Skip to main content

What Is the Modern Phillips Curve Debate?

The Phillips curve is one of the most studied—and contested—relationships in economics. Named after economist A.W. Phillips, who discovered it in 1958, the Phillips curve describes an inverse relationship between unemployment and inflation: when joblessness falls, inflation tends to rise, and vice versa. For decades, this trade-off guided central banks' decisions about interest rates and stimulus. But starting in the 1990s, and accelerating through the 2010s, the relationship broke down. The curve flattened, meaning unemployment could fall without triggering the wage and price increases economists expected. This shift sparked a fierce debate about whether the Phillips curve still matters, what drives inflation today, and how policymakers should set rates when the traditional guideposts fail.

Quick definition: The Phillips curve describes the historical inverse relationship between unemployment and inflation; the "flattening" refers to a weaker or disappearing link between job losses and price growth in recent decades, forcing economists to rethink inflation forecasting.

Key takeaways​

  • The Phillips curve showed that lower unemployment reliably preceded higher inflation, making it a central tool for Fed policy for 40 years.
  • Starting in the 1990s, the relationship weakened: unemployment fell without triggering proportional inflation, puzzling economists worldwide.
  • A flatter Phillips curve means the Fed faces less of a traditional trade-off between growth and inflation, but also harder forecasting.
  • Global supply chains, shifting worker expectations, and credible central bank inflation targets have all been proposed to explain the shift.
  • The COVID-19 pandemic briefly revived inflation concerns but didn't restore the old Phillips curve relationship.
  • Modern monetary policy relies less on unemployment levels and more on wage growth, slack in labor markets, and inflation expectations.

The Original Phillips Curve: A Stable Law​

In 1958, British economist A.W. Phillips examined 100 years of wage and unemployment data from the United Kingdom and found a remarkably consistent pattern: whenever unemployment was low, wage growth accelerated. Conversely, high unemployment coincided with flat or falling wages. Phillips plotted these pairs on a chart and drew a smooth curve through them. The relationship held so reliably that it seemed almost like a law of economics—a real trade-off built into the labor market itself.

In the 1960s, American economists Paul Samuelson and Robert Solow adapted Phillips's wage-unemployment relationship into a price-inflation relationship. They argued that if lower unemployment drives higher wages, and higher wages push up production costs, then firms will raise prices. The result: an inflation-unemployment trade-off. A policymaker could choose a point on the curve—accept more joblessness to keep inflation low, or accept higher inflation to reduce unemployment. This framing made the Phillips curve the foundation of macroeconomic policy. Central banks, including the Federal Reserve, used it to calibrate stimulus and tightening.

For the next 40 years, the relationship held broadly true. When the Fed pushed unemployment down in the late 1960s, inflation rose sharply. When the Fed tightened hard in the early 1980s to break inflation, joblessness climbed above 9%. The Phillips curve seemed to be etched into the economy's structure.

The Flattening Begins: The 1990s Puzzle​

In the early 1990s, something unexpected happened. The U.S. unemployment rate fell steadily—from 7.8% in 1992 to 3.9% by 2000—yet inflation remained subdued, hovering near 2% throughout most of the decade. By the old Phillips curve, inflation should have risen sharply as joblessness plummeted. Instead, it stayed flat. The flattening Phillips curve had begun.

Economists offered several explanations. The most credible centered on inflation expectations. In the 1970s and 1980s, workers and businesses expected inflation to stay high because the Fed had tolerated it. When unemployment fell, wage demands surged because everyone assumed prices would keep climbing. But by the 1990s, the Federal Reserve under Paul Volcker and Alan Greenspan had restored credibility: inflation targeting had become explicit policy, and workers believed the Fed would maintain price stability. With inflation expectations anchored, a tight labor market no longer automatically triggered wage-price spirals. Lower unemployment raised wages somewhat, but not explosively, because workers didn't expect those higher wages to lose purchasing power.

A second factor was global competition. The rise of offshoring and imports from low-wage countries like China and India meant that U.S. firms faced price pressure from abroad. Even with a tight labor market at home, companies could hold prices down by sourcing materials globally or shifting production overseas. A worker might demand higher wages, but if the firm could source the good cheaper from abroad, the price to the consumer stayed flat.

A third factor involved supply shocks that favored inflation. The 1990s saw rapid productivity growth, especially in information technology. Productivity gains mean firms can pay higher wages and still keep prices stable, because each worker produces more output per hour. Tech investments lowered costs across the economy, offsetting wage growth.

By 2003, Fed researchers had documented the flattening thoroughly, and the institution began relying less on unemployment levels as an inflation signal. The old Phillips curve wasn't quite dead—the relationship was weaker, not absent—but it no longer provided the reliable guidance it once did.

