Public Finance Seminar Spring 2015, Professor Yinger Sales Taxes.

Public Finance SeminarSpring 2015, Professor Yinger

Sales Taxes

Sales Taxes

Class Outline

Administration of the Sales Tax

Tax Incidence (Equity)

Tax Distortion (Efficiency)

Recent Research on the Sales Tax

Sales Taxes

The Basics

The sales tax is really a sales and use tax.

◦States levy a tax on the purchase or use of a commodity or service within their boundaries.

 ◦States cannot levy a tax on the purchase or use

of a commodity in another state. What counts is point of delivery. So wine shipped from one state to another is taxed where it is delivered.

Sales Taxes

The Supreme Court Ruling

State A cannot force firms in other states to collect sales tax even when they deliver to customers in State A—unless the firm has “nexus” in State A. This is based on the Supreme Court’s interpretation of the Interstate Commerce Clause in the U.S. Constitution.

  A lot of revenue is lost here, because this applies to catalogue

sales and to internet purchases.

Taxing the internet is a different issue. Congress has placed a moratorium on taxing access to the internet, which is a totally different thing than taxing products purchased over the internet.

Any state can tax internet purchases delivered by firms in their state to people in their state—and require the firms to collect the tax.

Sales Taxes

The Use Tax

A state can collect a use tax from its citizens once a product is delivered, even if the purchase is from an out-of-state firm, but this is virtually impossible to administer. New York has a line on the state tax forms requesting voluntary reporting of this tax.

 In some cases, states can solve the use-tax

enforcement problem through interstate agreements to trade credit card information.

Sales Taxes

Plugging the Leak

The definition of “nexus” is in doubt. New York went after Amazon because its affiliates have nexus in New York. This strategy was approved by the U.S. Supreme Court.

◦ Amazon has dropped its affiliates in some states to avoid collecting taxes.

◦ But in big states, Amazon discovered that a distribution center (=nexus!) quickened delivery times and gained customers.

There is an on-going effort to create standardized product definitions and even tax rates so that the “burden” identified by the Supreme Court disappears.

◦ About 20 states have signed onto this Interstate Compact, and it has passed the Senate (with a small-firm exception), but full Congressional approval is not assured.

Sales Taxes

Tax Incidence

Tax incidence has two meanings.

◦Legal incidence is about who has the legal obligation to pay a tax. With a sales tax, this is usually the seller. This is the meaning relevant for administration.

 ◦Economic incidence is about whose

opportunities are constrained by a tax, that is, who bears the burden. This is the meaning relevant for evaluating the fairness of a tax.

Sales Taxes

Analysis of Tax Incidence

An analysis of tax incidence has two steps.

◦ Figure out which side of the market, demand or supply, bears the burden of the tax.

 ◦ This is about tax shifting. Can the side on which the

legal incidence falls shift it to the other side? If the legal incidence is on the seller, does P rise enough to compensate for the tax?

 Given the answer to step 1, figure out the

distribution of the tax burden by income class.

Sales Taxes

Market Analysis

Standard incidence analysis begins by looking at the impact of a tax on the market for the taxed good or goods.

This analysis does not apply to the market for one taxed good if other goods are being taxed, too.

This analysis is based on relative prices, which do not change for two goods that are both taxed

Sales Taxes

Market Analysis, 2

Possibility 1: Consumers Pay

P

Q Q

P

S

S+tax

D

S

D

P1=P3

P2

P1=P3

S+tax

P2

Sales Taxes

Market Analysis, 3

Possibility 2: Firms Pay

P

Q Q

PSS+tax

D

S

DP3

P1=P2

P3

P1=P2

taxtax

Sales Taxes

Market Analysis, 2

Possibility 3: The burden is shared.

P

Q

SS+tax

D

P1

P2

P3

tax Burden on consumers

Burden on firms

Sales Taxes

A Simple Model of Incidence

Now consider the simplest possible supply and demand equations:

Since QS must equal QD in equilibrium, we can equate the two

right sides and solve for P:

( )

and

sS S

D D D

Q a b P T

Q a b P

D S S

DS

a a b TP

b b

Sales Taxes

Shifting

Now we can take the derivative of P with respect to T; i.e., we can ask how P changes as T changes.

