d log w F = 1(i = k) + d log A k b Ψe k + b Ψ α d log w

Σχετικά έγγραφα
Πανεπιστήµιο Κρήτης - Τµήµα Επιστήµης Υπολογιστών. ΗΥ-570: Στατιστική Επεξεργασία Σήµατος. ιδάσκων : Α. Μουχτάρης. εύτερη Σειρά Ασκήσεων.

1 Complete Set of Grassmann States

One and two particle density matrices for single determinant HF wavefunctions. (1) = φ 2. )β(1) ( ) ) + β(1)β * β. (1)ρ RHF

Multi-dimensional Central Limit Theorem

α & β spatial orbitals in

Constant Elasticity of Substitution in Applied General Equilibrium

Multi-dimensional Central Limit Theorem

2 Composition. Invertible Mappings

ΗΥ537: Έλεγχος Πόρων και Επίδοση σε Ευρυζωνικά Δίκτυα,

8.324 Relativistic Quantum Field Theory II

Variance of Trait in an Inbred Population. Variance of Trait in an Inbred Population

Phys460.nb Solution for the t-dependent Schrodinger s equation How did we find the solution? (not required)

Pricing of Options on two Currencies Libor Rates

CHAPTER 25 SOLVING EQUATIONS BY ITERATIVE METHODS

HOMEWORK 4 = G. In order to plot the stress versus the stretch we define a normalized stretch:

derivation of the Laplacian from rectangular to spherical coordinates

Ordinal Arithmetic: Addition, Multiplication, Exponentiation and Limit

3.4 SUM AND DIFFERENCE FORMULAS. NOTE: cos(α+β) cos α + cos β cos(α-β) cos α -cos β

8.1 The Nature of Heteroskedasticity 8.2 Using the Least Squares Estimator 8.3 The Generalized Least Squares Estimator 8.

Neutralino contributions to Dark Matter, LHC and future Linear Collider searches

The Simply Typed Lambda Calculus

Section 7.6 Double and Half Angle Formulas

Finite Field Problems: Solutions

Statistical Inference I Locally most powerful tests

EE512: Error Control Coding

Lecture 2. Soundness and completeness of propositional logic

C.S. 430 Assignment 6, Sample Solutions

Econ 2110: Fall 2008 Suggested Solutions to Problem Set 8 questions or comments to Dan Fetter 1

Homework 3 Solutions

Section 8.3 Trigonometric Equations

Other Test Constructions: Likelihood Ratio & Bayes Tests

Math221: HW# 1 solutions

Example Sheet 3 Solutions

Derivation for Input of Factor Graph Representation

Matrices and Determinants

ST5224: Advanced Statistical Theory II

Every set of first-order formulas is equivalent to an independent set

Notes on the Open Economy

Bounding Nonsplitting Enumeration Degrees

ΠΤΥΧΙΑΚΗ/ ΙΠΛΩΜΑΤΙΚΗ ΕΡΓΑΣΙΑ

LECTURE 4 : ARMA PROCESSES

The challenges of non-stable predicates

ΕΛΛΗΝΙΚΗ ΔΗΜΟΚΡΑΤΙΑ ΠΑΝΕΠΙΣΤΗΜΙΟ ΚΡΗΤΗΣ. Ψηφιακή Οικονομία. Διάλεξη 7η: Consumer Behavior Mαρίνα Μπιτσάκη Τμήμα Επιστήμης Υπολογιστών

Inverse trigonometric functions & General Solution of Trigonometric Equations

ANSWERSHEET (TOPIC = DIFFERENTIAL CALCULUS) COLLECTION #2. h 0 h h 0 h h 0 ( ) g k = g 0 + g 1 + g g 2009 =?

ΚΥΠΡΙΑΚΗ ΕΤΑΙΡΕΙΑ ΠΛΗΡΟΦΟΡΙΚΗΣ CYPRUS COMPUTER SOCIETY ΠΑΓΚΥΠΡΙΟΣ ΜΑΘΗΤΙΚΟΣ ΔΙΑΓΩΝΙΣΜΟΣ ΠΛΗΡΟΦΟΡΙΚΗΣ 19/5/2007

ΚΥΠΡΙΑΚΗ ΕΤΑΙΡΕΙΑ ΠΛΗΡΟΦΟΡΙΚΗΣ CYPRUS COMPUTER SOCIETY ΠΑΓΚΥΠΡΙΟΣ ΜΑΘΗΤΙΚΟΣ ΔΙΑΓΩΝΙΣΜΟΣ ΠΛΗΡΟΦΟΡΙΚΗΣ 6/5/2006

Chapter 6: Systems of Linear Differential. be continuous functions on the interval

Numerical Analysis FMN011

Generalized Fibonacci-Like Polynomial and its. Determinantal Identities

TMA4115 Matematikk 3

Instruction Execution Times

Exercises 10. Find a fundamental matrix of the given system of equations. Also find the fundamental matrix Φ(t) satisfying Φ(0) = I. 1.

A Class of Orthohomological Triangles

4.6 Autoregressive Moving Average Model ARMA(1,1)

6.1. Dirac Equation. Hamiltonian. Dirac Eq.

Nowhere-zero flows Let be a digraph, Abelian group. A Γ-circulation in is a mapping : such that, where, and : tail in X, head in

w o = R 1 p. (1) R = p =. = 1

Fractional Colorings and Zykov Products of graphs

DERIVATION OF MILES EQUATION FOR AN APPLIED FORCE Revision C

DESIGN OF MACHINERY SOLUTION MANUAL h in h 4 0.

