Vulnerable Groups and the Labor Market in Thailand Impact of the Asian Financial Crisis in the Light of Thailand’s Growth Process

Dipak Mazumdar and Hyun Hwa Son

University of Toronto October 2001

1. The Growth Process and Inequality The model of growth in agrarian economies proposed by Arthur Lewis (1954) and its subsequent elaborations stressed the transfer of labor from a low-productivity traditional sector to a high productivity ‘capitalist’ sector as an important element in the growth of per capita GDP. Labor in disguised unemployment in traditional agriculture, with very low marginal product, shares in the family pot, dragging down the per capita income in the rural economy. The transfer of a portion of such labor to the capitalist sector with a higher marginal product (and wage) of labor increases per capita income both in the developing sector as well as the traditional one. Kuznets (1954, 1963), however, noted that this process might increase inequality. With an elastic supply of labor wage rates do not increase much in the developing sector, and the rate of profit o that the share of labor income in the GDP is likely to fall. Furthermore, technical progress favors the modern sectors in the initial stages of development, and since the new techniques are more capital-intensive, the share of profits most likely increases over time. Hence the Kuznets hypothesis of the inverted U-shaped form of the relationship of inequality to growth in income. Only at a later stage of economic development, when the disappearance of disguised unemployment puts an upward pressure on wages, and technical progress becomes more widespread, does inequality fall. Poverty and Inequality It has not been sufficiently stressed in the literature that this view of the growth process, fuelled by transfer of labor, carries with it the implication that the rise in inequality, with increase in per capita income, goes hand in hand with a decrease in the incidence of poverty. Much of course depends on the pattern of rural-to-urban migration, and in particular if the migrants come from the lower rungs of the income ladder. There is much evidence in the migration literature that the lowest income earners, in fact, do not account for the bulk of migration. But even if this were generally true, the ultra-poor in the rural sector might not benefit directly from the transfer of labor, but the proportion below a reasonably defined poverty-line is very likely to fall as the disguisedly employed find jobs in the high wage modern sector.

1

2. The Thai Growth Process Between 1960 and 1980 Thailand achieved an average growth rate of around 7 per cent per annum which is high by most standards. There was a slowdown after the second oil crisis. Although the growth rate never plummeted (approximating 5 per cent in the first half of the eighties), problems in the form of high external debt ratio, budget and current account deficits began to emerge. The Thai government, with the assistance of the World Bank and the IMF, undertook a series of ‘structural adjustment’ reforms. “Partly because of the exchange rate adjustments, partly because of the sharp decline in oil prices after in 1986, and partly because of the transition of the Asian newly industrializing countries to more skilled and technologically intensive products, the growth of the Thai economy accelerated significantly after 1986. The growth was mainly driven on by a sharp increase in manufactured exports”. (Sussangkarn, p.589). This ‘take-off’ lasted until the crisis of the 1997-8 period, and will be the subject of the poverty reduction process in the next section. Several pecularities about the growth process in Thailand should be emphasized. (i)

First, although the growth process was fuelled by rapid growth of the ‘capitalist’ or modern sector, unlike the Lewis model the transfer of labor to the modern sector was limited. In the 1980-90 period the percentage of the labor force in agriculture fell from 71 to 64 per cent, while in the same period the same percentage was nearly halved in Korea from a much lower level—from 37 to 18 per cent. The rate of urbanization in Thailand was correspondingly slow.

(ii)

Secondly, the rapid growth in Thailand was accompanied by rising inequality. A World Bank study reports that the Gini index based on expenditure per capita increased by ten percentage points (from 36.4 to 46.2) over the 1975-92 period. No other country in the Asia-Pacific region registered an increase in the inequality index of this magnitude. At the end of the period the Gini index for Thailand was highest in the region (Ahuja et al, p.27). The index of inequality seems to have stabilized in the nineties, but the Asian crisis pushed it up a little more in 1997-99 (Son, Chapter V). The rise in inequality with economic growth was not associated with the Kuznets type of process of transfer of labor from the unproductive to the productive sector. Rather it was the slow release of labor from the less productive segments which seems to have been responsible for this extraordinary phenomenon. The most important of this set of factors is the retention of labor in agriculture and in the

2

poorer regions of the North and the North-East. The share of employment in agriculture fell at a significantly slower rate than the share of this sector in GDP. The ratio of income generated per worker in nonagriculture to that in agriculture increased from 5.88 in 1975 to 11.0 in 1992 (Sussangkarn, p.591 and World Development Indicators). In 1991 the relative income per worker in agriculture (relative, that is to say, to the economy-wide average) was the lowest in the region (table 1). Table 1. Relative income per worker in three broad sectors 1991

East Asian Countries

Agriculture

Manufacturing

Services

Korea

0.45

1.21

0.84

Thailand

0.21

2.38

1.66

Malaysia

0.61

1.46

0.73

Indonesia

0.38

1.87

1.17

Philippines

0.50

2.43

0.98

Source: World Development Report 1992. Associated with this low productivity agriculture was the very wide regional difference in income within the Thai economy. The development of the modern sector was concentrated in the Bangkok Metropolitan Area (BMA) and the regions surrounding it. This created a wide inter-regional difference in income levels—particularly between the North and North-East regions on the one hand and the Central and BMA areas on the other. (iii) A third important feature of the growth process in Thailand was that in spite of the high degree of inequality in Thailand, and increasing at least until the beginning of the nineties, the incidence of poverty has been on a significant declining trend for much of the time. The calculation of the official poverty line was refined and changed radically in 1988, making it impossible to construct a continuous long-term series. But looking at different periods before and after this date it is clear that the headcount ratio of poverty fell drastically between 1968/9 and 1981 (from 39.0 to 23.0 according to one

3

calculation), increased somewhat in the first half of the eighties (from 23.0 to 29.5 pre cent) as Thailand experienced adjustment problems as well as a serous drought in 1986 (Sussangkarn, p. 594), but resumed its downward trend until the onset of the crisis.

