Trade Adjustment and Productivity in Large Crises. Online Appendix May 2013. Appendix A: Derivation of Equations for Productivity



Similar documents
An Alternative Way to Measure Private Equity Performance

Answer: A). There is a flatter IS curve in the high MPC economy. Original LM LM after increase in M. IS curve for low MPC economy

Addendum to: Importing Skill-Biased Technology

How To Calculate The Accountng Perod Of Nequalty

Problem Set 3. a) We are asked how people will react, if the interest rate i on bonds is negative.

SPEE Recommended Evaluation Practice #6 Definition of Decline Curve Parameters Background:

Lecture 3: Force of Interest, Real Interest Rate, Annuity

Section 5.3 Annuities, Future Value, and Sinking Funds

Support Vector Machines

17 Capital tax competition

Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic

On the Optimal Control of a Cascade of Hydro-Electric Power Stations

Elements of Advanced International Trade 1

Chapter 7: Answers to Questions and Problems

Technical Memorandum Number 815. Bigger Slice or Larger Pie? Optimal Marketing Strategies for New Firms. John Angelis Moren Lévesque

The Development of Web Log Mining Based on Improve-K-Means Clustering Analysis

The OC Curve of Attribute Acceptance Plans

PRIVATE SCHOOL CHOICE: THE EFFECTS OF RELIGIOUS AFFILIATION AND PARTICIPATION

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ).

Causal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting

Chapter 15: Debt and Taxes

Depreciation of Business R&D Capital

Modelling the Cocaine and Heroin Markets in the Era of Globalization and Drug Reduction Policies Claudia Costa Storti and Paul De Grauwe

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek

Using Series to Analyze Financial Situations: Present Value

IS-LM Model 1 C' dy = di

Chapter 4 ECONOMIC DISPATCH AND UNIT COMMITMENT

Forecasting the Direction and Strength of Stock Market Movement

Sector-Specific Technical Change

The impact of hard discount control mechanism on the discount volatility of UK closed-end funds

Management Quality, Financial and Investment Policies, and. Asymmetric Information

Can Auto Liability Insurance Purchases Signal Risk Attitude?

A hybrid global optimization algorithm based on parallel chaos optimization and outlook algorithm

DEFINING %COMPLETE IN MICROSOFT PROJECT

7.5. Present Value of an Annuity. Investigate

A Probabilistic Theory of Coherence

HOUSEHOLDS DEBT BURDEN: AN ANALYSIS BASED ON MICROECONOMIC DATA*

Staff Paper. Farm Savings Accounts: Examining Income Variability, Eligibility, and Benefits. Brent Gloy, Eddy LaDue, and Charles Cuykendall

IMPACT ANALYSIS OF A CELLULAR PHONE

MATHEMATICAL ENGINEERING TECHNICAL REPORTS. Sequential Optimizing Investing Strategy with Neural Networks

Number of Levels Cumulative Annual operating Income per year construction costs costs ($) ($) ($) 1 600,000 35, , ,200,000 60, ,000

Section 5.4 Annuities, Present Value, and Amortization

Oligopoly Theory Made Simple

STAMP DUTY ON SHARES AND ITS EFFECT ON SHARE PRICES

SUPPLIER FINANCING AND STOCK MANAGEMENT. A JOINT VIEW.

WORKING PAPERS. The Impact of Technological Change and Lifestyles on the Energy Demand of Households

Leveraged Firms, Patent Licensing, and Limited Liability

BERNSTEIN POLYNOMIALS

Calculating the high frequency transmission line parameters of power cables

The Current Employment Statistics (CES) survey,

When Talk is Free : The Effect of Tariff Structure on Usage under Two- and Three-Part Tariffs

Bankruptcy and Aggregate Productivity

The Personalization Services Firm: What to Sell, Whom to Sell to and For How Much? *

Solution: Let i = 10% and d = 5%. By definition, the respective forces of interest on funds A and B are. i 1 + it. S A (t) = d (1 dt) 2 1. = d 1 dt.

