RESEARCH DOSSIER

Budget Constraints and Intertemporal Feasibility

Abstract. Intertemporal feasibility is the capacity of an economic agent to implement a dated and, when uncertainty is present, state-contingent allocation under a specified system of resources, prices, settlement rules and enforceable financial contracts. This monograph reconstructs the relationship between period-by-period budget constraints, present-value resources, borrowing restrictions, contingent claims, liquidity and solvency. It establishes the conditions under which a lifetime budget summarizes a system of sequential budgets and identifies the additional restrictions required when credit, asset market completeness and settlement capacity are limited. The analysis develops exact finite-horizon derivations, numerical counterexamples and a comparison of household, enterprise and public-sector applications. Empirical research on consumption smoothing, fiscal transfers, lottery winnings, microcredit, savings access and household portfolios is evaluated by its identifying variation, outcome definition and institutional setting. The resulting synthesis treats the value of resources, the timing of their availability and the distribution of feasible transfers across future states as distinct components of economic explanation. The research is a theoretical and comparative synthesis of established economics, supported by independently derived illustrative calculations and a bounded review of published evidence.

Keywords: Budget constraints; intertemporal choice; present value; dynamic feasibility; borrowing constraints; liquidity; solvency; incomplete markets; consumption smoothing; state-contingent claims; household finance; financial institutions

Foundations of dated economic choice

Economic objects and temporal structure

A budget constraint specifies the set of allocations whose acquisition and settlement are compatible with available resources and the applicable rules of transfer. An intertemporal budget constraint extends this specification across dates. It describes how a decision at one date changes the resources available at another through saving, borrowing, investment, contractual commitments and other mechanisms that connect economic periods.

The economic object is the feasible allocation, represented by quantities of goods, services, productive inputs or financial payments indexed by their delivery dates. A unit of consumption today and a unit of consumption next year are different dated commodities. Their physical similarity does not determine their relative economic value. A transfer between dates requires a technology, a financial arrangement or an institutional commitment capable of changing the delivery schedule. Time and enforceable transfer therefore enter the definition of the budget itself.

A feasible set belongs to a specified decision environment. Prices determine the resources required for transactions; endowments and income determine initial and subsequent availability; contracts determine permissible transfers; settlement rules determine when commitments must be fulfilled. Preferences rank the resulting feasible allocations. Expectations describe the uncertain environment in which plans are formed. Keeping these objects distinct is essential for interpreting evidence: an observed allocation can change because preferences change, because the anticipated environment changes or because the feasible set changes.

The historical reconstruction begins with Fisher (1930). His market line for dated income represents opportunities to exchange present and future income at an interest rate within a simplified environment of certainty. The separation of expenditure timing from the timing of income is a central economic consequence of borrowing and lending. A modern statement of consumption-saving problems appears in Angeletos (2013), whose consumption and saving lectures organize sequential resource constraints and optimality conditions within dynamic models.

The analysis here begins with the transfer mechanism and its accounting before introducing preferences. It then relaxes the assumptions under which one present-value inequality summarizes the entire allocation problem. This ordering makes it possible to distinguish a feasible allocation, an optimal allocation, a solvent position and an allocation that can actually be implemented at every settlement date.

Scope and research method

The formal analysis uses finite discrete time as its principal setting; an infinite-horizon extension identifies the boundary conditions needed to rule out perpetual unfunded borrowing. Consumption is measured in units of a single date-specific composite good unless otherwise specified. Asset positions and rates are defined explicitly. The numerical examples are constructed analytical examples, rather than estimates from household or national data.

The empirical discussion compares evidence at the level of study design and reported findings. It includes theoretical original articles, institutional reproductions, lecture material, major empirical studies and official international survey or policy documentation. Several empirical articles were consulted through their publisher abstracts; inference from those articles is consequently limited to the findings and design descriptions accessible there. No microdata re-estimation, systematic meta-analysis or claim of exhaustive coverage is made.

The comparison centers on mechanisms that restrict the timing and state distribution of resource access. Household behavior supplies the principal empirical evidence. Firms and governments are examined to establish the reach and limits of the accounting logic. Their legal positions, objectives and financing capacities remain economically distinct.

Sequential budgets and present-value resources

A finite-horizon economy

Let dates be t = 0, 1, …, T. Let c_t ≥ 0 denote consumption, y_t exogenous net income, a_t the value of financial assets available before date-t consumption and R_(t+1) > 0 the gross real return between t and t+1. Negative assets represent debt. The single asset can carry resources forward or, where borrowing is permitted, bring them forward through a negative position. Specify the sequential budget as

a_(t+1) = R_(t+1)(a_t + y_t − c_t).

The definition treats a_(t+1) as the contractual payoff at the next date. The amount invested at date t is a_(t+1)/R_(t+1). This convention avoids mixing the principal of a loan with its repayment obligation. Initial assets a_0 are given. Terminal assets satisfy a_(T+1) ≥ 0 when debts must be settled and a nonnegative terminal balance may be retained.

Define the date-zero price of one unit at date t by Q_0 = 1 and Q_t = 1/(R_1 … R_t). Multiplying each sequential constraint by Q_(t+1) yields

Q_(t+1)a_(t+1) = Q_t a_t + Q_t y_t − Q_t c_t.

Summation over dates cancels the intermediate asset positions. The resulting equality is

Σ_(t=0)^T Q_t c_t + Q_(T+1)a_(T+1) = a_0 + Σ_(t=0)^T Q_t y_t.

Define lifetime resources W = a_0 + Σ Q_t y_t. The terminal restriction gives the necessary condition Σ Q_t c_t ≤ W. Under the assumptions of one unrestricted borrowing-and-lending instrument, a common positive return, known incomes and no intermediate balance restrictions, it is also sufficient. To demonstrate sufficiency, start with the given a_0, construct the asset path from the sequential constraints for the proposed consumption sequence, and apply the equality above. The inequality ensures a_(T+1) ≥ 0. Every constructed position is admissible under unrestricted borrowing and lending.