3. The Impact of a Downturn in a Dual Economy The growth process in a dual economy like that of Thailand is generally fuelled by the export of manufactured goods. Much of this activity is located in the formal sector of the urban economy, as is the ancillary financial and commercial services. Furthermore, these manufactured firms and services are often concentrated in and around the region of the metropolitan city or cities—in the Bangkok Metropolitan Area (BMA) in the case of Thailand. The impact of the crisis would thus be felt disproportionately in the formal sector, and in the metropolitan area. With the limited transfer of labor from the traditional to the modern sector, it might have been expected that the impact of the crisis would reduce income of the relatively better off in the formal sector, in the metropolitan area and among the more skilled. Although some of these shocks might indeed be transmitted to workers outside the formal sector, as workers losing their jobs enter the informal sectors, the expectation would be that these effects would not have a serious impact in the rural economy or small towns. Thus household incomes in the lower half of the distribution might not be seriously depressed. But as we shall see the outcome in Thailand in the post-crisis period was exactly the opposite, with the incidence of poverty increasing in the more traditional and low-income parts of the labor market. A detailed examination of how the shock was transmitted to the poorer segments of the economy is outside the scope of this paper. But it is important to draw attention to at least two important issues which made this kind of impact possible. First is the nature of rural-to-urban migration. If the bulk of such migrants were of the permanent type during the growth process, most of them would not return to the rural areas since they had severed their links with the sector of origin. But in an economy like that of Thailand, we shall see, the incidence of temporary or circulatory migration is high. Thus the rural sector received a sizable number of migrants of this type back after the crisis. Secondly, a crucial factor is the way relative price changes help agriculture. Thailand had a large exportable surplus of agricultural commodities. The devaluation of the currency should have increased the relative prices of tradeables and helped agricultural incomes. In Ghana the food sector recovered strongly after adjustment both because of increased price incentives and of redirected infrastructure investment (Mazumdar 2002, Chapter 10), but, as we shall see, this need not have been the case in Thailand or in other East Asia in the post-crisis period. One

4

important point to note is that the world prices of commodities did not remain constant so that domestic prices did not increase as expected or as much with devaluation. The section below looks in detail at the changes in the incidence of poverty and how different segments of the labor market contribute to it. For this purpose the labor market is classified into a few groups, with successively different criteria of differentiation. A decomposition procedure of the aggregate change in poverty is suggested that enables us to measure the contribution of each segment of the labor market to the change. We study the factors affecting poverty reduction in the process of accelerated growth in the years preceding the crisis after the successful efforts at structural adjustment in the mid-eighties. It is followed by an analysis of the reversal of the process due to the impact of the crisis. But before coming to this analysis the definition of the poverty line and of the different measures of poverty is clarified in section 4. 4. Definition of the incidence of poverty The analysis is carried out using socio-economic surveys conducted in Thailand between 1988 and 1998. These are nationwide surveys that cover all households in urban and rural

areas in the country. In the current study, per capita income is used to measure the household welfare. Income used in the surveys encompasses both money and in-kind income. In measuring economic welfare, it is important to take into account different needs of each individual belonging to a household. Since households vary according to their size and composition, their needs are expected to be different. Hence, the measurement of individual welfare should reflect different needs of individuals. Kakwani and Krongkaew (1997) used a new approach to estimate poverty line for each household depending on its sex and age composition and also where it is located. The latter takes into account the regional variations in food consumption patterns, and also the spatial varistions in price indices. A household is classified as poor if its per capita income is less than the household specific poverty line. Suppose that xi is the per capita income (or consumption expenditure) of the ith household and zi is the household-specific poverty line. Then it is possible to define the welfare of the ith household as yi =

xi , which takes any value greater than zero. This measure of zi 5

welfare can be interpreted as the percentage of excess income (or expenditure) the ith household has over its poverty line. The ith household is characterized as poor if yi 1 . The FGT index measures the severity of poverty. In this study, ε = 2 is used to estimate the index. This measure will be referred here as the severity of poverty.

5. The Decomposition of the Change in Poverty into Three Elements. The labor market can be divided into different segments on the basis of different criteria. For each classification the change in the incidence of poverty can be broken down into three

6

elements: (i) any shift in population between the different segments with different degrees of poverty; (ii) the growth in income in each of the segments; and (iii) the change in the distribution of income, particularly at the lower end where the poor households are located. The methodology of such a decomposition is set out in the Appendix. To summarize the result, the percentage change in poverty for the whole economy can be expressed as:

f i Pi (∆Pi ) m f i Pi ( ∆Pi ) I P i f i  ∆f i  ∆P =∑ +∑ +∑   P P Pi P Pi P  fi  i i i =Growth Effect + Inequality Effect + Population Shift

(5)

where fi and Pi are the population share and poverty index of the ith group respectively;

fi =

f i1 + f i 2 P + Pi 2 and P i = i1 ; 2 2

the subscript m denotes the change in poverty due to mean income growth; and the subscript I gives the measure of poverty change due to change in inequality. As explained in the Appendix, this is an exact decomposition and therefore, there will not be any residual term. This decomposition does not require us to specify an inequality measure. It uses the idea of shift in that part of the Lorenz curve which affects the poor. Thus any change in poverty can be decomposed into growth effect, inequality effect, and the effect of a population shift between the segments with different incidence of poverty. This methodology is now applied to the decomposition of poverty change for different classifications of the labor market. Areas: urban and rural

We first divide the labor market into just two sectors, the urban and the rural. The process of urbanization is generally considered to be central to the process of growth and poverty reduction. In terms of the impact of the crisis, the urban sector, where the modern economy is located, could be expected to be hit more severely by the downturn. Was this shock imparted to the rural sector in the post-crisis years through return migration or other channels? Table 2 presents the estimates of population shares and various poverty estimates in Thailand for the pre-crisis period (1988-96) and the crisis period (1996-98). Although there has been a small shift in population between rural and urban areas, 7

almost 70 percent of total population in Thailand resides in rural areas. This, in turn, indicates that the Thai economy is largely engaged in agriculture and agricultural related activities. Urbanization is still in the process in Thailand. Table 2: Population shares and Poverty measures Areas

Population shares

Headcount ratio

Poverty gap ratio

FGT ratio

Pre-crisis period (1988-96) 1988

1996

1988

1996

1988

1996

1988

1996

Urban

27.8

29.6

12.56

3.06

3.42

0.70

1.43

0.25

Rural

72.2

70.4

40.31

14.91

13.08

3.73

5.83

1.40

Total

100.0

100.0

32.59

11.40

10.39

2.83

4.61

1.06

Crisis period (1996-98) 1996

1998

1996

1998

1996

1998

1996

1998

Urban

29.6

31.25

3.1

3.4

0.7

0.8

0.3

0.3

Rural

70.4

68.75

14.9

17.3

3.7

4.4

1.4

1.7

Total

100.0

100.00

11.4

13.0

2.8

3.3

1.1

1.2

In spite of the absence of any substantial shift of population to the urban areas, poverty declined dramatically from 1988 to 1996 in both sectors. For the headcount ratio the rate of decline was greater in the rural areas. But over the crisis years, while the slow transfer of population to the urban areas continued, there was a marked increase in poverty from 1996 to 1998. The relative contributions of population shift, growth and change in inequality (in the region of the poverty line) are computed based on the methodology presented in the previous section. They are set out in table 3. Table 3: Percentage change in poverty explained by changes in population, growth and inequality (urban and rural areas) Areas Explained by changes in Total % Explained by changes in Pop. Share Growth

