The Linkages between Growth, Poverty and Inequality in Vietnam: An Empirical Analysis. Hoi Quoc Le *

Transcription

1 DEPOCEN Workng Paper Seres No. 200/06 The Lnkages between Growth, Poverty and Inequalty n Vetnam: An Emprcal Analyss Ho Quoc Le * * Faculty of Economcs, Natonal Economcs Unversty, Vetnam The DEPOCEN WORKING PAPER SERIES dssemnates research fndngs and promotes scholar exchanges n all branches of economc studes, wth a specal emphass on Vetnam. The vews and nterpretatons expressed n the paper are those of the author(s) and do not necessarly represent the vews and polces of the DEPOCEN or ts Management Board. The DEPOCEN does not guarantee the accuracy of fndngs, nterpretatons, and data assocated wth the paper, and accepts no responsblty whatsoever for any consequences of ther use. The author(s) remans the copyrght owner. DEPOCEN WORKING PAPERS are avalable onlne at

2 The Lnkages between Growth, Poverty and Inequalty n Vetnam: An Emprcal Analyss Ho Quoc Le* Faculty of Economcs Natonal Economcs Unversty, Vetnam November 2008 Abstract Ths paper examnes how ntal nequalty and poverty rate are related to subsequent economc growth n the provncal level of Vetnam. The results show a robust negatve relatonshp between ntal poverty rate and subsequent economc growth. However, there s no lnk between ntal nequalty and subsequent economc growth. The results also show that lower nequalty leads to lower poverty rate and poverty reducton could help to reduce nequalty. Other determnants of nequalty and poverty reducton nclude human captal, nvestment, GDP growth rate and trade openness. The man polcy mplcaton that emerges from ths paper s that concentratng on poverty elmnaton wll help us to buld a more equtable socety wthout sacrfcng economc growth. Key words: poverty, nequalty, economc growth, Vetnam. *Author address: Ho Quoc Le, Faculty of Economcs, Natonal Economcs Unversty, 207 Ga Phong Road, Hano, Vetnam. Emal: holq@neu.edu.vn. Telephone:

3 . Introducton The lnks between growth, poverty and nequalty have long been nvestgated. Earler researchers lke Lews (954) and Kuznets (955) were prmarly nterested n the dstrbutonal consequences of economc development. In recent years, the renvgorated nterest n the endogenous growth theory has fuelled substantal research nto the exploraton of the mpact of nequalty on growth. Whle the emergng domnant vew s that nequalty s harmful for growth (Aghon al et. 999; Perott, 996; Persson and Tabelln, 994; Alesna and Rodrck, 994), some researchers have provded evdence to the contrary (L and Zou, 998; Partrdge, 997). Theoretcally, the channels through whch ths mpact s transmtted dffer n accordance wth the model used. Relable emprcal testng of these theores has been hampered by the lack of qualty data. A number of emprcal studes n ths feld s based prmarly on cross-secton country data, whch suffer from a lack of unformty. As the countres under consderaton are at dfferent stages of economc development, usng the avalable cross- country data to test the varous hypotheses of the lnkages among growth, poverty and nequalty does not yeld conclusve results. Ths paper ams to explore how ntal nequalty and poverty rate are related to subsequent economc growth n provnces of Vetnam. I consder both nequalty and poverty as they represent two dstnct areas of nterest. Whle the former deals wth the dsperson of ncome n each provnce, the latter focuses on the lower end of the ncome dstrbuton. Moreover, many of the theoretcal arguments relatng nequalty to growth are actually better suted to explan the relatonshp between poverty and growth. Therefore, ncludng both poverty and nequalty n the analyss ensures that the separate contrbutons of these two related, but dstnct, factors to economc growth are properly dentfed. Unlke most other related studes n the lterature, ths paper analyses a cross-secton of provnces rather than countres. Whle provnces are soco-economc enttes n ther own rght, they are partcularly relevant for studyng the relatonshp between growth, poverty and nequalty snce they are less arbtrary spatal-economc unts compared to cross-secton of countres, and they represent economes that are at relatvely smlar stages of development. In addton, as most of the relevant data for provnces are only obtaned from the General Statstc Offce of Vetnam, the ssue of non-unform methodologes and defnton does not pose a serous problem when dealng wth provnces. Hence, ths study on the relatonshp between ncome dstrbuton, poverty and growth s more relable than many of the studes n the lterature. The paper s organzed as follows. Secton 2 brefly revews the varous exstng theores on the mechansms through whch nequalty affects economc growth. Secton 3 presents the emprcal framework used to test the lnkages between growth, poverty and nequalty. 2

