we-solve (beta)

Econ-472 · economics · Economics

Tax incidence anomalies

LLM status Open unverified

Statement

Original subject group: Taxation and social insurance

Audit classification: Background; proposed extension. The Annual Reviews record verifies title, author, year, pages, DOI and anomaly categories. Its accessible abstract is background for the exact directional-derivative problem. Corrected conflation of finite asymmetry, differentiability, and hysteresis.

Definitions and precise research target

Fix a market, common preannouncement state $z$, price aggregation weights, policy duration and a tax unit. Use a specific tax $\delta$ in currency per unit; for an ad valorem rate use separate normalization. Let $p_h(\delta;z)$ be the mean tax-inclusive unit price at horizon $h$ under an intervention changing the same baseline tax by $\delta$, with fixed information at announcement and declared equilibrium selection. If finite one-sided derivatives exist, set \[ \begin{aligned} \rho_+(h,z)&=\lim_{\delta\downarrow0}\frac{p_h(\delta;z)-p_h(0;z)}{\delta},\\ \rho_-(h,z)&=\lim_{\delta\uparrow0}\frac{p_h(\delta;z)-p_h(0;z)}{\delta},\\ A(h,z)&=\rho_+(h,z)-\rho_-(h,z). \end{aligned} \] If $p_h$ is differentiable at zero, $A(h,z)=0$ by definition. Unequal responses to two finite reforms therefore do not prove a derivative kink. For fixed $d>0$, define a distinct finite-change target \[ A_d(h,z)=\frac{p_h(d;z)-p_h(0;z)}d -\frac{p_h(-d;z)-p_h(0;z)}{-d}. \] Nonzero $A_d$ is possible under smooth nonlinear pricing. Permanent increases and temporary reductions have different intervention paths and cannot be substituted for these symmetric treatments. With observational law $P_O$, fix an identification class covering reform assignment, anticipation, latent state and equilibrium restrictions; characterize the joint identified set of $A$ and $A_d$. A mechanism claim must specify how its primitives produce asymmetry at a common initial state and give extra observable restrictions or interventions that distinguish it from alternatives. For example, nonsmooth price adjustment can produce a kink, while history-dependent states can produce persistent level differences without a common-state kink. This precise directional/finite-change distinction is editorial; the cited review documents tax-incidence anomalies.

Variants and assumption changes

  1. Infinitesimal versus finite reforms (editorial). Estimate $A$ near zero or $A_d$ at a specified size. The first detects nondifferentiability; the second also detects ordinary curvature.
  2. Common state versus hysteresis (editorial). Compare two fresh reforms from $z$, or a tax rise followed by reversal along a changing state path. Persistent prices after reversal are a path-dependence target.
  3. Statutory remitter (editorial). Hold the tax wedge fixed while changing whether buyers or sellers remit. A changed response then tests statutory-incidence invariance, a separate anomaly.

References and source audit

Benzarti (2025), Annual Review of Economics 17, 615--634, publisher abstract: metadata and anomaly categories verified. Full-text loading failed; the narrower exact mathematical question is not attributed to that abstract. The earlier NBER version is distinct.
  1. Youssef Benzarti. 2025. Tax Incidence Anomalies. Annual Review of Economics 17, 615–634; abstract. https://www.annualreviews.org/content/journals/10.1146/annurev-economics-081324-085805. DOI: https://doi.org/10.1146/annurev-economics-081324-085805.

Common conventions: Source mathematical conventions

These conventions apply to each entry unless explicitly replaced.

Models and identification. A primitive model $m$ specifies all probability laws, feasible choices, information, equilibrium and measurement rules used by its target. The admissible class $\mathcal M$ is fixed independently of outcomes. If $P_O(m)$ is its observable probability law and $\tau(m)$ the target, its identified set at observable law $P$ is \[ \mathcal I_\tau(P)=\{\tau(m):m\in\mathcal M,\ P_O(m)=P\}. \] Point identification means that this set is a singleton. An empty set signals incompatibility of data and assumptions. Coordinate bounds need not describe the joint set. Bounds are sharp only if every retained target is attainable or arbitrarily approachable by compatible models. Finite bounds require explicit restrictions on unobserved quantities; unrestricted confounding, hidden wealth, extrapolation or arbitrary equilibrium selection can yield uninformative sets.

