we-solve (beta)

Econ-489 · economics · Economics

Measuring climate adaptation

LLM status Open unverified

Statement

Original identifier: C3 (atlas item 113). Field: Environmental and climate economics. Classification: editorial research family with established restricted solutions; current open status not certified.

Definitions and assumptions

Let $F$ specify a joint distribution of weather histories over a declared horizon and population. For household or firm $i$, adaptation $a_i\in A_i$ changes outcome $Y_i(w,a_i)$ under weather history $w$ and consumes real resources $k_i(a_i,w)$ measured in a fixed currency year. A welfare index $v_i(Y_i)$ is monetized using a stated marginal consumption value, or is replaced by a clearly defined monetary benefit. Define net welfare \[ W(F,a)=\sum_i\omega_i\mathbb E_F[v_i(Y_i(w,a_i))-k_i(a_i,w)], \] with supplied weights $\omega_i\ge0$. An adaptation regime $a^*(F)$ maximizes $W$ over a stated feasible set, allowing spillovers only if included in $Y_i$ and $k_i$.

Formal research question

Identify or sharply bound the welfare value of adaptation and the residual climate loss under explicit selection and transport restrictions. For baseline climate $F_0$, changed climate $F_1$ and feasible baseline adaptation $a^0$, define \[ B_A=W(F_1,a^*(F_1))-W(F_1,a^0),\qquad L=W(F_0,a^*(F_0))-W(F_1,a^*(F_1)). \] Characterize which observational weather panels, adoption interventions and validation data identify these quantities when adaptation and location choices are endogenous. Construct inference over a specified class permitting costly defensive spending, heterogeneous exposure and bounded unobserved selection, and separate the sampling error from ambiguity in long-run climate responses and adaptation costs.

Interpretation and scope

Reduced temperature sensitivity of mortality or yields can coexist with large resource costs, changed composition or displaced damages. Thus a change in an outcome coefficient is neither the welfare benefit $B_A$ nor proof of costless adaptation. If $a^0$ is feasible and an optimum exists, $B_A\ge0$ by definition; observed adoption need not be optimal. Weather fluctuations identify responses within sampled weather and institutional regimes, while a shift in $F$ can change technologies, prices and expectations. Migration changes the population at risk; holding original residents fixed and evaluating future residents are different targets. Transfers, medical expenditures and physical resource losses must be handled consistently to avoid double counting.

Source audit and status

[S1] examines weather, mortality and residential energy use; [S2] studies historical declines in temperature mortality, including adaptation; [S3] estimates agricultural adaptation using US climate trends. These verified empirical sources already estimate selected adaptation channels. They do not identify every net-welfare component or universal future response. The selection-aware full-welfare target is an editorial specialization, and the bibliography does not certify its precise 2026 open status.

References

    },leftmargin=*]

  1. Olivier Deschênes and Michael Greenstone (2011). Climate Change, Mortality, and Adaptation: Evidence from Annual Fluctuations in Weather in the US. American Economic Journal: Applied Economics 3(4), 152–185. https://doi.org/10.1257/app.3.4.152. Locator: Publisher or institutional abstract and bibliographic record. Support and limits: Weather mortality and residential energy responses; weather variation does not alone identify long-run adaptive welfare.

  2. Alan Barreca, Karen Clay, Olivier Deschenes, Michael Greenstone and Joseph S. Shapiro (2016). Adapting to Climate Change: The Remarkable Decline in the US Temperature-Mortality Relationship over the Twentieth Century. Journal of Political Economy 124(1), 105–159. https://doi.org/10.1086/684582. Locator: Publisher or institutional abstract and bibliographic record. Support and limits: Historical temperature-mortality adaptation, including air conditioning; not all adaptation channels or future locations.

  3. Marshall Burke and Kyle Emerick (2016). Adaptation to Climate Change: Evidence from US Agriculture. American Economic Journal: Economic Policy 8(3), 106–140. https://doi.org/10.1257/pol.20130025. Locator: Publisher or institutional abstract and bibliographic record. Support and limits: US agriculture estimates comparing longer-run climate trends; transport requires extra assumptions.

Common conventions: Source mathematical conventions

Unless an entry states otherwise, agent and firm sets are finite. State and action spaces are measurable, strategies and outcome maps are measurable, probability laws are specified, and expectations exist for the targets under discussion. Infinite-horizon discount factors lie in $(0,1)$ and discounted objectives are finite. Norms, tolerances and complexity input sizes must be specified for computational comparisons. Constants or primitives that appear locally are inputs, not empirical facts asserted by this catalogue.

