Education and Earnings in Lao PDR: Regional and Gender Differences. Phanhpakit ONPHANHDALA Terukazu SURUGA

Transcription

1 GSICS Workng Paper Seres Educaton and Earnngs n Lao PDR: Regonal and Gender Dfferences Phanhpakt ONPHANHDALA Terukazu SURUGA No. 4 July 2006 Graduate School of Internatonal Cooperaton Studes Kobe Unversty

2 Educaton and Earnngs n Lao PDR: Regonal and Gender Dfferences* Phanhpakt ONPHANHDALA 1 and Terukazu SURUGA 2 Abstract: Major economc reforms to move from a centrally planned to a market economy n Lao PDR snce 1986 have exerted a number of mpacts on the labor market, relatve earnngs, and returns to educaton. It s too soon to assess the full mpact of these reforms, but ths paper represents the frst analyses of the monetary benefts of schoolng n Lao PDR, on the bass of wage earnngs data collected n LECS 2 n 1997/98. The estmated rates of returns n Lao PDR are at approxmately the same level wth other transton economes, but relatvely low by nternatonal standards. There are sgnfcant dfferences n the returns to schoolng among regons and between genders. For example, the rate of return to schoolng s 4.5% n Ventane Captal, and only 2% n other provnces. Males are found to experence lower returns to schoolng than do females n Ventane Captal, but opposte results preval n other provnces. Key words: Educaton, Returns to Schoolng, Regonal dsparty, Gender, Lao PDR 1 Graduate School of Internatonal Cooperaton Studes, Kobe Unversty, Japan. 2 Professor of Economcs, Graduate School of Internatonal Cooperaton Studes, Kobe Unversty, Japan. Correspondng author, emal address: suruga@kobe-u.ac.jp.

3 I. Introducton The human captal theory vews wage earnngs of a worker as returns to hs or her knowledge captal that have developed through years of schoolng and work experence. Early works of Becker and Chswck (1966) lad the foundaton of ths lterature, and later Mncer (1974) proposed an emprcal approach to dstngush the contrbutons of schoolng and experence n wage earnngs. The model of Mncer, whch s well known as Human Captal Earnngs Functon, ncluded the log of ndvdual earnngs as the explaned varable, and schoolng and experence as explanatory varables. Recent studes of educaton and wage determnaton are almost always embedded n the framework of Mncer, and combned wth dfferent contexts hghlghtng the addtonal mportance of household and communty characterstcs, nsttutonal factors, and other observable attrbutes such as gender, race, and relgon. Wlls (1986) provdes a survey of ths lterature, and Card (1999) surveys the lterature on the causal relatonshp between educaton and earnngs. On the other hand, Schultz (1988) surveys the work done n the context of developng countres. A detaled account of educatonal attanment and earnngs across countres can be found n Psacharopolous and Patrnos (2002). Ths paper focuses on Lao People s Democratc Republc (Lao PDR). Snce the major economc reforms (the polcy known as Chntanakaan Ma or New Economc Mechansm) amng at transferrng from a centrally planned to a market economy n 1986, the economc transton has led to certan ssues n the labor market. The educaton systems that served n a command economy wll have to be adapted to serve the needs of a market economy. Ths s reflected n hgh levels of mprovement n adult lteracy and school enrollment, especally a hgh ncrease n female lteracy rate (15 years old and over) from 43 percent n 1990 to 61 percent n the early 2000s (Table 1), and a sharp ncrease n ether gross secondary school enrollment or tertary enrollment rate n both sexes from 1990 to Many adults who were llterate have had access to prmary school, whch can be observed from the hgh gross prmary school enrollment rato over 100 percent n both sexes n 2002 (Table 2). Openng up the economy has provded many new ncome-earnng opportuntes, thus ncreasng the opportunty cost of schoolng and perhaps reducng rates of return to schoolng. Table 1: Structure of Lteracy Rate (%) by Sex. Female Male Pupls Startng 50.0 a 65.0 b 56.0 a 64.0 b Grade 1 who Reach Grade 5 Lteracy Rate, 15 Years and Over Lteracy Rate Years Old Source: ADB, Key Indcators Note: a refers to b refers to

4 Table 2: Structure of School Enrollment Rato (%) by Sex. Female Male Gross Pre-Prmary School Gross Prmary School Gross Secondary School Gross Tertary a 7.0 Source: ADB, Key Indcators Note: a refers to Educaton s an mportant determnant of earnngs n market economes. The hgher one s level of educaton, the hgher that person s startng salary s and the steeper the rse n earnngs durng the early workng lfe. The evdence from centrally planned economes and transton economes s relatve lmted. Nevertheless, t has been shown that the rates of returns to schoolng are usually low n centrally planned economes, for example, n Chna (Xe and Hannum 1996, and We, Tsang, Wu and Chen 1999), and also n the early state of transton economes, for example, n Vetnam (Moock, Patrnos and Venkataraman 1998). But they tend to ncrease as market reforms take hold, for example, n Hungary (Varga 1995), Slovena (Orazem and Vodopvec 1995), and Czech Republc and Slovaka (Chase 1998). Ths paper examnes the frst analyses of the returns to educaton n Lao PDR on the bass of wage earnngs data collected n Lao Expendture and Consumpton Survey 2 (LECS 2) n 1997/98. It s attempted to estmate dfferental returns to schoolng for Ventane Captal 1 and other provnces, and for males and females. It s hoped to provde a bench mark to examne the effects of market lberalzaton and educaton reforms n Lao PDR. II. Recent Economc Developments, Educaton, and Earnngs A. Growth and Employment Snce the major economc reforms, the economy of Lao PDR has expanded remarkably wth an annual real growth rate of 6.2 percent from 1990 to Also, the Lao economy was undergong a notable degree of structural change (Table 3). The share of agrcultural sector n GDP decreased by 12.6 percent from 61.2 n 1990 to 48.6 n 2003, the share of ndustral sector ncreased sharply by 11.4 percent from 14.5 n 1990 to 25.9 n 2003, and the share of servce sector was almost unchanged at 26 percent. On the other hand, the agrcultural sector remans predomnantly subsstence n nature; nonetheless, t performed qute well wth an annual growth rate of 4.7 percent. The ndustral sector performed very well wth an annual growth rate of 11.5 percent, and the servce sector also performed well wth an annual growth rate of 6.1 percent. 1 Ventane Captal s the captal cty of Lao PDR, and s not located wthn Ventane Provnce. Ventane Provce s categorzed as one of the other provnces outsde Ventane Captal n ths study. 2