4 The emprcal results are presented n secton 4 and 5, and the fnal secton provdes concludng remarks. 2. Theoretcal lnks between poverty, nequalty and growth The work of Kuznets (955) s perhaps the startng pont for examnng the lnks between poverty, nequalty and growth. Kuznets hypotheszes that growth and nequalty are related n an nverted U-shaped curve. In the early stages of economc development, nequalty ncrease as a result of the shft of people from the large, relatvely poor and egaltaran agrcultural sector to the small, ndustral sector that s rcher but relatvely unequal. In the latter stages, however, as a bulk of the populaton shfts to the urban sector, there s an ncrease n the relatve wages of the poorer workers n both urban and rural sectors, and varous polcy measures are also mplemented to reduce ntra- and ntersectoral nequalty. Therefore, overall ncome nequalty n the economy decreases n the latter stages of development. One mplcaton of the Kuznets hypothess s that f, n early stages, economc growth leads to more nequalty, then poverty mght take many years to decrease n the developng world. In recent years, a number of theoretcal studes has tred to examne the lnks between nequalty and growth. Rather than focusng on the Kuznets hypothess, the renvgated nterest n the endogenous growth theory has provded substantal research nto the exploraton of the mpact of nequalty on growth. Although almost all the studes undertaken on the topc show a negatve effect of nequalty on growth, the channels through whch ths effect s transmtted dffer n accordance wth the model used. There are sx man famles of models whch explore the lnks between growth and nequalty: the poltcal economy model (PE), the captal market mperfecton model (CM), the ntegrated model (INT), the soco-poltcal nstablty model (PI), the fertlty/educaton ssue model (FE), and the socal comparsons model (SC). We now turn to brefly revew these models. PE model (Alesna and Rodrk, 994; Persson and Tabelln, 994; Bertola, 993; Perott, 992): Ths model tres to buld a brdge between theores of endogenous growth and theores of endogenous poltcal economy. In democratc socetes, the level of taxaton s decded by the medan voter. Taxaton s assumed to be proportonal to ncome, and publc expendture progressve as tax revenues are redstrbuted lump-sum to everyone. Hence, the beneft receved by the poor s greater than the benefts receved by the rch. Thus, the poor would prefer a hgh level of taxaton-redstrbuton. Snce n unequal socetes the ncome of the medan voter s slower than the mean ncome, majorty rule would dctate a hgh level of redstrbuton whch n turn dscourages nvestment by depressng ts net return and lowers growth. The negatve mpact of nequalty would be attenuated by the degree of wealth bas of the system aganst the poor. The more a socety moves away from the democratc archetype of one man, one vote, the less t s possble to reduce the level 3