Tridiagonal matrices. Gérard MEURANT. October, 2008

Section 9.2 Polar Equations and Graphs

( y) Partial Differential Equations

Lecture 2: Dirac notation and a review of linear algebra Read Sakurai chapter 1, Baym chatper 3

CRASH COURSE IN PRECALCULUS

Απόκριση σε Μοναδιαία Ωστική Δύναμη (Unit Impulse) Απόκριση σε Δυνάμεις Αυθαίρετα Μεταβαλλόμενες με το Χρόνο. Απόστολος Σ.

A Note on Intuitionistic Fuzzy. Equivalence Relation

2. Let H 1 and H 2 be Hilbert spaces and let T : H 1 H 2 be a bounded linear operator. Prove that [T (H 1 )] = N (T ). (6p)

Srednicki Chapter 55

Estimators when the Correlation Coefficient. is Negative

forms This gives Remark 1. How to remember the above formulas: Substituting these into the equation we obtain with

ΚΥΠΡΙΑΚΟΣ ΣΥΝΔΕΣΜΟΣ ΠΛΗΡΟΦΟΡΙΚΗΣ CYPRUS COMPUTER SOCIETY 21 ος ΠΑΓΚΥΠΡΙΟΣ ΜΑΘΗΤΙΚΟΣ ΔΙΑΓΩΝΙΣΜΟΣ ΠΛΗΡΟΦΟΡΙΚΗΣ Δεύτερος Γύρος - 30 Μαρτίου 2011

Concrete Mathematics Exercises from 30 September 2016

Reminders: linear functions

8.323 Relativistic Quantum Field Theory I

2. THEORY OF EQUATIONS. PREVIOUS EAMCET Bits.

ECE Spring Prof. David R. Jackson ECE Dept. Notes 2

department listing department name αχχουντσ ϕανε βαλικτ δδσϕηασδδη σδηφγ ασκϕηλκ τεχηνιχαλ αλαν ϕουν διξ τεχηνιχαλ ϕοην µαριανι

Μηχανική Μάθηση Hypothesis Testing

Congruence Classes of Invertible Matrices of Order 3 over F 2

[1] P Q. Fig. 3.1

Mean bond enthalpy Standard enthalpy of formation Bond N H N N N N H O O O

Mean-Variance Analysis

Approximation of distance between locations on earth given by latitude and longitude

Phasor Diagram of an RC Circuit V R

Right Rear Door. Let's now finish the door hinge saga with the right rear door

k A = [k, k]( )[a 1, a 2 ] = [ka 1,ka 2 ] 4For the division of two intervals of confidence in R +

Estimation for ARMA Processes with Stable Noise. Matt Calder & Richard A. Davis Colorado State University

557: MATHEMATICAL STATISTICS II RESULTS FROM CLASSICAL HYPOTHESIS TESTING

( ) 2 and compare to M.

SCHOOL OF MATHEMATICAL SCIENCES G11LMA Linear Mathematics Examination Solutions

IIT JEE (2013) (Trigonomtery 1) Solutions

ΗΜΥ 220: ΣΗΜΑΤΑ ΚΑΙ ΣΥΣΤΗΜΑΤΑ Ι Ακαδημαϊκό έτος Εαρινό Εξάμηνο Κατ οίκον εργασία αρ. 2

6.3 Forecasting ARMA processes

The Probabilistic Method - Probabilistic Techniques. Lecture 7: The Janson Inequality

2 Lagrangian and Green functions in d dimensions

Homomorphism of Intuitionistic Fuzzy Groups

Examples of Cost and Production Functions

Solutions for Mathematical Physics 1 (Dated: April 19, 2015)

Transcript:

B Proos Proo o Theorem 2.2. By Shephard s lemma, we now that, to a rst order, the productvty shoc A wll change the prces o any ndustry who purchases nputs, ether drectly or ndrectly, rom ndustry d log p = ( = ) + Rewrte ths n matrx orm to get F d log w F α F + Ω d log p. (36) d log p = (I Ω) ( α d log w e ) = Ψ( α d log w e ), (37) where e s the th standard bass vector. Let the household s aggregate consumpton good to be the numerare, so that the household s deal prce ndex P c s always equal to one. Then we now that Combne ths wth the prevous expresson to get Note that, b Ψ = λ and b Ψ α = Λ. Hence, d log P c = b d log p = 0. (38) b Ψe + b Ψ α d log w = 0. (39) λ + Λ d log w = 0. (40) Now, note that From ths, we now that Λ = w L P c C. (4) d log Λ = d log w + d log L d log C. (42) 76

Substtute ths nto the prevous expresson to get λ + Λ d log Λ Λ d log L + d log C = 0, (43) where we use the act that Λ =. Rearrange ths to get the desred result.to get an explct characterzaton or d log Λ / n terms o structural parameters o the model, wthout loss o generalty, assume that each good s produced rom a dstnct prmary actor (ths can be acheved by relabellng the nput-output matrx). Now note that λ = b + µ ω λ, (44) w = α p y µ G (w /P c ), (45) λ = p y P c C, (46) p = A C (p,..., p N, w )µ, (47) P c = p c C =. (48) We can derentate these to get to our answer. Denote the elastcty o substtuton between and or the total cost uncton o ndustry by ρ. Then we can wrte l d λ = d b + ρ d ω µ λ d λ + ω µ ω µ λ d log µ, (49) d b = b b ( ) ρ c d log p d log p, (50) d ω = ω ω ( ) d log p d log p + ω α ( ) d log p d log w, (5) d log w = ω ( ) d log w d log p + d log p + d log y d log µ, (52) ρ L ρ L 77

