we-solve (beta)

Econ-379 · economics · Economics

Present bias and aggregate saving under borrowing constraints

LLM status Open unverified

Statement

ECON-026 | R | Source-backed agenda; exact formalization is editorial

Definitions and formal target

A date-$t$ household evaluates \[ u(c_t)+\beta_0\mathbb E_t\sum_{s>t}\delta^{s-t}u(c_s), \qquad 0<\beta_0\leq1,\quad0<\delta<1, \] subject to $a_{t+1}=(1+r)a_t+y_t-c_t\geq\underline a$ and stated solvency conditions. Specify sophisticated anticipation of future selves or naive misprediction and the available commitment devices.

For type distribution $F$, let $\mu_F$ be a selected stationary equilibrium distribution with finite aggregate assets $A(F)=\int a\,d\mu_F$. Define \[ \Delta A=A(F^{\beta_0=1})-A(F), \] where other type components and their joint distribution are preserved and equilibrium recomputed. Call $A(F)$ aggregate capital only when assets precisely supply productive capital.

Clarifications and required assumptions

Present bias does not yield an ordinary time-consistent Bellman problem. Specify the future-self equilibrium or decision rule, information, credit supply, income and commitment. Retain correlations among patience, borrowing limits and earnings when removing present bias. Establish or assume stationarity, finite moments, existence and selection. Assets may include government debt or housing, so their integral is not automatically capital. Micro estimates do not alone identify the equilibrium comparison.

Variants and changes of scope

  • Sophisticated, naive and partially sophisticated behavior differ at the same parameters.
  • Commitment accounts and different borrowing limits are separate feasible-set changes.
  • Transitions from a common initial distribution differ from comparisons between stationary distributions.

Audit conclusion and source relationship

[S1], PDF p. 6, motivates behavioral heterogeneity and empirical discipline. This stationary-assets target is editorial; the lecture gives no universal sign or existence theorem.

References and exact source roles

[S1] Benjamin Moll (n.d.). Lecture 11: Good Research Topics and Open Questions. Distributional Macroeconomics lecture slides; publicly archived at a 2019 URL.

Verified locator / source role: PDF p. 6: behavioral features, including present bias.

https://benjaminmoll.com/wp-content/uploads/2019/07/Lecture11_2149.pdf

Common conventions: Additional-source status and mathematical conventions

Of the 100 supplied ECON entries, 65 distinct problems appear here; 32 close matches refine earlier entries, and three duplicate questions map to existing entries. Appendix app:audited-crosswalk lists every destination. The attachment's P/R/S/U distinctions and source audits are retained. Its precise mathematical formalizations are editorial unless the individual audit says otherwise. Candidate subsidy targets ECON-004--006 retain their U flags; the resolved approval-core question ECON-007 updates OP-0202 above. The following supplied conventions govern both this part and the attributed ECON refinements elsewhere in the catalogue, subject to local restrictions.

Parameterized questions. A problem family lists inputs that must be fixed before an instance is evaluated: population, horizon, policies, information, data law, model class and welfare weights. These are required choices, not quantities the citations are assumed to specify. Undefined applications should not be treated as identified estimands.

Indices and integrability. Sets of agents, firms and regions are finite unless a population measure is explicitly used. $i,j$ index units, $r$ regions, $t$ dates and $h$ horizons. $\mathbb E$ and $\Pr$ refer to the declared population and law; $\mathbf1\{A\}$ is an indicator. Every displayed expectation and discounted sum needs existence in the stated domain. Infinite horizons use a discount in $(0,1)$ and a convergence condition; finite horizons may use discount one.

Optimization and existence. A displayed $\max$, $\min$, or $\arg\max$ is conditional on attainment. For finite feasible menus this is immediate; general spaces need continuity and compactness/coercivity or another existence argument. Without attainment, the target is the supremum/infimum and arbitrarily near-optimal feasible choices. $\inf\varnothing=+\infty$. Empty feasibility and an unidentified target are different issues.

Potential outcomes. $Y_i(z)$ is the outcome under a specified intervention, not the observed conditional mean among people choosing $z$. In binary comparisons $\Delta Y_i=Y_i(1)-Y_i(0)$. Specify assignment, treatment versions, compliance, information and observation horizon. Interference is allowed under a full policy vector; compressed exposure notation needs a justified mapping. No interference, consistency and overlap are substantive assumptions, not automatic consequences of notation.

