MEXICO: EVOLUTIO ON F EARNINGS INEQUALITY AN...

M E X I C O : E V O L U T I O N O F E A R N I N G S I N E Q U A L I T Y A N D R A T E S O F

R E T U R N S T O E D U C A T I O N (1988-2002 )

G l a d y s López-Acevedo* Banco Mundial

Resumen: E s t e d o c u m e n t o a n a l i z a los f ac to res y m e c a n i s m o s q u e h a n a f e c t a d o l a

evolución de l a d e s i g u a l d a d e n Méx i co , se d e m u e s t r a q u e l a d e s i g u a l d a d

e d u c a t i v a e x p l i c a l a m a y o r p a r t e de l a variación e n l a d e s i g u a l d a d d e

ingresos . L a contr ibución de l a d e s i g u a l d a d e d u c a t i v a a l a d e s i g u a l d a d

d e ingresos e n Méx i c o es l a s e g u n d a m a s a l t a e n Lat inoamér ica . E l

i n c r e m e n t o e n l a d e s i g u a l d a d de ingresos n o parece ser e l r e s u l t a d o

d e u n d e t e r i o r o e n l a distr ibución de l a educación, a pesa r d e q u e e l

p e r f i l de ing r eso , e l c u a l e s t a r e l a c i o n a d o c o n l a t a s a d e r e t o r n o s a l a

educación se h a v u e l t o más p r o n u n c i a d o .

Abstract: T h i s p a p e r r e v i e w s t h e f a c t o r s a n d m e c h a n i s m s t h a t have b e e n d r i v i n g

i n e q u a l i t y i n M e x i c o a n d finds t h a t e d u c a t i o n a l i n e q u a l i t y a c c o u n t s f o r

b y f a r t h e l a r g es t s h a r e o f M e x i c o ? s v a r i a t i o n i n e a r n i n g s i n e q u a l i t y .

T h e c o n t r i b u t i o n o f i n e q u a l i t y o f e d u c a t i o n t o i n e q u a l i t y o f e a r n i n g s

i n M e x i c o is t h e s e cond h i g h e s t i n L a t i n A m e r i c a a f t e r B r a z i l . T h e

inc rease i n e a r n i n g s i n e q u a l i t y , howeve r , does n o t a p p e a r t o b e t h e

r e s u l t o f a w o r s e n i n g i n t h e d i s t r i b u t i o n o f e d u c a t i o n , a l t h o u g h t h e

i n c o m e p r o f i l e , w h i c h is r e l a t e d t o t h e r e t u r n s t o s c h o o l i n g , has b e c o m e

m u c h s t eepe r .

Clasificación J E L : 121, 131

Fecha de recepción: 24 VII 2000 Fecha de aceptación: 12 IV 2004

* E c o n o m i s t , L C S P P . T h i s r esearch was c o m p l e t e d as p a r t o f t h e E a r n i n g s

I n e q u a l i t y S t u d y a t t h e W o r l d B a n k R e p o r t N o . 19945. M E . T h e u s u a l d i s c l a i m e r

app l i e s , g a c e v e d o @ w o r l d b a n k . o r g

211

mailto:[email protected]

212 E S T U D I O S E C O N Ó M I C O S

1. I n t r o d u c t i o n

Achiev ing sustainable economic g rowth w i t h a more egal i tar ian d i s t r i b u t i o n of income is at the core of Mexico's development challenge. Yet the country does not per form well i n terms of equity when compared w i t h other L a t i n Amer ican countries. According to a recent s tudy developed by the In ter -Amer ican Development Bank (1998-1999), Mexico has the s ix th most unequal d i s t r i bu t i on of overall household i n come i n L a t i n Amer ica (and the t h i r d worst in urban areas). I n the broader internat iona l context, Mexico's rat io of income share accruing to the top 10 percent of the populat ion to the share accruing to the b o t t o m 40 percent is higher t h a n what is observed b o t h for the high-income countries and for the vast ma jo r i t y of low-income countries (see table 3.A3).

The per iod f rom the second hal f of the 1980s through the 1990s was an especially meaningful per iod for the Mexican economy, wh i ch sought to move f rom a protected economy driven by the publ ic sector t o a global ly integrated economy dr iven by the pr ivate sector. Th is s t ruc tura l change resulted in sizable economic g rowth , but Mexico's income d i s t r i bu t i on became increasingly unequal and failed t o respond either t o economic g rowth or to publ ic policy.

Most remarkable, the level, deter iorat ion, and resistance to po l icy of Mexico's inequal i ty over the past decade coexisted w i t h very r ap id progress in educat ional a t ta inment , b o t h in terms of coverage and d i s t r i bu t i on of schooling (de la Torre, 1997). This phenomenon, wh i ch has been observed in other developing countries as wel l as developed ones, is somewhat surpris ing, given the powerful equalizing properties generally a t t r i bu t ed to education.

Th is paper reviews the factors and mechanisms d r i v i n g inequali t y i n Mexico. More specifically, i t examines the expansion i n earnings inequal i ty w i t h emphasis on the role of educat ion, 1 establishes an ana ly t ica l framework tha t permits analysis of the interact ion between educat ion and the labor market , and examines the evo lut ion of earnings inequal i ty i n l ight of the macroeconomic and educational policies followed in the 1980s and 1990s.

The paper is organized as follows. Section 2 describes the evolut i on of t o t a l current income inequality, using in format ion contained in the Nat iona l Household Income and Expenditures Survey, ENIGH,

Wages a r e r e l a t e d d i r e c t l y t o i n d i v i d u a l c h a r a c t e r i s t i c s a n d d o n o t d e p e n d

o n f a m i l y s t r u c t u r e . Bes ides , t h e d i s t r i b u t i o n o f wages e x p l a i n s m u c h o f t h e

d i s t r i b u t i o n o f we l f a r e i n soc ie ty .

M E X I C O : E V O L U T I O N O F E A R N I N G S I N E Q U A L I T Y 213

and using household income per capi ta as the u n i t of analysis. Sect i o n 3 focuses on the evo lut ion of ind i v idua l earnings inequal i ty , using in fo rmat ion i n the Na t i ona l U rban Employment Survey, E N E U . Sect i o n 4 investigates how much of Mexico's earnings inequal i ty can be explained by educational inequality, as wel l as by other contro l var i ables, b o t h i n gross and marg ina l t e rms . 2 Section 5 analyzes the evo lut ion of educat ional a t ta inment . Section 6 relates changes in the d i s t r i bu t i on of educat ion t o changes i n earnings inequality. Section 7 examines the evo lut ion and s t ructure o f the rates o f re turns to educat ion by means of o rd inary least squares and quant i le regressions. The last section offers concluding remarks.

2. T h e E v o l u t i o n of T o t a l I n c o m e Inequa l i t y

The evaluat ion of income inequal i ty i n Mexico is based on in fo rmat ion available i n the E N I G H (see annex 1 for a br ie f descr ipt ion) . Th is survey captures t o t a l current income of households, inc lud ing nonmonetary income, earnings, and other sources of monetary income. The u n i t of analysis is the household, and the concept of income is household income per cap i ta . 3

The ma in results of th is evaluation are shown i n table 1, wh ich indicates a sizable deter iorat ion i n income d i s t r i bu t i on d u r i n g the per iod under review. Wh i l e the poorest 20 percent of the popu la t i on lost almost one-seventh of the i r income share (0.6 percentage po in t ) , the richest 10 percent increased theirs by something close t o one-seventh (5.2 percentage po ints ) . Moreover, the richest group was the only one t o gain over t h a t per iod, as not only the poorest b u t also those i n the middle lost i n relative terms.

Mexico i n the per iod f rom 1984 to 1996 was marked by a series of regressive income transfers f rom almost the entire spec t rum of the popu la t i on to the richest s t r a t u m . Accordingly, the most commonly used inequal i ty index points t o a worsening i n income inequal i ty over th is span of t ime . The G i n i coefficient, wh ich is especially sensitive

E d u c a t i o n a l a t t a i n m e n t has a n i m p a c t n o t o n l y o n i n c o m e b u t a lso o n o t h e r

o u t c o m e s t h a t a re i m p o r t a n t f o r a n i n d i v i d u a l ' s w e l l - b e i n g b u t t h a t a r e n o t nec

essa r i l y m e a s u r e d i n m o n e t a r y t e r m s . T h i s s t u d y , howeve r , does n o t c o n s i d e r t h e

n o n - m o n e t a r y i m p a c t s o f e d u c a t i o n .

3 T o t a l c u r r e n t i n c o m e o f t h e h o u s e h o l d d i v i d e d b y t h e n u m b e r o f h o u s e h o l d

m e m b e r s . T h a t i s , w e a re c o n s i d e r i n g t h e h o u s e h o l d as a u n i t c h a r a c t e r i z e d b y a

flow o f i n c o m e t r a n s f e r s a n d d i s r e g a r d i n g aspec ts r e l a t e d t o equ i va l ence scale .


t o changes in the middle of the d i s t r i bu t i on , rose f rom 0.473 in 1984 to 0.515 in 1996. The The i l T index, wh ich is extremely sensitive to changes in the upper and lower tai ls , rose f rom 0.411 in 1984 to 0.524 i n 1996.

T a b l e 1 Lorenz Curves for Total Current Income, 1984-1996

(accumulated income share; percent)

Population share 1984 1989 1992 1994 1996

10 1.66 1.39 1.32 1.39 1.39

20 4.47 3.88 3.68 3.76 3.89

30 8.19 7.29 6.92 6.98 7.29

40 12.85 11.65 11.09 11.08 11.63

50 18.76 17.05 16.26 16.28 17.08

60 26.15 23.78 22,83 22.79 23.86

70 35.51 32.25 31.13 31.10 32.39

80 47.64 43.12 42.14 41.93 43.44

90 64.53 58.75 58.32 57.68 59.33

92 68.79 63.06 62.81 62.03 63.61

94 73.73 68.03 68.03 67.26 68.68

96 79.38 73.82 74.47 73.70 74.95

98 86.68 81.60 82.81 82.49 83.32

100 100.0 100.0 100.0 100.0 100.0

B o t t o m 20% 4.5 3.9 3.7 3.8 3.9

Midd le 40% 21.7 19.9 19.2 19.0 20.0

Midd le -h igh 30% 38.4 35.0 35.5 34.9 35.5

Top 10 percent 35.5 41.3 41.7 42.3 40.7

G i n i coefficient 0.473 0.519 0.529 0.530 0.515

The i l T index 0.411 0.566 0.550 0.558 0.524

N o t e : T o t a l c u r r e n t i n c o m e is based o n h o u s e h o l d i n c o m e p e r c a p i t a .

Sou r c e : A u t h o r ' s c a l c u l a t i o n s based o n E N I G H .

The worsening of income d i s t r ibu t i on is indisputable, but two points must be stressed. The first one is tha t , according to the E N I G H

MEXICO: EVOLUTION OF EARNINGS INEQUALITY 215

survey, most of the deter iorat ion occurred i n the middle to late 1980s (1984-1989). There was l i t t l e var ia t ion i n earnings inequal i ty i n the early 1990s, except for a sl ight t r end toward deter iorat ion. F r om 1989 t o 1994, the income share accruing to the 20 percent poorest decreased s l ight ly ( f rom 3.9 t o 3.8 percent) , whereas the share accruing t o the richest 10 percent increased (by 1 percentage po in t ) ; those i n the middle also experienced losses.

The second fact is surpr is ing and hard t o explain: income d i s t r i b u t i o n improved between 1994 and 1996, an interva l of t ime i n wh i ch the Mexican economy experienced a severe f inancial cr is is . 4 Usual ly one wou ld expect inequal i ty t o rise dur ing t imes of recession, because the r ich have more ways of protect ing the i r assets t h a n the poor. Th is is especially t rue of labor, wh ich is basically the only asset of the poor (the labor-hoarding hypothesis). Nevertheless, d u r i n g th is t ime the 10 percent richest experienced relative losses ( their income share dropped 1.6 percentage points ) , and inequal i ty declined. The G i n i coefficient dropped f rom 0.534 0.530 i n 1994 to 0.515 i n 1996, whi le the The i l T index dropped f rom 0.558 to 0.524. I t could be argued t h a t the richest experienced severe capi ta l losses t h a t affected the i r t o t a l income more t h a n the poor, b u t th is hypothesis is not supported by the da ta presented i n table 2: monetary income other t h a n wages and salaries as wel l as f inancial income increased as a share of t o t a l income i n tha t t ime per iod, par t i cu lar ly i n u r b a n areas. Therefore, the fal l i n inequal i ty remains somewhat puzzl ing.

Table 3 displays the G i n i coefficient and The i l T index for urban and r u r a l areas using t o t a l current income. For b o t h indexes inequal i ty was lower i n ru ra l areas t h a n in urban areas and was remarkab ly stable u n t i l 1992. A f ter a smal l decrease i n 1994, ru ra l inequal i ty increased i n 1996, contrary to the aggregate result . I n l ight of these outcomes, the behavior of current income d i s t r i bu t i on i n Mexico seems to be dr iven by the trends i n urban areas.

3 . T h e E v o l u t i o n o f E a r n i n g s I n e q u a l i t y

How much of t o t a l income inequal i ty is due to earnings inequal i ty? Table 4 presents the results of t o t a l current income inequal i ty for each

4 I n 1994 , t h e c u r r e n t a c c o u n t d e f i c i t was $30 b i l l i o n , a b o u t 7 p e r c e n t o f gross d o m e s t i c p r o d u c t , G D P . T h e m a i n ef fects o f t h e f i n a n c i a l c r i s i s we r e ( a ) G D P a n d d o m e s t i c d e m a n d f e l l 6.2 a n d 14 p e r c e n t , r e spe c t i v e l y ; ( b ) t h e u n e m p l o y m e n t r a t e rose f r o m 3.7 p e r c e n t i n 1994 t o 6.2 p e r c e n t i n 1995; a n d (c ) G D P p e r c a p i t a decreased 7.8 p e r c e n t a n d w o r k e r s e x p e r i e n c e d a s i g n i f i c a n t r e d u c t i o n i n t h e i r r e a l wages , n e a r l y 17 p e r c e n t i n 1995.

216 ESTUDIOS ECONÓMICOS

O Í . 1 ' é i

CO CN CO CN

00 O

O rH CN rH

00 co

rH lO N l O (O n¡ N « 9

rH rH O

I O OÍ LO CO

CN L O 00 L O

05 CN tD O CO 00

00 CN CN

CN L O rH

O L O CN CN

g

|3 t> L O o

- 00 CN CN C0 CN

CO _ CN CN CO rH ^ CO

O O rH O

rH CO O CN CO CN

rH O Tjí O CD

CO Oí O CN CO ^ ^ rH rH O

CO

O) 00 N- CO rH es L O ^ o

CO rH t>«» i—I L O

CO

00 O rH

00 CO L O t"-

o CO 00 ró CN

O CN O CO

s o o sí

C3

S +3 u ce tí 03 03 >

b0 tí

O c3

tí ,o

tí «3 03 a o tí

O

03 a o

0> 0>

S o

tí O

CU" S tí CO

tí O o

en

tí cu a

cu tí 0 S 1 tí o

tí 03 *-<

cu

tí

O

bO .5 "53 tí O

i i

CJ C3 C

s;


of i ts components: earnings, 5 monetary income excluding earnings, and non-monetary income by urban and r u r a l areas. 6 Earnings cont r i b u t e t o most of the overall inequality, being responsible for a lmost hal f of inequal i ty at the nat iona l level. These figures clearly may be affected by the underrepor t ing of capi ta l gains, bu t unders tand ing the mechanisms t h a t produce earnings inequal i ty represents a large step toward understanding the behavior of t o t a l inequality. As l ong as labor is the ma in , i f not the only, asset of the poor, a better know l edge of earnings inequal i ty is a valuable i npu t for the assessment o f poverty and welfare issues.

T a b l e 3 Inequality Measures for Total Current Income, 1994-1996

Year Gini coefficient Theil T index

National Urban Rural National Urban Rural

1984 0.473 0.442 0.448 0.411 0.356 0.375

1989 0.519 0.498 0.444 0.566 0.526 0.361

1992 0.529 0.498 0.434 0.550 0.483 0.353

1994 0.534 0.508 0.419 0.558 0.499 0.325

1996 0.519 0.493 0.452 0.524 0.470 0.390 S o u r c e : A u t h o r ' s c a l c u l a t i o n s based o n E N I G H .

We use the E N E U household survey t o examine the behavior of earnings inequal i ty because i t is extremely r i ch i n household characteristics (see annex 2 ) . 7 Table 5 shows tha t the d i s t r i bu t i on o f

E a r n i n g s as d e f i n e d i n t h e E N I G H s u r v e y i n c l u d e sa lar ies a n d wages , p a i d o v e r - t i m e , t i p s , c o n t r a c t w o r k e r s ' e a r n i n g s , C h r i s t m a s o r N e w Yea r bonuses a n d o t h e r g i f t s , a n d o t h e r m o n e t a r y c o m p e n s a t i o n s ( n o n - r e g u l a r e a r n i n g s ) . E a r n i n g s as d e f i n e d i n t h e E N E U s u r v e y i n c l u d e sa lar ies a n d wages, s e l f - emp l oy ed w o r k e r s ' e a r n i n g s , c o n t r a c t w o r k e r s ' e a r n i n g s , a n d i m p l i c i t sa lar ies o f f i r m o w n e r s , as w e l l as n o n - m o n e t a r y e a r n i n g s .

6 A l t h o u g h t h e r e s u l t s a re s h o w n fo r t h e G i n i coe f f i c i en t , these a l so c o u l d have

b e e n o b t a i n e d f o r t h e T h e i l T i n d e x , as b o t h o f t h e m sa t i s f y t h e s i x p r o p o s i t i o n s

l i s t e d i n S h o r r o c k s ( 1980 a n d 1984) as w e l l as S h o r r o c k s a n d M o o k h e r j e e ( 1 9 8 2 ) .

