Skip to content

Feldstein-Horioka Puzzle

The anomaly that national saving and investment rates are strongly correlated across countries when frictionless capital mobility predicts near-zero — turning the regression slope into a continuous gauge of de facto capital-market integration.

Core Idea

The Feldstein-Horioka puzzle is an empirical anomaly in international macroeconomics: the cross-country correlation between national saving rates and national investment rates is strikingly and persistently high — approximately 0.89 in the original Feldstein and Horioka (1980) panel of OECD countries over 1960–1974, and substantially positive in modern replications — when standard open-economy theory predicts it should be near zero. The theoretical prediction follows from the logic of frictionless capital mobility: if global capital markets are well integrated, saving generated in one country should flow to wherever returns are highest rather than being retained domestically, so domestic investment should be funded from the global pool and domestic saving should bear little relationship to domestic investment within any single country. The high empirical correlation implies the opposite — that saving and investment within national borders are strongly coupled — and forces the conclusion that capital-market integration is far more limited and frictional than textbook arbitrage models assume. The puzzle's structural payoff is that it converts a fuzzy question — how integrated are global capital markets? — into a tractable diagnostic: the slope coefficient from a cross-country regression of the investment-to-GDP ratio on the saving-to-GDP ratio, sometimes called the Feldstein-Horioka coefficient, serves as a continuous measure of the gap between de jure capital mobility (the regulatory and institutional openness of the capital account) and de facto capital mobility (the actual responsiveness of investment to global returns). A coefficient near zero would indicate full integration; a coefficient near one indicates investment is largely domestically financed. The literature has proposed a range of candidate mechanisms to explain the residual correlation — country-size effects, investor home bias, current-account targeting by policy authorities, nontradable goods, intertemporal budget constraints, exchange-rate risk, and political-economy constraints on external imbalances — without any single mechanism fully closing the gap, and the modern consensus treats it as structural evidence that frictions systematically limit the degree of capital reallocation across borders.

Structural Signature

Sig role-phrases:

  • the open-economy panel — a cross-section of open economies and time periods, the unit of analysis
  • the national saving rate and the national investment rate — the two measured flows (as ratios to GDP) whose relationship is at issue
  • the frictionless-arbitrage benchmark — the standard open-economy prediction that, with integrated capital markets, saving flows to the highest global return, so the slope should be near zero
  • the Feldstein-Horioka coefficient — the slope of the cross-country regression of investment-to-GDP on saving-to-GDP, the diagnostic statistic pressed into service as a continuous gauge
  • the [0,1] level reading — near zero is the frictionless ideal; near one is investment overwhelmingly tethered to domestic saving, with the position metering de facto integration
  • the de-jure-vs-de-facto gap it meters — the divergence between regulatory openness on paper and investment's actual responsiveness to global returns, read off a high coefficient in an open country
  • the theory-filter (engineered bite) — the persistently steep slope rules out the entire class of frictionless-arbitrage models in one stroke, converting "what explains flows?" into "which frictions reproduce a slope this large?"
  • the unresolved-residual limitation — no single candidate friction (home bias, current-account targeting, nontradables, exchange-rate risk, political-economy limits) drives the coefficient to zero, so the construct measures the gap without naming its cause

What It Is Not

  • Not a free-standing empirical law. It is a puzzle only relative to one theoretical apparatus — open-economy macro's frictionless-mobility prediction of a near-zero slope. The high saving-investment correlation is "anomalous" because a specific model says it should not occur; absent that benchmark there is no puzzle, just a correlation.
  • Not a causal coupling of saving to investment. The high coefficient does not show that domestic saving funds or causes domestic investment within a country; it shows the two are statistically tethered more tightly than frictionless arbitrage predicts. The inference is about capital not reallocating freely across borders, not about a within-country financing mechanism.
  • Not a measurement of which friction is binding. The coefficient meters de facto integration and the de-jure/de-facto gap, but it does not name its cause. Home bias, current-account targeting, nontradables, exchange-rate risk, and political-economy limits are candidate explanations; the construct measures the gap without identifying which combination produces it.
  • Not closed by any single proposed mechanism. No candidate friction, on its own, drives the coefficient to zero — the standing result is precisely that integration frictions are multiple and compounding. Treating any one explanation as "the" resolution misreads a literature whose compact finding is that none singly suffices.
  • Not the home-bias puzzle. That sibling anomaly concerns portfolio holdings (investors holding mostly domestic assets); Feldstein-Horioka concerns saving-investment flows. The two are theoretically linked and often paired, but the measured quantity differs, and conflating them mistakes one open-economy puzzle for another.