Inequality

Change

Pop. share

Pre-crisis period (1988-96)

Growth

Inequality

Total % Change

Crisis period (1996-98)

Headcount ratio Urban

0.43

-9.53

1.15

-7.95

0.46

0.51

0.41

1.37

Rural

-1.53

-58.49

2.95

-57.08

-2.28

9.71

5.01

12.43

Total

-1.10

-68.02

4.10

-65.02

-1.82

10.21

5.42

13.81

Poverty gap ratio Urban

0.36

-8.86

1.36

-7.15

0.43

0.71

0.54

1.68

Rural

-1.46

-69.47

5.32

-65.62

-2.32

11.98

4.97

14.63

Total

-1.10

-78.33

6.67

-72.77

-1.88

12.69

5.50

16.3

FGT ratio

8

Urban 0.33

-8.58

1.25

-7.01

0.43

0.73

1.05

2.21

Rural

-1.42

-75.51

6.91

-70.02

-2.34

13.52

4.78

15.96

Total

-1.09

-84.10

8.16

-77.03

-1.91

14.25

5.83

18.2

The following are the major conclusions: (i) In both urban and rural areas, the growth effect dominates over the other two effects throughout the period—both in the decrease in poverty in the 1988-96 period and its increase in the crisis period. Partly because of the large share of the population in the rural areas, the contribution of the growth effect in the rural areas to the total reduction in poverty is dominant. In particular it is striking to note that 90 per cent of the increase in the headcount ratio in the crisis period (as indeed in the other poverty measures) is accounted for by the increase in rural poverty, and 70 per cent of it is accounted for by the effect of mean income growth in the rural sector. (ii) The effect of the change in the degree of inequality has been in the direction of increasing poverty both in the pre-crisis and the crisis period. But in the 1988-96 period its quantitative importance was small. In the crisis years the share of the inequality effect in the total poverty change has been much larger both in the rural and the urban areas. (iii) In terms of a shift in population from urban to rural sectors, the population shift led to a decline in poverty by 1.1 percent during the pre-crisis period and 1.82 percent during the crisis period. During the crisis, there was a continuous shift of different groups of people from urban to rural areas and vice versa, but it is important to note that the net effect was a net transfer from rural to urban areas. However, the increase in the population share of urban areas, although more important to the percentage change in aggregate poverty in this period compared to the growth period, still accounts for only 10 per cent of the increase in the latter. A similar analysis can be applied to other poverty measures, such as poverty gap ratio and FGT index. Results emerging from these measures indicate that as weights given to ultrapoor increase, both growth and inequality effects become increasingly more important. The importance of the rural sector in both the reduction of poverty in the growth process and its increase in the post crisis years prompts us to look more closely at the role of agriculture relative to other categories of employment. Thus we carry out the same decomposition analysis of poverty, this time dividing the labor market into its various occupational groups. Occupational Groups More than 40 percent of total population in Thailand is known to be agriculturalist. Yet, the share of agricultural households declined by 15.6 percent from 54.28 percent in 1988 to

9

45.81 percent in 1996 and further by 6.8 percent to 42.7 percent in 1998. Along with migration from rural to urban areas, a shift in occupation occurs towards sales and service industries away from backward agriculture. This process is noticeable in Table 4. The Table shows that households headed by agricultural and other primary sectors suffer the highest poverty, followed by households headed by economically inactive people and laborers. These two groups also account for the major part of the population in poverty because of their importance in the occupational distribution. Table 4: Population shares and poverty estimates (by occupational groups) Population Shares

Occupation

Headcount Ratio

Poverty Gap Ratio

FGT Ratio

Pre-crisis period (1988-96) 1988

1996

1988

1996

1988

1996

1988

1996

Professional & technicians

3.21

3.33

2.43

0.43

0.49

0.08

0.19

0.03

Executives & clerical

2.76

3.62

1.63

0.74

0.36

0.12

0.12

0.02

Sales workers

7.22

7.81

8.26

1.30

2.54

0.28

1.12

0.09

Services workers

3.91

4.13

5.97

1.08

1.09

0.22

0.40

0.06

Agriculturalists

54.28

45.81

45.60

18.76

14.81

4.70

6.65

1.75

Laborers

13.18

17.02

13.73

3.29

4.02

0.77

1.78

0.28

Economically Inactive

15.45

18.27

32.88

11.27

10.18

2.78

4.25

1.07

Total

100.00

100.00

32.59

11.40

10.39

2.83

4.61

1.06

Crisis period (1996-98) 1996

1998

1996

1998

1996

1998

1996

1998

Professional & technicians

3.33

3.56

0.43

0.20

0.08

0.07

0.03

0.04

Executives & clerical

3.62

4.14

0.74

1.44

0.12

0.25

0.02

0.07

Sales workers

7.81

9.38

1.30

2.54

0.28

0.60

0.09

0.22

Services workers

4.13

4.64

1.08

2.39

0.22

0.54

0.06

0.20

Agriculturalists

45.81

42.70

18.76

22.24

4.70

5.74

1.75

2.20

Laborers

17.02

16.89

3.29

4.60

0.77

0.95

0.28

0.30

Economically Inactive

18.27

18.69

11.27

12.21

2.78

3.13

1.07

1.21

Total

100.00

100.00

11.40

12.97

2.83

3.29

1.06

1.25

Results of the decomposition analysis are set out in Table 5. A striking result is the role played by agriculture in the changes in poverty. This is the group which accounted for most of the reduction of poverty in the 1988-92 periods, and it was also the group that was responsible

10

for the larger part of the increase in poverty in the crisis period. Yet movement out of the agricultural occupations—which is poverty reducing—has been very limited even though it continued in the crisis period. It is as though a confined body of households is trapped in the agricultural sector, and experiences the poverty incidence fall and rise with changing economic conditions. Again, the factor responsible for most of the change in poverty in this sector, as in the other occupations, is the change in mean income , with changes in the degree of inequality playing a minor role (although negative in both periods). Next in importance to agriculture is the ‘economically inactive’ group, particularly in the poverty reduction of the pre-crisis period. All the other groups, which received the small amount of households moving out of these two groups, individually and collectively accounted for a minor part of the change in poverty—less than 10 per cent of the total poverty reduction in over 1988-96 and 28 per cent of the poverty increase over 1996-98.

