Inequality Adjusted Growth Rates in Latin America A. Angeriz*, P. Arestis**, and S. P. Chakravarty***



Similar documents
How To Calculate The Accountng Perod Of Nequalty

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

Can Auto Liability Insurance Purchases Signal Risk Attitude?

An Alternative Way to Measure Private Equity Performance

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

Marginal Benefit Incidence Analysis Using a Single Cross-section of Data. Mohamed Ihsan Ajwad and Quentin Wodon 1. World Bank.

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

DEFINING %COMPLETE IN MICROSOFT PROJECT

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

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

17 Capital tax competition

Evaluating the Effects of FUNDEF on Wages and Test Scores in Brazil *

The OC Curve of Attribute Acceptance Plans

Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

7.5. Present Value of an Annuity. Investigate

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

Returns to Experience in Mozambique: A Nonparametric Regression Approach

1. Measuring association using correlation and regression

HOUSEHOLDS DEBT BURDEN: AN ANALYSIS BASED ON MICROECONOMIC DATA*

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek

Calculation of Sampling Weights

To manage leave, meeting institutional requirements and treating individual staff members fairly and consistently.

SIMPLE LINEAR CORRELATION

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

Multiple-Period Attribution: Residuals and Compounding

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

DO LOSS FIRMS MANAGE EARNINGS AROUND SEASONED EQUITY OFFERINGS?

World Economic Vulnerability Monitor (WEVUM) Trade shock analysis

Analysis of Premium Liabilities for Australian Lines of Business

Bank Credit Conditions and their Influence on Productivity Growth: Company-level Evidence

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

PRIVATE SCHOOL CHOICE: THE EFFECTS OF RELIGIOUS AFFILIATION AND PARTICIPATION

Overview of monitoring and evaluation

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

IIIS Dscusson & World Market Niche

Chapter 7: Answers to Questions and Problems

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

IS-LM Model 1 C' dy = di

Section 5.4 Annuities, Present Value, and Amortization

DEFINING AND MEASURING FAIRNESS IN FINANCIAL CONTRIBUTION TO THE HEALTH SYSTEM 1

Small pots lump sum payment instruction

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.

Financial Mathemetics

CHAPTER 5 RELATIONSHIPS BETWEEN QUANTITATIVE VARIABLES

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

How To Understand The Results Of The German Meris Cloud And Water Vapour Product

SUPPLIER FINANCING AND STOCK MANAGEMENT. A JOINT VIEW.

A household-based Human Development Index. Kenneth Harttgen and Stephan Klasen Göttingen University, Germany

Analysis of Demand for Broadcastingng servces

Energy prices, energy efficiency, and fuel poverty 1. Vivien Foster, Jean-Philippe Tre, and Quentin Wodon. World Bank. September 2000.

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

Texas Instruments 30X IIS Calculator

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

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12

ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING

Dynamics of Toursm Demand Models in Japan

The Application of Fractional Brownian Motion in Option Pricing

The Current Employment Statistics (CES) survey,

IDENTIFICATION AND CORRECTION OF A COMMON ERROR IN GENERAL ANNUITY CALCULATIONS

Intra-year Cash Flow Patterns: A Simple Solution for an Unnecessary Appraisal Error

An Interest-Oriented Network Evolution Mechanism for Online Communities

Support Vector Machines

Hollinger Canadian Publishing Holdings Co. ( HCPH ) proceeding under the Companies Creditors Arrangement Act ( CCAA )

Addendum to: Importing Skill-Biased Technology

The Effects of Tax Rate Changes on Tax Bases and the Marginal Cost of Public Funds for Canadian Provincial Governments

The demand for private health care in the UK

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

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

CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK. Sample Stability Protocol

Activity Scheduling for Cost-Time Investment Optimization in Project Management

8.5 UNITARY AND HERMITIAN MATRICES. The conjugate transpose of a complex matrix A, denoted by A*, is given by

What should (public) health insurance cover?

Traffic-light a stress test for life insurance provisions

LIFETIME INCOME OPTIONS

14.74 Lecture 5: Health (2)

Quasi-Hyperbolic Discounting and Social Security Systems

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

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

Faraday's Law of Induction

Using Series to Analyze Financial Situations: Present Value

SPECIALIZED DAY TRADING - A NEW VIEW ON AN OLD GAME

Criminal Justice System on Crime *

How Sets of Coherent Probabilities May Serve as Models for Degrees of Incoherence

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

The Short-term and Long-term Market

ADVERSE SELECTION IN INSURANCE MARKETS: POLICYHOLDER EVIDENCE FROM THE U.K. ANNUITY MARKET *

Working Papers SOCIAL EVALUATION OF ALTERNATIVE BASIC INCOME SCHEMES IN ITALY. R. Aaberge, U. Colombino, S. Strom

Proceedings of the Annual Meeting of the American Statistical Association, August 5-9, 2001

Chapter 4 ECONOMIC DISPATCH AND UNIT COMMITMENT

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

Depreciation of Business R&D Capital

Forecasting the Demand of Emergency Supplies: Based on the CBR Theory and BP Neural Network

