Recitation I Notes: Exact Consumer Surplus

14.471: Fall 2012: Recitation I: Exact consumer surplus September 13, 2012 Part I Outline/Take-aways 1. Goal: Measuring welfare loss/excess burden from taxation in an “exact way” in a empircally implementable way. 2. marshallian demand: In the first class we drew triangles. marshallian Excess Burden (EB) is appealing because it is easy and observable. But marshallian surplus is an ad-hoc measure and suffers from pathdependence. • Syndrom: path dependence • Conceptual issues 3. Hicksian demand: Solution: Introduce Hicksian EB and expenditure function to translate the utility loss into dollars • BUT: Hicksian demand is not observable and depends on utility measure h(q, u) • At what utility to measure Hicksian EB? 2 candidates: – Pre-tax utility: CV – Post-Tax utility: EV 4. Hicksian Excess burden measures without conceptual problems (e.g. Diamond-Mc Fadden) 1 Part II Content 1 Goal Measuring EB from taxes: • Rigorous, exact way consistent with micro-theory • Empirically implementable 2 marshallian surplus: Pros and Cons 2.1 Reminder: Definitions of marshallian/uncompensated demand. Individual solves max u(c) c s.t. q·c≤Z where q = p + t. With λ being the multiplier on the budget constraint, the FOC in ci is uci = λqi These condition implicitly define: • xi (q, Z): Marshallian (uncompensated) demand function • v(q, Z): the indirect utility function 2.2 Pro: Measuring EB with marshallian surplus is simple • EB is a simple triangular surface:EB = 12 dQdτ • Per-unit tax rate τ and (taking the total derivative of the) market equilibrium D(p + τ ) = S(p) leads to the following change in the producer price: dp ηD = p+τ (1) dτ η S − ηD p • Using that initial taxes are zero and (1) the ratio between the Harberger triangle DW L and raised revenues R equals: DW L 1 η S ηD = R 2 ηS − ηD 2 τ p (2) 2.3 Con: Path dependence problem when more than one price changes 2.3.1 Illustration with taxes on 2 goods A and B Let us compare: • The change in consumer surplus when we move: ( 0 0) ( 1 0) – first from qA , qB to qA , qB ( 1 0) ( 1 1) – and then from qA , qB to qA , qB 1 ˆqA CStop = 1 ˆqB 0 cA (qA , qB , Z)dqA + 1 cB (qA , qB , Z)dqB 0 qB 0 qA with • The change in consumer surplus when we move: ( 0 0) ( 0 1) – first from qA , qB to qA , qB ( 0 1) ( 1 1) – and then from qA , qB to qA , qB 1 ˆqB CStop = 1 0 cB (qA , qB , Z)dqB + 0 qB ˆqA 1 cA (qA , qB , Z)dqA 0 qA Class Question: Who knows under which condition they are equal? dcA B For CS1 = CS2 , we need equal cross-partials dq = dc dqA . This will not be satisfied for marshallian demand B functions unless there are no income effects (b/c income effects and initial consumption levels differ across goods). But they are equal for Hicksian (compensated) demand. Indeed, the Slutsky Matrix- containing the derivatives of the Hicksian demand function which holds utility constant- is symmetric. sAB = 2.3.2 ∂2e ∂qA ∂qB Why do we have a price dependence issue? Conceptual problems • Path-dependence problem reflects the fact that consumer surplus is an ad-hoc measure. – It is not derived from utility function or a welfare measure. • Question of interest: How much utility is lost because of tax distortions beyond the revenue transferred to government? • Need units to measure “utility loss” 3 3 Solution 3.1 Introduce expenditure function to translate the utility loss into dollars (money metric) • Hicksian/compensated demand hA (q, u) (“Demand of a consumer over a bundle of goods that minimizes their expenditure while delivering a fixed level of utility”) is the solution to the problem where you minimize costs: e(q, u) = min q · h h s.t. U (h) ≥ U • The value of the minimum is the expenditure function e(q, U ) • Now define individual’s loss from tax increase as: e(q 1 , u) − e(q 0 , u) • e(q 1 , u) − e(q 0 , u) is a single-valued function -> coherent measure of welfare cost, no path dependence issue 3.2 3.2.1 But where should u be measured? Measure utility at initial price level (before tax change): CV The compensating variation is the cost to a consumer of a shift in prices from q 0 to q 1 where the expenditure function e(q 1 , u0 ) is evaluated at the old utility level u0 : CV = e(q 1 , u0 ) − e(q 0 , u0 ) Intuition: How much compensation is needed to reach original utility level at new prices? 