RESEARCH DOSSIER

Business Cycles and Shock Propagation

Abstract. Business cycles arise from the interaction of economic disturbances with the dynamic structure of production, expenditure, contractual obligations, information, and policy. This monograph develops a comparative investigation of real, nominal, fiscal, financial, and expectations shocks; it distinguishes the impulse from endogenous transmission, amplification, persistence, and identification. The research integrates stock-flow accounting, standard dynamic representations, monetary and real models, financing and collateral constraints, sectoral networks, labor-market adjustment, international channels, and comparative historical analysis. It examines the Great Depression, the financial crisis of 2007–2009, and the COVID-19 contraction alongside evidence from localized supply-chain disruptions. Published research and institutional records support the analysis within the stated access and identification limits. Original analytical synthesis clarifies why the duration of a shock differs from the duration of its consequences and why similar aggregate outcomes require mechanism-specific empirical evidence. The monograph develops no new named theory and presents no original estimated data.

Keywords: Business cycles; macroeconomic shocks; shock propagation; financial accelerator; monetary transmission; aggregate demand; production networks; uncertainty; hysteresis; structural identification; local projections; stabilization policy

The scientific object of business-cycle analysis

The business cycle is the recurrent movement of aggregate economic activity through expansion, slowdown, contraction, and recovery. Its scientific significance lies in the coordinated movement of production, income, expenditure, employment, credit, and expectations through historical time. A cycle is a pattern of adjustment generated by an economy whose decisions are interdependent and whose resources, obligations, and information cannot all change simultaneously. It is recurrent without necessarily being periodic. A theory that expects a fixed calendar interval between recessions has imposed a temporal regularity that historical economies are not obliged to exhibit.

Shock propagation is the transmission of a disturbance through the decisions and institutional relations of the economy after its initial incidence. An impulse is the initiating innovation relative to a specified information set. Propagation concerns the subsequent pathway. Amplification concerns the magnitude of the total response relative to a stated benchmark without the mechanism under examination. Persistence concerns the duration of deviation from a specified reference path. These are distinguishable properties: a short disturbance may cause a long adjustment; a large initial disturbance may dissipate quickly; and an apparently small disturbance may accumulate substantial consequences when the exposed sectors occupy strategically important positions.

The purpose of this work is to explain how real, nominal, financial, fiscal, and informational disturbances enter a common economic process without assuming that the processes are observationally or institutionally identical. The methodological emphasis is comparative mechanism analysis, accounting, elementary formal derivation, historical evidence, and the evaluation of causal claims made in published research. The object is economic theory as applied to fluctuations, with aggregate macroeconomic analysis as its second scientific affiliation. It is a synthesis of established research and original interpretation of relationships between mechanisms; it contains no original estimated dataset or claim of a newly discovered general law.

Measurement, chronology, and the reference path

A recession is a broad contraction in economic activity identified against the immediately preceding expansion. A growth slowdown may occur while output continues rising. A negative output gap denotes activity below a benchmark of sustainable potential and may persist during an expansion after a trough. Consequently the growth rate, the level of output, the gap, and the cycle phase answer different questions.

The US National Bureau of Economic Research uses an assessment of magnitude, coverage, and duration rather than a mechanical two-quarter test. Its business-cycle dating procedure treats the criteria jointly. The official US turning-point chronology records the exceptionally short February–April 2020 recession. Monthly turning points and quarterly classifications must be kept distinct. The chronology identifies a contraction of two months; it does not assert that every economic consequence ceased at the trough.

Real GDP is an estimate of the volume of final production in a territory. Employment, hours, industrial production, and real income provide complementary observations. In the United States the Current Population Survey definitions distinguish unemployed individuals from those outside the labor force. A falling unemployment rate can coexist with declining participation. Output per worker and output per hour can diverge where hours change. An increase in nominal sales during inflation may conceal a decline in quantities. These discrepancies are economically informative, not statistical inconveniences to be averaged away.

The reference path deserves equal care. A pre-crisis trend extrapolation, a production-function estimate of potential, and a statistical filter produce different counterfactuals. Filtering can move part of a lasting shock into the estimated trend; a production-function estimate can itself be revised after investment or labor supply falls. Real-time vintages should be preserved when evaluating policy decisions. Revised data are appropriate for retrospective accounting but not for claiming what decision makers knew at the time.

A useful cumulative measure is the sum of real output differences from a declared no-shock baseline across a specified horizon. An estimated ten-percent shortfall for one quarter and a one-percent shortfall for ten quarters have the same simple sum if the denominator is fixed, although their welfare and distributional effects differ. The baseline is a hypothesis requiring justification; the recorded historical trough by itself does not supply it.

Historical genealogy of economic fluctuation

Different traditions emphasized different elements of the same scientific problem. Monetary approaches analyzed declining nominal expenditures, the purchasing power of money, and the banking system. Keynes emphasized the formation of effective demand, investment under uncertainty, and the interdependence of expectations and employment. In Notes on the Trade Cycle Keynes (1936) examined fluctuations in investment and the changing conditions of recovery. His analysis was situated in the institutions of a monetary production economy, not in a modern identified time-series regression.

Fisher’s (1933) debt-deflation analysis described the destructive interaction of nominal indebtedness and falling prices. Subsequent research developed financial mechanisms in terms of imperfectly enforceable debt, borrower balance sheets, and asymmetric information. The historical continuity lies in the recognition that past financial commitments influence the response to present disturbances.

The equilibrium approach supplied a different analytical discipline. Kydland and Prescott’s (1982) Time to Build and Aggregate Fluctuations integrated cyclical analysis with a dynamic growth framework. The calibration and comparison of predicted joint movements of economic time series became central tools. Capital accumulation and the time required to complete productive projects can transmit disturbances well beyond their arrival date. The existence of a model capable of reproducing selected cyclical moments does not alone establish that its assumed technology process is the exclusive historical cause.

Research on nominal rigidities sought to combine intertemporal optimization with delayed price and wage adjustment. Financial and network research extended the set of state variables and interactions. The traditions contribute distinct questions: which decision is distorted, which resource constrains adjustment, which commitment survives the impulse, and which observed pattern can discriminate between competing explanations. Their results can be compared within a unified accounting discipline while their assumptions remain explicit.

Accounting, stocks, and transmission through time

Begin with the expenditure identity in current prices: Y = C + I + G + X − M, where Y denotes domestic output, C private consumption, I gross private investment including inventory investment, G government purchases of goods and services, and X and M exports and imports. The identity describes recorded expenditure and production; it contains no behavioral coefficient. It is exact under a consistent national-accounts definition at current prices. Published chain-volume components need not add exactly because their weights vary over time.

