that promise to repay the money with interest. Each year’s deficit adds to the debt, which passed $35 trillion in 2024.

Who holds all that debt? Part is owed by one part of the government to another, such as the Social Security trust funds. The rest, called debt held by the public, is owned by American individuals, pension funds, banks, mutual funds, the Federal Reserve, and foreign investors and governments. Foreign holders own roughly 30% of the debt held by the public. When you buy a savings bond, or your pension fund buys Treasury bonds, you become one of the government’s lenders.

The best way to judge debt is against the size of the economy. Suppose a country’s debt is $1,000 billion and its GDP is $1,000 billion, a debt-to-GDP ratio of 100%. Next year it runs a $50 billion deficit and its GDP grows 5%. Debt becomes $1,050 billion and GDP becomes $1,050 billion, so the ratio stays at 100%. A growing economy can carry a growing debt. The trouble starts when debt grows faster than GDP year after year. And remember: cutting the deficit does not cut the debt. It only slows its growth.

Families borrow for sensible reasons: a mortgage on a house they will live in for thirty years, or a student loan that raises their earnings. Governments do too. Borrowing to fight a recession or a war, or for public investment in roads, schools and research that pay off for decades, can make sense, because the benefits come later and are shared by future taxpayers. Most economists agree on that much. The argument is about the size of the debt, what it pays for and when to bring it down.

Those worried about high debt make several points. The government’s net interest is a growing bill. If debt held by the public is $28 trillion and the average interest rate is 3%, interest costs $840 billion a year, money that cannot go to schools, defense or tax cuts. Heavy borrowing can crowd out private investment and slow long-run growth. A country with high debt has less room to borrow in the next emergency. And the bill passes to younger generations, who will pay the taxes to service it.

Others argue the danger is often overstated. The United States borrows in its own currency, and investors around the world treat Treasury securities as among the safest assets, so it can borrow at lower rates than most borrowers. When interest rates stay below the economy’s growth rate, the debt-to-GDP ratio can hold steady even with modest deficits. Cutting spending or raising taxes sharply in a weak economy can cause a recession, which shrinks revenue and can make the debt ratio worse. On this view, the priority is spending that raises future growth.

Economists across the range agree on a few things. A sustainable debt is one that does not grow faster than the economy over time; a debt that grows faster forever cannot last. The ways to slow it are the same for everyone: raise revenue, reduce spending, grow faster, or some mix. Each has a cost and a group that pays it. Where people land depends partly on evidence, such as how large crowding out really is, and partly on values, such as how big government should be. A good argument names its trade-offs.

The federal government can run a deficit for years. States mostly cannot. Almost every state has a balanced budget requirement in its constitution or laws: planned spending in its main operating budget cannot exceed expected revenue. States can still borrow by selling bonds for long-term projects like highways and buildings, which are kept in a separate capital budget. But the everyday budget for schools, prisons, health care and state workers has to balance.

See what that does in a recession. Suppose a state expects $50 billion in revenue and plans $50 billion in spending. A recession hits, incomes and sales fall, and revenue comes in at $46 billion, while more residents qualify for Medicaid. To balance, the state must close a $4 billion gap, maybe more, by cutting spending, raising taxes or both. Those moves take money out of the state’s economy exactly when it is weakest. Economists call this procyclical policy: it moves with the cycle and deepens it.

States soften the problem with a rainy day fund, savings set aside in good years to draw on in bad ones. A state that saved $3 billion in the boom can cover most of that $4 billion gap without deep cuts. The federal government also sends extra aid to states in major recessions, as it did in 2009 and 2020, partly so that state cuts do not cancel out federal stimulus. Local governments, such as cities, counties and school districts, rely heavily on the property tax, which changes more slowly than income or sales taxes.

A common mistake is to assume the federal government should balance its budget every year the way states do. The roles are different. The federal government borrows in dollars, the currency the nation’s own central bank issues, and it has the largest tax base in the country. A state cannot create money, and it competes with other states for residents and businesses, which is part of why its rules are stricter. Each rule brings its own trade-off.

Illinois’s state budget runs on a fiscal year that begins July 1. Its largest sources of money are the individual income tax, the sales tax and federal funds, especially for Medicaid. Illinois’s income tax is a flat tax: every individual taxpayer pays the same rate, 4.95% since 2017, on income after a small exemption. The state constitution requires a single rate. In 2020, voters rejected an amendment that would have allowed higher rates on higher incomes.

Work an example. A worker with $50,000 of taxable income pays 4.95% of $50,000, which is $2,475. A worker with $100,000 pays $4,950. Both pay the same percentage, so on that income the tax works as a proportional tax. The federal income tax, by contrast, is progressive. Now add the sales tax. The state rate on most goods is 6.25%, and local add-ons push the total above 10% in Chicago. Lower-income families spend a larger share of their income on taxed goods, which makes the sales tax a regressive tax.

The biggest long-run pressure is pensions, the retirement payments promised to teachers, state workers, university employees, judges and lawmakers. For decades Illinois put in less money than its pension systems needed, and the shortfall grew. Today the gap between what the systems owe and what they hold is well over $100 billion. A large and rising share of each year’s budget goes to pension payments instead of schools, roads or services. Deciding how to close that gap, and who bears the cost, is one of the state’s hardest fights.

Illinois also shows what happens when budget politics break down. From 2015 to 2017 the governor and the legislature could not agree, and the state went about two years without a full budget. Unpaid bills piled up past $15 billion, universities and social service agencies cut programs, and the state’s credit rating fell to one step above junk. A 2017 deal raised the income tax rate to 4.95%. The state’s finances and ratings have improved since, but the episode shows that budgets are never just arithmetic.