Statistical Methods to Develop Rating Models

What is Candidate Sampling

1. Fundamentals of probability theory 2. Emergence of communication traffic 3. Stochastic & Markovian Processes (SP & MP)

Risk-based Fatigue Estimate of Deep Water Risers -- Course Project for EM388F: Fracture Mechanics, Spring 2008

CHAPTER 14 MORE ABOUT REGRESSION

Labor Supply. Where we re going:

Forecasting the Direction and Strength of Stock Market Movement

ADVERSE SELECTION IN INSURANCE MARKETS: POLICYHOLDER EVIDENCE FROM THE U.K. ANNUITY MARKET

Transcription:

1 Inequalty Adjusted Growth Rates n Latn Amerca A. Angerz*, P. Arests**, and S. P. Chakravarty*** * a.angerz@qmul.ac.uk School of Busness and Management Queen Mary, Unversty of London Mle End Road, London E1 4NS, UK ** pa267@cam.ac.uk Cambrdge Centre for Economc and Publc Polcy, Department of Land Economy, Unversty of Cambrdge, 19 Slver Street, Cambrdge CB3 9EP, UK *** S.P.Chakravarty@bangor.ac.uk The Busness School Bangor Unversty Bangor, Gwynedd, Wales LL57 2DG, UK Paper presented at the ASSA/AEA Meetng n San Francsco n January 2009 and also dscussed at the Economcs Colloquum at Gregynog n March 2009

2 Inequalty Adjusted Growth Rates n Latn Amerca Abstract: The pace of economc growth n developng countres n Latn Amerca ganed momentum n the 1990s. However, doubts have been expressed n some crcles about the benefts to an economy dervng from hgher rates of GDP growth wthout at the same tme accountng for dstrbutonal effects. The seemng dfference n the nterpretaton of the evdence arses because some commentators focus on the mean ncome, whle others are concerned wth the dstrbuton around the mean. Ths paper offers a dstrbuton-senstve ndex, whch takes both nterpretatons nto account. It assesses the per capta ncome that would be needed, f ncome were equally dstrbuted to reach the same level of welfare, whch s avalable from current dstrbuton of the per capta GDP. Ths approach allows for the combnaton of both growth and dstrbuton n rankng the success of countres n ther attempt to rase the lvng standard of the populaton.

3 Inequalty Adjusted Growth Rates n Latn Amerca 1. Introducton The pace of economc growth n developng countres n Latn Amerca ganed momentum n the 1990s, but reservatons began to be expressed n some crcles about the benefts of growth to the economy as a whole. Ths paper addresses concerns about the dstrbutonal aspect of growth wthn the economy. The ssue s of specal nterest because how the benefts of growth are dstrbuted vares between polces adopted to encourage hgher growth. In ths paper, we measure the success of economc polcy n the attempt to rase the standard of lvng of the populaton, by reference to growth and dstrbuton taken together. For ths purpose, we examne recent evdence of economc growth wth a dstrbuton-senstve ndex. Ths measure assesses the per capta ncome that would be needed, f ncome were equally dstrbuted, to reach the same level of welfare avalable from current dstrbuton of the per capta GDP. Results are presented for all major Latn Amercan economes n the two decades followng the rather long economc crss of the early eghtes. Some of the lterature on growth and dstrbuton examnes the trckle down hypothess, whch argues that the poor become less poor as the growth rate of per capta GDP accelerates. Cross-country data on growth and poverty, where poverty s measured usng some absolute ncome concept, s provded n support of the argument (for example, Dollar and Kraay, 2002). An ncome equvalent of US$ 1.00 n purchasng power party at 1985 prces s one of the poverty lnes consdered n ths debate. If the Dollar-Kraay (op. ct.) lne of argument s followed, GDP growth wll elmnate poverty. Bhalla (2002, 2003) goes even further. Accordng to ths author, growth had already by the year 2000 reduced world poverty to the target set for the Mllenum Development Goal of halvng world poverty rate, measured by the percentage of the world populaton lvng below US$1 per day. World poverty measurements are subject to debate about how many people have to get by wth ncome below $1 a day, whether measured n 1985 or 1996 value of the US dollar or whether adjusted for dfferences between countres n ther purchasng power party. Ths dscusson, n our vew, s not partcularly helpful (see also Reddy and Pogge,