5 of nequalty through redstrbuton. CM model (Chou, 998; Aghon and Bolton, 997; Galor and Zera, 993; and Sant-Paul and Verder, 993): Ths model s based on the role played by mperfectons n the captal markets. Specfcally, n socetes where agents do not have free access to borrowng, nequalty mples that a relatvely large share of the populaton s below the threshold cost of educaton. Therefore, nvestment n captal human captal s low, and f growth s enhanced by nvestment n human captal, growth s low too. Redstrbuton ncreases total output and growth because t allows the poor to nvest n human captal. If captal markets tend to mprove as an economy develops, then the effects related to captal-market mperfectons are more mportant n poor economes than n rch ones. Therefore, the predcted effects of nequalty on economc growth would be larger n magntude for poor economes than for rch ones. It s also noted that the credt market mperfecton arguments are actually better suted to explan the relatonshp between poverty rate and economc growth. Whle hgher nequalty does not always mply that a larger fracton of the populaton s too poor to gan access to credt, a hgher poverty rate unambguously means that more people are credt-constraned. For example, nequalty n an economy could be hgh even though all the people n the economy are relatvely well off. Therefore, we should expect a negatve relatonshp between poverty rate and economc growth. INT model (Benabou, 996): Ths model provdes an ntegrated framework n whch the mpact of redstrbuton on growth s not necessarly lnear. There are two opposte effects. Redstrbuton s good f publc expendture goes to fnance educaton n a world wth mperfect captal markets, and bad f t only transfers ncome from the rch to the poor because t depresses the net return to nvestment of the rch. Therefore, growth s nverted- U shaped wth respect to redstrbuton and dstrbuton s U-shaped wth respect to nequalty. PI model (Alesna et al. 996; Benhabb and Rustchn, 996; Grossman and Km, 996; Fay, 993): Ths model emphasses the consequence of nequalty on poltcal nstablty and socal unrest. Accordng to the PI model, nequalty s an mportant determnant of soco-poltcal nstablty and ths has negatve effects on growth through lower expected returns to nvestment. Specfcally, nequalty exacerbates socal conflct whch n turn makes property rghts less secure and reduces growth. Moreover, the partcpaton of the poor n crme and other ant-socal actons represents a drect waste of resources because the tme and energy of the crmnals are not devoted to productve efforts. Defensve efforts by potental vctms represent a further loss of resources. FE model (Perott, 996): Accordng to the FE model, nequalty has a negatve effect on economc growth through the dstorton of the households decsons on educaton and fertlty. Parents have to optmse the use of the household s resources, alternatvely 4

6 through an mprovement n qualty (educaton) or n quantty (fertlty) of ther offsprng. Snce educaton has a cost equal to the ncome foregone whle at school, poor households do not nvest n human captal but n the quantty of chldren. However, growth s only enhanced by nvestment n human captal, therefore, ceters parbus, a socety n whch there s hgh nequalty presents a relatvely large number of poor households whch nvest n quantty rather than educaton. The hgh fertlty rate of ths socety leads to low growth. SC model (Knell, 998): Ths model s bult on the Benabou model (996), n whch ndvduals make socal comparsons. The model s based on the assumpton that maxmsaton of ndvdual utlty does not depend solely on own consumpton but also on the average consumpton of some reference group. In an unequal socety, poor households are tempted to conform to the norms and to fulfl socal needs and expectatons by nvolvng n hgher consumpton actvtes and by lowerng nvestment n human captal n order to reduce the gap wth rch households. These actvtes maxmse present welfare but go to the detrment of future welfare and growth. As the above dscusson shows, nequalty and poverty can affect growth through varous mechansms that often work n opposte drectons. It s not possble to predct whch mechansms are domnant by usng theory alone. Emprcal nvestgaton s therefore the key to understandng the relatonshp between nequalty, poverty and growth. 3. Emprcal framework The purpose of the paper s to explore the relatonshp between ntal nequalty, poverty and growth at the provncal level of Vetnam. As the prevous emprcal studes, the followng specfcaton s used: GROWTH = β + β 2 INEQ + β 3POV + β X + ε () Where GROWTH s the average growth rate of GDP n a provnce over the perod under consderaton, INEQ s a measure of ntal nequalty, POV s the rate of poverty and X s a vector of control varables. Poverty rate (POV) s defned as the fracton of people below the poverty lne. The poverty lne corresponds to the expendture (ncludng the value of home producton and adjusted regonal and seasonal prce dfferences) requred to purchase 200 Kcal per person per day usng the food basket of households n the thrd quntle, plus a non-food allowance equal to what households the thrd quntle spend on non-food tems. The percepton of nequalty depends on the nequalty ndex used. Indces are nether cardnally nor ordnally equvalent and some of them lack basc propertes that good ndces should have. We take ths problem nto consderaton by runnng the econometrc 5