7 I n o r d e r t o r e d u c e t h e h e t e r o g e n e i t y o f t h e s a m p l e a n d a lso a spec t s r e l a t e d

t o se l f - se l ec t i on , t h e p o p u l a t i o n u n d e r ana l y s i s i n c l u d e s i n d i v i d u a l s l i v i n g i n u r b a n

areas , b e t w e e n 16 a n d 65 years o l d , a n d w o r k i n g 20 h o u r s a week o r m o r e . I t does

n o t i n c l u d e seasona l w o r k e r s . A l s o t h e t w o h i g h e s t o b s e r v a t i o n s we r e d r o p p e d

f r o m t h e s a m p l e g i v e n t h e c l ea r e v i dence o f o u t l i e r s i n s ome years .


earnings has become more unequal i n recent times. The G i n i coefficient j u m p e d f rom 0.395 i n 1988 to 0.442 in 1997, after reaching a peak of 0.464 in 1996. Similar ly, the The i l T index increased f r o m 0.327 i n 1988 t o 0.372 i n 1997, w i t h 0.474 in 1996. Another index, the R 10/2o? which is the ra t io of the income share accruing to the richest 10 percent to t h a t accruing to the poorest 20 percent, increased f r o m 4.48 to 6.04 over the per iod, reaching a m a x i m u m of 6.74 i n 1996.

T a b l e 4 Decomposition of Total Current Income, 1984-1996

(percentage share in overall Gini)

Region Earnings Monetary income Non-monetary Total

and year excluding earnings current income

National

1984 46.0 32.9 21.0 100.0

1989 41.0 36.0 23.0 100.0

1992 42.9 31.9 25.2 100.0

1994 50.2 25.9 23.9 100.0

1996 46.7 29.4 23.9 100.0

Urban

1984 45.6 32.2 22.2 100.0

1989 38.6 37.3 24.1 100.0

1992 41.4 33.1 25.5 100.0

1994 50.0 26.0 24.0 100.0

1996 46.1 29.8 24.1 100.0

Rural

1984 30.7 49.5 19.8 100.0

1989 35.7 43.5 20.8 100.0

1992 29.6 42.2 28.2 100.0

1994 31.9 43.8 24.2 100.0

1996 35.7 41.2 23.1 100.0 Source : A u t h o r ' s c a l c u l a t i o n s based o n E N I G H .

There are two m a i n differences in the pa t t e rn shown by the dist r i b u t i o n of earnings and t o ta l current income. F i rs t , the gains were


not l im i t ed to the richest 10 percent. Those i n the seven-, eight-, a n d nine-tenths of the d i s t r i bu t i on also improved the ir relative earnings over the per iod by almost 2 percentage points; the biggest losers were the midd le 40 percent, who lost more t h a n 2 percentage po ints o f the i r income share. Second, the earnings d i s t r i bu t i on clearly worsened i n the 1990s up th rough 1996, a l though the inequal i ty associated w i t h t o t a l current income was moderate ly stable i n the 1990s a n d even improved i n 1996.

The behavior of t o t a l current income inequal i ty and earnings inequal i ty f rom 1994 to 1996 supports the idea tha t the poor , who rely most ly on labor as a source of income, are the least able t o protect themselves du r ing a recession. However, the substant ia l d rop i n earnings inequal i ty f rom 1996 to 1997 is, once more, a surpr i s ing finding. For example, the # 1 0 / 2 0 index declined f rom 6.74 i n 1996 to 6.04 i n 1997. I t is t rue tha t the Mexican economy as a whole had a s t rong and impressive performance i n 1997. The aggregate g rowth rate was around 7 percent, real investment grew 24 percent, exports grew 17 percent, indus t r i a l p roduct ion increased 9.7 percent, and the c i v i l construct ion sector, wh ich is h igh ly intensive i n less-skil led labor, grew close t o 11 percent. Under such a scenario, an improvement i n the d i s t r i bu t i on of earnings is not unlikely, b u t the magni tude and quickness of the recovery cal l for a detai led inspect ion o f the mechanisms responsible for i t .

Three broad hypotheses frequently are advanced to expla in the earnings inequal i ty experienced i n Mexico and other countries. 8 These l i nk the increase i n earnings inequal i ty to (a) increased openness of the economy, (b) i n s t i tu t i ona l changes i n the labor market , and (c) skill-biased technological change.

The first of these hypotheses argues t h a t as trade barriers are reduced, an economy is placed under heightened compet i t ive pressure t o specialize along i ts lines of comparative advantage. A developed country w i t h a relat ively abundant supply of high-ski l led workers, l ike the Un i t ed States, w i l l be induced to specialize i n act iv i t ies t ha t require a h igh level of sk i l l or education as i ts low-skil led industries come under increased compet i t ive pressure f rom countries w i t h an abundant supply o f low-skil led, low-wage workers.

Hanson and Harr ison (1995) examine the impact of Mex ican trade reform on the s t ructure of wages using in format ion at the firm level. They test whether trade reform shifted employment t oward i n -

See, f o r e x a m p l e , G o t t s c h a l k , P. et al. ( 1 9 9 7 ) , The World Bank Economic

Review ( 1 9 9 7 ) , s p e c i a l issue, a n d de F e r r a n t i et al., 2003 .


Os Os

Os Os

©i Os

I ^

Os

O} ©i Os

Os Os

Oj O} Os

OÍ

CO CN 00 Oí CN oó CN CO co

00 oo Os

1.1

o

00 oo oo Oí CN

có CO co CO o o

oo LO Oí CN

CÓ 00 oo CO o o

«3 cr

i 4

i<8


dustries t h a t are re lat ively intensive i n the use of ski l led labor ( the Stolper-Samuelson-Type, SST effect). They conclude t h a t the wage gap was associated w i t h changes w i t h i n industries and f i rms, w h i c h cannot be explained by the SST effect. Thus the increase i n wage inequal i ty was due t o other factors . 9 Hanson (1997) examines a t rade theory based on increasing returns, wh ich has impor tan t impl i ca t ions for regional economies, and concludes t h a t employment and wage pat terns are consistent w i t h the idea t h a t access t o markets is i m p o r t a n t for the locat ion of industry .

Th i s first hypothesis has several problems when appl ied to the Un i t ed States and becomes even less persuasive when applied t o Mexico. Mexico great ly l iberal ized i ts trade regime after 1984. However, the reduct ion o f i ts t rade barriers was most ly w i t h respect t o i m ports f r om the developed countries, notab ly the Un i t ed States and Canada, whose share of t o t a l Mexican merchandise impor ts increased f rom 68 percent i n 1985, to 73 percent i n 1993, and to almost 78 percent i n 1996. Since Mexico has an abundant supply of low-ski l led labor compared w i t h its nor thern neighbors, the l ibera l izat ion o f t rade could be expected to induce a pa t t e rn of specialization t h a t wou ld raise the relative demand (and hence wages) of the lesser-educated members o f the labor force. Th is d id not happen. Instead, the i n crease i n earnings inequal i ty observed i n Mexico followed the same pa t t e rn as t ha t observed i n the Un i t ed States: less-educated workers experienced real wage declines, whi le h ighly educated workers experienced real wage improvements. The trade-based explanat ion may s t i l l be relevant, however, t o the extent t ha t greater openness facilitates the transfer of ideas and technology. Th is is a more persuasive explanat ion of the increase i n earnings inequality. A var iant of the globalization-technology nexus advanced by Feenstra and Hanson (1996) involves outsourcing i n which mu l t ina t i ona l enterprises i n the developed country relocate their less skil l- intensive act iv i t ies t o the less ski l l -abundant developed countries. However, what is referred t o as a low-ski l l ac t i v i t y i n the Un i t ed States may be a h igh-sk i l l act ivi t y i n Mexico. Th i s could explain the s imi la r i t y i n the evo lut ion of earnings inequal i ty i n b o t h countries (de Ferrant i et al, 2003).

The second explanat ion revolves around ins t i tu t i ona l changes such as reductions i n the m i n i m u m wage, the weakening o f trade unions, and the decline of state-owned enterprises. The existence of a b ind ing m i n i m u m wage, for example, truncates the lower end of

T h e S t o l p e r - S a m u e l s o n ef fect a lso is e x a m i n e d u n d e r N A F T A i n B u r f i s h e r

a n d o t h e r s ( 1 9 9 3 ) .


the wage d i s t r i bu t i on . As the m i n i m u m wage is allowed t o erode -say, th rough i n f l a t i o n - i t becomes less b ind ing by mov ing far ther down the low end of the wage d i s t r i bu t i on , w i t h the result t h a t , ceteris paribus, a higher share of wages w i l l lie below the previous min imum-wage level. Th is translates into an increased dispersion i n wages and earnings. I n s t i tu t i ona l developments have not exerted a significant influence on the earnings d i s t r i bu t i on since the early 1980s (see Hernandez, Garro, and Llamas, 1997). The d i s t r i bu t i on of rea l wages, for example, does not reveal any significant d istort ions a round the m i n i m u m wage, which suggests tha t i t is not a b ind ing constra int . The fact tha t this m i n i m u m wage has continued to erode i n real value, therefore, seems to be irrelevant. Similar ly, the d i s t r i bu t i on of un i on wages is not s igni f icantly different f rom the d i s t r i bu t i on of nonun ion wages, once differences i n educational levels are taken into account. Th is also renders any erosion of union power irrelevant for the dist r i b u t i o n of earnings. I n conclusion, a l though the influence of i n s t i t u t i o n a l factors cannot be rejected entirely, i t does not appear to be the pr inc ipa l cause of the increase i n earnings inequality.

A persuasive explanat ion, b o t h for the Un i t ed States and for Mexico, seems to be one t h a t l inks earnings inequal i ty t o skil l-biased technological changes tha t raise the relative demand for higher-ski l led labor. Cragg and Epe lbaum (1996) examine the shift i n demand i n Mexico. They po int out tha t the major source of r is ing inequal i ty is a biased shift i n demand rather than a un i fo rm g rowth in demand when there are different labor supply elasticities. Meza (1999) also investigates shifts in demand and offers the hypothesis tha t the shi f t in demand toward a more educated labor force " w i t h i n " an economic sector explains the increase i n the ir p r emium when compared w i t h the shift in demand for less-educated workers "between" economic sectors. Tan and Ba t r a (2000) s tudy the skill-biased technical change hypothesis as a plausible explanat ion of wage inequal i ty using da t a at the f i rm level for Colombia, Mexico, and Taiwan (China) . They obta in the fol lowing results: (a) a f irm's investments i n technology have the largest impact on the d i s t r i bu t i on of wages for ski l led workers, (b) they have the smallest impact on wages paid to unski l led workers, and (c) wage premiums paid to skil led workers are led p r i mar i l y by the firm's investments i n research and development ( R & D ) and t ra in ing . Such conclusions seem to support the skill-biased technological change hypothes is . 1 0 Accord ing to the typology used by

T h e s e r e s u l t s s h o u l d b e c o n s i d e r e d ca r e fu l l y , s ince t h e a n a l y s i s is based o n

d a t a a t t h e f i r m l eve l a n d o n l y for t h e m a n u f a c t u r i n g i n d u s t r y .


Johnson (1997), the type of technological change t h a t drives wages up for the more h igh ly ski l led workers and drives wages down for the less-skilled workers (as occurred i n b o t h the Un i t ed States and Mexico) is extensive skill-biased technological change. Under th is type o f technological change, ski l led workers are more efficient i n jobs t h a t were t rad i t i ona l l y performed by unski l led workers (de Ferrant i el a/., 2003).

As shown i n figure 1, a l l series have the same t r end for a l l per i ods . 1 1 However, beginning i n 1990 condi t ional real earnings for workers w i t h a univers i ty educat ion increased substantial ly, whi le condit i ona l real earnings for workers w i t h low levels of education remained steady u p to 1994. A f ter t h a t , earnings differentials among workers of a l l educational levels remained constant. Th is suggests t h a t factors other t h a n the supply of new workers w i t h a basic educat ion drove earnings differentials by level of schooling. Th is issue is fur ther examined in section 7.

I n sum, demand and supply, interact ing w i t h i n a context of economic modern izat ion and global izat ion, generate the t r end toward greater wage disparity. However, none of these explanations deals exp l i c i t l y w i t h changes i n the d i s t r i bu t i on of education or w i t h the in teract ion between the educational policies t ha t induced t h e m and the workings of the labor market .

4. S t a t i c D e c o m p o s i t i o n

This section aims t o evaluate the cont r ibut i on to earnings inequal i ty i n Mexico of a set of variables, related either to ind iv idua l a t t r ibutes , such as schooling and age, or a form of par t i c ipa t ion i n the labor market , such as number of hours worked or status, for selected years f rom 1988 to 1997. The idea is t o measure the inequal i ty tha t is left unexplained after t ak ing into account the differences i n average earnings among workers i n different groups. W h e n the exercise is conducted for a single variable, th is reduct ion is said t o be the gross con t r i bu t i on of the variable to overall wage inequality. W h e n a var i able is added to a model tha t contains a l l the remaining variables, the change i n the gross cont r ibut i on of these two models is called the marg ina l con t r ibu t i on of the added variable. I n other words, the gross

M e d i a n r e a l h o u r l y e a r n i n g s a re e s t i m a t e d u s i n g q u a n t i l e r eg ress i on m o d e l s

(9 = 0.5) a n d c o n d i t i o n e d o n expe r i ence , gender , l a b o r m a r k e t s t a t u s , e c o n o m i c

sec to r , a n d r e g i o n (see a n n e x 1 f o r d e f i n i t i o n s ) .


con t r i bu t i on is the uncontro l led explanatory power of a given variable, and the marg ina l con t r ibu t i on is i ts explanatory power contro l led by a set o f other seemingly relevant variables. The methodology and corresponding appl icat ion are discussed below.

F i g u r e 1 Conditional Median Real Hourly Earnings

by Educational Level, 1988-1997 (1988 = 100)

20.00

18.00

U.W I | 1 1 1 1 1 1 1 1 1

1998 [ 999 1990 1991 1992 1993 1994 1995 1996 1997

— Primaryincorr^lete - - P r i m a r y c o m p l e t e - L s e c o n d a y c o m p .

- U. secondary comp. University

N o t e : M e d i a n s w e r e c a l c u l a t e d c o n d i t i o n a l o n expe r i ence , e x p e r i

ence s q u a r e d , g ender , e c o n o m i c sec tor , l a b o r m a r k e t s t a t u s , a n d r e g i o n .

Source : A u t h o r ' s e s t i m a t i o n s based o n E N E U survey .

4.1. Short Review

Before proceeding to the decomposit ion exercise, i t is w o r t h rev iewing the conclusions of other recent studies on the evolut ion of earnings inequal i ty and some variables tha t are impor tan t i n the process o f earnings format ion

Cragg and Epe lbaum (1996) show tha t bo th average wage and educat ion sk i l l p r em ium, defined as the percentage increase i n wages over those of the group w i t h p r imary schooling, have increased substant ia l ly for workers w i t h more education. I n other words, the higher


the level o f education, the larger the increase i n average wages, wh i ch i n t u r n leads to an increase i n inequality. They also examine whether the h igh demand for ski l led labor is indus t ry specific, task specific, or s imply the result o f general education. I n order to assess the marg ina l con t r i bu t i on of factors unrelated to education, these factors are cont ro l l ed by a set o f d u m m y variables t ha t describe the i n d u s t r y - a n d task-specific effects. The authors conclude t h a t the industry-speci f ic effect is smal l and t h a t the t a s k - specific effect (occupation variable) explains ha l f of the g rowth i n wage dispersion f rom 1987 t o 1993. Th is conclusion may not be correct, however, as occupat ion m i g h t be considered an endogenous variable, wh ich is determined by educat ion. As shown on table 3.A2, educational level and occupation variables are h igh ly correlated. I n contrast, the correlat ion between educat ion and other variables is low. Hence the occupation variable should be handled carefully i n any k i n d of analysis.

De Ferrant i et al, (2003, 2004), and Lopez-Acevedo and W a l t o n (2004) show tha t Mexico has a steep wage hierarchy, pa r t i cu la r l y w i t h respect t o schooling and spat ia l differences. Th is is pa r t l y because o f the long-run inheritance of unequal access and qua l i ty o f schooling. Furthermore , Mexico experienced r is ing differentials, especially for workers w i t h college education, i n the late 1980s and much of the 1990s, followed by decl ining differentials since about 1997. T h i s can be interpreted as a product of the interactions between changes in the relative supply and relative demand for different sk i l l categories. The educat ional expansion of the past two decades has led t o a gradual increase i n the relative supply of workers w i t h secondary educat ion and t e r t i a r y education-whose share i n the workforce rose by some 50 and 40 percent respectively between the late 1980s and late 1990s (de Ferrant i et al 2003, pp. 51 and 54).

Lopez-Acevedo and Wa l t on (2004) find t h a t returns t o t e r t i a ry educat ion of workers i n the labor market rose signif icantly i n the 90s. Th is is a t t r i b u t e d t o the skill-biased technical changes induced by the large-scale opening of the Mexican economy to trade and foreign d i rect investment, wh ich was consolidated under N A F T A . Chi le - w h i c h also went th rough a pro found opening to in ternat iona l marke t s - also experienced a large rise and modest fal l i n the wage p r e m i u m to college graduates, suggesting t h a t the effect of opening could be t rans i t i ona l , especially i f accompanied by vigorous expansion of educat ion (Montenegro, 1999).

4.2. Methodology

The approach i n th is paper uses inequal i ty measures known as "gen-


eralized entropy indexes." Bourguinon (1979), Cowell (1980), a n d Shorrocks (1980, 1984) have shown tha t such measures alone satisfy a l l the desirable properties for any inequal i ty measure and are addi t ive decomposable. 1 2

Assume tha t the popula t ion is d iv ided into groups (according t o educat ion, for instance). Then a measure of inequal i ty is said to be addi t ive decomposable (see Shorrocks 1980) when i t can be w r i t t e n as:

/ = l(0g,agt Ig) = IB{P9, <*9) + X > G f y , a ^ 9 (!) 9

where 0g is the f ract ion of the labor force employed i n group g, ag

is i ts relative mean income, and Ig represents the wage dispersion w i t h i n th is group as measured by the index I .

