Skip to content

Phillips Curve

The short-run inverse relation between unemployment and inflation — positioned by expected inflation, sloped by how anchored those expectations are, vertical at the natural rate in the long run, and displaced by supply shocks — whose exploitable trade-off dissolves once agents come to expect the inflation.

Core Idea

The Phillips curve is the macroeconomic relationship describing an inverse short-run association between the unemployment rate (or output gap) and the rate of price or wage inflation: when labour markets tighten — unemployment falls, output runs above potential — wage growth and then price inflation rise; when labour markets slacken the reverse holds. A. W. Phillips documented the empirical regularity for UK wage inflation and unemployment over 1861–1957; Samuelson and Solow generalised it to U.S. price inflation in 1960, and it became the canonical short-run policy menu — the rate of inflation a central bank could "buy" by accepting a given unemployment rate.

The simple curve was qualified in two decisive steps. Friedman (1968) and Phelps (1967) introduced the expectations-augmented version: the trade-off exists only for surprise inflation; if agents adapt their inflation expectations the short-run curve shifts up until, at the natural rate of unemployment (the NAIRU), there is no permanent trade-off — the long-run Phillips curve is vertical. Supply shocks — oil embargoes, cost-push pressures — shift the curve outward, combining higher inflation with higher unemployment (stagflation), as the 1970s demonstrated empirically. In the New Keynesian formulation the curve becomes a structural equation relating current inflation to expected future inflation and the output gap, forward-looking and derived from price-setting microfoundations. Since the 1990s the curve has flattened markedly in most advanced economies — well-anchored central-bank inflation expectations compress the inflation response to labour-market tightening — and its empirical instability, predicted by the Lucas critique (policy-conditional historical relationships shift when policy regimes change), is now a central fact of applied macroeconomics.

Structural Signature

Sig role-phrases:

  • the labour-market tightness axis — unemployment rate, output gap, or employment ratio, the horizontal variable the relation is read against
  • the inflation axis — price or nominal-wage inflation, the vertical variable that moves inversely with tightness in the short run
  • the short-run trade-off curve — the downward-sloping locus itself, whose slope sets how much inflation is "bought" per unit of unemployment reduction
  • the expectations position-shifter — expected inflation, which slides the whole curve up the page as agents adapt, so only surprise inflation rides the trade-off
  • the anchoring slope-control — how well-anchored expectations are, flattening the curve when credible (post-1990s) and steepening it when adaptive
  • the vertical long-run asymptote — the natural rate / NAIRU, where no permanent trade-off survives and any chosen inflation returns the economy
  • the supply-shock displacement — oil, terms-of-trade, cost-push forces that shift the whole curve outward, pairing high inflation with high unemployment (stagflation)
  • the reflexive non-stationarity — the deliberate Lucas-critique caution that the relation is policy-conditional, shifting when acted upon, so a curve fitted to past data is no stable menu

What It Is Not

  • Not a stable permanent menu of inflation–unemployment points. The historical scatter is not a fixed buffet a central bank can pick a point on and hold indefinitely — that was precisely the 1960s error. The exploitable trade-off exists only for surprise inflation; once agents come to expect a given inflation rate the short-run curve shifts up, so the long-run curve is vertical and any chosen inflation returns the economy to the natural rate.
  • Not refuted by 1970s stagflation. Simultaneous high inflation and high unemployment did not show the relationship "failed" — it cannot be a slide on any downward-sloping curve, so it diagnoses an outward shift: a supply shock displaced the whole curve. The expectations-augmented version absorbs stagflation as the same mechanism observed under a shock-displaced setting, not as a counterexample to it.
  • Not a permanent trade-off between inflation and jobs. There is no durable bargain in which accepting higher inflation buys permanently lower unemployment. The trade-off is short-run and conditional on the inflation being unanticipated; in the long run the curve is vertical at the natural rate / NAIRU, where the economy settles regardless of which inflation rate is chosen.
  • Not gone because it flattened. The post-1990s flattening — unemployment hitting historic lows while inflation barely stirs — is not the relationship vanishing but its slope compressing because well-anchored central-bank expectations damp the inflation response to labour-market tightening. The curve is read as flat under an anchored-expectations regime, not absent.
  • Not a structural law of nature. The curve is a policy-conditional relationship estimated under one policy regime, liable to shift when the regime changes — the Lucas critique made concrete. A curve fitted to past data is a regime-conditional correlation, not a stable structural equation; the very act of acting on it can move it, which is why it must not be extrapolated across regime changes as if it were invariant.
  • Not a generic "trade-off" or just the Lucas critique. It is one specific instance keyed to wage-price dynamics, the NAIRU, expectations-augmentation, and supply-shock shifters — its labour-and-price content is what makes it the Phillips curve rather than the bare two-objective frontier or the general regime-conditional-relationship pattern. Those broader structures carry the cross-domain lesson; the macroeconomic apparatus is what stays home.