5 Accordng to the Lao Labor Market Indcators (Syhalad, et al 2005), t shows that eght out of ten employed persons were workng n agrculture, huntng, forestry and fshng durng Ths dd not change very much over the three years. However, the percentage n 2003 (82 percent) was lower than n 1995 (86 percent). The proporton of persons employed n ndustry was almost the same as n servces at around 9 percent n The numbers of people employed n ndustry and servces ncreased slghtly over the three years. Thus, the agrcultural sector accounts for 49 percent of the GDP compared to 82 percent of the employed populaton. The result s low productvty and low ncomes placng a large number of the employed populaton among the workng poor. Table 3: Structure of GDP and Employment by Sector and Sex. Share of GDP (%) a Employment, Female (%) b Employment, Male (%) b Agrculture Industry Servce Source: a ADB, Key Indcators b Syhalad, et al 2005, Lao Labor Market Indcators However, t s worth notng that the data for the proporton of the employed n agrculture s somewhat msleadng snce many workers classfed as employed n agrculture have secondary jobs n off-farm actvtes such as household busnesses. Accordng to the LECS 2 (1997/98), 17 percent of rural households operate a busness. Over one-half of the household busnesses were operatng on trade actvtes, especally as retals on a mcro or small scale. Many rural households produce textles for sale. On the other hand, t s also unclear whether self-employed and unpad famly workers wll flow to whch sector on each survey. In LECS 2, t shows that there was 13 percent of the employed populaton classfed as self-employed, and another 10 percent as unpad famly workers. Only 10 percent was classfed as pad employment (Table 4). Moreover, accordng to the Lao Labor Market Indcators (Syhalad, et al 2005), 56 percent of the employed populaton was self-employed and another 26 percent was classfed as unpad famly workers n Many people work on famly farms and n the nformal sector. Together, self-employed workers and contrbutng famly members accounted for 82 percent of the employed populaton. Offcal estmates for 2003 show that only 14 percent were n pad employment, and a much smaller percentage (4 percent) were classfed as prvate employers. 3

6 Table 4: Economc Actvtes (Last 7 days) for Populaton 10+ by Regons n 1997/98. Pad Employee Employer Self- Employed Subsstence Farmer Unpad Famly Worker Ventane C Northern Central Southern All Source: LECS 2 n 1997/98. B. Poverty and Inequalty Transmttng the benefts of economc growth to the poor s essental to reduce poverty and rase the welfare of the most vulnerable. Whle poverty reducton requres economc growth, nternatonal evdence shows that growth alone s not suffcent to reduce poverty substantally. Gven the lack of data, there s a very lmted number of studes that concern the regonal dspartes n Lao PDR. Nevertheless, Bourdet (1998) has hghlghted the dspartes among the provnces and the regons, and evaluated the reform polcy n force n Laos snce the md-1980s. It s shown that there was a large dfference n GDP per capta between the provnces n The wealthest provnces of Ventane Muncpalty (or Ventane Captal n the present) and Ventane provnce were more than two tmes larger than that of the poorest provnces of Huanphanh, Luangnamtha, and Phongsaly n the northern regon. On average, the GDP per capta of the central regon s 7% greater than that of the southern regon and 62% larger than that of the northern one. A large part of the country s populaton lves outsde of the market economy and, as a result, the postve effects of growth were largely felt n urban areas and n lowlands. Evdence from LECS 1 n 1992/93 and LECS 2 n 1997/98 shows that the poverty ncdence (the percentage of the populaton wth consumpton below the poverty lne) dropped from 45.0 percent to 38.6 percent (Table 5). Ths happened at a tme when GDP growth averaged by 7.0 percent. In a smlar perod, the ncdence of poverty n Vetnam dropped from 58 percent to 37 percent, albet wth an economc growth rate of 9.0 percent. Durng the same perod nequalty ncreased n Lao PDR, wth the Gn ndcator ncreasng from a relatvely equtable 28.6 to Ths s stll qute equtable compared wth many other countres n the regon; for example, Thaland has a Gn ndcator around 50, ndcatng a substantal gap between wealther and poorer households. As economc growth ncreases, the poor n Lao PDR face the rsk of fallng nto a poverty trap. Households wth human captal are n the best postons to take advantage of the opportuntes generated by economc growth. Those wthout human captal may fnd themselves ncreasngly fallng behnd wealther households, lackng the resources to nvest n human captal for the future. Whle past economc growth n the country has benefted the poor, the bggest ganers have been the non-poor. Kakwan and Perna (2000) estmate the elastcty between economc growth and poverty reducton s approxmately 0.7, meanng that on average, a 1 percent ncrease n consumpton growth wll 4