4 2008). Seekng the connecton between GDP growth rates and the rate of declne n the numbers of people lvng below some arbtrary money measure, for example a dollar or two dollars a day, s not a partcularly meanngful method for countng the poor (see, for example, Angerz and Chakravarty, 2007, Ferrera et al., 2007). What the poor can buy depends not just on ther own money ncome but also on the dstrbuton of ncome n socety. The ncome needed even just to survve n a bg urban connurbaton such as Sao Paulo (Brazl) s greater than the ncome needed for bare survval n some small vllage n the same state of Sao Paulo. One aspect of the debate about growth s ts mpact on poverty, and ths debate s marred by problems of defnng poverty wthout reference to ncome dstrbuton (Prtchett 1997; Watkns 2002). Ths ssue s especally relevant n the context of latest trends showng growng nequalty around the world (see OECD, 2008). We bypass ths debate to provde a rankng of economc trends between countres not by GDP growth rates, but by the growth rates of an equvalent ncome measure. The latter s the equally dstrbuted per capta ncome requred to reach the same level of welfare currently acheved wth the actual dstrbuton patterns. The procedure adopted here s based on an approach developed by Atknson (1970) n dervng the Atknson Inequalty Index, whch nvolves a correspondng vew of socal welfare. The welfare measure ncorporates the strength of socety's preference for equalty through a subjectve nequalty averson parameter, whch appears n the welfare functon. The larger s the magntude of ths parameter, the hgher s the weght that socety attaches to ncome transfers at the lower end of the dstrbuton. Lkewse, lower magntudes of the parameter sgnfy greater senstvty of the welfare ndex to nequalty caused by transfers between ndvduals at the top end of the dstrbuton. A choce of the magntude of a subjectve nequalty averson parameter s then the bass for the calculaton of the recommended welfare ndex. Followng ths procedure, an nequalty-adjusted ndex s computed here n the attempt to measure an equally dstrbuted equvalent ncome. Varatons n ths ndex obtaned by defnng nequalty averson parameters are then correlated wth the GDP growth rates of Latn-Amercan countres. Ths s undertaken n an attempt to test whether GDP growth goes hand n hand wth nequalty-senstve welfare ndces. The data used n calculatng these ndces are derved from World Development Index

5 publshed by the World Bank. 1 explaned n secton 4 below. The salent features of ths source of data are Ths paper s organzed as follows. Secton 2 outlnes the proposed way of measurng poverty and dscusses the lnk between poverty and nequalty. Secton 3 explans the relevance of the socal welfare functon mplct n dervng the Atknson nequalty ndex. Secton 4 provdes and dscuses the estmates of the rate of change n the nequalty-adjusted growth ndex for varous economes. Fnally, secton 5 summarzes the argument and concludes. 2. Poverty, Inequalty and Growth Poverty and nequalty are mult-dmensonal deas, whch cannot be fully descrbed wthout detaled consderatons of opportuntes n the lfe that ndvduals wsh to lead. One of the dmensons of the dea of well-beng s current access to goods and servces. Though, even wthn ths narrow framework, dfferences are observed about how to set the poverty level ncome,.e. whether one uses ncome data or consumpton data. 2 As mentoned above, a strand n the current lterature measurng the mpact of growth on poverty bypasses the queston of nequalty and measures the number of the poor by reference to some partcular amount of money ncome n dollars, fxed all over the world (Bhalla 2003; Dollar 2002; Chen and Ravallon 2001). In ths exercse, we consder only ncome data, and translate ncome per capta as measured by gross domestc product per capta nto an equvalent equally dstrbuted ncome per capta. The latter would be needed to reach the same level of socal welfare by consderng a partcular socal welfare functon that s dstrbuton senstve, as explaned n secton 3 below. The ncome-based defnton of the poor entals countng those whose ncome falls below some poverty datum lne. There are two ways that ths problem s generally approached.

6. The frst s to defne some mnmum level by reference to physcal requrements, for example, nutrtonal requrements, for survval.. The second s "to endeavour to defne the style of lvng whch s generally shared or approved n each socety and fnd whether there s... a pont n the scale of the dstrbuton of resources below whch as resources dmnsh famles fnd t partcularly dffcult to share n the customs actvtes and dets comprsng ther socety's style of lvng" (Townsend, 1979). Both the above approaches rase dffcult conceptual ssues. How do we defne the physcal requrements for survval? (Sath, 2005; Sen, 2005). 3 How do we defne the style of lvng approved by socety? Desa and Shah (1988) attempt to resolve ths problem, rased n the debate between Pachaud (1981) and Townsend (1981), by redefnng the style of lvng approved by socety as the modal behavour (p. 518). By dong so, the authors make the socologcal vew of poverty emprcally measurable (Desa and Shah, op. ct., p. 158). Ther recommended method defnes the mnmum requrement for survval for a good as the constant term n a lnear expendture system of equatons. Then the mnmum amount of money requred for survval may be defned as the sum of the constant terms weghted by prces. Suppose that there are n goods and x amount of good s demanded, when ncome s M. Suppose further that s s the mnmum amount of the th good that s demanded and money left over after the mnmum demand for each of the goods s satsfed s spent on purchasng addtonal amount of goods by maxmzng the followng utlty functon subject to the budget constrant as follows: U x s subject to : n 1 n 1 p x M (2.1) The demand functon resultng from the above maxmsaton exercse turns out to be a lnear functon:

7 p x p s n M p s (2.2) 1 n 1 p s could now be regarded as the cost mnmum survval bundle of goods, whch are bought before addtonal amount of any good s purchased from the money that s left over (Green 1976, pp 142-143). Thel and Clements (1987, p. 10) call s the subsstence consumpton of the th commodty. The data requrements n gong down ths route for estmatng the poverty lne for comparson between countres and over tme are onerous. Instead, poverty measures used n the trckle down approach are based on some arbtrary dollar value of survval ncome (Dollar and Kraay, 2002), such as one or two dollars a day per person. Ths approach ams to express n money terms some vew of basc survval requrements for food and shelter. However, malnutrton and hunger mpacts on the qualty of lfe n other ways than by causng mmedate death or recorded llness. Atknson (1983), for example, ponts out that there s no unque level of food ntake defnng the subsstence level of nutrton. Instead, physcal effcency declnes n a number of ways due to malnutrton of dfferent knd. There s an addtonal dffculty n attemptng to set the mnmum survval requrements n money terms, such as dollar a day or two dollars a day. Prces of survval necesstes are not dentcal between dfferent regons that are at very dfferent stages of development. On reflecton, t appears that the dstncton between relatve and absolute poverty s not as sharp as t mght seem at frst sght. Relatve prces are not ndependent of the dstrbuton of ncome. As more people acqure cars and buses run wth empty seats, those who have to depend on buses for transport have to carry a greater fracton of the fxed cost of bus servce. Changes n ncome dstrbuton may thus lead to changes n relatve prces. Ths, n turn, may lead to a change n what and how much the poor can buy wth a fxed sum of money. Thus, the subsstence level of ncome, often thought of as some absolute level, s tself a concept that s not ndependent of the dstrbuton of ncome. 4

8 These vews support our choce for proposng to examne the mpact of growth on poverty. In the next secton, we provde the theoretcal underpnnngs of ths method, whch compares growth rates n GDP per capta wth growth rates n a dstrbutonadjusted welfare ndex followng from the Atknson nequalty ndex. 3. A Welfare Index To understand the nature of the welfare ndex reported here, t s necessary to hghlght some salent features of the Atknson nequalty ndex (Atknson, 1970). We begn wth a standard utlty functon U(y), where y stands for ncome. If utltes are assumed to be addtvely separable, then a socal welfare functon mght be constructed by takng a weghted average of the ndvduals utlty functons. The weghts are taken from the frequency densty functon of ncome y, f(y), where ncome s allowed to fluctuate from the mnmum ( 0) to some maxmum levels of ncome, be t nfnty, so that: W U ( y) f ( y) dy (3.1) 0 where W s the welfare ndex and the rest of the varables are as defned above. Next defne the cumulatve densty F(y) as follows: F( y) y f ( x) dx (3.2) 0 A preference for less nequalty s embedded n U, mplyng that the utlty functon exhbts dmnshng margnal utlty of ncome and t s thus concave,.e. U' >0 and U''<0, the frst and second dervatves of the utlty of ncome, U, respectvely. Then a dstrbuton f(y) wll provde a hgher value of W than a dstrbuton g(y) f and only f the followng condtons are satsfed (Atknson 1970, p. 246): 0 z F( y) G( y) dy 0 for z 0 z max (3.3)

9 and there exsts some value of y at whch F(y) G(y) (3.4) Note that F(y) and G(y) are cumulatve densty functons (see Equaton 3.2 above for a defnton) assocated wth the dstrbuton functons f(y) and g(y), respectvely. Atknson provdes an nterpretaton of the condton n Equaton (3.3) n terms of the Lorenz curve. The Lorenz curve plots on the vertcal axs the percentage of total ncome accrung to the poorest x percentage of the populaton recorded on the horzontal axs. For each value of F(y), as defned by equaton (3.2) above, we can wrte the Lorenz curve as follows: y 1/ x ( F) f ( x) dx (3.5) 0 where the parameter s the mean of the dstrbuton of ncome. The condton n equaton (3.3) wll be satsfed for two dstrbutons of dentcal mean value, for any concave utlty functon as descrbed above, f the Lorenz curve assocated wth f(y) s nested nsde,.e. les above the Lorenz curve for dstrbuton g(y), as shown n Fgure 1 below. where the parameter s the mean of the dstrbuton of ncome. The condton n equaton (3.3) wll be satsfed for two dstrbutons of dentcal mean value, for any concave utlty functon as descrbed above, f the Lorenz curve assocated wth f(y) s nested nsde,.e. les above the Lorenz curve for dstrbuton g(y), as shown n Fgure 1 below.

10 Fgure 1: Lorenz Curve Fracton of total ncome 1 1 Fracton of populaton Lorenz curve for g(y) Lorenz curve for f(y) Thus, f the Lorenz curve for f(y) s above that of g(y), for two mean preservng dstrbutons, socal welfare (see equaton 1) s hgher for dstrbuton f(y). If the Lorenz curves are not nested,.e.: when they cross as shown n Fgure 2 below, then the welfare functons assocated wth the dstrbutons f(y) and g(y) cannot be ranked.