For a closed economy under a simplified government budget, define private saving S_p = Y − T − C, where T denotes net taxes, and public saving S_g = T − G. Then national saving S_p + S_g equals investment ex post. In an open economy, the corresponding simplified relationship is S_p + S_g − I = X − M. A full national-income treatment distinguishes the current account from net exports by including relevant cross-border primary and secondary income. These equalities cannot establish whether a rise in desired saving causes higher investment, lower income, an external surplus, or some combination.

A productive stock changes through gross investment and depreciation: K_(t+1) = (1 − δ)K_t + I_t. Even an investment shock confined to one period can change capital available in later periods. Debt follows its own law of motion. Ignoring revaluation and write-offs, end-of-period debt equals initial debt plus new borrowing minus principal repayments. Cash payments of interest affect the borrower's flow resources; a change in an asset price can affect net worth without entering current GDP.

The distinction between a flow and a stock clarifies why the end of a disturbance is not the end of transmission. If a firm abandons a project because expected sales fall, the lost investment alters future capital. If a household loses an income stream while mortgage principal remains due, the outstanding liability shapes consumption. If a bank loses asset value while short-term funding comes due, the financing structure determines its immediate adjustment. Stock-flow consistency prevents the mistaken view that restoring an expenditure flow automatically restores solvency, capital, or employment relationships.

Dynamic representation of propagation

A compact representation makes the separation between impulse and response precise. Let x_t be a column vector of deviations from a local reference path containing the relevant predetermined state variables. Let ε_t denote structural innovations with conditional mean zero. A linear approximation has the form

xt+1=Axt+Bεt+1.x_{t+1}=A x_t+B\varepsilon_{t+1}.

Let an observed variable satisfy y_t = Cx_t. A one-time innovation of size u at date t, with all subsequent innovations fixed at zero and the initial deviation equal to zero, produces the horizon-h response

IRF(h)=CAhBu,h≥0.IRF(h)=CA^hBu,\qquad h\geq 0.

A is the propagation matrix, B maps shocks into exposed states, and C maps states into the measured outcome. If every eigenvalue of A has modulus below one, A^h tends to zero and a finite one-time impulse dissipates in this linear representation. Under that condition the undiscounted sum of responses is C(I − A)^(-1)Bu, provided the outcome is a deviation in a commensurate unit. Unit roots, changing regimes, and nonlinear constraints require different analysis.

For illustration only, suppose the quarterly output gap in percentage points follows x_t = 0.8 x_(t−1) + ε_t, and a shock reduces it by two points at impact. Responses at horizons zero, one, two, and three are −2, −1.6, −1.28, and −1.024. The infinite undiscounted sum is −10 percentage-point quarters. With persistence 0.2 and the same impact, it is −2.5 percentage-point quarters. These are arithmetic consequences of chosen coefficients, not estimates of any country. A model with two serially linked states can produce a response that increases for several periods before declining even if the initial shock is short-lived.

Structural shocks, reduced-form forecast errors, and residuals should not be equated. A reduced-form residual measures surprise relative to a statistical specification. A structural shock is an innovation with an economic interpretation justified by restrictions, external instruments, institutional timing, or an explicit structural model. A statistical fit of A and the covariance of residuals generally does not identify B uniquely. Identification and propagation are logically separate tasks.

Real shocks, capital, and productive capacity

A real disturbance changes technology, resource availability, preferences, relative production costs, or the physical conditions of production. A simple aggregate production relation is Y_t = Z_t F(K_t,L_t), with productivity Z_t, productive capital K_t, and labor input L_t. When capital is predetermined, a fall in Z_t initially reduces feasible output holding labor fixed. Subsequent substitutions, utilization, and investment responses depend on prices, expectations, and adjustment costs.

The Kydland–Prescott benchmark in time-to-build research (1982) treats the cycle as a dynamic process of household and producer decisions in a stochastic environment. Intertemporal substitution makes a temporary increase in the return to working or investing potentially affect activity today. Capital projects that require sequential inputs can create lags between orders and productive capacity. A long-lasting productivity change has different wealth effects from a temporary change; labor supply responses need not share a common sign under all preferences.

Measured total factor productivity is itself a constructed residual conditional on assumptions about factor shares and effective inputs. Omitted variations in capacity utilization, labor effort, input quality, and measurement error can enter it. Thus an observed residual need not be an independently exogenous technology shock. This is a critical limit when using productivity as a causal variable in cyclical research.

The real-shock approach is strongest when physical events and technological changes can be observed independently of the aggregate response. Crop failure, damaged infrastructure, and a closure of a productive facility are more directly interpretable than a residual computed from GDP. Yet their aggregate effects still depend on market substitution, inventories, financial arrangements, trade exposure, and demand. A productive shock and the institutions through which it travels belong to different levels of explanation.

Demand, expenditure, and the Keynesian mechanism

An autonomous demand disturbance changes desired purchases at the current price and income configuration. Under unused capacity and a given price level, firms can respond by increasing production and employment. The newly generated household income supports further purchases, creating an expenditure-income feedback. Under resource utilization near its limit, part of the same pressure is more likely to enter prices, imports, or displacement of other expenditure.

Consider a simplified closed-economy classroom derivation with fixed prices, fixed planned investment and government purchases, and consumption C = C_0 + c(Y − T), where 0 < c < 1 and lump-sum taxes T are fixed. Market clearing yields Y = (C_0 + I + G − cT)/(1 − c). The marginal response of equilibrium output to autonomous government purchases in this specification is 1/(1 − c). With c = 0.6, an additional 10 units of purchases gives 25 units of equilibrium output, provided the assumptions hold. Introduce proportional taxes, imports, interest-rate feedback or capacity limits and the coefficient changes. This illustration is not an empirical fiscal multiplier.

Keynes’s (1936) discussion of long-term expectations provides a microeconomic reason that planned investment can fall independently of immediate production capacity. When prospective returns become less credible, a project may cease to be desirable even at an unchanged interest rate. The induced decline in income affects firms that never held the original expectation. Demand propagation is therefore relational: one agent's expenditure constitutes another agent's receipts.

The composition of spending matters. A reduction in housing transactions affects construction, durable-goods purchases, commissions and financing differently from a comparable reduction in service consumption. Transfers affect spending through the recipients' balance sheets and consumption behavior. Aggregate expenditure identities clarify the sum, while heterogeneous behavioral relations determine the response. Neither the identity nor the elementary multiplier implies a fixed universal policy effect.