7 tests to alternatve specfcatons of the varable INEQ. In ths paper, we use two measures of nequalty - the GINI coeffcent and the rato of ncome shares gong to the top and bottom quntles of the populaton. The GINI coeffcent s the most wdely used aggregate measure of nequalty for the whole populaton n an economy. It s based on the Lorenz curve, whch descrbes the cumulatve dstrbuton of ncome (or expendture) as a functon of cumulatve dstrbuton of households (Cowell, 995). More specfcally, the GINI coeffcent s the area above the Lorenz curve and below the dagonal 45 degree lne dvded by the area under the dagonal lne. The Vetnam lvng standard household survey data (VHLSS) nclude the number of households n dfferent expendture ranges and the expendture of each ndvdual household. Usng nformaton about the proporton of people below dfferent levels of per capta expendture, the GINI coeffcent can be derved as follows: N GINI = 2 ( P + P + ) ( X + X + ) ( P + P ) (2) = 2 2 Where P s cumulatve share of the populaton for nterval and X s the cumulatve share of expendture for nterval. The frst term n the large parentheses s the heght of each slce, from the dagonal lne down to the Lorenz curve, whle the last term n small parentheses s the wdth of each slce. In ths paper, nequalty of consumpton expendture rather than ncome s analysed because consumpton s consdered a better measure of human well-beng. Also, relatve to the ncome varable, consumpton shows less volatlty as t s not so vulnerable to dosyncratc shocks. Accordng to Deaton (997), consumpton data are less nfluenced by measurement errors, partcularly for rural households. The man drawback of the GINI coeffcent s that because t s a measure of aggregate nequalty, t fals to completely capture certan changes n the underlyng ncome dstrbuton. Dennger and Squre (996) recommend reportng nformaton on the ncome shares by quntles to overcome ths problem. Drawng from ther suggeston, we supplement each GINI measure by the correspondng rato of the total ncome gong to the rchest and the poorest 20% (quntles) of the populaton. Apart from INEQ and POV varables, our vector of control varables follows a standard approach wdely used n the lterature and ncludes GDP per capta level (GDPPC) at the begnnng of the perod (to check for the convergence hypothess), the average rato of nvestment to GDP over the perod (INVEST), and a measure of human captal (HUMCAP). The best proxy for the measure of human captal has been consdered s the average schoolng years n the adult populaton. The other varables taken nto 6

8 consderaton such as the enrolment rato n prmary, secondary or hgher educaton can nstead be consdered as proxes for human captal. Observe that n the above framework, we are studyng how the ntal values of the ndependent varables INEQ, POV and GDPPC are related to subsequent GDP growth rate. Ths framework avods problems of endogenety or smultanety bas assocated wth these four varables. Whle t s true that GDPPC can affect subsequent nequalty and poverty rate, ths s a recursve relaton rather than a smultaneous equaton relaton and hence does not nfluence the estmaton of equaton. Whle several other varables have been suggested to be lnked to growth, we decde to keep the vector of control varables relatvely small, n the dffcult exercse of balancng the rsks of multcollnearty wth the rsks of omtted varable bas. However, we test the model wth several alternatve confguratons of the controls. The basc model only ncludes GDPPC, INVEST, and HUMAN. In fact, many of the other varables are found to be hghly correlated to GDP, to HUMAN, to POV or to INEQ. The data used n ths paper come from the General Statstc Offce of Vetnam. The data cover a total of 6 provnces of Vetnam. Data on GDP growth rate, nvestment, human captal, trade openness, GDP per capta at the provncal level are only avalable from 996 to Therefore, we are able to examne the relatonshp between ntal nequalty, poverty and growth for the perod of 996 to Data on poverty rate and the GINI coeffcent are computed by usng data from the four Vetnam lvng standard household surveys whch were mplemented by GSO n 993, 998, 2002, and The basc statstcs of the varables are provded n Table n Appendx. Accordng to the fgure n the table, whle mean GDP per capta and nequalty ncreased between 996 and 2004, the poverty rate decreased durng ths perod. Table 2 n Appendx provdes the correlaton matrx of varables used n ths paper. As shown, the correlaton of GDP growth rate wth ndependent varables s weak; there s a weak possblty that the sgns of these relatonshps may change when the regresson n equaton s estmated. 4. Emprcal results on the lnks between poverty, nequalty and growth The regressons, 2 and 3 n Table are based on equaton, but they test the relatonshp between growth rate of GDP (the dependent varable) and each of the two provncal characterstcs under consderaton ntal poverty rate and ntal nequalty one at a tme. All the estmatons are performed usng OLS wth Whte s correcton for heteroscedastcty. Regresson, whch consders only POVRATE 996, shows a negatve and statstcally sgnfcant relatonshp between POVRATE 996 and subsequent GDP growth rate. Ths mples that provnces wth hgher ntal poverty rate experence slower GDP growth rate. 7