Scope of Application

The Phillips curve lives across the subfields of macroeconomics concerned with inflation, labour markets, and monetary policy; its reach is within that domain — the "growth-versus-environment" or "security-versus-liberty" extensions are analogy whose real content is carried by the trade-off-frontier and Lucas-critique parents, not the wage-price machinery. The genuine habitats enumerate the in-domain contexts where the shift-parametrized inflation–unemployment relation actually organizes analysis.

  • Monetary policy analysis — the operational home: Federal Reserve dual-mandate deliberation, ECB and Bank of Japan policy, where the curve frames how much inflation a given degree of labour-market tightening implies and whether to ride or anchor it.
  • Inflation forecasting and central-bank modelling — Phillips-curve equations (output gap plus expected inflation) embedded in structural and reduced-form forecasting models, including the New Keynesian forward-looking formulation.
  • Labour economics and wage dynamics — the wage-Phillips-curve relating nominal wage growth to unemployment and the wage-price spiral, the original 1861–1957 UK regularity Phillips documented.
  • NAIRU and natural-rate estimation — empirical work locating the vertical long-run asymptote, the unemployment rate consistent with stable inflation, central to potential-output and slack measurement.
  • Business-cycle and stabilization theory — the short-run trade-off as the demand-side mechanism linking output gaps to inflation, and the expectations-augmented sequence explaining stagflation and disinflation episodes.
  • Macroeconometrics and the Lucas critique — the canonical case study in policy-conditional, non-stationary relationships, where the curve's instability under regime change is the worked example of why fitted historical relations may not survive being acted upon.

Clarity

Calling the inflation–unemployment relation a Phillips curve rather than just "the trade-off between inflation and jobs" forces the analyst to say which curve, and that demand is the whole clarifying payoff. A bare trade-off invites the fallacy that burned policymakers in the 1960s: that the historical scatter is a stable menu a central bank can pick a point on indefinitely. Naming the relationship as a curve with shift parameters makes its conditionality unavoidable — one must now specify the horizon (short-run trade-off versus vertical long-run), the inflation measure, the expectations regime (anchored versus adaptive), and whether supply shocks are displacing the curve. The sharper question stops being "how much inflation must we accept to cut unemployment?" and becomes "given today's expectations and supply conditions, where is the curve, how steep is it, and will riding it move it?"

The distinction it makes legible is movement along the curve versus movement of the curve — and, underneath that, the difference between a structural relationship and a regime-conditional correlation. A surprise that exploits a momentary trade-off is a slide along a fixed curve; an inflation that agents come to expect shifts the whole curve up, which is why the long-run version is vertical and the trade-off illusory. Recognizing this is what lets a macroeconomist read stagflation not as the curve "failing" but as a supply shock shifting it outward, and the post-1990s flattening not as the relationship vanishing but as anchored expectations compressing its slope. The Phillips curve is the canonical case in which the very act of exploiting an empirical regularity changes it — the concrete instance that makes the non-stationarity of policy-conditional relationships visible to the practitioner.

Manages Complexity

Decades of joint inflation and unemployment observations — across countries, policy regimes, oil shocks, and the quiet anchored-expectations era — present as a cloud of points that wanders, tightens, loosens, and at times inverts, with no obvious common law. The Phillips curve compresses that cloud into a single object: one downward-sloping short-run relation, governed by a handful of parameters the analyst can track in place of the raw scatter. The state of any inflation–unemployment episode reduces to a few quantities — the position of the curve (set by expected inflation), its slope (set by how anchored those expectations are), the location of the vertical long-run asymptote (the natural rate / NAIRU), and the size of any contemporaneous supply-shock displacement. With those in hand the macroeconomist reads the qualitative outcome off a small branch structure rather than re-deriving each decade from the underlying labour and price dynamics: expectations adapting upward walks the curve up the page until the trade-off disappears at the vertical long-run; a supply shock slides the whole curve outward, producing the otherwise-paradoxical pairing of high inflation with high unemployment; well-anchored expectations flatten the slope so tightening labour markets barely move inflation. The sprawl of historical episodes — the 1960s menu, 1970s stagflation, the post-1990s flattening, the 2021 surge — becomes the same curve seen under different settings of position, slope, asymptote, and shock, and the analyst predicts where inflation goes by tracking those few parameters and which of them today's policy will itself disturb. A wandering multi-decade data cloud collapses to one shift-parametrized relation whose movements are read off four scalars and a small set of named branches.