Nominal rigidities and monetary transmission

Money and nominal contracting introduce another distinction. An unexpected change in nominal financing conditions affects real borrowing costs when expected inflation does not move proportionately. For moderate rates, the expected real short rate is approximately the nominal rate less expected inflation over the same horizon. A change of policy rate thus changes the relative price of current and future expenditure to the extent that expectations, borrowing spreads and contract terms transmit it.

A canonical linear New Keynesian representation writes the output gap x_t and inflation π_t as

xt=Etxt+1−σ−1(it−Etπt+1−rtn),x_t=E_t x_{t+1}-\sigma^{-1}(i_t-E_t\pi_{t+1}-r_t^n),
πt=βEtπt+1+κxt+ut.\pi_t=\beta E_t\pi_{t+1}+\kappa x_t+u_t.

Here σ is a positive intertemporal preference parameter, β a discount factor, κ a positive slope in the basic specification, r_t^n the natural real interest rate, and u_t a cost-push disturbance. These equations are familiar model components, not a new theoretical contribution. A complete solution additionally requires a policy rule, the stochastic laws of disturbances, and equilibrium conditions. A policy rate increase relative to the natural-rate path can depress demand under the model's usual conditions; inflation and output responses cannot be computed from the two equations alone.

Christiano, Eichenbaum, and Evans (2001; journal version 2005) show in their nominal-rigidity model that a combination of wage contracts and variable capital utilization can help account for delayed inflation and persistent output responses. The result depends on an estimated or calibrated dynamic structure. It does not establish that prices are equally sticky in high-inflation economies, service sectors, or every policy regime.

Monetary transmission passes through retail and market interest rates, future expected rates, exchange rates, asset valuations, lending conditions, and expectations. Its strength is heterogeneous across fixed-rate and variable-rate debt, bank-financed and market-financed firms, and liquidity-rich and liquidity-constrained households. An observed increase in policy rates can also be a response to expected inflation. The causal effect of an unexpected policy innovation therefore differs from the descriptive correlation between rates and future activity. Romer and Romer’s (2004) monetary policy shock measure explicitly addresses this distinction by removing systematic policy responses to internal forecasts from an intended-rate series.

Fiscal shocks, announcements, and public balance sheets

Fiscal disturbances include unexpected changes in purchases, taxation, transfers, and the anticipated future budget. A purchase of a currently available domestic good creates immediate demand; a tax change operates through household resources, business incentives, and expectations; a transfer redistributes spending capacity. The expenditure date is not always the decision date. An announced future construction program can cause contractors to invest and hire before government purchases enter the national accounts.

Ramey (2011), in Identifying Government Spending Shocks, contrasts results obtained from standard VAR timing restrictions and narrative measures based on news. Her abstract reports different signs for consumption and real-wage responses under the approaches and documents anticipatory relationships. This evidence supports the methodological conclusion that a forecastable expenditure change cannot automatically be identified as a surprise at the moment the transaction is recorded. It does not determine the multiplier for every fiscal institution or episode.

Debt-financed spending can alter expected future taxes, sovereign risk premia, and central-bank responses. The degree of crowding out depends on capacity, the financial environment, monetary accommodation, and access to imported resources. In a severely depressed economy with idle resources, displaced expenditure may be small; in a capacity-constrained economy, resource displacement and inflation may be substantial. The burden of debt servicing is determined by maturity, interest rates, currency denomination, and the future primary balance.

Temporary transfers may prevent distress sales and maintain viable employment matches even when immediate measured output cannot rise. Their insurance value is conceptually distinct from the fiscal multiplier. That distinction is reinforced by Guerrieri and colleagues’ (2022) multisector supply-shock analysis, which shows under specified conditions that transfers can provide important insurance even where the spending multiplier is smaller. Fiscal evaluation should report the targeted objective and the fiscal counterfactual together.

News, uncertainty, and coordinated expectations

An expectation shock changes beliefs about future economic conditions before the corresponding physical or policy event occurs. A news shock may change the expected mean of future productivity or costs. An uncertainty shock changes the dispersion of possible outcomes, potentially with little change in the expected mean. In practice the two can arrive together. Expectations about others' responses create additional interactions among agents.

Keynes (1936), in The State of Long-Term Expectation, connects investment to prospective yields and confidence in the estimates underlying them. Expectations about an asset's resale value may affect the investment decision even when its engineering productivity is unchanged. This provides a route from information and confidence to real activity. Financing conditions and expectations are related because creditors evaluate borrowers' prospects as well as their existing assets.

For a firm facing costly reversal of investment, the option to wait can acquire greater value as uncertainty rises. If hiring requires search and training costs, postponing recruitment can likewise be rational. Bloom’s (2009) uncertainty shock model estimates adjustment costs and simulates a temporary interruption of hiring and investment followed by rebound. Its particular time profile is a model-dependent and empirically compared result. Uncertainty can also cause precautionary saving, widen credit spreads, and shift resources into activities perceived as safer.

A financial-market volatility index is helpful but is not a pure measure of exogenous uncertainty. It reflects expected variance, risk premia, market liquidity, and information about the economic outlook. A recession can raise uncertainty, while uncertainty can contribute to recession. Credible identification requires attention to both directions and to the distinct possibility that a common event causes each. The appropriate evidence includes survey dispersion, firm-level postponement decisions, event timing, and changes in expected cash flows.

Financial acceleration and collateral

Financial contracts convert changes in income and valuations into constraints on future action. A borrower with net worth N = V − D, where V is assets valued at the relevant market or economic value and D liabilities, can experience a disproportionately large percentage loss of net worth when leverage is high. For fixed D and an initial A-to-equity ratio of five, a ten-percent decrease in asset value reduces equity by fifty percent, absent other changes. This arithmetic does not imply the same change in regulatory capital, which uses legal and accounting definitions.

Bernanke, Gertler, and Gilchrist (1999), in the financial accelerator framework, formalize endogenous credit-market amplification. Bernanke's (2007) credit-channel explanation clarifies the external finance premium as the additional financing cost associated with obtaining external rather than internal resources, with monitoring and agency problems contributing to its dependence on borrower financial strength. A decline in cash flow can raise the premium, reducing investment and generating a further decline in income. This loop creates persistence without requiring a prolonged original shock.

Kiyotaki and Moore’s (1997) Credit Cycles develops a related mechanism in an economy in which lenders require enforceable security. For a simplified collateral constraint D_(t+1) ≤ m q_t K_t, with collateral price q_t and admissible fraction m between zero and one, a decline in q_t lowers borrowing capacity if the constraint binds. Asset sales may further reduce q_t when purchasers face similar constraints. With ample unencumbered assets and slack constraints the first-round mechanism is weaker.