This argument establishes precisely what a lifetime budget summarizes. It eliminates intermediate assets from an accounting system whose feasible asset positions are unrestricted until the terminal date. With additional interim restrictions, the same elimination gives a necessary resource condition that must be supplemented by further inequalities.

Timing and a numerical allocation

Consider two consumption dates, initial assets zero, income 20 at date zero and 110 at date one, and a gross real return of 1.10. Lifetime resources are 20 + 110/1.10 = 120. The allocation c_0 = 70 and c_1 = 55 exhausts those resources: 70 + 55/1.10 = 120. At date zero it requires a loan principal of 50 and a next-date repayment of 55. Date-one income then covers repayment and the proposed consumption.

The same allocation becomes infeasible as soon as initial borrowing is prohibited. Date zero offers only 20 in available resources, although lifetime resources are unchanged. The allocation (20, 110) remains feasible and also exhausts the lifetime budget. Consequently two consumption profiles may have the same present value while requiring different financial capabilities. Present-value equivalence is an economic property of a specified transfer market, rather than a universal guarantee of implementability.

The example also distinguishes positive wealth from liquid means at a particular time. A claim to sufficiently large future income can raise the present value of lifetime resources while leaving the date-zero payment capacity unchanged. Whether lenders recognize that future income as pledgeable collateral is an institutional question that the present-value calculation alone cannot settle.

Price units and the interest rate

The discount factor for a dated commodity is expressed in compatible purchasing-power units. If the model uses nominal cash flows, it requires nominal discount rates or an explicit conversion to real units. In the deterministic case with a price level P_t, a nominal gross return R^n between dates t and t+1 yields a gross real return R^n P_t/P_(t+1). Under uncertainty realized real returns depend on the state, and an expectation of inflation alone generally does not supply the entire pricing relationship.

Discounting future resources at a common market rate is appropriate in the simple certainty environment above. Credit spreads, inaccessible investment opportunities, taxes and risk change the relevant prices or introduce restrictions. A calculated present value is conditional on the discount convention and the contractual opportunities represented by it. Comparisons between quantities calculated with different units or discount conventions require explicit reconciliation.

Credit restrictions and the geometry of feasible consumption

Prefix budget conditions

Impose an intermediate borrowing bound a_(t+1) ≥ −B_(t+1), where B_(t+1) is a nonnegative maximum debt payoff. Telescoping the first t+1 sequential equations gives

Σ_(j=0)^t Q_j(c_j − y_j) = a_0 − Q_(t+1)a_(t+1).

The borrowing restriction implies

Σ_(j=0)^t Q_j(c_j − y_j) ≤ a_0 + Q_(t+1)B_(t+1).

These are prefix budget inequalities. At every date they bound cumulative discounted expenditure in excess of income by initial wealth plus the discounted debt that can still be carried forward. Together with nonnegative consumption and the terminal budget condition, they are sufficient for the sequential model with the specified deterministic debt bounds: the recursively constructed position satisfies each bound exactly when its corresponding prefix inequality holds.

The lifetime budget is the final member of this family with terminal debt disallowed. The intermediate members identify the restrictions that disappear when borrowing is unrestricted. The full feasible set is their intersection, so relaxing an intermediate bound enlarges the set only where the bound previously excluded an otherwise admissible allocation. This observation explains why additional credit may be irrelevant for a household far from its borrowing limit and valuable for another with the same lifetime income but an imminent payment.

Return to the example. If permitted repayment debt at date one is at most 22, the date-zero loan principal is at most 22/1.10 = 20. Consumption at date zero is therefore at most 40. The lifetime boundary stays unchanged; its accessible portion changes. At the extreme point c_0 = 40, date-one consumption is 110 − 22 = 88. These amounts satisfy both the present-value equation and the intermediate borrowing constraint.

A borrowing limit can itself depend on income, collateral, credit history, existing commitments or the realization of uncertainty. A fixed cap is an analytical representation of the permitted set conditional on those variables. When the cap depends on the agent's choices, the complete model must characterize that dependence and its effect on feasible transitions. A linear inequality with an externally given cap cannot capture every endogenous credit mechanism.

Different saving and borrowing rates

Let m denote liquid resources today and y future income. Let R_s be the positive gross saving return and R_b the positive gross borrowing cost, with R_b ≥ R_s. In a two-date model with no terminal debt, the boundary for future consumption is

c_1 = y + R_s(m − c_0) if c_0 ≤ m,

c_1 = y − R_b(c_0 − m) if c_0 ≥ m.

The boundary is continuous at c_0 = m and is piecewise linear. Its slope changes from −R_s to −R_b. With R_b ≥ R_s, it is concave; the associated feasible region with nonnegative consumption below the boundary is convex. A larger borrowing cost reduces future consumption for any given borrowed amount. A binding quantity limit removes the portion of the borrowing segment beyond the contractual cap.

The kink creates a range of preferences under which the optimal position involves neither borrowing nor saving. At the kink, the marginal utility trade-off can lie between the two market exchange rates. Observed zero net borrowing therefore admits several explanations: a deliberate choice at a price discontinuity, a binding quantitative restriction, the absence of a profitable use for a loan or an institutional inability to enter the market. The sign of aggregate borrowing cannot identify which explanation is operative.

Fees, minimum loan sizes and indivisible transactions produce further changes. A fixed loan fee can create a discontinuity between taking no loan and taking a positive loan. A minimum investment can remove intermediate allocations. An asset that can be sold only in indivisible units produces a discrete opportunity set. In these settings a local tangent condition can fail to characterize a global optimum; direct comparison of admissible alternatives becomes necessary.

Contract enforcement and equilibrium access

A household's ability to promise future repayment depends on the lender's ability and willingness to accept the promise. Future earnings may be uncertain, difficult to verify or legally unavailable as collateral. The value to the borrower of future income does not establish its market value as pledgeable security.

Stiglitz and Weiss (1981) demonstrate in their credit rationing analysis how interest rates can alter borrower selection and incentives. An increase in the contractual rate need not improve a lender's expected return monotonically. Their model establishes the possibility of equilibrium rationing even where borrowers would accept the prevailing terms or a higher rate. This is a mechanism for quantity restrictions arising from the structure of the credit market.