7 only reduce poverty by 0.7 percent. In the case of Lao PDR, economc growth does not contrbute substantally to poverty reducton and the postve effect s largely offset by ncreases n nequalty. Ths s low compared wth growth-poverty elastcty n other countres n the regon, such as Thaland. As n most countres, the poor are located predomnately n rural areas. The topography and the low populaton densty of Lao PDR make many rural areas especally solated and reduce the opportuntes that the poor have to escape poverty. In terms of the percentage of the populaton lvng n poverty, the northern regon s the poorest n the country. Not surprsngly, Ventane Captal and the central regon have the lowest ncdence of poverty. Not only does Ventane Captal have the lowest ncdence of poverty n the country, t also saw the largest declne n poverty between the two LECSs. Poverty n Lao PDR s largely a problem located outsde of the major ctes. Although poverty has declned n all regons of the country, ths declne has been slow. Table 5: Percentage of the Populaton Lvng n Poverty, 1992/93 and 1997/98 Regon 1992/ /98 Urban Rural Total Urban Rural Total Ventane C Northern Central Southern All Source: Kakwan et al C. Educaton and Government Expendture Accordng to ADB key ndcator 2005, a large amount of government expendture has been expended on economc servces, whch consst of transport and communcaton, ndustry, agrculture, etc. It s concentrated on road mprovement and other economc nfrastructure. A very small amount of budget has been allocated to socal nfrastructure. Only about 1 percent of the budget was spent on the educaton sector n the early 1990s. From 1997 to 2001, the expendture on educaton has ncreased drastcally to an average of 8 percent. The educaton system n Lao PDR s categorzed by 5 years of prmary educaton, whch chldren generally enter at age 6; 3 years of lower secondary and another 3 years of upper secondary educaton. A vocatonal educaton program s generally 2 years, and a techncal educaton program s generally 3 years; both lead to a dploma. A hgher educaton (unversty) lasts 4 to 6 years, dependng on the program. A post-graduate system dd not exst untl recent years. Accordng to LECS 2, n all regons outsde of Ventane Captal, the average number of years of schoolng s below 5 whch means that most students do not complete prmary educaton. Although most vllages have ther own prmary school, less than half of the vllages have a complete prmary school and only 11 percent have a lower secondary school. The 5

8 nsuffcent educatonal servces are a hard obstacle for those wthout completed prmary school n rural regons. Males tend to have both a hgher lteracy rate and longer schoolng years n all regons (Table 6). In order to acheve unversal prmary educaton, the government of Lao PDR has spent a large amount of the budget on prmary educaton. Accordng to Lao PDR publc expendture revew (2002), about one half of the government expendture on the educaton sector was concentrated on prmary educaton durng the 1990s. Another one fourth was spent on secondary educaton (Table 7). Gven the lack of data, measurng the effect of government spendng on the welfare of the poor s dffcult. However, a study of the health and educaton needs of ethnc mnortes n Lao PDR showed that educaton spendng dsproportonately benefts urban students. Estmates are that a unversty student receves a subsdy that s 20 tmes larger than the subsdy for a prmary student. Lkewse the ncrease n government captal expendtures probably tends to be based towards wealther households. In addton, t s beleved that ths caused the schoolng gap between Ventane Captal and other provnces ncrease, contrbutng to the overall ncrease n nequalty. Table 6: Educatonal Servces and Educaton by Regons and Sex n 1997/98. Prmary Sch. n Completed Lower Schoolng Lteracy15+ Vllage Prmary Sch. Secondary F M F M Ventane C North Center South All Source: LECS 2 n 1997/98. Table 7: Government Expendture on Educaton by Levels (%) / /00 Pre-Prmary Educaton Prmary Educaton Lower Secondary Educaton Upper Secondary Educaton Techncal and Occupatonal Tranng Teacher Tranng Hgher Educaton Mantenance Total Source: WB, IMF, ADB (2002), Lao PDR Publc Expendture Revew: Country Fnancal Accountablty Assessment. D. Employment and Wages Lttle quanttatve data s currently avalable on wages and salares (Table 8). Government salares appear to be well below the market level and salary ncreases are largely gven as admnstratve rewards rather than as adjustments to market condtons. Durng the most recent bout of hgh nflaton (1998 to 6

9 1999), publc salares were only adjusted once and n real terms fell dramatcally. Evdence from qualtatve surveys durng ths perod show that salares n state-owned enterprses and the prvate sector are substantally above those n the government, and that these salares ncreased substantally faster than those n the publc sector. The salary scale n the government s qute flat, wth the salary of top offcals about twce that of the low pad ndvduals. A top government offcal mght earn only one tenth of the salary pad for a smlar poston n a prvate enterprse. Table 8: Range of Monthly Wages 2 n Selected Occupatons, Ventane, Occupaton Monthly Salary/Wage n Kp Prvate Sector Garment Workers 29,000 90,000 Restaurant Workers 20,000 70,000 Motor Vehcle Mechancs 30,000 75,000 Unsklled Constructon Workers 30,000 50,000 Sklled Workers 40,000 60,000 Brewery Workers 30,000 40,000 Supervsory Workers 60,000 80,000 Techncan 40,000 80,000 Managers, Large Frm 140, ,000 Publc Sector Government Employee 18,000 33,000 Producton Worker, State Frm 30,000 70,000 Source: World Bank, Snce the LECS 2 data that s used n ths paper was undertaken from March 1997 to February 1998, t s necessary to note about the mpact of the Asan Fnancal Crss. Although the level of nflaton was hgh compared wth rates reported n other Southeast Asan countres and probably has some harmful effect on economc growth, nflaton was not a serous problem n Lao PDR untl md Moreover, the wages and salares were not adjusted durng the perod of the survey. Therefore, t s beleved that there s no mpact of the regonal fnancal crss on the earnngs data. III. Data Despte the urgent need for labor market nformaton, the current statstcs n Lao PDR are very lmted. Only two urban labor force surveys were conducted durng 1992 and The 1995 populaton census contans useful data for employment and unemployment. It s also possble to obtan nformaton from the Lao Expendture and Consumpton Surveys of , and Whle the LECS 1 was combned wth a large module of socal ndcators, the LECS 2 and LECS 3 versons focused on economc actvtes of the households. These surveys do not follow nternatonal standards for collectng 2 Average market exchange rate n 1993 was at 716 kp/dollar. 7