9 Regressons 2 and 3 n Table show that coeffcents on both GINI and TOPBOTTOM are negatve, but not sgnfcant. Ths mples that ntal nequalty does not affect subsequent economc growth. Ths can be explaned by the fact that there are many opposng mechansms through whch nequalty can affect growth. The results presented here ndcate that, on the whole, the mechansms through whch nequalty leads to hgher growth are offset by the opposng mechansms through whch nequalty harms growth. Regressons 4 and 5 n Table nclude varables of all provncal characterstcs as ndependent varables n equaton. Lke the results n regressons, 2 and 3, the results n regressons 4 and 5 do not show the relatonshp between ntal nequalty and growth, but they support the fndng that hgh poverty rate s negatvely assocated wth subsequent growth rate of GDP. In all the regressons, GDPPC 996 s postvely and sgnfcantly related to the subsequent growth rate of GDP. Ths ndcates that on the average, provnces wth hgh ntal GDP per capta have a hgher GDP growth rates. Ths result s surprsng n that ts does not support the theory of condtonal convergence among economes (Barro, 99; Mankw et al., 992; Sala-- Martn, 996). Ths can be explaned by the fact n Vetnam that durng the transton perod from a centrally plannng economy to a market economy, economc opportuntes are emboded by rch provnces wth hgh level of ncome per capta and resource advantages. As a result, these provnces often sustan hgher GDP growth rate than the others. Smlarly, the vtal role of human captal and nvestment s shown n all the regressons as the coeffcents on HUMCAP and INVEST are postve and statstcally sgnfcant. Table : The mpact of poverty and nequalty on growth Explanatory varable POVRATE (0.0)** (0.0)** (0.0)** GINI (3.96) (3.9) TOPBOTTOM (0.7) (0.7) GDPPC (0.00)* (0.00)** (0.00)** (0.00)* (0.00)** INVEST

10 (0.0)*** (0.0)** (0.0)** (0.0)*** (0.0)** HUMCAP (0.03)** (0.03)* (0.03)* (0.03)** (0.03)** Intercept (3.76) (4.7) (3.92) (4.06) (3.82) N R squared The dependent varable s the average growth rate of GDP between 996 and Standard errors are n parentheses. *, **, and *** denote sgnfcance at 0%, 5% and % level respectvely. 5. Determnants of poverty and nequalty The prevous secton sheds lght on how ntal nequalty and poverty rate are related to subsequent GDP growth rate. Ths secton turns to examne the determnants of poverty and nequalty n Vetnam. As the prevous emprcal studes, the followng specfcaton s used: POVRATE = δ + α X + ε INEQUALITY = δ + α X + ε Where POVRATE and INEQUALITY are the rate of poverty and nequalty n provnce respectvely. X s vectors of control varables. Our vector of control varables follows a standard approach wdely used n the lterature and ncludes the growth rate of GDP per capta level (GDPPC), a measure of ntal nequalty (INEQ), the rato of nvestment to GDP over the perod (INVEST), trade openness (OPENNESS) and a measure of human captal (HUMCAP). The best proxy for the measure of human captal has been consdered the average schoolng years n the adult populaton. The other varables taken nto consderaton such as the enrolment rato n prmary, secondary or hgher educaton can nstead be consdered as proxes for human captal. Trade openness s measured as the rato of exports and mports to GDP. The percepton of nequalty depends on the nequalty ndex used. Indces are nether cardnally nor ordnally equvalent and some of them lack basc propertes that good ndces should have. We take ths problem nto consderaton by runnng the econometrc tests to alternatve specfcatons of the varable INEQ. We use two measures of nequalty - the GINI coeffcent and the rato of ncome shares gong to the top and bottom quntles of the populaton (TOPBOTTOM). 9