d log y = λ The proo or the case wth marups s very smlar. d λ d log p + d log P cc, (53) Proo o Proposton 2.6. Tae all dervatves wth respect to A and normalze P c =. In ths case, the system reduces to λ d λ = 0, (54) d log w = d log p + d log y (55) d log y = d log p + d log C (56) d log p = ( = ) + α d log w + ω d log p, (57) d log C = w l C d log w + π ( d log λ + d log C ). (58) C We also now that b d log p = 0 by normalzaton. Hence, combnng the second and thrd relatonshp we get that d log w = d log C, (59) n other words, all wages move together there s no pecunary externalty on the actors n response to a shoc. Next, we now the prcng equaton s d log p = (I Ω) (α d log C e ). (60) Combne ths wth b d log p = 0 to get the desred result that d log C = b (I Ω) e, (6) ollowng the act that b (I Ω) α = or, wthout marups, the sum o actor payments would add up to GDP snce there are no pure prots. Proo o Proposton 3.. Frst, observe that d log p = ( = ) + ω d log p + α d log w. (62) Hence, d log p = e (I Ω) α d log w e (I Ω) e = d log w ψ. (63) 78

Snce, b d log p = d log w λ = 0, ths s d log p = λ ψ. (64) Ths completely characterzes the eect on prces, and the real wage s equal to output, we are done. However, n equlbrum, the real wage s not equal to output. To get the eect on output, we need to tae nto account how the labor share o ncome changes. Denote the labor share by Λ L. Usng Theorem 2.2, we have that d log C = λ Λ L d Λ L. (65) Hence, we need to characterze d Λ L. We now that d Λ L = α µ λ ( θ ) [ d log w d log p ( = ) ] + Whch smples to Now, note that d Λ L = α µ λ ( θ ) ( ψ ( = ) ) + d λ = b ( θ 0 ) d log p + ω µ ( θ ) ( d log p d log p ( = ) ) + Substtute n prces, and solve ths lnear system n d λ to get α µ d λ. (66) α µ d λ. (67) ω µ d λ. d λ m = b ( θ 0 )( λ ψ ) + ω µ λ ( θ )( ψ ψ ) + ω µ λ (θ ) ψ m. Ths can be smpled to d λ m = b ( θ 0 ) λ ψ m b ( θ 0 )ψ m ψ + ω µ ( θ )λ ψ ψ m,, ω µ ( θ )λ ψ ψ m (68) (69) 79

Whch can be urther smpled to Hence, m + ω µ λ (θ )ψ m. (70) d λ m = λ m λ ( θ 0 ) b ( θ 0 )ψ m ψ + ω µ ( θ )λ ψ ψ m,, ω µ ( θ )λ ψ ψ m + λ (θ ) [ ψ m (m = ) ]. (7) = (θ 0 )cov b ( Ψ (), Ψ (m) ) + (θ )λ cov ( Ψ µ Ω (), Ψ (m) ) + (θ )λ ω ψ µ ω ψ m, ω µ (θ )λ ψ ψ m + λ (θ )(ψ m (m = )), = (θ 0 )cov b ( Ψ (), Ψ (m) ) + (θ )λ cov ( Ψ µ Ω (), Ψ (m) ). (72) α m µ m d λ m = (θ 0 )cov b ( Ψ (), Puttng ths altogether, we get d Λ L = α µ λ ( θ ) ( ψ ( = ) ) + (θ 0 )cov b ( Ψ (), α m µ m Ψ (m) ) + m m α m µ m Ψ (m) )+ To nsh the proo, note that m µ m α m Ψ m = Ψ L. Then d Λ L = α µ λ ( θ ) ( ψ ( = ) ) + (θ 0 )cov b ( Ψ (), α m µ m Ψ (m) ) + m (θ )λ cov ( Ψ µ Ω (), (θ )λ cov ( Ψ µ Ω (), m m α m µ m Ψ (m) ). (73) α m µ m Ψ (m) ). (74) (θ )λ cov ( Ψ µ Ω (), Ψ (L) ), (75) 80

= (θ 0 )cov b ( Ψ (), α m µ m Ψ (m) ) + m (θ )λ cov ( Ψ µ [Ω,a ] (), Ψ (L) ), (76) where note that Ψ LL =. Proo o Proposton 3.2. From Theorem 2.2, we now that d log C/ d log µ = λ Λ L d λ L. (77) For a quantty m, we have d λ m = b ( θ 0 )(d log p ) + Ω µ ( θ )λ [d log p d log p ] ( = )θ Ψ m. Note that Hence, Snce, b d log p = d log w + λ = 0, ths s (78) d log p = ( = ) + ω d log p + α d log w. (79) d log p = d log w + ψ. (80) d log p = λ + ψ. (8) Substtutng ths bac nto (78) and set m = L to get d λ L = b ( θ 0 )( ψ λ )ψ L + λ ( θ )µ Rearrange ths to get Ω [ ψ ψ ]Ψ L θ µ λ d λ L =( θ 0 )Cov b ( Ψ (), Ψ (L) ) + λ ( θ )µ Cov Ω ()( Ψ (), Ψ (L) ) µ ( θ )λ Ω Ψ L θ µ λ Ω Ψ L. Ω Ψ L. (82) Combne ths wth (77). 8