11

Table 5: Changes in total poverty explained by population share, growth effect, and inequality effect (by occupation) Occupation

Explained by changes in Pop. share

Growth

Total %

Inequality

Change

Explained by changes in Pop. share

Pre-crisis period (1988-96)

Growth

Inequality

Total % Change

Crisis period (1996-98)

Headcount ratio Professional & technicians

0.01

-0.24

0.04

-0.20

0.01

0.01

-0.08

-0.06

Executives & clerical

0.03

-0.28

0.19

-0.06

0.05

0.13

0.10

0.29

Sales workers

0.09

-1.44

-0.16

-1.52

0.26

0.34

0.59

1.20

Service workers

0.02

-0.60

-0.01

-0.58

0.08

0.15

0.36

0.58

Agriculturalists

-8.36

-44.95

3.73

-49.57

-5.58

12.04

1.50

7.96

Laborers

1.00

-4.26

-0.58

-3.84

-0.04

0.85

1.10

1.90

Economically Inactive

1.91

-13.10

1.92

-9.26

0.43

0.09

1.43

1.95

Total

-5.29

-64.86

5.14

-65.02

-4.80

13.60

5.01

13.81

Poverty gap ratio Professional & technicians

0.00

-0.14

0.01

-0.13

0.01

0.03

-0.03

-0.00

Executives & clerical

0.02

-0.24

0.16

-0.05

0.03

0.09

0.09

0.22

Sales workers

0.08

-1.33

-0.30

-1.55

0.25

0.22

0.74

1.21

Service workers

0.01

-0.38

0.05

-0.32

0.07

0.16

0.33

0.56

Agriculturalists

-7.95

-53.65

4.97

-56.62

-5.73

14.76

1.45

10.49

Laborers

0.88

-3.87

-0.86

-3.84

-0.04

0.87

0.25

1.08

Economically Inactive

1.76

-14.78

2.76

-10.25

0.44

0.39

1.92

2.75

Total

-5.18

-74.38

6.80

-72.77

-4.98

16.52

4.76

16.30

FGT ratio Professional & technicians

0.00

-0.12

0.01

-0.11

0.01

0.02

0.01

0.04

Executives & clerical

0.01

-0.22

0.15

-0.05

0.02

0.07

0.09

0.19

Sales workers

0.08

-1.26

-0.42

-1.59

0.23

0.24

0.82

1.30

Service workers

0.01

-0.35

0.05

-0.28

0.06

0.17

0.41

0.64

Agriculturalists

-7.72

-58.85

5.67

-60.90

-5.80

16.78

1.92

12.90

Laborers

0.86

-3.69

-1.24

-4.08

-0.04

0.90

-0.55

0.31

Economically Inactive

1.63

-15.28

3.64

-10.01

0.45

0.43

1.92

2.80

Total

-5.13

-79.77

7.86

-77.03

-5.06

18.62

4.61

18.17

The importance of agriculture and the ‘economically inactive’ groups in the process of changes in the incidence of poverty induced us to look more closely at the composition and

12

sources of income of these groups. First, who are the ‘economically inactive’? The various groups included in this category and their relative importance in the group are given in Table 6. Note that unskilled workers, not elsewhere classified, are also included in this group. Table 6: Population share within economically inactive groups Economically

Years

Inactive groups

1988

1996

1998

Unskilled workers

20.1

11.2

5.7

Housewives

15.3

13.7

15.4

Students

1.4

1.8

2.6

Retired workers

51.5

66.0

67.8

Disabled workers

4.3

4.6

4.6

Unemployed

1.0

0.5

3.0

Beggars

0.0

0.0

0.0

Not reported & no

6.4

2.2

0.9

100

100

100

Secondary occupation Total EIG

The share of economically inactive groups has increased from 15.4 percent in 1988 to 18.3 percent in 1996 and further to 18.7 percent in 1998. A large proportion of the population in this group is found to be retired workers. It is interesting to note that the share of housewives and retired workers increased significantly between 1996 and 1998. This may be explained by the so called ‘discouraged worker effect’ during the crisis period. We next consider the components of the income accruing to agriculture households. While such households are defined to be those whose major source of income is farming, it is well known that most of these households obtain their income from multiple sources. Has there been any significant change in the shares of different types of income over the period we are considering? It seems that there has been no dramatic change in the shares of different types of income over the crisis period. There has been a drop on the relative share of ‘wages and salary’ and the result is an increase in the share of farm income. This as is to be expected since the impact of the crisis is felt more directly on the wage sector. But it is little surprising to find that the share of remittances show only a marginal decline. It is possible that if we isolate the regions from which individual migrants usually come to work in the wage economy of Bangkok and

13

other areas, and remit money back to the family left in the villages of their origin, the change in the share of remittances would be larger. Table 7. Share of the different sources of income of agricultural households (% of total)

Income sources

1988

1996

1998

Wages & Salaries

25.0

22.6

Non-farm Income

8.7

8.5

Farm Income

50.2

52.3

Remittances

12.1

11.8

Property Income

1.3

1.1

Other money income

2.9

3.5

TOTAL

100.0

100.0

To come back to the question of the increase in the incidence of poverty among agricultural households: if the composition of the income of these households changed only slightly over the crisis years, we can conclude that the increase in poverty in this group over the crisis is due to changes in farm income—both a decline in mean income and an increase in inequality in the lower end of the distribution. The result shows that the price effect, which the devaluation might have caused, increasing the terms of trade of agriculture, was either non-existent or swamped by the drastic reduction of village-to-city migration. One important point to note is that the world prices of commodities did not remain constant so that domestic prices could increase with devaluation. In fact “currency devaluation seems to have increased the competitiveness of the crisis countries, thereby contributing to increases in supply of several commodities “ (World Bank 2000, p.108). For example: prices of rubber (Indonesia, Malaysia and Thailand account for 70 per cent of global exports) fell by nearly one-third in the two years following the start of the crisis in July 1997 (although they had already halved in the two years prior to this). Similarly rice price (Thailand accounted for 23 per cent of world exports) dropped sharply after July 1997. Regions We turn now to another important dimension of poverty differences – that relating to different regions of the country. The Northeast is the poorest region in Thailand, and Bangkok and its vicinity the richest. The gap in income levels between the two is enormous. Statistics produced by the National Economic and Social Development Board (1994) shows that per capita income in Bangkok was four times the average in the Northeast. Part of the reason for the