Vasicek s Model of Distribution of Losses in a Large, Homogeneous Portfolio

Mean Molecular Weight

Activity Scheduling for Cost-Time Investment Optimization in Project Management

High Correlation between Net Promoter Score and the Development of Consumers' Willingness to Pay (Empirical Evidence from European Mobile Markets)

A Model of Private Equity Fund Compensation

How To Compare Frm To An Isac

The Cross Section of Foreign Currency Risk Premia and Consumption Growth Risk

An Empirical Study of Search Engine Advertising Effectiveness

Modelling the World Oil Market Assessment of a Quarterly Econometric Model

On the allocation of resources for secondary education schools

Kiel Institute for World Economics Duesternbrooker Weg Kiel (Germany) Kiel Working Paper No. 1120

Results from the Dixit/Stiglitz monopolistic competition model

Asia-Pacific Research and Training Network on Trade. Working Paper Series, No. 81, July Truong P. Truong

Is Thailand s Fiscal System Pro-Poor?: Looking from Income and Expenditure Components. Hyun Hwa Son

Recurrence. 1 Definitions and main statements

Internet companies extensively use the practice of drop-shipping, where the wholesaler stocks and owns the

WORKING PAPER. C.D. Howe Institute. The Effects of Tax Rate Changes on Tax Bases and the Marginal Cost of Public Funds for Provincial Governments

Logistic Regression. Lecture 4: More classifiers and classes. Logistic regression. Adaboost. Optimization. Multiple class classification

Student Performance in Online Quizzes as a Function of Time in Undergraduate Financial Management Courses

) of the Cell class is created containing information about events associated with the cell. Events are added to the Cell instance

Structural Estimation of Variety Gains from Trade Integration in a Heterogeneous Firms Framework

Multiple-Period Attribution: Residuals and Compounding

Why Do Cities Matter? Local Growth and Aggregate Growth

Two Faces of Intra-Industry Information Transfers: Evidence from Management Earnings and Revenue Forecasts

A Master Time Value of Money Formula. Floyd Vest

Time Value of Money Module

ECONOMICS OF PLANT ENERGY SAVINGS PROJECTS IN A CHANGING MARKET Douglas C White Emerson Process Management

Price Competition in an Oligopoly Market with Multiple IaaS Cloud Providers

Heterogeneous Paths Through College: Detailed Patterns and Relationships with Graduation and Earnings

Lecture 3: Annuity. Study annuities whose payments form a geometric progression or a arithmetic progression.

Substitution Effects in Supply Chains with Asymmetric Information Distribution and Upstream Competition

Transcription:

Trade Adjustment Productvty n Large Crses Gta Gopnath Department of Economcs Harvard Unversty NBER Brent Neman Booth School of Busness Unversty of Chcago NBER Onlne Appendx May 2013 Appendx A: Dervaton of Equatons for Productvty Dervaton of Equaton 8 The dervaton s very smlar to Basu Fernald 2002 wth the dstncton that we have labor that s used n fxed costs. The producton functon for each frm s gven by equaton 3. Snce frms are prce takers n the prmary factor ntermedate nput markets set prces as a constant markup 1/ over margnal cost we have: Y p w L p p Y r K p Y P X X. To measure the growth rate of value added we use the conventon dvsa ndex formula: ln Y V A ln Y s Y X ln X 1 s Y X ln Y sy X ln X 1 s Y ln Q X where s Y X s the revenue share of ntermedates sy X P X X P Y whch s equal to the constant µ. We can then wrte: Y Y K + Y L p K L + Y X + Y A p X A 1 s Y X 1 s Y X ln Y s k ln K + ω L ps L ln L p + sy X ln X + F A A Y P ln A. Ths follows from the relaton s k sy k ω L 1 s Y p L p X L. Rearrangng we get: ln Y V A ln Y sq X 1 s Q X 1 µ ln X ln Y 1 µ [s k ln K + w L ps L ln L p ] + µ [ ] 1 µ 1 µ 1 µ 1 ln X ln Y + F A A 1 µ Y p ln A. 1