Proo o Proposton 3.3. By Shephard s lemma, d log p = ( = ) + Ω d log p + α d log w. (83) Invert ths system to get d log p = Ψe + Ψ d log w, (84) where Ψ = (I Ω) α s a N K matrx o networ-adusted actor ntenstes by ndustry. Now consder a actor L, we have d Λ L = b ( θ 0 )[ Ψ + Ψ d log w ]Ψ L, Smply ths to + ( θ )λ µ Ω [ Ψ + Ψ d log w + Ψ Ψ d log w ]Ψ L + (θ )λ µ Ω Ψ L. d Λ L =(θ 0 ) b Ψ Ψ L b Ψ L Ψ d log w, + (θ )λ µ Ω Ψ Ψ L Ω Ψ Ψ L ( ) + ( θ )λ µ Ω Ψ Ψ d log w Ψ L + (θ )λ µ Ω Ψ L, =(θ 0 ) b Ψ Ψ L b Ψ L Ψ d log w, + (θ )λ µ Ω Ψ Ψ L Ω Ψ L Ω Ψ + ( θ )λ µ ( ) Ω Ψ Ψ d log w Ψ L, 82

=(θ 0 ) b Ψ Ψ L b Ψ L Ψ d log w, + (θ )λ µ Cov Ω ()( Ψ (), Ψ (L) ) + ( θ )λ µ Ω Ψ Ψ L Ω Ψ L Ω Ψ d log w, =(θ 0 ) b Ψ Ψ L b Ψ L Ψ d log w, + (θ )λ µ Cov Ω ()( Ψ (), Ψ (L) ) + ( θ )λ µ Cov Ω ()( Ψ ( ), Ψ (L) ) d log w, =(θ 0 ) b Ψ Ψ L b Ψ L Ψ d log w, + (θ )λ µ Cov Ω ()( Ψ () Ψ ( ) d log w, Ψ (L) ) =(θ 0 ) b Ψ L Ψ Ψ d log w, + (θ )λ µ Cov Ω ()( Ψ () Ψ ( ) d log w, Ψ (L) ) =(θ 0 )Cov b ( Ψ () Ψ ( ) d log w, Ψ (L) ) + (θ 0 )( λ Λ d log w )λ L, + (θ )λ µ Cov Ω ()( Ψ () Ψ ( ) d log w, Ψ (L) ). Hence, or a productvty shoc, lettng Λ L be demand or actor L, and ndexng all actors by, we have d λ L =(θ 0 )Cov b Ψ () Ψ ( ) d log w, Ψ (L) + (θ )µ λ Cov [Ω,α ] Ψ () Ψ ( ) d log w, Ψ (L) 83

+ (θ 0 ) λ λ d log w λ L. (85) Combne ths wth the observaton that λ L (d log w L + d log L L d log C) = d λ L. (86) Fnally, we now that d log C = λ + λ d log λ. (87) Set d L L = 0, and we have a lnear system wth F + equatons and F + unnowns where F s the total number o actors. Substtute bac nto networ ormula to get λ d λ + d log C = d log w (88) Λ L d log λ L =(θ 0 )Cov b Ψ () Ψ ( ) d log λ, Ψ (L) + (θ )µ λ Cov [Ω,α ] Ψ () Ψ ( ) d log λ, Ψ (L) + (θ 0 ) λ λ d log λ d log C λ L. (89) Use Theorem 2.2 to urther smply ths λ L d log λ L =(θ 0 )Cov b Ψ () Ψ ( ) d log λ, Ψ (L) + (θ )µ λ Cov [Ω,α ] Ψ () Ψ ( ) d log λ, Ψ (L). (90) 84

Proo o Proposton 3.4. From Theorem 2.2, we now that d log C/ d log µ = λ Λ L d λ L. (9) For a marup shoc, we can derentate demand or a quantty m to get d λ m = b ( θ 0 )(d log p ) + Ω µ ( θ )λ [d log p d log p ] ( = )θ Ψ m. By Shephard s lemma, (92) d log p = ( = ) + Ω d log p + α d log w. (93) Invert ths system to get d log p = Ψe + Ψ d log w, (94) where Ψ = (I Ω) α s a N K matrx o networ-adusted actor ntenstes by ndustry. Substtutng ths bac nto (92) and set m = L to get d Λ L = b ( θ 0 )[ Ψ + Ψ d log w ]Ψ L, + ( θ )λ µ Ω [ Ψ + Ψ d log w Ψ Ψ d log w ]Ψ L θ µ λ Ω Ψ L. Smply ths to d Λ L =( θ 0 ) b Ψ Ψ L + ( θ 0 ) b Ψ L Ψ d log w, + ( θ )λ µ Ω Ψ Ψ L Ω Ψ L Ψ + ( θ )λ µ Ω Ψ d log w Ψ L Ω Ψ L θ λ µ Ψ L, Ψ d log w 85

