Milton Friedman
MV = PY: inflation is made by money
friedman-quantity-theoryfriedman-monetary-phenomenonfriedman-price-controlsMoney growth vs inflation, by episode (log scales, compressed above 100k%)
click a dotMV = PY, as growth rates
(1+P) = (1+M)(1+V)/(1+Y) · bar lengths on a log scale
Try: start at the k-% rule (money grows about as fast as output, so prices are flat). Raise velocity alone: you get a one-off burst, but velocity cannot rise 20% every year forever. Now drag money growth up through the 1970s and Brazil 1993: the blue dot slides up the 45° line, right where the historical episodes sit. Across four orders of magnitude, sustained inflation tracks money growth.
Brazil 1980–2006: six plans, one that worked
annual CPI, log scaleTry: click each plan. The freezes attacked the price level directly while money kept growing, and inflation returned within a year. Friedman's point: a price control treats the thermometer, not the fever. (The Real's designers stressed inertial indexation too; the URV broke that inertia, while fiscal and monetary restraint kept inflation from coming back.)
The Great Contraction, 1929–33: the Fed that stood aside
friedman-great-contractionfriedman-lender-of-last-resort360 banks
140 failedMoney, prices and output (Aug 1929 = 100)
M = H · (1+c) / (r+c)
Runs push c up (people hold cash) and r up (banks hoard reserves). Both shrink the multiplier. H rose a little; M fell anyway.
Try: play the history and watch the waves of red in late 1930, autumn 1931 and early 1933; the money stock ends about a third lower. Then switch to “lender of last resort”: the same weak banks fail, but runs on sound banks are met with cash, so the panics never spread. Add open-market purchases and money keeps growing. Same 1929 crash, an ordinary recession.
The natural rate: why printing money stops buying jobs
friedman-natural-ratefriedman-expectations-phillipsfriedman-stagflationUnemployment vs inflation
1959Year by year
Try: play a decade with “hold u below natural” at 4%. Each year the short-run curve shifts up and inflation ratchets higher: the accelerationist result. Fire an oil shock, then switch to “accept natural rate”: high inflation and natural-rate unemployment together, which the old Phillips curve said could not happen. Finally “disinflate”: a painful recession, the Volcker years, until expectations follow inflation down. Raise λ and everything happens faster.
Fine-tuner (discretion)
k-% rule (money +5%/yr)
Nominal spending growth, % per year
quarter 80When money acts
Share of a money change that reaches spending each quarter afterwards. Friedman estimated six months to two years, varying from episode to episode.
Try: with a 5-quarter lag, push the reaction past 1: the fine-tuner turns the hot tap all the way, nothing happens, turns it more, then gets scalded and swings to freezing. Drop the lag to 1 and the delay to 0: now gentle discretion beats the rule. The case for a k-percent rule rests on lags being long and unpredictable, and on policymakers not knowing the economy well enough to time their moves.
Permanent income: why a one-off check is mostly saved
friedman-permanent-incomeIncome and consumption, $k per year
Wealth of the permanent-income consumer, $k
Base earnings $50k a year, 40-year horizon, zero interest, for clarity.
Try: a one-off +$20k bonus barely moves the blue line, while the red Keynesian line jumps by $15k. Switch to a permanent raise and both lines move fully. That is why temporary tax rebates (US 1975, 2001, 2008) stimulate less spending than a simple consumption function predicts. Then set a −$20k one-off with “can't borrow” on: credit constraints are the main reason measured MPCs out of windfalls are often higher than pure theory says.
Four ways to spend money, and programs that never end
friedman-four-waysfriedman-free-lunchfriedman-temporary-programsThe 2 × 2 (average of 300 purchases)
rows: whose money · cols: for whomNet = value to the person served minus what was spent. Best achievable on average: 19.
