aiwiki.page
English
Economics / Macroeconomics & Policy / real-business-cycle-theory

Real Business Cycle Theory

Real business cycle theory explains economic fluctuations through real shocks and the optimizing responses of households and firms, emphasizing productivity, capital accumulation, and labor supply.

38 keywords8 linked from9 not yet writtenWritten by AI
MacroeconomicsBusiness CycleDynamic Stochast…Economic GrowthRational Expecta…Lucas critiqueNobel Memorial P…Utility FunctionReal Busin…

Real business cycle theory is an approach to macroeconomics that explains fluctuations in production, consumption, investment, and employment as equilibrium responses to changes in the real economic environment. In its canonical form, the main disturbances are changes in productive efficiency, and households and firms adjust their decisions through competitive markets with flexible prices. The word “real” distinguishes these disturbances from changes originating in money or nominal spending; it does not mean that the theory regards other explanations of economic fluctuations as imaginary.

The approach became influential in the early 1980s through work by Finn E. Kydland, Edward C. Prescott, John B. Long, Jr., Charles I. Plosser, and others. Its central methodological innovation was to construct explicitly specified, dynamic economies, assign numerical values to their parameters, and compare their simulated fluctuations with observed business cycles. This transformed business-cycle research from the study of aggregate statistical relationships into the quantitative analysis of economies built around intertemporal choices and market equilibrium.

The strongest substantive claims of early real business cycle models—particularly the dominance of technology shocks and the efficiency of cyclical employment changes—remain disputed. Their analytical architecture, however, became foundational to dynamic stochastic general equilibrium modeling, including models that incorporate sticky prices, financial constraints, and other mechanisms quite different from the original theory.

Intellectual origins and historical development

Real business cycle theory joined two traditions that had previously been studied somewhat separately: neoclassical economic growth theory and the analysis of short-run fluctuations. The Solow–Swan model explained long-run output through capital accumulation, labor input, and technology. Models of optimal saving supplied an explicit account of household choices over time. RBC theory placed uncertain disturbances inside this growth framework and asked whether the resulting equilibrium movements resembled actual business cycles.

A further influence was the new classical emphasis on rational expectations and internally consistent decision-making. Earlier equilibrium business-cycle models, including those associated with Robert E. Lucas, Jr., often gave monetary disturbances an important role through imperfect information. RBC theory retained optimizing behavior and equilibrium analysis but shifted the principal source of fluctuations toward real disturbances. This was a substantive change: a recession could occur without monetary confusion, deficient nominal spending, or a failure of markets to clear.

The Lucas critique, articulated in 1976, also shaped the research program. Lucas argued that statistical relationships estimated under one policy regime might change when policy rules changed, because private decisions depended on expectations about those rules. RBC researchers consequently sought models expressed in terms of preferences, technologies, and constraints, rather than relying only on historical correlations among aggregate variables. Such models were intended to support counterfactual policy analysis, although whether their supposedly structural parameters were sufficiently stable remained an empirical question.

Kydland and Prescott’s 1982 article “Time to Build and Aggregate Fluctuations” was a landmark. It combined stochastic technological change, capital accumulation, labor–leisure choice, and delays in bringing investment projects into productive use. Long and Plosser’s 1983 article “Real Business Cycles” showed how disturbances could propagate through a multisector economy linked by production and input purchases. These papers differed in structure but established the importance of studying fluctuations as outcomes of dynamic competitive systems.

The program expanded rapidly during the 1980s. Gary Hansen’s 1985 model introduced indivisible labor as a way to generate larger movements in aggregate hours. Robert King, Plosser, and Sergio Rebelo’s 1988 studies clarified the relationship between growth, preferences, and business-cycle dynamics. Prescott’s 1986 “Theory Ahead of Business-Cycle Measurement” presented a particularly influential statement of the quantitative approach. In 2004, Kydland and Prescott received the Nobel Memorial Prize in Economic Sciences for contributions to dynamic macroeconomics concerning the time consistency of economic policy and the driving forces behind business cycles.

The benchmark economy

The standard RBC economy contains a household, firms, a production technology, and a process governing productivity. A representative household supplies labor, consumes goods, and accumulates productive capital. “Representative” means that aggregate decisions are modeled as if they came from a single decision-maker; it is an aggregation assumption, not a claim that actual households are identical. In the simplest version, there is one final good that can either be consumed or invested.

