Harberger approximation to deadweight loss from a small ad valorem tax

Writes the triangle in terms of spending before the tax and the price elasticity of demand. With constant marginal cost the wedge is tau times p and quantity falls by about eps times tau times q, so the loss is half their product. Zero elasticity gives zero loss, and doubling the tax rate roughly quadruples it. The expression is the single-tax case of the second-order approximation that Auerbach and Hines attribute to Harberger, with the quantity response written through the elasticity.

Signature

DWL = 0.5 * eps * tau^2 * p * q
Inputs
InputsDefinitionUnit
epsAbsolute value of the price elasticity of demand at the pre-tax price and quantity, compensated in principlenone
tauTax or wedge as a proportion of the pre-tax price p, zero or aboveproportion
pPrice before the tax, equal to the constant marginal costcurrency per unit
qQuantity bought before the taxunits per period
Output
DWLApproximate excess burden of the taxcurrency per period

Function

Deadweight loss of a wedge between marginal value and marginal cost

Maps a price distortion, such as a tax, a subsidy, a price above cost or insurance that lowers the price at the point of use, to the net surplus it destroys: the value of trades it prevents in excess of their cost, or the cost of trades it adds in excess of their value. Money that only changes hands, such as tax revenue or an insurer's payments, is netted out. With straight-line curves the area is a triangle whose height is the wedge and whose base is the change in quantity. The notation follows the Deadweight Loss article.

Try this function

Implementations

  • Excel

    Harberger deadweight loss from named elasticity, tax rate and spending

    With the absolute elasticity in Elasticity, the tax as a proportion in TaxRate and the pre-tax price and quantity in PreTaxPrice and PreTaxQty, the formula returns the approximate loss, held in a cell named HarbergerDWL.

    =0.5*Elasticity*TaxRate^2*PreTaxPrice*PreTaxQty

Assumptions

  • Small tax on a good supplied at constant marginal cost

    The producer price stays fixed, so the whole tax passes to buyers, and the tax is small enough for the second-order approximation to hold. On linear demand the expression is exact; on curved demand it is an approximation.

  • Compensated elasticity and no other distorted market

    The elasticity is the compensated one; Hines reports that ordinary triangles usually approximate compensated measures closely, with exceptions. Other markets are undistorted, so no cross-market terms arise.

Worked examples

  • Twenty per cent tax at the article's full price

    At the article's full price of 50 pounds and 5,000 sessions, demand q = 10000 minus 100p has a point elasticity of 1.0. A 20 per cent tax raises the price to 60 pounds and cuts use by 1,000, a loss of 5,000 pounds, which the approximation reproduces exactly because demand is linear (computed here for illustration).

    eps = 1; tau = 0.2; p = 50; q = 5000; DWL = 5000
  • Doubling the tax rate quadruples the Harberger loss

    At a 40 per cent rate on the same curve the loss is 20,000 pounds, four times the loss at 20 per cent (computed here for illustration).

    eps = 1; tau = 0.4; p = 50; q = 5000; DWL = 20000
  • Elasticity of 0.2 on the less responsive curve

    On the article's less responsive curve, q = 6000 minus 20p, the elasticity at 50 pounds is 0.2, and the same 20 per cent tax costs 1,000 pounds (computed here for illustration).

    eps = 0.2; tau = 0.2; p = 50; q = 5000; DWL = 1000

Common errors

  • Applying the approximation to a large tax on curved demand

    With a constant elasticity of 1 instead of linear demand, a 40 per cent tax on the same starting point gives a Marshallian loss of about 12,690 pounds, while the approximation gives 20,000 (computed here for illustration). For large wedges the area is computed from the demand curve itself.

  • Entering the tax per unit in place of the tax rate

    Tau is a proportion of the price. Entering a tax of 10 pounds per session as tau = 10 instead of 0.2 multiplies the result by 2,500.

Sources

  • Second-order approximation to the excess burden of a single tax

    Auerbach AJ, Hines JR Jr. Taxation and economic efficiency. In: Auerbach AJ, Feldstein M, editors. Handbook of Public Economics. Vol 3. Amsterdam: Elsevier; 2002. p. 1347-1421 (read as NBER Working Paper 8181, 2001). Section 2.2: the second-order approximation around the undistorted point approximates the total excess burden (Harberger 1964a); for a single tax with producer prices fixed, the excess burden is minus one half of the tax change times the compensated demand response to it times the tax change, so excess burden increases with the square of a tax. Section 2.3: Harberger derived the approximation used to measure deadweight loss.

    View source →

  • Loss rising with the square of the tax rate and compensated measures

    Hines JR Jr. Three sides of Harberger triangles. Journal of Economic Perspectives. 1999;13(2):167-188 (read as NBER Working Paper 6852, 1998). Dupuit observed that the welfare loss triangle is generally a function of the square of the tax rate; Harberger's 1964 papers derived and applied the triangle method; Harberger triangles usually approximate compensated measures closely.

    View source →

Canonical Identity