The two mechanisms can coexist. One emphasizes costly external financing and borrower incentives; the other emphasizes limited pledgeability and the collateral value of durable assets. To distinguish them empirically, debt contracts, cash-flow measures, collateral valuations, lending spreads, covenant changes, and quantities of credit need to be observed jointly. A rise in spreads could reflect increased default risk, liquidity compensation, lender market power, or expected monetary tightening. It is not by itself a measure of one theoretical friction.

The causal mechanism also has a counterfactual benchmark. Amplification means the difference between responses with and without the relevant borrowing friction, holding other assumptions and the initial shock comparable. Comparing a leveraged economy with a wholly different economy without controlling how its productive and financial structure differs cannot isolate the mechanism.

Funding liquidity, bank runs, and debt deflation

A bank or financial intermediary may finance long-lived or difficult-to-sell assets with claims that are repayable on short notice. An institution can be solvent under an orderly valuation and unable to meet simultaneous withdrawal requests. Conversely, emergency funding cannot restore positive economic net worth if losses exceed the value of capital and feasible future earnings. Funding liquidity and solvency should be separated analytically even though a disorderly sale can turn the first problem into the second.

Diamond and Dybvig (1983), in Bank Runs, Deposit Insurance, and Liquidity, show within a model of private liquidity needs how deposit contracts can permit efficient risk sharing while also allowing a run equilibrium. The expected actions of depositors affect the incentive to withdraw. A panic can therefore cause destruction of economic value rather than merely reveal preexisting losses. Deposit insurance is a possible institutional response in the model; its real-world design also involves credibility, incentives, supervision, and allocation of losses.

Fisher’s (1933) debt-deflation theory identifies the importance of nominal obligations when prices decline. If debt principal is 100 monetary units and the price index declines from 100 to 80, the real principal measured in units of the original consumption basket increases from 100 to 125, a 25 percent increase rather than a 20 percent increase. Debt restructuring, inflation expectations, income changes, and interest-rate resets affect the actual burden.

The macroeconomic feedback can combine distressed sales, declining collateral, payment difficulties, tightened credit, lower investment, and weak demand. However, the decline in a particular asset price does not have the same effect as broad price deflation. A reallocation of wealth to debt-free buyers can offset part of the first-round expenditure loss. The net result depends on differences in spending propensities and on whether financial intermediaries can continue to perform their functions. A liquidity provision policy can arrest a self-fulfilling funding panic while failing to resolve an underlying capital deficiency.

Production networks and sectoral interdependence

Aggregate disturbances can originate in a sector whose share of final demand understates its role as an intermediate supplier. An input-output system makes the distinction transparent. Let z be the vector of gross output, d final demand, and A a matrix whose element a_ij denotes the units of supplier i's output used per unit of purchaser j's output, in consistent reference units. The fixed-coefficient relation is

z=Az+d,z=(I−A)−1d.z=Az+d,\qquad z=(I-A)^{-1}d.

The second expression is valid when the inverse exists; with nonnegative A and spectral radius below one it has a convergent series I + A + A² + …. This representation traces intermediate requirements for an imposed final-demand change. It is not by itself a model of a capacity shortage. A production limit or a disrupted supplier requires additional rationing, substitution, inventory, or price-adjustment conditions.

For a two-sector illustration, set a_12 = 0.4, a_21 = 0.2, and diagonal coefficients to zero. An increase in sector 2 final demand of 10 output units requires sector 2 gross output of 10/0.92, approximately 10.87, and sector 1 gross output of approximately 4.35. The induced gross-output requirements cannot simply be added to GDP because intermediate goods would be double counted. A loss of available sector 1 inputs requires a different calculation of feasible sector 2 output.

Acemoglu and colleagues (2012), in The Network Origins of Aggregate Fluctuations, establish under their model conditions how asymmetry in supplier influence makes some idiosyncratic disturbances important at the aggregate level. Sparse network connectivity alone is not a sufficient statistic. Carvalho and colleagues’ Great East Japan Earthquake study (2014 working-paper version; 2021 journal article) exploits the regional shock in 2011 to investigate transmission through customer and supplier links. The evidence supports the relevance of upstream and downstream propagation.

Network exposure must be evaluated with substitutability, the time to qualify new suppliers, geographic correlation, transport, inventories, and financial constraints. An economy with several nominal suppliers who depend on the same upstream facility may be less diversified than supplier counts suggest. Short-run capacity constraints can be severe even if long-run substitution is easy. A pre-shock input-output matrix can therefore be informative about first-round exposure while providing an incomplete account of the entire adjustment.

Supply disturbances that generate demand shortages

The aggregate distinction between demand and supply remains useful when it specifies a counterfactual. Supply describes feasible production under technology, resource availability and constraints; demand describes desired expenditure conditional on prices, incomes, expectations and financing. The categories become misleading if an investigator presumes that an event belongs permanently to one side of the economy.

Consider two sectors, one temporarily unable to operate and one still physically capable of producing. Workers in the first sector lose income; if consumption insurance is incomplete, their purchases from the second may fall. Households unable to substitute easily toward the available sector may save rather than spend. The second sector then reduces production even though its own technological capacity did not fall. This differs from a one-sector account in which the same decrease in potential production simply creates scarcity everywhere.

Guerrieri, Lorenzoni, Straub, and Werning (2022) analyze this mechanism in Can Negative Supply Shocks Cause Demand Shortages?. Their model identifies the combination of limited substitution across sectors, intertemporal choices, and incomplete markets as important conditions. The article also examines firm exit and job destruction as routes of further amplification. This is a conditional result about multisector economies, not an assertion that every adverse supply shock creates a net demand shortfall.

The diagnosis has practical significance for economic analysis. An indiscriminate increase in expenditure cannot produce a service that remains physically unavailable; stabilizing the income of displaced workers can prevent avoidable declines elsewhere. The distinction requires sector-level quantities and prices, not GDP alone. In sectors with binding capacity, increased demand may be inflationary, while other sectors experience insufficient sales. A successful aggregate explanation must be able to accommodate both phenomena simultaneously.

Oil, relative prices, and real-income redistribution

Energy costs affect production, transport, heating, household budgets, and external income. A rise in the observed oil price can result from a physical supply reduction, stronger global economic activity, or changed expectations about oil availability. These mechanisms imply different movements in other variables. Treating the price increase as an autonomous negative supply disturbance in every episode embeds an unwarranted causal restriction.

Kilian’s (2009) oil-market shock decomposition distinguishes oil supply innovations, demand innovations related to global activity, and oil-specific demand innovations. His findings highlight why an estimated relationship between oil prices and aggregate activity may change as the underlying mix of shocks changes. A common price movement is not a common structural treatment.