Scope of Application

Because the Feldstein-Horioka coefficient is at bottom a constructed diagnostic statistic — a regression slope, not a causal mechanism — its habitats are wherever its precondition holds: national saving rates and national investment rates measured across a panel of open economies, read against the frictionless-arbitrage benchmark of a near-zero slope. That precondition confines it to international macroeconomics and open-economy finance, and the contexts below are genuine literal uses of the identical statistic; the loose insight that "apparent integration can hide frictions" belongs to the parent friction/integration primes, not here. - Capital-mobility measurement — the canonical use: the slope serves as a continuous indicator of de facto capital-market integration, comparable across country samples and across decades. - De jure vs. de facto integration analysis — the same coefficient meters the gap between regulatory openness on paper and investment's actual responsiveness to global returns, read off a high slope in a country with a liberalized capital account. - Eurozone-integration studies — movement of the coefficient tracks financial integration across eurozone members before, during, and after the sovereign-debt crisis, turning "is integration advancing?" into the sign of the slope's change. - Open-economy macro modeling and DSGE calibration — the steep slope rules out frictionless-arbitrage models in one stroke and disciplines the calibration of frictions in models that include international capital flows. - Current-account sustainability analysis — the within-border tethering of investment to saving doubles as a ceiling on how far external imbalances travel, bounding the size of sustainable current-account deficits. - The open-economy "puzzles" family — it is reasoned alongside its sibling anomalies (the home-bias puzzle on portfolio holdings, the consumption-correlation puzzle, the Backus-Kehoe-Kydland puzzle, and the broader Obstfeld-Rogoff six puzzles), which share its international-flows footprint and benchmark logic.

Clarity

Naming the Feldstein-Horioka puzzle converts a vague and almost unfalsifiable question — how integrated are global capital markets? — into a single tractable statistic that the field can measure, compare, and argue over: the slope coefficient from a cross-country regression of the investment-to-GDP ratio on the saving-to-GDP ratio. That is the puzzle's main clarifying act. Without it, capital-market integration is assessed by an unwieldy mixture of regulatory openness, anecdote, and flow data; with it, an international macroeconomist has a continuous gauge whose endpoints are interpretable — a coefficient near zero would mark the frictionless ideal in which domestic saving flows to the highest global return, a coefficient near one marks investment that is overwhelmingly domestically financed. Tracking that number across decades or across a deepening currency union (the eurozone before, during, and after the sovereign-debt crisis) turns "is integration advancing?" into a question with an answer.

Its second contribution is to sharpen the distinction between de jure and de facto capital mobility — between the legal and institutional openness of the capital account and the actual responsiveness of investment to global returns. The high correlation makes vivid that the two can diverge sharply: a country can be open on paper while its investment remains tethered to its own saving, so the regression coefficient becomes a measure of exactly that gap. The puzzle also disciplines theory by acting as a filter on admissible models. A flat saving-investment relationship is what frictionless-arbitrage open-economy models predict; the persistently steep one rules them out and forces any candidate account — home bias, current-account targeting, nontradables, exchange-rate risk, political-economy limits on external imbalance — to confront a number none of them, singly, has been able to drive to zero. The sharper question a practitioner can now ask is not whether frictions exist but which frictions, and in what combination, can reproduce a coefficient this large.