The t e r m I B on the r ight side of equation 1 corresponds t o the inequal i ty between groups ( that is, the amount of inequal i ty t h a t wou ld be observed i n the case of an earnings red i s t r ibut i on w i t h i n each group, in such a way tha t , at the end, a l l workers in a group wou ld receive the same earnings). The second t e r m in the r i gh t -hand side (I\v) reflects the inequal i ty within groups; tha t is, the share o f overall inequal i ty associated w i t h factors other t h a n those involved in the par t i cu lar p a r t i t i o n under study. I t represents the degree o f inequal i ty tha t would be observed i f a l l groups had the same average earnings. Notice tha t Iw is a weighted average of the in terna l i n equalities, the weights, w(/3g,ag), being a funct ion of the popu la t i on share and average earnings of each group.

One can thus estimate the cont r ibut i on of a (the) given var i able (s) to the overall earnings inequal i ty at a given po in t i n t ime as the f ract ion of th is inequal i ty tha t would be e l iminated i f the average wage of a l l groups formed by tha t (those) variable(s) were equalized, whi le keeping the internal dispersions unchanged. The rationale beh i n d th is exercise is tha t the effect of th is (these) variable(s) is (are) captured by differences i n average earnings at the group level.

A m o n g the most commonly used inequal i ty indexes, the The i l T is one of the few tha t is addit ive decomposable. 1 3 The general stat istics needed for the decomposit ion by age, sector, level of schooling, hours worked, and status f rom 1988 to 1997 are shown in table 6.

Using equation (6) i n annex 2 and the year 1988 we have the fo l lowing results for the The i l :

A n n e x 2 e x t e n s i v e l y r e v i e w s d e c o m p o s i t i o n m e t h o d s .

F o r t h e d e c o m p o s i t i o n o f t h e T h e i l T , see R a m o s (1990 ) a n d a n n e x 2.


55

g l

0>

oo oo

h o

6 6

S 8

v '35

8 S

3 d

io eo ö d

o d

eo d

ci ci

V. o

h o

228 E S T U D I O S E C O N Ó M I C O S 19

91

Thei

l

0.33

1

0.40

7

0.33

2

0.25

5

0.31

7

0.26

9

0.37

2

0.33

3

0.33

1

0.44

4

0.37

2

0.42

8

0.34

0

0.17

5

1991

8 0.88

0.89

1.79

1.06

1.25

0.58

0.86

!

0.98

1.13

2.15

0.79

0.48

1991

«O.

0.05

7

0.18

0

0.02

7

0.06

8

0.29

3

0.13

6

0.17

2

0.54

0

0.28

8

0.04

6

0.16

7

0.15

0

1996

Thei

l

0.38

2

0.48

4

0.40

7

0.34

4

0.37

3

0.27

4

0.46

4

0.39

9

0.42

1

0.53

5

0.46

4

0.56

1

0.37

7

0.17

4

1996

0.91

0.90

1.90

1.03

1.23

0.58

1

0.84

0.96

1.16

2.18

0.75

0.47

1996

0.06

0

0.18

8

0.02

4

0.06

4

0.29

4

0.13

6

0.17

4

0.52

5

0.30

1

0.04

8

0.17

4

0.14

7

1992

Thei

l

0.40

9

0.41

5

0.38

4

0.31

0

0.38

0

0.29

1

1 0.

395

0.39

1

0.33

2

0.48

3

0.39

5

0.46

3

0.35

4

1992

1.06

0.92

1.77

1.12

1.12

0.70

0.87

0.95

1.20

2.44

0.89

0.54

1992

0.06

4

0.19

6

0.02

7

0.06

9

0.26

1

0.12

5

0.17

4

0.56

6

0.26

0

0.04

8

0.14

9

0.14

0

1988

Thei

l

0.22

4

0.41

5

0.23

0

0.19

1

0.28

0

0.38

5

0.32

7

0.27

8

0.28

0

0.43

8

0.32

7

0.54

9

0.33

8

0.21

0

1988

ö 0.91

1.01

1.39

1.12

1.10

0.73

0.89

0.96

1.20

2.32

0.97

0.58

1988

0.05

8

0.17

8

0.03

0

0.06

6

0.25

3

0.12

2

0.20

1

0.58

1

0.21

8

0.04

6

0.15

8

0.12

2

Var

iabl

e

Non

man

uf.

indu

s.

Com

mer

ce

Fina

nc s

er o

r re

nt

Tran

s or

com

mun

Soci

al s

ervi

ces

Oth

er s

ervi

ces

Tbt

al

Hou

rs

wor

ked

20-3

9

40-4

8

49+

Tot

al

Stat

us

Empl

oyer

Self-

empl

oyed

Info

rmal

sal

arie

d

M E X I C O : E V O L U T I O N O F E A R N I N G S I N E Q U A L I T Y 229 19

97

Thei

l

0.31

1

0.26

8

0.37

2

1997

ö 1.13

0.79

1997

0.56

7

0.07

0

1996

Thei

l

0.41

2

0.30

2

0.46

4

1996

ö 1.14

0.77

1996

0.55

8

0.07

2

1992

Thei

l

0.34

2

0.29

7

0.39

5

1992

ö 1.03

0.96

1992

0.60

2

0.06

2

1988

Thei

l

0.24

0

0.23

0

0.32

7

1988

ö O) CO Oi Oi d d 19

88

0.60

9

0.06

4

Var

iabl

e

Form

al s

alar

ied

Con

trac

t

Tot

al

er

I f g co s .8 § U

i l 9 . .. ^ u CP O Ui o t5 g


G G

T = J2a90glna9^Yla9f39T9 i s T = 0.323673 9=1 9=1

= 0.063887 (between groups) + 0.25978577 ( w i t h i n group ) .

Thus, 20% of the inequal i ty was due to inequal i ty between groups i n the case of education.

4.3. Results

The results for the exercise of stat ic decomposit ion are shown on table 7 . 1 4 We appl ied equat ion (6) in annex 2 to calculate the gross and marg ina l contr ibut ions in table 7. For example, consider again the case of education: the marg ina l con t r ibut i on is equal to the between groups effect (percentage) using education, sector or ac t iv i ty , and status i n the labor market as the p a r t i t i o n - minus the percentage between groups using only sector of ac t i v i t y and status i n the labor market as par t i t ions . The gross con t r ibu t i on is equal to the between-groups effect (percentage) using only education as the p a r t i t i o n .

Educat ion (the result of the interact ion between demand and supply) is the variable t h a t accounts for by far the largest share o f earnings inequal i ty i n Mexico, i n terms of b o t h gross and marg ina l contr ibut ions . The gross cont r ibut i on - t h a t is, the variable's explanato ry power when i t is considered alone-amounted to one- f i f th of t o t a l inequal i ty i n 1988 and one- th ird in 1997. 1 5 This result comes f r om using the inequal i ty between groups from the The i l above. The sl ight difference between the share presented above and the one i n table 8 may be due to missing values or the round ing of numbers. T h e marg ina l con t r ibu t i on - t h a t is, the increase in the explanatory power when the variable is added to a model tha t already has the other var iables- was remarkably stable and meaningful, staying around 21 percent throughout the per iod. The difference between the two cont r i bu t i ons has been growing over t ime, ind ica t ing tha t the degree of

S ince t h i s exerc i se is v e r y i n t e n s i v e i n t h e n u m b e r o f o b s e r v a t i o n s ( w h i c h

c o n s t i t u t e s i t s m a i n h a n d i c a p ) , t h e v a r i a b l e " h o u r s w o r k e d " was d r o p p e d i n o r d e r

t o a v o i d t h e p r o b l e m o f h a v i n g ce l ls w i t h t o o few o b s e r v a t i o n s . T h e d e c i s i o n w a s

m a d e t h r o u g h t h e c o m p a r i s o n a m o n g d i f f e r e n t c o m b i n a t i o n s o f v a r i a b l e s , w h e r e

h o u r s w o r k e d e n d e d u p b e i n g t h e least r e l e v a n t .

1 5 I n m o s t e a r n i n g s e q u a t i o n s for a n y c o u n t r y , t h e set o f m e a s u r a b l e o b s e r v a b l e

v a r i a b l e s e x p l a i n s a t m o s t 60 p e r c e n t o f t h e t o t a l v a r i ance . I n t h e U n i t e d S t a t e s ,

e d u c a t i o n a c c o u n t s f o r 10 p e r c e n t o f t h e t o t a l v a r i ance .

M E X I C O : E V O L U T I O N O F E A R N I N G S I N E Q U A L I T Y 2 3 1

O i

O

I l

.1

o>

•e*

I I

00 00 01

«0

i l

«3

I S !

es rH ^ ^ ^ <M LO

CS ^ CS M r f csi OD i d I s -

CD CD T—I

CO ^ rH <N <7> CN CO tO 0(3

CS O

CO r-i

00 00*

CO I C CS

s o ö o ü

w

O i O i

I 00 QO 01

C3

CO

00 CD £

ES

0 •e>a -fo

«S

1 CD

o

3

o

O i O i

I i

O i O i

O i o>

oo 00 01

I O

I es

CS CO

i l l

•e*

¡1

<3 e

I l

i l

' 1

1 2

cd CS

a o

o

IO CO

O i ^ CS co

Iß

lO cd

CO « ° ! id I

CO CS ® P <=? CS ^ CO

I s -

00 W H Tf W N

« °? CO

n* oo co iß CO N

oo I CS

I (-4

3 o


correlat ion w i t h other variables has been increasing. Th is means t h a t the " ind i r ec t " effects are becoming more impor tan t .

The other variables considered seem to be much less i m p o r t a n t . A l l three of them-but par t i cu lar ly economic sector and status i n the labor market-display an upward t rend i n the ir gross con t r i bu t i on and a decl in ing t r end in the i r marg ina l cont r ibut i on . Th is can be in ter preted as evidence tha t the interact ion between these variables and educat ion has become more intense. T h a t is, the workers ' skil ls are becoming increasingly more relevant to the determinat ion of the i r type of par t i c ipa t ion i n the labor market as wel l as to the i r pos i t i on across different economic segments of the economy. The same pa t t e rn holds when number of hours worked instead of sector is considered (see table 8).

There is an analogy between the stat ic decomposit ion and the regression analysis. However, there is an impor tan t difference. I n stat ic decomposit ion, the cont r ibut i on refers to the percentage in i n equal i ty whi le i n regression refers to the percentage in the variance. B o t h the variance and the The i l index are measures of inequal i ty b u t the The i l has some special properties tha t make i t preferable to the variance (see Ramos, 1990).

The ma in difference between the Fields' methodology and the methodology used in this paper is tha t the former assumes a funct iona l form. I n other words, the Fields' methodology assumes an earnings funct ional model such Y = X B whi le decomposit ion makes a p a r t i t i o n of the indiv iduals based on certain variables such as educat ion -sector of a c t i v i t y - status in the labor market bu t i t does no t assume a funct ional form. Therefore, the gross and marg ina l con t r i but ions i n decomposit ion do not have to coincide w i t h the R2. As discussed in annex 2, Fields (1996) uses the fol lowing equat ion:

A,-[/(.)] =

to determine the cont r ibut ion , where the SJ are the weights i n the inequal i ty measure and the sum of the SJ is equal to the R2.

The analysis of these results leads to the conclusion t h a t educat iona l inequal i ty is a key variable for understanding earnings inequali t y i n Mex i c o . 1 6 Though remarkable to some extent, th is finding

1 6 A d d i t i o n a l e v i d ence is t h a t t h e e x p l a n a t o r y p o w e r o f t h e c o m p l e t e m o d e l was 42 .5 p e r c e n t i n 1988, 45 .0 p e r c e n t i n 1992, 45 .5 i n 1996, a n d 48 .3 p e r c e n t i n 1997. T h i s m e a n s t h a t t h e m a r g i n a l c o n t r i b u t i o n o f e d u c a t i o n was a l m o s t e q u a l t o t h e j o i n t c o n t r i b u t i o n o f age, e c o n o m i c sec tor , a n d s t a t u s i n t h e l a b o r m a r k e t .


comes as no surprise i n the L a t i n Amer ican context. The results for some countries i n the region, where s imi lar exercises have been conducted, are reported i n table 9. Mexico stays i n the average range

T a b l e 9 Contribution of Education to Earnings Inequality:

International Comparison

Country Author(s) and reference Period Gross contri

bution (period)

Latin America Altimir and Pinera (1982) 1966-1974 17-38

Argentina Fiszbein, 1991 1974-1988 16-24

Brazil Ramos and Trindade, 1991 1977-1989 30-36

Vieira, 1998 1992-1996 30-35

Colombia Reyes, 1988 1976-1986 29-35

Moreno, 1989 1976-1988 26-35

Costa Rica Psacharapoulos et al, 1992 1981-1989 23-26

Perù Rodrfguez, 1991 1970-1984 21-34

Uruguay Psacharapoulos et al, 1992 1981-1989 10-13

Venezuela Psacharapoulos et al, 1992 1981-1989 23-26

for L a t i n Amer i can countries and displays a s i tuat ion close t o t h a t observed i n Colombia and Peru. However, education seems to be more impo r t an t for inequal i ty i n Braz i l and much less impor t an t i n Ar gentina and Uruguay. Th is is a comparison i n relative terms. Given t h a t i n Colombia and Peru, where education has a s imilar explanatory power, there is a lower degree of inequal i ty than in Mexico, the absolute con t r i bu t i on of education is higher i n Mexico. I n absolute terms, the con t r i bu t i on of education to inequal i ty i n Mexico is the second highest i n L a t i n America, after Braz i l . Moreover, wha t seems t o be

Szelcely ( 1 9 9 5 ) app l i e s t h e s t a t i c d e c o m p o s i t i o n o f t h e T h e i l t o t h e E N I G H f o r t h e

y ea rs 1984 , 1989 , a n d 1992 , u s i n g e d u c a t i o n , o c c u p a t i o n , r e g i o n , e c o n o m i c sec to r ,

a n d j o b s t a t u s as c o n t r o l v a r i ab l e s . T h e m a i n finding is t h a t t h i s set o f v a r i a b l e s

e x p l a i n s 55 , 58 , a n d 6 4 p e r c e n t o f i n c o m e d i s p e r s i o n , r espec t i v e l y , f o r e ach year ,

w i t h e d u c a t i o n a n d j o b s t a t u s b e i n g t h e r e l e v a n t va r i ab l e s .


par t i cu la r l y interest ing in the Mexican experience is the fact tha t the significance of education has been increasing over t ime . Therefore, the evo lut ion of educat ional d i s t r i bu t i on and the income prof i le associated w i t h i t , as wel l the l ink between changes in th is d i s t r i b u t i o n and changes in earnings inequality, are addressed in the nex t section.

5. T h e E v o l u t i o n o f E d u c a t i o n a l A t t a i n m e n t

Levels of educational a t ta inment have increased rap id ly i n most developing countries since the 1950s (Schultz 1988). A l t h o u g h Mexico also par took of t ha t development, there was a significant lag i n i ts educat ional indicators. Londono (1996), for example, po ints to an "educat ion def ic i t " , according to wh ich L a t i n Amer ican countries i n general, and Mexico in part icular , have approximate ly two years less educat ion than would be expected for their level of development. Elias (1992) finds t h a t education was the most impor tan t source o f improvement i n the qua l i ty of labor i n L a t i n Amer ica between 1950 and 1970, a l though such improvements d id not take place to the same extent i n Mexico as i n other countries i n the region. This changed dramat i ca l l y i n the 1980s. Figure 2 shows tha t , a l though Mexico's educat ional atta inment increased steadily after the 1970s, i t remained below the in ternat iona l t r end l i n e . 1 7 I n the 1980s, however, the g r ow th of educat ional a t ta inment i n Mexico accelerated, p e rm i t t i n g i t to catch up w i t h in te rnat iona l standards by 1990, where its placement i n figure 2 is s l ight ly above the t rend line.

The closure of Mexico's education gap vis-a-vis the rest of the wor ld was hastened in part by the country 's economic stagnat ion. Mexico's real G D P per capi ta i n the mid-1990s was roughly the same as i t had been i n the first hal f of the 1980s. Nevertheless, th is should not detract f rom the remarkable increase in schooling t h a t occurred dur ing the 1980s. Wh i l e the level of average schooling i n Mexico increased by roughly a year per decade dur ing 1960-80 ( f rom 2.76 to 4.77 years), i t increased by two years in the decade of the 1980s.

T h e s c a t t e r d i a g r a m is based o n 317 o b s e r v a t i o n s f r o m five years . T h e t r e n d

l i n e r ep r e s en t s t h e l east squares regress ion l i n e g i v e n b y

S = - 1 3 . 1 7 + 2.28Ln(GDPcap) A d j u s t e d R2 = 0.68

(—18.7) (26.0) t- va lues i n pa ren theses

T h e a p p l i c a t i o n o f R a m s e y ' s R E S E T t e s t t o t h i s r eg ress i on e q u a t i o n f a i l e d

t o d e t e c t a s p e c i f i c a t i o n e r r o r , u n l i k e w i t h t h e a l t e r n a t i v e s p e c i f i c a t i o n o f t h e

f o l l o w i n g t y p e : S — a -f bX -+- cX2.


This acceleration i n schooling was the product of concerted efforts to increase the coverage of basic education, combined w i t h advances made i n the reduct ion of p r imary school repet i t ion and d ropou t rates. The observations per ta in ing t o Mexico, ordered by date, are shown i n table 10.

F i g u r e 2 Cross-Country Relation between Educational

Attainment and GDP, 1960-1990

Ln (GDP per capita)

W i t h respect t o changes i n the d i s t r i bu t i on of schooling by socioeconomic groups, there are several aspects to be considered. I n par t i cu lar , three are examined here: the changes i n this d i s t r i bu t i on t h a t are related t o gender, economic sector, and age.