A net importer of energy experiences a transfer of purchasing power abroad when the relative import price increases, holding volumes and other prices fixed. A net exporter can experience an income gain and increased fiscal resources while consumers and non-energy firms still face costs. Exchange rates, taxes, subsidies, production contracts and market structure determine the distribution. Domestic real GDP and real gross domestic income may consequently move differently in a terms-of-trade shock.

Adjustment depends on the duration agents expect. Inventories cushion a brief interruption but cannot indefinitely replace a permanently constrained flow. Long-lived investments in more energy-efficient equipment require a credible expected return over their useful life. A price cap or subsidy can redistribute the cost among users, producers and taxpayers but cannot eliminate the underlying scarcity. The economic consequences must be traced through both contemporaneous income and future adaptation.

International transmission and financial exposure

The world economy connects domestic cycles through external demand, intermediate goods, cross-border ownership, securities, bank funding, remittances, commodity prices, and monetary expectations. A foreign recession lowers export demand for exposed producers, while a disturbance to an internationally specialized supplier can change domestic production even if foreign final demand is unchanged. The strength of each channel depends on gross exposures rather than net exports alone.

Consider a firm with 100 units of unhedged foreign-currency debt when one unit of foreign currency costs one local unit. If that exchange rate rises to 1.2, the local value of its foreign debt rises to 120, other things equal. Export revenue denominated in foreign currency may naturally hedge part of the exposure. A firm whose receipts are domestic and liabilities foreign experiences a deterioration of net worth unless another asset or financial hedge offsets it.

Rey (2013), in her author account of the global financial cycle, relates common movements in capital flows, leverage and asset valuations to international financial conditions. This analysis challenges the idea that a flexible exchange rate alone guarantees full insulation from global credit conditions. The BIS (2025) financial conditions report likewise describes increasingly connected markets and exchange-rate effects through the valuations of lenders. These studies do not imply that all monetary autonomy disappears or that every country has the same exchange-rate sensitivity.

Trade adjustment can oppose financial adjustment. Depreciation may improve the competitiveness of an exporter while raising the domestic value of foreign debt and imported intermediate costs. In an economy with large unhedged liabilities, financial contraction may dominate over the relevant horizon; in a relatively hedged economy with responsive trade volumes, the balance differs. Countries with distinct debt currencies, financial depth, policy credibility and trade links therefore respond differently to the same external innovation.

A cross-country study should specify common shock exposure, local response variables, initial external balance sheets, currency composition and domestic policy. It should distinguish a common global cause from genuine bilateral transmission. A synchronized contraction is evidence of co-movement, not proof of a single causal route.

Labor, household heterogeneity, and persistent scars

Employment relationships embody accumulated knowledge, search effort, and organization. When demand falls, firms may delay hiring, reduce hours, postpone wage adjustments, or dismiss workers. The response depends on labor contracts, financing, institutions, and expectations about recovery. Recovery in output can precede recovery in employment when existing firms raise utilization or reorganize tasks before expanding payrolls.

The household sector is not a representative balance sheet. A family with liquid savings, stable earnings and fixed-rate debt can smooth a temporary loss more easily than a household with limited liquidity and variable-rate obligations. If high-spending households suffer large income losses, a given decline in total labor income may reduce purchases more than the same aggregate loss concentrated among households able to draw down savings. An income transfer changes the distribution of resources; the aggregate demand effect depends on that distribution.

Blanchard and Summers (1986) discuss hysteresis in unemployment, the possibility that the equilibrium unemployment rate itself is affected by the history of unemployment. A temporary reduction in vacancies can lead to lasting effects if skills depreciate, labor-force attachment weakens, or wage-setting and hiring institutions respond to the changed state. Such mechanisms are candidates to explain hysteresis rather than proof of a permanent loss in every episode.

A statistical unit root in unemployment does not uniquely identify labor-market hysteresis, since persistent exogenous shocks and structural breaks can generate similar evidence. Conversely, an estimated mean-reverting aggregate unemployment rate may conceal long-lasting losses for individual cohorts. Empirical assessment should track participation, duration of unemployment, transitions into work, hours, earnings, and cohort outcomes. The labor-force definitions identify why unemployment alone is not sufficient. Aggregate recovery and recovery of the affected people are connected but distinct scientific objects.

State dependence, asymmetry, and nonlinear amplification

The response to a shock may depend on the economy's initial position. A firm with enough reserves to cover several months of fixed costs can absorb a temporary sales loss; a similar firm near a debt covenant or payment deadline may be forced to cut investment immediately. When a constraint switches from slack to binding, an otherwise smooth behavioral relationship acquires a kink. The financial system may contain many such thresholds with clustered exposures.

A scalar linear approximation is useful near a stable operating point, but large departures may require nonlinear methods or explicit regime dependence. For example, x_(t+1) = a x_t + b ε_(t+1) may be a reasonable local description of an unconstrained economy; if collateral falls below a threshold, the relevant coefficient or set of state variables changes. The response from a recession cannot be presumed equal to the negative of a response from an expansion of the same magnitude.

Jordà, Schularick, and Taylor (2012; journal version 2013) investigate credit accumulation and recession paths using historical observations and local projections. The study illustrates why accumulated credit is valuable as a state variable in the analysis of downturns. Its observational estimates do not amount to randomly assigning economies to high- and low-debt states. Pre-crisis credit can correlate with asset booms, liberalization, risk taking and changes in policy.

Nonlinearity also complicates prediction. A fitted average response may disguise a nearly insignificant response under normal liquidity and a large response during distress. Estimating separate regimes reduces the effective sample size; model selection and thresholds can introduce inference problems. A robust investigation states the regime definition in advance, reports uncertainty and sensitivity, and avoids interpreting a visually striking episode as proof of a universal threshold.

Identification, counterfactuals, and empirical design

An economic causal question specifies what changes, which conditions are held fixed, who receives the treatment, and the response horizon. Identification uses variation that can credibly distinguish the hypothesized shock from concurrent disturbances. In a vector autoregression, reduced-form residuals capture innovations relative to lagged observed variables. If their covariance matrix is Σ and structural shocks have identity covariance, decompositions B satisfying BB′ = Σ need not be unique. Additional restrictions are substantive assumptions about economics or information.

Recursive restrictions impose a specified within-period ordering. Sign restrictions restrict directions of responses over chosen horizons and may leave a set of admissible structural models. Long-run restrictions use theoretical neutrality or permanent-effect assumptions. Blanchard and Quah (1989), in aggregate demand and supply disturbances, distinguish disturbances with lasting output effects from those without such effects. Their economic labels follow the restriction; the labels cannot be validated by the restriction alone. Hysteresis and persistent demand effects are important reasons to test its applicability.