Manages Complexity

Capital-market integration is, before the puzzle, a sprawling and ill-posed object: a researcher wanting to know how integrated global markets are confronts capital-account regulations that differ line by line across jurisdictions, gross and net flow data that move for a dozen reasons, withholding taxes, exchange-rate regimes, the depth of cross-border banking, anecdotes of home bias, and no agreed way to weigh any of it into a verdict — and the verdict would still differ by country, by asset class, by decade. The Feldstein-Horioka puzzle collapses that entire question to a single scalar the analyst can actually track: the slope coefficient from a cross-country regression of the investment-to-GDP ratio on the saving-to-GDP ratio. The endpoints fix its reading — a coefficient near zero is the frictionless ideal in which domestic saving flows to the highest global return and investment is funded from the world pool; a coefficient near one is investment overwhelmingly tethered to domestic saving — so the high-dimensional "how integrated are the markets?" problem reduces to where on that [0,1] interval the number falls, and "is integration advancing?" becomes the slope of that number through time (across decades, or across a deepening currency union before, during, and after a sovereign-debt crisis). The same coefficient simultaneously meters the otherwise-slippery gap between de jure mobility (openness on paper) and de facto mobility (investment's actual responsiveness to global returns), so two distinct constructs collapse onto one tracked quantity.

The compression also operates on the theory side, where the coefficient works as a filter that spares the analyst from evaluating each candidate model on its own terms. A flat saving-investment relationship is the unique prediction of frictionless-arbitrage open-economy models; the observed steep one therefore rules that whole class out in a single stroke, and converts the open-ended question "what explains international capital flows?" into the bounded one "which frictions, in what combination, can reproduce a coefficient this large?" Every candidate account — country-size effects, home bias, current-account targeting, nontradables, intertemporal budget constraints, exchange-rate risk, political-economy limits on external imbalance — is forced to confront the same number, and the standing finding that none singly drives it to zero is itself a compact summary of a large literature. So rather than re-adjudicate the integration question case by case or model by model, the practitioner tracks one regression coefficient, reads de facto integration and the de-jure/de-facto gap off its level, reads the integration trend off its movement, and reads model admissibility off whether a candidate can generate it.

Abstract Reasoning

The Feldstein-Horioka puzzle turns one regression coefficient into an inferential instrument, and every reasoning move it licenses works by reading that single number — its level, its movement, or its mere existence above zero.

The foundational move is diagnostic measurement of a hidden quantity. Capital-market integration is not directly observable, but the puzzle infers it from a surface signature: the slope of the cross-country regression of the investment-to-GDP ratio on the saving-to-GDP ratio. The reasoning runs from "investment within national borders tracks domestic saving" to "capital is not in fact reallocating freely across borders," because under frictionless arbitrage domestic saving would flow to the highest global return and the slope would be near zero. The analyst thus reads de facto integration off where the coefficient falls on the [0,1] interval — near zero is the frictionless ideal, near one is investment overwhelmingly tethered to domestic saving — and infers an unobservable structural property of the world from a measurable statistical one.

A closely paired move is diagnosing the de jure/de facto gap. Because a country can be fully open on paper while its investment stays bound to its own saving, the coefficient meters precisely the divergence between regulatory openness and actual capital responsiveness. Reasoning from a high coefficient in a country with a liberalized capital account, the analyst concludes that legal openness has not produced behavioral integration — that frictions other than formal barriers are binding — a conclusion invisible without the number, since neither the regulations nor the flow data alone reveal it.

The third move is theory-filtering by a single decisive prediction. A flat saving-investment relationship is the unique signature of frictionless-arbitrage open-economy models; the persistently steep one therefore rules that entire class out in one stroke, without case-by-case adjudication. This converts the open-ended "what explains international capital flows?" into the bounded "which frictions, in what combination, can reproduce a coefficient this large?" — and forces every candidate account (home bias, current-account targeting, nontradables, exchange-rate risk, intertemporal budget constraints, political-economy limits on external imbalance) to confront the same number as a hard target. The standing result that none singly drives it to zero is itself a compact inference about the world: integration frictions are multiple and compounding, not reducible to one cause.

The fourth move is predictive about integration trends through time. If markets are genuinely integrating, the analyst predicts the coefficient should fall, so its trajectory becomes a forecastable indicator: tracking the slope across decades, or across a deepening currency union before, during, and after a sovereign-debt crisis, turns "is integration advancing or reversing?" into a question answered by the sign of the coefficient's movement. A coefficient that fails to fall despite formal liberalization is read as evidence that de facto integration has stalled even as de jure openness advanced.