Abstract Reasoning

The Phillips curve licenses a distinctive set of moves in macroeconomic analysis, organized around the master distinction between sliding along the curve and shifting it.

Diagnostic (read the source of an inflation–unemployment episode from where the cloud sits and moves). Confronted with a joint inflation–unemployment observation, the macroeconomist infers the hidden configuration that produced it. Inflation rising as unemployment falls along a stable trace says the economy is sliding up a fixed short-run curve — a demand-side movement exploiting the momentary trade-off. Inflation and unemployment rising together (stagflation) cannot be a slide on any downward-sloping curve and therefore diagnoses an outward shift: a supply shock has displaced the whole relation. Unemployment falling to historic lows with inflation barely stirring diagnoses a flat slope, which in turn indicts well-anchored expectations. Inflation accelerating with no clear labour-market move points away from the curve's domain toward a pure supply or expectations disturbance. The reasoning runs from the shape and motion of the scatter to the hidden parameter that must have changed — position (expected inflation), slope (anchoring), asymptote (the natural rate), or a contemporaneous shock — because those four are the only things that can move the relation.

Interventionist (predict the effect of a policy by asking what it does to the curve, not just where it lands on today's curve). The central interventionist subtlety the concept enforces is that policy can move the object it is trying to ride. A central bank tightening or easing predicts a slide along the short-run curve — but only transiently, and only to the extent the move is a surprise; the same policy, once anticipated, shifts the curve up by raising expected inflation, so the predicted long-run effect on unemployment is nil at the vertical natural rate. Forward guidance and an inflation target are read as interventions on expectations — they pin the curve's position and, by anchoring, flatten its slope, predicting that subsequent labour-market tightening yields less inflation. Supply-side policy is read as a favourable shift of the whole curve inward. Each instrument is mapped to the parameter it moves (slide vs. position vs. slope vs. shock-offset), and the predicted outcome differs accordingly — which is why "ride the trade-off" and "anchor expectations" are opposite bets even though both touch inflation.

Boundary-drawing (when the trade-off may be invoked, and the horizon at which it dissolves). The governing boundary is horizon: a trade-off may be claimed in the short run and must not be claimed in the long run, where the curve is vertical and any chosen inflation rate returns the economy to the natural rate. A second boundary is the expectations regime — the exploitable trade-off exists only for unanticipated inflation, so the move "pick a permanent point on the historical menu" is ruled out of bounds. The deepest boundary the concept draws is reflexive: the relation is policy-conditional, estimated under one policy regime and liable to shift when the regime changes, so a curve fitted to past data may not survive being acted upon. This is the boundary that distinguishes a structural relationship from a regime-conditional correlation, and it is precisely what tells the analyst when the curve may be extrapolated and when it may not.

Order-of-events / predictive sequence. The expectations-augmented version supplies a characteristic temporal narrative: a demand expansion first buys lower unemployment at the cost of surprise inflation (slide up the short-run curve); agents then revise expectations upward; the short-run curve shifts up; unemployment returns to the natural rate at the new, higher inflation. The analyst predicts not a static point but this sequence — initial trade-off, expectational catch-up, return to the vertical long-run at a worse inflation level — which is what turns the 1970s stagflation, the post-1990s flattening, and the 2021 surge into the same mechanism observed at different settings rather than into a relationship that keeps failing.