An external instrument for a structural innovation needs a first-stage relationship with that innovation and credible exclusion from other structural disturbances. Narrative information, policy announcements, weather variation, and localized disasters can supply relevant variation in suitable settings. But a policy announcement may reveal news about the central bank's forecast as well as its policy stance, and a disaster can simultaneously affect local demand and supply. The exclusion assumption remains a scientific argument requiring evidence.

Romer and Romer (2004), through their narrative monetary shock construction, use intended policy movements and forecasts to reduce confounding. Ramey (2011) demonstrates why the timing of fiscal news matters. The methods share an emphasis on the information set, but they answer different questions about different shocks. An investigator should state the event window, anticipation treatment, residualization procedure and possible omitted information before interpreting the response.

In local projections, introduced by Jordà (2005) in Estimation and Inference by Local Projections, responses are estimated separately for each horizon. A simplified specification may regress y_(t+h) on an identified shock ε_t, controls predetermined with respect to the shock, and an intercept. In cumulative growth specifications the dependent variable can instead be the change from the pre-shock level. Overlapping horizons can induce serial correlation; inference should account for the data structure. Local projections are a method of estimating dynamic responses, not a substitute for shock identification.

The Lucas (1976) policy evaluation critique is a reminder that historical behavioral coefficients can change if the policy regime changes and agents adjust their expectations. The scope of a historical estimated response is therefore tied to its institutional environment. A stable description of past comovement does not by itself justify a new policy counterfactual. Structural modeling, reduced-form identification and historical investigation contribute complementary evidence.

Historical comparison: the Great Depression

The Great Depression demonstrates the interaction of monetary contraction, financial fragility, nominal obligations, institutional constraints, and weak aggregate demand. The Federal Reserve's historical account of the Depression (2013) describes the onset in 1929, recurrent banking distress in the early 1930s, the national banking holiday in March 1933, and the renewed contraction in 1937. These landmarks distinguish the onset, propagation, trough, and incomplete recovery.

The stock market crash was a visible event, but the subsequent evolution included changes in bank deposits, the availability of credit, prices, and obligations. Falling prices increased real debt burdens, a channel emphasized by Fisher's (1933) debt-deflation account. The gold-standard regime restricted monetary responses and linked financial conditions across national borders. International transmission interacted with national banking and political institutions.

Several mechanisms make distinct observable predictions. A monetary explanation should be assessed against monetary aggregates, prices, policy instruments, and the timing of monetary regime shifts. A credit-intermediation explanation should examine borrower access, bank failures, spreads, and lending conditions in addition to the quantity of money. A demand explanation should trace income, expenditure and expected profitability. These causal processes may reinforce each other; evidence for one does not logically exclude the others.

The historical account cannot provide a randomized counterfactual for the entire Depression. Its value for this monograph is the observed temporal sequence and the consistency of institutional mechanisms with that sequence. Comparative research across the timing of gold-standard exit, local bank exposure, and policy decisions can supply further discrimination. This study does not calculate a percentage of the Depression attributable to a single cause.

Historical comparison: the financial crisis of 2007–2009

The financial crisis and Great Recession illustrate a long buildup of vulnerability followed by a rapid contraction. The Federal Reserve's history of the Great Recession (2013) traces the housing and mortgage-credit expansion, market funding stress in 2007, intensified distress after Lehman Brothers' bankruptcy in September 2008, and the slow recovery after the economy reached its trough in 2009.

The initial loss on residential assets affected borrowers, institutions holding mortgage-linked securities, and counterparties dependent on the value and liquidity of those assets. A deterioration of confidence in collateral valuation could impair funding even among counterparties whose final losses remained uncertain. Financial institutions adjusted exposures, borrowers faced worse financing terms, and private demand and investment weakened. Some feedback operated through actual defaults and others through uncertainty about the distribution of future losses.

The episode shows why the condition of the economy before the impulse matters. A long expansion financed by debt could sustain consumption and investment while making the system more exposed to collateral declines. The financial accelerator discussed by Bernanke (2007) in The Financial Accelerator and the Credit Channel supplies a theoretical explanation of why lower borrower financial strength can raise borrowing costs and prolong contractions. It is one analytical channel within a complex historical episode.

The public interventions addressed funding-market functioning, aggregate demand, capital and other objectives through different institutions. Emergency loans are conceptually distinct from transfers or capital injections and from reductions in the general level of interest rates. The policy mix and the institutional response altered the observed propagation path. The downturn therefore cannot be used as an estimate of what would have happened without interventions.

The contrast with the Depression is institutionally informative: policy regimes, banking protections, monetary arrangements, and cross-border securities markets differed. Yet both episodes show how existing nominal obligations and financial relationships can transmit disturbances into future output. The point is similarity in conditional mechanisms, not an assertion of identical crises.

Historical comparison: the pandemic and supply-chain shocks

The pandemic contraction began with health risks, behavioral changes, restrictions on contact, and changes in physical and organizational conditions of production. The official NBER business-cycle chronology dates the US peak to February 2020 and the trough to April 2020. The short measured contraction coexisted with prolonged sectoral adjustment. These dates should not be generalized to the world economy; countries faced different timing and responses.

Sectoral heterogeneity was central. In-person services faced severe constraints, while demand for certain goods and remote services could rise. Income losses and uncertainty could reduce demand in still-operating sectors. Supply-chain exposure, inventories and shifts in household expenditure affected the subsequent reopening. The multisector mechanism analyzed by Guerrieri and colleagues (2022) in their pandemic-related model explains how a negative supply shock can under stated conditions cause excess capacity in other sectors.

An independent comparative example is the 2011 Great East Japan earthquake. Carvalho and colleagues' supply-chain disruption study investigates how a regional physical disturbance affected connected firms. Its setting offers a more localized initiating event than a worldwide pandemic, which makes the distinction between initial exposure and propagation particularly instructive. The findings cannot be transferred mechanically to a global event with simultaneous demand and health shocks.

The pandemic episode and the earthquake demonstrate different configurations of physical loss, expectations, network links and public responses. Reconstruction after physical damage can raise demand for investment even as welfare and wealth have fallen. Temporary health-related closures can generate deferred or redirected consumption even if some physical capital remains intact. Measured aggregate output is therefore an incomplete measure of disaster loss. A historical comparison must preserve these distinctions when evaluating recovery.

Comparative implications and mechanism diagnostics

A productive shock, a demand shock, a financial shock, and a change in expectations can generate the same sign of movement in aggregate output. The explanatory task is to identify different intermediate variables, temporal orderings and counterfactuals. The following comparisons are hypotheses derived from familiar mechanisms rather than universal sign restrictions.