Finally, the puzzle supports a boundary-drawing move on external imbalances. Because saving and investment co-move strongly within borders, the coefficient bounds how far current-account deficits typically travel — a large slope implies that domestic investment cannot be financed far out of line with domestic saving, so the analyst reads a constraint on sustainable external imbalance directly off the same correlation that meters integration. The reasoning is that the empirical tethering of investment to saving is also a ceiling on how persistently a country can run its investment ahead of its saving on the global market.

Knowledge Transfer

The Feldstein-Horioka puzzle is at bottom a diagnostic statistic — a regression coefficient pressed into service as a continuous gauge of capital-market integration — wrapped in the theoretical surprise of an anomaly, so its transfer is governed by an instrument's precondition rather than by "mechanism within / metaphor beyond." The construct is computable, and means what it is supposed to mean, wherever its two ingredients exist: national saving rates and national investment rates measured across a panel of open economies, read against the frictionless-arbitrage benchmark of a near-zero slope. Wherever both hold, the coefficient transfers literally and as itself.

Within international macroeconomics and open-economy finance it travels broadly on exactly that footing. The Feldstein-Horioka coefficient is the canonical capital-mobility indicator, comparable across country samples and decades; its movement is the tracked signal in eurozone-integration studies (before, during, and after the sovereign-debt crisis); it disciplines the calibration of frictions in open-economy DSGE models; it bounds current-account sustainability (the within-border tethering of investment to saving is also a ceiling on how far external imbalances travel); and it sits in a tight family of sibling open-economy anomalies it reasons alongside — the home-bias puzzle (portfolio holdings rather than saving-investment flows), the consumption-correlation puzzle, the Backus–Kehoe–Kydland puzzle, the broader Obstfeld–Rogoff "six puzzles." Across all of these the same level-reading (where the slope falls on [0,1]), the same de-jure-versus-de-facto gap metering, the same theory-filter (the steep slope rules out frictionless-arbitrage models in one stroke), and the same trend-reading (a falling coefficient signals deepening integration) carry without translation. This is instrument transfer in the strict sense: the same statistic, computed the same way, meaning the same thing, wherever cross-border saving and investment are measured.

Beyond international financial flows the reach is essentially nil, and honesty requires saying so flatly: the puzzle exists only relative to one theoretical apparatus (open-economy macro's frictionless-mobility prediction), and there is no second substrate — in physics, biology, computing, or elsewhere — where "the Feldstein-Horioka puzzle" arises in any non-metaphorical sense. Invoking a "Feldstein-Horioka coefficient" for, say, the correlation between locally-raised and locally-spent funds in some non-monetary system would be borrowing the shape of the statistic while lacking the saving/investment/capital-mobility content that gives it meaning — over-reading, not the instrument traveling. What does generalize, and what should be carried when the lesson is wanted elsewhere, is not the puzzle but the general fact it measures — apparent market integration can conceal substantial frictions — which is already carried at the prime level by frictions, market_integration, and arbitrage_finance. Strip the national-accounting vocabulary and the residual is just "two quantities are more correlated than a frictionless model predicts," i.e. frictions matter — a parent-level concept, not the Feldstein-Horioka puzzle. So the honest split is between instrument-reach (the coefficient is computable, and informative, wherever cross-border saving and investment are measured) and over-reading (there is no genuine cross-substrate analogue, and the portable insight belongs to the friction/integration primes). The full split is drawn in Structural Core vs. Domain Accent.

Examples

Canonical

Feldstein and Horioka's original 1980 study built the instrument. They took a cross-section of sixteen OECD countries and, for each, averaged its investment-to-GDP ratio and its saving-to-GDP ratio over 1960–1974, then regressed the investment ratios on the saving ratios across countries. Under frictionless capital mobility the slope should be near zero — a country's saving should fund the highest-return projects anywhere, so domestic investment need not track domestic saving. Instead they found a slope of roughly 0.9 (a saving–investment correlation near 0.89): almost every extra dollar of national saving stayed home as national investment. The result was startling precisely because these were exactly the rich, open economies textbook theory expected to be integrated, and it launched a literature by turning "how integrated are capital markets?" into one measurable coefficient.

Mapped back: The sixteen OECD countries over 1960–74 are the open-economy panel; the two averaged ratios are the national saving rate and national investment rate. The near-zero prediction is the frictionless-arbitrage benchmark, and the estimated ≈0.9 slope is the Feldstein-Horioka coefficient landing near one on the [0,1] level reading — investment tethered to domestic saving. Its steepness ruling out frictionless-mobility models in one stroke is the theory-filter (engineered bite).