14

income-gap between the regions is that the poorer regions are less urbanized-- and incomes tend to be higher in urban areas in all regions. The regional income disparities have been traditionally mitigated by internal migration on large scale, but this migration is more of a seasonal rather than a permanent type. The National Migration Survey (NMS) of 1995 (Appendix table 6) showed that rates of net inmigration were generally small, with gains ranging from 4.1 per thousand for Bangkok to 12.8 per thousand for the South region over the two-year period. By contrast short-term population flows were strongly affected by migration. Net losses as a result of seasonal and other temporary migration were nearly 150 per thousand for Bangkok in the 1995 wet season and 50 per thousand for the Central region. At the receiving end net gains of 33 and 68 per 1000 are recorded for the North and the North-East respectively for this season. There are two major types of seasonal migration in Thailand, both involving the North and North-East on the one hand, and Bangkok and the Central-South regions on the other. One important stream involves migration into Bangkok from the North and North-East during the dry season. In the period of enumeration of the NMS in the wet season these migrants show up as urban-to-rural migrants in the North and North-East. The second stream is rural-to-rural migration to meet the peak demands for labor in the busy season. Table 8 shows that the share of population residing permanently in each region has remained pretty constant over time. More than 30 percent of total population is found to live in the Northeast region in Thailand. While the population share has declined slightly in Northeast, Bangkok has shown a little increase in its population for the last decade. Table 8 also shows that the Northeast is the region with the highest incidence of poverty, followed by the South and the North. Bangkok has remained the most affluent region that enjoyed the largest reduction in poverty during the period 1988-96: the incidence of poverty in Bangkok alone fell sharply from 6.1 percent in 1988 to 0.6 percent in 1996, almost 90 percent reduction in the head-count ratio. The Northeast, the poorest, also shared in the substantial reduction in poverty during 1988-96. In relative terms the percentage decline in the poverty measures in the North-East was somewhat larger than the economy-wide average, though not as large as in the Bangkok region—the headcount rate, for example, falling by 40 per cent against 35 per cent for the country—but because of the large incidence of poverty in the initial period, poverty remained significantly higher in this region than in any other. The crisis years changed all this. Poverty increased in all regions with the exception of Bangkok. Thus the disparity between Bangkok and the other regions--including the Northeast— increased even further after the crisis.

15

Table 8: Population shares and poverty estimates (by regions) Regions

Population shares

Headcount ratio

Poverty gap ratio

FGT ratio

Pre-crisis period (1988-96) 1988

1996

1988

1996

1988

1996

1988

1996

Central

16.7

16.8

26.6

6.3

8.1

1.5

3.6

0.6

North

19.8

18.6

32.0

11.2

9.5

2.7

3.9

1.0

Northeast

34.3

34.1

48.4

19.4

16.3

4.8

7.4

1.8

South

13.0

13.3

32.5

11.5

10.2

3.2

4.4

1.3

Bangkok

16.3

17.3

6.1

0.6

1.7

0.1

0.8

0.0

Total

100.0

100.0

32.6

11.4

10.4

2.8

4.6

1.1

Crisis period (1996-98) 1996

1998

1996

1998

1996

1998

1996

1998

Central

16.8

16.6

6.3

7.6

1.5

2.0

0.6

0.8

North

18.6

18.1

11.2

9.1

2.7

2.3

1.0

0.9

Northeast

34.1

33.4

19.4

24.0

4.8

5.9

1.8

2.2

South

13.3

13.2

11.5

14.6

3.2

4.0

1.3

1.6

Bangkok

17.3

18.6

0.6

0.6

0.1

0.2

0.0

0.1

Total

100.0

100.0

11.4

13.0

2.8

3.3

1.1

1.2

Table 9 presents the decomposition of changes in total poverty by regions. Because of its high initial value, changes in poverty in Northeast have contributed the most to the change in total poverty in Thailand during the growth period. The very limited impact of population shifts is brought out in these figures. The growth effect dominates in the accounting of poverty reduction. Inequality increase had a small negative effect on poverty decline. The crisis period saw a significant reversal of the trends. But it can be seen that in quantitative terms almost all of the net increase in poverty in the kingdom is accounted for by the increase in the North-East. Again it is the reversal of the growth effect which is responsible for the post-crisis increase. It is important to note that population shift continued to be away from the regions of grater poverty incidence. The negative effect of population shifts on poverty change was in fact larger in the crisis years than in the period of growth. Thus the evidence by itself would contradict the hypothesis that the negative impact of the crisis was imparted to the rural areas in the low income regions through a reversal of the migration flows from the poorer to the richer regions. But such a conclusion needs more careful assessment. There are different types of migration involved in the process of labor transfer—permanent migration of more skilled or productive labor is mixed up with the movement of less skilled temporary or seasonal migrants.

16

It is possible that while the former type to the Bangkok area might have increased in the crisis period, the movement of the latter might have been reversed. In fact, it is the low marginal productivity of this latter type of labor which might have been partly responsible for the negative growth rate of the North-East noted in the figures of Table 9. Table 9: Changes in poverty explained by population share, growth effect, and inequality effect (by regions) Regions

Explained by changes in Pop. Share

Growth

Total %

Inequality Change

Explained by changes in Pop. Share

Pre-crisis period (1988-96)

Growth

Total %

Inequality Change

Crisis period (1996-98)

Headcount ratio Central

0.04

-9.77

-0.64

-10.36

-0.11

0.88

1.04

1.81

North

-0.81

-13.06

0.82

-13.05

-0.41

-1.51

-1.94

-3.86

Northeast

-0.23

-33.37

2.97

-30.63

-1.20

13.20

0.28

12.28

South

0.19

-8.60

0.14

-8.27

-0.04

0.76

2.84

3.57

Bangkok

0.11

-2.35

-0.47

-2.71

0.07

0.07

-0.13

0.01

Total

-0.69

-67.16

2.83

-65.02

-1.68

13.40

2.09

13.81

Poverty gap ratio Central

0.04

-10.25

-0.37

-10.59

-0.11

0.86

1.76

2.51

North

-0.71

-13.94

1.39

-13.25

-0.40

-1.47

-0.85

-2.72

Northeast

-0.22

-41.55

3.94

-37.83

-1.20

16.84

-3.45

12.19

South

0.18

-9.99

1.14

-8.68

-0.04

1.02

2.88

3.86

Bangkok

0.09

-1.98

-0.53

-2.42

0.07

0.12

0.28

0.46

Total

-0.62

-77.71

5.56

-72.77

-1.68

17.36

0.62

16.30

FGT ratio Central

0.04

-10.61

-0.35

-10.93

-0.11

0.93

2.20

3.01

North

-0.65

-14.13

2.00

-12.78

-0.42

-1.55

0.46

-1.51

Northeast

-0.22

-46.13

4.36

-41.99

-1.18

18.94

-6.02

11.75

South

0.17

-10.71

1.72

-8.81

-0.04

1.23

2.87

4.06

Bangkok

0.09

-1.82

-0.79

-2.52

0.07

0.12

0.66

0.85

Total

-0.57

-83.40

6.94

-77.03

-1.68

19.67

0.17

18.17

Education The final classification of labor markets we look at is by levels of education. Along with industrialization and structural transformation of the economy with a growing share of the modern sector, the demand for skilled labor increases. Skill embodied in the labor force can be obtained through education or various forms of training. Education improves job prospects and