One purchase: the menu, and what each spender picks
Try: with default settings, only the top-left spender reliably lands on the item with the best value for money. The expense-account luncher buys the most expensive thing; the bureaucrat buys something pricey that may not suit the recipient at all. Raise “care about someone else's money” to 1 and “know the recipient” to 1: the gap closes. Friedman's argument is that institutions, not virtue, set those two sliders.
“Nothing is so permanent as a temporary government program”
A program created for an emergency builds a constituency that outlasts the emergency. Each one below must win a renewal vote at every sunset date.
12 temporary programs, created in one emergency (budget, log scale)
Dotted lines are renewal votes. ✕ = program actually ended.
Try: lower “concentrated benefits” toward 0 and most programs die at their first vote. Push it up and they survive and compound. Short sunsets help, because they force a vote before the constituency has grown, which is why public-choice economists favour them.
The negative income tax: a floor that never punishes work
friedman-negative-income-taxfriedman-poverty-trapIncome after taxes and benefits vs earnings
Effective marginal tax rate: share of the next dollar earned that is lost
The NIT trade-off
Guarantee = rate × break-even. A generous floor with a gentle phase-out reaches far up the income scale and costs more. You can pick any two of the three; not all three.
Try: with 100% withdrawal, the red line is flat until the benefit runs out: someone earning $0 and someone earning $11k take home the same. That is the poverty trap. Turn on the cliff and earning more past 26k actually lowers income. The blue NIT line always slopes up, so work always pays. Now lower the NIT rate to 25% and watch the break-even and the cost climb.
School vouchers: money follows the student
friedman-school-vouchersAssigned by address
8 schoolsVouchers: money follows the student
8 schoolsAverage quality of the school a child attends
Try: play 30 years. Under assignment, a bad school keeps its pupils whatever it does, so quality just drifts. Under vouchers, families leave weak schools, which must improve or close, and entrants copy what works. Raise “mind distance”: in a sparse or rural town choice is weaker, and so is the pressure. (The model assumes families can judge quality; that assumption is where most of the real debate lies.)
- takeawaySustained inflation comes from sustained money growth, and depressions can come from money collapsing. Whoever controls money is responsible for both, so central banks should follow stable, predictable rules rather than discretion.
- takeawayExpectations limit what policy can do. Unemployment can be held below its natural rate only by inflation that keeps surprising people, and households spend on their long-run income, so short-lived stimulus achieves less than it seems to.
- takeawayIncentives depend on whose money is spent and on whom. Friedman's policy proposals (the negative income tax, vouchers, sunset clauses) all try to put the choice, and the cost, back with the person affected.
Key concepts · 20
Milton Friedman- Quantity theory of money (MV = PY)
Money times its velocity equals the price level times real output, so in growth rates inflation ≈ money growth + velocity growth − output growth.
Friedman restated the old identity as a theory of the demand for money: because people's desired money holdings are fairly stable, sustained changes in the money stock show up in nominal income, and in the long run in prices.
Studies in the Quantity Theory of Money (1956)
US 1970s: ~10% money growth, ~7% inflationSwitzerland: steady money, steady prices↑ see it in the visualization - "Always and everywhere a monetary phenomenon"
Sustained inflation happens only when the quantity of money grows faster than output.
Oil shocks, unions or greedy firms can raise some prices or the price level once, but cannot keep it rising year after year unless money accommodates them. That puts responsibility for inflation on whoever controls the money supply. Critics note money demand became unstable after the 1980s, weakening the short-run link.
Money Mischief (1992); first stated in the 1963 lecture "Inflation: Causes and Consequences"
Weimar Germany 1923Zimbabwe 2008Venezuela 2018Brazil 1993: ~2,500% a year↑ see it in the visualization - Price controls do not cure inflation
Freezing prices while money keeps growing suppresses the symptom, producing shortages, and inflation returns when the freeze ends.
Friedman opposed Nixon's 1971 wage and price controls for exactly this reason. Brazil's repeated freezes in 1986–91 fit the pattern; the 1994 Plano Real worked once fiscal and monetary restraint backed it.