The fixed borrowing bounds used in the mathematical reconstruction are compatible with several institutional origins. They may represent lender screening, collateral requirements, regulatory rules, contractual covenants or limits on enforceability. Their common role is to restrict an admissible asset transition. The economic consequences depend on the rule's actual design, so evidence of a constrained allocation does not by itself establish the Stiglitz–Weiss mechanism.

Optimal choice within a restricted feasible set

Euler conditions and shadow values

For two dates, let b denote net saving in date-zero units, m current resources and y_1 future income. Then c_0 = m − b and c_1 = y_1 + Rb. Let preferences be represented by u(c_0) + β E[u(c_1)], where β > 0 is the date preference parameter and u is increasing, differentiable and concave. Assume strictly positive consumption in all states considered and a fixed positive return R. A credit cap takes the form b ≥ −B.

For an interior solution with respect to consumption and no binding credit cap, a small increase in b sacrifices u'(c_0) marginal utility today and adds βR E[u'(c_1)] expected discounted marginal utility tomorrow. The optimality condition equates the two:

u'(c_0) = βR E[u'(c_1)].

When the lower cap on b binds, the constrained first-order condition becomes

u'(c_0) = βR E[u'(c_1)] + μ,

with μ ≥ 0 and μ(b + B) = 0. This follows from the multiplier on b + B ≥ 0 in the maximization problem. Under concavity and an admissible interior consumption domain, these conditions characterize the solution together with feasibility. A stochastic asset payoff replaces R E[u'(c_1)] with E[R u'(c_1)].

The multiplier represents the marginal value of relaxing the cap in date-zero units. It can be positive when the marginal utility of current consumption exceeds the discounted marginal utility obtainable through saving. The economic significance of a restriction thus depends jointly on its position and the agent's objective. A legal ceiling on debt can coexist with a zero shadow value for a household that would choose a smaller loan.

Euler equations are derived from permissible marginal substitutions. If a transfer is contractually impossible, its unrealized marginal benefit has no direct market transaction through which it can be observed. In empirical work an Euler inequality, a discrete-choice comparison or a model with adjustment costs may be necessary to interpret the observed allocation.

A complete worked optimization

Take the earlier two-date economy and let u(c) = ln(c), β = 1, and R = 1.10. With unrestricted transfers and positive consumption, the first-order condition is 1/c_0 = 1.10/c_1. It follows that c_1 = 1.10c_0. The present-value equality becomes 2c_0 = 120, giving c_0 = 60 and c_1 = 66. The necessary loan principal is 40 and its repayment is 44.

With the maximum repayment cap of 22, the maximum loan principal is 20. The unconstrained allocation is infeasible. The strictly concave objective has its constrained maximum at the boundary c_0 = 40, c_1 = 88. The multiplier for the principal cap equals 1/40 − 1.10/88 = 0.0125 utility units per additional unit of admissible principal at the boundary. Relaxing the cap initially increases attainable utility at this rate; the multiplier declines to zero at the unrestricted optimum.

The two optima use the same present value of consumption. Their utility levels differ because of the particular preference ordering and the inability to change allocation timing. Calculations of the financial opportunity set and calculations of welfare therefore require separate input. Higher current consumption is desirable in this example because of the given utility and the position on the feasible boundary. Other objectives, precautionary motives or risks can select different positions.

Preferences, expectations and the observational problem

Consumption responses to expected future income are often used to assess whether households can smooth consumption. Such responses also depend on preferences, income uncertainty, the anticipated persistence of shocks, demographic change and asset adjustment costs. Identification requires evidence capable of distinguishing those channels.

Hall (1978), in his stochastic consumption analysis, derives benchmark implications for predictable marginal utility changes and tests related empirical restrictions. A useful limiting case is βR = 1, in which the conditional expected future marginal utility equals current marginal utility under the model assumptions. An exact random-walk implication for consumption itself requires further conditions, such as linear marginal utility. The benchmark is a conditional theoretical result rather than a universal characterization of observed consumption.

Zeldes (1989) uses Euler implications of liquidity constraints to investigate household consumption with panel data. The analysis connects a potential inequality in intertemporal optimality to differences in household resources. Empirical classification by asset holdings is informative about possible constraints, but it is endogenous to past income, preferences and choices. An inferred wedge deserves investigation of these alternative mechanisms before it is attributed to a single credit restriction.

Carroll (1997) develops buffer-stock saving under uncertainty and sufficient impatience in the model. Precautionary accumulation can arise because adverse future income realizations are consequential. A household may keep a reserve without a presently binding cap; a household close to a cap may consume little today to reduce the chance of a costly future shortage. Levels of liquid wealth and marginal propensity to consume are thus joint outcomes of a dynamic decision environment.

Uncertainty, contingent claims and market incompleteness

State prices and the feasibility of insurance

With uncertainty, an allocation specifies both a date and a state. Suppose the second date has a finite set S of possible states, with date-zero prices q_s > 0 for securities delivering one unit of consumption in state s and zero in the others. Let current resources be m, future incomes y_s and desired consumption c_s. If all state transfers are available in unrestricted quantities, the present-value budget is

c_0 + Σ_(s∈S) q_s c_s ≤ m + Σ_(s∈S) q_s y_s.

To see sufficiency, choose the state payoff c_s − y_s for every state, purchase or sell these contingent securities at their quoted prices, and settle current consumption from m. Free disposal or retained current balances accounts for slack. Negative security positions must actually be permitted for this construction.

State prices are the prices of economic transfers across states. They need not coincide with the probabilities of those states. If a risk-free one-unit payoff is available as the combination of one unit in every state, its price equals Σ q_s in the complete-market setting with consistent prices. Risk preferences, endowments and market equilibrium can influence the state prices. Replacing them with physical probabilities without adjustment changes the economic budget.

Arrow (1964) analyzes securities and risk-bearing through a theoretical setting in which contingent transfers support risk allocation under the specified assumptions. The result identifies a deep connection between the financial span of available securities and the feasible distribution of future consumption. Market completeness concerns what payoff vectors can be obtained; it does not assert unlimited contractual enforceability in every actual economy.