Table 11 shows the d i s t r i bu t i on of schooling by gender f rom 1988 to 1997. Even though there were clear improvements for b o t h males and females, wh ich signify an upgrade of educational a t ta inment , women achieved a better performance dur ing tha t per iod, especially at the t o p of the d i s t r i bu t i on . Improvements for males, i n contrast, were spread more evenly over the entire d i s t r ibu t i on . Nevertheless,


in 1997 women were undoubted ly more educated than men, as the i r cumulat ive d i s t r i bu t i on dominated tha t of men (see figure 3 ) . 1 8

T a b l e 10 Years of Schooling and GDP per-capita in Mexico,

(1960-1990)

Year Average schooling Ln (GDP per-capita in

US dollars)

1960 2.76 7.95

1970 3.68 8.29

1980 4.77 8.71

1985 5.20 8.63

1990 6.72 8.67

Source : A u t h o r ' s c a l c u l a t i o n s based o n B a r r o a n d Lee

d a t a set . T h e W o r l d B a n k .

T a b l e 11 Educational Distribution by Gender, 1988 and 1997, %

Educa,

group

Primary

incomplete

Primary

complete

Lower

secondary

complete

Upper

secondary

complete

Univer.

complete

1988

M a l e 19.0 30.1 24.5 14.6 11.8

F e m a l e 17.3 22.2 23.2 29.1 8.2

T o t a l 18.5 27.7 24.1 18.9 10.7

1997

M a l e 13.0 25.7 28.4 18.0 14.9

F e m a l e 12.2 20.0 22.3 30.1 15.5

T o t a l 12.7 23.7 26.3 22.1 15.1

Source : A u t h o r ' s c a l c u l a t i o n s based o n E N E U s u r v e y ( t h i r d q u a r t e r ) .

T h i s is t r u e f o r t h e o v e r a l l d i s t r i b u t i o n i n 1997 r e l a t i v e t o t h a t i n 1988.


F i g u r e 3 Cumulative Educational Distribution

by Gender, 1997

University Upper sec. Lower sec Pr im com Pr imincom

W i t h respect t o the d i s t r i bu t i on of schooling by economic sector, table 12 shows large heterogeneity i n the d i s t r i bu t i on of schooling across sectors f rom 1988 to 2002. The results suggest tha t w i t h i n th is heterogeneity, the financial sector uses more h igh ly skil led labor force. I t seems t h a t the p r imary sector, together w i t h other sectors employ a more low-ski l led labor force. T h i r d , commerce is very heterogeneous i n i ts labor force composit ion.

Ano ther relevant observation is tha t educational a t ta inment by age group also improved, as the d i s t r i bu t i on by educational level was higher i n 1997 t h a n i t was i n 1988 (table 13). I n an a t t empt t o reach a better understanding of th is event, i t is interest ing to contrast the t ime and cohort effects. 1 9 I n order to do th is , one can look at the first age groups, 16-25 and 26-34, l ike synthetic cohorts. Namely, the 26-34 age group i n 1997 can be compared direct ly to the 16-25 age group i n 1988, and, t o a lesser extent, the 35-49 age group i n 1997 can be compared to the 26-34 age group i n 1988. From 1988 t o 1997,

T h e t i m e ef fect re fers t o t h e c o m p a r i s o n o f t h e same age g r o u p i n t w o d i f

f e r en t p o i n t s o f t i m e .


-2> «3 •cS> -+0

tí

i o

1^3

, 3

"*3

<3

00 oo OÍ

<3 ^ I

I ¡

§• g

°? o

2 °°

^ oo ^ o ^ M h H i ñ ^ i r l H (N °

rH 10 CM 1 0 <M ^ rH

M O O N O O i C O N l N ' - |

O CO LO l>-H có oó oó CN CO CN CM

S ?D ^> N l O N ¿ N CO rH CO CM

i— I O H CO CO Tí* rH 00

3 tí

S

OH 2

-tf CO 00 L O

^ H o i oo rH rH CO rH

S § 8 $ 8 .a .a > >

s 6 o

O

«3

oo 2 en

O i (N O O oo o TJ<

rH rH CS|

h N « 5 | N (N 0 0 6 rH CO rH C0

N ^ N O H CD (N CN CO ( N

rH O CO ^ oó H oo d

H M H

3 C!

2 <̂ S S a S o o

M E X I C O : E V O L U T I O N O F E A R N I N G S I N E Q U A L I T Y 239 U

nive

rsity

com

plet

e

35.6

8.0

30.0

2.4

15.1

2002

4.6

7.8

9.0

7.8

39.1

8.4

31.9

3.2

14.0

Upp

er

seco

ndar

y

com

plet

e

40.3

23.9

29.6

11.1

22.1

2002

7.1

11.7

7.7

17.5

34.5

16.2

16.2

8.9

13.7

Low

er

seco

ndar

y

com

plet

e

16.1

32.2

21.1

24.6

26.3

2002

23.2

40.7

27.0

39.5

21.6

39.4

31.8

32.8

34.9

Pri

ma

ry

com

plet

e

5.4

26.8

13.2

35.7

23.7

2002

35.0

30.6

34.8

25.4

4.3

28.6

14.9

36.2

26.1

Pri

ma

ry

inco

mpl

ete

2.7

9.1

6.0

26.2

12.7

30.1

9.3

21.5

9.8

0.6

7.5

5.3

18.9

11.3

Ed

uca

tion

al

grou

p

and

year

Fin

ance

ser

. or

ren

t

Tra

nsp

or

com

muni

Soc

ial

serv

ices

Oth

er s

ervi

ces

Tot

al

Pri

mar

y

Man

ufa

cturi

ng

Non

man

uf

indust

ry

Com

mer

ce

Fin

ance

ser

. or

ren

t

Tra

nsp

or

com

muni

Soc

ial

serv

ices

Oth

er s

ervi

ces

Tota

l

T3


the percentage of persons i n the category of incomplete p r i m a r y schoo l ing decreased, and th is decline was higher than tha t experienced by the 16-25 age group (who were i n the 26-34 age group i n 1997). The opposite t ook place for the highest level of ins t ruc t ion . I n other words, improvements throughout the educational process i n Mexico were signif icant, b o t h for those entering the system (higher coverage) and for those already in i t (higher efficiency).

T a b l e 13 Educational Distribution by Age Group, 1988 and 1997, %

Age Primary Primary Lower Upper Univer.

group incomplete complete secondary secondary complete

complete complete

1988

16-25 8.5 26.5 36.7 23.7 4.6

26-34 12.6 23.7 23.1 22.5 18.2

35-49 24.0 33.3 16.8 14.3 11.6

50-65 46.1 27.2 9.9 9.0 7.8

T o t a l 18.5 27.7 24.1 18.9 10.7

1997

16-25 5.8 23.8 38.7 25.5 6.2

26-34 6.9 19.5 28.1 27.0 18.5

35-49 14.8 25.8 19.5 19.1 20.7

50-65 37.3 27.6 11.5 10.6 13.0

T o t a l 12.7 23.7 26.3 22.1 15.1 Source : A u t h o r ' s c a l c u l a t i o n s based o n E N E U ( t h i r d q u a r t e r ) .

Also concerning the interact ion between age and education, one can argue tha t developments i n the educational system have more impact on the new generations than on the elderly. To investigate th is , i t is necessary to contrast the behavior of inequal i ty between different age groups to tha t of inequal i ty w i t h i n synthetic cohorts and in re lat ion to education. As seen, the younger cohorts are, i n fact, better educated t h a n the older ones. A t the same t ime, the " w i t h i n " income dispersion for the youngest cohorts seems to increase over t ime , compared w i t h the internal The i l in 1997 and 1988 (see table


6). Thus i t becomes easier to understand why the gross c o n t r i b u t i o n of age t o inequal i ty has been r is ing, whi le its marg ina l c o n t r i b u t i o n has been decreasing. I n other words, differences i n b o t h educat ional a t ta inment and d i s t r i bu t i on among cohorts have become pronounced i n recent t imes, leading t o a higher (negative) correlat ion between educat ion and age.

6. T h e D y n a m i c D e c o m p o s i t i o n

I n order t o address the re lat ionship between education (the result o f the interact ion between supply and demand) and earnings inequal i ty , i t is necessary t o expla in how the labor market determines the earnings differentials among workers w i t h different educational a t t r ibutes . Th is re lat ionship can be viewed as determined by two elements: (a) the d i s t r i bu t i on of educat ion itsel f and (b) the way the labor market rewards educat ional a t ta inment . The first element reflects a preexis t ing social s t rat i f i cat ion t h a t already entails some inequal i ty due to reasons other t h a n the workings of the labor market itself. T h e second is associated w i t h the degree t o wh ich this preexist ing inequal i ty grows in t o earnings inequal i ty due t o the performance of the labor market ( tha t is, demand behavior) .

F igure 4 shows the d i s t r i bu t i on of education i n the hor i zonta l axis (mt is an indicator of the average schooling of the labor force, and it represents i ts dispersion), whi le the vert ica l axis presents the d i s t r i bu t i on of earnings. The first quadrant depicts the interact ion between the preexist ing condit ions (the d i s t r i bu t i on of education) and the workings of the labor market , th rough the steepness $t o f the income profi le related to education. Therefore, at a certa in po in t i n t ime we see (a) the higher rat, the larger the average earnings; (b) the lower it, the smaller the earnings inequal i ty ; and (c) the higher st, the higher the g rowth of preexist ing disparit ies, and, accordingly, the higher the earnings inequality. As these indicators change over t ime, they w i l l induce changes i n the income d i s t r i bu t i on : changes i n it, assuming st constant, w i l l change earnings inequal i ty due t o changes in the composi t ion of the labor force (the so-called a l locat ion-populat ion effect), whereas changes i n st w i l l alter the earnings differentials (the income effect).

Barros and Reis (1991) develop three synthetic measures for the indicators mt (average schooling), it (schooling inequal i ty ) , and s t

( income profi le) , based direct ly on the def in i t ion of the The i l T index (see annex 2). The figures for Mexico f rom 1988 to 1997 are presented i n table 14. Average schooling improved somewhat, b u t the


inequal i ty of the d i s t r i bu t i on of education deteriorated, whereas the income profile, wh i ch is related to the returns to schooling, became much steeper. Th is means tha t there was a shift in demand t oward h igh ly skil led labor tha t was not met by an increase i n supply. T h i s probably occurred as a result of the accelerated pace of ski l l -biased technological change faci l i tated by the increased openness of the Mex ican economy. The same pat t e rn observed for the overall sample holds for the 16-25 age group: the mt rose f rom 0.561 i n 1988 t o 0.574 i n 1997; the i t increased f rom 0.0196 to 0.0218, whereas the st doubled, r is ing f rom 0.0196 t o 0.0383.

F i g u r e 4 A Stylized View of Education and

Labor Market Interaction

6.1. Methodology

The dynamic decomposit ion analysis is a suitable too l for t rans la t ing this styl ized view in quant i ta t i ve results, g iv ing one a bet ter unders tanding of the socioeconomic transformations responsible for changes i n the earnings d i s t r ibu t i on . Besides pe rm i t t i ng ident i f icat ion of the


relevant ind i v idua l variables, i t also helps i n understanding the nature of the con t r i bu t i on o f each variable t o the evolut ion of earnings inequal i ty over t ime .

T a b l e 14 Synthetic Indicators of Schooling Distribution and Income Profile

1988-1997

Year 1988 1992 1996 1997

Tilt 0.476 0.491 0.511 0.510

H 0.066 0.069 0.076 0.075

St 0.066 0.102 0.122 0.111

Source : A u t h o r ' s ca l cu la t ions based on E N E U

( th i rd qua r te r ) .

Ramos (1990), fo l lowing Shorrocks (1980), shows t h a t i t is possible t o break down the change i n inequal i ty between two points i n t ime. Th i s is done according to whether the change can be a t t r i b u t e d t o changes i n the socioeconomic groups relative to income, t o group size, or t o internal inequal ity, th rough use of the The i l T index. I n generic terms, as shown before i n a s l ight ly different way, for a given p a r t i t i o n o f the popu la t i on , the inequal i ty indexes of th is class can be w r i t t e n as:

I = I(ag,l3gJg) (2)

where ag is the ra t i o between the average income of group g and the average income of the whole populat ion , (3g is the p ropor t i on of the popu la t i on i n group and Ig is the internal dispersion of incomes i n group g.

Of course, the as are related to the indicator st i n the previous p ic ture , and the 0s refer t o mt and it. I n th is context, the population or allocation effect corresponds to the var iat ion induced i n the inequal i ty index I by modif icat ions i n the al location of the populat i o n among the groups (changes i n the /?s), w i t h no direct changes in the group's relative incomes ( a s ) . 2 0 The income effect corresponds to

T h e difference between th i s a n d w h a t K n i g h t a n d Sabot (1983) c a l l the " compres s i on " effect is tha t i n the present exercise we are inc lud ing the ind i rect change i nduced in / th rough the va r i a t ion in the weights of the Igs. O f course,


the changes i n J induced by changes in group incomes (as), w i t h o u t changing the groups' shares of the popu la t ion (/?s), and the in t e rna l effect is the change in the inequal i ty caused only by modi f icat ions in dispersions at the group level (the Igs).21 The expressions corresponding to the The i l T index are derived in annex 2.

6.2. Results

The results of the decomposit ion of the variations in the The i l T i n dex for different intervals of t ime are shown in table 15. F i r s t , when the variables are considered alone, education made the highest gross con t r ibu t i on t o the changes in earnings d i s t r ibu t i on . Second, b o t h the a l locat ion and the income effect were positive i n a l l periods. T h i s means tha t changes in the d i s t r i bu t i on of education and i n the re l ative earnings among educational groups were always i n phase w i t h alterat ions in the earnings d i s t r ibu t i on . Namely, when the income profile related to education became steeper and the inequal i ty of educat ion grew, the earnings d i s t r i bu t i on worsened (as i n the 1988-1992, 1992-1996, and 1988-1997 periods) and vice versa (as in the 1996-1997 per iod) .

T h i r d , the income effect is always prevalent. I f one considers, for instance, the 1988-1997 per iod, changes in relative earnings among educational groups alone would have generated a larger deter iorat ion in the earnings d i s t r i bu t i on than the one observed. To a lesser extent, the same holds t rue for the other pe r i ods . 2 2 Even the decrease i n i n equal i ty observed between 1996 and 1997 is par t ia l l y explained by the changes i n relative earnings (the income profile related to educat ion became less steep i n this per iod, as shown i n table 15). Therefore, i t seems reasonable to conclude tha t the income effect is the leading

t h e i n d i v i d u a l ' s a s change as t h e (3s change , s ince t h e o v e r a l l average i n c o m e

is a l t e r e d . T h i s i n d i r e c t i m p a c t is a lso c o m p u t e d i n t h e c o m p o s i t i o n effect (see

a n n e x 2 ) .

2 1 T h e m e t h o d o l o g y a p p l i e d b y F i e l d s (1996 ) a n d B o u i l l o n , L e g o v i n i , a n d

L u s t i g ( 1998 ) m a k e s i m p o r t a n t a s s u m p t i o n s . I n c o n t r a s t , Szeke ly ( 1 9 9 5 ) , i n o r d e r

t o e x p l a i n t h e changes i n i n e q u a l i t y b e t w e e n t w o p o i n t s i n t i m e , a p p l i e s a m e t h o d

o l o g y t h a t d i f f e rs d r a s t i c a l l y f r o m t h e d y n a m i c d e c o m p o s i t i o n s ince he does n o t

c o n t r o l f o r t h e ef fects t h a t a r i se f r o m changes i n t h e p o p u l a t i o n d i s t r i b u t i o n a n d

f r o m changes i n t h e r e l a t i v e e a r n i n g s o f i n c o m e g r o u p s c ons i d e r ed i n t h e p a r t i t i o n

o f t h e p o p u l a t i o n (see a n n e x 2 ) .

2 2 O f course , t h e e x p l a n a t i o n for s u c h a p h e n o m e n o n is t h a t changes i n t h e

o t h e r v a r i a b l e s a t t e n u a t e d t h e changes i n t h e r e w a r d s t o e d u c a t i o n .


force beh ind the increase i n inequality, and this , i n t u r n , suggests t h a t the workings of the labor market , and i ts interact ion w i t h educat ional policies, should be thoroughly examined.

F o u r t h , the significance of changes i n the d i s t r i bu t i on of educat i o n remains h igh even when one controls for changes i n other relevant var iab les . 2 3 As a mat te r of fact, w i t h the exception of the 1996-1997 t rans i t i ona l per iod, the marg ina l con t r ibu t i on of age, economic sector , and status i n the labor market is usually negative. Th i s means t h a t changes i n these variables reduced the effects induced by changes related t o education, as most of the t ime they reduced inequal i ty after the influence of educat ion is taken in to account.

T a b l e 15 Results of the Dynamic Decomposition, 1988-1997

Period and variable Allocation Income Gross Marginal

1988-1992

E d u c a t i o n 11.4 58.8 70.2 30.5

A g e -1.8 21.9 20.2 -5.2

E c o n o m i c s ec t o r -0.6 7.8 7.1 -17.7

S t a t u s 3.9 15.1 19.0 -7.4

1992-1996

E d u c a t i o n 23.9 32.8 56.7 27.6

A g e 11.1 10.5 21.6 10.5

E c o n o m i c s ec t o r -5.4 25.4 20.0 10.5

S t a t u s 1.2 12.4 13.6 -4.2

Szeke l y ( 1995 ) c o n c l u d e s t h a t , f o r t h e 1984-1989 p e r i o d , t h e v a r i a b l e s t h a t

c o n t r i b u t e d s i g n i f i c a n t l y t o e x p l a i n i n g i n e q u a l i t y we r e e d u c a t i o n a n d e c o n o m i c

sec to r , w h i l e e d u c a t i o n a n d j o b s t a t u s we r e s i g n i f i c a n t i n t h e 1984-1992 p e r i o d .

T h e se l ec t ed v a r i a b l e s we r e e d u c a t i o n , o c c u p a t i o n , r e g i o n , e c o n o m i c s ec t o r , a n d

j o b s t a t u s . B o u i l l o n , L e g o v i n i , a n d L u s t i g ( 1 9 9 8 ) , a p p l y i n g B o u r g u i g n o n ' s m e t h o

d o l o g y t o t h e E N I G H , find t h a t t h e r e t u r n ef fect t o t h e h o u s e h o l d c h a r a c t e r i s t i c s

( age/gender , e d u c a t i o n / a g e , assets) e x p l a i n e d 49 p e r c e n t o f t h e inc r ease i n t h e

G i n i b e t w e e n 1984 a n d 1994 , e d u c a t i o n b e i n g t h e m o s t i m p o r t a n t e x p l a n a t o r y

v a r i a b l e . T h e r e g i o n ef fect ( u r b a n / r u r a l ) was 9 p e r c e n t , t h e s o u t h ef fect was 15

p e r c e n t , a n d t h e p o p u l a t i o n ef fect was 23 p e r c e n t .