10 labor statstcs. Lao labour market ndcators s the only survey that follow nternatonal standards for labor statstcs. Ths, n turn, makes t dffcult to pant a pcture over tme. In ths paper, we attempt to apply the data of LECS 2 as a bench mark to examne the returns to schoolng among wage earners (pad employees) on the focus of regonal and gender dfferences n Lao PDR durng In vew of lmted data on labor statstcs for ths perod, ths survey s deemed very useful to study the mpacts of schoolng on ncome n the post-reform era. Also, as mentoned n the prevous secton, the data collecton of LECS 2 s not affected by the Asan Fnancal Crss. Hence, ths paper s the frst of ts knd to study the return to schoolng for ths crtcal perod the pre-crss perod n the transton. The LECS 2 was undertaken from March 1997 to February 1998, n whch the sample was made up of 8,882 households, 57,624 persons from 450 vllages. In ths survey, there were 1,488 samples classfed as pad employment n the last 12 months. About 2 percent of samples were found to be no educaton workers. These samples were dropped from the analyss because t was felt that the sample was too low to allow a precse estmaton of any effects. After clearng the data, we fnally have 1,315 observatons: 484 from Ventane Captal and 831 pooled from the remanng 17 provnces. 376 observatons were female. There are some lmtatons n the observatons that need to be descrbed. Frst, there were many observatons that dd not complete a level of schoolng. Thus, we categorzed the educaton levels by the followng: 1 to 5 years of schoolng as prmary level, 6 to 11 years of schoolng as secondary level, and 12 years of schoolng or over as post-secondary level. The post-secondary educaton dd not dstngush between vocatonal, techncal or unversty level. Second, the ncome earnngs data n LECS 2 were classfed by the status of pad employment and other characterstcs. Thus, t can not be dstngushed whether pad employment was n the publc sector or prvate sector. Ths causes a problem because publc sector wages typcally do not reflect market wages. Of course, n many countres although fewer now than n the past the majorty of unversty graduates end up n publc sector employment. The concentraton of graduates n publc sector employment s dentfed as a problem n growth studes (Pssardes 2000). Hence, t s expected that the rates of returns to schoolng n ths paper are low by ncludng a part of the publc employment. However, publc employment pay-based rate of return estmates are useful n prvate calculatons regardng the ncentves set by the state to nvest n educaton. Summarzng the data of LECS 2, Table 9 presents the characterstcs of the samples ncludng, age, earnngs (n thousand kp), schoolng years, educaton levels, and experences. The samples are also classfed nto two geographcal areas namely Ventane Captal and the rest of the country. It s worth notng that Ventane Captal alone comprses roughly one thrd of the sample sze, whch could justfy the classfcaton. On average, the schoolng years were roughly 10 years n all regons because the ntervews manly covered pad employment n the urban cty n each provnce. Smlarly, for the same reason, educaton at the prmary and secondary levels vared between 14% and 19%, and 49% and 63%, respectvely. However, 8

11 the hgher the educaton level, the larger the gap was between Ventane Captal and other provnces. Specfcally, the proporton of males n the captal accounted for 34%, whereas that of males n other provnces was only 25%. The average age of the ntervewees was about 35 years old, whch resulted n comparable experence. The earnngs, on the other hand, recorded a sgnfcant dfference between employees n the captal and the rest of the country despte the above-mentoned comparable characterstcs, namely the average monthly earnng n Ventane Captal was 78,000Kp, whereas ths amount was only 62,000 n other provnces. Lke n many other economes, females earn lower rates than women. For example, males earned 28% more than females n Ventane Captal, 24% n other provnces. Table 9: Means of Selected Varables by Regon and Sex. Varable Ventane C. Male Female All Provnces Male Female All All Male Female All Earnngs 3 (1,000Kp) Schoolng Educaton Level (0,1) Prmary (1-5) Secondary (6-11) Post-secondary (12+) Experence Age Observatons N ,315 Source: LECS 2 n 1997/98. Smlar to the gross amount, earnngs classfed accordng to educaton levels also show a clear dfference between regons and among the educaton levels themselves (Table 10). In Ventane Captal, the average earnng of an employee wth prmary educaton was recorded at 63,000Kp, whle ths fgure was 54,000 for the other regons. As expected, the hgher educaton level one acheved, the hgher the wage ncome. Specfcally, for a pad employee n Ventane Captal, the salary ncreased to 71,000 and 99,000 as he/she completed secondary and post-secondary educaton, respectvely. In other provnces, however, all salary levels were about the same as the prmary level n Ventane Captal, regardless of the educaton level reached. In terms of gender dspartes, on average a male employee receved 18% to 33% more than a female employee n each educaton level. Male workers n Ventane Captal earned the hghest wage at 103,000Kp per month, 33% more than male workers n other regons. The dscrepancy s slghtly hgher, 38%, between female employees n the two regons. Wth respect to age-earnngs profles, the data for males n Ventane Captal showed a comparable trend wth the nternatonal standard (Table 11). The earnngs ncreased wth the experences (age), peakng at 3 Average market exchange rate n 1997 was at 1,260 kp/dollar. 9