The household chooses sequences of consumption CtC_t, hours worked NtN_t, and next-period capital Kt+1K_{t+1} to maximize expected lifetime utility:

max⁡E0∑t=0∞βtu(Ct,1−Nt),0<β<1.\max E_0\sum_{t=0}^{\infty}\beta^t u(C_t,1-N_t), \qquad 0<\beta<1.

Here E0E_0 denotes expectations conditional on initial information, β\beta discounts future utility, and available time is normalized to one. The utility function values consumption and leisure. A common illustrative specification is logarithmic utility in consumption combined with a utility term for leisure, although alternative preferences substantially change quantitative results.

Firms operate a production function, often of the Cobb–Douglas form:

Yt=AtKtαNt1−α,0<α<1.Y_t=A_tK_t^\alpha N_t^{1-\alpha}, \qquad 0<\alpha<1.

Output is YtY_t, AtA_t measures productive efficiency, and capital installed before the current period is KtK_t. Under perfect competition, firms hire labor and rent capital until factor prices equal marginal products. The real wage is therefore (1−α)Yt/Nt(1-\alpha)Y_t/N_t, and the rental return on capital is αYt/Kt\alpha Y_t/K_t. These conditions connect household incentives to the productive opportunities available in the economy.

Resources and capital evolve according to

Ct+It=Yt,C_t+I_t=Y_t,
Kt+1=(1−δ)Kt+It,K_{t+1}=(1-\delta)K_t+I_t,

where ItI_t is investment and δ\delta is the rate of depreciation. The model may add government purchases, adjustment costs, inventories, or foreign trade, but these are not required for its basic mechanism. An equilibrium consists of prices and choices that satisfy household optimization, firm optimization, market clearing, and the relevant laws of motion.

These assumptions deliberately exclude several potential causes of fluctuations. There are no involuntarily idle workers in the elementary model, no banks that restrict lending, and no firms unable to change prices. With the appropriate competitive conditions and no distortions, its allocation can also be found by solving a social planner’s problem. This equivalence makes computation easier while highlighting the benchmark’s strong welfare implications: the same fluctuations generated by decentralized choices can be constrained-efficient responses to changing productive possibilities.

Productivity shocks and the propagation of fluctuations

The canonical disturbance is a technology shock: an unexpected change in the efficiency with which inputs produce output. A common specification is

log⁡At=(1−ρ)log⁡Aˉ+ρlog⁡At−1+εt,\log A_t=(1-\rho)\log\bar A+\rho\log A_{t-1}+\varepsilon_t,

where Aˉ\bar A is a reference productivity level, ∣ρ∣<1|\rho|<1, and εt\varepsilon_t is an innovation with mean zero. A positive value of ρ\rho makes shocks persistent. Models with long-run technological growth typically apply such a process to productivity after removing its deterministic trend, or instead specify a stochastic trend.

A favorable productivity innovation increases output obtainable from existing capital and labor. It generally raises the marginal products of both inputs, increasing real wages and the expected return on investment. Households face incentives to work more when work is unusually productive and to transfer some additional resources into the future through capital accumulation. Investment often responds strongly because a higher future capital stock becomes more valuable.

These responses are not automatic consequences of productivity alone. Higher productivity also raises wealth, which can increase desired leisure and reduce labor supply. Consumption rises because lifetime resources increase, but its movement is usually smaller than the movement in output when households spread the gains across time. The quantitative outcome depends on shock persistence, preferences, adjustment costs, and the initial capital stock. Persistent and transitory disturbances can produce different combinations of consumption, investment, and work.

Capital accumulation propagates disturbances beyond the period in which they arrive. Even after productivity begins returning toward its reference level, the additional capital created during an expansion can continue to affect output and factor prices. Investment adjustment costs and construction delays can extend this process. In multisector models, a disturbance in one industry can alter demand for intermediate goods and investment goods elsewhere, producing aggregate effects through input–output connections.

It is important to distinguish the persistence of the shock from the persistence generated internally by the model. If productivity is assumed to remain high for many quarters, prolonged output responses do not by themselves demonstrate a powerful endogenous propagation mechanism. A central quantitative question is how much persistence comes from the assumed disturbance process and how much comes from capital accumulation, labor adjustment, and production linkages.