Knowledge Transfer

Within macroeconomics the Phillips curve transfers as mechanism, and its reach across the field is rich. The same shift-parametrized relation — a downward-sloping short-run trade-off between labour-market tightness and inflation, positioned by expected inflation, sloped by how anchored those expectations are, anchored long-run at the vertical natural rate (NAIRU), and displaced by supply shocks — organizes Federal Reserve dual-mandate analysis, ECB and Bank of Japan policy, the post-1990s flattening, and the 2021–2022 surge. The diagnostics carry intact: distinguish sliding along the curve from shifting it, read stagflation as an outward supply-shock displacement rather than a "failure," read the flattening as anchored expectations compressing the slope, and map each instrument (output-gap targeting, inflation targeting, forward guidance, supply-side and wage-bargaining policy) to the parameter it moves. The vocabulary — the short-run/long-run distinction, expectations-augmentation, NAIRU, the wage-price spiral, the slope and the shifters — moves with the machinery wherever there is a labour market, an inflation process, and a central bank.

Beyond macroeconomics the honest reading is the shared-abstract-mechanism case (B), and unusually it draws on two distinct parent patterns rather than one. First, the curve's surface form — two desired states standing in inverse relation — is one instance of the general two-dimensional trade-off frontier family, which genuinely recurs across domains (Pareto frontiers, multi-objective optimization, the risk-return trade-off, the fast-cheap-quality triangle) and is already housed at the prime level (trade_offs, pareto_efficiency, multiobjective_optimization, risk_return_tradeoff). Second, and deeper, the curve's most distinctive lesson — that the relation shifts when you act on it — is one instance of the Lucas-critique pattern: a policy-conditional historical correlation, estimated under one regime, that breaks when the regime changes. That pattern recurs as genuine co-instances far from economics: factor returns reversing after academic publication (McLean and Pontiff), distribution shift under model deployment in machine learning, broken-windows policing whose effect is contingent on conditions, treatment effects that fail to generalize in medicine. Both general patterns travel and are the things that should carry the cross-domain lesson — the trade-off-frontier family for the static shape, the Lucas-critique / regime-conditional-relationship pattern for the reflexive instability — not "the Phillips curve" as named.

The home-bound cargo is the labour-and-price machinery that gives the curve its empirical content: wage-price spirals, the NAIRU, the expectations-augmentation formal apparatus, supply-shock shifters keyed to oil prices and terms of trade, and the monetary-policy framing. None of that survives extraction — there is no wage bargaining, no natural rate, no central bank in a generic trade-off — so invoking "a Phillips curve" for growth-versus-environment, innovation-versus-stability, or security-versus-liberty borrows the inverse-relation-that-shifts shape while dropping the wage-price-expectations mechanism, and should be marked as analogy with the genuine content carried by the two parents. A caution travels with the concept and is worth stating in any cross-domain use, because it is the concept's own deepest point: the relation is non-stationary under intervention, so reading a curve fitted to past data as a stable menu one can pick a permanent point on is exactly the error that burned 1960s policymakers — and importing the curve elsewhere as if it were structural risks repeating that error one substrate removed. Mechanism within macroeconomics, two-parent recurrence (trade-off frontier plus Lucas-critique) plus metaphor beyond — the profile Structural Core vs. Domain Accent makes precise.

Examples

Canonical

The original finding is A. W. Phillips's 1958 study. Examining nearly a century of British data on the rate of change of nominal wages against the unemployment rate — roughly 1861 to 1957 — Phillips found a stable, nonlinear inverse relationship: years of low unemployment coincided with rapid wage growth, and years of high unemployment with flat or falling wages, tracing a downward-sloping curve steepening sharply as unemployment approached very low levels. Samuelson and Solow generalized it to U.S. price inflation in 1960 and cast it as a policy menu — the inflation a government could "buy" with a chosen unemployment rate. This clean empirical regularity, fitted across many decades, became the foundational object; its very stability over that historical window is what tempted 1960s policymakers to treat it as a permanent trade-off they could pick a point on.

Mapped back: Phillips's unemployment variable is the labour-market tightness axis and wage growth is the inflation axis; the fitted downward-sloping locus is the short-run trade-off curve itself, whose steepening at low unemployment prefigures its slope structure. That the curve looked stable across 1861-1957 is exactly what made the reflexive non-stationarity invisible until policy tried to exploit it.