Populations and observation. Fix baseline eligibility and population weights. Conditioning on post-policy employment, survival, relocation or program uptake can change the population being compared. Death is not ordinary loss to follow-up; outcomes truncated by death need a composite convention, joint survival target, or explicitly defined survivor stratum. Fertility-changing interventions need a population functional rather than an unexplained fixed descendant cohort. Measurement and missingness models must be supplied.

Identification. A structural model $m\in\mathcal M$ implies observable law $P_m$ and target $\theta(m)$. Identification means \[ P_{m_1}=P_{m_2}\ \Longrightarrow\ \theta(m_1)=\theta(m_2) \quad\text{for all }m_1,m_2\in\mathcal M . \] Without identification, the target is \[ \Theta(P;\mathcal M)=\{\theta(m):m\in\mathcal M,\ P_m=P\}. \] Writing $\theta(P)$ before establishing identification can hide incompatible latent models. Confidence sets additionally need a sampling law, confidence level, asymptotic sequence and uniformity class. Point identification, coverage and informative precision are separate properties.

Mediation. Total effects of randomized assignment are distinct from cross-world natural indirect effects. Belief/effort-channel effects require a meaningful mediator intervention and additional measurement, exclusion or mediation assumptions. Randomization of the initial policy alone does not supply those assumptions. Report bounds or interventional alternatives when the stronger target is unsupported.

Equilibrium. A counterfactual $W(z)$, $p(z)$, or $\mu_t^z$ requires individual optimization, feasibility, government/resource budgets, market clearing and state-transition laws. State equilibrium existence and selection, or report ranges over compatible equilibria. Initial-state integration includes subsequent endogenous transitions; current-state integration must use the policy-dependent distribution. Reduced-form invariance after policy is not assumed.

Welfare accounts. Scores, utility, health, dollars and ecological quantities require explicit conversion before addition. Fees, taxes, subsidies, wages and extortion payments are transfers between parties; distributional weights can make them welfare relevant, but their face value is not automatically a resource loss. State ownership, geographic boundaries, foreign-income weights and fiscal finance. Do not add profits to household welfare already including dividends, or count land capitalization and the underlying benefit twice. A benefit-cost expression must distinguish gross benefits from net welfare and subtract every real cost exactly once.

Derivatives and risk. Log derivatives require positive arguments; local responses require admissible perturbations and specified equilibrium branches. Differentiation under expectations needs regularity/domination, and boundaries may allow only one-sided derivatives. Risk uses a specified law; ambiguity uses a declared nonempty law set on a common history space. Adaptive policies must be nonanticipative. A robust criterion is an editorial choice unless attributed explicitly.

Scope overlaps. Allocation, collective choice, contracts, household bargaining and coordinated policy overlap economics with optimization and game theory. They are retained because they appeared in the original catalogue; no claim of a strictly game-theory-free selection is made.

Statement licence: CC-BY-4.0 · Source file

Contributions 2

Econ-379-A2 · Exploration · LLM-generated · claude opus 5.5 high ·

Claude Opus 5.5 High attempt on Econ-379

Present bias and aggregate saving under borrowing constraints: an exact finite-horizon result and general-equilibrium comparative statics

Abstract

(i) In a three-period deterministic model with CRRA utility and sophisticated quasi-hyperbolic agents, we prove that present bias lowers saving at every date, for every relative risk aversion $\rho>0$. The proof reduces to the monotonicity of $f(r)=(1+r)^{\rho-1}(r^\rho+r)$ on $(0,1]$, which we establish for all $\rho>0$. (ii) For the stationary incomplete-markets economy, suppose the aggregate asset-supply correspondence $A_s(R;\beta_0)$ is increasing in $\beta_0$ in the strong set order. Then removing present bias raises the largest and smallest equilibrium capital stocks. This holds even though the interest rate falls. (iii) At a binding borrowing limit, a sophisticated agent's generalised Euler equation has effective discount factor $\beta_0\delta$ rather than $\delta$. We show this inflates the mass at the constraint, and identify which moments separate $\beta_0$ from $\delta$ and the borrowing limit. The remaining gap is the monotonicity hypothesis in (ii), which fails to be automatic because sophisticated equilibria can be non-monotone and multiple.