=( θ 0 ) ( ) Cov b ( Ψ (), Ψ (L) ) + λ Λ L ( ) + ( θ )λ µ CovΩ ()( Ψ (), Ψ (L) ) ( = )Ψ L + ( θ 0 ) Cov b Ψ ( ) d log w, Ψ (L) + Λ d log w Λ L + ( θ )λ µ Cov Ω Ψ () ( ) d log w, Ψ (L) ( = )Ψ L θ λ µ Ψ L, =( θ 0 ) Cov b( Ψ () + Ψ ( ) d log w, Ψ (L) ) + λ Λ L + ( θ )λ µ Cov Ω ()( Ψ () + Ψ ( ) d log w, Ψ (L) ) ( = )Ψ L + ( θ 0 ) λ + Λ d log w Λ L λ µ Ψ L, =( θ 0 ) Cov b( Ψ () + Ψ ( ) d log w, Ψ (L) ) + λ Λ L + ( θ )λ µ Cov Ω ()( Ψ () + Ψ ( ) d log w, Ψ (L) ) λ Ψ L. The nal lne ollows rom the act that ( λ + ) Λ d log w = b d log p = d log P c = 0. Fnally, substtute d log w = d log Λ + d log C nto the expresson above to get d Λ L =( θ 0 ) Cov b( Ψ () + Ψ ( ) (d log Λ + d log C), Ψ (L) ) + λ Λ L + ( θ )λ µ Cov Ω ()( Ψ () + Ψ ( ) (d log Λ + d log C), Ψ (L) ) λ Ψ L. 86

To complete the proo, note that d log C Ψ ( ) = d log C. (95) In other words, ths s a vector o all ones multpled by the scalar d log C, and hence t drops out o the covarance operators, snce the covarance o a vector o ones wth any other vector under any probablty dstrbuton s always equal to zero. Hence, d Λ L =( θ 0 ) Cov b( Ψ () + Ψ ( ) d log Λ, Ψ (L) ) + λ Λ L + ( θ )λ µ Cov Ω ()( Ψ () + Ψ ( ) d log Λ, Ψ (L) ) λ Ψ L. Ths can be combned wth Theorem 2.3 to complete the proo. Proo o Example 3.6. We start wth the matrces lsted n (25) and (26), so that ( θ 0 )Cov b ( Ψ (L), Ψ (L) ) ( θ 0 )Cov b ( Ψ (K), Ψ (L) ) Γ = ( θ 0 )Cov b ( Ψ (K), Ψ (L) ) ( θ 0 )Cov b ( Ψ (K), Ψ. (96) (K) δ () = ( θ 0 )Cov b ( Ψ (), Ψ (L) ) ( θ 0 )Cov b ( Ψ (), Ψ (K). (97) From the structure o the problem, we can explctly wrte the value o Γ as ollows: Γ = ( θ 0 ) b 3 (b µ + b 2 µ ) 2 b 3(b µ + b 2 µ ) 2 b 3 ( b 3 )µ 3 b 3 ( b 3 )µ 3. (98) Frst we loo at the cases or =, 2 (the case where a actor o producton s shared). We note that the case s symmetrc or =, 2 by the structure o the networ and the problem. For these values, we have that δ () = ( θ 0 ) b (µ (b µ + b 2 µ 2 )) b b 3 µ 3. (99) 87

Generally (or all cases), n order to solve the system n equaton (24), wrte (Λ Γ) = Invert ths to get (Λ Γ) = det Γ The determnant s (b µ + b 2 µ )( 2 0)b 3 ) ( θ 0 )b 3 (µ b + b 2 µ ) 2 ( θ 0 )b 3 ( b 3 )µ 3 b 3 µ ( 3 0)( b 3 )) b 3 µ ( 3 0)( b 3 )) ( θ 0 )b 3 (µ b + b 2 µ ) 2 ( θ 0 )b 3 ( b 3 )µ 3 (b µ + b 2 µ )( 2 0)b 3 ) det Γ =(b µ + b 2 µ 2 )( ( θ 0)b 3 )b 3 µ 3 ( ( θ 0)( b 3 )) ( θ 0 ) 2 b 2 3 ( b 3)µ 3 (µ b + b 2 µ 2 )), = Λ L Λ K (( ( θ 0 )b 3 )( ( θ 0 )( b 3 )) ( θ 0 ) 2 b 3 ( b 3 )). (00). (0) = Λ L Λ K ( ( θ 0 )b 3 ( θ 0 )( b 3 ) + ( θ 0 ) 2 b 3 ( b 3 ) ( θ 0 ) 2 b 3 ( b 3 )) = Λ L Λ K ( ( θ 0 )(b 3 + b 3 )) = Λ L Λ K θ 0. Plug ths bac nto (0) and smply, (Λ Γ) Λ K ( ( θ 0 )( b 3 )) ( θ 0 )b 3 Λ L = Λ L Λ K θ 0 Λ K ( θ 0 )( b 3 ) Λ L ( ( θ 0 )b 3 ). (02) Returnng to the specc case where =, (Λ Γ) δ () = ( ( θ 0 )( b 3 )) (θ 0 )b 3 θ 0 Λ L Λ K θ 0 ( θ 0)( b 3 ) ( θ 0 )b 3 θ 0 Λ L Λ K θ 0 b (θ 0 )(µ Λ L ) b (θ 0 )Λ K. (03) Multplyng the values n (03), and usng the dentty n (24), d log Λ =(Λ Γ) δ (), = b (θ 0 ) θ 0 ( µ Λ L )θ 0 + b 3 ( θ 0 ) µ Λ L + b 3 (θ 0 ) b 3 (θ 0 ) (θ 0 )( b 3 )( µ Λ L ) ( ( θ 0 )b 3 ). 88