12 the age of 35-44, and decreasng afterward. The only excepton s that the peak appears to arrve relatvely earler than n many other countres. In contrast, for other provnces, age-earnngs profles showed a small dfference n both sexes. Females acheved the hghest level when a person frst entered the labor force (at age of 15-24) and decreased steadly durng her workng age. Perhaps younger persons were more adaptable to the change n market systems, partcularly among the ncreasng opportuntes n the servce sector. The combned data showed a smlar trend n Ventane Captal due to ts domnance. Table 10: Educaton Level-Earnngs Profles by Regon and Sex. Varable Ventane C. Male Female All Provnces Male Female All All Male Female All Earnngs (1,000Kp) Prmary Secondary Post-secondary Observatons N ,315 Source: LECS 2 n 1997/98. Table 11: Age-Earnngs Profles by Regon and Sex. Varable Ventane C. Male Female All Provnces Male Female All All Male Female All Earnngs (1,000Kp) Age Age Age Age Age 55 or above 82 -* * * 74 Observatons N ,315 Source: LECS 2 n 1997/98. Note: * An nsuffcent number of observatons. IV. Emprcal Model Our basc model s taken from the human captal earnngs functon (Mncer 1974): lny = c + α S + βex + γex + u (1) 2 where Y represents the monthly earnngs of the th ndvdual, S s the years of completed schoolng, Ex represents the number of years an ndvdual has worked snce completng schoolng, and u s a statstcal resdual. In the absence of drect nformaton on experence, Mncer proposed the use of potental experence : the number of years an ndvdual of age A could have worked, assumng he started school at age 6, fnshed S years of schoolng n exactly S years, and began workng mmedately thereafter: 10

13 X A S 6. In ths sem-log specfcaton the coeffcent on S (α ) s nterpreted as the average rate of return to one addtonal year of schoolng, regardless of the level of schoolng. The earnngs functon method s used to estmate average rate of returns to manly dfferent levels of schoolng by convertng the contnuous years of schoolng varable S nto a seres of dummy varables representng the dfferent levels of schoolng, and other ndvdual s characterstcs. After fttng the extended earnngs functon: ln Y = c + α Sec + α Post + βex + γex + θx + u (2) 1 2 where Sec and Post are secondary and post-secondary educaton by ndvdual, and X s dummy varable ndcatng female, marred status, and regons. The omtted category for level of educaton s prmary educaton. Fnally, we attempt to estmate the rate of returns to schoolng per year for each academc year from the age of 11 to 18, wth the completed prmary educaton of 5 years beng the reference pont. After fttng the extended earnngs functon: ln 18 n n 2 Y = c + S + βex + γex + n= 11 α u (3) Snce the data set does not dstngush between vocatonal, techncal and unversty, we opt for ths approach to compare the rate of returns for each academc year. Thus, the rate of returns to schoolng per year for each academc year λ are calculated as dvdng each schoolng coeffcent by the dfferent n between academc year and prmary educaton of 5 years or λ =. n 5 2 α In the emprcal lterature, t s customary to estmate the return to schoolng by ordnary least squares (OLS) or Instrumental Varable (IV) technques. In terms of estmatng the returns to schoolng, Grlches (1977) provdes a landmark survey of the frst-wave lterature. Grlches ponts out that n optmzng models the ablty bas need not be postve and shows that measurement errors n schoolng would lead to a downward bas n the OLS estmate of the effect of schoolng n earnngs (The actual attenuaton of an OLS may be less than 10 percent). Card (2001) revews a set of recent studes that have attempted to measure the causal effect of educaton on labor market earnngs, wth regard to the propertes of OLS and IV estmators. Card ponts out that IV estmated return s as bg as or bgger than OLS (between 20% and 40% above). In ths paper, gven the lack of data - such as famly background nformaton and school qualty - that can be used to ether drectly control for unobserved ablty or as an nstrumental varable for completed educaton, we opt to use only OLS for the estmatons. To correct for possble selecton bas n the earnngs equatons, as seen n many lteratures, we also apply Maxmum Lkelhood (ML) estmates. Snce ML yelds strongly smlar results wth OLS; hence, we smply show OLS results. 11

14 V. Estmaton Results A. Regonal Dfferences The results of econometrc analyss on rate of returns to schoolng are presented n Table 12. Except for experence-squared n other provnces, all the varables are statstcally sgnfcant at the 5% and 1% levels. The results of estmatng a smple earnngs functon show educaton to be a margnal nvestment. An addtonal schoolng year would yeld 4.5% and 2% more earnngs n Ventane Captal and other provnces, respectvely. The returns to one addtonal year of experence would ncrease earnngs by 3.5% n Ventane Captal and 1.2% n other provnces. The estmates for the whole sample are upward based due to the large number of observatons n Ventane Captal as mentoned n Secton III. As can be noted, the regonal dfferences are sgnfcant n these two varables due to the economc structure and workng envronment. Lke many captal ctes n the world, Ventane Captal s the poltcal, economc and cultural center of the country wth the base of the government. The level of development and busness actvtes s by far hgher than any other urban ctes. Ths mples that a person wth a hgher educaton background and experence s lkely to have better opportuntes n fndng a well pad job n a prvate enterprse or an nternatonal organzaton. The salary pad largely reflects the educaton level and experences. On the other hand, wth a relatvely small prvate sector, the publc sector employs a sgnfcantly large porton of the labor force n other provnces, partcularly those n more remote areas. By law, cvl servants n Lao PDR receve the same wage rate regardless of locaton. Ths wage, however, does not reflect the market wages. Therefore, the upper end of educaton level (post-secondary) would not lead to an ncrease n earnngs; rather mply an over-qualfcaton n many cases. The estmates for Lao PDR are low compared wth the returns to educaton estmated for other developng countres. Worldwde, another year of schoolng ncreases earnngs by about 10% (Psacharopoulos and Patrnos 2002). However, estmated returns to schoolng are low n general n centrally planned and transton economes. For example, n Chna the estmates of the returns to schoolng were 3.1% n 1988 (Xe and Hannum 1996), 4.8% n 1991 (We, Tsang, Wu and Chen 1999), and 4.8 percent n 1992/93 n Vetnam (Moock, Patrnos, and Venkataraman 1998). Scarce over-tme evdence, however, seems to be showng that successful reform wll eventually lead to hgher returns. An emprcal analyss of changes n the wage structure n Slovena between 1987 and 1991 reveals that the returns to human captal rse dramatcally durng transton. Workers wth four years of unversty educaton ganed the most n relatve earnngs, followed by those wth two years of unversty. The educaton group that ganed the least, however, relatve to the least educated, were holders of vocatonal degrees (Orazem and Vodovec 1995). In Hungary, the prvate rate of return almost doubled n secondary between 1971 and 1993, and there was a three-and-a-half-fold ncrease n the returns to hgher educaton (Vargaa 1995). Moreover, n Czech Republc and Slovaka between 1984 and 1993, especally n the returns to male workers from 2.4% to 5.2% and 2.8% to 4.9% n Czech Republc and Slovaka, respectvely (Chase 1998). 12