Applied / In Practice

Central banks put the shift-parametrized curve to work interpreting the 2021-2022 inflation surge. For decades the curve had appeared nearly flat — unemployment fell to historic lows through the 2010s with inflation barely moving — which policymakers read as anchored expectations compressing the slope, not the relationship disappearing. When inflation then jumped sharply in 2021 amid pandemic supply-chain disruptions, an energy shock, and a tight labour market, the framework structured the debate precisely: was this a temporary outward displacement of the curve by supply shocks (transitory, self-reversing), or were inflation expectations beginning to de-anchor and shift the curve's position up (persistent, requiring aggressive tightening)? The Federal Reserve's pivot to rapid rate hikes reflected a judgment that the second risk was rising — that letting high inflation persist would move expectations and steepen the trade-off — exactly the reasoning the expectations-augmented curve prescribes.

Mapped back: The pandemic energy and supply-chain disruptions are the supply-shock displacement pushing the curve outward; the worry that persistent inflation would raise the expectations position-shifter and unwind the anchoring slope-control is the diagnostic distinction between shifting the curve and sliding along it. The Fed's tightening was an intervention aimed at the parameter — expectations — rather than merely at today's point on the curve.

Structural Tensions

T1: Slide along versus shift of the curve (the master distinction that the data hide). The whole analytical content turns on separating a movement along a fixed short-run curve (a demand-side slide exploiting the momentary trade-off) from a movement of the curve (expectations or supply shifting the whole relation). But a single inflation-unemployment observation does not announce which it is; only the shape and motion of the scatter, plus a judgment about expectations and shocks, distinguish them. The tension is that the two produce overlapping data yet demand opposite readings — a slide is exploitable and self-limiting, a shift is not — so an analyst who mistakes a shift for a slide reads a permanent menu where there is none. The distinction is indispensable and never directly observable; it must be inferred from the very parameters whose movement it is trying to detect. Diagnostic: Is this inflation-unemployment change a slide along a fixed short-run curve (exploit the trade-off) or a shift of the curve itself (expectations or a supply shock moving the whole relation)?

T2: Real short-run trade-off versus vertical long-run (the horizon at which the menu dissolves). There genuinely is a short-run trade-off — surprise inflation buys lower unemployment for a while — and that transient reality is what tempts treating it as durable. But the trade-off holds only for unanticipated inflation, so as agents adapt, the curve shifts up until, at the natural rate, the long-run curve is vertical and any chosen inflation returns the economy to the same unemployment. The tension is that the short-run gain is real and immediate while the long-run neutrality arrives later and diffusely, so a policymaker optimizing on the visible horizon is structurally drawn toward exploiting a trade-off that the model says will only ratchet inflation with no lasting employment gain. Both horizons are true; they simply disagree about whether the trade-off exists. Diagnostic: Is the trade-off being claimed over a horizon short enough that inflation stays a surprise, or extended into the long run where the curve is vertical and the trade-off illusory?

T3: Exploitable regularity versus self-destroying under exploitation (the Lucas-critique reflexivity). The curve is the canonical case in which acting on an empirical regularity changes it: a relation fitted under one policy regime shifts when policy tries to ride it, because agents re-form expectations around the new policy. This is the concept's deepest lesson and its deepest hazard — the very stability that made the 1861-1957 curve look like a permanent menu is what dissolved once the menu was exploited. The tension is that a well-estimated historical curve is simultaneously the best available guide and a booby-trap: it describes the past accurately and predicts the future only until it is acted upon. Treating a fitted curve as structural invites exactly the reflexive collapse that burned 1960s policymakers. Diagnostic: Would acting on this fitted relationship change the expectations that generated it — and is the curve being extrapolated as if exploitation would leave it intact?

T4: Anchoring as policy success versus the trade-off it erases (a flat curve is both). Well-anchored expectations flatten the curve, which is a genuine achievement — credible central banks damp the inflation response to labour-market tightening, so tight labour markets no longer stoke inflation. But the same flattening erases the exploitable trade-off: a flat curve means monetary policy buys little inflation and offers little employment leverage through the inflation channel, and it can be misread as the relationship having vanished. The tension is that the slope carries opposite meanings — a steep curve is an exploitable menu but an inflation-prone regime; a flat curve is a stable regime but a dead menu — so the very success of anchoring removes the trade-off that the curve was originally prized for describing. Reading the flat post-1990s curve as "the Phillips curve is gone" mistakes a slope compression for a disappearance. Diagnostic: Is a flat curve here evidence the relationship has vanished, or evidence anchored expectations have compressed its slope — the signature of policy success, not of the mechanism's absence?

