Every serious science eventually agrees on its own foundations. Economics never has. Its five biggest arguments, whether markets are rational, whether government spending helps or hurts, whether inflation and jobs trade off against each other, whether a government can ever truly run out of money, whether free trade beats protection, have been running for over a century, and neither side has ever fully won. That is not because economists are bad at their jobs. It is because each theory keeps getting proven right in one decade and proven wrong in the next. Here are all five fights, with a real case study on each side.
Chapter 1Do markets know best, or do they lose their minds on schedule?
Eugene Fama's efficient market hypothesis says asset prices already reflect everything knowable at any given moment. If a mispricing existed, someone faster or smarter would have already traded it away. So nobody can reliably beat the market without taking on extra risk.
The proof. SPIVA, the industry's own scorecard on this exact question, tracks how many actively managed funds beat their benchmark index over time. Over a fifteen year stretch, roughly 90% of actively managed large cap US funds fail to beat the S&P 500. Pay someone to pick winners for you, and nine times out of ten you would have done better in a boring index fund. That is efficient markets working almost exactly as advertised.
The disproof. Robert Shiller flagged the dotcom bubble in 2000 and the housing bubble in 2005, both years before they burst, using nothing more exotic than valuation ratios that had drifted far from their historical average. If prices already reflected all available information, that gap should not have been visible in advance, let alone visible for years. Long Term Capital Management is the sharper version of the same story. The fund's founders had literally helped build the Nobel winning theory behind efficient pricing, and the fund still lost about 92% of its value in 1998 when a liquidity panic broke every correlation their models assumed was stable.
The live test right now. A handful of AI and semiconductor names have carried a wildly disproportionate share of global index returns for two years. In the first week of July 2026, that trade cracked hard. VanEck's Semiconductor ETF fell 4.5% in a single session, Teradyne dropped 13.6%, and South Korea's Kospi fell almost 8% on the same story. Efficient market believers call this a rational repricing of genuinely uncertain future earnings. Behavioural economists call it a crowded trade meeting its own weight. Both explanations fit the same three days of price action.
Does government spending rescue an economy, or just borrow growth from tomorrow?
Keynes argued that during a genuine demand shortfall, private spending can get stuck low because everyone is waiting for everyone else to spend first, and only government deficit spending can break that standoff.
Milton Friedman's monetarism says the opposite mechanism usually wins. Government borrowing pulls savings out of the pool available for private investment, pushing up interest rates and crowding out the very private activity the stimulus was meant to spark.
The proof of Keynes. The 2020 stimulus response is about as clean a test as this argument gets. Massive fiscal and monetary support kept unemployment from spiralling the way it did in past pandemics or depressions, and the US economy recovered its lost output faster than after 2008, when the stimulus was smaller and slower.
The proof of Friedman. Britain in 1976 is the textbook counter case. Prime Minister James Callaghan told his own party's conference that Britain could no longer spend its way out of a recession by cutting taxes and boosting government spending, because that had only ever worked, to the extent it worked at all, by injecting inflation into the economy, followed by a bigger dose of unemployment as the next step. Britain went to the IMF for an emergency loan that same year. It remains one of the few moments a sitting head of government has publicly renounced Keynesian orthodoxy while still in office.
The complication neither side saw coming. Interest rates stayed near zero for over a decade after 2008 despite enormous government borrowing. Simple crowding out theory predicts rates should rise when the government borrows heavily, not fall for ten straight years. Neither camp has a fully satisfying answer for that stretch on its own.
Chapter 3Is there really a tradeoff between inflation and jobs, or is that a trap?
The original Phillips curve, built from close to a century of British wage data, showed a clean pattern. Lower unemployment came with higher inflation. Policymakers through the 1960s treated this as a dial they could turn, accept a bit more inflation, buy a bit less unemployment.
Friedman and Edmund Phelps argued in 1967 and 1968 that this only works as long as people are surprised by inflation. Once wage contracts start expecting it, the tradeoff disappears and all you are left with is permanently higher inflation and no lasting jobs gain.
The proof of the original curve. 1962 US data lines up almost perfectly with it. Unemployment at 5.6% came with inflation near 1%.
The proof of Friedman and Phelps. By 1979, unemployment sat at a similar 5.6%, but inflation had climbed past 11%. By 1980, inflation peaked near 13.5% with unemployment at 7%, high inflation and high unemployment together, the exact combination the original curve said should not be possible. Paul Volcker's Federal Reserve broke that cycle from 1979 to 1982 by deliberately driving unemployment above 10%, a brutal, direct demonstration that the tradeoff Friedman described was real, not a coincidence.