15 Table 13 shows the estmated results of an extended verson of the Mncer model. As dscussed n the model descrpton, dummy varables for varous educaton levels, sex, martal status and regons were ncluded. Except for the experence and martal status n other provnces, all other coeffcents of ndependent varables are statstcally sgnfcant at least at the 10% level. The dummes for secondary and post-secondary levels ndcated a sgnfcant ncrease n earnngs as the educaton level rses. In partcular, t s nterestng to observe that the ncrement n the earnngs at the post-secondary level for Ventane Captal s about 36% compared to the prmary level, and more than double the amount for the same level n other provnces. At the lower level, the fgure vared between 14% and 10% for Ventane Captal and the other regons respectvely. For the combned data, these results are smlar wth the cases of Czech Republc and Slovaka (Chase 1998), and Cote d Ivore (Vjerberg 1993), but a relatvely lower than n Indonesa (Deolalkar 1993). Wth regard to gender dfferences, on average a female would earn roughly 18% to 20% less than a man regardless of her locaton. Ths ssue wll be treated n more detals n the next secton. As expected, marred employees tend to earn more than unmarred counterparts (27% for Ventane Captal). A more detaled categorzaton of data nto the northern, central and southern regons has confrmed our expectaton, that employment outsde Ventane Captal would yeld a lower wage ncome. More specfcally, a worker n the northern, central and southern regons would earn about 28%, 16%, and 21% lower than hs/her counterpart n the captal. These results lend support to the argument of Bourdet (1998). Table 12: Earnngs Functons by Regon. (Dependent varable s the natural log of monthly earnngs) Varable Ventane C. Provnces All Constant *** (105.54) *** (130.78) *** (165.99) Years of Schoolng *** (7.26) *** (4.18) *** (8.43) Years of Experence *** (4.82) ** (1.99) *** (3.99) Experence-squared *** (-3.14) (-1.15) ** (-2.32) R-squared Observatons N ,315 Note: t-statstcs n parentheses. ** Statstcally sgnfcant at the 5% level; *** at the 1% level. 13

16 Table 13: Extended Earnngs Functons by Regon. (Dependent varable s the natural log of monthly earnngs) Varable Ventane C. Provnces All Constant *** (105.14) *** (138.32) *** (171.27) Secondary * (1.85) ** (2.10) ** (2.54) Post-Secondary *** (4.33) *** (3.03) *** (5.04) Experence ** (2.02) (1.51) ** (2.10) Experence-squared * (-1.70) (-1.26) * (-1.72) Female *** (-3.61) *** (-4.54) *** (-5.57) Marred *** (4.33) (-0.90) *** (2.81) Northern *** (-6.20) Central *** (-4.56) Southern *** (-5.53) R-squared Observatons N ,315 Note: t-statstcs n parentheses. The omtted category for level of educaton s prmary educaton, and the omtted category for regonal dummy s Ventane Captal. * Statstcally sgnfcant at the 10% level; ** at the 5% level; *** at the 1% level. B. Gender Dfferences The dfference between males and females s sgnfcant n Ventane Captal, whch s the same wth nternatonal standards (Psacharopoulos and Patrnos 2002), but t cannot be observed n others regons (Table 14). For the captal, the coeffcents of schoolng years and experence are statstcally sgnfcant at the 5% and 1% levels. For the years of schoolng, the rate of returns for females s about 1.5% hgher than that for males, comparng the fgure of 5.29% for females to 3.82% for males. In terms of returns to experence, the dfference between the genders s neglgble. The dscrepancy between the two sexes for other provnces s just the opposte,.e. the rate of returns to schoolng s statstcally sgnfcant and hgher for males (1.71%), whereas that for women s statstcally not sgnfcant and smaller n magntude (1.18%). Agan, as mentoned above, the level of development and busness actvtes n other provnces s by far behnd than the captal. It s lkely that there was lack of opportuntes for persons wth a hgher educaton background and experence to fnd a well pad job. As can be seen n table 10, a margnal ncrease n earnngs between persons wth a secondary and post-secondary level n other provnces s almost neglgble. 14