The 2010s: A Flatter Phillips Curve Solidifies​

The Great Recession of 2008–2009 and the slow recovery that followed offered another test. Unemployment peaked at 10% in late 2009, then fell gradually over a decade-long expansion. By 2019, joblessness had dropped to 3.5%—a 50-year low—yet inflation remained around 2%, exactly where the Fed wanted it. By the old Phillips curve logic, an unemployment rate that low should have triggered 4% or 5% inflation. Instead, price growth barely budged.

This puzzled Fed officials and academic economists alike. Janet Yellen, who chaired the Federal Reserve from 2014 to 2018, gave speeches questioning whether the Phillips curve was "dead." She noted that the relationship had flattened so much that unemployment alone was a poor predictor of inflation. Other variables—inflation expectations, the behavior of wage growth, "slack" in labor markets—seemed to matter more than the jobless rate itself.

The 2010s solidified the view that the Phillips curve had shifted permanently. When the Fed raised interest rates from 2015 to 2018, unemployment fell to historic lows without inflation spiking. When the Fed cut rates again in 2019, unemployment stayed low and inflation still didn't accelerate. The curve had flattened so much it was nearly horizontal.

Why Did the Phillips Curve Flatten? The Competing Theories​

Economists have proposed several mechanisms to explain the shift. None is universally accepted, but each captures part of the story.

Anchor expectations theory holds that the Fed's credible commitment to a 2% inflation target has convinced workers and businesses that inflation will stay near that level, removing the self-fulfilling prophecy of wage-price spirals. If everyone expects 2% inflation, they'll ask for wages that reflect that expectation, but won't demand the exponential raises that used to follow a tight labor market.

Global slack theory emphasizes that supply chains have made labor markets truly global. A U.S. firm facing tight domestic labor markets can hire from abroad, use telecommuting workers in cheaper locations, or relocate production. This caps wage growth even when unemployment is low. As long as global labor markets have slack—high unemployment in Europe, Asia, or emerging markets—U.S. firms can moderate wage pressures.

Declining worker power theory points to the fall of union membership, the rise of gig work, and stricter immigration enforcement, all of which have weakened workers' ability to demand higher wages. Even with low unemployment, workers lack the bargaining power to push for wages that exceed productivity growth.

Measurement issues theory suggests that the official unemployment rate is less informative than it once was, because participation rates have fallen and many workers have given up looking. The real slack in labor markets is larger than the jobless rate suggests, which is why we don't see inflation even when unemployment looks low.

Productivity paradox theory notes that despite decades of investment in digital technology, measured productivity growth remained slow in the 2010s. Slow productivity growth should theoretically mean firms can't afford high wage growth without raising prices. If productivity is the binding constraint, then tight labor markets won't produce inflation until productivity accelerates.

Each theory has evidence. Surveys show inflation expectations have remained anchored near 2% for years. Global trade does link U.S. wages to global labor markets. Union membership has fallen from 35% in the 1950s to 10% today. Participation rates are lower than they'd be if the labor market were as tight as headline unemployment suggests. And productivity growth, while higher in 2020–2022, was genuinely sluggish in the 2010s.

The flattening Phillips curve likely reflects all of these forces at once, not just one.

The Phillips Curve and Modern Monetary Policy​

The flattening has forced policymakers and forecasters to rely on different signals. The Federal Reserve, in practice, now pays more attention to wage growth, labor force participation trends, and inflation expectations than to the jobless rate alone. When the Fed raises rates, it's not because unemployment hit some magic threshold; it's because wage growth has accelerated, labor market slack has tightened, or inflation expectations have begun to drift upward.

This shift is visible in Fed speeches and policy statements. Officials speak of "maximum employment" as a multidimensional goal, not a single unemployment target. They point to wage data from the U.S. Bureau of Labor Statistics, labor force participation from the Census Bureau, and survey-based inflation expectations from the University of Michigan. The unemployment rate remains in the Fed's dashboard, but it's one instrument among many, not the primary gauge.

The shift also affects how economists forecast inflation. In the old Phillips curve framework, you'd estimate what unemployment rate corresponds to stable inflation (the "natural rate" or NAIRU—Non-Accelerating Inflation Rate of Unemployment—typically estimated around 4.5–5%). Once you knew the NAIRU, you could predict inflation by checking whether the actual rate was above or below it. Today, that method is less reliable. Even unemployment rates well below historical NAIRUs have failed to produce the inflation surge theory would predict.

Instead, modern inflation forecasts use approaches that treat inflation expectations as exogenous and focus on wage growth relative to productivity growth and changes in slack. If wages grow faster than productivity, and slack tightens, inflation may rise; but low unemployment alone doesn't seal the deal.