Combne ths wth (3) gves ( θ 0 [ µ d log C = b ( b 3 )b ] ) θ 0 + b 3 ( θ 0 ) µ θ 0 Λ L Λ L [ θ 0 b 3 b (θ 0 )( b 3 ) ( µ θ 0 whch urther smples to Λ L ) ( ( θ 0 )b 3 ), d log C ] = b + b (θ 0 ) [ ( b 3 ) µ. (04) d log A Λ L Ths gves the desred result or rm. Note, as mentoned beore, a symmetrc result holds or rm 2. In the case o rm 3, we have that δ (3) = (θ 0 ) b 3 (b µ + b 2 µ 2 ) b 3 µ 3 ( b 3) ].. (05) The results rom (02) gve the blueprnt or solvng or the value o d log C d log A 3 ths, we can conclude that d log C = b 3 d log A 3 b + b 2 b 3 whch urther smples to T ( ( θ 0 )( b 3 )) θ 0 Λ L (θ 0 )b 3 Λ K θ 0 ( θ 0)( b 3 ) θ 0 Λ L ( θ 0 )b 3 Λ K θ 0 b 3 (b µ + b 2 µ 2 ) b 3 µ 3 ( b 3) as well. From, (06) = b 3 b 3 b 3 T [ b 3 ( θ 0 ) θ 0 ( θ0 )( b 3 ) ] + (θ 0 ) 2 θ 0 b 3 ( b 3 ) (θ 0 ) 2 θ 0 b 3 ( b 3 ) + [ ( θ 0 )( b 3 ) ] ( b 3 )(θ 0 ) θ 0. (07) Ths smples to gve the requred result d log C d log A 3 = b 3. (08) 89

Proo o Proposton 5.. The ndustry level prce s then gven by p = ( M M ( c(w, p) µ m = c(w, p) µ ( (m E A ) ε Φ(z, m ) d m d z A z ) ε. z ) ε ) ε, By Shephard s lemma, we now that, d log p = ( = ) + F d log w F α F + Ω d log p, (09) where α F and Ω are rm-level expendtures on actor F and ndustry as a share o costs (excludng entry costs). Rewrte ths n matrx orm to get d log p = (I Ω) ( α d log w e ) = Ψ( α d log w e ), (0) where e s the th standard bass vector. Let the household s aggregate consumpton good to be the numerare, so that the household s deal prce ndex P c s always equal to one. Then we now that Combne ths wth the prevous expresson to get Note that, b Ψ = λ and b Ψ α = Λ. Hence, d log P c = b d log p = 0. () b Ψe + b Ψ α d log w = 0. (2) λ + Λ d log w = 0. (3) Now, note that Λ = w L F P c C. (4) 90

From ths, we now that d log Λ = d log w + d log L d log C. (5) Substtute ths nto the prevous expresson to get λ + Λ d log Λ Λ d log L + d log C = 0, (6) where we use the act that Λ =. Proo o Proposton 5.2. Let the cost uncton o entrant n ndustry producng y unts be gven by c(w, p)h(y /A), (7) where A s ndustry level shoc and subscrpts have been suppressed. Free entry then mples that µc(w, p)h (y/a) y c(w, p)h(y/a) = c(w, p), (8) A where s the entry cost n unts o the nput good. We can smply ths to µh (y/a) y h(y/a) =. (9) A Ths equaton pns down the ecent scale o operaton y = Ay (, µ). The cost o producng q unts o ndustry output s then gven by nc(w, p) (y (µ, )) + n c(w, p), (20) such that ny = q. Substtute the constrant nto ths to get C(w, p) q A = Hence the ndustry s cost uncton s lnear q/a as needed. Fnally, q y A c(w, p) (y (µ, )) + q c(w, p). (2) Ay C(w, p)q/a = q c(w, p) (y (µ, )) + q c(w, p) = nl w y A w Ay + n l w. (22) 9

Thereore, the ndustry level cost uncton obeys Shephard s Lemma and we can replcate the rest o the proo rom Proposton 5. Proo o Proposton 5.3. ( ) d log p = ( = ) d log A + s (z) α (z) d F(z) d log w z ( ) ( ) + s (z) ω (z) d F(z) d log p + s (z) β (z) d log r (z) d F(z) z ε s (z) (z ) d log z. where s (z) s rm s share o sales n ndustry, and β (z) s the cost share o rm on ts cttous xed actor. We can solve ths system to get z d log p = (I Ω) ( e d log A + α d log w + ξ κ ), (23) where ξ = ( s z (z) β (z) d log r (z) d F(z) ) and κ = ε s (z) (z ) d log z, and the elements o Ω and α are dened approprately. Let output be the numerare so that b d log p = 0. (24) Hence, λ + Λ d log w + λ ξ λ κ = 0. (25) For each actor, we now that d log w = d log Λ + d log C. (26) Substtute ths n to get λ + Λ d log Λ + Λ d log C + λ z s (z) β (z) d log Λ r (z) d F(z) + Fnally, observe that Λ + λ z s (z) β (z) =. The proo or marup shocs s very smlar. z λ s (z) β (z) d log C λ κ = 0. Proposton B.. Consder an economy wth a cash-n-advance constrant, and nomnal rgdtes. 92