Candidate mechanism Intermediate observations Distinguishing questions
Productive constraint Input availability, utilization, sector quantities, relative prices Was the loss of capacity independently observable; could substitution restore output?
Demand contraction Orders, inventories, idle capacity, income and expenditure Did purchases fall in sectors with feasible spare capacity?
Nominal tightening Unexpected policy news, real rates, refinancing and spreads Did borrowing terms change beyond responses to expected macroeconomic news?
Collateral or funding stress Asset valuations, loan covenants, deposit outflows, credit terms Were vulnerable borrowers disproportionately affected conditional on demand?
News or uncertainty Forecast revisions, dispersion, deferred hiring and investment Did decisions change before the anticipated event; was the expected mean or dispersion altered?
Network transmission Preexisting input-output links, delivery interruption and downstream output Did effects follow the exposure topology after accounting for common shocks?

These observations are not mutually exclusive. A physical input shortage can reduce borrower cash flows and collateral; a bank funding interruption can force customers to delay purchases; a change in exchange rates can both revalue debt and alter relative prices. The appropriate analysis follows the sequence of these links rather than selecting one aggregate label and assigning the entire event to it.

The temporal indicators have to be measured at the appropriate frequency. High-frequency market prices may adjust on the announcement date, while published production and employment data are monthly or quarterly. The observed ordering can reflect data publication and aggregation rather than the economic sequence. Survey expectations, transaction dates and balance-sheet reporting periods help align the observations. Distinguishing genuine propagation from the delayed observation of the impulse is one of the most important empirical tasks.

Stabilization policy and its limits

Stabilization policy is the attempt to reduce socially costly adjustment while preserving the conditions for productive allocation, financial functioning, and sustainable prices. Its evaluation depends on the disturbance, the affected agents, and the channels through which an instrument operates. A policy that restores financial-market functioning may prevent a secondary contraction even when it cannot reverse the initial physical loss.

Monetary instruments affect general financing conditions and expectations. The effectiveness of reductions in short-term interest rates depends on pass-through, the expected future policy path, the term structure, financing spreads, and the state of balance sheets. Financial-market liquidity operations may address disrupted funding markets; supervision, resolution and recapitalization address distinct institutional problems. Fiscal instruments can replace income, purchase output, support public investment, or redistribute resources. Their effects depend on recipients, resource constraints, financing and timing.

The scientific assessment should compare feasible policies, not an actual intervention with a frictionless ideal. If markets are temporarily unable to match buyers and sellers because information is scarce, the design of a liquidity facility matters. If resources are physically unavailable, attempts to restore aggregate demand may shift the burden into prices. If viable firms face temporary revenue losses, a bridge may preserve productive relationships; prolonged support for persistently unviable projects can delay reallocation.

Policy can also influence future vulnerability. Insurance and guarantees improve the resilience of funding if credible; poorly priced or weakly supervised guarantees may change risk-taking incentives. Predictable countercyclical institutions can improve expectations, while an arbitrary change in commitments can raise uncertainty. No individual historical intervention establishes a universally optimal response. The analytical criterion is whether an instrument addresses a documented mechanism under the actual institutional and resource conditions, accounting for welfare and distribution as well as aggregate output.

Original analytical synthesis and research agenda

The principal conclusion of this comparative investigation is that the economic consequences of a shock depend jointly on its first incidence, the inherited structure of commitments and exposures, and the ability of the economic system to rearrange resources over time. These dimensions are established elements of economic theory; the contribution here is their explicit integration in the comparative interpretation of business-cycle evidence.

The first implication is that the same innovation may generate different cumulative responses across economies with different inherited states. A country with stronger borrower net worth, diverse suppliers, credible monetary institutions, and flexible access to replacement inputs may adjust differently from one with leveraged balance sheets and concentrated exposures. This is a conditional analytical inference, not a proposed universal ranking of countries. The choice of relevant state variables depends on the mechanism being tested.

The second implication is that distinctions between transitory shocks and persistent losses require two separate investigations. One concerns the duration of the exogenous innovation; the other concerns the evolution of capital, debt, labor relationships and expectations after it. Temporary disturbance to an income flow can destroy a relationship that takes time to rebuild. If the lost relationship alters future productive opportunities, the original disturbance and its long-run consequences belong to different causal stages.

The third implication concerns evidentiary hierarchy. A model establishes that specified assumptions can produce a response. A historical chronology establishes the order of observed events. An identified empirical intervention or innovation supplies conditional evidence on a causal response. Accounting provides consistency constraints across these forms of evidence. Their combination can support a stronger explanation than any one type alone. Independent replication, robustness and explicit treatment of counterfactual uncertainty remain essential.

Several concrete research questions follow. First, measure firm exposure to a disruption using pre-event supply links and test whether downstream losses are stronger among firms without alternative suppliers after controlling for industry and geography. Second, estimate whether the same identified external policy innovation produces different output and investment responses across firms or countries with different pre-event currency mismatches. Third, examine whether an income-replacement intervention protects employment relationships beyond the period of the original income shock. Fourth, assess the role of prior credit expansion using predetermined balance-sheet variables, multiple recession definitions, and alternative identification assumptions. These are proposed empirical tests, not completed empirical results.

A final implication is editorial and scientific: taxonomy should follow causal questions. “Oil-price shock” is a useful description of an observation but an insufficient causal interpretation. “Financial crisis” may identify a historical event that contains distinct initiating disturbances and propagation loops. “Recession” refers to an outcome and phase rather than one underlying mechanism. Maintaining this vocabulary reduces ambiguity and makes research results reusable across economic theory and macroeconomics.

Conclusion

Business cycles are historically realized paths of interdependent decisions under changing economic conditions. Their explanation requires the separation of the initiating impulse from the economic structure that carries its effects through time. Productive capacity, expenditure, nominal contracts, credit, expectations, networks, and international exposures condition these paths. Each mechanism offers specific restrictions and observable implications; their interactions explain why output alone is insufficient to identify a shock.

An initiating event can be brief while the consequences endure because investment, balance sheets, employment matches, and expectations are states carried into the future. Conversely, an industrial or financial exposure may be mitigated by substitutability, effective insurance, resilient funding, or policy. Multiple causal processes can coexist within the same recession. Policy is evaluated according to the documented mechanism, initial conditions, resource constraints, institutional environment and welfare effects rather than a fixed universal multiplier.

The study's original analytical contribution consists in the explicit comparative integration of these established mechanisms, their evidentiary requirements, and their historical implications. Its formal examples are transparent derivations with illustrative coefficients. Empirical claims are confined to the reported scope of the cited publications; the historical discussion supplies comparative interpretation, not newly estimated causal shares. Future research can discriminate among channels through improved measurement of exposures, independently justified shock identification, and matched counterfactuals. This permits business-cycle research to explain both the shared dynamics and the distinctive institutional histories of the world economy.

