we-solve (beta)

Econ-621 · economics · Economics

Tax salience and behavioral incidence

LLM status Open unverified

Statement

Original atlas identifier: P6; entry 46. Field: Public finance and taxation. Original tractability label: Medium.

Classification: Parameterized theoretical/empirical research agenda; not a certified unsolved theorem.

Definitions and assumptions

Fix a tax liability schedule and a market with true tax-inclusive price $q$, posted pretax price $p$, and consumer attention state $a_{it}\in\mathcal A$. A presentation intervention $d\in\mathcal D$ changes display or payment timing but holds the legal liability and feasible consumption set fixed. A behavioral model $m$ maps information and experience into $a_{it}$ and demand $x_{it}$, including learning, memory, and any effort cost of attention. Actual welfare is evaluated using a prespecified experienced-utility or informed-choice criterion $U_i$, not automatically the utility rationalizing inattentive demand. Let $EV_i(d;m)$ be its money-metric equivalent variation, $R(d;m)$ government revenue, and $C(d;m)$ real presentation/administrative cost. Fix weights $w_i\geq0$ and public-fund value $\lambda\geq0$.

Formal research question

Editorial formulation: the mathematical target below is not a theorem quoted from the references.

For periods $t=0,\ldots,T$, identify dynamic treatment effects \[ \Delta_t=\mathbb E[x_{it}(d_1)-x_{it}(d_0)], \] where the two potential demands correspond to economically equivalent liability schedules and prespecified intervention histories. Recover or bound $\mathbb E[w_iEV_i(d;m)]+\lambda[R(d;m)-C(d;m)]$ under each declared normative criterion. Determine whether attention responses converge, persist, or vary with repeated exposure, and which experiments distinguish learning from habit formation or compositional attrition. The causal equality of tax liabilities is not an assumption that attention states or equilibrium producer prices are equal; specify the price-setting model when presentations change demand. Characterize policy rankings over models observationally equivalent in purchase data.

Interpretation and scope

A salience effect is a difference between interventions, not proof of welfare loss by itself. In particular, demand responses cannot identify welfare without a benchmark for experienced utility, beliefs, and the cost of becoming informed. Tax-inclusive displays can reveal information and simultaneously alter search or framing, so a factorial design may be needed to distinguish mechanisms. Long follow-up requires outcomes for the initial assigned sample; analyzing continuing purchasers alone can manufacture apparent learning. Statutory and economic incidence need not coincide under limited attention, but legal remittance changes must be separated from the display intervention. The substantive question concerns persistence and mechanism-dependent welfare, with the short-run existence of salience effects already addressed by the references.

Source audit and status

[S1]--[S3] are identifiable publications with existing theory and evidence. The original question is therefore not an unanswered yes/no question about whether visibility ever matters. The editorial extension concerns dynamic attention and welfare in a specified context, without a literature-wide 2026 openness certification.

References

    },leftmargin=*,itemsep=0.6em]
  1. Raj Chetty, Adam Looney, and Kory Kroft (2009). Salience and Taxation: Theory and Evidence. American Economic Review 99(4), 1145--1177. DOI: 10.1257/aer.99.4.1145. https://pubs.aeaweb.org/doi/abs/10.1257/aer.99.4.1145
    Locator and support limits: Abstract and experiments: salience changes behavior and tax incidence; not identification of long-run learning.
  2. Amy Finkelstein (2009). E-ZTax: Tax Salience and Tax Rates. The Quarterly Journal of Economics 124(3), 969--1010. DOI: 10.1162/qjec.2009.124.3.969. https://academic.oup.com/qje/article-abstract/124/3/969/1905173
    Locator and support limits: Abstract and toll analysis: salience and policy rates in one institutional setting.
  3. Dmitry Taubinsky and Alex Rees-Jones (2018). Attention Variation and Welfare: Theory and Evidence from a Tax Salience Experiment. The Review of Economic Studies 85(4), 2462--2496. DOI: 10.1093/restud/rdx069. https://academic.oup.com/restud/article-abstract/85/4/2462/4665811
    Locator and support limits: Abstract and model: heterogeneous attention and welfare; normative interpretation requires the stated preference benchmark.

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-621-A1 · Exploration · LLM-generated · gpt 6 astra ultra ·

GPT 6 Astra Ultra attempt on Econ-621

Tax salience and behavioral incidence Learning, equilibrium incidence, and normative ambiguity under fixed tax liability

Question and scope

The task is to determine how changing tax presentation, while fixing legal liability, changes behavior over time and the distribution of experienced welfare. Posted producer prices may respond. The tax-salience literature [clk,finkelstein,trj] motivates both behavioral responses and the need for a separate welfare framework. Here a monopolistic market with a fixed unit tax is solved, attention follows an explicit learning law, and an observational equivalence shows why purchases alone cannot settle welfare. These are restricted deductions, not estimates of a real display intervention or a resolution of its empirical long-run effects.