A counterexample with one bond

Consider two equally priced future state claims, q_H = q_L = 0.5, date-zero resources of 50, and future endowments y_H = 100 and y_L = 0. The proposed allocation c_0 = c_H = c_L = 50 satisfies the complete-market budget. The required net claims are −50 in H and +50 in L. Their date-zero value is zero. The arrangement can be interpreted as exchanging some high-state resources for low-state consumption.

Now restrict the market to a risk-free bond that delivers the same amount in both states. If current consumption remains 50 and there are no other current transactions, the bond position must be zero because its price is positive. The resulting future consumptions are 100 and 0. Even if current consumption were allowed to vary, a single bond payoff could not make both future consumptions 50 given the stated incomes: it would have to equal −50 in one state and +50 in the other.

The desired allocation satisfies the scalar state-price budget but lies outside the attainable payoff space. This is a different mathematical obstruction from an intermediate borrowing cap. The cap restricts quantities of transfers that are available in principle; an incomplete market restricts the directions of possible transfers. Both can prevent implementation of a resource-valued allocation.

More generally let A be a matrix whose columns are the future state-payoff vectors of traded assets and x the vector of positions. With unrestricted portfolios, an exact desired payoff vector v is attainable only when v belongs to the column space of A; its current cost must also fit the current budget. Full row rank supplies payoff spanning in the finite model, subject to consistent prices and unrestricted positions. With position constraints, collateral rules or transaction costs, even full rank may be insufficient. The admissible portfolios form a restricted subset of the vector space and their image under A determines the attainable payoffs.

Nonanticipativity and state-dependent borrowing

A dynamic plan must be measurable with respect to the information available when the relevant action is taken. A date-zero portfolio can be contingent in its contracted future payoff, while the choice of that portfolio is made with date-zero information. In a scenario tree, paths that share the same information history must share decisions made at that history. Allowing a decision to depend on a future state before it becomes observable creates an artificial feasibility expansion.

Debt repayment may also be feasible in some states and impossible in others. With a single risk-free bond, no terminal default and nonnegative future consumption, the two-date specification requires y_s + Rb ≥ 0 for every state. Consequently b ≥ −min_s(y_s)/R, assuming the minimum exists and the incomes are nonnegative. If the lowest income is zero, a risk-free debt position is infeasible under these assumptions, even when expected income is large.

The example gives a natural borrowing limit induced by the repayment and consumption requirements. Actual institutions can permit default, restructure claims, use collateral or offer income-contingent repayment. Such mechanisms change the set of promised and realized transfers, together with the welfare and incentive consequences. Their introduction demands a definition of the outcomes after default and of the claims that creditors can enforce.

Ex ante feasibility and ex post adjustment

A plan may be formulated as a policy mapping observable states into actions. Feasibility is checked at every reachable history under the specified transitions. Ex ante resources and ex post available balances then enter different parts of the decision problem. An agent can have a valuable expected income stream while experiencing a severe shortage in one realized state.

An unexpected shock can also invalidate a plan that was feasible relative to earlier forecasts. Updating income expectations, prices or the set of enforceable transfers changes the current opportunity set. Reoptimization is a response to the revised environment. A rigorous dynamic model records which constraints and information remain in force at the time of revision, including commitments inherited from previous decisions.

Long horizons, solvency and dynamic consistency

The boundary condition in infinite time

The finite-horizon accounting identity continues to hold up to every finite truncation of an infinite sequence:

Σ_(t=0)^T Q_t c_t = a_0 + Σ_(t=0)^T Q_t y_t − Q_(T+1)a_(T+1).

To derive an infinite-horizon resource bound, specify the limiting behavior of assets. One sufficient no-Ponzi condition for this deterministic formulation is liminf_(T→∞) Q_(T+1)a_(T+1) ≥ 0. If the discounted income sum converges and nonnegative consumption has a well-defined discounted sum, the finite identities yield Σ_(t≥0) Q_t c_t ≤ a_0 + Σ_(t≥0) Q_t y_t. With finite resources and the stated boundary condition, consumption cannot be financed by a permanently negative discounted terminal position.

Omitting the condition allows a misleading sequence. Let initial assets and all incomes equal zero, let R be constant and greater than one, and choose consumption one at date zero and zero thereafter. The initial loan is rolled over indefinitely, producing a_(T+1) = −R^(T+1) for every T. The discounted terminal position remains −1. Every local accounting equation holds, yet the sequence violates the stated no-Ponzi condition. This demonstrates why local bookkeeping and overall solvency are different mathematical requirements.

Transversality conditions arise from optimization under particular objectives and admissibility assumptions. They often restrict the limiting value of accumulated resources in an optimal allocation. A solvency restriction defining admissible debt paths enters the feasibility problem even before preferences are introduced. The two conditions can be related under a specified model, but an optimality condition should not be substituted for a missing contractual feasibility rule.

Dynamic programming and inherited obligations

A dynamic program describes a current state containing all variables needed to determine available actions and future transitions. These can include liquid assets, illiquid wealth, outstanding obligations, collateral, income, market access and relevant information. Let F(s) denote the feasible action set at state s and let s' follow from the selected action and next shock. A Bellman objective takes the form

V(s) = sup_(x∈F(s)) {u(s,x) + β E[V(s') | s,x]}.

This representation presupposes the required state sufficiency, well-defined preferences and integrability; it does not by itself establish existence or uniqueness of a solution. Feasibility requires a definable transition for each admissible action and compliance with future constraints. A recursively computed choice can otherwise lead into a state in which no allowed continuation exists.

A maturity structure is economically important because the inherited state is a vector of contractual obligations, rather than a single present-value number. Equal-value liabilities with different maturities expose borrowers to different funding situations. Replacing the vector with a scalar total assumes that all relevant timing differences can be neutralized by accessible transfers.

Time consistency concerns the relation between a plan chosen today and decisions taken by future decision makers. Under stable exponential discounting, a fixed model and the standard dynamic programming assumptions, an optimal contingent policy admits recursive implementation. Changes in information can produce new actions within the same policy. Preference changes, commitment limitations or renegotiation can require a different analysis. A budget-feasible plan and a dynamically implementable plan therefore answer related but distinct questions.