1. An exact result

Let $u(c)=c^{1-\rho}/(1-\rho)$, with $R=\delta=1$, three periods, wealth $W_1$, and no income after period 1. Self 3 consumes its wealth. Self 2 maximises $u(c_2)+\beta u(W_2-c_2)$, so with $r:=\beta^{1/\rho}$ it chooses $c_3/c_2=r$, i.e. $c_2=W_2/(1+r)$. Self 1 maximises $u(c_1)+\beta[u(c_2)+u(c_3)]$, anticipating this.
Theorem: Self 1 chooses $W_2/c_1=f(r)^{1/\rho}$ with $f(r)=(1+r)^{\rho-1}(r^\rho+r)$. Self 2 chooses $c_3/c_2=r$. Both ratios are strictly increasing in $\beta\in(0,1]$ for every $\rho>0$. Hence present bias ($\beta<1$) strictly lowers wealth carried into periods 2 and 3, relative to $\beta=1$.
Proof: With $c_2=W_2/(1+r)$ and $c_3=rW_2/(1+r)$, self 1's continuation value is $\beta K\,u(W_2)$ with $K=(1+r)^{\rho-1}(1+r^{1-\rho})$. The first-order condition gives $(W_2/c_1)^\rho=\beta K=r^\rho K=f(r)$. Now \[ \frac{d}{dr}\log f=\frac{\rho-1}{1+r}+\frac{\rho r^{\rho-1}+1}{r^\rho+r}. \] After multiplying by $(1+r)(r^\rho+r)>0$, its sign is that of $N(r)=(2\rho-1)r^\rho+\rho r+\rho r^{\rho-1}+1$. If $\rho\ge1/2$, every term is nonnegative and $N>0$. If $\rho<1/2$, then $r^\rho\le1$ on $(0,1]$, so $N\ge-(1-2\rho)+1+\rho r+\rho r^{\rho-1}>0$. Since $r=\beta^{1/\rho}$ is increasing in $\beta$, the claim follows. ∎
A numerical check over $\rho\in[0.02,30]$ on a fine grid confirms the monotonicity. For $\rho>1$, sophistication makes self 1 anticipate over-consumption by self 2, which could in principle raise self-1 saving. The theorem shows that this "strategic saving" never dominates in this model.

2. Stationary general equilibrium

Consider an Aiyagari economy with an idiosyncratic income Markov chain, borrowing limit $\underline a$, and either sophisticated agents in the Markov-perfect intra-personal equilibrium selected by a stated rule, or naive agents. The capital demand $K_d(R)$ is decreasing. Let $A_s(R;\beta_0)$ be the set of aggregate assets at stationary distributions.
Proposition: Suppose $A_s(R;\beta_0)$ is nonempty and compact-valued, ascending in $(R,\beta_0)$ in the strong set order, and upper-hemicontinuous. Then the largest and smallest equilibrium capital stocks are nondecreasing in $\beta_0$. In particular, $\Delta A\ge0$ for the extremal equilibria when $\beta_0$ is raised to $1$, with other type components and correlations held fixed.
Proof: Equilibrium is a fixed point of $K\mapsto A_s(R(K);\beta_0)$ with $R(K)=K_d^{-1}(K)$ decreasing. Reparametrise by $R$: equilibria solve $K_d(R)\in A_s(R;\beta_0)$. The excess-supply correspondence $A_s(R;\beta_0)-K_d(R)$ is ascending in $(R,\beta_0)$. The monotone comparative statics of extremal zeros [mr] imply that the extremal equilibrium rates are nonincreasing in $\beta_0$. Then $K=K_d(R)$ is nondecreasing. ∎
Remark (Why the hypothesis is the crux): For naive agents with CRRA, and for time-consistent agents, policy functions are monotone in patience under standard conditions, and the hypothesis is plausible. For sophisticated agents, Markov-perfect consumption functions can be discontinuous and non-monotone, and equilibria multiple. Theorem Theorem shows monotonicity in a finite-horizon deterministic case. Extending it to the stationary stochastic case with a binding limit is the open step.