Applied / In Practice

Economists studying European financial integration use the coefficient as a live gauge. Through the run-up to and early years of the euro, cross-country and within-eurozone Feldstein-Horioka regressions showed the saving-investment slope falling — evidence that monetary union and a common financial market were loosening the domestic tether, letting capital flow to where returns were highest (the peripheral investment booms funded by core saving). Then the 2010–12 sovereign-debt crisis reversed it: as cross-border lending retrenched and capital "came home," measured coefficients rose again, registering financial fragmentation. The single number tracked the deepening and then partial unwinding of integration that regulatory and flow data alone reported only murkily.

Mapped back: The eurozone members across pre-crisis and crisis years are the open-economy panel; the falling-then-rising slope is the Feldstein-Horioka coefficient whose movement reads the integration trend. A slope that fell despite no single formal barrier change, then rose in the crisis, meters the de-jure-vs-de-facto gap — behavioral integration advancing and retreating independently of the legal capital account, exactly the hidden quantity the instrument is built to surface.

Structural Tensions

T1: An anomaly versus a contested benchmark (the puzzle exists only if the frictionless model predicts what it claims). Feldstein-Horioka is a puzzle strictly relative to one prediction — that integrated capital markets imply a near-zero saving-investment slope. Absent that benchmark there is only a correlation, and the benchmark is not beyond dispute. A standing critique is that even under perfect capital mobility saving and investment should be positively correlated, because an intertemporal budget constraint forbids a country from running unbounded external deficits forever, so long-run saving and investment must co-move regardless of frictions. If that is right, a substantial slope is partly what mobility predicts, and the "anomaly" is inflated by an over-strong null. The tension is that the construct's entire status as a puzzle — and the interpretation of a high coefficient as evidence of frictions — is hostage to a theoretical prediction that may itself overstate how low the slope should be. Diagnostic: Is the high coefficient being read as a friction because the frictionless model truly predicts near-zero, or because a solvency-consistent benchmark that predicts a positive slope has been ignored?

T2: A clean scalar gauge versus an uninterpreted residual (measures how much, names no cause). The coefficient's great virtue is collapsing the sprawling "how integrated are capital markets?" into one tracked number on [0,1]. But it meters de facto integration without identifying which friction produces it: home bias, current-account targeting, nontradables, exchange-rate risk, and political-economy limits are all candidates, and the standing result is that none singly drives it to zero. So the gauge is a thermometer, not a diagnosis — it tells a policymaker the gap is large but not what to change to close it. The tension is that the compression which makes integration measurable is exactly what strips out the causal content needed to act, so the more cleanly the number summarizes the state of integration, the less it says about its mechanism. Diagnostic: Is the coefficient being used to gauge the magnitude of the integration gap, or over-read as if it identified which friction is binding — a cause it does not contain?

T3: The slope as a mobility gauge versus a slope with many generators (the confounded instrument). The coefficient is interpreted through a single channel — capital-market integration — but the same slope can be manufactured by processes unrelated to frictions. Common shocks that move saving and investment together (productivity, demographics, the business cycle) inflate the correlation; large countries mechanically show higher coefficients because their own returns move the world rate; and current-account targeting means governments may actively hold saving and investment in line, producing a high slope by policy rather than by immobility. A coefficient that falls or rises can therefore reflect changing shock structure or policy behavior rather than changing integration. The tension is that the instrument claims to surface one hidden quantity while the observable slope is a convolution of several, so reading integration off it presumes away confounds the number cannot separate. Diagnostic: Is the slope's level or movement being attributed to capital mobility, or could common shocks, country size, or deliberate current-account targeting be generating the same coefficient?