11 Fgure 2: Lorenz Curve wth crossng dstrbutons Fracton of total ncome 1 1 Fracton of populaton Lorenz curve for g(y) Lorenz curve for f(y) A way around ths problem, f we retan the addtvely separable welfare functon gven by equaton (3.1), s to mpose a partcular concave functonal form on the utlty functon U(y). To arrve at a functonal form, we start by askng for the level n per captal ncome X e needed, f ncome were equally dstrbuted, to acheve the same level of welfare obtaned gven the current dstrbuton f(y) of ncome. We can wrte the current level of welfare as: U ( X ) U ( y) f ( y) dy 0 e (3.6) where X e s analogous to a certanty equvalent ncome when ncome s randomly dstrbuted. To compare ths certanty equvalent ncome level wth the actual mean of the per capta ncome, µ, Atknson (1970) derves a functonal form of a concave utlty functon, consstent wth a vew of nequalty averson:

12 U ( y) A B y or U ( y) ln( y), (1 ) (1 ), f 1 f 1 (3.7) Note that A and B are constants. Also note that the assumpton of concavty of the utlty functon entals holdng 0 n the above expresson for U(y), where s postve, whch s ndcatve of averson to nequalty. The parameter can be nterpreted as a measure of the degree of nequalty averson. For dscrete dstrbutons of ncome, the equvalent ncome s as follows (Atknson 1970, p. 257): X e (1 ) n y f ( y ) 1 1 (1 ) (3.8) Equaton (3.8) above s not drectly presented n Atknson (op. ct.), but n the context of an nequalty ndex, whch s defned as follows: I X e 1 (3.9) An nterpretaton, whch may be emerge by reference to equatons (3.5) and (3.9) s that the value of I allows us to ascertan how much greater the mean ncome has to be than the equally dstrbuted equvalent ncome, n order to obtan the same level of welfare. For example, f I = 0.3, t allows us to say that f ncomes were equally dstrbuted, then we should need only 70% of the present natonal ncome to acheve the same level of welfare (accordng to the partcular welfare functon [chosen here]) (Atknson 1970, p. 250). Referrng to equatons 3.8 and 3.9, we may note that I depends on the nequalty averson parameter,.e.. 4. Growth Trends Adjusted for Inequalty In ths secton we compare growth rates n GDP between dfferent perods for some countres and between countres for smlar perods wth the correspondng growth rates of computed values of welfare preservng equally dstrbuted equvalent

13 ncomes. We fnd that perods of hgher GDP growth are not necessarly perods of hgher growth n the above equvalent ncomes. Some of the countres have acheved hgher growth rates at the expense of not only reducng the share of the lower quntle groups, but sometmes by also reducng the absolute per capta ncome correspondng to some of these groups. The data on GDP and quntle dstrbutons are obtaned from the World Development Indcators, an extensve dataset publshed by the World Bank (see footnote 1), whch covers economc, socal and fnancal data, along wth nformaton about natural resources and envronmental ndcators. GDP growth comparsons wth welfareequvalent ncome growth are produced by selectng a group of countres from Latn Amerca for the perod 1980-2004. Ths span s long enough to cover the experence of recoverng from the 1982 fnancal crss n the regon, and the years of expanson, whch took place n the 1990s; followed by the crss n the early 2000s.. For most of larger economes varous data ponts are avalable n these years, thereby provdng a good coverage of what happened n the perod, although only untl 2004. For each of these countres we calculate the welfare-equvalent equally dstrbuted ncome, Xe. The latter s defned as per capta GDP n constant dollars. Data for GDP and quntle dstrbuton are not avalable for all years, but the growth rates are annualzed growth rates between the years for whch data are avalable 5. All the results reported n ths paper are computed for the nequalty averson parameter = 1.5. Such magntude s a large enough value to ensure that the nequalty ndex and also the welfare-equvalent equally dstrbuted ncome are partcularly senstve to changes n the share of ncome accrung to the poor, the lowest quntle. Due to ncreasng nequalty durng many of the ntervals, the growth rate of the equalty equvalent ncome s found to be lower than the growth rate n per capta GDP over much of the perod. In a quarter of the cases, equvalent ncome even declnes despte per capta GDP growth beng postve. For large values of the nequalty averson parameter, lowerng nequalty s no less mportant n mprovng the standard of lvng of the poor than faster economc growth. An llustraton s provded n Table 4.1 usng data for Argentna.

14 In the table below, we report Xe, per capta GDP n constant (Year 2000) dollars, and also the Atknson nequalty ndex for Argentna, calculated for the avalable years. An nequalty ndex of 0.46 n 1992 means that, f ncome were equally dstrbuted, we could acheve the same level of welfare as we do wth the gven per capta GDP for less than a half of that ncome. Table 4.1: Per Capta Income and Inequalty Year GDP (Yr 2000 dollars) Argentna Per Capta Equalty Equvalent Income 1986 6543 4173 0.36 1992 6860 3675 0.46 1996 7489 4268 0.43 1998 8213 4524 0.45 2001 7288 3677 0.50 2002 6431 3240 0.50 2003 6932 3508 0.49 2004 7486 3838 0.49 Inequalty ndex Notce that the equalty equvalent ndex suggests a negatve rate of growth between 1986 and 1992 notwthstandng an ncrease of 5 per cent n the per capta GDP durng the same perod. In ths emblematc case, certan dstnctve perods can be dentfed after 1992. Up to 1996, and followng hstorcal trends, hgh per capta GDP growth rates are concurrent wth decreasng nequalty ndces. Then, nequalty starts ncreasng, but such escalaton s not enough to prevent a rse n the equally dstrbuted equvalent ncome untl the onset of the 1999 economc downturn. Argentna observed a dramatc economc crss startng n 1999, whch clearly dstngushes the subsequent perod. Note that ths tme there s a drop n almost 10% n the per capta GDP, whch concdes wth an assocated ncrease n the nequalty ndex, culmnatng n a trend of reduced equally dstrbuted equvalent ncome untl 2002. Then both per capta GDP and the above equvalent ncome begn to recover. The recovery n the equvalent ncome s, however, slower than t mght have been f nequalty had also decreased to the levels seen n the earler years of the same decade. We note from Table 4.1 that, between 1992 and 2003, per capta GDP n Argentna