Liquidity, portfolios and financial institutions

The settlement dimension of wealth

Liquidity is the capacity to mobilize resources at the date and under the conditions required for a transaction. It depends on the available instruments, conversion costs, contractual rights and time needed to obtain spendable funds. Solvency concerns whether resources cover obligations under the stipulated valuation and admissibility rules. A balance-sheet surplus can coexist with a settlement shortage.

Consider an agent with wealth of 120 measured at quoted current valuations, comprising 10 in liquid funds and 110 in an asset that cannot be converted before an imminent bill of 25 comes due. With no alternative financing, the agent faces a payment shortfall of 15. Total valued assets exceed the bill, yet the present transaction is infeasible. Removing the conversion restriction would alter the opportunity set without requiring a change in the quoted value of aggregate wealth.

Illiquid assets can generate higher returns, commitment benefits, housing services or other advantages. Selling them may incur taxes, fees, price concessions or loss of productive capacity. In a multicomponent budget, an asset's transfer into the liquid account is governed by its own conversion rule. The values of two portfolios on a consolidated balance sheet can be equal while their accessible transfer sets differ.

Kaplan, Violante and Weidner (2014) document wealthy hand-to-mouth households in a study using household portfolio evidence from eight countries: the United States, Canada, Australia, the United Kingdom, Germany, France, Italy and Spain. The economic relevance is the coexistence of relatively substantial illiquid holdings with limited liquidity. Cross-country classifications depend on the assets and survey conventions included. Their evidence supports studying the composition of wealth, rather than identifying constraints solely through its total value.

Financial intermediation and collective liquidity

An intermediary can pool funds and issue claims with a maturity or redemption arrangement that differs from the timing of its assets. This allows some depositors to access resources when the underlying investment has not matured. It also creates dependence on the intermediary's ability to meet simultaneous withdrawals.

Diamond and Dybvig (1983) establish in their model of bank runs how deposit contracts can furnish liquidity insurance in the presence of uncertain individual consumption timing and how the same arrangement can admit a run equilibrium. The model is useful because it connects individual demand for early consumption to a collective constraint on settlement. The institutional details of any actual banking system, including liquid reserves, access to emergency funding and deposit protection, determine how the analogy applies.

The relevant distinction is between economic value at the planned realization date and resources available under acceleration of payments. A sequence of individually admissible claims may place stress on an intermediary's joint ability to pay if withdrawals occur at the same time. Conversely, pooling independent timing needs can widen the attainable set. The aggregate benefit and exposure both arise from the transfer technology and contract structure.

The institutional cost of constrained adjustment

Households can respond to short-term scarcity through formal borrowing, transfers from relatives, delaying bills, selling inventories, postponing consumption, taking additional work or using savings. The effective costs and feasible quantities of these alternatives differ. A measured expenditure response represents choices within this broader adjustment environment.

Credit access may be legally present but practically costly. Savings access may be formally available yet expensive to use or insecure. Payment instruments may be functional only within a network of merchants and financial providers. The empirical meaning of liquidity is consequently tied to the settlement technology and the horizon of the measured outcome. An asset definition appropriate for an annual wealth survey may be inadequate for an analysis of payments due tomorrow.

Empirical identification of intertemporal constraints

Observable outcomes and identifying variation

An intertemporal constraint is a restriction on the feasible set. Its presence and shadow value generally cannot be read directly from one observed allocation. A compelling empirical design identifies a change in income timing or financial access, measures the resulting adjustment and specifies which alternative explanations are addressed by the identifying variation.

Income shocks differ in anticipation, persistence, size, risk and payment timing. A transfer announced long before receipt can influence prior saving or borrowing. A household that anticipates a payment but cannot borrow may still spend at receipt. A household with unrestricted credit might also respond at receipt because the information contained in the payment changes expectations, or because its durable purchase technology is discrete. The timing of expenditure offers evidence about mechanisms when the information environment and outcome definition are established.

An experiment or quasi-experiment that identifies the causal effect of access or cash receipt establishes a response to that intervention in its population and period. Translating the estimated effect into a structural borrowing cap, a marginal utility wedge or a welfare calculation requires an additional maintained model. The distinction preserves the inferential value of the study while avoiding overinterpretation.

Temporary income and fiscal-payment evidence

Johnson, Parker and Souleles (2006) examine variation in the timing of US rebates through the 2001 tax rebate study. The reported estimate is that households spent approximately 20–40 percent of rebate amounts on nondurables in the three-month receipt period. The reported cumulative response rises through the following quarter. The study's timing variation is important because ordinary correlations between income and spending can reflect joint economic conditions or household characteristics.

Parker, Souleles, Johnson and McClelland (2013) examine the 2008 stimulus payments. Their publisher abstract reports nondurable spending responses of about 12–30 percent during the three-month receipt period, and larger estimates when durable goods are included. Expenditure on a durable purchases a stream of future services; a measured durable outlay is not equivalent to current-period consumption of that entire value. The outcome definition materially changes the estimated spending response.

The two US studies offer evidence of the timing and magnitude of spending adjustments around distinct public interventions. The ranges describe reported results rather than pooled structural parameters. Differences in samples, economic conditions, payment programs, timing and specifications constrain direct comparisons. The figures identify expenditure responses at stated horizons and should not be treated as permanent measures of households' borrowing capacity.

Lottery winnings and wealth composition

Fagereng, Holm and Natvik (2021) study MPC heterogeneity and balance sheets with Norwegian administrative lottery evidence. The reported expenditure response varies with prize size, liquid holdings and life-cycle circumstances. The design offers a source of income variation different from an ordinary wage fluctuation. The measured responses show why a representative propensity to spend can obscure the distribution of underlying household situations.

Prize size matters because a given windfall can cross different liquidity thresholds. A small transfer may be absorbed entirely by an immediate shortfall; a larger transfer may exceed that shortfall and permit saving or investment. At the same time, prize size and liquid assets can be associated with different household groups. A subgroup comparison cannot be interpreted as a causal effect of liquidity alone without the relevant identification assumptions.