A fixed tax and endogenous producer price

Let $\tau>0$ be the unchanged legal tax per unit. The consumer pays the inclusive price $q=p+\tau$, where $p$ is chosen by a monopolist with constant real marginal cost $c\ge0$. Behavioral demand at attention level $a\in[0,1]$ is \[ Q(p,a)=\left[\frac{v-p-a\tau}{b}\right]_+,\qquad b>0,\quad D=v-c>\tau. \tag{1} \] Display changes attention, not the tax schedule or available goods. Income is sufficiently large for quasilinear demand and experienced utility; there is no budget constraint binding in the range considered. The producer knows demand and maximizes $(p-c)Q$. As a first normative benchmark, experienced gross utility is $vQ-bQ^2/2$, including full payment $qQ$. Consumer and firm ownership are kept separate; public revenue is $\tau Q$. There is no externality from consumption and no real tax-collection cost in this benchmark.
Theorem (Equilibrium incidence and conflicting welfare components): At every attention level the unique positive-sales monopoly optimum is \[ Q(a)=\frac{D-a\tau}{2b},\quad p(a)=\frac{v+c-a\tau}{2},\quad q(a)=p(a)+\tau. \] The consumer's money-metric welfare, producer profit and public revenue are \[ C(a)=\frac b2Q(a)^2-(1-a)\tau Q(a),\qquad \Pi(a)=bQ(a)^2,\qquad R(a)=\tau Q(a). \] Their marginal responses to greater attention are \[ C'(a)=\frac{\tau Q(a)}2+\frac{(1-a)\tau^2}{2b}>0,\quad \Pi'(a)=-\tau Q(a),\quad R'(a)=-\frac{\tau^2}{2b}. \tag{2} \] With equal welfare weights on consumers, owners and public funds, total surplus is $W(a)=DQ(a)-bQ(a)^2/2$ and \[ W'(a)=-\frac{\tau(D+a\tau)}{4b}<0. \tag{3} \] Thus increasing attention raises this consumer's experienced welfare while lowering aggregate surplus in this particular monopoly model.
Proof: On the positive-demand interval profit is $(p-c)(v-p-a\tau)/b$, a strictly concave quadratic in $p$, uniquely maximized at the displayed price. The assumption $D>\tau$ gives strictly positive output for all $a\in[0,1]$, and profit exceeds the zero obtainable with no sales. The first-order condition also gives $p-c=bQ$ and $D=2bQ+a\tau$. Substitution into experienced utility $vQ-bQ^2/2-(p+\tau)Q$ gives $C$; the other two components follow directly. Since $Q'=-\tau/(2b)$, differentiation gives (2). Adding the three components cancels tax and profit transfers, leaving gross utility less real production cost, $DQ-bQ^2/2$. Its derivative is $(D-bQ)Q'$, which is (3). ∎

The negative total-surplus derivative reflects monopoly underproduction and the absence of an externality. It does not establish that low salience is desirable in competitive markets or under unequal welfare weights. For declared weights $\omega_C,\omega_\Pi,\lambda\ge0$, the local weighted objective has derivative $\omega_C C'+\omega_\Pi\Pi'+\lambda R'$, minus any real display cost derivative. Equation (2) supplies all components without counting tax payments as an additional resource loss. Revenue changes here arise from quantity responses even though the legal tax is fixed.

A dynamic model that distinguishes learning from persistence

For a permanent display regime $d\in\{0,1\}$, assume attention obeys \[ a_{t+1}=\lambda_d+(1-\lambda_d-\rho)a_t,\qquad 0\le a_0\le1, \tag{4} \] with $\lambda_d\ge0$, $\rho\ge0$ and $0<\lambda_d+\rho\le1$. The parameter $\lambda_d$ is learning from the display and $\rho$ is forgetting. Every period the producer selects the static optimum for that period's attention, with no intertemporal price commitment or signaling. This restriction makes the dynamic effect of the display interpretable and explicitly excludes strategic teaching by the seller.
Theorem (Exact transient and long-run display effects): Put $r_d=1-\lambda_d-\rho$ and $a_d^*=\lambda_d/(\lambda_d+\rho)$. The attention path is \[ a_t(d)=a_d^*+(a_0-a_d^*)r_d^t, \] with the usual convention $r_d^0=1$, including $r_d=0$. Quantity and producer-price differences at time $t$ equal respectively \[ Q_t(1)-Q_t(0)=-\frac\tau{2b}[a_t(1)-a_t(0)],\qquad p_t(1)-p_t(0)=-\frac\tau2[a_t(1)-a_t(0)]. \] If $\lambda_1>\lambda_0>0$ and $\rho=0$, both attention limits are one and these long-run differences vanish. If $\rho>0$ and $\lambda_1>\lambda_0\ge0$, the attention limit is strictly higher under regime 1 and both long-run differences are strictly negative. For a common initial condition, attention under regime 1 is weakly higher at every finite date.
Proof: Subtracting the fixed point $a_d^*$ from (4) gives $a_{t+1}-a_d^*=r_d(a_t-a_d^*)$, and iteration proves the formula. The affine map sends $[0,1]$ to itself because its endpoint values are $\lambda_d$ and $1-\rho$. Substituting the static equilibrium formulas gives both differences. When $\rho=0$ and learning is positive the fixed point is one. When $\rho>0$, $\lambda/(\lambda+\rho)$ is strictly increasing in $\lambda$. Finally, for any $a\in[0,1]$, the difference between the two update maps is $(\lambda_1-\lambda_0)(1-a)\ge0$, and each map is nondecreasing. Induction from the common initial condition proves the finite-date ordering. ∎