References

  • Acemoglu, Daron, Vasco M. Carvalho, Asuman Ozdaglar, and Alireza Tahbaz-Salehi. 2012. “The Network Origins of Aggregate Fluctuations.” Econometrica 80 (5): 1977–2016. DOI 10.3982/ECTA9623. MIT repository. Research record.
  • Bank for International Settlements. 2025. “Financial Conditions in a Changing Global Financial System.” Annual Economic Report 2025, chapter II. Basel: BIS. Research record.
  • Bernanke, Ben S. 2007. “The Financial Accelerator and the Credit Channel.” Speech, Federal Reserve Bank of Atlanta Conference on the Credit Channel of Monetary Policy, June 15. Board of Governors of the Federal Reserve System. Research record.
  • Bernanke, Ben S., Mark Gertler, and Simon Gilchrist. 1999. “The Financial Accelerator in a Quantitative Business Cycle Framework.” In Handbook of Macroeconomics, vol. 1C, edited by John B. Taylor and Michael Woodford, 1341–1393. Amsterdam: Elsevier. Research record.
  • Blanchard, Olivier J., and Danny Quah. 1989. “The Dynamic Effects of Aggregate Demand and Supply Disturbances.” American Economic Review 79 (4): 655–673. RePEc record. Research record.
  • Blanchard, Olivier J., and Lawrence H. Summers. 1986. “Hysteresis in Unemployment.” NBER Working Paper 2035. DOI 10.3386/w2035. Research record.
  • Bloom, Nicholas. 2009. “The Impact of Uncertainty Shocks.” Econometrica 77 (3): 623–685. Stanford GSB institutional record. Research record.
  • Carvalho, Vasco M., Makoto Nirei, Yukiko U. Saito, and Alireza Tahbaz-Salehi. 2014. “Supply Chain Disruptions: Evidence from the Great East Japan Earthquake.” RIETI discussion-paper summary. Published version: Quarterly Journal of Economics 136 (2), 2021: 1255–1321. DOI 10.1093/qje/qjaa044. Research record.
  • Christiano, Lawrence J., Martin Eichenbaum, and Charles L. Evans. 2001. “Nominal Rigidities and the Dynamic Effects of a Shock to Monetary Policy.” Federal Reserve Bank of Cleveland Working Paper 01-07. DOI 10.26509/frbc-wp-200107. Journal version: Journal of Political Economy 113 (1), 2005: 1–45. Research record.
  • Diamond, Douglas W., and Philip H. Dybvig. 1983. “Bank Runs, Deposit Insurance, and Liquidity.” Journal of Political Economy 91 (3): 401–419. DOI 10.1086/261155. University of Chicago repository. Research record.
  • Fisher, Irving. 1933. “The Debt-Deflation Theory of Great Depressions.” Econometrica 1 (4): 337–357. FRASER, Federal Reserve Bank of St. Louis, digitized full-text edition. Research record.
  • Guerrieri, Veronica, Guido Lorenzoni, Ludwig Straub, and Iván Werning. 2022. “Macroeconomic Implications of COVID-19: Can Negative Supply Shocks Cause Demand Shortages?” American Economic Review 112 (5): 1437–1474. DOI 10.1257/aer.20201063. Research record.
  • Jordà, Òscar, Moritz Schularick, and Alan M. Taylor. 2012. “When Credit Bites Back: Leverage, Business Cycles, and Crises.” Federal Reserve Bank of San Francisco Working Paper 2011-27, October revision. Published as “When Credit Bites Back,” Journal of Money, Credit and Banking 45 (S2), 2013: 3–28. Research record.
  • Jordà, Òscar. 2005. “Estimation and Inference of Impulse Responses by Local Projections.” American Economic Review 95 (1): 161–182. DOI 10.1257/0002828053828518. Research record.
  • Keynes, John Maynard. 1936. The General Theory of Employment, Interest and Money. London: Macmillan. Chapter 12, “The State of Long-Term Expectation.” HET electronic edition. Research record.
  • Keynes, John Maynard. 1936. The General Theory of Employment, Interest and Money. London: Macmillan. Chapter 22, “Notes on the Trade Cycle.” HET electronic edition. Research record.
  • Kilian, Lutz. 2009. “Not All Oil Price Shocks Are Alike: Disentangling Demand and Supply Shocks in the Crude Oil Market.” American Economic Review 99 (3): 1053–1069. DOI 10.1257/aer.99.3.1053. Research record.
  • Kiyotaki, Nobuhiro, and John Moore. 1997. “Credit Cycles.” Journal of Political Economy 105 (2): 211–248. DOI 10.1086/262072. University of Edinburgh research record. Research record.
  • Kydland, Finn E., and Edward C. Prescott. 1982. “Time to Build and Aggregate Fluctuations.” Econometrica 50 (6): 1345–1370. RePEc bibliographic record. Research record.
  • Lucas, Robert E., Jr. 1976. “Econometric Policy Evaluation: A Critique.” Carnegie-Rochester Conference Series on Public Policy 1: 19–46. DOI 10.1016/S0167-2231(76)80003-6. RePEc record. Research record.
  • National Bureau of Economic Research. n.d. “Business Cycle Dating Procedure: Frequently Asked Questions.” Business Cycle Dating Committee. Research record.
  • National Bureau of Economic Research. n.d. “US Business Cycle Expansions and Contractions.” Business Cycle Dating Committee. Research record.
  • Ramey, Valerie A. 2011. “Identifying Government Spending Shocks: It’s All in the Timing.” Quarterly Journal of Economics 126 (1): 1–50. DOI 10.1093/qje/qjq008. Research record.
  • Rey, Hélène. 2013. “Dilemma Not Trilemma: The Global Financial Cycle and Monetary Policy Independence.” VoxEU/CEPR, August. Author’s account of the 2013 Jackson Hole paper. Research record.
  • Richardson, Gary. 2013. “The Great Depression.” Federal Reserve History, November 22. Federal Reserve System. Research record.
  • Romer, Christina D., and David H. Romer. 2004. “A New Measure of Monetary Shocks: Derivation and Implications.” American Economic Review 94 (4): 1055–1084. DOI 10.1257/0002828042002651. Research record.
  • U.S. Bureau of Labor Statistics. n.d. “Concepts and Definitions (CPS).” Current Population Survey. Research record.
  • Weinberg, John A. 2013. “The Great Recession and Its Aftermath.” Federal Reserve History, November 22. Federal Reserve System. Research record.