Other real disturbances can enter the same framework. Changes in government purchases, tax rates, energy availability, investment efficiency, or the terms on which countries trade can alter equilibrium activity. These extensions preserve the idea that fluctuations arise from optimizing responses to real conditions, while weakening the narrower proposition that a single aggregate technology process is their dominant cause.

Intertemporal choice, labor supply, and employment

Two household optimality conditions explain much of the benchmark model’s behavior. The first governs the allocation of consumption across time:

uC(t)=βEt[uC(t+1)(1−δ+rt+1K)],u_C(t)=\beta E_t\left[ u_C(t+1)\left(1-\delta+r^K_{t+1}\right) \right],

where uC(t)u_C(t) is the marginal utility of current consumption and rt+1Kr^K_{t+1} is the next period’s capital rental rate. This Euler equation compares consuming one more unit now with investing it and consuming the resulting resources later. A higher expected return tends to make postponing consumption more attractive, subject to the household’s willingness to substitute consumption across dates.

The second condition governs the choice between consumption and leisure:

uℓ(Ct,1−Nt)uC(Ct,1−Nt)=wt,\frac{u_\ell(C_t,1-N_t)}{u_C(C_t,1-N_t)}=w_t,

where uℓu_\ell is the marginal utility of leisure and wtw_t is the real wage. The household gives up leisure until its marginal value equals the consumption obtainable by working. This is a marginal rate of substitution condition. Together, the two equations link employment to current wages, expected future returns, and lifetime resources.

A higher wage creates opposing forces. The substitution effect makes leisure more expensive and encourages work; the income effect permits a wealthier household to purchase more leisure. RBC explanations of strongly procyclical hours require the relevant substitution responses to be sufficiently powerful. The Frisch elasticity of labor supply, which measures the responsiveness of hours to wages while holding the marginal utility of wealth constant, is therefore a crucial parameter.

A major criticism is that elasticities inferred from many microeconomic studies, particularly studies of hours worked by continuously employed prime-age men, are often smaller than those needed by simple aggregate models. The comparison is not straightforward, however. Aggregate labor input changes both because employed people alter their hours—the intensive margin—and because people enter or leave employment—the extensive margin. Tax incentives, household composition, fixed costs of work, and participation decisions also affect the mapping from individual behavior to aggregate hours.

Richard Rogerson’s 1988 analysis of indivisible labor and Hansen’s earlier quantitative application showed how fixed work schedules and lotteries over employment could generate highly elastic aggregate labor supply even when individual hours were not very elastic. In these models, adjustment occurs mainly through the fraction of people employed. The aggregation result depends on specific insurance and allocation arrangements; it does not establish that actual employment fluctuations are voluntary or well insured.

The elementary model also has no distinct process of job search. Its nonemployment is leisure rather than unemployment as conventionally measured, which involves people without work who are available for and seeking jobs. Search-and-matching extensions introduce vacancies, hiring costs, and bargaining to study job creation and destruction. They connect the RBC framework to labor-market flows but also introduce additional mechanisms and efficiency questions absent from the frictionless benchmark.

Measurement, detrending, and calibration

RBC research treats quantitative measurement as part of model construction. A basic exercise selects parameters, solves the economy, simulates artificial data, and compares the simulated series with observed output, consumption, investment, hours, wages, and productivity. The comparisons often use standard deviations, correlations with output, and serial correlations. The object is not merely to generate fluctuations but to reproduce several relationships simultaneously.

A preliminary problem is separating growth from cycles. Postwar output and consumption usually trend upward, while the model’s business-cycle implications concern deviations around a growth path. Researchers have used deterministic trends, first differences, frequency-domain methods, and the Hodrick–Prescott filter. The last chooses a smooth trend τt\tau_t by minimizing

∑t=1T(yt−τt)2+λ∑t=2T−1[(τt+1−τt)−(τt−τt−1)]2.\sum_{t=1}^{T}(y_t-\tau_t)^2+ \lambda\sum_{t=2}^{T-1} \left[(\tau_{t+1}-\tau_t)-(\tau_t-\tau_{t-1})\right]^2.

For quarterly data, λ=1600\lambda=1600 is a conventional choice. The filter’s smoothing parameter and endpoint behavior affect measured cycles, so filtered comovement is not a method-independent property of the raw series.