17

expands an individual’s earning capacity. `Table 10 shows that even after its impressive growth, more than 70 percent of the Thai population is concentrated at the primary level. Although the share of population in secondary education has increased gradually over time, it seems to be far lower than it is in neighboring Asian countries. This points to a problem associated with the present educational system in Thailand. It has been note by many observers and has been singled out as one of the major causes of the limited ability of the Thai economy to diversify its production sector technologically (see, for example, Sussangkarn 1994). The low level of secondary education may result from a perception that future returns from pursuing education beyond the primary level will be low, based on the expectation of finding a job in the informal sector where the majority of Thailand’s less-educated workforce is employed. Alongside this perception of low expected returns from education is the relatively high cost of further education. This combination discourages many families, particularly poor households, from pursuing secondary education. We shall now see how the growth process rewarded different education groups and what the differential impact of the downward has been. Table 10: Population shares and poverty estimates (by education) Population Shares

Education

Headcount Ratio

Poverty Gap Ratio

FGT Ratio

Pre-crisis period (1988-96) 1988

1996

1988

1996

1988

1996

1988

1996

No formal education

10.69

9.26

43.70

20.75

14.21

5.64

6.31

2.30

Primary

76.38

74.19

35.49

12.24

11.27

2.98

4.99

1.09

Secondary

6.06

8.65

9.35

4.11

2.92

1.01

1.33

0.38

University

6.87

7.90

3.60

0.52

1.20

0.12

0.53

0.04

Total

100.00

100.00

32.59

11.40

10.39

2.83

4.61

1.06

Crisis period (1996-98) 1996

1998

1996

1998

1996

1998

1996

1998

No formal education

9.26

7.83

20.75

21.33

5.64

6.62

2.30

2.95

Primary

74.19

71.44

12.24

15.21

2.98

3.73

1.09

1.37

Secondary

8.65

10.93

4.11

3.34

1.01

0.79

0.38

0.30

University

7.90

9.79

0.52

0.75

0.12

0.19

0.04

0.08

Total

100.00

100.00

11.40

12.97

2.83

3.29

1.06

1.25

As would be anticipated, poverty is the most severe among those who have not received

18

any formal education. As indicated by the table, the degree of poverty tends to decline with the level of education. The growth of the economy in the 19988-96 period helped the decline in poverty at a higher rate among those with only primary education, than in the group with secondary education or with no education. This, together with the fact that as much as threequarters of the laborforce was accounted for by this group, implied that the bulk of poverty reduction in this period could be attributed to this group, as can be seen from the results reported in Table 11. Table 11: Changes in total poverty explained by population shares, growth effect, Education

and inequality effect (by education) Explained by changes in Pop. share

Growth

Total %

Inequality Change

Explained by changes in Pop. share Growth

Pre-crisis period (1988-96)

Total %

Inequality Change

Crisis period (1996-98) Headcount ratio

No education

-1.42

-7.94

0.91

-8.44

-2.63

1.72

-1.29

-2.20

Primary

-1.60

-56.65

2.96

-55.30

-3.31

19.69

-0.74

15.64

Secondary

0.54

-1.70

0.52

-0.65

0.75

0.70

-1.36

0.09

University

0.06

-0.56

-0.14

-0.63

0.11

0.06

0.12

0.28

Total

-2.42

-66.85

4.25

-65.02

-5.09

22.17

-3.27

13.81

Poverty gap ratio No education

-1.37

-9.88

1.66

-9.59

-3.09

2.12

0.84

-0.12

Primary

-1.50

-65.00

4.89

-61.61

-3.26

23.62

-4.21

16.15

Secondary

0.49

-1.78

0.43

-0.86

0.72

0.71

-1.48

-0.05

University

0.07

-0.60

-0.17

-0.71

0.10

0.05

0.17

0.33

Total

-2.31

-77.26

6.81

-72.77

-5.52

26.50

-4.67

16.30

FGT ratio No education

-1.34

-11.09

2.41

-10.02

-3.54

2.67

2.59

1.71

Primary

-1.45

-69.44

5.63

-65.25

-3.19

26.19

-6.98

16.02

Secondary

0.48

-1.79

0.28

-1.03

0.74

0.72

-1.51

-0.05

University

0.06

-0.62

-0.18

-0.73

0.10

0.06

0.32

0.49

Total

-2.24

-82.93

8.14

-77.03

-5.90

29.65

-5.58

18.17

It is apparent from the figures presented that very little of the production could be attributed to the educational upgrading of the laborforce. In fact it is the growth of mean income per capita which is dominant in the process depicted here, as in the other classifications of the labor market. Again increase in inequality among the all the educational groups (except the university educated) reduced, but by a small amount, the decrease in the incidence of poverty. The crisis period saw a sharp reversal of the trend—and again the increase in poverty by

19

all three measures was higher for the group with primary education, than for those with secondary education or no education. The percentage increase in poverty incidence was indeed the highest for the university educated but it is such a small group that its share in the total poverty increase was minimal. Educational upgrading continued at a slow rate during the crisis years, as in the earlier growth period, but although it mitigated the impact of the fall in mean income somewhat, the importance of the latter was dominant.