3.2.2 Measure utility at new price level (after tax change): EV The equivalent variation is the cost to a consumer of a shift in prices from q 0 to q 1 where the expenditure function e(q 1 , u1 ) is evaluated at the new utility level u1 : EV = e(q 1 , u1 ) − e(q 0 , u1 ) Intuition: Lump sum amount agent willing to pay to avoid tax? 3.3 3.3.1 Can we derive empirically implementable formulas Hicksian demand as we did graphically with marshallian demand? Build some intuition on graphical representation of Hicksian demand • marshallian demand is more elastic than Hicksian because a drop in prices q – 1) substitution: pushes x up 4 – 2) income (if normal): pushes x up Thus marshallian reaction of quantities to prices is bigger than Hicksian effect restricted to substitution only. Figure 1: marshallian demand is more elastic than Hicksian demand • EV is the area to the left of the Hicksian Demand Curve related to NEW utility. Figure 2: EV graphical 5 • CV is the area to the left of the Hicksian Demand Curve related to OLD utility. Sketch of the proof: – CV = e(q 1 , u0 ) − e(q 0 , u0 ) – Suppose only q1 changes. The integral from q 0 to q 1 of ´ q1 q0 h(u0 )dq1 – Use now that h= – to get that ´ q1 q0 ∂e ∂q1 h(u0 )dq1 = e(q 1 , u0 ) − e(q 0 , u0 ) = CV Figure 3: CV graphical • With one price change EV < M arsSurplus < CV but this is not true in general. 6 Figure 4: marshallian surplus graphical 4 Precise EB • Deadweight burden: change in consumer surplus less taxes paid • Equals what is lost in excess of taxes paid (“net cost measure of intervention”) • Diamond and McFadden (1974) measure the deadweight burden of a tax-induced price change from q0 to q1 = q0 + τ as follows by subtracting compensated revenue (not observed): EB(u0 ) = CV − (q 1 − q 0 )h(q 1 , u0 ) • Mohring (1971) works at new utility levels: EB(u1 ) = EV − (q 1 − q 0 )h(q 1 , u1 ) • In general, CV and EV measures of EB will differ (equal when quasilinear utility) – CV is useful if you have multiple alternatives because you always look at old utility. 7 Figure 5: CV and EV measures of EB can differ • marshallian measure overstates excess burden because it includes income effects – Income effects are not a distortion in transactions – Buying less of a good due to having less income is not an e!ciency loss; no surplus foregone b/c of transactions that do not occur • Thus: Hicksian EB is appealing because no path dependence. But unappealing because not observable and depends on utility measure h(q, u) 8 Part III Appendices: 4.1 Symmetry Slutsky • Hicksian demand hA (q, u) is the solution to the problem where you minimize cost: e(q, u) = min q · h h U (h) ≥ U • The value of the minimum is the expenditure function e(q, U ) • By envelope (the change in the value function is given by the partial derivative of the Lagrangian with respect to choice variable ) Hicksian demand equals hA = ∂e ∂qA • now the Slutsky matrix is defined as the price derivative of hicksian demand sAB = ∂hA ∂qB • If we now use (??), we get the symmetry of the Slutsky result: sAB = 4.2 ∂2e ∂qA ∂qB Reminder: Slutsky & Roy identity • Slutsky based on deriving h(q, U ) = g(q, e(q, u)) wrt q into a direct and indirect effect. ∂gA ∂gA ∂gA ∂e ∂gA ∂hA = + = + hB ∂qB ∂qB ∂qB ∂e ∂qB ∂R • Roy’s identity: ∂V ∂q gi = − ∂Vi ∂R is based on taking ratio of ∂V =λ ∂R and ∂V = −λgi ∂R 9 4.3 References • Auerbach (1985) on Path dependence problem • Duality theory and welfare evaluation of economic changes: – Quick: Salanié, B., 2011. The Economics of Taxation, The MIT Press: Appendix A: “Some Basic Microeconomics” – Depth: MWG section 3.I “welfare evaluation of economic changes 3.1” • Lecture notes Raj Chetty 10