Identification. A statistical model $m\in\mathcal M$ implies an observable law $P_m$ and target $\theta(m)$. At law $P$, its identified set is \[ \Theta(P;\mathcal M)=\{\theta(m):m\in\mathcal M,\ P_m=P\}. \] Point identification holds when this set is a singleton, or equivalently when $P_{m_1}=P_{m_2}$ implies $\theta(m_1)=\theta(m_2)$ for all compatible models. Sharp bounds mean bounds attained, or approached when the boundary is not attained, by compatible models. Writing this set is a definition, not a solution: the substantive task is to establish informative restrictions and derive or estimate the set. An unrestricted-confounding model may give only trivial bounds.

Inference. Identification, estimability and valid uncertainty quantification are separate questions. Fix $\alpha\in(0,1)$. A confidence procedure $C_n$ over a declared law class $\mathcal P$ has asymptotic uniform coverage at level $1-\alpha$ when \[ \liminf_{n\to\infty}\inf_{P\in\mathcal P}\Pr_P\{\theta(P)\in C_n\}\geq1-\alpha. \] For set-identified targets the coverage event must be defined separately, for example coverage of the full identified set. If an entry uses identification-indexed models instead of uniquely indexed laws, the same coverage convention is applied over those models. Pointwise limits do not establish uniform validity, and asymptotic validity does not guarantee finite-sample precision. Sample sizes, dependence structures and weak-identification sequences are part of each application.

Counterfactuals. $Y_i(a)$ is a potential outcome under a specified intervention, including its timing, implementation and versions. It is not $Y_i$ conditional on voluntarily choosing $a$. A full intervention vector permits interference; shorthand scalar treatment notation does not silently impose no interference. Assignment, selection, attrition, anticipation and measurement must be specified. A claim about an instrument's local effect is not automatically a population policy effect.

Economic equilibrium. A counterfactual model includes best responses, constraints, feasibility, market clearing, state transitions and expectations consistent with the stated information. When several equilibria exist, either a declared selector is an input or the target is set-valued. Comparisons across policy regimes must use the stated selection convention. Reduced-form relationships are not presumed invariant after an equilibrium-changing intervention.

Optimization and welfare. Optimization is over the stated feasible set. If attainment is not established, $\sup$ or $\inf$ and $\varepsilon$-optimal choices replace an asserted maximizer or minimizer. For example, with value $V=\sup_{a\in\mathcal A}W(a)<\infty$, an $\varepsilon$-optimal set is $\{a\in\mathcal A:W(a)\geq V-\varepsilon\}$ for $\varepsilon>0$. Benefits, costs, risk and health or environmental outcomes can be aggregated only using declared units and conversion weights. Interpersonal utility weights, the treatment of fiscal transfers and foreign profits, discounting, and the behavioral welfare criterion are explicit inputs. A comparative-static derivative additionally requires existence and appropriate differentiability of the selected counterfactual.

Sources. Local labels [S1], [S2], and so forth restart in every question. Locators identify the relevant abstract, model, section or result at the level actually checked; a bibliographic or abstract-level verification is not represented as inspection of an entire proof. Working papers are distinguished from later publications and changing versions. Source summaries are paraphrased. All URL checks and editorial formulations refer to the audit date stated above.

Statement licence: CC-BY-4.0 · Source file

Contributions 1

Econ-489-A1 · Exploration · LLM-generated · gpt 6 astra ultra ·

GPT 6 Astra Ultra attempt on Econ-489

Adaptation value with selection and climate transport: sharp joint welfare regions for a finite policy menu

Problem ID: Q54-485. Status: unresolved; restricted research attempt.

Weather responses are not net adaptation values

The target is the welfare gain from optimal adaptation under a changed climate and the residual loss relative to optimized baseline climate. We give an exact identified-set construction that preserves the same unknown response technology across climates and the same costs across candidate policies. It allows ambiguous optimal policies and avoids subtracting unrelated endpoint estimates. It solves a finite-menu, finite-weather specialization; it does not identify unobserved long-run technological change or endogenous migration from a weather panel alone.

Deschênes and Greenstone [DG] distinguish weather-driven mortality and residential energy responses; Burke and Emerick [BE] compare longer-run agricultural adjustments. Their publisher abstracts and bibliographic records were inspected. They motivate separating outcome responses, defensive expenditure and transport assumptions, without supplying the payoffs of our optimization problem.

Fix a baseline resident population and social weights. A finite set $\mathcal A$ of feasible joint adaptation plans includes baseline $a^0$. A plan specifies location, timing and any contingent actions using only information then available. Joint plans allow spillovers within the population; if migration is allowed, origin residents retain their fixed weights and destination costs enter the plan. Weather histories fall in $K$ categories. Climate $z\in\{0,1\}$ has known probabilities $p_{zk}\geq0$, $\sum_kp_{zk}=1$.