T4: Statistical tethering versus causal financing (the correlational instrument read as a within-country mechanism). The construct's legitimate inference is cross-border and correlational: investment tracks domestic saving more tightly than frictionless arbitrage predicts, so capital is not reallocating freely across borders. But the vivid gloss — "national saving stays home to fund national investment" — slides into a within-country causal claim the coefficient does not support: it shows tethering, not that domestic saving finances or causes domestic investment. The two readings diverge sharply, because a high correlation is consistent with saving and investment both responding to a third driver with no financing link between them. The tension is that the instrument's entire payload is a statement about the absence of cross-border reallocation, yet its natural narrative reaches for a domestic financing mechanism it is not licensed to assert. Diagnostic: Is the claim that investment is statistically tethered to saving (what the coefficient shows), or that domestic saving causally funds domestic investment (what it does not)?

T5: Autonomy versus reduction (a named open-economy puzzle or the friction/integration primes it measures). As a diagnostic statistic the Feldstein-Horioka coefficient transfers literally wherever its two ingredients exist — national saving and investment rates across a panel of open economies read against the near-zero benchmark — and within international macro it travels broadly as itself, reasoned alongside its sibling anomalies (home-bias, consumption-correlation, Backus-Kehoe-Kydland). But beyond international financial flows its reach is essentially nil: the puzzle exists only relative to open-economy macro's mobility prediction, and there is no second substrate where it arises non-metaphorically. What generalizes is not the puzzle but the general fact it measures — apparent market integration can conceal substantial frictions — already carried at the prime level by frictions, market_integration, and arbitrage_finance. The tension is between a genuinely useful, literally-computable instrument in situ and the recognition that its portable insight belongs to those friction/integration primes, with the saving/investment/capital-account content being home-bound accent. Diagnostic: Resolve toward frictions/market_integration/arbitrage_finance when carrying the "integration hides frictions" lesson outside cross-border capital flows; toward the Feldstein-Horioka coefficient when gauging de facto integration from an actual panel of national saving and investment rates.

Structural–Framed Character

The Feldstein-Horioka puzzle sits at mixed on the structural–framed spectrum — evaluatively neutral, resting on a genuine empirical regularity, yet pinned unusually hard to its home domain by two framing facts: it is a puzzle only relative to a contested theoretical benchmark, and its cross-domain reach is, by the entry's own flat statement, essentially nil. The criteria split. On evaluative weight it is structural: the Feldstein-Horioka coefficient praises and blames nothing — it is a gauge of de facto capital-market integration, and the "puzzle" is a theoretical surprise, not a normative verdict on any agent, so it is nothing like the fallacies elsewhere in this corpus. But on human-practice-bound it leans framed in a specific, telling way: the phenomenon lives on a human economic-institutional substrate (national accounts, capital markets, capital-account regimes, saving and investment flows) rather than in observer-free nature, and its very status as an anomaly is theory-dependent — as T1 makes explicit, absent open-economy macro's frictionless-mobility prediction there is only a correlation, and under a solvency-consistent benchmark a positive slope is partly what mobility predicts. A construct whose puzzle-hood is hostage to a chosen model is framed on this axis in a way a mind-independent mechanism like isostasy is not.

On institutional origin it is likewise framed at the level that matters: while the saving-investment tethering is a real regularity of economies, the puzzle-framing and the diagnostic coefficient are analytical artifacts of the economics discipline (Feldstein and Horioka 1980, the near-zero null, the [0,1] level reading). On vocab-travels it is decisively framed: the operative vocabulary — saving/investment rates, capital mobility, de jure versus de facto integration, current-account sustainability, DSGE calibration — presupposes an international-macro substrate and has no referent outside it. And on import-vs-recognize the entry is blunt: within international finance the coefficient is instrument-transfer in the strict sense (the same statistic, computed identically, meaning the same thing across country samples, decades, and the eurozone), but beyond that substrate there is no genuine analogue — invoking a "Feldstein-Horioka coefficient" elsewhere would borrow the shape while lacking the saving/investment/capital-mobility content, which is over-reading, not the instrument travelling.

The portable structural skeleton is thin and lives entirely in the parents: two quantities are more correlated than a frictionless model predicts, so frictions matter — apparent market integration can conceal substantial frictions, already carried at the prime level by frictions, market_integration, and arbitrage_finance. That skeleton is what the puzzle instantiates from its umbrella, not what makes "Feldstein-Horioka" itself travel: the cross-domain reach (such as it is) belongs to those friction/integration primes, while the saving/investment/capital-account content, the frictionless-arbitrage benchmark, and the whole open-economy-puzzles family stay home. Its character: an evaluatively neutral, empirically grounded diagnostic statistic that is nonetheless a theory-relative puzzle about a human economic-institutional substrate with essentially no cross-domain reach — structural in its neutrality and its literal in-domain instrument-transfer, framed in its theory-dependence and home-bound vocabulary, leaving it mixed rather than a free-floating prime.