T a b l e 15 (continued)

Period and variable Allocation Income Gross Marginal

1996-1997

E d u c a t i o n 2.2 15.5 17.7 24.2

A g e -0.4 5.9 5.5 12.5

E c o n o m i c s e c t o r 0.4 1.0 1.4 18.4

S t a t u s 1.4 6.1 7.5 7.8

1988-1997

E d u c a t i o n 35.8 108.4 144.1 33.7

A g e 7.4 32.7 40.1 -19.9

E c o n o m i c sec to r -6.6 43.2 36.6 -40.6

S t a t u s 9.0 20.2 29.2 -35.6

Sourc e : A u t h o r ' s c a l c u l a t i o n s based o n E N E U s u r v e y ( t h i r d q u a r t e r ) .

The last per iod, f rom 1996 to 1997, deserves special comment . F i rs t , inequal i ty was substant ia l ly reduced. Second, once more, a l terat ions were associated w i t h education, now work ing i n the other d i rect ion, and such alterations appear to be the ma in factor responsible for the reduct ion in inequality. As can be seen f rom the synthet ic indicators, a smal l improvement i n the d i s t r i bu t i on of schooling t ook place d u r i n g the per iod, as wel l as a sizable decrease i n the steepness of the income profile related to education. A l l other variables, as observed for other periods, also contr ibuted to an improvement i n earnings inequality.

Table 16 shows the results of the same k ind of decomposit ion for Braz i l , Argent ina , and Peru. The significance of educat ion as an explanat ion of changes in inequal i ty seems to be a common pa t t e rn i n L a t i n Amer ican countries. Moreover, the relevance of the income effect over the al locat ion (populat ion) effect is shared by a l l countries where a s imilar analysis was carried out . I n the Mex ican case, however, the figures are higher than those for other countries (and in a shorter per iod of t ime ) . Th is means tha t changes i n the s tructure of supply and demand for labor, which are greatly affected by the educational and macroeconomic policies followed by the country or by the i r interact ion w i t h the workings of the labor market , were par t i cu lar l y relevant for the earnings d i s t r ibu t i on .

MEXICO: EVOLUTION OF EARNINGS INEQUALITY

N CO CO rH ^ ò 00 4 1-1 co co

t o ^ <^ ^ ir» CO

0 0 ^ 00 0 0 0 5 0 5

^ O

& I s

^3

N

bp

CO .5 a


7. T h e E v o l u t i o n a n d S t r u c t u r e o f t h e R a t e s o f R e t u r n s t o E d u c a t i o n : A n A p p l i c a t i o n o f Q u a n t i l e R e g r e s s i o n

The increase in earnings inequal i ty is not the result of a worsening in the d i s t r i bu t i on of education, whereas the income profile, wh i ch is related to the returns to schooling, is much steeper. I n l i ght of th i s evidence, this section analyzes the s t ructure and evolut ion of the ra te of re turns to education. A l t h o u g h this is a common procedure, i t is also an i m p o r t a n t caveat: the in ternat iona l comparison becomes c u m bersome because the s t ructure of the educational process i n Mexico is different t h a n tha t of other countries.

7.1. Quantile Analysis

Average rates of returns hide major variations. I t is necessary t o take a pre l iminary look at the relat ionship between the d i s t r i b u t i o n of earnings and educational a t ta inment i n Mexico. For th is purpose, real hour l y earnings by quant i le (0.10, 0.25, 0.50, 0.75, and 0.90) and the mean are c o m p u t e d . 2 4

As can be seen f rom figures 5 th rough 7, the curves do not cross each other for a l l educational categories or for al l periods. Th i s suggests tha t there is a s t r i c t dominance of the education variable throughout the earnings d i s t r ibu t i on . I n other words, there is a posit ive re lat ion between educational level and hour ly earnings th rough out the d i s t r ibu t i on . Those figures also show t h a t the difference among quantiles ( that is, f rom the t en th to the twenty- f i f th percenti le, f rom the twenty- f i f th to f i f t i e th percentile, and so for th ) changes throughout educational levels (the greater the level of educat ion, the larger the difference among quantiles of hour ly earnings). I n add i t i on , the difference between quantiles also changes over t ime . These pat terns may provide empir ica l evidence tha t there are differences in the increase in real hour ly earnings throughout educational d i s t r i b u t i o n and t ime . The quanti le analysis provides a complete assessment of the impact of many variables (education, age, gender, economic sector, labor market status, region, and so forth) throughout the earnings dist r i b u t i o n . F inal ly , for a l l educational categories, real average hour l y earnings are greater t h a n the median, and the d i s t r i bu t i on of hour l y earnings is always right-skewed.

^ T h e t h i r d q u a r t e r o f t h e E N E U d a t a for 1988, 1992, a n d 1996 is used . T h e

s a m p l e is d e s c r i b e d i n t h e a p p e n d i x .


F i g u r e 5 Real Hourly Earnings by Quantile

and Educational Level, 1988

8 W \ ft \

M tUi i

1 U

I M

/ /

/ /

/

few* fria m LsKvsee Upset



* H i

ft

t in i

I M

I

/

/ /

/

/ /

'




tmkm hw m Imm [ § » fcá?

Source : A u t h o r ' s e s t i m a t i o n s based o n E N E U .

I n sum, these results suggest tha t a quanti le method of es t imat ion is needed to provide a better understanding of the rate of returns t o educat ion.

7.2. Methodology

I n th is paper, we used the basic Mincer model and then added var i ables such as region, then sector, and then status in the labor market one at a t ime to assess the ir effect on the returns to educa t i on . 2 5

Then the earnings funct ion can be described as follows:

log Y t - at + Stbt + Xtct + utit = 1988, 1992, 1997, 2002 (3)

where

2 5 P r i m a r y i n c o m p l e t e ( e d u c a t i o n ) a n d M e x i c o C i t y ( r e g i on ) a c t e d as re f e rence

v a r i a b l e s .


Yt Vector of i nd i v i dua l hour ly earnings i n t ime t,

at L o g a r i t h m of the mean real hour ly earnings of the reference g roup

i n t i m e t, bt Earnings dif ferential associated w i t h education i n t ime £, 2 6

ct Vector of earnings differentials related to the contro l variables i n

t i m e £, St Vector of educat ional a t ta inment i n t ime t, Xt M a t r i x of contro l variables for t ime t, ut Vector of residual terms for t ime t[E(ut = 0 and E(ut,ut = 0 ] . 2 7

Figure 8 shows average rates o f r e tu rn based on the basic Mincer model . Th i s figure shows average h igh rate of returns t o educat ion, especially for t e r t i a r y education, bu t returns fell for most categories since around 1997.

These earnings functions can be f i t ted using least squares estimat ion . However, a new technique of est imat ion has been developed recently: quant i le regression. Th is technique has usually been appl ied to analyze the determinants of wage s t ructure as wel l as the rate o f returns t o investment i n education throughout the earnings d i s t r i b u t i on . Buchinsky (1994, 1995, 1998) applies this technique t o the U S labor market i n order to assess the wage structure and its changes. Other authors also used quanti le regression to s tudy the pa t t e rn o f U S wage differentials between state and local government employees and the i r pr ivate counterparts. The quanti le regression analysis also has been applied t o other countries: Shultz (1998) and Mu l l e r (1998) i n Canada, Abadie (1997) i n Spain, and Montenegro (1999) i n Chi le . Th is methodology has never been applied i n Mexico. Th i s paper fo l lows closely the methodology proposed by Buchinsky (1994, 1995, 1998 ) . 2 8

The quant i le regression models have some desirable characteristics, especially for analyzing a certain variable throughout i ts dist r i b u t i o n . The m a i n features of these models can be summarized as follows:

A s t h i s is a c a t e g o r i c a l v a r i a b l e , one has , i n f a c t , a v e c t o r (bit) o f e a r n i n g s

d i f f e r e n t i a l s , w i t h e a c h o f i t s c o m p o n e n t s r e p r e s e n t i n g t h e e a r n i n g s d i f f e r e n t i a l

b e t w e e n t h e i t h e d u c a t i o n a l g r o u p a n d t h e re ference g r o u p ( p r i m a r y i n c o m p l e t e )

i n t i m e t. 27

I n a d d i t i o n , o n e has t o a s sume t h a t t h e r e s i d u a l t e r m is n o t c o r r e l a t e d w i t h

t h e u n o b s e r v e d d e t e r m i n a n t s o f i n d i v i d u a l e a r n i n g s ( f a m i l y b a c k g r o u n d , n a t u r a l

a b i l i t y , a n d so f o r t h ) . 28

T h e a u t h o r shows t h a t t h e q u a n t i l e m e t h o d is r o b u s t even i n t h e presence

o f p o s s i b l e se l f - se l ec t i on .


• The model can be used to characterize the entire cond i t i ona l dist r i b u t i o n of the dependent variable,

• The quanti le regression objective funct ion is a weighted sum o f absolute deviations, wh ich gives a robust measure of locat ion, so t h a t the est imated coefficient vector is not sensitive t o out l i e r observations of the dependent variable,

• W h e n the error t e rm is non-normal , quanti le regression est imators may be more efficient than least squares estimators,

• Different solutions at d is t inct quantiles may be interpreted as differences i n the responses of the dependent variable t o changes i n the independent variables at various points i n the cond i t i ona l d i s t r i bu t i on of the dependent variable (see Buchinsky 1998),

• The earnings funct ion (equation 3) can be r ewr i t t en as a quant i l e regression model . Then we have logY^ = Xij3$ + w i t h Quante(logYi\Xi) = Xi0o(I = l , . . . , n ) , where 0Q and X{ are K x 1 vectors, and Xu = 1. Q uante (log Y\X) denotes the 0 t h condi t iona l quant i le of Y given X. Also let fUe(-\X) denote the density of fie given X. I t follows tha t Quant(fio\X) = 0.

The X{ vector includes the set of explanatory d u m m y variables as wel l as the controls. For an extensive review, see Buchinsky (1998).

B y using the regression coefficients, one can compute the different ials and marg ina l value related t o each level of education. Accord ing to the specification of the earnings functions, for least squares as we l l as for quant i le regression, the exponential of the di f ferential associated w i t h the j t h category of the ith variable, exp(c^ ) , corresponds to an estimate of how much higher, on average, the earnings of an ind i v idua l i n tha t category are relative to the earnings of an i n d i v i d ual i n the reference group for tha t variable, a l l other a t t r ibutes be ing i d en t i c a l . 2 9 The marg ina l value of some educational level j in t ime t(MV?duc) can be interpreted as the earnings differential for th is level relative to the previous one, as fo l lows: 3 0

MVftduc = bjt/bU-1)t for j > l (4)

and M V e d u c = b j t for j = z l

2 9 I f t h e d i f f e r e n t i a l is c lose t o zero , t h e n i t c a n be i n t e r p r e t e d as b e i n g a p

p r o x i m a t e l y e q u a l t o t h e average p e r c e n t a g e increase i n e a r n i n g s a s soc i a t ed w i t h

a m o v e m e n t f r o m t h e re ference g r o u p t o j t h ca t ego ry , ceteris paribus. 3 0 S i m i l a r l y , t h e d e f i n i t i o n app l i e s t o t h e r e s u l t s o f t h e q u a n t i l e r e g r ess i on a p

p r o a c h . T h e o n l y d i f f e r ence is t h a t i n t h i s s i t u a t i o n one needs a n a d d i t i o n a l

s u b s c r i p t (0) t o ass ign t h e q u a n t i l e


4 - Piirary complete -*- Lower secondary complete -k- Upper secondary complete -fc- Tertiary education

N o t e : T h e y e a r l y r a t e o f r e t u r n r ep resen ts t h e p e r

c en tage o f inc rease over t h e base wage f o r a n a d d i t i o n a l

y ea r o f a c e r t a i n l eve l o f e d u c a t i o n . A l l t h e coe f f i c i ents

a r e s t a t i s t i c a l l y s i g n i f i c a n t a t 5 % l eve l , a n d c o n d i t i o n e d t o

age, s q u a r e d age, g ender , r e g i o n ( N o r t h , C e n t e r , S o u t h , a n d

M e x i c o C i t y ) . Source : A u t h o r ' s e s t i m a t i o n s u s i n g t h i r d

q u a r t e r o f E N E U f r o m 1988 t o 2 0 0 1 a n d t h i r d q u a r t e r a n d

u r b a n s e c t i o n o f E N E T 2002 .

7.3. Empirical Results

B o t h ord inary least squares and quanti le regression models are estim a t e d . 3 1 However, before analyzing the quanti le rate of re turns to educat ion, i t is w o r t h invest igat ing the role o f each explanatory var i able i n the determinat ion of earnings. For th is purpose, several regressions are f i t t ed add ing the explanatory variables one at a t ime . Th is exercise has two advantages: (a) i t allows us to assess the marg ina l

3 1 T h e 6s p a r a m e t e r s i n t h e q u a n t i l e r eg ress i on were 0 . 1 , 0 .25, 0 .5 , 0 .75 , a n d

0 .9 , f o l l o w i n g a c o m m o n p r o c e d u r e i n t h e l i t e r a t u r e .


cont r ibu t i on of each explanatory variable, and (b) i t shows the exp lanatory power of each variable throughout the condi t ional earnings d i s t r i bu t i on . Cragg and Epe lbaum (1996) perform a s imi lar exercise as wel l as other authors such as Meza (1999), and Lopez-Acevedo and Wa l t on (2004). Nevertheless, the occupation variable was left out of th is study, since as s t ructured in E N E U - I N E G I questionnaires, i t is h igh ly correlated to the indiv idual ' s level of education. As shown in table 2.A3, educat ional level and occupation are h ighly correlated, whi le educat ion and the rest of the explanatory variables are weakly correlated.

Educat ion is the most impor tan t variable in the explanat ion o f earnings inequality. However, we can assess the importance of other explanatory variables using the estimates of differentials i n educat iona l l eve l I f the changes in such differentials, in a given per iod of t ime , have been smoothed by some other explanatory variable, then tha t variable is a measure of some specific sk i l l . For do ing such an assessment, we compute the relative change i n the differentials by educational level i n 1988-1992 and 1992-1997 periods. The estimates are presented below.

Table 17 shows tha t earnings differentials were reduced by i n t r o duc t i on of the economic sector variable in the regression for the 1992-1997 per iod, par t i cu lar l y for t e r t i a ry education, whi le the reduct ion was very smal l for the 1988-1992 per iod (see Cragg and Epe lbaum 1996). Labor market status seems to have the same reduct ion effect on earnings differentials as the economic sector variable. These results suggest tha t the degree of correlat ion between educat ion and economic sector, as wel l as labor market status, increased t h r o u g h t ime. Table 17 also shows tha t region had an almost insignif icant effect on earnings differentials.

A t this po int , one tentat ive conclusion emerges: the reduct ive effect on the earnings differentials of bo th the economic sector and the labor market status variables was signif icantly larger i n 1992-1997 t h a n i n 1988-1992 (before the trade agreement). Th is means t h a t the re lat ionship between education and the types of specific skil ls acquired th rough such variables changed in the labor market . Thus a worker's insert ion in to the labor market and economic sector variables were a consequence of skills differentials and a t t r i bu t ed not solely t o education.

I n order t o have a precise assessment of the marg ina l value of educational level, we estimated several models as discussed earlier, s t a r t ing f rom the basic Mincer equation and then adding variables one at a t ime (region, status in the labor market, and sector). The


Eco

nom

ic s

ecto

r

regi

on a

nd s

tatu

s

1992

1997

0.01

0.00

0.00

0.04

Eco

nom

ic s

ecto

r

regi

on a

nd s

tatu

s

1988

1992

-0.0

2

-0.0

3

0.01

0.15

Eco

nom

ic s

ecto

r

and

stat

us

1992

1997

0.02

0.00

0.00

0.04

Eco

nom

ic s

ecto

r

and

stat

us

1988

1992

-0.0

2

-0.0

3

0.01

0.15

Sta

tus 19

92

1997

0.03

0.03

0.04

1

0.09

Sta

tus

1988

1992

-0.0

4

-0.0

6

-0.0

2

0.12

Eco

nom

ic s

ecto

r

1992

1997

0.02

0.03

0.04

i

0.08

Eco

nom

ic s

ecto

r

1988

1992

-0.0

1

-0.0

5

0.02

0.15

Con

trol

ling

for

none

1992

1997

0.05

0.08

iro 0.

18

Con

trol

ling

for

none

1988

1992

-0.0

3

-0.0

6

-0.0

2

!

0.14

Ed

ucat

ion

leve

l

Pri

mary

com

ple

te

Low

er s

ec.

com

ple

te

Upper

sec

.

com

ple

te

Un

iver

sity

com

ple

te

CD

I o

CO — y Q. U

¡1 cu

8

P w

ö o

'S

a o

1 g

8 1

3 <

2 o

CO


results presented next are based i n the basic Mincer equat ion. I n general, the ord inary least squares ( O L S ) estimates are qu i te

s imi lar t o the ones obtained by the quanti le regression approach for 0 — 0.5, 0.75. However, estimates obtained by the quanti le regression technique tend to increase as one moves f rom the r ight to the left o f the condi t iona l earnings d i s t r i bu t i on , par t i cu lar ly for the upper levels of educat ion. I n summary, the results have three strong impl i ca t ions : (a) educat ion does play a crucial role in the process of earnings format i on , (b) its effect is not the same throughout the condi t iona l earnings d i s t r i bu t i on , and (c) the marg ina l value of education has not changed signi f icantly i n basic education.