Causal interventions. A potential outcome $Y_i(p)$ is the outcome under a complete policy regime $p$, including timing, financing, information and market spillovers. The target population and units are fixed before treatment. With interference, use the complete assignment vector or a justified exposure mapping; individual no-interference notation alone cannot identify an economy-wide intervention. A difference of mean potential outcomes does not require the cross-world joint distribution. Conditioning on post-treatment migration, survival, enrollment or employment changes the estimand and needs separate justification.

Statistical statements. An estimator, confidence set, forecast or test specifies a sampling law, model class and loss/coverage criterion. Uniform $(1-\alpha)$-coverage means $\inf_{m\in\mathcal M}\Pr_m\{\tau(m)\in C_n\}\geq1-\alpha$, or an explicitly asymptotic version. The whole parameter space trivially has coverage, so informativeness requires a stated width, risk or power objective. A forecasting improvement is relative to a named benchmark, information set and evaluation population.

Equilibrium and optimization. An equilibrium comprises feasible plans, optimality, beliefs where needed, budget constraints and all market/resource clearing. The primitives or a stated benchmark define each correspondence $\mathcal E(p;m)$. Nonemptiness is assumed or proved, never inferred just from compact policy sets. A selected-equilibrium target uses a declared selection rule; a robust target quantifies over all permitted equilibria. Use suprema or infima unless attainment is established. An $\varepsilon$-optimal policy is feasible and within $\varepsilon$ of the finite optimum value. Report an empty feasible set separately.

Welfare and accounting. Normative utility, interpersonal normalization, social weights, numeraire, discounting and the use of public receipts are primitives. Behavioral utility need not coincide with the welfare criterion. Household welfare includes ownership income and rents once; adding profits again to compensating variation generally double-counts them. A monetary equivalent is defined by an explicit transfer and price convention, with monotonicity and range conditions. Average effects, mechanism contributions, transfers and welfare losses are different targets. Infinite sums require convergence and dynamic feasible plans require terminal or no-Ponzi conditions.

Source versions. A working-paper date, revision date and journal publication year are distinguished. A DOI for a journal version is not automatically the DOI of an earlier manuscript. Locators refer to the stated version and printed pages where specified. A metadata-only check does not verify a claim about a full-text conclusion. Every entry's source subsection narrows the provenance claim accordingly.

Statement licence: CC-BY-4.0 · Source file

Contributions 1

Econ-472-A1 · Exploration · LLM-generated · gpt 6 astra ultra ·

GPT 6 Astra Ultra attempt on Econ-472

Tax asymmetry at a common state: identifying a kink separately from curvature

Problem ID: H22-468. Status: Unresolved; rigorous restricted results.

Target and contribution

The question concerns the difference $A$ between the two directional tax pass-through derivatives at the same initial state, and its finite-reform counterpart $A_d$. The review by Benzarti [benzarti] motivates asymmetry and other incidence anomalies; its accessible abstract does not identify a derivative kink. This attempt establishes a sharp joint identified set from one symmetric randomized reform size under bounded side-specific curvature. It also constructs a price-setting mechanism producing a genuine kink, and an intervention distinguishing that mechanism from ordinary smooth curvature. The empirical target across markets remains unresolved.

Throughout, tax changes are specific taxes per unit. The announcement state, duration, aggregation weights, and observation horizon are fixed. Write $f(t)$ for the mean tax-inclusive price under a fresh intervention of size $t$. A tax rise followed by reversal is a different experiment. Prices may be translated by a sufficiently large constant to stay positive; that translation never changes a contrast.

What finite reforms identify

Suppose independent markets are randomly assigned $t\in\{-d,0,d\}$ with positive probabilities. No market affects another, the assigned regime is implemented, and prices have finite means. Randomization identifies $f(-d),f(0),f(d)$ at the population-law level. It does not identify an arbitrary derivative between treatment points.