The COVID Shock and Brief Phillips Curve Revival​

The COVID-19 pandemic and the policy response created a moment when the Phillips curve briefly seemed to reassert itself. In 2021 and early 2022, as unemployment fell from 6.7% to 3.4%, inflation climbed from 1.4% to 8.6%, the highest in 40 years. Unemployment was falling, inflation was rising—exactly the old Phillips curve relationship. Fed skeptics argued that skepticism about the flattening had been premature.

But a closer look reveals that the 2021–2022 inflation surge was supply-driven, not demand-driven. Inflation jumped because of disrupted supply chains, shortages of semiconductors and container shipping, energy price shocks (especially after Russia invaded Ukraine), and fiscal stimulus concentrated during the pandemic. The inflation wasn't caused by tight labor markets driving up wage-price spirals; it was caused by constrained supply. Wages did accelerate, but they lagged inflation—real wages actually fell. And once the Fed tightened rates and supply bottlenecks eased, inflation came down sharply in 2023–2024, even though unemployment stayed low.

This experience suggests that the Phillips curve remains flattened. The 2021–2022 episode shows that tight labor markets can coexist with surging inflation (the old Phillips curve's prediction), but it also shows that high inflation can exist without a tight labor market being the root cause. The relationship is no longer stable and predictable.

If the Phillips curve has flattened, what relationship has replaced it? Many researchers now focus on nominal wage growth versus productivity growth. The logic is straightforward: if wages grow faster than productivity, unit labor costs rise, and firms have incentive to raise prices. If wage growth tracks productivity, costs stay stable and inflation stays contained.

From 2000 to 2019, nominal wage growth averaged around 2–3% per year, while productivity growth was around 1–2%. Wage growth modestly exceeded productivity, but by such a small margin that inflation remained low. This relationship held even when unemployment fell to very low levels, supporting the "anchored expectations" view: workers and businesses expected 2% inflation, so even in a tight labor market, wage growth stayed moderate.

In 2021–2023, wage growth accelerated to 4–5% briefly, outpacing productivity more noticeably, and that coincided with inflation. But the relationship was weaker than old Phillips curve models would predict. A 4.5% wage growth rate would have generated vastly more inflation under 1970s dynamics; instead, it produced an inflation spike that receded as supply recovered.

The implication is that wage growth matters, but it's neither sufficient nor necessary for inflation. Supply shocks, energy prices, inflation expectations, and global slack all interact with wage growth to determine inflation outcomes. A 4% unemployment rate with 3% wage growth might produce 2% inflation if expectations are anchored and global slack is high; the same conditions might produce 4% inflation if a supply shock has occurred and expectations have drifted.

Debates and Open Questions​

The modern Phillips curve debate remains unresolved. Here are the key unresolved questions:

Is the flattening permanent? Some economists argue that credible inflation targeting and global integration have permanently altered the inflation-unemployment relationship. Others suggest the flattening is cyclical and that the Phillips curve could re-steepen if inflation expectations become unanchored.

How much slack exists in labor markets? The official unemployment rate is one measure, but underemployment, labor force participation, and job-mismatch are harder to quantify. If hidden slack is larger than the unemployment rate suggests, a flatter Phillips curve makes sense.

Do expectations drive inflation, or just constrain it? If inflation expectations are fully anchored, monetary policy becomes more powerful (expectations about future policy matter, not just current unemployment), but less predictable (small shifts in expectations can have large effects).

What will the next recession reveal? Recessions have always been reset events for the Phillips curve. If unemployment spikes in the next downturn, will inflation fall as predicted, vindicating the flattening? Or will inflation prove sticky, surprising observers again?

Real-world examples​

The U.S. experience from 2010 to 2019 is the clearest example. Unemployment fell from 10% to 3.5%, a historic drop, yet inflation stayed near the Federal Reserve's 2% target the entire time. Under old Phillips curve logic, such a decline should have generated cumulative inflation of 4–6%. Instead, it didn't happen.

The euro area offers a parallel case. After the 2008 financial crisis, euro unemployment stayed much higher—above 10%—for a full decade. Yet inflation also remained low, well below the 2% target. The ECB raised rates modestly despite double-digit unemployment, a strategy unthinkable under the old Phillips curve framework. This "secular stagnation" period in Europe further convinced researchers that the Phillips curve had fundamentally changed.

Japan's experience is even starker. For 30 years, Japanese unemployment has been low (often <3%), yet inflation has remained near zero. If the Phillips curve held, Japan should have experienced steady inflation instead of deflation. Japan is an extreme case—structural issues like an aging population, labor market inflexibility, and deeply anchored deflationary expectations matter—but it illustrates how far the Phillips curve can flatten under certain conditions.

