14.123 Spring 2005 Peter Diamond page 1 of 6 Handout on taxing externalities with uniform taxes Let A , mnemonic for atmosphere, be the only route for externalities. A = ∑ a i x1i (1) i Note A is the same for all consumers. Note linearity is not important. This could be done with a vector of different atmospheres. I. Quasi-linear preferences, 2 goods, additive externality u h = x0h + f h ( x1h ) + g h ( A) ; f ′h > 0, f ′′h < 0 (2) x0h + qx1h = I h Consumers ignore atmospheric feedback. Interior solution f ′h ( x1h ) = q → x1h ( q ) , x1′h ≤ 0 (3) Social Welfare assuming linear technology with producer price p for good one. ∑h u h =∑h x0h + ∑h f h ⎡⎣ x1h ( q )⎤⎦ + ∑h g h ⎡⎣ A⎤⎦ ⎡ ⎤ = ∑ x0h + ∑ f h ⎡⎣ x1h ( q ) ⎤⎦ + ∑ g h ⎢ ∑ ai x1i ( q ) ⎥ h h h ⎣ i ⎦ (4) ⎡ ⎤ ⎣ ⎦ = ∑ e0h − p ∑ x1h ( q ) + ∑ f h ⎡⎣ x1h ( q ) ⎤⎦ + ∑ g h ⎢∑ ai xi ( q )⎥ h h h h i 14.123 Spring 2005 Peter Diamond page 2 of 6 First order condition from differentiating (4) with respect to q : ⎛ ⎞⎛ ⎞ 0 = − p ∑ x1′h + ∑ f ′h x1′h + ⎜ ∑ g ′h ⎟ ⎜ ∑ a i x1′i ⎟ ⎠⎝ i ⎠ ⎛ ⎞⎛ ⎞ = − p ∑ x1′h + q ∑ x1′h + ⎜ ∑ g ′h ⎟ ⎜ ∑ a i x1′i ⎟ ⎝ h ⎠⎝ i ⎠ h h h ⎝ h h (5) ∑ a x′ ⎛ ⎛ h ⎞⎛ i i⎞ ⎛ i⎞ h⎞ q − p = t = − ⎜ ∑ g ′ ⎟ ⎜ ∑ a x1′ ⎟ / ⎜ ∑ x1′ ⎟ = − ⎜ ∑ g ′ ⎟ ⎝ h ⎠⎝ i ⎠ ⎝ i ⎠ ⎝ h ⎠ ∑ x′ i i 1 i i (6) 1 i II. Quasi-linear preferences, 2 goods, externality not necessarily additive u h = x0h + f h ( x1h , A) ∂f h ( x1h , A) ∂x1h =q → x1h ( q, A ) Solve (8) and (1) simultaneously. Note ∂x1 / ∂q | A different from h (7) d h x1 ( q, A ( q ) ) . dq (8) 14.123 Spring 2005 III. Peter Diamond page 3 of 6 Quasi-linear preferences, 3 goods, additive externality ⎛ ⎞ u h = x0h + f h ( x1h , x2h ) + g h ⎜ ∑ ai x1i ⎟ ⎝ i ⎠ (9) f1h ( x1h , x2h ) = q1 f 2h ( x1h , x2h ) = q2 (10) xih ( q1 , q2 ) Social welfare function ∑h eh − p1 ∑h x1h ( q1, q2 ) − p2 ∑h x2h ( q1, q2 ) + ∑h f h ⎡⎣ x1h ( q1, q2 ) , x2h ( q1, q2 )⎤⎦ + ∑h g h ( A) First order conditions with respect to q1 and q2 i ∂x1h ∂x2h ⎛ h ⎞⎛ i ∂x1 ⎞ 0 = t1 ∑ + t2 ∑ + ⎜ ∑ g′ ⎟ ⎜ ∑ a ⎟ ∂q1 ⎠ h ∂q1 h ∂q1 ⎝ h ⎠⎝ i i ∂x1h ∂x2h ⎛ h ⎞⎛ i ∂x1 ⎞ ′ + t2 ∑ + ⎜∑ g ⎟⎜∑ a 0 = t1 ∑ ⎟ ∂q2 ⎠ h ∂q2 h ∂q2 ⎝ h ⎠⎝ i If, at the optimum we have ∑ ai i (11) ∂x1i ∂x i ≠ ∑ a i 1 , then at the optimum t2 ≠ 0 . ∂q1 ∂q2 i (12) 14.123 Spring 2005 IV. Peter Diamond page 4 of 6 General preferences, additive externality, lump sum redistribution, two goods Max ⎧⎪ ⎡ ⎩ p ⋅ ∑ x ( q, I h s.t. ⎤ ⎫⎪ ∑h ⎨⎪vh ( q, I h ) + g h ⎢⎣∑i ai x ( q, I i )⎥⎦ ⎬⎪ h h i 1 ⎭ )=B (13) Normalizations: p0 = q0 = 1; p1 = p; q1 = q Alternative writing of the resource constraint: ∑ { x ( q, I h ) + px ( q, I h )} = ∑ e h h 1 0 h h 0 (14) h FOC: h i ⎧⎪ ∂x0 h ∂v h ∂x1 ⎫⎪ ⎛ h ⎞⎛ i ∂x1 ⎞ ′ − + g a λ + p ∑h ∂q1 ∑h ⎪⎨ ∂q1 ∂q1 ⎪⎬ ⎝⎜ ∑h ⎠⎟ ⎜ ∑i ∂q1 ⎟ = 0 ⎝ ⎠ ⎩ ⎭ (15) Using Roy’s identity and moving the last term: h i ⎛ ∂x0 h ⎛ ∂v h ∂x1 ⎞ h ⎞⎛ i ∂x1 ⎞ ′ − λ∑⎜ −∑ x1 +p ⎟ = −⎜ ∑ g ⎟⎜∑ a ⎟ ∂I ∂q1 ⎠ ∂q1 ⎠ h h ⎝ ∂q1 ⎝ h ⎠⎝ i h (16) From the individual budget constraints we have: x0h (q, I ) + qx1h (q, I ) = I x1h (q, I ) + ∂x0h ∂x h +q 1 =0 ∂q ∂q (17) ∂x0h ∂x h + q 1 =1 ∂I ∂I With q = p + t , we have: ∂x0h ∂x h ∂x h + p 1 = −t 1 − x1h (q, I ) ∂q ∂q ∂q ∂x0h ∂x h ∂x h + p 1 = 1− t 1 ∂I ∂I ∂I Substituting in (16) we have: (18) 14.123 Spring 2005 Peter Diamond page 5 of 6 −∑ x1h h ⎛ ⎛ ⎞⎛ ∂x h ⎞ ∂xi ⎞ ∂v h + λ ⎜ ∑ x1h + t ∑ 1 ⎟ = − ⎜ ∑ g ′h ⎟ ⎜ ∑ ai 1 ⎟ ∂I ∂q1 ⎠ h ∂q1 ⎠ ⎝ h ⎠⎝ i ⎝ h (19) h We have FOC for each I : h ∂v h ∂x0h ∂x1h ⎛ h ⎞ h ∂x1 ′ +p )+⎜∑ g ⎟a − λ( =0 ∂I ∂I ∂I ∂I ⎝ h ⎠ (20) Substituting from the budget constraints: ∂v h ∂x h = λ − λt 1 ∂I ∂I ⎛ ⎞ − ⎜ ∑ g ′h ⎟ a h ⎝ h ⎠ ∂x1h ∂I (21) The Slutsky equation s1h = ∂x1h ∂x h + x1h 1 ∂q1 ∂I (22) Substituting from the Slutsky equation in (19): h i ⎞ ⎛ ⎛ ⎞ ⎛ ⎛ h ∂v h h h⎞ i⎛ i h ∂x1 ⎞ i ∂x1 ⎞ −∑ x + λ ⎜ ∑ x1 + t ∑ ⎜ s1 − x1 ⎟ ⎟ = − ⎜ ∑ g ′ ⎟ ⎜⎜ ∑ a ⎜ s1 − x1 ⎟⎟ ∂I ∂I ⎠ ⎠ ∂I ⎠ ⎟⎠ h h ⎝ ⎝ ⎝ h ⎠⎝ i ⎝ h h 1 (23) Rearranging terms: i ⎛ ⎛ h ∂x1h ⎞ ⎛ ∂v h h h⎞ h ⎞⎛ i i⎞ ⎛ h ⎞⎛ i i ∂x1 ⎞ −∑ x + λ ⎜ ∑ x1 + t ∑ s1 ⎟ = − ⎜ ∑ g ′ ⎟⎜ ∑ a s1 ⎟ + ⎜ ∑ g ′ ⎟ ⎜ ∑ a x1 ⎟ + λt ∑ ⎜ x1 ⎟ ∂I ⎠ ∂I ∂I ⎠ h h h ⎝ ⎝ h ⎠ ⎝ h ⎠⎝ i ⎠ ⎝ h ⎠⎝ i h 1 (24) Substituting from (21): −∑ x1h h ⎛ ⎛ ∂v h ⎛ ⎞⎛ ⎞ ⎞⎞ ⎛ ⎞ ∂v h + λ ⎜ ∑ x1h + t ∑ s1h ⎟ = − ⎜ ∑ g ′h ⎟ ⎜ ∑ a i s1i ⎟ − ∑ ⎜ x1h ⎜ − λ ⎟⎟ ∂I h ⎝ h ⎠ ⎠⎠ ⎝ h ⎠⎝ i ⎠ h ⎝ ⎝ ∂I (25) 14.123 Spring 2005 Peter Diamond page 6 of 6 Simplifying, we have: ⎛ ∑ g ′h ⎞ ⎛ ∑ a h s1h ⎞ ⎟ ⎟⎜ h t = − ⎜⎜ h h ⎟ ⎜ ⎟ ⎜ λ ⎟ ⎜ ∑ s1 ⎟ ⎝ ⎠⎝ h ⎠ (26) h ∂v h ∂x1h ⎛ h ⎞⎛ h ∂x1 ⎞ ∑h ∂I − λ H = −λt ∑h ∂I − ⎜ ∑h g ′ ⎟ ⎜ ∑h a ∂I ⎟ ⎝ ⎠⎝ ⎠ (27) Adding (21) over h: Substituting for λt : λH − ∑ h ⎛ ⎞⎛ ⎛ ⎞⎛ ∂x h ⎞ ∂v h ⎞ ⎛ ∂x h ⎞ = − ⎜ ∑ g ′h ⎟ ⎜ ∑ a h s1h ⎟ ⎜ ∑ 1 ⎟ / ∑ ( s1h ) + ⎜ ∑ g ′h ⎟ ⎜ ∑ a h 1 ⎟ ∂I ∂I ⎠ ⎠ ⎝ h ∂I ⎠ h ⎝ h ⎠⎝ h ⎝ h ⎠⎝ h (28) or ⎧ ⎛ h ∂x1h ⎞ ⎛ ∂x1h ⎞ ⎛ ⎛ ∂v h ⎞ ⎛ ∂x1h ⎞ ⎛ ⎞ ⎛ ⎞ ⎫⎪ h ⎞⎪ ′ λ H − = g ∑h ∂I ⎟ ⎜ ∑h ∂I ⎟ ⎜⎝ ∑h ⎟⎠ ⎨⎪∑h ⎜ a ∂I ⎟ / ⎜ ∑h ∂I ⎟ − ⎜⎝ ∑h a h s1h ⎟⎠ / ⎜⎝ ∑h s1h ⎟⎠ ⎬⎪ ⎜ ⎝ ⎠ ⎝ ⎠ ⎠ ⎝ ⎠ ⎩ ⎝ ⎭ (29) h ⎧ ⎫ h ∂x1 a a h s1h ⎪ ∑ ∑ h h ⎪ ⎛ ∂x ⎞ ⎛ ∂v ⎞⎪ ⎪ ∂I + ∑ ⎜ 1 ⎟ ⎜ ∑ g ′h ⎟ ⎨ h − h h ⎬ λH = ∑ h ∂x1 h ∂I h ⎝ ∂I ⎠ ⎝ h ⎠⎪ ∑h s1 ⎪ ∑ ⎩⎪ h ∂I ⎭⎪ (30) or