6. Conclusions on the Impact of the Thai Boom and Bust This paper has studied the impact of the growth process and of the post-crisis downturn on the incidence of low household income in Thailand. For this purpose the incidence of poverty was selected as the center of attention. Poverty is reduced during the growth process in three ways: (i) a shift of population from the low income segments of the labor market to the more productive growing sector; (ii) an increase in mean per capita income in both the low and high income sectors; and (iii) a reduction in the inequality of distribution in the region of the poverty lines. We proposed a decomposition analysis which quantifies the contributions of these three elements to the over-all poverty reduction, and the methodology was applied to a set of different classifications of the Thai labor market. The most important and perhaps distinguishing characteristic of the Thai growth process in the period of accelerated growth 1988-96 was that shift of population from the low to the high income segment played a very minor role in the substantial poverty reduction which took place. Most of the decrease in poverty was due to mean income growth, particularly in the low income segments of the market. Distribution of income played a minor but adverse role in poverty reduction. This conclusion goes against some of the implications of growth models visualized for agrarian economies based on the historical evidence of large population transfers from low productivity sectors. The impact of the downturn was to increase poverty as is to be expected. But in spite of the fact that the crisis hit the formal modern sector first and foremost, the impact of the downturn on low income households was more important in the low productivity segments of the labor market. It is not only that the share of the poor located in these sectors was higher so that these groups would account for a larger share of poverty increase. The relative incidence of poverty also seems to have increased more in some of the low-income sectors. In all four classifications of the labor marker those who benefited most from the boom also suffered the most in relative terms. The population in the rural areas, those with agriculture as the principal occupation, the poorer regions of the economy like the North-East, and the less educated are the groups which reveal a larger incidence of poverty increase in the 1996-8 period. Again the increase in poverty was not due to increase in the share of the population of the vulnerable groups. The percentage

20

decrease in mean income was by far the most important factor. The way the shock of the crisis was transmitted from the modern sector to the vulnerable groups in agriculture, in the poor North-East, and to those with only primary education merit more detailed research. We have suggested two factors which might have been of some importance. First, is the return of temporary migrants from the rural North-East in particular who participated in the boom in the Bangkok and surrounding regions. The share of the population of the North-East, as already seen did not increase, but the composition of the population might have changed, with more of the less skilled labor with lower than average marginal productivity returning to their area of origin. A second factor was the decline in the prices of some important agricultural export commodities, e.g., rice. The downward trend in the world prices of these commodities was accentuated by the slowdown in the Asian region as a whole, and this factor overwhelmed any gain farmers might have got from the shift in the terms of trade in their favor because of the devaluation.

7. Policies to help the vulnerable groups The finding that the groups hit most strongly by the downturn are not those directly involved in the formal economy also makes the problem of devising programs to help them a difficult one. The bottom line is that the structural problems of the Thai economy have left the most vulnerable groups stuck in pockets of low income in agriculture, in the rural areas, and in the depressed regions of the country. A surprising result is that although these groups seemed to have been helped by the growth process in so far as the incidence of poverty has fallen relatively faster in these parts of the labor market, these were precisely the group which suffered most from the downturn in falling below the poverty threshold. It is difficult to devise programs meant to help vulnerable groups affected by a downswing of the cycle, when the problem is basically a structural one. A second difficulty is that a vast majority of these workers affected by the downswing are not wage employees but self-employed, working on their own farms or small businesses. Unlike more industrialized countries like Korea the impact of the crisis, therefore, did not lead to a large increase in the number of the openly unemployed. To be sure, the rate of unemployment increased by a large percentage—almost as much as Korea in relative terms. But while in Korea the unemployment rate climbed from the pre-crisis level of around 2.2 per cent to 8.4 per cent in the first quarter of 1999 (Betcherman and Islam, chapter 3) the Thai unemployment rate peaked at 5.2 per cent in the same time period (World Bank 2000b, p.8). Employment loss was concentrated in the wage sector, but those displaced from jobs did not all become totally

21

unemployed. Rather they were absorbed in self-employment (both own-account workers and unpaid family labor). Kakwani’s analysis of the Labor Force data shows that both these categories showed a major reversal of declining trend established in the decade before 1996. But this absorption of labor in the self-employed categories was accompanied by a very much larger decline in earnings than in the wage sector (Kakwani and Pothong).1 Much of the discussion in post-crisis Thailand among policy makers and their advisers has been directed to ameliorating the conditions of those affected by the downturn on the wage sector. These include formal labor protection programs like severance pay, social security coverage (including unemployment insurance), and minimum wages. Some of these concerns are no doubt relevant to large numbers of workers affected by the crisis. After all the wage sector account for more than a third of the workforce. Furthermore, if we look at changes in real wage levels and employment the impact of the crisis is clearly regressive even within the wage sector. Younger and less educated workers, those outside the Bangkok area, those in smaller establishments or in lower paid occupations were affected relatively more by the decline. (World Bank 2000b, pp 3-4 and chapter 4). Thus focusing on such measures as increasing the coverage of social security to low paid workers is an important part of the measures to help low income earners. But as the analysis of this paper shows these labor protection measures are unable to deal with the major impact of the downturn on the self-employed, and particularly in the subsectors where the impact has been felt most strongly. The most important approach of addressing the problems of the low-income group outside the wage sector has been the job creating programs of small-scale public works type. The expenditures involved in these programs were financed by external borrowing. They included the so-called Miazawa package and the Social Investment package (SIP) supported by a number of international agencies including the Japanese government. These programs came rather late in the downturn, beginning to be implemented only in April 1999. Apart from the timing questions have been raised about its efficiency in targeting. The projects under these schemes were expected to pay at least the minimum wage in the region. But in Thailand a very large persons of the labor force (as much as 30 pr cent) had earnings below the minimum wage (World Bank 2000, Figure 28, p.34). Apart from the problems of targeting, although the project was on a large scale, and seems to have produced a perceptible dent in the unemployment rate in 1999, its limitation as a 1

See Mazumdar and Horton (pp 456-57) for a summary of the data on the change in the crisis period of employment and earnings, relative to the trend lines established before the crisis, for different employment categories. In 1998-3 the percentage deviation of income per earner from its specific trend line was –5.8 for wage earners in non-agriculture, but –35.8 for own-account workers in business and –36.0 for ownaccount workers in farming.

22

short-run measure needs to be emphasized. It could hardly be expected to address the basic problem of labor being locked in large pockets of low-income segments of the economy outlined in this paper. In fact, by temporarily augmenting the resources available for targeted spending it might indeed have distracted attention from the long-run need for the Thai fiscal system to provide much needed resources to provincial authorities to promote local and regional development. The high degree of centralization of the Thai fiscal system has been one of the factors responsible for the failure of the Thai economy to develop infrastructure and other supportive services which would encourage a less concentrated growth of industry. (See, e.g., Yin and Lao).