Fnally defne the welfare relevant frm-level productvty usng the modfed Solow Resdual: ln PR t ln Y V A s k ln K s l ln L ln Y V A s k ln K s l ω L p ln L P s l 1 ω L p ln L F. Substtutng for ln Y V A we arrve mmedately at equaton 8. Dervaton of Expressons 9-12 We present the dervaton n the followng steps: Step 1: We express ln X ln Y as a functon of γ. It follows from equatons 4 5 p C / gven fxed w r that: ln X ln Y ln p ln P X µ 1 ln P X. Followng the defnton of γ we wrte: P Z [ p 1 ] 1 d rα w 1 α 1 µ ɛ [ rα w 1 α 1 µ P µ Z ɛ [ γ P ZZ PZ 1 P X X P X C P Z γ 1 µ 1 P X P Z γ ] 1 1 d 1 A 1 d r α w 1 α 1 µ P µ X ɛa ] 1 [ ] 1 1 µ γ 1 A 1 d } {{ } Q 1 γ where ɛ µ µ 1 µ 1 µ α α 1 α 1 α 1 µ. We can then wrte: P Z rα w 1 α ɛ 1 1 µ P X rα w 1 α ɛ 1 1 µ Q 1 1 µ γ Q 1 1 µ γ 1 γ 1 d ln P X 1 µ 1 ln Q γ + 1 ln γ. 13 Step 2: The frms decson for use of L F 1 s related to ts decson on Ω. The frm 2

maxmzes: Π Π wl F 1 P Y wl F subject to: We defne: wrte: Y g + z j dj j 1 D P G p P G 1 1 G + 1 1 G + j 1 P Z j p P Z 1 1 1 P 1 Z Xj dj. P Xj 1 1 1 P 1 Z Xj dj P Xj Π 1 p Y 1 p 1 D. The FOC for Ω s: whch gves the followng expressons: Π Ω w LF Ω 1 Π Π Ω wl P wl F LP L F ln Π ln1 + ln p 1 Ω Π Ω Π µ1 γ Ω 1 ln p + ln D µ ln P X 1 Ω wf v λω λ 1 Π µ1 γ wf v λω λ wλl F wl F λ 1 1 P Y µ1 γ wl P 1 µ1 αp Y 1 µ1 αp Y λ 1 1 P Y µ1 γ L P L F 1 µ1 α λ 1 1 µ1 γ. PM P X 1 1 Ω 1 µ1 α λ 1 1 µ1 γ As γ λ ncrease so does the share of labor that s used for producton. Ths s used to arrve at the expresson for ω L P : 1 ω lp L F L wl F wl F + wl P λ 1 1 µ1 γ λ 1 1 µ1 γ + 1 µ1 α. 3

Step 3: Express ln F as a functon of P X. We wrte: ln L F λ ln Ω λ 1 ln P M ln p m ln1 γ 1 [ln P M ln P X ] ln P M ln P X 1 ln1 γ ln L F λ ln Ω λ ln P X 1 1 ln1 γ ln p m s L wlp + wl F P V A Y V A 1 µ 1 µ s l 1 ω lp ln L F λ 1 1 µ1 γ + 1 µ1 α 1 µ s l 1 ω lp ln L F λ 1 1 µ1 γ ln L F 1 µ 1 µ 1 µ µ1 γ 1 µ λ 1 1 µ1 γ ln L F 1 µ [ ln P X 1 ln1 γ ln p m Step 4: Replace the expresson for ln P X from equaton 13 n the precedng equaton. Replacng the above terms n the expresson for frm-level productvty equaton 8 aggregatng over all usng frm value-added shares ω we arrve at an expresson for aggregate productvty. The last step s to relate changes n ln Q γ to changes n ω γ. We start wth an expresson for the value-added weghts whch should relate market shares of each frm to technologes trade shares: ω p p j j 1 1 dj 1 We then substtute n usng our expresson for Q: [ Q γ ] γ µ 1 1 1 A 1 d j γ j µ 1 γ µ 1 ω 1 A 1 1 Aj 1 1 dj A γ µ 1 1. ]. wrte: ln Q γ 1 ω ln ω + µ 1 ω ln γ gven ln A 0. 4