From this it is clear that

1

1

S

D DS

S

bdPdT b b b

b

1if = then half shifted2

if = then 0 not shifted at all

if then 1 fully shifted

D S

D

S

dPb bdT

dPbdT

dPbdT

Sales Taxes

Shifting, 2

This simple example refers to slopes. A more general version would refer to price elasticities. The degree of shifting depends on the relative price elasticities of demand and supply (in absolute value).

The intuition is about responsiveness. A more responsive curve, which suggests more alternatives, implies that a side of the market is better able to escape the tax.

Sales Taxes

Incidence, 2nd Step

The second step is to translate to the distribution of income. This obviously depends on the incomes of suppliers versus demanders. Either one could be higher.

The result is usually summarized with a graph with T/Y on the vertical axis and Y on the horizontal axis. A flat curve indicates a proportional system; a rising curve indicates a progressive system, and a falling curve indicates a regressive system.

Sales Taxes

Incidence by Income

In a graph, these concepts are:

Y

TY

Progressive

ProportionalRegressive

Sales Taxes

Equity Principles

In a policy setting, the positive analysis of tax incidence is combined with judgments about what is fair.

The ability to pay principle says that people with a higher income should pay more, but does not necessarily support progressivity.

The benefit principle says that people who benefit from the funded service should pay, but this does not rule out progressivity within this set of people.

Sales Taxes

Tax Distortion

Taxes affect behavior. This implies that taxes move outcomes away from the efficient level that is generated by a competitive private market (under some circumstances)—i.e. that taxes cause distortion.

As before, the basic analysis applies to taxes markets relative to untaxed markets—not to one of many taxed markets.

Sales Taxes

Excess Burden

Consider an elastic supply curve. The initial equilibrium is (P1, Q1). Because the tax is fully shifted, the new price, P2, equals P1(1+t), or ΔP=tP1.

Now if we draw the picture, we see that consumer surplus drops because of the tax revenue that is collected and because of a lost triangle = excess burden.

Sales Taxes

Excess Burden, 2

In a graph:

P

Q

S

S+tax

D

P1

P2

Q2 Q1

Excess Burden

ΔP = t

ΔQ

Government Revenue

The Market for Taxed Goods

Sales Taxes

Excess Burden, 3

We do not consider the tax revenue as a loss of consumer surplus. Tax analysis usually compares the burden from alternative taxes, so this revenue will be lost no matter what we do.

◦Moreover, even if we were just evaluating a single possibility, the revenue collected presumably results in benefits in the form of public services.

The triangle, called distortion or excess burden or EB, equals (½)ΔQΔP = (1/2)ΔQ(tP1).

Sales Taxes

Excess Burden, 4

Now the absolute value of the price elasticity of demand, say e, equals (P1ΔQ)/(Q1ΔP). So

ΔQ = (eQ1)(ΔP/P1)

= (eQ1)(tP1)/P1 = (eQ1)t

and

EB= ½ (e)(t2)(P1Q1)

This is a central result in tax analysis. Remember it!

Sales Taxes

Excess Burden, 5

This result yields three important lessons in here:

◦ 1. EB increases with the size of the market being taxed (PQ).

◦ 2. EB increases with the price elasticity of demand. Lesson: tax products with a low price elasticity—that leads to minimal distortion (but may not lead to greatest fairness!).

◦ 3. EB increases with the square of the tax rate. The lesson: Have a diversified revenue system.

Sales Taxes

Excess Burden, 6

Why the square of t?

P

Q

S

S+t

D

P1

P2

Q2 Q1

S+(2×t)P3

Q3

Sales Taxes

Excess Burden, 7

An additional important result arises in a market with an externality.

An example is the market for gasoline.

◦ People buy gasoline in order to drive, and when they drive, they cause highway deterioration, pollution, and congestion.

◦ Generally, gas taxes are set to cover maintenance only.

◦ Raising the tax further would actually reduce distortion (i.e. lower excess burden).

Sales Taxes

Excess Burden, 8

Excess Burden with an Externality

P

Q

S = PMC

S + Tax = PMC+SMC

D

P1

P2

Q2 Q1

Excess Burden Avoided with Tax = SMC

ΔP = t

ΔQ

Government Revenue

The Market for Taxed Goods

Sales Taxes

Recent Literature on Sales Tax Shifting

Several recent studies regress P on T in some form or another to see if P goes up dollar for dollar with T.

