The Atkinson-Stiglitz theorem is a pivotal result in public economics, particularly in the realm of optimal taxation. Established in 1976 by economists Anthony Atkinson and Joseph Stiglitz, the theorem explores the conditions under which an optimal tax system can be achieved without resorting to commodity taxes. This derivation adopts Kaplow's analytical approach, focusing on constructing distribution-neutral tax reforms to evaluate their efficacy in achieving optimal taxation.
Consider an economy composed of individuals indexed by \( i \), each endowed with a type \( z_i \) representing innate characteristics such as earning ability or taste preferences. The utility function for individual \( i \) is given by:
\[ U_i(c, l; z_i) = u(c) + v(l; z_i) \]
Here, \( c \) denotes consumption, and \( l \) represents labor (or leisure if preferred). The utility function is assumed to be separable, meaning that consumption and labor/leisure decisions are independent of each other.
The economy consists of multiple commodities. One commodity is designated as the numeraire with a normalized price of 1. All other commodities face ad valorem taxes \( t_j \), altering their relative prices to \( p_j = (1 + t_j) \cdot p_j^0 \), where \( p_j^0 \) is the pre-tax price.
The government's objective is to raise a fixed amount of revenue \( R \) while minimizing economic distortions. It can achieve this through a combination of non-linear income taxes \( T(z_i) \) and commodity taxes \( t_j \). The central question is whether incorporating commodity taxes alongside income taxes can lead to a more efficient tax system.
Begin by introducing a small perturbation \( \delta t_j \) to the commodity tax \( t_j \). This alteration changes the relative price of commodity \( j \), influencing individual consumption choices. The new price becomes:
\[ p_j' = (1 + t_j + \delta t_j) \cdot p_j^0 \]
To ensure that the perturbation does not alter the distribution of after-tax incomes, adjust the income tax schedule \( T(z_i) \) accordingly. The adjustment \( \delta T(z_i) \) is calibrated such that for each individual:
\[ (w_i \cdot l_i - T(z_i)) = (w_i \cdot l_i - (T(z_i) + \delta T(z_i))) \]
This ensures that the individual's disposable income remains unchanged despite the change in commodity taxes.
The reform must be revenue-neutral, meaning that the increase in revenue from the adjusted commodity taxes must be offset by the decrease in revenue from the altered income taxes. Mathematically, this condition is expressed as:
\[ \sum_j \delta t_j \cdot q_j + \int \delta T(z_i) \, dF(z_i) = 0 \]
Where \( q_j \) is the consumption quantity of commodity \( j \), and \( F(z_i) \) is the distribution of types.
The change in commodity taxes alters the relative prices, driving individuals to substitute between taxed and untaxed goods. Given the separable utility function, the demand for taxed goods responds to the new prices, modifying consumption patterns without directly affecting labor supply decisions.
Individuals maximize their utility subject to their budget constraints. The introduction of \( \delta t_j \) alters the budget constraint as follows:
\[ c_i = w_i \cdot l_i - T(z_i) - \sum_j (1 + t_j + \delta t_j) \cdot p_j^0 \cdot q_j \]
The adjustment \( \delta T(z_i) \) ensures that the utility level remains constant despite the change in commodity taxes.
The key question is whether the introduction of \( \delta t_j \) introduces any welfare gains or losses. Under the assumption of separable preferences, the non-linear income tax can fully capture the redistribution objectives without needing to distort relative commodity prices through taxes.
Introducing commodity taxes leads to inefficiencies by distorting consumption choices. These distortions represent a welfare loss that is not counterbalanced by any additional redistributive benefits, given the optimal design of the non-linear income tax.
Since the income tax adjustments \( \delta T(z_i) \) maintain the distribution of disposable incomes, any potential redistributive gains from commodity taxes are nullified. Therefore, the only net effect is the efficiency loss from the distorted commodity prices.
A tax system is Pareto optimal if no individual can be made better off without making another worse off. The Atkinson-Stiglitz theorem examines whether incorporating commodity taxes alongside optimal income taxes can achieve Pareto improvements.
By introducing \( \delta t_j \) and adjusting \( \delta T(z_i) \) to maintain distribution neutrality, we observe that commodity taxes do not provide any additional Pareto improvements. Instead, they introduce inefficiencies without corresponding welfare gains.
Commodity taxes do not offer any extra screening mechanisms beyond what non-linear income taxes can achieve. Since the income tax is already optimally designed to address incentive issues and redistribution, commodity taxes fail to contribute meaningfully to these objectives.
The presence of commodity taxes introduces distortions in relative prices, leading to inefficient consumption choices. These distortions are unnecessary because the non-linear income tax already fulfills the redistribution and incentive requirements efficiently.
The analysis concludes that under the assumptions of separable preferences and the availability of a non-linear income tax system, commodity taxes do not enhance the efficiency or equity of the tax system. Instead, they introduce avoidable inefficiencies.
The Atkinson-Stiglitz theorem thus asserts that in environments where preferences are separable between consumption and labor/leisure, and where non-linear income taxation is feasible, optimal taxation does not require commodity taxes. Non-linear income taxes alone can achieve the desired redistribution without incurring the inefficiencies associated with commodity taxes.
Policymakers aiming for optimal tax systems should focus on designing progressive, non-linear income tax schedules tailored to income distributions rather than relying on commodity taxes. This approach ensures efficient resource allocation while achieving equitable income redistribution.
The Kaplow-style derivation of the Atkinson-Stiglitz theorem demonstrates that under the assumption of separable utility functions, optimal taxation can be effectively achieved through non-linear income taxes alone. Introducing commodity taxes, even when adjusted to maintain distribution neutrality, does not offer additional benefits and only serves to introduce inefficiencies into the economic system. Therefore, the theorem provides a strong theoretical foundation for prioritizing income taxation over commodity taxation in the design of optimal tax systems.
For further reading and detailed formalizations, please refer to the following resources: