Thread by @EtienneLehmannE

I take the opportunity of posting the slides of my short (9h) course in optimal tax theory for a thread to discuss

What are the most important and policy-relevant principles to learn?

What are the main current limits and the next steps for further research?

1/

To me, the theory of optimal taxation aims at better understanding the determinants of the optimal various tax rates. In this “applied theory” & “normative” perspective, I wish to answer:

On which assumptions and estimable statistics depend the optimal various tax rates?

2/

I start with optimal nonlinear income tax, because an optimal tax system does not exist in asolute terms.

dropbox.com/s/waf7j0ubjs1t…

At best, there exists an optimal tax system GIVEN a well-defined but always politically disputable social objective 3/

In my course, I (almost) forgive the Mirrlees (1971) mechanism design approach for the tax perturbation approach initiated by @PikettyLeMonde (RFE 1997) and Saez (REStud 2001), which is far more intuitive. 4/

This approach is now 100% rigorous

By differentiating taxpayers foc with respect to the magnitude of any smooth perturbations (as in #Golosov-Tsivinsky @nicolas_werquin), one can test the social desirability of any tax reform 5/

and then deduce the necessary condition for the optimal tax system.

The trick is to redefine substitution and income effects taking into account that a tax reform
=> labor supply responses
=> Change in marginal tax rates
=> further labor supply responses
=> etc.
6/

In practice, these total responses are close to usual ones except at kinks.

The tax perturbation approach requires slightly different assumptions (no jumping is the most crucial). It is flexible enough to treat multidimensional heterogeneity and have many other advantages 7/

For instance, one can use this approach not only to test whether a given tax schedule is Pareto efficient (Stiglitz (1982), @IvanWerning (2007)) but also to identify which tax reform leads to unanimous improvement @Dominik__Sachs, @pierrecboyer @APeichl and their coauthors 8/

I also teach optimal commodity taxation, because it tells us a lot about the optimal structure of tax. dropbox.com/s/w70kwjcmn0w1…

I start with the Ramsey formula for pedagogical reasons and for its implication to the optimal pricing of multi-product public monopoly 9/

But I think the most insightful result is Atkinson-Stiglitz (1976), and more specifically its proof by

To explain this result with words, I think the fundamental question of optimal tax theory is the following one :

When does a tax distortion (in the sense of perturbation between supply and demand due to to tax or transfers) appears socially desirable?

11/

One obvious responses relies on situations where “market frictions” prevents the decentralization by the market of the social optimum.

Many candidates, especially, externalities #Pigouvian_Tax, inatention, …

Call this the corrective role of taxation.

12/

Otherwise, government expenditures, whatever their social necessity, do not justify alone tax distortions. Because they can be in the theory financed by a non-distorsive lump sum tax.

13/

So, tax distortions are socially desirable either for corrective purposes or to make taxpayers’ liabilities better connected to their heterogeneous and unobservable ability to pay.

In Atkinson Stiglitz (1976), the assumption of weakly separable preference implies that the difference in the structure of consumption are only due to difference in income.
14/

In such a context, a higher tax rate on commodities on which the richest spend a larger share of their income is only an indirect way to differentiate total tax liabilities as a function of income 15/

Differentiated commodity taxation thus induce the same labor supply distortions AND equity gains as Income taxation, but it adds USELESS distortions on commodity choices 16/

Can we then intuitively explain production efficiency /no tax on intermediate inputs? When there is no profits or when profits are 100% taxed, distorting input price for firms does not help making households tax liabilities being better correlated with their ability to pay. 17/

Should capital be taxed? Yes if labor income taxation does not capture difference in ability to pay due to differences in
i) propensity to save,
ii) wealth
iii) financial literacy and access to high financial returns.
iv) ...
@wwwojtekk @PikettyLeMonde @gabriel_zucman #Saez ...

@_basjacobs @KevinSpiritus @daniwaldenstrom @spencerbastani, Robin Boadway, Helmuth Cremer, Pierre Pestieau, JC Rochet, ....

How to tax carbon emissions? By combining Pigouvian tax with an adequate income tax reform that neutralize the incentive and redistributive effects of the Pigouvian tax.

This feasible only when preferences are weakly separable and sub-utility is homogeneous.

There many limits to optimal tax theory. First, my presentation does not give any justice to dynamic public finance.

Apologizes to many great scholars, including @IvanWerning, #Golosov #Tsyvinsky #Kocherlakota ...

Anyway, I think another very promising research avenue is to plug optimal tax models within macro / general equilibrium models @Florian_Scheuer #CaseyRotshild @Dominik__Sachs @nicolas_werquin with relevant market frictions e.g. @S_Stantcheva

To close this thread. Deepest apologize to the very numerous scholars that deserved being quoted for their great contributions and that I did not mention here. Please feel free to fil the gaps.