15 only ncreased margnally, whlst, at the same tme, the equalty equvalent ncome declned as nequalty ncreased. Equally, notce that there was vrtually no change n per capta ncome between 1996 and 2004, but equvalent ncome declned by 10 per cent. If poverty were to be addressed n Latn Amerca, these results would ndcate that a more actve polcy for ncome redstrbuton was needed. Indeed t s not the case that ncome nequalty s essental for hgher growth. Table 4.2: Per Capta GDP (Yr 2000 dollars) and Equalty Equvalent Income Year Brazl Chle GDP Equalty Equvalent Inequalty Index GDP Equalty Equvalent Inequalty Index 1984 3175 1438 0.55 n/a n/a n/a 1985 3355 1422 0.58 n/a n/a n/a 1990 3374 1346 0.60 3093 1554 0.50 1992 3285 1416 0.57 3616 1840 0.49 1993 3386 1433 0.58 n/a n/a n/a 1995 3613 1353 0.63 n/a n/a n/a 1996 3634 1496 0.59 4543 2297 0.49 1997 3699 1499 0.59 n/a n/a n/a 1998 3644 1816 0.50 4862 2407 0.51 1999 3601 1526 0.58 n/a n/a n/a 2000 n/a n/a n/a 4917 2499 0.49 2001 3695 1558 0.58 n/a n/a n/a 2002 3738 1621 0.57 n/a n/a n/a 2003 3733 1639 0.56 5215 2719 0.48 2004 3892 1748 0.55 n/a n/a n/a 2005 3951 1833 0.54 n/a n/a n/a We now turn our attenton to Brazl and Chle (relevant data reported n Table 4.2) between 1992 and 2003. Brazl performed worse on both counts, growth n per capta GDP and also n the growth of equvalent ncome. Inequalty remaned hgh all across ths perod n ths counry. There was also a smlar resstance n the movement n the nequalty ndex for Chle durng ths tme, although the level of nequalty was lower n Chle. Furthermore, a combnaton of lower GDP growth and no relef from nequalty ensured that the equvalent ncome between Brazl and Chle contnued to dverge durng ths decade. Brazlan per capta GDP rose by 14 per cent and the equvalent ncome ncreased by 16 per cent durng ths perod. The correspondng fgures for Chle are 44 and 48 per cent, respectvely. It s worth pontng out that, despte regsterng lower growth rates over the decade, Brazl could have managed to

16 keep the percentage dfference n equvalent ncome unchanged wth Chle f at least t had lowered the nequalty ndex from 0.57 to the Chlean fgure of 0.48 by 2003. If that mprovement n the dstrbuton of ncome had taken place, the equvalent ncome n Brazl would have gone up by 37 per cent, nstead of the meager 16 per cent that was wtnessed, durng the 1992-2003 perod. For an llustraton of the general pont, consder the separate effects of growth and dstrbuton n the gap between the equvalent ncome for both countres and the two perods of tme. A country that mantans a hgher level of nequalty has to grow faster to keep up wth the rate of growth n equvalent ncome for countres that enjoy a lower level of nequalty. Consder the followng example as an llustraton of the pont. Suppose that n 1992, the per capta GDP n countres A and B were equal, say at $1000. The nequalty ndex (for an nequalty averson parameter =1.5) n the case of Country A s 0.5, but t s 0.6 for Country B. Assume dentcal per capta growth rates n both countres, at 5 per cent per annum for 10 years. In such case, per capta GDP levels are dentcal between the countres, but the dfference n Xe ncreases from $100 to $163. For ths dsparty to shrnk, the dfference n nequalty between the countres wll have to narrow over tme. If nequalty falls from 0.6 to 0.575 n Country B, whlst the ndex remans at ts orgnal value of 0.5 n Country A, then the above ntal dfference of $100 n the ntal year goes up to only $122 at the end of the perod. Ths dfference, however, contnues to wden over tme f Country B remans more unequal. A corollary of the pont rased above s that the countres, whch start out wth a hgher level of nequalty and contnue to reman more unequal, would have to grow faster for ther equvalent ncome not to fall further behnd. Ths s consstent wth the fndngs reported n Watkns (2002): Countres wth low levels of ncome nequalty can expect to regster far hgher rates of poverty reducton than hghly unequal countres. Whle economes wth hgher levels of per capta GDP also sometmes exhbt hgher levels of equvalent ncome (Fgure 4.1), ths correspondence dsappears when a comparson of the growth rates of per capta GDP and Xe s made, as n Fgure 4.2.