For analysis of transfers, define a measured response as the change in expenditure attributable to an intervention divided by the transfer amount, over a specified horizon. This ratio is sensitive to measurement of durable purchases, debt repayment, informal transfers and changes in other income. It is an empirical statistic rather than a primitive preference parameter. Its relationship to a borrowing constraint depends on the model used to connect observed expenditure to feasible choices.

A quantitative interpretation of heterogeneous responses

Consider a purely illustrative population with two equally sized groups receiving the same one-time transfer. Suppose the first spends 80 percent of its transfer during the observation window and the second spends 20 percent. The aggregate spending response is 50 percent of the total transfer. If the same transfer budget is instead allocated 75 percent to the first group and 25 percent to the second, with behavior held fixed as an explicit counterfactual assumption, the response is 65 percent.

The calculation illustrates weighting by transfer dollars. It provides no empirical estimate of either response and no evidence that reallocating transfers would leave behavior unchanged. It clarifies why an aggregate statistic does not uniquely identify the distribution of constraints, and why the allocation of payments can affect a measured total response even if its group-specific components are fixed.

Credit and liquidity effects also evolve across horizons. A household may initially save a transfer and spend it when a bill comes due. Another may use it for debt repayment, reducing later interest or principal payments and raising later feasible consumption. A short observation window may classify both responses as low spending. Longitudinal tracking of assets, debt and expenditure is needed to understand the complete adjustment.

International evidence and institutional heterogeneity

Expanding borrowing opportunities

Banerjee, Duflo, Glennerster and Kinnan (2015) analyze randomized microfinance access in Hyderabad, India. Their reported experiment introduces microcredit access through randomized treatment neighborhoods. The publisher abstract reports an 8.4 percentage-point increase in take-up, effects on some business outcomes, and no statistically significant effect on aggregate consumption in the reported comparison.

This design distinguishes the offer of a financing opportunity from universal borrowing. The treatment modifies access; actual take-up remains an individual decision influenced by the household's circumstances and the terms of the product. The causal effect of being offered access is therefore a central result. Estimating the causal effect of borrowing itself from the offer requires additional assumptions concerning compliance and the channels through which the offer affects outcomes.

The lack of a statistically significant aggregate consumption effect is compatible with several economic situations, including heterogeneous uses of credit, limited take-up, adjustment through business investment and differences in the time needed for returns to materialize. It does not establish that the effect is exactly zero for each borrower or that borrowing access is irrelevant to all feasible plans.

Expanding saving opportunities

Dupas and Robinson (2013) examine savings access in Kenya using randomized access to noninterest-bearing deposit accounts. The publisher abstract reports increased saving, productive investment and expenditure among market women offered the accounts, despite withdrawal fees. A corresponding impact was not detected among bicycle taxi drivers.

The study concerns the capability to retain and mobilize resources over time. A financial intervention can improve intertemporal feasibility through secure or effective saving arrangements even without enlarging a borrowing limit. A household with opportunities to accumulate small balances can finance a future indivisible outlay that would otherwise remain outside its attainable set.

The comparison between market women and bicycle taxi drivers belongs to the study's occupational and demographic setting. It cannot isolate a universal gender effect because occupation and gender are intertwined in the stated groups. The important economic result is that access to the same general type of financial instrument can have different consequences depending on the activity, income process and usable transfer technology.

International measurement of access and resilience

The World Bank's Global Findex Database 2025 describes survey evidence from approximately 148,000 adults in 141 economies, collected in 2024. Its coverage of account ownership and use, savings and borrowing, financial resilience and related digital dimensions makes it useful for comparative institutional research.

Survey measures identify experiences and reported capabilities that complement expenditure-response evidence. An account ownership indicator concerns formal access to a financial instrument. A question about raising emergency funds concerns reported capacity under a particular hypothetical event. Actual settlement capacity for a dated obligation depends on the amount, time limit, available networks and realized circumstances. Each indicator represents a specific observation rather than the entire feasible set.

International comparisons require attention to survey wording, sampling and weighting, currency units, price levels, financial infrastructure and social arrangements. The same nominal value of emergency resources has different purchasing power in different economies. Informal financing can be economically material even when it is not captured by formal-credit measures. Descriptive differences in indicators can motivate hypotheses about transfer constraints; they do not automatically identify the causal effect of a country's institutional design.

Comparative evidence register

Research design Identifying variation or evidence Observed outcome Inference boundary
US rebates and stimulus payments Variation in payment timing Expenditure within specified quarters Timing responses, not a universal debt cap
Norwegian lotteries Lottery windfalls and administrative balance sheets Heterogeneous expenditure responses Prize and wealth differences require care in structural interpretation
Hyderabad microfinance Randomized neighborhood access Take-up and household or business outcomes Access effect differs from effect of borrowing for every person
Kenyan savings accounts Randomized account access Saving, investment and expenditure Study-specific occupational and demographic heterogeneity
Global Findex International survey responses Financial access, use and resilience Descriptive comparisons rather than identified causal effects

The studies support a comparative approach in which the financial mechanism is specified before the outcome is interpreted. Cash receipt, credit access and saving access modify different aspects of the transfer environment. Their effects therefore need not coincide even if they are measured through an expenditure outcome.

Production, public finance and aggregation

Investment opportunities and enterprise financing

A productive enterprise faces intertemporal constraints because inputs often require payment before the resulting output can be sold. Let an indivisible investment cost 80 today and deliver a certain net payoff of 110 next period. At a gross discount rate of 1.10, its net present value is 110/1.10 − 80 = 20. If the firm has 20 in accessible funds and a maximum available credit principal of 30, the most it can mobilize today is 50. The project is economically valuable under the valuation assumptions but infeasible under the specified funding rule.

Funding constraints interact with the production set. If an input can be purchased in smaller quantities, the firm may choose a reduced scale. An indivisible investment instead demands a threshold of available financing. The resulting change in production can be discontinuous when a small increase in the credit limit crosses that threshold.