T5: A framework that absorbs every episode versus its own falsifiability. The shift-parametrized curve accommodates the entire historical record: the 1960s menu is a stable slide, 1970s stagflation is an outward supply shift, the post-1990s flattening is anchored expectations compressing the slope, the 2021 surge is a shock plus incipient de-anchoring. This is genuine explanatory reach — one curve at different settings, not a relationship that keeps failing. But the same flexibility is a hazard: with four parameters (position, slope, asymptote, shock) free to move, almost any inflation-unemployment pattern can be rationalized after the fact, so the framework risks explaining everything and predicting little. The tension is that the power to absorb anomalies by re-setting a parameter is inseparable from the difficulty of ever being surprised by the data. Diagnostic: Is the framework making a falsifiable forward prediction here, or absorbing an observed episode after the fact by attributing it to whichever parameter moved?

T6: Autonomy versus reduction (a macro relation or an instance of two distinct parents). The Phillips curve is a named, content-rich macro relation — wage-price spirals, the NAIRU, expectations-augmentation, oil-keyed supply shifters, the monetary-policy framing — and within macroeconomics it transfers as mechanism richly. But its cross-domain content splits into two parents rather than one: the surface form (two desired states in inverse relation) belongs to the trade_offs/pareto_efficiency frontier family, and its deepest lesson (a policy-conditional relation that shifts when acted upon) belongs to the Lucas-critique / regime-conditional-relationship pattern, which recurs in published factor returns decaying, ML distribution shift under deployment, and non-generalizing treatment effects. Neither is "the Phillips curve." The tension is between a macro relation that earns its own standing and the recognition that its static shape and its reflexive instability are carried by two different, more general patterns, with the wage-price machinery staying home. Diagnostic: Resolve toward the trade-off-frontier family for the static inverse-relation shape and toward the Lucas-critique pattern for the shifts-when-acted-upon insight; toward the named Phillips curve when reasoning about inflation, unemployment, and expectations in an actual economy.

Structural–Framed Character

The Phillips curve sits in the middle of the structural–framed spectrum — best read as mixed — further toward framed than a mixed-structural mechanism like isostasy, but well short of the framed pole where the named fallacies sit, because it is a descriptive relation rather than a verdict. On evaluative weight it reads structural: the curve names an inverse association between labour-market tightness and inflation, a neutral relation that convicts nothing; the normative caution attached to it ("do not treat it as a permanent menu") is advice about using the object, not a judgment baked into it. But human-practice-bound pulls hard toward framed, and does so in an unusually deep way: the relation is constituted by expectation-forming agents, wage bargaining, and a central bank, and it is reflexively non-stationary — a policy-conditional correlation that shifts the moment policy acts on it (the Lucas critique made concrete). Remove the judging, expecting, policy-setting agents and there is no curve at all; nothing in nature traces it. Institutional origin is mixed: the underlying wage-price dynamics are real behaviour among economic agents, but the curve as a modelled object — the NAIRU, the expectations-augmentation apparatus, the vertical long-run asymptote — is theoretical furniture erected by Phillips, Samuelson–Solow, and Friedman–Phelps within macroeconomics. On vocab-travels it scores low: wage-price spiral, natural rate, expectations-augmentation, supply-shock shifters, and the monetary-policy framing are all pinned to the domain. And import-vs-recognize is bimodal — within macroeconomics the shift-parametrized relation transfers as recognized mechanism across Fed, ECB, and BoJ analysis; beyond it, "a Phillips curve" for growth-versus-environment or security-versus-liberty is import-by-analogy that renames every part.

Unusually, the curve offers two structural-looking skeletons rather than one, and neither lifts it off its position because both are what it instantiates from parents. Its static surface form — two desired states in inverse relation — is one instance of the two-objective trade-off frontier carried by trade_offs, pareto_efficiency, and multiobjective_optimization; its deepest lesson — a fitted relation that shifts when acted upon — is one instance of the Lucas-critique / regime-conditional-relationship pattern, which recurs far from economics (published factor returns decaying, distribution shift under model deployment, non-generalizing treatment effects). Both patterns genuinely travel, but the travelling belongs to them; the labour-and-price machinery that makes it "the Phillips curve" rather than a bare frontier is exactly the part that stays home. Its character: an evaluatively neutral but heavily practice-bound macroeconomic relation, reflexively constituted by the very agents it describes, structural only in the two general patterns — a trade-off frontier and a regime-conditional relation — it instantiates and dresses in wage-price vocabulary that does not travel.