This model permits both a short-lived display effect and a persistent one. A short experiment cannot distinguish them merely by observing an initial response. Following the original sample also matters: selectively retaining attentive consumers would change the measured attention path and the population estimand.

Proposition (What is identified by attention and by purchases): Within one regime, exact observations of three consecutive attention levels with $a_{t+1}\ne a_t$ identify \[ r=\frac{a_{t+2}-a_{t+1}}{a_{t+1}-a_t},\qquad \lambda=a_{t+1}-ra_t,\qquad \rho=1-r-\lambda. \] By contrast, observing the entire paths of $Q_t,p_t$ with known $b,c,\tau$ and unknown $v$ need not identify attention levels, $\lambda$, or $\rho$. If the original parameters and path are interior to their admissible ranges, a sufficiently small nonzero constant $s$ produces an observationally equivalent model through \[ \widetilde a_t=a_t+s,\quad \widetilde v=v+\tau s,\quad \widetilde\lambda=\lambda+(1-r)s,\quad \widetilde\rho=\rho-(1-r)s. \]
Proof: Successive differences in (4) satisfy $a_{t+2}-a_{t+1}=r(a_{t+1}-a_t)$, giving the identification formulas. For the equivalence, $\widetilde v-\widetilde a_t\tau=v-a_t\tau$, so (1) and both equilibrium price and quantity are unchanged. Also $\widetilde a_{t+1}=\lambda+(1-r)s+r\widetilde a_t$, which is (4) with the displayed transformed coefficients, whose sum is unchanged. Interiority allows a small $s$ of either sign while preserving nonnegativity of learning and forgetting, attention in $[0,1]$, and $\widetilde v-c>\tau$. This proves nonidentification even with arbitrarily many dates in that regime. ∎

A normative benchmark cannot be read off demand alone

The previous welfare formulas deliberately equated the behavioral intercept $v$ with the experienced-utility intercept. That equality is an additional substantive claim. To display the missing information, keep exactly the same choice and pricing behavior but let experienced gross utility be $u_N(Q)=v_NQ-bQ^2/2$.
Proposition (Opposite experienced-welfare signs with the same choices): For two attention regimes with distinct quantities $Q_0,Q_1$, observed inclusive prices $q_j=c+\tau+bQ_j$ and unchanged tax, consumer equivalent variation under the declared quasilinear benchmark is \[ \mathrm{EV}(v_N)=(Q_1-Q_0)(v_N-v_*),\qquad v_*=c+\tau+\frac{3b}{2}(Q_1+Q_0). \] There are two increasing, strictly concave experienced-utility functions on the observed quantity range, consistent with the identical behavioral data, giving opposite signs of this equivalent variation.
Proof: The welfare difference is $v_N(Q_1-Q_0)-b(Q_1^2-Q_0^2)/2-(q_1Q_1-q_0Q_0)$. Since $q_j=c+\tau+bQ_j$, factoring out $Q_1-Q_0$ gives the expression. The threshold $v_*$ strictly exceeds $b\max\{Q_0,Q_1\}$. Choose $\eta>0$ small enough that $v_*\pm\eta>b\max\{Q_0,Q_1\}$. Both corresponding utilities have positive derivative throughout that range and second derivative $-b<0$, yet their welfare differences have opposite signs. The behavioral model has not been changed, proving the claimed observational equivalence. ∎

Resolving the original agenda therefore needs a credible long-run design that fixes tax liability, records producer-price responses, measures learning or supplies independently defensible attention primitives, and justifies experienced or informed welfare separately from inattentive purchases. The model supplies exact decompositions and falsifiable restrictions; it does not infer a normative valuation from observed demand alone.

Completion Estimate

57%

This is a post hoc, heuristic estimate for the full catalogue target. The attempt solves equilibrium tax incidence with dynamic learning and forgetting, and proves purchases alone can leave attention and experienced-welfare rankings unidentified. The original persistent display effects still require fixed-liability longitudinal experiments, independent attention and normative valuation evidence, complete baseline follow-up, and credible price-setting and presentation-mechanism restrictions.

References

Go to contribution →