The 2021–2022 U.S. inflation spike briefly suggested the Phillips curve might be returning. But economists noted that inflation measures that exclude energy and food (which were supply-shocked) remained more moderate, and that nominal wages, while rising, were not keeping pace with inflation—real wages fell. Once supply bottlenecks cleared in 2023, inflation came down rapidly despite tight labor markets, again showing that the Phillips curve relationship is weaker than before.

Common mistakes​

Mistake 1: Assuming the Phillips curve never changed. Some economists and policymakers cling to 1960s-style thinking, in which any unemployment rate below 4.5% guarantees an inflation spiral. This misses decades of evidence that the relationship has shifted.

Mistake 2: Assuming the Phillips curve is completely dead. Others argue that the Phillips curve is now completely irrelevant and unemployment tells us nothing about inflation. This overstates the case. Wage growth, which is loosely tied to unemployment, does affect inflation via unit labor costs. The relationship is weaker and less predictable, not absent.

Mistake 3: Conflating inflation with wage growth. Wages and prices are linked through labor costs, but they're not the same. Inflation in the 2010s was low even as wage growth accelerated. Inflation in 2021–2022 spiked due to supply shocks, not primarily wage growth. The two move together in equilibrium, but can diverge in transition.

Mistake 4: Ignoring supply shocks. The Phillips curve describes demand-driven inflation (low unemployment → high demand → higher prices). But inflation can spike from supply shocks (oil embargoes, pandemics, supply-chain breaks), in which case unemployment may be high or low. Modern inflation forecasting requires accounting for both demand and supply.

Mistake 5: Treating inflation expectations as exogenous. Inflation expectations don't just happen; they're shaped by the Fed's track record, communication, and realized inflation. If the Fed lets inflation run above target for years, expectations will drift. An explicitly anchored 2% target is only as credible as the Fed's commitment to defending it.

FAQ​

Has the Phillips curve permanently flattened?​

Possibly, but it's too early to be certain. The relationship has been flat for 25+ years, which suggests something structural changed. But if expectations become unanchored (due to years of inflation above target, for example), the Phillips curve could re-steepen. The best answer is: the relationship has shifted and is likely to remain weaker than before, but "permanent" is an unfalsifiable claim.

Does low unemployment no longer predict inflation?​

Low unemployment alone is a weak predictor now, but in combination with other variables—wage growth, inflation expectations, global slack, supply conditions—it still provides useful information. A 2% unemployment rate would be concerning for inflation; a 5% rate is much less so. The relationship is weaker, not absent.

Why do central banks still care about unemployment?​

Because low unemployment does correlate with wage growth, and wage growth (relative to productivity) does correlate with inflation. The correlation is just weaker than it was in the 1960s–1980s. Additionally, the Fed has a dual mandate including "maximum employment," so unemployment matters for policy even if its inflation implications are uncertain.

Could the Phillips curve re-steepen in the future?​

Yes. If global labor markets tighten, if worker bargaining power increases, if inflation expectations become unanchored, or if credible targeting erodes, the Phillips curve could steepen. The shift to a flatter curve required changes in expectations, global conditions, and institutional credibility; reversing it would require those conditions to shift back.

What do modern central banks use instead of the Phillips curve?​

Modern monetary policy frameworks emphasize inflation expectations (from surveys and asset prices), wage growth data, broad labor market slack measures, and forward-looking inflation expectations. The Fed publishes a "dot plot" of rate expectations rather than relying mechanically on an unemployment threshold. Real-time data like initial jobless claims, job openings, and quits (from the JOLTS report) matter more than the headline unemployment rate alone.

Did the Phillips curve break, or did economists misunderstand it?​

A bit of both. The Phillips curve is real—lower unemployment does tend to correlate with wage growth. But economists overstated how tight and predictable that relationship was. They also underestimated the role of expectations and global factors. The curve didn't break; economists simply learned it was less powerful and more fragile than they'd assumed.

Summary​

The Phillips curve—the historical inverse relationship between unemployment and inflation—was one of economics' most reliable laws for 40 years. But starting in the 1990s, the curve flattened: unemployment fell without triggering proportional inflation. This shift reflected changing inflation expectations, global competition, and possibly shifting worker power, but the exact causes remain debated. Modern monetary policy relies less on the Phillips curve and more on wage growth, slack measures, and inflation expectations. The curve is not dead, but it is much weaker and less predictable than in its heyday, forcing central banks to look beyond jobless rates when setting policy.

Next​

→ The JOLTS report explained