Rewards t o educat ion display log-convexity for al l years invest igated tha t is they increased at a r is ing rate for higher levels o f educat ion, par t i cu lar l y i n the upper t a i l of the condi t ional earnings d i s t r i bu t i on (table 18). Th is log-convexity, however, became more pronounced in the 1988-1997 per iod along the condi t ional earnings d i s t r i bu t i on . I n 1988, a person w i t h t e r t ia ry education in the top t en percent of the condi t ional earnings d i s t r ibu t i on earned on average 34 percent more income compared to a person w i t h upper secondary complete education. By 1997, a person w i t h t e r t i a ry educat ion in the top par t of the d i s t r i bu t i on earned on average 67 percent more compared t o a person w i t h upper secondary complete education. F rom 1997 to 2002, th is upward t r end continued only for ind iv iduals i n the top t a i l of the condi t iona l earnings d i s t r ibu t i on , reaching a marg ina l value to education of 73 percent. For those in the middle and lower t a i l of the d i s t r i bu t i on , the marg ina l value to education fell f rom 75 to 67 pe rcen t . 3 2

I n sum, the returns to education increased in Mexico f rom 1988 to 1997, especially for higher levels of education and in the upper t a i l of the condi t iona l earnings d i s t r ibu t i on . However, there was a reversal t o th is t r end after 1997, especially for higher levels of educat ion and in the middle and lower tai ls of the condit ional earnings d i s t r i bu t i on . Th is reflects a s t ruc tura l development, i f expanding relative supplies of school leavers are offsetting the secular tendency of r is ing relative demand for skil ls especially at t e r t i a ry level (see de Ferrant i et ai, 2003).

Wh i l e observed changes in wages are a product of the interact ion

T o t e s t t h e r o b u s t n e s s o f these t r e n d s we e s t i m a t e d t h e f o l l o w i n g m o d e l s :

1) t h e bas i c m o d e l o n l y i n c l u d e d age, s q u a r e d age, a n d gende r . 2 ) T h e s e c o n d

m o d e l was t h e bas i c m o d e l p l u s r e g i o n . 3 ) T h e t h i r d m o d e l was t h e s e cond m o d e l

p l u s s t a t u s i n t h e l a b o r m a r k e t . T h e l a s t m o d e l i n c l u d e d a l l these v a r i a b l e s p l u s

s e c t o r o f a c t i v i t y .


<5>

Ob

00

8 « ?

r H I O CM N

0 0 CM CO .

I O r H LO I O a CM CM I V

^ 0 0 C M 92 © rH ^ «O

CM I O r— rH O 1—« rH CO

I O N N ^ 0 0 CM CO CO

i o rj« LO r— C 5 C M C M

CO 0> "tf CM O r H C M I O

o> oo oo O r H rH

rH rH rH CO rH CM rH ^

CD

"a

t 8 a

R ce

o

S

Ì11

CD a a

P

I O Tt« CO o CM c o r -

0 0 rH oo t* -O CM C M 0 0

. CO rH rH . rH CM O )

I O I O C M *1 r H r H 0 0

I ß CO h rH rH rH CO

LO O C M N -o c o ^ <o

rH CM LO CO rH CO CO CO

CO CO LO rH CM CM CTi

CO rH CM r H rH CM CM 0 5

CM O LO I O rH rH rH I s -

'a

i I c o

CD

P

^ 2

O CD +3 ' u

g CD

TS °

o

a £

'5 §

S o

bO g

i I

8

3

M

O

a> rH P W

cd CD

"g

bo çy

-+-» •a bO •s .S

3 ä

T3 .5 O O

•5 se

CO S

o H N J3 M W

258 E S T U D I O S ECONÓMICOS

between supply and demand changes, i t is possible to derive the implicit shifts on the demand side by mak ing assumptions on the extent t o wh ich different sk i l l categories can subst i tute for each other i n the p roduc t i on process (see de Ferrant i et a/., 2003). The results of such an analysis is presented i n figure 9, this shows that there were large increases i n the relat ive demand for college graduates i n the 1990s, followed by a modest decline after 1997.

F i g u r e 9 The Evolution of the Relative Demand for Workers with Tertiary Education,

Mexico Urban Areas, 1988-2001

l i

11 I I

06000

0.4000

02000

0 *

ooooo mm

mi 1993 m m m m 1997 2000 2001

-0=1-§-0=20=3

1. T h e e s t i m a t i o n o f r e l a t i v e d e m a n d f o l l ows de Fer

r a n t i et al. ( 2003 ) m e t h o d o l o g y a s s u m i n g t h r e e leve ls o f

e l a s t i c i t y o f s u b s t i t u t i o n (s) b e t w e e n u p p e r s e c o n d a r y a n d

t e r t i a r y e d u c a t i o n w o r k e r s ; 2. T h e e v o l u t i o n o f d e m a n d c o n

s iders t h r e e y ea r moving average e s t i m a t i o n s , i n c l u d i n g 12

q u a r t e r s p e r o b s e r v a t i o n . Source : A u t h o r ' s e s t i m a t e s using

t h i r d q u a r t e r o f E N E U f r o m 1988 t o 2 0 0 1 .

7.4. Rate of Returns to Education and Inequality

This subsection further explores the shifts i n labor demand t h a t help


t o expla in inequal ity. The "between" probab i l i t y is the mob i l i t y o f unski l led and ski l led workers between j and k economic sectors . 3 3 B y contrast, the " w i t h i n " mob i l i t y depicts workers who move across sub-sectors or occupations. Table 19 presents the t rans i t i on probabi l i t ies for the respective periods. O n the one hand, the f inancial services sector shows a clear t r end to subst i tute unski l led labor for ski l led labor: the p robab i l i t y o f workers changing t o another economic sector is much higher for unski l led workers (70 percent) t han for ski l led workers (21 percent) . The p r imary sector follows the same t r end only at the end of the 1980s. O n the other hand, non-manufactur ing indus t ry is subs t i tu t ing ski l led for unski l led workers. Final ly , manufac tur ing i n dust ry and t ranspor ta t i on and communicat ions do not have a clearly dominant probab i l i t y o f h i r i n g either ski l led or unski l led workers.

Using shifts b o t h " w i t h i n " and "between" economic sectors, one can explore the effect of these shifts on the relative wage of ski l led and unski l led workers. Table 19 also shows th a t , for a l l periods considered, the "between" probab i l i t y of having a skil led versus an unski l led labor force is substant ia l ly higher; conversely, the " w i t h i n " probab i l i t y of hav ing a ski l led labor force is signif icantly lower t h a n t h a t o f having an unski l led one. Therefore, one might infer t ha t the relat ive wage of unski l led labor relative to ski l led labor increased, derived f rom shifts w i t h i n economic sectors. However, th is effect might have been par t i a l l y offset by the decrease i n relative wages of unski l led labor relative to ski l led labor, derived f rom the shift between economic sectors. Given the rate of returns to education, i t is plausible t o infer t ha t the shifts i n relat ive demand w i t h i n economic sectors dominated the shifts i n relative demand between sectors.

W i t h the goal of p u t t i n g the rate of returns i n perspective, ta ble 20 shows the percentage of earnings differentials for other L a t i n Amer ican countries. Mexico is above the average, second on ly t o Braz i l ( the country w i t h the highest inequal i ty i n L a t i n Amer ica ) . Once more, th is indicates t h a t educational policies must be at the core of any effort aimed at reducing inequal i ty and, by extension, poverty i n Mexico. These findings are s t i l l consistent w i t h de Fer-r a n t i et a/., 2004.

T h e t r a n s i t i o n p r o b a b i l i t i e s d e s c r i b e t h e s h i f t s o f s k i l l e d a n d u n s k i l l e d w o r k

ers w i t h i n a n d across sec tors . T h e t r a n s i t i o n p r o b a b i l i t i e s a r e t h e c o n d i t i o n a l

p r o b a b i l i t y o f f i n d i n g a w o r k e r i n e c o n o m i c s ec t o r k a t t h e e n d o f t h e p e r i o d

g i v e n t h a t t h e w o r k e r b e g a n i n s ec t o r j. T h i s p r o b a b i l i t y g ives us t h e m o b i l i t y o f

less - a n d h i g h - s k i l l e d w o r k e r s b e t w e e n a n d k e c o n o m i c sec tor . S k i l l e d w o r k e r s

a r e those i n d i v i d u a l s w i t h m o r e t h a n 12 years o f s c h o o l i n g .


Oí

¿̂ 0>

©0

o>

OÍ OÍ

Oí

0 0 0 0 01

-tí

tí

05

*° I

o ó o ó c ó o o o S o i o Ó N c ó C O i - H T f C M C N r - l r - I C O C M

co LO

có Tj¡ oí £ CM oó 2

O O O Í O C O C O L O O ' Í L O C M

Tf 00 CM LO CO O CM CM CM LO CM

O Oi CO CM CM CM

~ . CO co co ^ ^ w ^

00 C i

H LO 00 CO CO LO

00 CO O O LO T-H

CM LO LO co co í > -

o i d d i o o o ó c Í H o o

LO 00 CM

CM CO

H N ^ 00 ^ co 53 OÓ £

rJJ r-- co o CO CM LO 00 ^ LO Tí* ^

T-H o r-- O Í LO LO Oí ^t4

CM CO LO LO LO

CO Oi

CM CM

CO LO LO

co co co H ^ CÓ LO CM CO

"Ü« LO CM rH

CO IV O r-i LO LO

N 00 H 00 N ^ LO CM CO

O ° í H o o ó

CO CO IV r-H CM* tV

^ lO

3

62

3

es a

¡3

I

M E X I C O : E V O L U T I O N O F E A R N I N G S I N E Q U A L I T Y 261 Y ch

ange

Bet

wee

n

33.0

15.1

24.4

10.3

42.0

18.7

1996

-199

ri

Sec

tor

Wit

hin

I

15.4

18.6

6.1

18.2

1.4

17.4

No

chan

ge

51.6

66.2

69.6

71.5

56.5

64.0

chan

ge

Between

32.4

24.7

17.0

13.0

43.0

21.8

1992

-199

'c

Sec

tor

Wit

hin

|

15.4

6.1

8.3

23.2

0.5

19.1

No

chan

ge

52.2

69.2

74.7

63.8

56.5

59.1

chan

ge

Bet

wee

n

44.4

21.6

30.2

10.9

50.4

23.8

1988

-198

$

Sec

tor

Wit

hin

1

13.6

1.3

18.9

34.0

4.4

25.1

No

chan

ge

41.9

77.0

50.9

55.1

45.1

51.1

Leve

l of

edu

cati

on

and

sect

or

Com

mer

ce

Fin

an

ser

or

rent

Tra

ns o

r co

mm

un

Soc

ial

serv

ices

Oth

er s

ervi

ces

Wei

ghte

d av

erag

e

.5 T3

i

» co U O O •«

co Q)

% 8 a m ? ?

H 2 S

& c 'S cp H3 § o P ,„

•2 S

« eg O -g

"S) il. a o CP f>

1

o

£ Si


T a b l e 20 Earnings Differentials in Latin America, by Country

(percent)

Level of

education

Latin

America

Mexico Brazil Argentina Peru

Primary complete

50 100 100 35 40

Upper secondary complete

120 170 170 80 80

University complete

200 260 280 160 145

N o t e : Re fe rence g r o u p is n o s c h o o l i n g . Source : I D B (1998 - 1 9 9 9 ) .

8. C o n c l u s i o n s

Even though the levels of educational a t ta inment expanded very rapidly, Mexico has experienced a pronounced increase i n the degree of income inequal i ty over the per iod of analysis. Most of the deter i o ra t ion i n the d i s t r i bu t i on of t o t a l current income happened in the middle to late 1980s (1984-1989). The early 1990s displayed l i t t l e change in t o t a l current income inequal i ty except for a slight t r end t o ward deter iorat ion. The trends in the d i s t r ibu t i on of earnings differ f rom the trends in the d i s t r i bu t i on of current income i n two ways. F i rs t , the gains are not l im i t ed to the richest 10 percent, as those i n the seven-, eight-, and nine-tenths of the d i s t r ibu t i on improved the i r relative earnings over the per iod by almost 2 percentage points. Second, the d i s t r i bu t i on of earnings clearly worsened in the 1990s u n t i l 1996, a l though the inequal i ty associated w i t h t o t a l current income was moderately stable i n the 1990s, displaying an improvement i n 1996. Differences i n the behavior of t o ta l current income and labor earnings inequalities f rom 1994 to 1996 support the idea tha t the poor, who rely the most on labor as a source of income, are the least able to protect themselves dur ing a recession.

Educat iona l inequal i ty is the variable tha t accounts for by far the largest share of earnings inequal i ty i n Mexico, b o t h in terms o f gross and marg inal cont r ibut i on . The cont r ibut ion of education t o earnings inequal i ty i n Mexico is the second highest i n L a t i n Amer ica . Moreover, what seems to be par t i cu lar ly interesting in the Mex ican

MEXICO: EVOLUTION OF EARNINGS INEQUALITY 2 6 3

experience is the fact t h a t the significance of education has been i n creasing over t ime .

The increase i n earnings inequality, however, does not appear t o be the result o f a worsening i n the d i s t r i bu t i on of education, whereas the income profile, wh i ch is related to the returns to schooling, has become much steeper. Th i s means tha t there was a shift i n demand toward high-ski l led labor t h a t was not met by an increase i n supply. Th is probably occurred as a result o f the rap id rate of ski l l -biased technological change, whose transmission to Mexico was fac i l i ta ted by the economy's increased openness.

Re fe r ence s

A b a d i e , A . ( 1 9 9 7 ) . " C h a n g e s i n S p a n i s h L a b o r I n c o m e S t r u c t u r e D u r i n g t h e 1980 ' s : A Q u a n t i l e Reg r e s s i on A p p r o a c h " , Investigaciones Económicas, v o l . 2 1 , n o . 2, p p . 253 -272 .

A l t i m i r , O . a n d S. P i n e r a ( 1 9 8 2 ) . "Anál is is de descomposición de las d e s i g u a l dades de i n g r e s o e n l a A m é r i c a L a t i n a " , El Trimestre Económico, v o l . 4 9 , n o . 196 , p p . 813 -860 .

B a r r o s , R . a n d J . A l m e i d a ( 1 9 9 1 ) . " W a g e I n e q u a l i t y a n d t h e D i s t r i b u t i o n o f E d u c a t i o n : A S t u d y o f t h e E v o l u t i o n o f R e g i o n a l D i f f e rences i n I n e q u a l i t y i n M e t r o p o l i t a n B r a z i l " , Journal of Development Economics, v o l . 36 , p p . 117-143 .

B o u i l l o n Buendía, C , A . L e g o v i n i , a n d N . L u s t i g ( 1 9 9 9 ) . " C a n E d u c a t i o n E x p l a i n I n c o m e I n e q u a l i t y C h a n g e s i n M e x i c o ? " , h t t p : / / w w w . i a d b . o r g / s d s / d o c / 1488eng .pd f . ( 1 9 9 9 ) . " R i s i n g I n e q u a l i t y i n M e x i c o : R e t u r n s t o H o u s e h o l d C h a r a c t e r i s t i c s

a n d t h e C h i a p a s E f f e c t " , h t t p : / / p a p e r s . s s r n . c o m / s o l 3 / p a p e r s . c f m 7 a b s t r a c t _ i d = 1 8 2 1 7 8 .

B o u r g u i g n o n , F . ( 1 9 7 9 ) . " D e c o m p o s a b l e I n c o m e I n e q u a l i t y M e a s u r e s " , Econo-metrica, v o l . 4 7 , n o . 4 , p p . 901 -920 . , M . F o u r n i e r , a n d M . G u r g a n d ( 1 9 9 8 ) . " L a b o r I n c o m e s a n d L a b o r S u p p l y i n t h e C o u r s e o f T a i w a n ' s D e v e l o p m e n t , 1 9 7 9 - 1 9 9 4 " , W o r l d B a n k , ( m i m e o ) .

B u c h i n s k y , M . ( 1 9 9 8 ) . " R e c e n t A d v a n c e s i n Q u a n t i l e Reg ress i on M o d e l s : A P r a c t i c a l G u i d e l i n e f o r E m p i r i c a l R e s e a r c h " , Journal of Human Resources, v o l . 3 3 , n o . 1, p p . 8 8 - 1 2 6 . ( 1 9 9 5 ) . " Q u a n t i l e Reg ress i on , B o x - C o x T r a n s f o r m a t i o n M o d e l a n d t h e U S

W a g e S t r u c t u r e , 1 9 6 3 - 1 9 8 7 " , Journal of Econometrics, v o l . 65 , p p . 109-154 . ( 1 9 9 4 ) . " C h a n g e s i n t h e U S W a g e S t r u c t u r e 1963-1987 : A p p l i c a t i o n o f

Q u a n t i l e R e g r e s s i o n " , Econometrica, v o l . 6 2 , n o . 2, p p . 405 -458 . B u r f i s h e r , M . et al ( 1 9 9 3 ) . Wage Changes in a US-Mexico Free Trade Area:

Migration Versus Stolper-Samuelson Effects, W P 645 , U C , B e r k e l e y . C o w e l l , F . ( 1980 ) . " O n t h e S t r u c t u r e o f A d d i t i v e I n e q u a l i t y M e a s u r e s " , Review

of Economic Studies, v o l . 47 , p p . 5 2 1 - 5 3 1 . C r a g g , M . , a n d M . E p e l b a u m ( 1 9 9 6 ) . " W h y H a s t h e W a g e D i s p e r s i o n G r o w n i n

M e x i c o ? Is i t t h e I n c i d e n c e o f R e f o r m s o r t h e G r o w i n g D e m a n d f o r S k i l l s ? " , Journal of Development Economics, v o l . 5 1 , n o . 47 , p p . 99 -116 .

http://www.iadb.org/sds/doc/

http://papers.ssrn.com/sol3/papers.cfm7abstract_


D e F e r r a n t i , D . et al. ( 2 0 0 4 ) . Inequality in Latin America: Breaking with History?, W o r l d B a n k . et al. ( 2 0 0 3 ) . Closing the Gap in Education and Technology, W o r l d B a n k .

D e l a T o r r e , R. ( 1 9 9 7 ) . Desigualdad, pobreza y polarización social en México, ( m i m e o ) .

E l i a s , J . ( 1 9 9 2 ) . Sources of Growth: A Study of Seven Latin American Economies, I n t e r n a t i o n a l C e n t e r for E c o n o m i c G r o w t h , ( m i m e o ) .