The fnal expressons 9-12 are arrved at through substtuton regroupng these terms usng the approxmaton ln1 γ γ 1 γ ln γ whch s vald for small shocks. In the smulaton secton we do not use ths approxmaton because we study large shocks. Appendx B: Numercal Algorthm The algorthm works as follows. Frms start wth an ntal assumpton about the prces of the domestc nput bundle PZ 0 the fnal good P G 0. Snce the mportng behavor of each frm determnes ts margnal cost thereby nfluences P Z P G ths assumpton s effectvely equvalent to takng as gven all other frms mportng decsons. Holdng these prce aggregates fxed each frm smultaneously chooses the optmal number of mported varetes Ω 1. 48 Wth ths new set of mport varety choces {Ω 1 } we must solve a fxed pont problem to fnd a consstent set of new prces {p 1 } because each frm s margnal cost s a functon of all other frms prces due to roundabout producton. In partcular we terate the system: P 1 µ V p 1 1 1 A µ µ 1 µ 1 µ PZ 1 p 1 1 d 1 [ P 1 1 Z + 1 ] p µ 1 m Ω 1 1 for all frms untl the set of prces {p 1 } s consstent wth the domestc nput prce ndex PZ 1 wth all frms choces of mported varetes {Ω1 }. We then repeat ths algorthm wth frms takng as gven the prce ndces PZ 1 P G 1 generate a new set of prces { mport varetes {p 2 Ω 2 } prce ndces {PZ 2 P G 2 }. We contnue ths process untl p j } { } Ωj p j 1 Ω j 1 up to a very small tolerance. Appendx C: Comparatve Statcs of the Frm s Trade Response In ths appendx we evaluate how each frm s response to the terms of trade shock wll dffer based on ts pre-shock level of total mports. The ntent here s to derve an expresson that provdes some ntuton for the results n the text as such we do not provde a formal proof. We have shown that as long as λ s suffcently hgh the number of mported varetes s ncreasng n the frm s exogenous technology A. Gven ther relatve cost advantage frms wth hgher A have lower prces p consequently sell more have hgher Y. These are also the frms wth the lowest γ snce P M /P Z s lower the hghest M. The elastcty of the response n γ to the mport prce change s a functon of the ntal 48 Though our frms have fnte market shares they gnore the mpact of ther own prce changes on the aggregate prce ndex. Ths s not problematc because the largest frm n our benchmark calbraton has a market share of only 5 percent. 5

γ. Usng the defnton of γ we can show that: ln γ 1 γ ln PM ln P Z 1 1 γ 1 ln P Z + 1 ln Ω > 0 1 ln Ω µ 1 1 [ ] + 1 γ + ln D + 1. γ γ ln P Z λ µ + µ 1 For the second order condtons for an nteror soluton to Ω to hold the denomnator must satsfy λ µ + µ 1 γ 1 > 0. As long as the numerator s negatve ln P Z < 1 whch s not always the case frms ncrease the share spent on domestc nputs γ when mport prces ncrease. To see how ths elastcty vares across exstng mporters we wrte: ln γ γ 1 1 ln P Z + 1 As long as the parameters are such that ln P Z depends on ln Ω ln Ω 1 ln Ω 1 γ < 1 ln Ω ln Ω γ. < 0 the sgn of ths expresson / γ whch measures how the elastcty of the sub-extensve margn vares wth γ. If / γ > 0 ndcatng that the elastcty of the sub-extensve ln γ margn decreases wth the ntal γ then we know that / γ < 0 mplyng that larger mporters wll change ther mport share by a greater percentage followng an mport ln Ω prce shock. If on the other h / γ < 0 then the net effect depends on whether the drect effect of a lower γ on rasng the percent change n γ exceeds the ndrect effect that rases the relatve prce of the optmal mport bundle relatve to domestc nputs by less. ln Ω We can wrte / γ as: ln Ω 1 γ κ 2 1 µ [ µ λ 1 ln P ] Z + 1 µ ln D + 1 1 where κ λ µ + µ 1 1 γ. ln P Z ln D do not vary wth γ. As long as ln P Z < 1 λ s suffcently large the senstvty to γ depends on whether s postve or negatve. µ 1 1 6