Newsweek columns (1966–84); Free to Choose (1980), ch. 9
Nixon controls 1971–74Plano Cruzado 1986Plano Collor 1990Plano Real 1994↑ see it in the visualization - The Great Contraction
From 1929 to 1933 the US money stock fell by about a third as waves of bank runs shrank deposits, and nominal income collapsed with it.
Friedman and Schwartz argued the Depression was not proof that markets are unstable, but of a monetary failure: the Fed had the tools to stop the decline and did not use them. Ben Bernanke, as Fed governor in 2002, told Friedman: "You're right, we did it. We're very sorry."
A Monetary History of the United States, 1867–1960 (1963, with Anna Schwartz), ch. 7
Bank of United States failure, Dec 1930Fed discount-rate hike, Oct 1931Bank holiday, Mar 1933↑ see it in the visualization - Lender of last resort
A central bank that lends freely to solvent but illiquid banks in a panic, so runs cannot spread from weak banks to sound ones.
In the Friedman–Schwartz account, runs raised the public's currency/deposit ratio and banks' reserve ratio, so the money multiplier collapsed while high-powered money barely moved. Supplying reserves would have offset it. The 2008 Fed response was explicitly shaped by this lesson.
A Monetary History of the United States (1963)
Fed passivity 1930–33Fed liquidity facilities 2008Walter Bagehot, Lombard Street (1873)↑ see it in the visualization - Natural rate of unemployment
The unemployment rate the economy settles at when expectations are correct, set by real factors such as job search, skills mismatch and labor-market rules.
Monetary policy can push unemployment below it only by fooling people, and only temporarily. To lower the natural rate itself you need real reforms, not more money.
"The Role of Monetary Policy", AEA presidential address (1968)
↑ see it in the visualization - Expectations-augmented Phillips curve
The trade-off between inflation and unemployment exists only for inflation people did not expect; there is no lasting trade-off.
As people come to expect the higher inflation, wage demands rise, the short-run curve shifts up, and holding unemployment down requires ever faster inflation (the accelerationist result). Friedman and Edmund Phelps predicted this before the 1970s confirmed it.
"The Role of Monetary Policy" (1968); Nobel lecture "Inflation and Unemployment" (1976)
US 1965–1980UK 1970s↑ see it in the visualization - Stagflation
High inflation and high unemployment at the same time.
The original Phillips curve said this could not happen. With expectations built in it is natural: once inflation is expected, it persists even at the natural rate, and supply shocks or disinflation add unemployment on top. Ending it took the painful Volcker disinflation of 1979–82.
Nobel lecture "Inflation and Unemployment" (1976)
US 1974–75 and 1979–80Volcker disinflation 1979–82↑ see it in the visualization - Long and variable lags
Monetary policy affects spending only after a delay that is long (roughly 6 to 18 months or more) and that varies unpredictably.
A policymaker reacting to current data is like a fool in the shower: by the time the hot water arrives he has turned the tap too far, and he scalds himself. Activist fine-tuning can amplify the cycle it is trying to smooth.
A Program for Monetary Stability (1960); "The Lag in Effect of Monetary Policy" (1961)
↑ see it in the visualization - k-percent rule
Grow the money supply at a fixed, announced rate every year, roughly the economy's long-run real growth rate.
A rule cannot overreact, cannot be pressured, and gives everyone a stable expectation. Money demand turned unstable in the 1980s and targets were dropped, but the idea that central banks should follow predictable rules lives on in inflation targeting and the Taylor rule.
A Program for Monetary Stability (1960)
Bundesbank money targets 1975–98Fed and Bank of England monetarist experiments 1979–82↑ see it in the visualization - Permanent income hypothesis
People base their spending on their long-run expected income, not on this year's income, so they save windfalls and borrow through bad years.