F e e n s t r a , R . a n d H . H a n s o n ( 1996 ) . " G l o b a l i z a t i o n , O u t s o u r c i n g , a n d W a g e I n e q u a l i t y " , AEA Papers and Proceedings, v o l . 86 , n o . 2, p p . 2 4 0 - 2 4 5 .

F i e l d s , G . ( 1 9 9 6 ) . Accounting for Income Inequality and Its Change, C o r n e l l U n i v e r s i t y , ( m i m e o ) .

F i s z b e i n , A . ( 1 9 9 1 ) . An Essay on Labor Markets and Income Inequality in Less Developed Countries, P h . D . , U C , Be rke l e y .

H a n s o n , H . ( 1 9 9 7 ) . " I n c r e a s i n g R e t u r n s , T r a d e , a n d t h e R e g i o n a l S t r u c t u r e o f W a g e s " , Economic Journal, v o l . 107, no . 440 , p p . 113-133. , a n d A . H a r r i s o n ( 1 9 9 5 ) . Trade, Technology, and Wage Inequality, W P S 5 1 1 0 , N B E R , C a m b r i d g e .

Hernández-Laos , E . , N . G a r r o , a n d I . L l a m a s ( 1997 ) . Productividad y mercado de trabajo en México, ( m i m e o ) .

I N E G I . Encuesta nacional de ingreso y gasto de los hogares ( E N I G H ) , C D r o m s , s eve ra l years , M e x i c o . . Encuesta nacional de empleo urbano ( E N E U ) , C D r o m s , s eve ra l y ea r s , M e x i c o . . Encuesta nacional de empleo trimestral, ( E N E T ) , severa l y ea rs , Méx i c o .

I n t e r - A m e r i c a n D e v e l o p m e n t B a n k ( 1998 -1999 ) . Informe: América Latina frente a la desigualdad: progreso económico y social en América Latina.

J o h n s o n , G . ( 1997 ) . " C h a n g e s i n E a r n i n g s I n e q u a l i t y : T h e R o l e o f D e m a n d S h i f t s " , Journal of Economic Perspectives, v o l . 1 1 , no . 2, p p . 4 1 - 5 4 .

J u h n C. , K . M u r p h y , a n d B . P i e r ce ( 1993 ) . " W a g e I n e q u a l i t y a n d t h e R i se i n R e t u r n s t o S k i l l " , Journal of Political Economy, v o l . 3, n o . 3, p p . 4 1 0 - 4 4 8 .

K n i g h t , J . a n d R. S a b o t ( 1 9 8 3 ) . " E d u c a t i o n a l E x p a n s i o n a n d t h e K u z n e t s E f f e c t " , American Economic Review, v o l . 73 , p p . 1132-1136.

K r u e g e r , A . B . (ed. ) ( 1997 ) . " S y m p o s i u m o n W a g e I n e q u a l i t y " , Journal of

Economic Perspectives, v o l . 1 1 , n o . 2, p p . 21-96 . L e i b b r a n d t , M . , C . W o o l a r d , a n d I . W o o l a r d ( 1996 ) . The Contribution of Income

Components to Income Inequality in South Africa, W P 125, W o r l d B a n k . Londoño , J . ( 1 9 9 6 ) . Poverty, Inequality, and Human Capital Development in

Latin America, 1950-2025, W o r l d B a n k , ( m i m e o ) . L ópe z -Ace v edo , G . a n d M . W a l t o n ( 2 0 0 4 ) . Mexico Poverty Assessment, W o r l d

B a n k . L u s t i g , N . a n d M . Székely ( 1996 ) . México: evolución económica, pobreza y

desigualdad, ( m i m e o ) . M e z a , L . ( 1999 ) . " C a m b i o s en l a e s t r u c t u r a s a l a r i a l e n Méx i co e n el p e r i o d o 1988-

1993 y e l a u m e n t o e n e l r e n d i m i e n t o de l a educación s u p e r i o r " , El Trimestre Económico, v o l . 49 , no . 196, p p . 813 -860 .

M o n t e n e g r o , C . ( 1 9 9 9 ) . " T h e s t r u c t u r e o f Wages i n C h i l e 1960 -1996 : A n A p p r o a c h o f Q u a n t i l e Reg r e ss i ons " , Economics Letters, v o l . 60 , n o . 2 , p p . 2 2 9 -235 .


M o r e n o , A . ( 1 9 8 9 ) . " L a distr ibución d e l i n g r e so l a b o r a l u r b a n o e n C o l o m b i a : 1 9 7 6 - 1 9 8 8 " , Desarrollo y Sociedad, v o l . 24 .

M u l l e r , R . ( 1 9 9 8 ) . " P u b l i c - P r i v a t e Sec to r W a g e D i f f e r e n t i a l s i n C a n a d a : E v i d e n c e f r o m Q u a n t i l e Reg r e s s i ons " , Economics Letters, v o l . 60 , n o . 2, p p . 2 2 9 - 2 3 5 .

P a n u c o - L a g u e t t e , H . a n d M . Székely ( 1996 ) . " I n c o m e D i s t r i b u t i o n a n d P o v e r t y i n M e x i c o " , i n V . B u l m e r - T h o m a s ( ed . ) , The New Economic Model in Latin America and Its Impact on Income Distribution and Poverty, N Y , S t . M a r t i n ' s Press, p p . 185-222 .

P s a c h a r o p o u l o s , G . ( 1 9 9 3 ) . Returns to Investment in Education: A Global Update, P o l i c y Resea r ch P a p e r 1067 , W o r l d B a n k . , et al. ( 1 9 9 2 ) . Poverty and Income Distribution in Latin America: The Story of the 1980s, W o r l d B a n k , ( m i m e o ) .

R a m o s , L . a n d C . T r i n d a d e ( 1 9 9 2 ) . " E d u c a g a o e d e s i g u a l d a d e d e s a l a r i o s n o B r a s i l : 1 9 7 7 - 1 9 8 9 " , Perspectivas da Economía Brnsileira, I P E A .

R a m o s , L . ( 1 9 9 0 ) . The Distribution of Earnings in Brazil: 1976-1985, P h . D . d i ss . , U . o f C a l i f o r n i a , Be rke l e y .

Re is , J . a n d R. Paes de B a r r o s ( 1989 ) . Um estudo da evolugáo das diferencas regionais da desigualdade no Brasil, R J , I P E A , ( m i m e o ) .

Reyes , A . ( 1 9 8 8 ) . "Evo luc ión d e l a distribución d e l i ng r e so e n C o l o m b i a " , Desarrollo y Sociedad, v o l . 2 1 .

Rodr í guez , J . ( 1 9 9 1 ) . Desigualdad salarial y educación en Lima metropolitana: 1970 a 1984, M a s t e r ' s thes i s , P U C / R J .

S c h u l t z , T . ( 1 9 8 8 ) . " E d u c a t i o n I n v e s t m e n t s a n d R e t u r n s " , i n H . C h e n e r y a n d T . S r i n i v a s a n (eds . ) , Handbook of Development Economics, v o l . 1, E l s e v i e r , p p . 543 -629 .

S h o r r o c k s , A . ( 1 9 8 4 ) . " I n e q u a l i t y D e c o m p o s i t i o n b y F a c t o r C o m p o n e n t s " , Eco-nometrica, v o l . 5 2 , n o . 6 , p p . 1369-1384 . ( 1 9 8 0 ) . " T h e C lass o f A d d i t i v e l y D e c o m p o s a b l e I n e q u a l i t y M e a s u r e s " , Eco-

nometrica, v o l . 4 8 , n o . 3 , p p . 613 -625 . , a n d J . Fos t e r ( 1 9 8 5 ) . " T r a n s f e r Sens i t i v e I n e q u a l i t y M e a s u r e s " , Review of Economic Studies, v o l . 5 , p p . 105-138.

S h o r r o c k s , A . , a n d D . M o o k h e r j e e ( 1982 ) . " A D e c o m p o s i t i o n A n a l y s i s o f t h e TVend i n U K I n c o m e I n e q u a l i t y " , Economic Journal, v o l . 92 , p p . 8 8 6 - 9 0 2 .

S h u l t z , P. a n d E . M w a b u ( 1 9 9 6 ) . " S y m p o s i u m Issue o n H o w I n t e r n a t i o n a l E x c h a n g e , T e c h n o l o g y , a n d I n s t i t u t i o n s A f f e c t W o r k e r s " , The World Bank Economic Review, v o l . 1 1 , n o . 1.

S y r q u i n , M . (ed. ) ( 1 9 9 7 ) . A Symposium Issue on How International Exchange, Technology, and Institutions Affect Workers, The World Bank Economic Review, v o l . 1 1 , n o . 1, sp e c i a l issue.

Székely, M . ( 1995 ) . " A s p e c t o s d e l a d e s i g u a l d a d e n M é x i c o " , E l Trimestre Económico, v o l . 6 2 , n o . 246 , p p . 201 -243 .

T a n , H . , a n d G . B a t r a ( 2 0 0 0 ) . " T e c h n o l o g y a n d F i r m S i ze -Wage D i f f e r e n t i a l s i n C o l o m b i a , M e x i c o , a n d T a i w a n ( C h i n a ) " , The World Bank Economic Review, v o l . 1 1 , n o . 1 , p p . 59 -83 .

V i e i r a , M . ( 1 9 9 8 ) . A relacao entre educacao e desigualdade no Brasil, M a s t e r ' s d i ss , E P G E , R J .

W o r l d B a n k ( 1996 ) . World Development Report 1996, W a s h i n g t o n , O x f o r d U n i v e r s i t y Press.


A n n e x 1. D a t a Sources

The Na t i ona l Household Income and Expendi ture Survey, E N I G H , and the Na t i ona l U rban Employment Survey, E N E U , were used i n th is study.

ENIGH

The Nat iona l Household Income and Expenditures Survey is collected by the Instituto Nacional de Estadística, Geografía e Informática, I N -E G I . Th i s survey is available for 1984, 1989, 1992, 1994, 1996, 1998 . 3 4

Each survey is representative at the nat iona l level, for urban and r u ra l areas. For 1996, the E N I G H is also representative for the states o f México, Campeche, Coahuila, Guanajuato, Hidalgo, Jalisco, Oaxaca and Tabasco.

For each year the survey design was strat i f ied, mult istage and clustered. The final sampl ing u n i t is the household and a l l the members w i t h i n the household were interviewed. I n each stage, the select ion probab i l i t y was propor t i ona l to the size of the sampl ing u n i t . Thus, i t is necessary to use we i gh ts 3 5 i n order to get suitable estimators. The table below shows the sample size for each year.

T a b l e 1.A1 Sample Size by Year

Year Number of households Number of persons

1984 4,735 23,756

1989 11,531 56,727

1992 10,530 50,378

1994 12,815 59,835

1996 14,042 64,359

The available in fo rmat ion can be grouped into three categories:

T h e s a m p l e i n a g i v e n y ea r is i n d e p e n d e n t f r o m a n o t h e r .

3 5 T h e w e i g h t s s h o u l d b e c a l c u l a t e d a c c o r d i n g t o t h e s u r v e y d e s i g n a n d c o r r e

s p o n d t o t h e inve rse o f t h e p r o b a b i l i t y i n c l u s i o n .


• Income and consumpt ion: the survey has monetary, no mone tary and f inancial items.

• I nd i v i dua l characteristics: social and demographic, i.e., age, schooling attendance, level o f schooling, pos i t ion at work, etc.

• Household characteristics.

Category Selection

For the purpose of the analysis, the indiv iduals i n the sample were classified according t o the i r educational level, pos i t ion i n occupat ion, sector o f ac t i v i t y and geographical region i n the fol lowing categories:

Educat iona l level

a) P r imary incomplete: no education and p r imary incomplete (one to five years of p r imary ) ,

b) P r imary complete: p r imary complete and secondary incomplete (one or two years),

c) Secondary complete: secondary complete and preparatory incomplete (one or two years),

d) Preparatory complete: preparatory complete and univers i ty i n complete,

e) Univers i ty complete: universi ty complete ( w i t h degree) and postgraduate studies.

Posi t ion i n occupat ion

a) Worker or employee, b) Employer, c) Self employed.

Sector of ac t i v i t y

a) Agr i cu l ture , b) Manufac tur ing , c) Construct ion , d) Commerce, e) Services,


f) Other (ut i l i t i es , ex t ract ion, t ransports , financial services, communicat ions, etc) .

Geographical regions

a) N o r t h : Baja Cal i fornia, Ba ja Cal i fornia Sur, Coahuila, Ch ihuahua, Durango, Nuevo León, Sinaloa, Sonora, Tamaul ipas, and Zacatecas,

b) Center: Aguascalientes, Col ima, Guanajuato, Hidalgo, Jalisco, México, Michoacán, Morelos, Nayar i t , Puebla, Querétaro, San Luis Potosí, and Tlaxcala,

c) South: Campeche, Chiapas, Guerrero, Oaxaca, Qu in tana Roo, Tabasco, Veracruz, and Yucatán,

d) D i s t r i t o Federal.

Group Selection

The labor force was l im i t ed to indiv iduals who are:

a) work ing as employee, employer or self employed , 3 6

b) between 12 and 65 years o ld, c) l i v ing i n urban areas, d) work ing 20 hours or more per week, e) w i t h posit ive income, f) hav ing the a t t r ibutes of interest defined.

The number of persons in the survey tha t are part of the labor force is shown in the next table.

T a b l e 2 . A 1 Sample Size for the Labor Force

Year Number of persons % of the total sample

1984 3,892 16.4

1989 10,401 18.3

1992 8,752 17.4

1994 10,982 18.4

1996 12,996 20.2

T h e r e s p e c t i v e ca tegor i es : w o r k e r s w i t h o u t p a y m e n t a n d c o o p e r a t i v e m e m

be r s we r e e x c l u d e d because o f t h e s a m p l e size.


Accord ing t o the groups mentioned we have tha t ,

T a b l e 3 .A1 Sample Size by Variable and Year

Year 1984 1989 1992 1994 1996

Education level

P r i m a r y i n c o m p l e t e 1,246 1,951 1,879 2,387 2,736

P r i m a r y c o m p l e t e 1,299 3,006 2,501 2,975 3,411

S e c o n d a r y c o m p l e t e 803 2,875 2,489 3,014 3,734

P r e p a r a t o r y c o m p l e t e 389 1,614 1,168 1,617 1,915

U n i v e r s i t y c o m p l e t e 245 955 715 989 1,200

Position in occupation

E m p l o y e e 3,175 8,604 7,188 8,843 10,207

E m p l o y e r 126 311 393 450 610

S e l f e m p l o y e d 681 1,486 1,171 1,689 2,179

T o t a l 3,982 10 ,401 8 ,752 10,982 12 ,996

ENEU

This s tudy uses in fo rmat ion f rom the Nat iona l Urban Employment Survey, E N E U , wh ich is also a micro-level data set collected by I N -

E G I and contains quarter ly wage and employment data over the past 10 years (1987-1997). Accord ing to I N E G I ' s methodology document on the E N E U , the data are representative of the 41 largest u rban areas i n Mexico, covering 61 percent of the popula t ion i n u rban areas w i t h at least 2,500 inhabi tants and 92 percent of the popu la t i on l i v ing i n met ropo l i t an areas w i t h 100,000 or more inhabi tants . I n 1985 the E N E U included 16 urban areas: Mexico City, Guadalajara, Monterrey, Puebla, León, San Luis Potosí, Tampico, Torreón, Ch i huahua, Orizaba, Veracruz, Mérida, C iudad Juárez, T i juana , Nuevo Laredo, and Matamoros , covering 60 percent of the urban popu la t i on for t ha t year. I n 1992, 18 more urban areas were inc luded i n the survey: Aguascalientes, Acapulco, Campeche, Coatzacoalcos, Cuer-navaca, Culiacán, Durango, Hermosil lo, Morel ia , Oaxaca, Salt i l lo , Tepic, Toluca, T u x t l a Gutiérrez, Vi l lahermosa, Zacatecas, Co l ima, and Manzani l lo . I n 1993 and 1994 Monclova, Querétaro, Celaya, I r a -puato , and T laxca la entered the E N E U . Final ly , Cancún and L a Paz


j o ined the survey i n 1996. Accord ing to I N E G I , the E N E U always has covered about 60 percent of the nat iona l urban popula t ion .

The data are f rom household surveys, which ful ly describe fami ly composi t ion, human capi ta l acquisit ion, and experience i n the labor market (the variables contain in fo rmat ion about social household characteristics, ac t i v i t y condi t ion , pos i t ion in occupat ion, unemployment, ma in occupat ion, hours worked, earnings, benefits, secondary occupat ion, and search for another j ob ) . As w i t h the E N I G H , the samp l ing design was strat i f ied in several stages (where the final selection u n i t was the household) and w i t h propor t i ona l probab i l i t y t o s i ze . 3 7

This stat is t ica l construct ion allowed us to make comparisons among different years. Moreover, th is survey is s t ructured to generate a panel da ta set tha t conforms w i t h a ro ta tor or r o ta t ing panel (a fifth o f the t o t a l sample goes out and a new one comes i n every quar ter ) . Hence, the panel data follow the same household throughout five quarters .

Category Selection

The indiv iduals i n the sample were classified according to the i r educat ional level, age, sector of act iv i ty , pos i t ion in occupat ion, hours worked, and geographic region in the fol lowing categories:

Educat ional level

a) P r imary incomplete: no education and p r imary incomplete (one to five years of p r imary ) ,

b) P r imary complete: p r imary complete and secondary incomplete (one or two years),

c) Secondary complete: secondary complete and preparatory incomplete (one or two years),

d) Preparatory complete: preparatory complete and univers i ty i n complete,

e) Univers i ty complete: university complete ( w i t h degree) and postgraduate studies.

Age

a) 12 to 25 years o ld,

3 7 Fo r t h i s i t was necessary t o use w e i g h t s o r e x p a n s i o n f a c t o r s .


b) 26 t o 34 years o ld , c) 35 t o 49 years o ld , d) 50 t o 65 years o ld .