Structural Core vs. Domain Accent

This section decides why the Feldstein-Horioka puzzle is a domain-specific abstraction and not a prime — and it is an unusually stark case, because the portable core is thin and the domain accent is nearly the whole construct.

What is skeletal (could lift toward a cross-domain prime). Strip the international-macro apparatus and only a thin relational structure survives: two quantities are more tightly correlated than a frictionless model predicts, so the residual coupling is evidence that frictions limit reallocation — apparent integration can conceal substantial friction. The portable pieces are abstract — a benchmark that predicts decoupling under frictionless flow, an observed coupling that violates it, and the inference that unobserved frictions must be binding. That skeleton is genuinely substrate-portable and is exactly what the puzzle instantiates from its parents — frictions (reallocation is not costless), market_integration (the degree to which separate markets behave as one), and arbitrage_finance (returns should equalize when capital flows freely). But that is the core the puzzle shares — and, notably, it is almost all of the portable content there is.

What is domain-bound. Nearly everything that makes this specifically the Feldstein-Horioka puzzle is international-macro furniture and none of it survives extraction. Its worked content is national-accounting and open-economy-finance vocabulary: national saving and investment rates as ratios to GDP, a panel of open economies, the frictionless-arbitrage benchmark of a near-zero slope, the Feldstein-Horioka coefficient itself as the regression statistic, the [0,1] level reading, the de-jure-versus-de-facto capital-mobility gap, current-account-sustainability bounds, DSGE-friction calibration, and the sibling open-economy puzzles (home-bias, consumption-correlation, Backus-Kehoe-Kydland). The empirical cases (the 1980 OECD panel, the eurozone integration studies) are drawn from it. The decisive test is unusually blunt here: there is no second substrate — in physics, biology, computing, or elsewhere — where "the Feldstein-Horioka puzzle" arises non-metaphorically, and invoking a "Feldstein-Horioka coefficient" for the correlation of locally-raised and locally-spent funds in some non-monetary system borrows the shape of the statistic while lacking the saving/investment/capital-mobility content that gives it meaning. The construct is also, at bottom, a theory-relative puzzle: its very anomaly-status is hostage to a chosen benchmark, which pins it further to the discipline that holds that benchmark.

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 Feldstein-Horioka puzzle's transfer is bimodal and sharply lopsided. Within international macroeconomics and open-economy finance it travels literally, as itself — the same statistic, computed the same way, meaning the same thing wherever national saving and investment are measured across a panel of open economies: capital-mobility measurement, de-jure/de-facto analysis, eurozone-integration tracking, DSGE calibration, current-account sustainability, and the reasoning alongside its sibling anomalies all carry without translation. Beyond that substrate the reach is essentially nil — not merely analogy but no genuine analogue at all, because the construct exists only relative to open-economy macro's mobility prediction. And when the bare structural lesson is wanted elsewhere — apparent market integration can conceal substantial frictions — it is already carried, in more general form, by frictions, market_integration, and arbitrage_finance. The cross-domain reach belongs to those parents; "Feldstein-Horioka," as named, is a home-bound instrument whose saving/investment/capital-account content should stay in international finance.

Relationships to Other Abstractions

Local relationship map for Feldstein-Horioka PuzzleParents 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.Feldstein-HoriokaPuzzleDOMAINPrime abstraction: Calibration Anomaly — is a decomposition ofCalibrationAnomalyPRIMEPrime abstraction: Correlation — is a decomposition ofCorrelationPRIMEPrime abstraction: Frictionless Benchmark Reasoning — is a decomposition ofFrictionless Be…PRIME

Current abstraction Feldstein-Horioka Puzzle Domain-specific

Parents (3) — more general patterns this builds on

  • Feldstein-Horioka Puzzle is a decomposition of Calibration Anomaly Prime

    A model predicts a saving-investment slope near zero, observation produces a persistent slope near one, and the magnitude of that gap excludes the frictionless model class while constraining the missing-friction search.