Structural Core vs. Domain Accent

This section decides why the Phillips curve is a domain-specific abstraction and not a prime, and it carries the case for its domain-specificity. What is distinctive here is that the portable core is doubled — the curve offers two skeletons, not one — so being exact about both, and about the wage-price machinery neither of them includes, is the whole task.

What is skeletal (could lift toward a cross-domain prime). Strip the macroeconomics and two thin relational structures survive, each portable in its own right. The first is the curve's static surface form: two desired states standing in inverse relation, so that gaining on one is paid for in the other — a two-dimensional trade-off frontier. The second, and deeper, is the curve's reflexive lesson: a fitted, policy-conditional relation, estimated under one regime, that shifts the moment it is acted upon. Both are genuinely substrate-portable. The frontier shape recurs as Pareto frontiers, multi-objective optimization, the risk-return trade-off, the fast-cheap-quality triangle; the shifts-when-acted-upon pattern recurs far from economics — published factor returns decaying after publication, distribution shift under model deployment, non-generalizing treatment effects. Each is the core the curve shares, and each is exactly why it recurs as one of the two families of parent prime the curve instantiates — but neither is what makes it the Phillips curve.

What is domain-bound. Everything between the two skeletons — the empirical content that makes the object "the Phillips curve" rather than a bare frontier or a bare regime-conditional correlation — is macroeconomic furniture that does not survive extraction. The labour-market tightness and inflation axes; the wage-price spiral; the expectations position-shifter and the anchoring slope-control; the natural rate / NAIRU at which the long-run curve goes vertical; the supply-shock displacement keyed to oil prices and terms of trade; the expectations-augmentation apparatus and the whole monetary-policy framing. These are the worked vocabulary, the instruments, and the empirical cases — the 1960s menu, 1970s stagflation, the post-1990s flattening, the 2021 surge — and they are all specific to an economy with wage bargaining, an inflation process, and a central bank. The decisive test: there is no wage bargaining, no natural rate, and no central bank in a generic two-objective trade-off, so remove the labour-and-price substrate and what is left is a bare inverse relation that shifts — no longer the Phillips curve, but one of its two parents wearing none of its accent.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose cross-domain transfer is recognition of the same mechanism, not analogy. The Phillips curve's transfer is bimodal. Within macroeconomics the shift-parametrized relation travels intact as recognized mechanism — the same slide-versus-shift diagnostics, NAIRU, and expectations-augmentation organize Fed, ECB, and Bank of Japan analysis across the flattening and the 2021–22 surge alike. Beyond macroeconomics it travels only by analogy: "a Phillips curve" for growth-versus-environment, innovation-versus-stability, or security-versus-liberty borrows the inverse-relation-that-shifts shape while renaming every part and dropping the wage-price-expectations mechanism that gives the object its content. And when the bare structural lesson is needed cross-domain, it is already carried, in more general and correctly-split form, by the two families of parent the curve instantiates: the static inverse-relation shape belongs to the trade-off-frontier family (trade_offs, pareto_efficiency, multiobjective_optimization, risk_return_tradeoff), and the reflexive instability belongs to the Lucas-critique / regime-conditional-relationship pattern. Both parents travel; the travelling belongs to them. One caution rides along and is worth stating in any cross-domain use, because it is the concept's own deepest point: the relation is non-stationary under intervention, so importing "a Phillips curve" as if it were a stable structural menu risks repeating, one substrate removed, exactly the error that burned 1960s policymakers. The cross-domain reach belongs to the two parents; "the Phillips curve," as named, carries the macroeconomic baggage that should stay home.

Relationships to Other Abstractions

Local relationship map for Phillips CurveParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.Phillips CurveDOMAINDomain-specific abstraction: Inflation — is part ofInflationDOMAINDomain-specific abstraction: Natural Rate of Unemployment — is part ofNatural Rateof UnemploymentDOMAINPrime abstraction: Trade-offs — is a decomposition ofTrade-offsPRIME

Current abstraction Phillips Curve Domain-specific

Parents (3) — more general patterns this builds on

  • Phillips Curve is part of Inflation Domain-specific

    The Phillips curve contains inflation as its vertical measured variable, expectations object, surprise term, and long-run policy outcome.

  • Phillips Curve is part of Natural Rate of Unemployment Domain-specific

    The expectations-augmented Phillips curve contains the natural rate as its vertical long-run asymptote and stable-inflation unemployment anchor.