11 The results n Table 2 show that the coeffcent on the growth rate of GDP per capta (GDPPC) s negatve and statstcally sgnfcant n both regressons, and the correspondng standardzed coeffcent n each case s also larger than the standardzed coeffcents for most of the other ndependent varables. Thus, these regressons ndcate that a hgher growth rate of GDP per capta s assocated wth lower rate of poverty n the future, a fndng that s consstent wth the wdely held vew among economsts. The coeffcents on the GINI coeffcent (GINI) and the rato of ncome shares gong to the top and bottom quntles of the populaton (TOPBOTTOM) are postve and sgnfcant. Ths mples that provnces wth a hgher level of nequalty wll have a hgher rate of poverty n the future. Ths result also supports the dea that poverty reducton depends on the level of nequalty. Another result of the regressons s that provnces wth hgher level of human captal and hgher level of openness have, on the average, lower end-of-perod poverty rate. The postve mpact of nvestment on poverty rate mples that provnces wth hgher poverty rate have a hgher rato of nvestment to GDP. Table 2: Determnants of poverty Explanatory varable 2 GDPPC GINI 996 TOPBOTTOM 996 INVEST HUMCAP OPENNESS Intercept N R squared (0.9)* 0.30 (0.8)* 0.23 (0.04)*** (0.05)* (0.0)*** 5.0 (.02) (0.20)* 0.58 (0.28)* 0.23 (0.04)*** -0. (0.06)* (0.0)*** 4.27 (8.93)

12 The dependent varable s the rate of poverty n Standard errors are n parentheses. *, **, and *** denote sgnfcance at 0%, 5% and % level respectvely. The regresson results n Table 3 present the determnants of nequalty. As shown, the coeffcents on GINI 996 and TOPBOTTOM 996 are sgnfcant and have a standardzed value that s much larger than all other standardzed coeffcents. Thus, ntal level of nequalty s, by far, the most mportant determnants of future nequalty. Smlarly, as we can see from the Table 3, the ntal poverty level s clearly the mportant determnant of end-of-perod nequalty. Also, as shown n the prevous secton, hgher ntal nequalty contrbutes to hgher end-of-perod poverty. These fndngs suggest that a reducton n poverty could help reduce future nequalty and vce versa. Another fndng s that provnces wth hgher ntal GDP per capta have a hgher nequalty n the future. Trade openness s postvely and sgnfcantly related to nequalty, mplyng that provnces wth hgher level of openness have a hgher future nequalty. Table 3: Determnants of nequalty Explanatory varable GINI 2004 TOPBOTTOM 2004 GINI (0.)** TOPBOTTOM GDP POVRATE 996 INVEST GDPPC 996 HUMCAP OPENNESS (0.00) 0.6 (0.05)*** 0.0 (0.02) (0.000)** 0.0 (0.05)** 0.03 (0.0)** (0.07)** (0.05) 0.05 (0.02)** 0.00 (0.009) 0.00 (0.0005)** 0.0 (0.0) 0.03 (0.0)**