A positive net present value is conditional on the relevant cash flows and discount rate. It does not establish a lender's recoverable payoff, the division of surplus between financier and entrepreneur or the availability of required early inputs. The condition for investment combines productive opportunity, funding timing and contractual implementation. An enterprise can face a valuable opportunity while lacking the instruments needed to undertake it.

The distinction also matters for working capital. A firm may hold inventories and receivables of considerable value while facing wages or supplier bills before receivables settle. Converting the receivables into cash requires an enforceable claim, a buyer or lender, and acceptable terms. An accounting measure of total current assets can therefore be supplemented by maturity information and actual payment arrangements.

Sovereign debt and fiscal arithmetic

For a simplified government budget, let D_t denote debt outstanding at date t, R the gross interest factor between dates, and S_(t+1) the primary surplus at the next date, expressed in the same units as debt. Assume no valuation changes, privatization receipts, contingent liability realizations or other stock-flow adjustments. Then

D_(t+1) = R D_t − S_(t+1).

If GDP grows by gross factor 1 + g and d_t = D_t/Y_t, s_(t+1) = S_(t+1)/Y_(t+1), the debt ratio satisfies

d_(t+1) = [R/(1 + g)] d_t − s_(t+1).

Writing R = 1 + r, the primary surplus relative to next-date GDP that stabilizes the debt ratio is

s = [(r − g)/(1 + g)] d_t.

Rates and GDP must be expressed on a consistent nominal or real basis. As an illustrative arithmetic example, if d_t = 0.80, r = 0.05 and g = 0.02, the stabilizing primary surplus is approximately 0.02353, or 2.35 percent of GDP. This is a conditional calculation, not a fiscal recommendation or forecast.

Adrian, Gaspar and Gourinchas (2024), in their fiscal risk discussion, address the significance of debt burdens, financing costs and growth for fiscal conditions. Debt dynamics can incorporate variations in interest rates, maturity, currency denomination and economic growth. Public budgets also face taxes, spending commitments and institutional limits whose effects cannot be captured by a single constant-rate equation.

A sovereign's economic options depend on its tax capacity, currency regime, debt structure, legal obligations and monetary and financial institutions. A single household budget is therefore an insufficient model of sovereign finance. Individual monetary obligations can be settled in environments with sovereign monetary authority, while aggregate real resources remain limited. Debt in foreign currency introduces a further settlement requirement involving access to that currency. Fiscal feasibility must specify which institutional and resource constraints apply.

Aggregation across interconnected balance sheets

A financial asset held by one domestic sector may be a liability of another. Consolidation cancels matching internal claims under consistent definitions, while the underlying timing obligations remain consequential for individual members of the consolidated system. The external position of a nation concerns net claims on the rest of the world, but its domestic allocation depends also on the distribution of assets and debts among households, enterprises, intermediaries and government.

A scarcity of liquid means in one sector can constrain transactions even when another sector has substantial reserves. Transfers across sectors depend on financial institutions, contractual terms and policy arrangements. A consolidated balance-sheet statistic does not describe all of the distributional and timing restrictions within the economy.

Aggregating consumption responses likewise requires knowledge of the distribution of resources, shocks and feasible actions. The illustrative weighting calculation in the empirical chapter demonstrates one elementary aggregation issue. In a full economy, prices, incomes and interest rates may respond to changes in demand and credit access, so the total effect can differ from a fixed-price summation. General equilibrium introduces the requirement that the plans of agents are jointly consistent with markets, settlement, production and resource availability.

Analytical synthesis and research implications

The relation between valuation and implementation

The formal results establish a hierarchy of economic questions. A present-value condition determines compatibility with a specified system of intertemporal prices and total valued resources. Sequential inequalities determine compatibility with the dated borrowing limits and settlement rules. Contingent payoff restrictions determine which state-dependent allocations are transferable through available financial instruments. Optimization ranks the allocations in the resulting set.

These questions can be examined through explicit counterexamples. The deterministic consumption allocation (70, 55) satisfies its lifetime present-value constraint and fails immediate implementation under a zero borrowing bound. The contingent allocation (50, 50, 50) satisfies the state-price resource condition and cannot be obtained with a risk-free bond alone. The productive investment with a positive present value of 20 cannot proceed with accessible funding of only 50 against a cost of 80. Each example maintains economic value under its stated assumptions while identifying a distinct obstruction to implementation.

The analytical contribution of juxtaposing these cases is to make precise a recurrent source of error in applied reasoning: a scalar valuation can summarize a feasible set only when the relevant transfer and settlement assumptions justify that reduction. When restrictions affect timing, portfolio conversion or payoff spanning, the omitted variables carry explanatory information. The argument follows from established sequential accounting and finite-dimensional linear algebra. It is an original organization and comparative interpretation of those tools, without a claim to a new economic theorem or a named theory.

Empirical consequences and measurement

A study of intertemporal feasibility is strengthened by collecting dated cash inflows and outflows, debt maturity and repayment terms, liquid holdings, conversion costs for other assets, collateral rules, access to formal and informal instruments, and the contingencies under which claims can be enforced. These variables allow a proposed allocation to be compared with actual financing capabilities instead of relying exclusively on wealth or income.

Empirical tests should specify the time horizon, the state of information, the outcome and the variation that changes the opportunity set. The response to a cash payment has a different interpretation from the response to a loan offer or a savings instrument. The comparison of the US fiscal-payment studies, Norwegian lottery evidence and experiments in India and Kenya demonstrates the importance of this specificity.

Where constraints are latent, research can compare the implications of candidate mechanisms. An experimentally relaxed borrowing cap may change behavior among households that would otherwise choose positions beyond it. A reduction in savings conversion costs may affect future investment even in the absence of demand for borrowing. An income-contingent contract may alter responses to bad states without the same effect in good states. These are implications under specified alternatives, not unconditional empirical predictions.

A careful evaluation also separates the welfare value of access from the volume of transactions undertaken. Expanding a feasible set weakly raises the maximum of an unchanged objective if old choices remain available and no external effects intervene. This elementary set-inclusion result establishes an opportunity value. Actual welfare can depend on information, transaction costs, default consequences, changes in prices, strategic behavior and household decision structures. Measuring these dimensions requires further evidence.