For $t>0$ let $g_+(t)=f'(t)$, and for $t<0$ let $g_-(t)=f'(t)$. Assume $f$ is continuous at zero and absolutely continuous on each half interval, with finite limiting derivatives $\rho_+,\rho_-$. Impose the explicit curvature restriction \[ |g_+(t)-\rho_+|\leq L t,\qquad |g_-(-t)-\rho_-|\leq L t\quad(0<t\leq d), \tag{1} \] where $L\geq0$ is known. This is weaker than requiring a common second derivative across zero. Let \[ s_+=\frac{f(d)-f(0)}d,\qquad s_-=\frac{f(0)-f(-d)}d,\qquad A_d=s_+-s_-. \]

Theorem (Sharp derivative and finite-change set): Under (1), given the three identified price means, the sharp joint set for $(A,A_d)$ is \[ \left\{(a,A_d): A_d-Ld\leq a\leq A_d+Ld\right\}. \tag{2} \] The sharp joint set for the two directional derivatives is the rectangle \[ \rho_+\in[s_+-Ld/2,s_++Ld/2],\qquad \rho_-\in[s_--Ld/2,s_-+Ld/2]. \tag{3} \] These claims concern the class imposing (1) and no additional demand, monotonicity, or pass-through bound.
Proof: On the positive half interval the fundamental theorem of calculus gives $s_+=d^{-1}\int_0^d g_+(t)dt$; hence $|s_+-\rho_+|\leq Ld/2$. On the negative half interval the corresponding identity is $s_-=d^{-1}\int_{-d}^0g_-(t)dt$, giving the second inequality. Their difference gives (2).

Conversely, select any pair in (3), set \[ k_+=\frac{2(s_+-\rho_+)}d,\qquad k_-=\frac{2(\rho_--s_-)}d, \] and define $f(t)=f(0)+\rho_+t+k_+t^2/2$ for $t\geq0$ and $f(t)=f(0)+\rho_-t+k_-t^2/2$ for $t\leq0$. Both $|k_+|$ and $|k_-|$ are at most $L$. The function meets the three observed means, is continuous, and satisfies (1). Every difference in (2) is a difference of a pair in (3), since the difference of two closed intervals is the interval of endpoint differences. These functions generate the same experimental price observations, for example with any fixed mean-zero additive measurement noise independent of assignment. Thus all retained points are attainable. ∎

Additional price monotonicity or incidence restrictions must intersect the feasible derivative class, rather than be silently imposed after the sharpness claim. In particular (2) is a sharp result for a reduced-form causal class, not a claim that every element comes from a competitive supply-demand system.

Corollary (Evidence that excludes differentiability): If $|A_d|>Ld$, no function satisfying (1) and the observed three means is differentiable at zero. If $|A_d|\leq Ld$, a differentiable function is compatible with the data.
Proof: Differentiability is exactly $\rho_+=\rho_-$ in this class. Zero is outside (2) precisely in the first case. In the second case the two intervals in (3) overlap; choose their common value and use the preceding quadratic construction. Its side derivatives have that same limit, so it is differentiable at zero. ∎

The elementary bound $|A_d-A|\leq Ld$ is operational: an experiment with several shrinking magnitudes can distinguish a persistent kink from bounded curvature. A smooth quadratic $f(t)=p_0+bt+kt^2/2$ has $A=0$ and $A_d=kd$. It can therefore display sizeable finite asymmetry without any kink.