Poterba (one of the stars of empirical public finance,

an economist at MIT) does this for a few broad categories of goods in a few cities and finds approximately full forward shifting onto consumers.

Besley and Rosen (two more stars; one an editor of the American Economic Review, the other the author of a popular public finance textbook) conduct this analysis for many narrow categories of goods and find overshifting.

Sales Taxes

Terminology

This research is a bit hard to get into because of terminology.

 The key is to distinguish between marginal cost (MC),

the price “inclusive of tax” (Q), and the price “exclusive of tax” (P). The price inclusive of tax is what one observes in a market once the tax is imposed. The price exclusive of tax is what the firm receives.

So with no taxes and competitive markets, P=Q=MC. With taxes, competitive markets, and full shifting:

 Q = P(1+t) = MC(1+t).

Sales Taxes

Poterba

Poterba’s approach is to take market prices, Q, divide by (1+t) to get P, and regress P on t. With full shifting, the coefficient of t will be zero:

 Q/(1+t) = P = MC

 or, in regression form,

P = MC + βt

The predicted value of β is zero; a value above zero indicates overshifting onto consumers.

Sales Taxes

Besley/Rosen

The Besley/Rosen approach is to regress Q on the determinants of MC and t.

Q = MC(1+t)

or

ln(Q) = ln(MC) + βln(1+t) ≈ ln(MC) + γt

In this case, the expected value of γ is 1.0. A value above 1.0 indicates shifting.

How could there be over-shifting? The most plausible explanation is market power. With an elastic demand curve, a monopolist will raise the price by more than the tax.

Sales Taxes

Poterba’s Results

Poterba uses data on categories of spending, namely, women’s and girl’s clothing, men’s and boys’ clothing, and personal care items. One data set is for 1947-77, another is for 1925-39.

  He finds full shifting.  He estimates in difference form, with the dependent variable

the “inflation rate” for the spending category in the city and a controls for the city and national “inflation rate.” Because it is in difference form, the key dependent variable is the change in the tax rate.

  He has lagged values to pick up the phase-in of tax effects

and finds small but significant lagged impacts. The total impacts imply full shifting.

Sales Taxes

Besley/Rosen Results

They use data for the last 30 years for specific products (a Big Mac, a box of Kleenex, a Monopoly game, and a package of 3 boys cotton briefs.) Their regressions include city and time effects and measures of real rental, wage, and energy costs—to capture MC.

The find full shifting for some (including Monopoly, Big Macs, and Kleenex) and overshifting for others.

They also do some dynamic analysis and find that the impact of taxes shows up very rapidly. They also reject Poterba’s differencing approach.

Sales Taxes

Young and Bielinska-Kwapisz

This paper is in the NTJ in 2002.

They find over-shifting of excise taxes on beer and wine.

So taxes are shifted and there is considerable evidence for market power!

Sales Taxes

Hawkins

Finally, Hawkins, (NTJ, December 2002) provides some interesting results on the distortion caused by various forms of the sales tax.

Hawkins estimates a demand system for all broad categories of consumption (nondurables, services, alcohol, gasoline, tobacco, home food, utilities, and autos and furnishings). This is done by assuming a pretty general functional form that yields structural parameters, that is, the parameters of the underlying utility functions--which makes it possible to calculate the EB from various forms of the sales tax.

He looks at 8 cases. The base case has no exemptions and no taxation of business inputs (which he calls pyramiding). The other cases add exemptions and/or taxation of (non-exempted) inputs: Here they are, with the excess burden that results (as a percent of revenue raised):

Sales Taxes

Hawkins’s Results

Case Excess Burden1. Base --2. Services + home food 23.3%

3. Case 2 + utilities 30.6%4. Case 3 + Gasoline 38.5%5. Base + Pyramiding 0.4%

6. Case 2 + Pyramiding 17.9%

7. Case 3 + Pyramiding 22.1%

8. Case 4 + Pyramiding 26.7%

Sales Taxes

Lessons from Hawkins

The exemptions do cause excess burden, as expected. Services and home food, which are usually exempted, have a significant burden. The equity gains must be worth it!

Adding pyramiding actually lowers EB (except in case 5)! This is a surprise. But it happens because the taxed inputs produce goods that have inelastic demands and that tend to be exempt themselves. So ironically, pyramiding actually boosts efficiency by indirectly taxing exempt goods!