Then, where and d log C Λ d log L = λ d log A λ e s d log µ + d H( Λ, Λ), d log µ = (e s Ψe s ) e s Ψ(d log A α d log w), (27) d log w = d log Λ + d log M d log L. In the specal case where some racton δ n ndustry are lexble. Then, λ (s) d log µ = ( b (I Ω) b (I δ Ω) δ ) d log A ( b (I δω) δα) d log w, (28) = ( λ b (I δ Ω) δ ) d log A ( b (I δω) δα) d log w. Proo o Proposton B.. Order the producers so that the rst s producers are the ones wth stcy prces. For a vector x, denote x (s) = e sx. From the cash n advance constrant, we now that d log C = d log M d log P c, = d log M b d log p, = d log M b Ψ ( α d log w d log A ) + b Ψe s d log µ, = d log M Λ d log w + λ d log A λe s d log µ, = d log M Λ ( d log Λ + d log M d log L ) + λ d log A λe s d log µ, = d log M Λ ( d log Λ d log L ) d log M + λ d log A λe s d log µ, whch mples that d log C Λ d log L = λ d log A λ (s) d log µ Λ d log Λ. (29) To get the marups necessary to eep p (s) stcy, we mpose d log p (s) = d log µ + e s Ω log p d log A (s) = 0. (30) Ths mples d log µ = e s Ω d log p α (s) d log w + d log A (s). (3) 93

On the other hand, we have d log p = Ψ ( α d log w d log A ) + Ψe s d log µ. (32) Substtutng ths bac nto the prevous expresson gves d log µ = e s Ω Ψ α d log w e s Ω Ψe s d log µ + e s Ω Ψ d log A e s α d log w + e s d log A. (33) Solve ths to get d log µ = (e s(i + Ω Ψ)e s ) e s(i + Ω Ψ) ( d log A d log w ), = (e s Ψe s ) e s Ψ ( d log A d log w ). Proo o Proposton 4.. The labor-lesure condton and the cash-n-advance condton mply that Hence, or Thereore, L /(ν) = ( ) ( w w =. (34) P c C M) d log L = d log w d log M = d log Λ d log L + d log M d log M, (35) ν d log L = To nsh, apply Propostons 3. and 3.2. ν d log Λ. (36) + ν d log w = d log Λ + d log M (37) ν + Proo o Proposton 4.2. Wth log utlty n consumpton and nnte Frsch elastcty o labor supply we have that d log w d log P c = d log C, (38) or n other words, substtutng n the cash n advance constrant d log w = d log M. (39) 94

Furthermore, the cash n advance constrant mples that d log C = d log M d log P c, = d log M Λ d log w + λ d log A λ e s d log µ, = d log M d log M + λ d log A λ e s d log µ, = λ d log A λ e s d log µ, substtutng equaton (27) rom Proposton B. completes the proo = λ d log A λ (s) (e s Ψe s ) e s Ψ(d log A α d log w), = λ d log A λ (s) (e s Ψe s ) e s Ψ(d log A α d log M). In the specal case where some racton δ n ndustry are lexble. Then, λ (s) d log µ = ( b (I Ω) b (I δ Ω) δ ) d log A ( b (I δω) δα) d log w, (40) = ( λ b (I δ Ω) δ ) d log A ( b (I δω) δα) d log w. At the ndustry level, equaton (40) shows that the changes n marups can be nterpreted as some racton o the rms n each ndustry change ther marup n response to shocs. Proo o Proposton 5.4. On the other hand, On the other hand, d log A ( ) e ( = s = s µ ) = s µ 0. (4) d τ e d 2 2 log A ( = (σ )Var d τ 2 s (µ ) s µ )2 = (σ 2)Var s (µ ) s µ. (42) Consder an ndustry where: all rms use the same upstream nput bundle wth cost C; rms transorm ths nput nto a rm-specc varety o output usng constant return to returns to scale technology; each rm has productvty a and charges a marup µ ; the varetes are combned nto a composte good by a compettve downstream ndustry accordng to a CES producton uncton wth elastcty σ on rm. 95

We denote the quantty o composte good produced as Frm charges a prce Q = [ b σ The resultng demand or rm s varety s where the prce ndex s gven by Total prots are gven by p σ σ ] σ σ. (43) p = µ a C. (44) q = ( p P ) σ b Q, (45) P = [ ] b p σ σ. (46) Π = (p C)( p P ) σ b Q. (47) We solve out the prce ndex and prots explctly and get P = b ( µ a ) σ σ C, (48) σ ( µ Π = ) a a µ a [ ( ] ) σ σ µ b a b CQ. (49) For completeness we can also solve or the sales o each rm as a racton o the sales o the ndustry λ = p q PQ = b ( µ a ) σ b ( µ a ) σ. (50) We want to understand how to aggregate ths ndustry nto homogenous ndustry 96

wth productvty A and marup µ. These varables must satsy Π = P = µ C, (5) A ( µ A A) CQ. (52) Ths mples that A and µ are the solutons o the ollowng system o equatons µ A = b ( µ a ) σ σ, (53) σ ( µ A A) ( µ = ) a a µ a [ ( ] ) σ σ µ b a b. (54) The soluton s A = [ ( ] ) σ σ µ ( b a µ ) µ a ( ) σ µ σ b a σ µ a b, (55) µ = [ ( µ ) σ] σ b a [ ( ] ) σ σ µ ( b a µ ) µ a ( ) σ µ σ b a σ µ a b. (56) We can also rewrte ths n a useul way as A = [ ( ] ) σ σ µ b a µ ( µ a ) σ b b ( µ a ) σ = [ ( ] ) σ σ µ b a, (57) µ λ 97

