NOTES

К оглавлению
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 
34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 
51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 
68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 
119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 
136 137 138 139 140 141 142 

1. More recent World Development Reports broadened the defi nition of poverty to include

various non-income dimensions (World Bank, 2000) and stressed problems of delivering

services to the poor (World Bank, 2004). The discussion of corruption and clientelist

politics in the latter is relevant to the governance problem associated with targeting

implementation.

2. Meyer (2002) gives a useful survey of the diffi culties of assessing the impact of microfi

nance programs on poverty reduction.

3. The only exception to the use of offi cial estimates is in the case of India, where in Figure

1.6 the fi nal year observation comes from the estimates of Deaton (2001).

4. We can write the poverty gap (PG) as

PG

1

1 n

z y

z

i

i

p

where z is the poverty line, yi is income of individual i, p is the number of poor and n is

the total population.

5. Hence the squared poverty gap (SPG) is

The theory behind SPG is set out in Foster et al. (1984).

6. Schemes with budgets of below Rs 1 billion are excluded from this total.

7. For example, a poverty map has just been completed on a trial basis for three provinces in

Indonesia (Suryahadi et al., 2003). There are doubts however as to whether this approach

can apply at the village level. Hentschel et al. (2000) explain the poverty mapping approach

and illustrate how it can improve on the use of simple basic indicators to identify the

poor.

8. For evidence of undercoverage a World Bank survey in Uttar Pradesh, one of the poorest

states, found that 56 per cent of the lowest income quintile did not get identifi cation

cards to enable them to access the public distribution system (Srivastava, Chapter 2 in

this volume).

9. Perdana and Maxwell (Chapter 3 in this volume) report evidence that in the late 1990s

those in the top four quintiles of household expenditure were three quarters of the

recipients of subsidized rice and that only roughly half of the poor families under the

offi cial criteria were recipients (see Table 1.2).

10. In the late 1980s in only one third of the classifi ed poor counties was income per capita

actually below the original norm of RMB 150 set by the Leading Group for Poverty

Reduction of the State Council (Wang, Chapter 4 in this volume).

11. The poverty-mapping model is in Balisacan et al. (2002).

12. This scheme received a great deal of attention internationally and was commented on

very favorably in World Bank (1990). Its performance declined substantially after 1979

following a large increase in the wage offered, thus weakening the self-selection by the

poor (Gaiha, 1996).

13. The targeting count gap (TCG) is defi ned as

TCGt = 1/N {Iit1(Pit = 0, Yit < Zt) + Iit2 (Pit = 1, Yit > Zt)}

where N is the total number of counties, indexed by i, and t is a time period. Iit1 is an

indicator of undercoverage (type 1 error) and equals 1.0 if a county is not designated

as poor (Pit = 0), but its income per capita (Yit) is below the poverty line (Zt). Iit2 is an

indicator of leakage (type 2 error) that equals 1.0 if a county is designated as poor (Pit

= 1), but its income per capita is above the poverty line.

14. Weiss (2003) fi nds that the key factor infl uencing rural poverty reduction in PRC across

provinces has been the growth of grain production and to a lesser extent the trend in

farm-gate prices.

15. Perdana and Maxwell (Chapter 3 in this volume) enter the qualifi cation that fi rstly these

results may not be representative of the national picture and secondly that the IDT scheme

was revised to address the undercoverage problem.

16. van de Walle (1998b) reports this result for Indonesian data from the late 1980s, although

the bias is much greater for hospitals where gains to the top decile of the income strata in

monetary terms are roughly seven times those to the bottom decile. World Bank (2004)

Figure 2.5 reports estimates for Indonesia in 1989 and 1990 showing the gains to the

poor (the bottom 20 per cent) from public spending to exceed the gains to the rich (the

top 20 per cent) for primary health care and primary education expenditure, whilst the

reverse holds for aggregate expenditure under both headings.

17. In the analysis, variation in growth explains about 40 per cent of the variation in poverty

reduction between countries. It is well known that there can be major regional variations

in the growth–poverty relation within countries. Datt and Ravallion (1998) provide an

analysis across Indian states and fi nd that poor initial conditions in terms of the rural

sector and weak human resource development lower the impact of a given rate of nonfarm

growth on poverty.

18. The authors hypothesize that the lower poverty elasticity in the Philippines as compared

with Indonesia is due to the relatively more agriculture and labor-based growth pattern

in the latter.

19. The Chinese data can be interpreted in different ways. As Stern (2001: 109) points out

‘For poor people in most countries of the world, of course, an average income growth

of 4 per cent annually [sic for the 1990s] would be a great improvement – but in China

that rate was just one-third of the 12 per cent growth rate that the wealthiest enjoyed.’

20. Cross-country analysis such as Dollar and Kraay (2004) has tended to fi nd that as an

average relation the income poverty elasticity is around unity, implying that the poor’s

income rises by the same proportion as the average of the country concerned. The country

cases discussed above do not support this result.

21. The exception here is micro-fi nance funds, which have become signifi cant in some

countries, particularly Indonesia of those covered here.

22. Using a different approach Jalan and Ravallion (1998) fi nd a similar return to the poor

area development program in Southwest China.

23. Given that the poverty loan variable is not statistically signifi cant it is unclear whether

much meaning can be placed on the impacts for poverty loans. However it should be noted

that the structure of the model appears to ensure that other categories of expenditure

will have a greater impact on poverty reduction than will poverty loans. This is because

these loans only impact on poverty directly (in equation 1 of the system), whilst other

expenditure categories enter indirectly through their effect on growth in productive sectors

(equations 2 to 11).

24. Here the coeffi cient is statistically signifi cant. The poverty expenditure enters into the

equation for non-agricultural employment (equation 6) and the latter is one of the terms

in the poverty equation (equation 1).