  • Feldstein-Horioka Puzzle is a decomposition of Correlation Prime

    The observable payload is systematic saving-investment covariation whose regression slope supports prediction but does not identify a causal financing relation or which hidden friction generated it.

  • Feldstein-Horioka Puzzle is a decomposition of Frictionless Benchmark Reasoning Prime

    The puzzle begins from a frictionless-capital benchmark with a sharp near-zero saving-investment slope, then treats the observed residual coupling as a coordinate for cataloguing and measuring omitted frictions.

Hierarchy paths (3) — routes to 3 parentless roots

Not to Be Confused With

  • Home-bias puzzle. The sibling open-economy anomaly that investors hold overwhelmingly domestic assets when diversification predicts large foreign holdings. It shares Feldstein-Horioka's benchmark logic — frictionless integration predicts one thing, the data show another — but meters a different quantity: portfolio stocks (asset holdings) rather than saving-investment flows. The two are theoretically linked and habitually paired, which is exactly why the measured object gets swapped. Tell: is the anomalous quantity the composition of a portfolio's asset holdings (home-bias), or the correlation of national saving and investment rates (Feldstein-Horioka)?

  • Consumption-correlation / risk-sharing puzzle (Backus-Smith). The anomaly that consumption growth is far less correlated across countries than full international risk-sharing predicts. Like Feldstein-Horioka it exposes incomplete integration against a frictionless benchmark, but its measured object is cross-country consumption comovement, not the saving-investment slope. Tell: does the anomaly live in how national consumption tracks across borders (risk-sharing puzzle), or in how domestic investment tracks domestic saving (Feldstein-Horioka)?

  • Backus-Kehoe-Kydland quantity puzzle. The sibling finding that cross-country output correlations exceed cross-country consumption correlations — the reverse of what risk-sharing models imply. It is a different statistic on the same open-economy-puzzles footprint; conflating it with Feldstein-Horioka swaps an output-versus-consumption ranking for a saving-versus-investment slope. Tell: is the comparison between output and consumption comovements across countries (BKK), or between saving and investment rates within them (Feldstein-Horioka)?

  • The Obstfeld-Rogoff "six major puzzles" family (super-set). The broader catalogue of open-economy anomalies of which Feldstein-Horioka is one member, grouped by their shared international-flows footprint and frictionless-benchmark logic. The relation is part-to-whole: Feldstein-Horioka is the specific saving-investment entry, not the roster. Tell: is the reference the whole family of open-macro anomalies (the six puzzles), or specifically the saving-investment correlation (Feldstein-Horioka)?

  • The saving-investment accounting identity. In a closed economy, saving equals investment by national-accounting identity, so their equality is trivially true and uninformative. Feldstein-Horioka is precisely not that identity: it regresses investment on saving across open economies, where the current account lets the two diverge, so a high slope is a substantive finding about frictions rather than a bookkeeping tautology. Tell: is the equality forced by closed-economy accounting (the identity), or an empirical cross-country regression coefficient that could have been near zero under free capital flow (the puzzle)?

  • Price-based capital-mobility tests (interest-parity, real-interest equalization). Alternative gauges of de facto integration that read it off the equalization of returns or asset prices across borders rather than off quantity flows. Feldstein-Horioka infers integration from the saving-investment quantity relationship, which can diverge from what price tests report about the same markets. Tell: is integration being measured by whether returns or prices converge across countries (parity tests), or by whether investment decouples from domestic saving (Feldstein-Horioka)?

  • The friction / integration parent primes (frictions, market_integration, arbitrage_finance). The substrate-neutral patterns the puzzle instantiates — reallocation is not costless, separate markets behaving as one, returns equalizing under free flow — which carry the portable lesson "apparent integration can conceal substantial frictions," treated more fully above. Feldstein-Horioka is the international-finance instrument that meters this, not the general pattern. Tell: strip the saving/investment/capital-account content and the near-zero-slope benchmark, and what remains — two quantities more correlated than a frictionless model predicts, so frictions matter — is these parents, not the named puzzle.

Neighborhood in Abstraction Space

Feldstein-Horioka Puzzle sits in a crowded region of the domain-specific corpus (32nd 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