17 Fgure 4.1. GDP per capta and Xe GDPpc 8000 6000 4000 2000 0 par95(5) par02(3) par03(1) par98(3) par99(1) par90(1) bo99(2) bo02(3) bo97(6) bo86(6) bo91(5) n93(1) n98(5) n01(3) br89(2) br95(2) br96(1) br97(1) br87(2) br99(1) br01(2) br02(1) br04(1) br05(1) ch94(2) ve03(5) br03(1) br98(1) ch92(2) br90(1) br92(2) br85(1) br93(1) br84(3) br81(1) ch90(3) ja02(2) ja03(1) ja99(3) ja04(1) ja00(1) ja96(3) ch87(7) ja89(1) ja90(1) ja91(1) ja93(2) ja88(8) ar98(2) ar04(1) ar96(4) ar01(3) ar03(1) ar92(6) uy98(2) ar02(1) ar86(6) uy04(1) uy00(2) uy01(1) uy96(7) ve81(1) uy03(2) ch03(3) uy81(1) ch98(2) ch00(2) ve87(6) ve98(1) ve97(1) ve93(4) ve96(1) ve95(2) uy89(8) ch96(2) ve89(2) 0 1000 2000 3000 4000 5000 Xe Note: Indvdual cases are dentfed by country abbrevaton, correspondng year and dfference n years wth prevous perod. Fgure 4.2 40 Xe (per cap) growth rate 30 20 10 0-15 -10-5 0 5 10-10 Seres1-20 -30 GDP (per cap) growth rate

18 The ndvdual cases n the above fgures are dentfed by country abbrevaton, correspondng year and dfference n years wth prevous perod. Whlst the detals n the fgures are not mmedately obvous to the naked eye, the pont that emerges s that hgher rates of aggregate growth are not necessarly related to hgher rates of growth n ncome for the poor. There s no room for complacency about poverty reducton through the beneft of economc growth trcklng down to the poor. We compare growth rates n GDP between dfferent perods for some countres and between countres for smlar perods wth the correspondng growth rates of computed values of welfare preservng equally dstrbuted equvalent ncomes. We note that perods of faster growth are not necessarly perods of hgher growth n the alternatve measure of equvalent per capta ncome. Indeed, the correlaton coeffcent for the rates of growth of per capta GDP and equalty equvalent ncome s only 0.14 n the data examned here. In most countres, the nequalty ndex s over 0.5, ndcatng that the actual level of welfare could be acheved, f we were to take a strong averson to nequalty and poverty, wth less than half the reported per capta ncome. Perods of hgher growth rates n per capta GDP are not necessarly accompaned by perods of hgher growth n equvalent ncome wthn a country. Between 1984 and 1985, Brazl wtnessed per capta growth n GDP of 5.7 per cent per annum, but the equvalent ncome fell by 2.7 per cent n that year. Yet between 2003 and 2004, when per capta GDP growth rate was lower, only 4.3 per cent, equvalent ncome rose by 6.1 per cent. Lkewse, countres that record faster rates of growth n per capta GDP do not necessarly regster hgher rates of growth n the nequalty adjusted ncome than countres where the per capta growth rates n GDP are slower. Table 4.3: Per Capta GDP (Yr 2000 dollars) and Equalty Equvalent Income Year Bolva Honduras GDP Equalty Equvalent Inequalty Index GDP Equalty Equvalent Inequalty Index 1986 2909.623 1348.838 0.536422 2713.301 1100.658 0.594347 1997 3480.55 1085.622 0.688089 3043.904 1294.771 0.574635 2002 3537.809 883.3714 0.750306 n/a n/a n/a 2003 n/a n/a n/a 3140.854 1330.445 0.576407

19 Fnally, Table 4.3 ncludes two cases representatve of low ncome per capta countres. The ssues are the same as those observed for mddle ncome countres that are examned earler. Bolva managed to regster a greater ncrease n per capta GDP between 1986 and 2002 than Honduras managed n the slghtly longer perod between 1986 and 2003. However, whle per capta GDP rose by 22 per cent but equvalent ncome declned by 35 per cent durng 1986-2002 n Bolva. Honduras, on the other hand, regstered only 16 per cent growth n per capta GDP between 1986-2003, but could only manage a slow growth rate of 21 per cent n equvalent ncome durng that perod. Faster rates, per se, of growth n the GDP do not flter down to mprovements n the ncome of the poor. 5. Summary and Conclusons Whlst t s wdely recognzed that an aggregate measure such as the GDP leaves out many relevant varables ndcatve of well beng, growth rates n per capta GDP remans popular as an ndex of success of economc polcy. Consder the followng statement by Mchael Camdessus, one tme Managng Drector of the IMF: Our prmary objectve s growth. In my vew, there s no longer any ambguty about ths. It s towards growth that our programs and ther condtonalty are amed (as reported n Prezeworsk and Vreeland, 2000, p. 385). In recent years there has also been heghtened concern about poverty reducton. Whlst hgher rates of growth and hgher rates of poverty reducton are not ncommensurate, growth does not necessarly address the dstrbutonal ssues entaled n reducng poverty. In ths paper, we propose an alternatve to per capta GDP, an ndex, whch s senstve to ncome dstrbuton. Fgures for GDP are modfed followng an approach that s derved from Atknson s semnal paper on ncome nequalty (Atknson, 1970). Usng a subjectve nequalty averson parameter, a measure of equvalent ncome s obtaned from the publshed data on per capta GDP and ts quntle dstrbuton. The parameter s chosen hgh enough for the equvalent ncome to be especally senstve to dstrbuton towards or away from the poorest quntle. Ths method allows us to compare not just the growth rate of GDP over tme