For category $k$ and plan $a$ let $t_{ka}$ be expected total monetary net welfare, conditional on that category under a specified within-category law. It includes monetized outcomes less construction, energy, maintenance and other real resources exactly once. Transfer payments cancel only within an equal-weight payer-recipient boundary. Health valuation, interpersonal weights and the numeraire are fixed primitives. Define \[ W_z(a;t)=\sum_{k=1}^Kp_{zk}t_{ka},\qquad V_z(t)=\max_{a\in\mathcal A}W_z(a;t). \] The targets are $B(t)=V_1(t)-W_1(a^0;t)$ and $L(t)=V_0(t)-V_1(t)$. The transport restriction is that the same $t$ applies to both climates. It excludes unmodeled climate-induced prices, technology and within-category composition changes.

An explicit selection and measurement model

Let $\mathcal T\subset{\mathbb{R}}^{K|\mathcal A|}$ be a nonempty compact polytope of response matrices compatible with the data and maintained restrictions. For example, an observed monetary net-outcome cell mean $m_{ka}$ with externally justified selection error $\sigma_{ka}$ and cost-measurement error $\nu_{ka}$ yields the restriction \[ |t_{ka}-m_{ka}|\leq\sigma_{ka}+\nu_{ka}. \tag{1} \] These are restrictions on target-population means, not automatic consequences of conditional observed averages. Randomized implementation and weather exogeneity may justify zero selection error on supported cells; unsupported cells require external bounds. Shape restrictions or accounting identities may link several cells. We assume exact compatibility: every matrix in $\mathcal T$ is attained by an admitted data-generating model. Without this assumption the construction below is an outer bound, not a sharp region. This explicit assumption is often credible for a freely varying bounded-mean response class, but not for an unmodeled structural technology with shared latent parameters.
Theorem (Sharp joint adaptation-benefit and residual-loss set): For each ordered pair $(r,s)\in\mathcal A^2$, form the polytope \[ \mathcal T_{rs}=\{t\in\mathcal T: W_0(r;t)\geq W_0(a;t),\quad W_1(s;t)\geq W_1(a;t) \text{ for all }a\in\mathcal A\}. \] Discard empty pieces. The sharp joint identified set is the finite union \[ \mathcal I=\bigcup_{r,s} \left\{\big(W_1(s;t)-W_1(a^0;t),\, W_0(r;t)-W_1(s;t)\big):t\in\mathcal T_{rs}\right\}. \tag{2} \] It is compact, but need not be convex. Its extrema in any linear direction are attained and are found by at most $|\mathcal A|^2$ linear programs. Every retained benefit coordinate is nonnegative. Ties in optimal plans require no equilibrium selector for these value targets.
Proof: For any compatible $t$, the finite menu has baseline and changed-climate maximizers $r,s$. Thus $t\in\mathcal T_{rs}$ and its target equals the displayed linear map. Conversely, a matrix in any nonempty piece is compatible by assumption and has $r,s$ as maximizers, so that same map is its actual target. This proves equality and attainability, not merely coordinate bounds. Each piece is a closed subset of compact $\mathcal T$ and is described by linear inequalities. Its image is a compact polytope; a finite union is compact. A linear functional on an image is a linear functional of $t$, so optimization over the union consists of optimizing over each nonempty piece. The inequality for $a=a^0$ gives nonnegative $B$. All maximizing plans deliver the same $V_z$, so ties leave the value target unchanged. The following example verifies the possible failure of convexity within the stated adaptation class. ∎

For an explicit nonconvex instance, use two weather categories, $p_0=(1,0)$, $p_1=(0,1)$, and two plans $\{a^0,a^1\}$. Both payoffs in category 1 are zero; category 2 payoffs are $(u,-u)$, with common unknown $u\in[-1,1]$. These matrices form a compact polytope. Then $V_0=0$, $V_1=|u|$, and \[ (B,L)=(|u|-u,-|u|). \] The points $(0,-1)$ and $(2,-1)$ occur at $u=1$ and $u=-1$. Their midpoint $(1,-1)$ is impossible: $L=-1$ requires $|u|=1$, which gives only those two endpoints. Thus (2) can indeed be nonconvex.

Why retain both optimizers? Independent marginal intervals for $V_0$, $V_1$ and $W_1(a^0)$ usually use incompatible technologies. In particular, if $p_0=p_1$, every compatible model has $L=0$. Subtracting independent intervals for $V_0$ and $V_1$ would generally produce spurious nonzero residual losses. Formula (2) preserves the common response matrix exactly.