13 Intercept N R squared 0.06 (0.06) (0.47) The dependent varables are GINI 2004 and TOPBOTTOM Standard errors are n parentheses. *, **, and *** denote sgnfcance at 0%, 5% and % level respectvely. 6. Concludng remarks Ths paper has analysed the relatonshp between growth, poverty and nequalty n Vetnam from 996 to 2004 by usng the provncal data and data from household lvng standard surveys. The most robust result has been the negatve assocaton between poverty rate and subsequent GDP growth rate. Moreover, hgher ntal poverty level leads to hgher nequalty n the future. The emprcal results show no evdence of the relaton between nequalty and the growth rate of GDP. As dscussed n secton 4 of the paper, there are a number of arguments we can make to explan the observed relatonshp. More specfcally, there are many opposng mechansms or channels through whch nequalty can affect growth. The results presented here ndcate that, on the whole, the mechansms through whch nequalty leads to hgher growth are offset by the opposng mechansms through whch nequalty harms growth. The next step n ganng a deeper understandng of the nequalty-growth relatonshp would nvolve solatng and testng these varous mechansms to determne ther relatve mportance to the growth process. Such an undertakng, however, s beyond the scope of ths paper, and must be set asde for future research. Gven emprcal fndngs n ths paper, we can pont out some polcy mplcatons. Frst, t s necessary to pont out that no relatonshp between nequalty and GDP growth rate does not mean that redstrbutve polces are undesrable. But t does mply that when GDP growth s major objectve, we should pursue redstrbutve polces wth cauton and choose polces that are least dstortonary n the economc sense. Second, the results present good news to those who see poverty elmnaton as an ntegral part of economc development polcy. They ndcate that reducng poverty, apart from helpng the most vulnerable segments of socety, s benefcal from the perspectve of mprovng the future dstrbuton of ncome and promotng economcs growth. In other words, there s defntely no trade-off between poverty reducton and economc growth. Thrd, the results suggest that lower nequalty leads to lower poverty rate and poverty reducton could help to reduce nequalty, therefore concentratng on poverty elmnaton wll also help us to buld a more equtable socety wthout sacrfcng economc growth. Pursut of ant-poverty polces, 2

14 therefore, appears to be a very desrable poltcal objectve. References Aghon, P., Carol, E., and Garca-Penalosa, C. 999, Inequalty and economc growth: The perspectve of the new growth theores, Journal of Economc Lterature, 37. Alesna, A and Perott, D. 996, Income dstrbuton, poltcal nstablty and nvestment, European Economc Revew, 40 (6). Alesna, A and Rodrck, D. 994, Dstrbutve poltcs and economc growth, Quarterly Journal of Economcs, 09. Barro, R. 999, Inequalty, growth and nvestment, NBER workng paper, 7038, Massachusetts. Benabou, R. 996, Inequalty and growth, NBER Macroeconomcs Annual. Benhabb, J and Speegel, A. 996, Socal conflc and growth, Journal of Economc Growth, (). Bertola, G. 993, Factor shares and savngs n endogenous growth, Amercan Economc Revew, 83(5). Chu, W. 998, Income nequalty, human captal accumulaton and economc performance, The Economc Journal, 08 (446). Deaton, A. 997, The nanlyss of household surveys: a mcro-econometrc approach to development polcy, Johns Hopkns Unversty Press, Baltmore and London. Dennger, K. and Squre, L. 998, New ways of lookng at old ssues: Inequalty and growth, Journal of Development Economcs, 57. Galor, O and Zera, J. 993, Income dstrbuton and macroeconomcs, Revew of Economc Studes, 60. Knell, M. 998, Socal comparsons, nequalty and growth, Mmeo, unversty of Zurch. Kuznets, S 955, Economc growth and ncome nequalty, Amercan Economc Revew, 45 (). L, H and Zou, H. 998, Income equalty s not harmful for growth: theory and evdence, Revew of Development Economcs, 2. Perott, M. 993, Poltcal equlbrum, ncome dstrbuton and growth, Revew of Economc Studes, 60(4). Persson, T and Tabelln, G. 994, Is nequalty harmful for growth? Amercan Economc Revew, 84(3). 3

15 Sala--Martn, X. 997, I just run two mllon regressons, Amercan economc revew, Proceedng and Papers, 87(2). Vetnam Government Statstc Offce, varous year, Vetnam Lvng Standard Survey, Hano. Appendx Table : Descrptve Statstcs Varable Mean Mnmum Maxmum Std. Dev. GDP GDPPC POVRATE POVRATE GINI GINI TOPBOTTOM TOPBOTTOM INVEST HUMCAP OPENNESS

16 Table 2: Correlaton matrx for the varables used X X2 X3 X4 X5 X6 X7 X8 X9 X0 X X: GDP X2: GDPPC X3: POVRATE X4: POVRATE X5: GINI X6: GINI X7: TOPBOTTOM X8: TOPBOTTOM X9: INVEST X0: HUMCAP X: OPENNESS