20 or between countres, but also the rate of growth of the equvalent ncome. Ths measure allows us to combne growth rates and nequalty of ncome, and thus to comment on the success of the twn objectves of growth and poverty reducton usng one sngle measure. Examnng recent data on Latn Amerca untl 2004, we notce that some of the countres reduced nequalty, f only margnally, but the general pcture was at the tme not encouragng. In some countres, nequalty has even worsened as austerty measures, followng fscal and monetary crses that appear to be endemc n the regon, have been ntroduced. These trends are captured n our equvalent ncome measure of economc well-beng. References Angerz, A. and Chakravarty, S. (2007), Changng patterns of UK poverty, 1997-2004, Cambrdge Journal of Economcs, 31(4), 995-1006. Atknson, A. (1970), On the measurement of nequalty, Journal of Economc Theory 2, 244 263. Atknson, A. (1983), The Economcs of Inequalty. 2 nd edton Oxford: OUP Arrow, K. (1982), Why People Go Hungry?, New York Revew of Books, XXIX(12), 15 July, pp. 24-26. Bhalla, S.S. (2002), Imagne, There s No Country: Poverty, Inequalty and Growth n the Era of Globalsaton, Washngton, D.C.: Insttute of Internatonal Economcs Bhalla, S.S. (2003), "Cryng Wolf on Poverty", Economc and Poltcal Weekly, 5 July, 2843-2856. Chen, S. and Ravallon, M. (2001), "How Dd the World's Poorest Fare n the 1990s?" Revew of Income & Wealth, 47(3), 283-301. Desa M.J. and Shah A. (1988), "An Econometrc Approach to the Measurement of Poverty", Oxford Economc Papers, vol. 40, pp. 505-522. Dollar, D. and Kraay, A. (2002), Growth Is Good for the Poor, Journal of Economc Growth, 7(3), 195-225. Ferrera, F.H.G., Lete, P.G. and Ravallon, M. (2007), Poverty Reducton wthout Economc Growth? Explanng Brazl s Poverty Dynamcs, 1985-2004, World Bank Polcy Research Workng Paper Seres, No. 4431, Development Research Group Poverty Team, Washngton D.C.: World Bank.

21 Green, H. A. J. (1976), Consumer Theory. Basngstoke: Macmllan OECD (2008), Growng Unequal? Income Dstrbuton and Poverty n OECD Countres, Pars: Organsaton of Economc Co-operaton and Development. Pachaud, D. (1981), Peter Townsend and the Holy Gral, New Socety, 10 September. Prezeworsk, A. and Vreeland, J.R. (2000), "The Effect of IMF Programs on Economc Growth", Journal of Development Economcs, 62(3), 385-321. Prtchett, L., 1997. "Dvergence, Bg Tme," Journal of Economc Perspectves, Amercan Economc Assocaton, vol. 11(3), pages 3-17, Summer. Reddy, S and Pogge, T (2008), How Not to Count the Poor, n Stgltz, J., Anand, S. and Segal, P. (eds.), Debates on the Measurement of Global Poverty, Oxford: Oxford Unversty Press. Sath, S. (2005), Poverty Lnes versus the Poor: Method versus Meanng, Economc and Poltcal Weekly, 22 October, 4601-4610. Sen, P. (2005), Of Calores and Thngs: Reflectons on Nutrtonal Norms, Poverty Lnes and Consumpton Behavour n Inda, Economc and Poltcal Weekly, 25 October, 4611-4618.. Thel, H. and K. W. Clements (1987), Appled Demand Analyss, Cambrdge, Mass: Ballnger. Townsend, P. (1979), Poverty n the Unted Kngdom, Harmondsworth: Pengun. Townsend, P. (1981), Reply to Pachaud, New Statesman, 17 September. Townsend, P. (1985), A Socologcal Approach to the measurement of Poverty A Rejonder to Professor Amartya Sen, Oxford Economc Papers, 37(4), 659-668. Watkns, K. (2002), Makng Globalzaton Work for the Poor, Fnance and Development, 39(1), http://www.mf.org/external/pubs/ft/fandd/2002/03/watkns.htm

22 1 The relevant URL s: http://publcatons.worldbank.org/wdi/ndcators 2 Awareness has been rased wth respect to the dstncton between measures of poverty and nequalty derved usng ncome and consumpton data. The ssues rased here, however, are ndependent of that dstncton. 3 The settng of the Brtsh supplementary benefts n levels after the war was nformed by the Beverdge Report of 1942. It provded an estmaton of the subsstence level of ncome needed for survval. Ths approach, though, presents a problem n that people can reman alve for qute a number of years even wth ncredbly low levels of nutrton. 4 Arrow (1982) provdes an llustraton. He explans how the sudden engagement of htherto unemployed workers nto the labour force and the consequent changes n ncome dstrbuton caused meat prces to rse even though supply ncreased. 5 The avalable data can be obtaned on request from the authors.