Sector o f ac t i v i t y

a) P r i m a r y sector (includes agr icul ture, forestry, f ishing, and m i n ing ) ,

b) Manu fac tur ing industry, c) Non-manufactur ing indust ry (includes construct ion and u t i l i t i e s ) , d) Commerce, e) Finance services and rent , f) T ranspor ta t i on and communicat ion, g) Social services ( tour ism, education, heal th, publ ic admin i s t ra t i on ,

embassy), h) Other services.

Labor market status

a) Employer, b) Self-employed, c) In fo rma l salaried: people who work in an enterprise w i t h 15 or

fewer workers and do not receive social security ( IMSS, ISSSTE, pr ivate , and so f o r th ) ,

d) Formal salaried: people who work in an enterprise w i t h 16 or more workers or receive social security ( IMSS, ISSSTE, pr ivate , and so f o r th ) ,

e) Contract .

Hours worked

a) 20 t o 39 hours a week, b) 40 t o 48 hours a week, c) A t least 49 hours a week.

Geographic regions

a) N o r t h : Ba ja Cal i fornia, Ba ja Cal i fornia Sur, Coahuila, Chihuahua, Durango, Nuevo Leon, Sinaloa, Sonora, Tamaul ipas, and Zacatecas,


b) Center: Aguascalientes, Col ima, Guanajuato, Hidalgo, Jalisco, México, Michoacán, Morelos, Nayar i t , Puebla, Querétaro, San Luis Potosí, and Tlaxcala,

c) South: Campeche, Chiapas, Guerrero, Oaxaca, Qu in tana Roo, Tabasco, Veracruz, and Yucatán,

d) D i s t r i t o Federal.

Group Selection

Analogous to the E N I G H , the sample consists of indiv iduals who are:

a) Between 16 and 65 years o ld,

b) L i v i n g in u rban areas (localities w i t h at least 2,500 inhab i tants ) , c) W o rk in g regular ly (non-seasonal workers), d) Wo rk ing 20 hours or more a week, e) Hav ing posit ive earn ings , 3 8

f) Hav ing the a t t r ibutes of interest defined.

T a b l e 4 . A 1 Sample Size, 1988-1997

(number of persons)

Year Labor force Total

1988 124,322 45,870

1989 125,820 47,630

1990 127,387 48,109

1991 126,262 48,080

1992 235,696 91,279

1993 239,394 90,860

1994 246,906 102,105

1995 252,563 100,838

1996 262,478 108,159

1997 272,356 116,559

I n t h i s s u r v e y a n a d d i t i o n a l a d j u s t m e n t h a d t o be m a d e : i f t h e w o r k e r g o t

a b o n u s a t t h e e n d o f t h e yea r (aguinaldo), t h e n t h e wage was e x p a n d e d ( w e

a s s u m e d t h a t t h i s b e n e f i t was e q u i v a l e n t t o 30 days o f wages a y e a r ) .


A n n e x 2. Me thodo l og i c a l No t e

Gini Index

The G i n i index is defined by

G / = 2 c o v [ r , F ( F ) ] ( 1 )

where Y is the d i s t r i bu t i on of per capita income Y = ( y i , y n ) , where yi is the per capi ta income of ind i v idua l i , I = l , . . . , n ; j / i s the mean per capi ta income; F(Y) is the cumulat ive d i s t r i bu t i on o f t o t a l per cap i ta income i n the sample ( that is, F(Y) = [/(y 1 ) , . . . , / ( y n ) ] » where f(yi) is equal t o the rank of yi d iv ided by the number of observations [ n ] ) . 3 9

Equat i on 1 can be r ewr i t t en and expanded into an expression for the G i n i coefficient t ha t captures the " con t r ibu t i on to inequa l i t y " o f each o f the K components of income (see Le ibbrandt and others, 1996).

K

GI = Y . R k G k S k (2)

where Sk is the share of source k of income i n t o t a l group income ( that is, Sk = Vk/v), Gk is the G i n i coefficient measuring the inequal i ty i n the d i s t r i bu t i on of income component k w i t h i n the group, and Rk is the G i n i coefficient of income f rom source k w i t h t o t a l i ncome . 4 0

The larger is the product of these three components, the greater is the con t r i bu t i on of income f rom source k to t o t a l inequality.

Theil T Index41

This index is calculated as fo l lows: 4 2

3 9 B o t h t h e c o v a r i a n c e a n d c u m u l a t i v e d i s t r i b u t i o n a r e c o m p u t e d u s i n g t h e

h o u s e h o l d w e i g h t s .

*> Rk i s d e f i n e d as: Rk = ffijl 4 1 T h e T h e i l T i n d e x i s s ens i t i v e t o changes a t t h e b o t t o m a n d t h e t o p t a i l o f

t h e d i s t r i b u t i o n .

4 2 T h e m a t h e m a t i c a l n o t a t i o n s i n t h i s s e c t i o n a n d t h e n e x t f o l l o w s R a m o s ( 1 9 9 0 ) .


where Yj is the income of the zth ind i v idua l , Y is average income, and n is popu la t i on size.

Stat ic decomposit ion of the The i l index. I f the popu la t i on is d iv ided into G groups w i t h ng observations each, i t is then possible to w r i t e equation 3 as:

g=l 2=1 v / v /

where Y{g is the income of the zth ind i v idua l of the gth popu la t i on subgroup.

I f we now define (3g = ng/n and Zg = Yg/fc where Yg is the average income of the gth group and A; is a reference income, i t is possible t o show, after some algebraic man ipu la t i on , tha t T can be expressed as:

G G

\ / g-l ^ ' g=l

where k — ^2/3gZg and Tg is the The i l index for the gth group. The first two terms on the r i ght -hand side of equat ion 5 corre

spond to the between group inequality, and the t h i r d corresponds one to the w i t h i n group inequality.

Choosing the mean income as the reference income - t h a t is, Zg =

ag — ^s/y- expression 5 simplifies to :

G G

T = z J 2 a 9 f 3 9 l n a 9 + J^a9 09 ? 9 ( 6 ) 9=1 9=1

The first t e rm in equation 6 is the between group inequality, and the second t e rm is the w i t h i n group inequality.

Dynamic decomposit ion analysis. By to ta l l y di f ferent iat ing equat i on 6, we have:

^ dT „ A dT , ^ dT , dT=y - — d / ? f l + > ~ dag+y -—dTq (7)

g=l 9 g=i y g=l J


The first t e r m on the r i gh t -hand side is the popu la t i on a l locat ion effect (changes i n T caused exclusively by popula t ion shi f ts ) . The second t e r m is the income effect (changes i n T induced exclusively by changes i n standardized mean incomes), and the t h i r d one is the in terna l effect (changes i n T caused by changes i n interna l dispersion) .

I t can be shown tha t :

Replacing equations 8, 9, and 10 in to equation 7 and s impl i fy ing , we ob ta in

G

d T = z J 2 a 9 Qn<*g + Tg-T-l)dl3g

G G

+ Y^09(lna9 + T9-T)dag-rJ2(<*909)dTg (11) 9=1 9=1

The three terms on the r ight -hand side of equation 11 correspond to the al locat ion, income, and interna l effects, respectively.

For es t imat ion purposes, equation 11 must be approx imated. The convention used i n the empir ica l exercises is to evaluate the expression at the middle points.


Level, Inequality, and the Indicator of Steepness of the Income Profiles in Educational Level

Ramos (1990) uses three synthetic measures for the indicators mt

(average schooling), i t (schooling inequal i ty ) , and st ( income prof i le ) , based d irect ly on the de f in i t ion of the The i l index.

The calculations of the pr inc ipa l parameters ag, (3g, and Tg (5) could determine the changes in the d i s t r i bu t i on by level of educat ion (g groups in this category). These parameters allow us to analyze the t r end in educat ional income differentials, the d i s t r i b u t i o n of the popula t ion in each educational level, and the inequal i ty among t h e m .

Three synthet ic measures are used t o summarize the changes related to education: mt is the average level of schooling for the year t, it is the degree of inequal i ty in the d i s t r ibu t i on of educat ion for year t, st is the var ia t ion i n the income ratios associated w i t h educat ion for year t.

These measures can be calculated as follows:

9

It =

£ > X l o g ( a p

- l o g l^Tafö

£ a « / ? ! l o g ( a ' )

9H9

where a * is the standardized income of educational category g for the reference year, 0* is the fract ion of the labor force i n the gth educational category i n year t, and ¡3* is the value (3g i n the reference year, st can be understood as an indicator of the relative steepness of the income profiles related t o education. I f one fixes the f ract ion o f the labor force i n each educational group, i t follows tha t the steeper is the income profile, the larger is the between group inequality, it corresponds to the The i l T index tha t would prevail i n a popu la t i on w i t h no inequal i ty w i t h i n the educational groups and where the group incomes are propor t i ona l to the group average incomes i n the base year.


Methods of Decomposition Analysis

The decomposit ion analysis is a useful t oo l for assessing the impact o f certain factors on the evo lut ion of income d i s t r i bu t i on . I n general, the different decomposit ion methods follow two definit ions (Fields 1996):

• Inequal i ty i n the popu la t i on can be decomposed in to different elements such t h a t the sum of the parts is equal t o t o t a l inequal i ty .

• Inequal i ty i n the popu la t i on can be decomposed as a weighted sum of inequal i ty w i t h i n and between groups.

Fields (1996) and Bourguinon, Fournier, and Gurgand (1998) employ the f irst method of decomposit ion. Fields decomposes t o t a l popu la t i on inequal i ty i n a sum of different variables or elements, each being the explanatory variable i n the earnings funct ion. Th i s helps us t o answer two questions: how much income inequal i ty is explained by each right-hand-side variable i n a given po int in t ime? A n d how much o f the difference i n inequal i ty between groups or dates does each variable explain? Th is technique assumes tha t we know the correct model specification".

Formally, th is methodology can be w r i t t e n as Y = Z ' B , where Y = l n ( W ) is the vector of the l ogar i thm incomes, Z = ( 1 , X i , X j , e) is the m a t r i x of explanatory variables, and error t e r m B = (a,/?i,

1 ) ' is the regression coefficient vector. Then ,

_ cavQgj-Zj . r ) ^(Z^COTTJZ^Y) S j ~ a 2 ( F ) - a(Y) { L Z )

where sj is the relative factor weight, and Yl sj = ^ 2 (de terminat ion coefficient).

The con t r ibu t i on of factor j to the change i n the inequal i ty measure /(.) between t ime 0 and t ime 1 is

A i [/(•) ] =

where sj is the relative weighted factor for year 0, and Sj is the relat ive weighted factor for year 1.

Fields also proposes a change breakdown i n the factor's cont r i bu t i on in to the fol lowing: the change i n the coefficient of the factor or variable, the change of the standard deviat ion of the variable, and the change i n the correlat ion between the variable and earnings.


Bourguignon, Fournier, and Gurgand (1998) decompose the effects of changes i n an entire d i s t r i bu t i on rather than on a scalar summary stat ist ic . Th i s methodology was or ig inal ly proposed by Barros and Reis (1991) and Juhn , Murphy , and Pierce (1993) and later generalized by Bourguignon, Fournier, and Gurgand.

The methodology, by means of micro simulations, decomposes the changes in income d i s t r i bu t i on into different effects. Bou i l l on , Legovini , and Lus t i g (1998) use this technique in the case of Mexico to decompose the change into the r e tu rn effect, the popu la t i on effect, the error t e rm effect, and the residual effect.

Th i s can be expressed as follows: let D(y) = D(@,X,e) be the income d i s t r i bu t i on measure and define y = Xf3 + e, where X is the set of demographic variables, ¡3 is the set of prices, and e is the error terms.

I f y is the income i n year 0 and y' is the income in year 1, the change i n income d i s t r i bu t i on can be expressed as:

A = D(y) - D(y) = P{x\e) +X(0,s)

+e(0,x') + [e(p,X) - e{p\x)] (13)

where 0(x\e) = £>(/?', X\ e) - D(j3X\e') is the r e tu rn effect, X(0,e) = D{0, X' ,e)-D((3,X,e) is the popula t ion effect, e(p\ X) = D(0\x\e) - D{p ,X ,e) is the error t e rm effect, and [e{/3,x') -e(/3',X )] is the residual effect.

The analysis makes the fol lowing assumptions:

Income is correct ly expressed as a linear combinat ion.

I n order to compute D(/3,X , e), the residuals i n the second year are rescaled to the second year of reference by a constant such t h a t the variance in t ha t year is the same as the variance of the residuals in the first year. Th is , i n t u r n , implies tha t the d i s t r i bu t i on of e and e j us t differs by the variance.

Bou i l l on , Legovini, and Lus t i g (1998, 1999) use th is methodo l ogy. I n these documents, a l though the assumption of unchangeable dispersions of the regression error terms does not signif icantly restr ict the model 's results, using the variance instead of a proper inequal i ty index is questionable. Th is means tha t one measure is used for the w i t h i n inequality, and another is used for the between inequality.

M igue l Szekely (1995), i n order to explain the inequal i ty changes between two points i n t ime, applies the fol lowing formula:


w - y 1 < I 4 »

where n is the p a r t i t i o n or div is ion of the populat ion, TB(ir) is the The i l index between groups i n year 1, TB{^) is the The i l index between groups i n year 0, CB(K) is the percentage of the change i n inequal i ty explained by the variables i n 7r, T is the The i l index i n year 1, and T is the The i l index in year 0.

Th i s methodology does not allow us to separate the income f rom the a l locat ion effect.


S o o

<3

o

2?

ti

fc O o

'S 3

(3

me

neo

"ti Gin

o o

3

O C5 CO CN ^ CO

O rH CN

^ CN 0 0 CO rH LO CN rH O o

o

O o CM

co o o

o LO 1>-

0 0 '

oo <y¡ CO rH ^ CO O O

O

CN O

0 0 0 0 b -<M «O CO Tf LO O ÍO h CO o o o

a .s O tí O M tí d

. t í cu bJO

¡3 .9 s i : o o ¡S CU

cu . t í tí ^

S s 2 o

rH CO ^ 0 0

0 0 co 0 0 rH 0 0

CN rH O o

CN O CO i—i LO ^

O o

LO cr> co o CO rH

CO O O

CO CN

0 0 o CN

0 0

0 0

o

CO LO CN CN

0 0 LO rH CN 0 0 ^ rH rH LO CO 0 0 CO

tí

s •§ O tí O ^ tí

• tí CD

>> bJO a .s 4J 'TÍ s ?

o


5 s

* 1

ft, "

3 8 1

3

.8 S

s !

3

CM r H c o

io CM

0 0 rH

CM CO CO CM rH r H Ö Ö Ö

O C O CTJ O ) H "tf

CO ^ 0 0

ö ö ö

rH CO r H r H LO 0 0 CO

CM CM Ö O Ö

O O O) Tf CM 1> H O

co oo co o © ö

a £ O G O ^ .3 (Ü

S .s s i O ü

*H O

co

CM O i ö io l O CM

CO CM s S 1

© oo oo r— co c o CM CM r H r H O Ö Ö

o o CO co I O

CM CM

CM I

c o i o CM CM CO rH 0 0 CO CO CO CO Ö O Ö

O o o o o OO

co

0 0 CM O co tO r-1 C i LO I O

CM CM Ö Ö Ö

O

o

l O CM

0 00 r H 01 LO CO Ci © CO i > co Ö Ö Ö

0 0 CO co LO

LO h o

0 bO

a .a .S g

s .s Ö £ O CJ

CO ü

Ö

s gl o

"3 o I

bjO


-CS

2

S 1

a, "

1 ß

5

C* e •<s> -se C ö

"<s> 9

6 -5

CO CM

o CM

LO CM co 0 0

CM

CM

CO CM O CO

0) bO

a .a O tí Ö cö .tí

tí ^ O o

tí o a o

-o tí

'£ ¡ 2

o Ö

CM C5)

o 6 tí e o

o O e

o

tí

«0 «3 o tí o

CJ cu «0

C¿ B

s rH 0.

04

o o CJ

e •S

o O co a. o rH o

cu ° 0

QO O Ö

c¿ O Í O *H

o OO LO oo <3 o co O o L O

rH o O Ö Ö

•2 o o C3 ü i-i 5> tí CP CO

tí tio "tí

.2 tio

S g CÖ CU O

Yea

r UC

cu

on

Stat

u

Yea

r

Ed Oc

Ec Stat

u

Sp


CO o o -to

Ö o p rH rH CO

©

CJ CO

S <J o o S

o p

S rH Ö rH O 1 © 1

o CJ

•S o CN 00 o 05

Cl. o O p p O O

CU\

rH Ö Ö rH Ö © cj © >

O

e

o CO CO 00 o O CT» rH CN o CO p o co O p p rH p rH Ö ö Ö ò rH O Ö O O

C3 ' £ to

r

ho

:tor

ho

C3 s S-i Ì6 M

S> ti V CO

CO Ö 8 (0 •X3 C O cj Ö .2 CJ e <3

.2 Vu

ati

s cö g .2

IUI CO

«3 g

ca O a co cô

o UD

no CO

cl

Yea

Edu

Occ

Eco

;

Sta

t

Spei

Edu

Occ

Eco

Sta

i

Spe

1 ï

S a 5 z cB H

6 § co "


o q CN

co CO LO

LO C5 LO

CO 51

C3

9

•«s> 5 •§

C T

^ o

. 8 ^

0 « t

2

o ^

e <=>

o

e

E 1 ^

1-3

i « 3

o CJ

cá

o

5

SO

o

ce

ce tí

S I I ^ ü

ex

w

1 « o •s

«3

1 «

' 1

S

É?1

oí

e (-i CP

O

CN CN CN

ce o co w>l co

ce

I tí

ce l e

CP

tí , ce

PU

•3 cu

¡I ce

tí CP

CP

CO

ce

p CN

tí , tí

o fcuO tí

-4-3 'tí

ce CO

•g -+-> tí

¡3

CO Tt5

-O ce

s I N

PQ Q

co O» Oí

tí ce

MEXICO: EVOLUTIO ON F EARNINGS INEQUALITY AN...

Documents

Transcript of MEXICO: EVOLUTIO ON F EARNINGS INEQUALITY AN...