Consumption is smoother than income, and temporary tax rebates stimulate much less than a simple Keynesian consumption function predicts. Measured responses to windfalls are often larger than the pure theory says, largely because many households cannot borrow.
A Theory of the Consumption Function (1957)
1975 US tax rebate2001 and 2008 rebate checks↑ see it in the visualization - Four ways to spend money
Your money on yourself (care about cost and value), your money on others (cost), others' money on yourself (value), others' money on others (neither).
Most government spending falls in the fourth box, where nobody involved has a strong reason to economize or to get good value. The problem is incentives and information, not bad people.
Free to Choose (1980, with Rose Friedman), ch. 4
↑ see it in the visualization - "Nothing is so permanent as a temporary government program"
Programs launched for an emergency acquire beneficiaries, staff and lobbies that keep them alive after the emergency ends.
Concentrated benefits organize; diffuse costs do not. Each renewal vote pits a motivated clientele against taxpayers who each lose a few dollars. Friedman himself helped design wartime income-tax withholding at the Treasury in 1942–43, and later regretted that it outlived the war.
Tyranny of the Status Quo (1984, with Rose Friedman)
Federal telephone excise tax: wartime 1898, reimposed 1914, fully ended 2006NYC rent control since 1943Wool and mohair subsidy 1954–95, revived 2002↑ see it in the visualization - There's no such thing as a free lunch
Everything has a cost; if you do not pay for something directly, someone pays for it, through taxes, inflation or forgone alternatives.
Programs that look free hide their costs in the fourth box of spending. Friedman used the old saloon phrase as the title of a 1975 book of essays.
There's No Such Thing as a Free Lunch (1975)
↑ see it in the visualization - Negative income tax
Replace the patchwork of welfare programs with a single cash payment that tapers off gradually as earnings rise.
It guarantees a floor, treats recipients as adults who can spend cash, cuts bureaucracy, and, unlike benefits withdrawn dollar for dollar, always leaves people better off for working more. The US EITC is a partial descendant. The catch: any two of a generous floor, a low taper and a low cost; not all three.
Capitalism and Freedom (1962), ch. 12
US EITC (1975)US NIT experiments 1968–82Brazil's Bolsa Família (cash, not in-kind)↑ see it in the visualization - Poverty trap (effective marginal tax rate)
When benefits are withdrawn as earnings rise, the effective tax on extra earnings can reach 100% or more.
If earning $5,000 more leaves you no better off, staying on welfare is rational. Friedman argued such schemes punish exactly the work that leads out of poverty.
Free to Choose (1980), ch. 4
Benefit cliffs in US Medicaid and housing aidUK benefit withdrawal before Universal Credit↑ see it in the visualization - School vouchers
Government funds education, but gives the money to families as a voucher they can spend at any approved school.
Financing schooling does not require running the schools. When money follows the student, schools compete for pupils and poor families get the choice richer ones already have by moving house. Evidence from Chile, Sweden and US programs is mixed and hotly debated.
"The Role of Government in Education" (1955); Capitalism and Freedom (1962), ch. 6
Milwaukee 1990Chile 1981Sweden 1992Arizona ESAs 2022↑ see it in the visualization - Economic freedom as a condition of political freedom
A competitive market economy disperses power, which makes political freedom possible.
When the state is the only employer, dissent has nowhere to earn a living. Markets let people cooperate without agreeing on everything, which reduces how much has to be decided politically.
Capitalism and Freedom (1962), ch. 1
Hollywood blacklist: blacklisted writers kept working under pseudonymsChile and Taiwan: markets first, democracy later - Floating exchange rates
Let currency prices be set by the market instead of pegging them.
A fixed rate forces domestic money and prices to adjust to defend the peg, often through recession; a floating rate absorbs shocks and frees monetary policy. The Bretton Woods system ended in 1971–73, as Friedman had urged since 1953.
"The Case for Flexible Exchange Rates" (1953)
End of Bretton Woods 1971–73Brazil's float, January 1999UK leaves the ERM 1992