The live test right now. June 2026's US jobs report added only 57,000 jobs against a forecast of 113,000, a genuinely weak print. A simple Phillips curve reading says cut rates immediately. But core inflation was running at 4.1% in May, more than double the Fed's target, and the Fed held its policy rate steady in June with one more hike still pencilled in for later this year. A weak jobs market next to stubborn inflation and a central bank refusing to ease is Friedman's natural rate framework playing out in real time, not the simple 1960s version of the curve.
Chapter 4Can a government that prints its own money ever actually go broke?
Modern Monetary Theory argues that a government borrowing only in its own currency cannot be forced into an involuntary default, because it can always create the currency needed to pay its own debt. The real limit under this view is not the debt ratio at all. It is inflation.
Ricardian equivalence, from Robert Barro, argues households see through deficit spending in advance. They expect higher taxes later to pay for it, so they save the stimulus rather than spend it, quietly cancelling out the policy before it even works.
The proof that currency issuers have real room. Japan has run a government debt to GDP ratio above 200% for years, among the highest in the world, without a currency collapse or a failed bond auction. If the traditional debt ceiling logic were strictly true everywhere, Japan should have broken long before now.
The disproof, and it is a sharp one. Britain's September 2022 mini budget is the cleanest real world test either side of this argument has ever produced. The government announced large unfunded tax cuts, and UK government bond yields spiked so violently within days that the Bank of England had to step in with emergency bond purchases to stop pension funds from collapsing. The Prime Minister who announced the budget was gone within weeks. Britain borrows in its own currency, the pound, exactly the condition MMT says should protect a government from this kind of market discipline. The market punished it anyway, in days, not years.
The live number to watch. US federal debt to GDP sat at roughly 123% in May 2026, with total debt above 39 trillion dollars and annual interest payments now exceeding 1 trillion dollars a year, a bigger line item than defence spending. The Penn Wharton Budget Model's June 2026 estimate puts an outer limit on US debt at roughly 210% of GDP, beyond which markets stop being able to rationally absorb further borrowing, currency issuer or not.
Does free trade make everyone richer, or does protection build tomorrow's winners?
Ricardo's 1817 theory of comparative advantage says countries get richer by specialising in whatever they make most efficiently and trading freely for the rest. Any tariff that blocks this makes both trading partners poorer on average, even if it helps one protected industry.
The infant industry tradition, from Alexander Hamilton and Friedrich List, and its modern version in Paul Krugman's new trade theory, argues that in industries with strong scale advantages, being first and biggest can lock in a lasting edge a free market might never let a late arriving country's firms reach. A temporary government push can be justified even inside free trade logic, in this narrow case.
The proof of comparative advantage. Vietnam, running tariffs at roughly half of India's average level and signing trade deals India has spent nearly two decades failing to conclude, grew its exports to nine times India's size over about a decade, despite having a fraction of India's population.
The proof of infant industry protection, done properly. South Korea protected its domestic market from the 1960s onward, but tied that protection directly to export performance. Miss your export target, lose your cheap credit and your tariff shelter. Manufacturing went from 14.3% of Korea's economy in 1962 to 30.3% by 1987, and growth climbed toward 9% a year.
The disproof of protection done carelessly. Argentina and much of Latin America ran the same import substitution playbook from the 1930s onward, protecting domestic industry for decades with no export conditionality attached. The result was chronic stagnation rather than a Korea style takeoff, and it is the reason economists now treat "protection" and "protection with an exit test" as two entirely different policies rather than variations on one idea.
Chapter 6So which one is actually right?
None of them, as a permanent law. All of them, as a description of one specific moment. Efficient markets describe most of most markets most of the time, right up until a crowded trade or a liquidity panic hands the argument to behavioural economics for a few brutal weeks. Keynesian spending works when an economy has real spare capacity sitting idle. Monetarist crowding out concerns start to bite once that spare capacity runs out and inflation, not unemployment, becomes the actual constraint, which is closer to where the US sits today. The Phillips curve is real in the short run and disappears in the long run exactly as Friedman said, and June 2026's weak jobs report next to sticky inflation looks like his framework reasserting itself live. MMT is right that a currency issuer has more room than a household. Britain's 2022 gilt crisis is the reminder that more room is never the same as infinite room. Every theory on this list has been completely right at least once and completely wrong at least once, often in the same country a decade apart. The only real skill in economics is knowing which decade you are standing in.