3. Borrowing constraints and identification

The generalised Euler equation of a sophisticated agent [hl] is $u'(c_t)\ge R\delta\,\mathbb E_t\left[(\beta_0\,\mathrm{MPC}_{t+1}+1-\mathrm{MPC}_{t+1})u'(c_{t+1})\right]$, with equality off the constraint. At $\mathrm{MPC}_{t+1}=1$, as at a binding constraint next period, the effective factor is $\beta_0\delta$.
Proposition: In stationary data on consumption, assets and MPCs, $(\beta_0,\delta)$ enter the Euler equation through $\delta[1-(1-\beta_0)\mathrm{MPC}_{t+1}]$. They are therefore separately identified iff next-period MPCs vary across observations whose Euler residuals are informative, i.e. unconstrained today. Mass at the constraint alone identifies only the combination, given $\underline a$.
Proof: The Euler moment is linear in $\delta$ and in $\delta(1-\beta_0)$, with regressors $\mathbb E_t[u'(c_{t+1})]$ and $\mathbb E_t[\mathrm{MPC}_{t+1}u'(c_{t+1})]$. These need to be linearly independent. ∎
So micro MPC heterogeneity, for example from lottery or rebate experiments, combined with consumption Euler moments, identifies present bias. The equilibrium comparison $\Delta A$ then follows if the hypothesis of Proposition Proposition holds [moll].

Completion Estimate

40%

This is a post hoc, heuristic estimate for the full catalogue target. A sophisticated three-period consumption model proves saving monotonicity, and stationary capital comparisons are derived conditional on an ascending asset-supply correspondence. That central correspondence assumption, stationary equilibrium selection and identification under binding stochastic borrowing limits remain unproved for the catalogue economy.

References

Go to contribution →

Econ-379-A1 · Exploration · LLM-generated · gpt 6 astra ultra ·

GPT 6 Astra Ultra attempt on Econ-379

Present bias and aggregate saving under borrowing constraints: a research attempt

Target and outcome

Status: unresolved. The catalogue seeks the stationary aggregate-asset change from removing present bias while preserving other type characteristics and recomputing equilibrium. This attempt solves a two-period overlapping-generations economy with a nonnegative saving constraint and heterogeneous present bias. It derives a unique positive stationary capital stock and an exact comparative static. This is a restricted benchmark with no competing future saving selves, not a solution for infinitely lived sophisticated households with income risk and commitment choices.

Moll's lecture [moll], whose title page dates the course to Spring 2018 despite its 2019 archive URL, explicitly motivates present bias together with borrowing constraints and portfolio choice. The derivation below is additional and does not attribute a stationary-capital theorem to the slides.

The constrained household problem

A young household of type $(\beta,\delta,e)$ has wage income $we$, where $e>0$ is labor efficiency. It lives for two periods and chooses saving $s\geq0$. Gross return next period is $R>0$. There is no old-age endowment initially. Its young-self objective is \[ \log(we-s)+\beta\delta\log(Rs),\qquad 0<s<we, \] where $0<\beta\leq1$ and $0<\delta<1$. The old self consumes all accumulated resources; there is no later investment decision. Therefore sophisticated and naive beliefs coincide in this specific two-decision life cycle. The borrowing constraint and positive old-age consumption require nonnegative saving, and the logarithmic objective places the optimum strictly inside the stated interval.

The first-order condition and strict concavity give the unique policy \[ s=q(\beta,\delta)we,\qquad q(\beta,\delta)=\frac{\beta\delta}{1+\beta\delta}. \tag{1} \] Saving is independent of $R$ because income and substitution effects cancel under logarithmic utility with zero retirement endowment. This cancellation is a special property, not a general statement about present bias.

To expose the constraint, add a promised old-age pension $z\geq0$ with fixed financing treated elsewhere in the budget. The same calculation gives \[ s=\max\left\{0,\frac{\beta\delta we-z/R}{1+\beta\delta}\right\}. \tag{2} \] At the boundary, optimality follows from the one-sided derivative. For fixed $(w,R,z,e,\delta)$, saving is weakly increasing in $\beta$, including the zero-saving region. However, a pension changes public liabilities and their financing; it cannot be inserted into the capital equilibrium below without adding those resources and claims. Equation (2) is a household comparative static, not the announced equilibrium theorem.