Open research directions and limitations

Several research directions follow from the present synthesis. Matched transaction and balance-sheet data could establish how the timing of obligations interacts with liquid wealth and borrowing eligibility. Studies of contingent repayment contracts could measure how different pledgeability rules alter choices across realized states. International research could compare formal and informal adjustment instruments while harmonizing the timing and amount of emergency-resource indicators.

Further work could link microeconomic transfer constraints to sectoral settlement networks and maturity mismatches. Such analysis would require explicit balance-sheet consistency, institutional mechanisms and joint feasibility conditions. A separate investigation of counterparty networks could examine why individual default or illiquidity changes the resource access of other agents.

The numerical cases are proofs by construction or illustrations of conditional arithmetic. They are not calibrated economies. The empirical comparison is selective and includes publisher-abstract evidence, which limits conclusions to reported designs and outcomes. Reproduction of estimates, confidence intervals, full identifying-assumption audits and broader cross-country evidence would be required for a quantitative synthesis. The monograph establishes the general accounting and analytical distinctions within its specified setting and evaluates the reach of selected empirical findings.

Conclusion

Intertemporal feasibility is determined by the structure of admissible resource transfers across dates and states. A lifetime budget is an exact description of the sequential feasible consumption set under the explicitly stated conditions of unrestricted deterministic borrowing and lending, consistent prices and terminal settlement. Intermediate constraints add prefix inequalities. In a contingent economy, available payoff vectors and restrictions on portfolio positions determine whether a proposed state-dependent allocation can be obtained.

The distinction between feasibility and optimality establishes the role of preferences and shadow values. The distinction between valued resources and settlement capacity establishes the economic significance of liquidity and portfolio composition. The distinction between expected resources and attainable contingent transfers establishes the importance of insurance, enforceability and market completeness. Infinite-horizon analysis requires a solvency condition that prevents local accounting identities from supporting perpetual unfunded borrowing.

The reviewed research shows several routes by which financial conditions influence economic behavior: equilibrium credit rationing, precautionary accumulation, portfolio illiquidity, temporary income receipt, access to credit and access to saving. Their empirical manifestations vary with the institutional setting, the identifying intervention and the measured horizon. Financial feasibility therefore belongs to the actual architecture of an economy's contracts, assets, information and settlement systems.

Economic research gains precision when it specifies that architecture before interpreting an allocation's value or observed behavior. The dated budget, the permissible transfer, the realized funding position and the attainable contingent allocation together describe the resource environment within which economic choice takes place.

References

Fisher, Irving. 1930. The Theory of Interest, as Determined by Impatience to Spend Income and Opportunity to Invest It. New York: Macmillan. Original digital edition.

Angeletos, George-Marios. 2013. Consumption and Saving. Lecture notes, 14.05 Intermediate Macroeconomics, Spring 2013. MIT OpenCourseWare. Lecture notes.

Arrow, Kenneth J. 1964. The Role of Securities in the Optimal Allocation of Risk-bearing. Review of Economic Studies 31 (2): 91–96. DOI: 10.2307/2296188. Institutional article copy.

Stiglitz, Joseph E., and Andrew Weiss. 1981. Credit Rationing in Markets with Imperfect Information. American Economic Review 71 (3): 393–410. JSTOR: 1802787. Original article copy.

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. Institutional article copy.

Hall, Robert E. 1978. Stochastic Implications of the Life Cycle–Permanent Income Hypothesis: Theory and Evidence. Journal of Political Economy 86 (6): 971–987. DOI: 10.1086/260724. Author-hosted article.

Zeldes, Stephen P. 1989. Consumption and Liquidity Constraints: An Empirical Investigation. Journal of Political Economy 97 (2): 305–346. DOI: 10.1086/261605. Institutional article copy.

Carroll, Christopher D. 1997. Buffer-Stock Saving and the Life Cycle/Permanent Income Hypothesis. Quarterly Journal of Economics 112 (1): 1–55. DOI: 10.1162/003355397555109. Publisher abstract.

Kaplan, Greg, Giovanni L. Violante, and Justin Weidner. 2014. The Wealthy Hand-to-Mouth. Brookings Papers on Economic Activity, Spring 2014: 77–153 (including discussion). Original research paper.

Johnson, David S., Jonathan A. Parker, and Nicholas S. Souleles. 2006. Household Expenditure and the Income Tax Rebates of 2001. American Economic Review 96 (5): 1589–1610. DOI: 10.1257/aer.96.5.1589. Publisher article record.

Parker, Jonathan A., Nicholas S. Souleles, David S. Johnson, and Robert McClelland. 2013. Consumer Spending and the Economic Stimulus Payments of 2008. American Economic Review 103 (6): 2530–2553. DOI: 10.1257/aer.103.6.2530. Publisher article record.

Banerjee, Abhijit, Esther Duflo, Rachel Glennerster, and Cynthia Kinnan. 2015. The Miracle of Microfinance? Evidence from a Randomized Evaluation. American Economic Journal: Applied Economics 7 (1): 22–53. DOI: 10.1257/app.20130533. Publisher article record.

Dupas, Pascaline, and Jonathan Robinson. 2013. Savings Constraints and Microenterprise Development: Evidence from a Field Experiment in Kenya. American Economic Journal: Applied Economics 5 (1): 163–192. DOI: 10.1257/app.5.1.163. Publisher article record.

Fagereng, Andreas, Martin B. Holm, and Gisle J. Natvik. 2021. MPC Heterogeneity and Household Balance Sheets. American Economic Journal: Macroeconomics 13 (4): 1–54. DOI: 10.1257/mac.20190211. Publisher article record.

World Bank. 2025. The Global Findex Database 2025. Washington, DC: World Bank. Report and survey overview. Official report.

Adrian, Tobias, Vitor Gaspar, and Pierre-Olivier Gourinchas. 2024. The Fiscal and Financial Risks of a High-Debt, Slow-Growth World. IMF Blog, March 28. Official IMF commentary.