  • Phillips Curve is a decomposition of Trade-offs Prime

    Removing wage-price vocabulary from the Phillips curve leaves a short-run inverse relation in which improvement on one desired axis costs movement on another.

Hierarchy paths (3) — routes to 2 parentless roots

Not to Be Confused With

  • Okun's law. A different empirical macro relation on the same labour-market variable: it links the unemployment rate to the output/GDP gap — how much output an economy forgoes per point of unemployment above its natural rate. It shares the Phillips curve's tightness axis (unemployment, the output gap) but pairs it with real output, not with the inflation axis. Confusing the two collapses the real side (output lost to slack) into the nominal side (inflation bought by tightness). Tell: does the relation put unemployment against inflation (Phillips curve, a nominal trade-off) or against real output/GDP (Okun's law, a real-side accounting)? One prices tightness in inflation, the other in lost output.

  • Beveridge curve. Another eponymous, downward-sloping labour-market curve — hence the easy mix-up — but it relates job vacancies to unemployment, mapping the efficiency of labour-market matching, not the inflation consequences of tightness. It lives entirely on the tightness side of the Phillips relation and never touches the inflation axis. Tell: is the vertical variable inflation (Phillips curve) or the vacancy rate (Beveridge curve)? Both slope down and both are named after economists; only the Phillips curve is about what tightness does to prices.

  • The NAIRU / natural rate (a part, not the whole). The natural rate of unemployment is one feature of the curve — the location of its vertical long-run asymptote, the unemployment rate to which the economy returns regardless of chosen inflation — not the relation itself. Treating "the NAIRU" as synonymous with "the Phillips curve" mistakes a single reference point for the whole shift-parametrized locus (position, slope, asymptote, and shock displacement). Tell: is the object a single unemployment rate at which inflation is stable (the NAIRU, the vertical asymptote), or the entire downward-sloping short-run relation that swings around it as expectations and shocks move (the Phillips curve)? The natural rate is where the curve goes vertical, a component of it.

  • The Lucas critique (the reflexive-instability parent). The general lesson that a policy-conditional historical relationship, estimated under one regime, shifts when the regime changes — and the Phillips curve is its canonical worked example, not the pattern itself. The critique recurs far from economics (published factor returns decaying after publication, distribution shift under model deployment, non-generalizing treatment effects); the Phillips curve is the single macro instance keyed to inflation, unemployment, and expectations. Tell: is the claim the general "fitted relations shift when acted upon" (the Lucas critique, the portable pattern) or specifically the inflation–unemployment relation with its NAIRU and wage-price content (the Phillips curve, one instance)? The reflexive instability travels under the parent; the wage-price machinery stays home.

  • The trade-off / Pareto frontier family (the static-shape parent). The other umbrella the curve instances — two desired states standing in inverse relation, so gaining on one is paid for in the other (trade_offs, pareto_efficiency, multiobjective_optimization, risk_return_tradeoff). The Phillips curve's surface form is one member of this family, but a bare frontier has no expectations-shifter, no natural rate, and no supply-shock displacement, and — unlike the Phillips curve — it does not shift when you act on it. Tell: is it a fixed menu of attainable trade-off points (the frontier family, static) or a relation that moves as expectations adapt and policy is exploited (the Phillips curve, whose whole content is the shift)? Importing "a Phillips curve" for growth-versus-environment or security-versus-liberty borrows only this static shape and drops the mechanism — the genuine content is carried by the frontier family, treated more fully elsewhere.

  • Short-run aggregate supply (SRAS). The near-dual macro relation: SRAS plots the price level against real output, and the short-run Phillips curve is essentially its re-expression with the axes recast as inflation against unemployment (via Okun's link between output and unemployment). They move together and encode the same demand-side mechanism, which invites treating them as one, but they are stated over different variable pairs and belong to different diagrams. Tell: are the axes price level and output (SRAS) or inflation and unemployment (Phillips curve)? Same underlying short-run mechanism, two coordinate systems — reading a result proved in one as automatically holding in the other silently swaps the axes.

Neighborhood in Abstraction Space

Phillips Curve sits in a crowded region of the domain-specific corpus (22nd percentile for distinctiveness): several abstractions share nearly its structure, so a description that fits it tends to fit its neighbors too.

Family — Macroeconomic Puzzles & Long-Run Relations (5 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-07-12