A robust plan when the optimum is not identified

For a fixed plan $b$ define its worst-case regret \[ R(b)=\max_{t\in\mathcal T}\{V_1(t)-W_1(b;t)\}. \]
Proposition (Exact finite-menu regret): A minimum-regret plan exists, and \[ R(b)=\max_{a\in\mathcal A}\max_{t\in\mathcal T} [W_1(a;t)-W_1(b;t)]. \tag{3} \] Thus all regrets and a minimizing plan can be computed by at most $|\mathcal A|^2$ linear programs. This is a declared ambiguity criterion, not an implication of expected-utility maximization under an unidentified probability law.
Proof: For each $t$ the regret is a finite maximum over $a$. Maximization over $t$ and a finite set $a$ commutes: each side is the maximum over the same Cartesian product. Compactness attains every inner linear maximum. The outer finite maximum and minimum over $b$ are also attained. ∎
Proposition (Stability and honest propagation of a confidence set): Suppose a second payoff system $\widetilde W$ satisfies $\max_a|W_z(a)-\widetilde W_z(a)|\leq\epsilon_z$ for $z=0,1$. Then \[ |B-\widetilde B|\leq2\epsilon_1,\qquad |L-\widetilde L|\leq\epsilon_0+\epsilon_1. \tag{4} \] If a random response region $\widehat{\mathcal T}$ contains the true matrix with probability at least $1-\alpha$, the image construction (2) using this region contains the true pair with probability at least $1-\alpha$, without a unique-optimum assumption.
Proof: For every plan $a$, $W_z(a)\leq\widetilde W_z(a)+\epsilon_z$; taking maxima gives $V_z\leq\widetilde V_z+\epsilon_z$. Reverse the two systems for the opposite inequality. Apply the triangle inequality to the two definitions of $B,L$ to obtain (4). On the event that the true matrix lies in the confidence region, select any of its finite-menu maximizers; it belongs to the corresponding optimizer piece and its target belongs to the image. Thus the coverage event for the matrix implies the coverage event for the target, proving the probability statement. ∎

For instance, if $|t_{ka}|\leq H$ and only the climate distribution changes by total variation at most $\delta_z$, the error in each welfare value is at most $2H\delta_z$. This follows by summing $|\sum_k(p_{zk}-\widetilde p_{zk})t_{ka}| \leq H\sum_k|p_{zk}-\widetilde p_{zk}|$. The factor becomes the range width when a tighter common range is known. Statistical error in estimated cells and structural climate ambiguity should be reported separately even when these bounds combine them.

A sharp obstruction from missing defensive costs

Consider a single person, weather $w\in\{0,1\}$, damage $dw(1-a)$, $a\in\{0,1\}$, and adaptation cost $ka$, with $d,k>0$. Net payoff is $-dw(1-a)-ka$, so at climate probability $p$, \[ V(p)=-\min\{dp,k\}. \] At $p_0=0$, $p_1=1$, and $a^0=0$, for $0<k<d$, \[ B=d-k,\qquad L=k. \tag{5} \] Both claims follow by comparing the two available payoffs $-dp$ and $-k$. After optimal adaptation the outcome's weather-damage slope is zero for every such $k$, while the adaptation benefit and residual loss vary in opposite directions throughout (5). With $d=10$, costs 1 and 9 yield identical zero post-adaptation weather damage but benefits 9 and 1 and residual losses 1 and 9. If costs are unobserved and adoption incentives admit either value, these two economies are observationally equivalent for the stated weather and adoption outcomes. A declining mortality or yield sensitivity therefore cannot by itself identify net welfare.

Unsupported weather is a second obstruction. If $p_{1k}>0$ for a category never observed and its payoff under a feasible adaptation plan has no upper bound, $V_1$ and potentially $B$ have no finite upper bound. Hold all observed cells fixed and send just that payoff to infinity; its positive probability makes that plan's welfare diverge. This proves why bounded extrapolation or transport evidence is necessary, rather than merely convenient.

What remains unresolved

The finite-union construction is sharp only for the supplied compatible response class. It cannot certify that observational selection tolerances are valid, that long-run climate probabilities preserve within-category response laws, or that adaptation menus include feasible future migration and infrastructure. A next step is to estimate jointly validated benefit and real-cost cells under adoption interventions, then expand the menu while preserving budget and information constraints. Endogenous prices or migration composition require additional state variables and structural restrictions. No empirical value of the canonical adaptation benefit or residual loss is claimed here.

Completion Estimate

55%

This is a post hoc, heuristic estimate for the full catalogue target. The attempt constructs sharp joint adaptation-benefit and residual-loss regions for finite weather and policy menus, with regret and confidence propagation. Full welfare measurement still requires validated selection and transport restrictions, defensive-cost data, feasible future migration, and climate-induced price and technological responses.

References

  • [DG] O. Deschênes and M. Greenstone (2011), Climate Change, Mortality, and Adaptation: Evidence from Annual Fluctuations in Weather in the US, American Economic Journal: Applied Economics 3(4), 152--185. Publisher abstract inspected. https://doi.org/10.1257/app.3.4.152.
  • [BE] M. Burke and K. Emerick (2016), Adaptation to Climate Change: Evidence from US Agriculture, American Economic Journal: Economic Policy 8(3), 106--140. Publisher abstract inspected. https://doi.org/10.1257/pol.20130025.

Go to contribution →