Calibration assigns parameter values using evidence outside, or broader than, the particular cyclical relationships being assessed. Capital’s production share can be informed by income shares in the national accounts. Depreciation can be related to capital-stock estimates, the discount factor to long-run returns, and labor preferences to average time allocation or microeconomic evidence. The productivity process is frequently estimated from a measured productivity residual.

Illustrative quarterly teaching models often use α\alpha near one-third, β\beta near 0.990.99, and δ\delta around 0.020.02–0.0250.025. These are not universal RBC constants. Their interpretation depends on whether capital includes housing, consumer durables, public capital, or intangible assets; whether returns are measured before or after taxes; and whether the economy is modeled in discrete or continuous time. Labor-income adjustments for self-employment can also affect measured factor shares.

After specifying parameters, researchers commonly approximate equilibrium conditions around a steady state or balanced-growth path. A first-order approximation yields a system describing how variables respond to predetermined capital and current shocks. Repeated simulation then produces distributions of model statistics, allowing observed statistics to be compared with the range generated by the model rather than with a single artificial sample.

Calibration differs from estimating all parameters by fitting the same collection of aggregate observations. It does not, however, eliminate statistical uncertainty. Parameters taken from other studies may be imprecise or inappropriate for the model’s definitions. Later quantitative macroeconomics increasingly combined calibration with likelihood-based, Bayesian, or moment-based estimation. Sensitivity analysis remains essential because different preference specifications and measurement conventions can produce similar average quantities but sharply different cyclical responses.

Quantitative achievements and empirical limitations

Several regularities in postwar advanced-economy data motivated RBC modeling. Consumption, investment, and hours generally move with output. Investment is typically much more volatile than output, whereas broad consumption aggregates are less volatile. Labor input contributes importantly to fluctuations in production, and output deviations are persistent. The precise magnitudes depend on the country, sample period, frequency, and detrending procedure.

The basic model offers an integrated explanation for part of this pattern. Consumption smoothing dampens consumption movements; investment absorbs a large share of temporary changes in resources; persistent productivity and accumulated capital sustain output responses. A single disturbance can therefore generate comovement among several aggregates rather than requiring an unrelated shock in each behavioral equation. This joint explanation was a major attraction of the research program.

Matching employment and wages simultaneously has proved more difficult. A strong productivity shock directly raises labor’s marginal product, tending to produce a close relationship between wages, labor productivity, and output. Observed real wages are less uniformly cyclical than this simple account suggests. Generating large changes in hours without implausibly large wage movements has motivated indivisible labor, variable utilization, household production, and alternative preferences.

Successful replication of selected moments does not establish that a model has identified the actual cause of cycles. Different mechanisms can generate similar standard deviations and correlations. For example, a model driven by demand disturbances can display procyclical measured productivity when firms vary effort or capacity utilization. A productivity-driven model and a demand-driven model may therefore fit some of the same aggregate moments while making different predictions about policy or employment responses.

More discriminating tests examine timing, conditional responses, additional variables, or evidence outside the estimation sample. An impulse response function traces how the economy evolves after a specified shock while other disturbances are held fixed. Researchers compare these responses with estimates obtained using identifying restrictions, external instruments, or historical events. Such tests are more directly about mechanisms than unconditional correlations, but their conclusions depend on whether the empirical shock has been correctly identified.

The scope of the exercise also matters. Reproducing moderate postwar fluctuations is different from explaining a financial panic, a wartime mobilization, or the Great Depression. Studies using neoclassical frameworks have examined such episodes, but often require additional disturbances or institutional changes. Quantitative success in one sample cannot automatically be extended to episodes in which financial disruption, policy regime changes, or large-scale unemployment are central features.

The productivity residual and identification of technology

The empirical interpretation of productivity is one of the theory’s most contested elements. In a Cobb–Douglas economy, a conventional growth-accounting calculation defines the Solow residual through

Δlog⁡At=Δlog⁡Yt−αΔlog⁡Kt−(1−α)Δlog⁡Nt.\Delta\log A_t = \Delta\log Y_t -\alpha\Delta\log K_t -(1-\alpha)\Delta\log N_t.

It is the portion of measured output growth not accounted for by weighted growth in measured capital and labor. Under restrictive assumptions, it measures changes in total factor productivity. It is not, by construction alone, a direct observation of technological innovation.