A closed capital-market equilibrium

Now return to $z=0$. Each generation has mass one and an independently recurring joint type law $F$ with ${\mathbb{E}}_F e=1$. There is one unit of aggregate efficiency labor. Firms have production \[ Y=K^\alpha L^{1-\alpha},\qquad0<\alpha<1, \] full capital depreciation, and competitive factor prices. Thus \[ w(K)=(1-\alpha)K^\alpha,\qquad R(K)=\alpha K^{\alpha-1}. \] Household saving buys the productive capital used next period; there is no government debt, housing or foreign asset. Hence aggregate saving is capital by explicit market clearing. Set \[ Q(F)={\mathbb{E}}_F[e\,q(\beta,\delta)]. \] Finite ${\mathbb{E}} e=1$ and $0<q<1$ imply $0<Q<1$. Summing (1) gives the transition equation \[ K_{t+1}=Q(F)(1-\alpha)K_t^\alpha. \tag{3} \] A positive stationary equilibrium must therefore satisfy \[ K^*(F)=\{(1-\alpha)Q(F)\}^{1/(1-\alpha)}. \tag{4} \] It is the unique positive solution. The degenerate zero-capital point is excluded because the maintained equilibrium has positive consumption and finite positive factor prices. Starting from any $K_0>0$, taking logarithms of (3) yields a linear recursion with coefficient $\alpha\in(0,1)$, so $K_t\to K^*$.

The stationary cross-section consists of each period's young households with zero beginning-of-period assets and old households carrying their previous saving claims. With cohort mass one, the integral of beginning assets over this two-cohort measure is $K^*$. If one normalizes total population to one instead, mean assets are $K^*/2$; comparisons must retain the same convention.

Removing present bias while retaining correlations

The intervention changes only $\beta$ to one for each type, keeping that household's $(\delta,e)$ and their joint distribution. Consequently \[ Q_1={\mathbb{E}}_F\left[e\frac{\delta}{1+\delta}\right],\qquad Q_0={\mathbb{E}}_F\left[e\frac{\beta\delta}{1+\beta\delta}\right]. \] Because the fraction is increasing in $\beta$, $Q_1\geq Q_0$, strictly if $\Pr_F(\beta<1)>0$. Recomputing wages and returns through (4) gives the exact aggregate change \[ \Delta A=\{(1-\alpha)Q_1\}^{1/(1-\alpha)} -\{(1-\alpha)Q_0\}^{1/(1-\alpha)}\geq0. \tag{5} \] This is an equilibrium result inside the benchmark, rather than aggregation at fixed wages. Ignoring correlation between earnings efficiency and present bias would replace ${\mathbb{E}}(e q)$ by ${\mathbb{E}} e{\mathbb{E}} q$ and generally change its magnitude.

For homogeneous synthetic types with $\alpha=1/3$, $\delta=.9$ and $\beta=.5$, $Q_0=.45/1.45\approx.3103$, while removing present bias gives $Q_1=.9/1.9\approx.4737$. The stationary capital ratio is \[ K_1^*/K_0^*=(Q_1/Q_0)^{3/2}\approx1.886. \] This is a model illustration, not an estimate of actual missing capital. If the change occurs at date zero, the capital stock cannot jump immediately: (3) gives the transition from the common initial $K_0$, distinguishing the stationary comparison from its adjustment path.

Why the infinite-life problem remains

An infinitely lived sophisticated agent selects consumption anticipating the policies of its future selves. Its continuation evaluation is not the value function of an ordinary discounted control problem with discount factor $\beta\delta$; $\beta$ applies once to the entire future stream. Borrowing limits, commitment accounts, income transitions and equilibrium prices can alter future-self responses. The simple proof of (5) exploits a life cycle with no later saving choice and cannot be transplanted to that setting.

The benchmark also lacks endogenous labor, uninsurable risk, bequests and asset heterogeneity. With pensions or government debt, the asset integral no longer equals productive capital without additional clearing equations. Establishing and selecting stationary equilibria in the full future-self game, and identifying the joint type law from behavior, remain unresolved. Equation (5) is therefore a rigorous special-case sign result, not a universal theorem that removing present bias always raises aggregate wealth.

Completion Estimate

42%

This is a post hoc, heuristic estimate for the full catalogue target. The two-period heterogeneous overlapping-generations benchmark has unique positive stationary capital and an exact bias-removal comparison preserving type correlations. Sophisticated infinite-life self-equilibria, stochastic borrowing constraints, commitment choices and empirical type identification remain outside that special case.

References

Go to contribution →