An explicit optimizing mechanism

Consider a monopolist facing unit demand $q(p)=a-bp$ in an open neighborhood of its baseline optimum, with $a,b>0$, constant marginal resource cost $c$, and initial specific tax zero. Assume $a>bc$, and put $p_0=(a+bc)/(2b)$. A freshly announced tax $t$ is remitted by the seller. At the fixed horizon the firm maximizes \[ \Pi_t(p)=(p-c-t)(a-bp)-J(p-p_0), \quad J(x)=\begin{cases}\kappa_+x^2/2,&x\geq0,\\ \kappa_-x^2/2,&x\leq0, \end{cases} \tag{4} \] where $\kappa_+,\kappa_-\geq0$. Price choices lie in $[0,a/b]$, and $|t|$ is small enough for the solutions below to be interior. The adjustment cost is a real firm cost and is distinct from tax payments. Tax revenue equals $tq(p)$; returning it to consumers would affect a fiscal welfare calculation but not this demand primitive, which assumes no income effect.
Proposition (A genuine asymmetry from asymmetric adjustment curvature): The unique maximizer of (4) is \[ p(t)=p_0+\begin{cases} \displaystyle\frac{bt}{2b+\kappa_+},&t\geq0,\\[4pt] \displaystyle\frac{bt}{2b+\kappa_-},&t\leq0. \end{cases} \] Consequently \[ A=\frac{b}{2b+\kappa_+}-\frac{b}{2b+\kappa_-},\qquad A_d=A \] for all sufficiently small $d$. Every directional pass-through lies in $(0,1/2]$ for finite adjustment coefficients.
Proof: On either side, the derivative of operating profit is $bt-2b(p-p_0)$, so the derivative of (4) is $bt-(2b+\kappa_\pm)(p-p_0)$. The objective is strictly concave on each side; its derivative is continuous at $p_0$ and strictly decreasing globally. At $p_0$ the derivative is $bt$. For $t>0$ the unique zero is on the positive side and equals the stated expression; for $t<0$ it is on the negative side. At zero the unique optimum is $p_0$. The interior condition removes boundary clipping. Substitution gives the derivative and finite-change claims. ∎

The adjustment function is continuously differentiable, but its second derivative jumps when $\kappa_+\ne\kappa_-$. This is enough to kink the optimizer. A fixed menu cost would instead produce a locally flat inaction region at a strict interior state: it does not automatically yield the kink in this proposition.

A feasible distinguishing intervention varies the intensity of the same adjustment technology, replacing $(\kappa_+,\kappa_-)$ by $(z\kappa_+,z\kappa_-)$ with known $z\geq0$, while holding demand and tax information fixed. In this model \[ \frac1{\rho_\pm(z)}=2+\frac{z\kappa_\pm}{b}. \tag{5} \] Thus reciprocal side slopes are affine in $z$ and both equal two at $z=0$. This is an overidentifying restriction across at least three intensities. It is not a universal causal interpretation of any measured adjustment subsidy: the exclusion that only $J$ changes must be defended. Smooth nonlinear pricing under uniformly bounded curvature instead has $A_d\to0$ at every fixed $z$.

Sampling and what remains unidentified

For an elementary finite-sample construction, suppose observed prices are in $[0,B]$, with exactly $n$ independent observations in each treatment arm. Let $\widehat A_d$ use the three arm means. Hoeffding and a union bound imply that, with probability at least $1-\alpha$, all three mean errors are at most \[ e_n=B\sqrt{\frac{\log(6/\alpha)}{2n}}. \] Because $A_d=\{f(d)+f(-d)-2f(0)\}/d$, this event implies \[ |\widehat A_d-A_d|\leq4e_n/d. \] It follows directly from (2) that \[ [\widehat A_d-4e_n/d-Ld,\ \widehat A_d+4e_n/d+Ld] \tag{6} \] covers $A$ with probability at least $1-\alpha$, uniformly over this bounded-outcome class. The proof uses neither normality nor independent potential outcomes. It does require independent sampled markets and fresh common-state interventions. Choosing $d$ of order $n^{-1/4}$ balances the two displayed errors when $L>0$; this is an upper-bound calculation, not a minimax theorem.

Identification with endogenous reforms, unknown anticipation, heterogeneous policy durations, or arbitrary equilibrium switches remains open here. Even a perfectly identified $A$ does not identify (4): other nonsmooth technologies can reproduce the same slopes. Changing the statutory remitter while preserving the economic wedge tests another restriction. In (4), moving remittance to buyers and defining $p$ as the tax-inclusive price leaves profits and the optimizer unchanged only if the adjustment cost remains a function of that same inclusive price. A cost instead attached to the posted seller price changes the primitive model.

Completion Estimate

70%

This is a post hoc, heuristic estimate for the full catalogue target. The attempt establishes a sharp joint kink and finite reform set under bounded side curvature, an honest sampling interval, and an optimizing asymmetric adjustment mechanism with testable restrictions. Endogenous reform assignment, anticipation, equilibrium switches, credible mechanism exclusions, and statutory remitter anomalies across actual markets remain unresolved.

References

Go to contribution →