17 Table 14: Earnngs Functons by Gender. (Dependent varable s the natural log of monthly earnngs) Varable Ventane C. Males Females Constant *** *** (81.88) (62.28) Years of Schoolng *** *** (4.98) (5.01) Years of Experence *** ** (3.90) (2.36) Experence-squared *** (-2.93) (-1.58) R-squared Observatons N Note: t-statstcs n parentheses. Provnces Males Females *** *** (100.49) (87.91) *** (3.02) (1.52) (1.26) (0.93) (-0.87) (-0.96) All Males Females *** *** (128.09) (103.48) *** *** (6.01) (5.13) *** ** (2.65) (2.25) * (-1.77) (-1.56) * Statstcally sgnfcant at the 10% level; ** at the 5% level; *** at the 1% level. Table 15: Extended Earnngs Functons by Gender. (Dependent varable s the natural log of monthly earnngs) Varable Ventane C. Provnces All Males Females Males Females Males Females Constant *** (83.75) *** (64.74) *** (109.89) *** (100.71) *** (139.63) *** (111.21) Secondary (1.05) * (1.80) * (1.79) (1.04) ** (2.10) ** (1.98) Post-Secondary *** (2.90) *** (3.57) *** (2.61) (1.47) *** (3.91) *** (3.72) Years of Experence ** (2.10) (1.05) (1.26) (1.26) * (1.86) * (1.81) Experence-squared * (-1.88) (-0.92) (-1.04) (-1.29) (-1.56) * (-1.73) Marred *** (2.82) *** (3.46) (-0.53) (-1.00) ** (2.14) * (1.76) Northern *** (-7.10) (0.24) Central *** (-3.67) *** (-3.30) Southern *** (-5.18) *** (-2.34) R-squared Observatons N Note: t-statstcs n parentheses. The omtted category for level of educaton s prmary educaton, and the omtted category for regonal dummy s Ventane Captal. * Statstcally sgnfcant at the 10% level; ** at the 5% level; *** at the 1% level. 15

18 Table 15 llustrates the estmaton results of the extended Mncer model for males and females, whch yelded statstcally sgnfcant outcomes for Ventane Captal (compared wth results n Table 13). Of partcular nterest s the estmated coeffcent of women and men n the captal. A woman wth secondary and post-secondary educaton s lkely to earn about a 20% and 50% hgher salary than a woman wth a prmary educaton background, respectvely. The fgure for male workers at the post-secondary level s about 30%. In the provnces, the gap n earnngs between the secondary and post-secondary level for males s 10% and 18%, respectvely. In terms of martal status, marred men and women n Ventane Captal tend to earn 24% and 31% more as compared to unmarred counterparts, but t cannot be observed n other regons. C. Returns to Schoolng by Educaton Level In the prevous secton, the rate of returns on years of schoolng has been estmated for the entre samples and males and females (Table 12 and 14). In ths secton, we wll address the rate of returns to schoolng for each academc year from the ages of 11 to 18, wth the completed prmary educaton of 5 years beng the reference pont. Snce the data set does not dstngush between vocatonal, techncal and unversty, we opt for ths approach to compare the rate of returns for each academc year. The estmated coeffcents are presented n Table 16. As the table shows, the estmaton results are somewhat mxed for both Ventane Captal and other provnces. For Ventane Captal, the rate of returns to schoolng showed a fluctuatng trend over the perod under study. Intally t ncreased from 3.5% for 11 years of schoolng to 5.1% for 13 years (Vocatonal level), and started to declne 2.4% for 15 years (Techncal level), and ncreased agan to 6% for 18 years of schoolng (Unversty level). The frst ncrease for 13 years of schoolng mght be attrbutable to vocatonal educaton, whle the second ncrease mght be returns to unversty educaton. These results are smlar wth the case found n Czech Republc and Slovaka, that rates of returns to schoolng were hgh n vocatonal and unversty (also post graduate) educaton level, whereas a relatvely lower n the techncal educaton level (Chase 1998). On the contrary, for other provnces, ths rate showed a decreasng trend from 2.7% for 11 years to 2% for 18 years of schoolng. Ths downward trend s the same wth nternatonal standards (Psacharopoulos and Patrnos 2002), and t also lends support to the argument descrbed n the prevous secton that the prvate sector beng too small to absorb sklled labor and the wage rate of the publc sector beng flat. Hence, the hgher end of educaton level tends to yeld a lower return and s often consdered overqualfed. Perhaps, a secondary or vocatonal educaton level s well enough for servng the needs of the market. 16

19 Table 16: Earnngs Functons wth schoolng levels. (Dependent varable s the natural log of monthly earnngs) Varable Ventane C. Provnces All Constant *** (76.54) *** (111.27) *** (132.91) Years of schoolng S= * (1.83) [0.0346] *** (2.40) [0.0273] *** (2.73) [0.0276] Years of schoolng S= (0.77) [0.0188] Years of schoolng S= *** (2.73) [0.0506] Years of schoolng S= *** (3.26) [0.0467] Years of schoolng S= * (1.72) [0.0238] Years of schoolng S= (1.16) [0.0184] Years of schoolng S= ** (2.12) [0.0328] Years of schoolng S= *** (4.91) [0.0599] Experence *** (2.96) Experence-squared ** (-2.12) R-squared Observatons N 238 Note: t-statstcs n parentheses. Per year returns to schoolng n brackets ** (2.12) [0.0266] (0.47) [0.0063] (1.11) [0.0112] * (1.88) [0.0193] (1.4062) [0.0173] * (1.88) [0.0203] ** (2.10) [0.0211] (1.58) (-1.09) The omtted category for level of educaton s completed prmary educaton S=5. * Statstcally sgnfcant at the 10% level; ** at the 5% level; *** at the 1% level * (1.87) [0.0219] *** (2.43) [0.0272] *** (3.22) [0.0272] *** (2.66) [0.0224] * (1.88) [0.0186] *** (2.90) [0.0264] *** (5.58) [0.0437] *** (2.65) * (-1.72) VI. Conclusons The returns to schoolng n Lao PDR are low by nternatonal standards, but relatvely the same wth other transton economes. The estmates come from earnngs data for 1997/98. It s very useful from the lmted data currently avalable, but may be dstorted for a number of reasons. Frst, ntervewees are classfed by number of schoolng years rather than a clear educaton level, and they are not dstngushed by the publc or prvate sectors. Also, we have lmted the analyss to wage earners, whle over 80 percent of the Lao labor force s self-employed and many have multple jobs. 17