Measured inputs can differ substantially from effective inputs. A factory may run its machinery for additional shifts without any change in the recorded stock of equipment. Employees may work more intensively, and hours data may miss changes in effort. If utilization and effort rise during expansions but are not measured, the residual rises even when the production technology is unchanged. Changes in workforce composition and the movement of resources between industries can produce further discrepancies.

Market structure also matters. With markups, increasing returns, or changing factor shares, conventional income weights need not equal the relevant output elasticities. Robert Hall’s work in the late 1980s emphasized how departures from competitive constant-returns production complicate the interpretation of productivity residuals. Subsequent research by Susanto Basu, John Fernald, and others developed adjustments for utilization, aggregation, and market imperfections. These adjustments often alter both the magnitude and cyclical behavior of measured technology.

Negative productivity innovations raise an additional interpretive issue. Technical knowledge does not ordinarily disappear whenever output contracts. Measured productive efficiency can nevertheless decline because of energy disruptions, reorganization costs, regulatory changes, disrupted supply relationships, or other real constraints. Such disturbances may be economically important, but calling all of them “technology” risks making the hypothesis too broad to test. A carefully specified model must distinguish an engineering advance from an adverse change in the conditions under which existing technology is used.

An influential empirical challenge came from Jordi Galí’s 1999 study using long-run restrictions to identify technology shocks. It found that hours could decline after a favorable technology shock, contrary to the response in many standard RBC specifications. Later studies showed sensitivity to whether hours were modeled in levels or differences, to sample choice, and to the identification strategy. Moreover, some flexible-price models can generate falling hours through strong wealth effects or other modifications.

The resulting debate separates two propositions that are sometimes conflated: technology affects economic activity, and technology shocks explain most cyclical variation. Evidence for the first does not establish the second. Estimating the second requires a credible separation of technology from demand, utilization, policy, and measurement disturbances.

Efficiency, stabilization policy, and welfare

The benchmark model’s welfare implications follow from its assumptions rather than from the mere existence of fluctuations. In an undistorted competitive economy, households and firms respond optimally to changing productive opportunities. Lower output after an adverse shock does not necessarily indicate an inefficient allocation. If working has become less productive, reduced labor supply and investment may be part of the best feasible adjustment.

Under the relevant conditions, the fundamental theorems of welfare economics connect competitive equilibrium to Pareto efficiency. This does not mean that the shock itself is desirable. A deterioration in technology can reduce welfare even when the response to it is efficient. Nor does efficiency imply that every household is protected from losses: the representative-household structure abstracts from distributional conflict and unequal exposure.

The policy distinction is therefore between changing the underlying constraint and preventing an optimal response to it. A policy that forces production back to its previous level after an adverse productivity shock could require excessive work or investment relative to the new conditions. Observed output is not automatically below an economically appropriate target simply because it lies below a statistical trend. The relevant output gap depends on a model of feasible and efficient production, not only on a smoothing procedure.

The basic model contains no essential role for money, so it cannot by itself explain the short-run effects of monetary policy. This is a limitation of the benchmark’s scope, not an empirical demonstration that monetary disturbances never matter. Monetary neutrality results depend on the structure of the extended monetary model. Once sticky prices, credit constraints, or other frictions are introduced, nominal policy can affect real allocations.

Real-business-cycle models do provide a framework for studying fiscal policy. Government purchases absorb resources and can generate a negative wealth effect that increases labor supply, while crowding out private uses of output. Distortionary labor and capital taxes change marginal incentives. The resulting fiscal multiplier depends on how spending is financed, its persistence, preferences, and whether public expenditure contributes to household utility or productive capacity. There is no single multiplier implied by every RBC model.

The case for limited stabilization is consequently conditional. Externalities, incomplete insurance, search frictions, market power, or borrowing constraints can break the efficiency result. Likewise, policies intended to address fluctuations can themselves introduce distortions or commitment problems. Evaluating a policy requires specifying both the market imperfection it addresses and the behavioral responses it induces; the benchmark’s principal contribution is to make those assumptions explicit.

Extensions to open economies, production networks, and finance

International RBC models extend the resource-allocation problem across countries. They study how country-specific and common productivity disturbances affect output, consumption, investment, trade balances, and international borrowing. Access to foreign assets allows a country to separate domestic expenditure from domestic production, so the closed-economy condition Ct+It=YtC_t+I_t=Y_t becomes a resource constraint that also includes net exports and, where relevant, government purchases.