Appendx D: Addtonal Emprcal Analyses In ths appendx we consder two addtonal emprcal analyses. Frst we study the crosssecton of manufacturng ndustres. Next we consder changes n the frequency wth whch dropped nput varetes are permanently dropped. Startng wth the cross-sectonal analyss we focus on wthn-manufacturng varaton because trade may plausbly account for mportant varaton n productvty. Dfferences n the productvty declnes of the fnance government sectors for example would lkely have lttle to do wth trade. The Argentne Annual Manufacturng Census ncludes nformaton on value-added the number of salared workers n roughly 20 2-dgt ndustres. We combne ths wth nformaton also from the Argentne natonal statstcs on producer prces n those same 2-dgt ndustres to construct growth n real value-added per worker. Ths s a proxy for total factor productvty at the sector level. Next we use the nformaton from the Captal IQ database on each frms Prmary Sector to classfy some of them as belongng to these 2-dgt ndustres some sectors do not match are excluded. Fnally we combne 1 the mpled growth n mports for each manufacturng subsector from our data wth 2 nformaton on ntal sector levels of γ from the 1997 nput-output table wth 3 growth n total sectoral ntermedate spendng from the census to obtan a tme-seres for sectoral γ. Ths s essentally the same method used to determne changes n γ for the overall manufacturng sector n our calbraton. Fgures 14a 14b show the resultng relatonshp between changes n ths measure of productvty n the share of nput spendng on mports 1 γ for the perod 2000-2002. Fgure 14a shows ths plot for all sub-sectors wth avalable data but we note that some of these data ponts can reflect as few as two frms each. Fgure 14b shows ths relatonshp when ncludng only those sectors wth at least 20 matched frms. Both fgures suggest that shfts n ntermedate spendng from mported nputs toward domestc nputs correlate wth measured productvty declnes though the relatonshp s clearly nosy senstve to rules on the treatment of outlers. Gven ths senstvty gven ths s not our preferred welfare-relevant productvty measure we consder ths evdence less compellng robust than the results n the prmary paper for that reason only nclude them n ths Appendx. Next we consder the share of dropped varetes that are permanently dropped as a way to address concerns that nventored varetes are used for producton. In the paper n part to deal wth ths concern we focus on a 2 year perod. Ths s a horzon much longer than that typcally used by forecasters to descrbe the nventory cycle. But we addtonally fnd t useful here to measure the share of all frm mport varety combnatons from the prevous perod that are permanently dropped.e. permanent sub-extensve margn adjustment n the current perod. A varety s defned as HTS10xCountry s consdered to be permanently dropped f the frm does not agan mport t through 2008 we cannot go back earler n the analyss due to the gaps n our data n 1999 as dscussed n the paper. In ths analyss we only consder frms that mported at least 1 dollar of some good sometme after 2006 therefore exclude all frms whch permanently exted trade. Ths s a conservatve treatment of these extensve margn adjusters whch f ncluded would ncrease the set of permanently dropped goods. The share of varetes that are permanently dropped s plotted n Fgure 15. Early n the crss the lne jumps above 0.2 ndcatng that more than one-ffth of all prevous 7

Change n Value-Added per Worker Percent Per Year -.6 -.4 -.2 0.2.4 TV Equp Comp Equp Leather Metals Auto Ol Food Chemcals Paper Plastcs Wood Tobacco Med Equp Machnes Trans Equp PublshngApparel Textles Furnture -.25 -.2 -.15 -.1 -.05 0.05.1.15 Change n 1-g Percentage Ponts Per Year a All Sectors Change n Value-Added per Worker Percent Per Year -.6 -.4 -.2 0.2.4 Comp Equp Auto Med Equp Machnes Trans Equp Metals Ol Food Chemcals Paper Wood Textles -.25 -.2 -.15 -.1 -.05 0.05.1.15 Change n 1-g Percentage Ponts Per Year b Sectors wth at Least 20 Frms Fgure 14: Sectoral Input Sourcng Productvty 2000-2002 mporter-varety combnatons are permanently dropped a level far n excess of anythng seen before or after the crss. Our analyss does not requre that varetes are permanently dropped but ths nonetheless confrms that not only dd the number of dropped varetes spke dramatcally upward durng the begnnng quarters of the crss as shown n the paper but many of those dropped varetes are permanently dropped. Whle ths does not elmnate a possble role for nventores n smoothng the use of nput varetes t lmts the extent to whch nventores could have substtuted for these dropped varetes. 8

Fgure 15: Share of Varetes that are Permanently Dropped by Contnung Frms 9