20 Ths study has mplcatons for the fnancng of educaton n Lao PDR, even though t s based on admttedly poor data. Lower educaton (prmary and secondary), the most proftable sub-sector judgng from the estmated rate of return results, especally outsde of Ventane Captal, s much less subsdzed than hgher levels. In fact, the hgh subsdy levels for hgher educaton contrbute to the low rates of return for these sub-sectors. Famly contrbutons to drect cost fnancng at the prmary level are a heavy burden, especally for the poor, and ths s nether socally optmal nor equtable. Lao s polcy makers may consder a more drect subsdzaton of poor prmary school age chldren to ensure that they enroll n school and reman enrolled. It s very mportant to montor future earnngs and labor market trends. It would be desrable to update these estmates as soon as recent earnngs data are avalable. The Hgher Educaton Graduate Tracer Study, for example, n Vetnam (World Bank 1997) suggests that the labor market s changng rapdly. Therefore, future updates of ths analyss, based on more recent data, could not only provde more robust estmates, but also provde evdence on whether the mpact of labor market reforms s ncreasng over tme, more rapdly and deeply. * Ths research was fnancally supported by Grant-n-Ad for Scentfc Research of Japan Socety for the Promoton for the Scence (Research Projects Integrated Research on Constructon of Insttutons for Recentralzaton and Methods for ts Evaluaton n Developng Countres (Chef: Prof. J. Matsunam) and Progress, Problems and Polces n Market Formaton and Japanese Cooperaton n Transton Countres (Chef: T. Suruga). References: ADB Labor Market Issues: Lao People s Democratc Republc. Asan Development Bank. ADB Country Economc Revew: Lao People s Democratc Republc. Asan Development Bank. ADB Key Indcators Asan Development Bank. Becker, G. S. and B. R. Chswck Educaton and the Dstrbuton of Earnngs. Amercan Economc Revew, Proceedngs 56: Bourdet, Y The Dynamcs of Regonal Dspartes n Laos: The Poor and the Rch. Asan Survey. 38(7): Card, D The Causal Effect of Educaton on Earnngs. n O. Ashenfelter and D. Card (eds), Handbook of Labor Economcs, Vol. 3, Amsterdam: North-Holland,

21 Card, D Estmatng the Returns to Schoolng: Progress on Some Persstent Economc Problems. Econometrca, 69(5): Chase, R. S Markets for Communst Human Captal: Returns to Educaton and Experence n the Czech Republc and Slovaka. Industral and Labor Relatons Revew, 51(3): Deolalkar, A. B Gender Dfferences n the Returns to Schoolng and n School Enrollment Rates n Indonesa. The journal of Human Resource, 28(4): Grlches, Z Estmatng the Returns to Schoolng: Some Econometrc Problems. Econometrca, Vol. 45, Kakwan, N. D. and E. M. Perna What s Pro-poor Growth? Asan Development Revew, 18 (1): Kakwan, N. D., G. B. Ssouphanhthong, P. Souksavath, & Wang, L Poverty n Lao PDR Durng the 1990s, mmeo, Manla: Asan Development Bank. Mncer, J Schoolng, Experence and Earnngs. New York: Natonal Bureau of Economc Research. Moock, P., H. Patrnos and M. Venkataraman Educaton and Earnngs n a Transton Economy (Vetnam). World Bank Polcy Research Paper Orazem, P. F. and M. Vodopvec Wnners and Losers n Transton: Returns to Educaton, Experence, and Gender n Slovena. World Bank Economc Revew 9(2): Pssardes, C. A Human Captal and Growth: A Synthess Report. OECD Development Centre, Techncal Papers No Psacharopoulos, G. and H. A. Patrnos Returns to Investment n Educaton: A Further Update. World Bank Polcy Research Paper Research Trangle Insttute Health and Educaton of Ethnc Mnortes n the Greater Mekong Sub-Regon: Lao PDR Country Report. North Carolna. Schulz, P. T Educaton Investments and Returns. n H. Chenery and T. N. Srnvasan (eds), Handbook of Development Economcs, Vol. I, Amsterdam: North- Holland, Syhalad, B., S. Keolangsy, S. Amnthalad and N. Dhanasakd Lao Labour Market Indcators Internatonal Labour Organzaton. Varga, J Returns to Educaton n Hungary. Acta Oeconomca 47 (1-2): Vjverberg, W. P. M Educatonal Investments and Returns for Women and Men n Cote d Ivore. The journal of Human Resource, 28(4): We, X., M. C. Tsang, W. Wu and L. K. Chen Educaton and Earnngs n Rural Chna. Educaton Economcs 7(2). 19

22 Wlls, R. J Wage Determnants: A Survey and Renterpretaton of Human Captal Earnngs Functon. n O. Ashenfelter and R. Layard (eds), Handbook of Labor Economcs, Vol. I, Amsterdam: North-Holland, World Bank Lao PDR: Country Economc Memorandum. World Bank: Washngton, D.C., World Bank Vetnam: Educaton Fnancng. (A World Bank Country Study). Washngton, D.C. World Bank. World Bank, IMF, ADB Publc Expendture Revew: Country Fnancal Accountablty Assessment. Vol.2, Washngton, D.C. Xe, Y. and E. Hannum Regonal Varaton n Earnngs Inequalty n Reform-Ear Urban Chna. Amercan Journal of Socology. 101(4):