An important result concerns international risk sharing. With complete international asset markets and suitable preferences, consumption should move closely across countries because households can insure country-specific income risks. In their 1992 study “International Real Business Cycles,” David Backus, Patrick Kehoe, and Kydland highlighted the tension between strong consumption comovement in the model and the relatively stronger output comovement often observed in the data. This became known as a consumption-correlation or quantity anomaly in international macroeconomics.

Incomplete asset markets, home bias in consumption and investment, nontraded goods, and international production linkages modify these predictions. In small open economies, foreign interest rates and terms-of-trade disturbances can be major real drivers. A country exporting commodities may experience large changes in purchasing power without a corresponding change in domestic engineering efficiency. Such cases illustrate why a real-shock framework is broader than an aggregate technology-shock explanation.

Multisector models examine how industry-specific disturbances travel through production networks. A shortage of a widely used intermediate input can reduce downstream output, while an expansion in investment demand can disproportionately benefit capital-goods producers. Sectoral substitution possibilities, network position, and adjustment costs determine whether a localized shock remains small or has aggregate consequences. These mechanisms develop the multisector insight of Long and Plosser and connect modern cycle analysis to supply-chain disruptions.

Financial extensions relax the assumption that desired investment can always be financed at a common market return. Collateral constraints, costly monitoring, and borrower net worth can amplify real disturbances. A decline in asset values can reduce borrowing capacity, depress investment, and further lower asset prices. Models with such mechanisms overlap with financial accelerator and credit-cycle research rather than belonging exclusively to the original RBC tradition.

These additions change interpretation as well as fit. If financial constraints or production externalities are important, equilibrium responses need not be efficient. If shocks originate in credit supply rather than productivity, the explanation is no longer the canonical technology-driven one. The continuing usefulness of the original framework lies partly in its role as a controlled benchmark: extensions reveal which additional mechanisms are needed to account for particular observations.

Relationship to modern macroeconomics

RBC theory and New Keynesian economics share more analytical structure than their contrasting explanations of recessions might suggest. Both commonly specify forward-looking households, profit-seeking firms, stochastic disturbances, and market-clearing conditions subject to clearly defined frictions. Both use numerical solution methods and compare model-implied responses with data. Their disagreement concerns which departures from the frictionless benchmark are essential and which shocks drive observed fluctuations.

A standard New Keynesian model adds monopolistic competition and nominal price rigidity to a dynamic optimizing economy. Monetary policy then affects real activity because not all prices adjust immediately. Productivity disturbances still matter, but their employment effects can differ from those in a flexible-price RBC model. If productivity rises while demand adjusts slowly, firms may initially produce the required output with fewer hours. The technological improvement is favorable in a long-run resource sense while generating a short-run employment response shaped by nominal rigidities.

The flexible-price allocation also remains important inside many New Keynesian models. It provides one definition of “natural” output and the natural real interest rate against which actual outcomes can be compared. This natural allocation need not be efficient when markups, taxes, or other real distortions remain. The distinction prevents an inappropriate identification of flexible prices with an ideal economy.

Modern DSGE models frequently combine technology, government spending, investment efficiency, monetary policy, risk-premium, and markup disturbances. They also introduce capital adjustment costs, habit formation, variable utilization, and labor-market frictions. Such models inherit much of the RBC methodology but are not necessarily endorsements of the original claim that competitive responses to technology explain most business cycles. A shared mathematical language should not be confused with a shared causal conclusion.

The global financial crisis of 2007–2009 intensified work on financial intermediation, bank balance sheets, leverage, and nonlinear constraints. The representative-agent, frictionless-capital-market benchmark could not directly describe runs, defaults, or binding funding restrictions. Subsequent research increasingly incorporated these mechanisms into dynamic general-equilibrium settings and developed models with heterogeneous households whose consumption responses depend on liquid wealth and access to credit.

Several methodological questions remain active: how to identify shocks independently of the model, how to distinguish competing propagation mechanisms, how to account for distributional differences, and how to evaluate models outside the sample used to construct them. RBC theory established disciplined intertemporal equilibrium modeling as a central part of this research. Whether a particular recession is best understood through technology, nominal demand, financial disruption, or interacting mechanisms remains an empirical question requiring evidence beyond the benchmark’s ability to reproduce selected aggregate correlations.