23

REFERENCES Ahuja, V., B.Bidani, F. Ferreira and M. Walton (1997). Everyone’s Miracle? Revisting Poverty and Inequality in East Asia. World Bank, Washington D.C. Betcherman, Gordon and R. Islam (eds) (2001). East Asian Labor Markets and the Economic Crisis: Impacts, Responses and Lessons. The World Bank, Washington D.C. and the International Labor Organization, Wsahington D.C. Kakwani, N. and J. Pothong ( 1998). ‘Indicators of Well-being and Poverty Analysis’, vol.2, No.4, October, NESDB, Bangkok. Kakwani, N. and Medhi Krongkaew (1998). ‘Poverty in Thailand: Defining, Measuring and Analyzing’, Working Paper No.4, Asian Development Bank and Dvelopemnt Evaluation Division, National Economic and Social Development Board (NESDB), Bangkok. Kuznets, Simon (1955). ‘Economic growth and Income Inequality’, American Economic Review, 45,1: 1-28. ____________ (1963). ‘Quantitative Aspects of the Economic growth of nations; VIII. Distribution of Income by Size’, Economic Development and Cultural Change, 12: 1-80. Lewis, W. A. (1954). ‘Economic Development with Unlimited Supply of Labor’, Manchester School, 22, 2: 139-91. Mazumdar, Dipak ( with Ata Mazaheri) (2002). Wages and Employment in Africa, Ashgate Academic Publishers, Aldershot, UK, forthcoming. Mazumdar, Dipak and Susan Horton (2000). ‘Impact of the Asian Finacial Crisis on Labor Markets, Income Distribution and Poverty: A Comparative Study of Five Countries’ Indian Journal of Labor Economics, Special Issue on East Asian Labor Markets, 43, 2, 451-465. Son, Hyun Hwa (2001). Economic Growth, Inequality and Poverty: Korea and Thailand, Unpublished Ph.D. Dissertation, School of Economics, University of New South Wales, Sidney. Sussangkarn, Chalongphob (1994). ‘Thailand’ in Suasn Horton, Ravi Kanbur and Dipak Mazumdar (eds), Labor Markets in an Era of Adjustment, Vol. 2, EDI Development Studies, World Bank, Washington D.C. World Bank (2000a). Global Economic Prospects. Washington D.C. __________(2000b). Thailand Social Monitor. July, World Bank Thailand Office, Bangkok. __________(various years). World Development Report. Washington D.C.

24

APPENDIX .

Poverty Decomposition Methodology

Suppose that the total population is divided into k mutually exclusive socioeconomic and demographic groups. For decomposable poverty measures, then, we can write the total poverty as the weighted average of poverty within each group. P = ∑ f i Pi i

where fi and Pi are the population share and poverty index of the ith group, respectively. Further, define the change in poverty between two periods as ∆P = P2 − P1

(1)

where P1 = ∑ f i 1 Pi 1 and P2 = ∑ f i 2 Pi 2 . Equation (1) can be written as i

∆P =

i

1  1  f i 1 ( Pi 2 − Pi 1 ) + ∑ f i 2 ( Pi 2 − Pi 1 ) + ∑ Pi1 ( f i 2 − f i 1 ) + ∑ Pi 2 ( f i 2 − f i 1 ) ∑  2 i i i  2 i 

which shows that the change in total poverty can be written as the sum of two components. The first component measures the effect on total change in poverty due to changes in within-group poverty and the second component estimates the change in total poverty due to possible shifts in population between groups. The percentage change in total poverty, thus, can be written as follows: f P  ∆P ∆P = ∑ i i  i P P  Pi i

where f i =

 Pi fi  + ∑  i P

 ∆f i   fi

  

(2)

f i1 + f i 2 P + Pi 2 and P i = i1 . Note that the first term in Equation (2) 2 2

estimates the percentage change in total poverty explained by changes in poverty within groups. The second term estimates the percentage change in total poverty due to a shift in population between groups. The shift in population is deemed pro-poor if the second term is negative because it leads to a reduction in poverty. This situation is likely to occur if migration occurs from rural to urban areas. If migration takes place from urban to rural areas, on the other hand, the second component is likely to make a positive 25

contribution to poverty. In this case, the population shift is not pro-poor. Kakwani (2000) has proposed a decomposition, which explains the percentage change in poverty as a sum of two components: one is the growth effect, measuring the change in poverty when mean income changes but inequality remains fixed and the other component is the inequality effect, which measures changes in poverty when inequality changes but the mean income remains constant. This methodology can now be applied within each group. A general poverty measure is characterized as P = P( z , µ , L ( p )) where z is the poverty line, µ is the mean income of society, and L(p) is the Lorenz curve. The Lorenz curve measures the effect of inequality on poverty. Following from Kakwani (2000), the percentage change in poverty can be written as ∆P = ( ∆P) m + ( ∆I ) I = Mean effect + Inequality effect

(3)

where ( ∆P) m is the change in poverty if mean income changes from µ1 in period 1 to

µ 2 in period 2 but the Lorenz curve remains fixed. Thus, ( ∆P) m can be written as

( ∆P) m =

1 [ P( z, µ 2 , L1 ( p)) − P( z, µ1 , L1 ( p)) + P( z, µ 2 , L2 ( p) − P( z , µ1 , L2 ( p))] 2

where L1(p) and L2(p) are the Lorenz curves in periods 1 and 2, respectively. Note that in deriving the mean effect, we can either fix the Lorenz curve for the initial period or for the terminal period. Because we do not know a priori which period of the Lorenz curve we should fix, we have taken the average of the two periods. Similarly, the inequality component can be derived as 26

( ∆I ) I =

1 [ P( z, µ1 , L2 ( p)) − P( z, µ1 , L1 ( p)) + P( z, µ2 , L2 ( p) − P( z, µ2 , L1 ( p))] 2

which estimates the change in poverty if inequality measured by the Lorenz curve changes from L1(p) in the initial period to L2(p) in the terminal period but mean income is fixed between the two period. The sum of the mean and inequality effects gives rise to the total changes in poverty. We apply the decomposition in (3) within each group, which results in ∆Pi ( ∆Pi ) m ( ∆Pi ) I = + Pi Pi Pi

(4)

for i = 1, 2, 3, …….., k. From (2) and (4), the percentage change in total poverty can be expressed as f i Pi (∆Pi ) m f i Pi ( ∆Pi ) I P i f i  ∆f i  ∆P   =∑ +∑ +∑ P P Pi P Pi P  fi  i i i =Growth Effect + Inequality Effect + Population Shift

(5)

Therefore, any change in poverty can be decomposed into growth effect, inequality effect, and a shift in population shares. This is an exact decomposition and therefore, there will not be any residual term. This decomposition does not require us to specify an inequality measure. It uses the idea of shift in that part of the Lorenz curve, which affects the poor.

27