µ = µ ( µ a ) σ b b ( µ a ) σ =. (58) µ λ Dene the ecency o each rm to be e = µ. Then, consder a steady state where s µ = s e =. Consder a transormaton e = τ + ( τ)e. Ths transormaton eeps µ constant. On the other hand, On the other hand, d log A ( ) e ( = s = s µ ) = s µ 0. (59) d τ e d 2 2 log A ( = (σ )Var d τ 2 s (µ ) s µ )2 = (σ 2)Var s (µ ) s µ. (60) C Relabellng As mentoned prevously, our results can be appled to any nested CES economy, wth any arbtrary pattern o nested substtutabltes and complementartes among ntermedate nputs and actors. For concreteness, we descrbe the relabelng or one specc example. Let the household s consumpton aggregator be C C = b ( c c ) σ σ σ σ. (6) Suppose each producer produces usng a CES aggregator o value-added VA and ntermedate nputs X: y y = A α ( VA VA ) θ θ + ( α ) ( X X θ ) θ θ θ. 98

Value-added conssts o derent actors VA VA = F l ν = where l s actor o type used by. Intermedate nput conssts o nputs rom other producers X X = N = l η η η η ε ( ) ε ε x ε ω x. Denote the matrx o ν usng V and the matrx o ω as Ω. We rewrte ths economy n the standard orm we requre, and put hats on the new expendture shares. The new nput-output matrx ˆΩ has dmenson (+3N+F) (+3N+F). The rst ndustry s the household s consumpton aggregator, the next N ndustres are the orgnal ndustres, the next N ndustres produce the value-added o the orgnal ndustres, the next N ndustres produce the ntermedate nputs o the orgnal ndustres, and the nal F ndustres correspond to the actors. Under the relabelng, we have ˆb = (, 0,..., 0),, ˆΩ = b 0 0 0 0 0 0 dag(a) dag( a) 0 0 0 0 0 V Ω 0 0 0 0 0 0 0 0 0, (62) wth and ˆθ = (σ, θ, ν, ε, ), (63) ˆµ = (, µ,,..., ). (64) 99

D Aggregaton o cost-based Domar weghts In ths Appendx we show that recoverng cost-based Domar weghts rom aggregated data s, n prncple, not possble. We also show how the Basu-Fernald decomposton can detect changes n msallocaton even n acyclc economes (where the equlbrum s ecent, and there s no possblty o reallocaton o resources). Example 2. also shows the alure o the aggregaton property mpled by Hulten s theorem. The easest way to see ths s to consder aggregatng the nput-output table or the economy n Example 2.. For smplcty, suppose that marups are the same everywhere so that µ = µ or all. Snce there s no possblty o reallocaton n ths economy, and snce marups are unorm, ths s our best chance o dervng an aggregaton result, but even n ths smplest example, such a result does not exst. Suppose that we aggregate the whole economy S = {,..., N}. Then, n aggregate, the economy conssts o a sngle ndustry that uses labor and nputs rom tsel to produce. In ths case, the nput-output matrx s a scalar, and equal to the ntermedate nput share o the economy Ω SS = µ N µ N, (65) and the aggregate marup or the economy s gven by µ. Thereore, λ S constructed usng aggregate data s λ S = (I µω) = µ N µ. (66) µ However, we now rom the example that d log C d log A = S λ = N λ S = µ N µ, (67) µ except n the lmtng case wthout dstortons µ. Thereore, even n ths smplest case, wth homogenous marups and no reallocaton, aggregated nput-output data cannot be used to compute the mpact o an aggregated shoc. To compare our decomposton wth that o Basu and Fernald (2002), consder the acyclc economy n Fgure 3 wth two actors L and L 2. Now, we now that d log C d log A 2 = λ 2 = ( α ). (68) 00

The Basu-Fernald decomposton or ths economy gves d log C d log A 2 = λ 2 + R M + µ v R M = α µ + µ v R M, (69) where the rst term (n ths case, the sales-share o 2) s the pure technology eect, and R M s the reallocaton o ntermedate nputs, even though n ths economy, there s no capacty or reallocatng resources and the equlbrum s ecent. L L 2 HH 2 Fgure 3: Acyclc economy where the sold arrows represent the low o goods. The low o prots and wages rom rms to households has been suppressed n the dagram. The two actors n ths economy are L and L 2. E Extra Examples Example E.. Next, consder the mnmal example wth two elastctes o substtuton, whch demonstrates the prncple that changes n msallocaton are drven by how each node swtches ts demand across ts supply chan n response to a shoc. To ths end, we apply Proposton 3. to the economy depcted n Fgure 4. d log C d log A 3 = λ 3 Λ L ( (θ0 ) [ b (ω 3 µ (ω 3µ 3 + ω 4 µ +(θ )µ λ [ ω3 µ 3 ω 3 ( ω3 µ 3 + ω 4 µ 4 4 )) ω 3b Λ L ] The term multplyng (θ 0 ) captures how the household wll sht ther demand across and 2 n response to the productvty shoc, and the relatve degrees o msallocaton n and 2 s supply chans. The term multplyng (θ ) taes nto account how wll sht ts demand across 3 and 4 and the relatve amount o msallocaton o labor between 3 and 0 )]).

L 3 4 2 HH Fgure 4: An economy wth two elastctes o substtuton. 4. Not surprsngly, nstead we shoc ndustry, then only the household s elastcty o substtuton matters, snce ndustry wll not sht ts demand across ts nputs n response to the shoc to ndustry 2: d log C d log A =b Λ L (θ 0 ) [ b µ (ω 3µ 3 + ω 4 µ 4 ) b Λ L ]. Ths llustrates the general prncple n Proposton 3. that an elastcty o substtuton θ matters only s somewhere downstream rom. 02