DATA SOURCES FOR TARGETING

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 

Accurate, reliable and up-to-date data about poverty are vital for both

program planners and implementers as they seek to locate the precise

whereabouts of the poor and ensure that program benefi ts are delivered

to those who are most deserving of special assistance. Before we begin

to consider the targeting of particular poverty alleviation programs in

Indonesia in any detail, it is important to make some general observations

about the particular problems associated with the data that have been used

for the purposes of identifying poverty in Indonesia, both at a geographic

level and at the household level.

In an archipelagic country as geographically diverse as Indonesia there

have been considerable diffi culties in accounting for poverty on a geographic

basis. The offi cial poverty statistics that are presently available provide

poverty estimates to the provincial level and on an urban/rural divide, but

do not break down poverty data beyond that point, for example to the

district (kabupaten) and sub-district (kecamatan) levels.

In the immediate post-Crisis period, a precise account of the social

impacts of the Crisis was not immediately available to the central

government and its senior offi cials who tended to regard the whole of the

country as being equally affected. In fact, by late 1998 information was

appearing to show that the Crisis was quite heterogeneous in its impacts

and that urban areas were hardest hit, as was Java-Bali, and much of eastern

Indonesia.11 However, there was a tendency for many of the emergency

poverty alleviation measures that made up the government’s Social Safety

Net program to apply a safety blanket over the entire country, relying on

local offi cials and program implementers to ‘fi ne-tune’ the targeting at lower

levels of administrative authority.

Regarding information about poverty at the household level, the only

available source of data that covers the entire country in a thoroughly

comprehensive manner has been that produced by the National Family

Planning Coordinating Board (BKKBN), which produces a classifi cation of

family welfare (kesejahteraan) compiled from a national registration based

on the work of an army of village-based, family planning cadre. One of the

major criticisms of the BKKBN registration is that, in marked contrast to

the Central Board of Statistics (or BPS) socio-economic surveys and census

that employ paid and trained data collectors and enumerators, the village

cadre who carry out the BKKBN registration are unskilled and do the work

in an honorary capacity. Although this is substantially true and must have

some infl uence on the accuracy of the registration process, this must also be

balanced against the advantages of local knowledge.12 The data is collected

on a regular basis and the results are updated annually.

The variables upon which this classifi cation was originally based covered

food consumption patterns, the type of health care family members were

able to access, the possession of alternative sets of clothing, the material

and size of the fl oor of the family home, and the ability of household

members to practice their religion. Families that failed to meet certain

minimum standards in any one of these fi ve areas were registered in the

lowest welfare category, referred to as ‘pre-prosperous families’ (keluarga

pra-sejahtera or KPS).13

As a result of widespread criticisms from various quarters about the

suitability of applying such a classifi cation to determine socio-economic

status for the purposes of targeting poverty programs, during 1999 BKKBN

were persuaded to distinguish two additional categories of families based

on particular ‘economic’ criteria. Consequently, from that year BKKBN

have been producing a separate classifi cation of those families in the two

lowest welfare categories, ‘pre-prosperous families’ (keluarga pra-sejahtera

or KPS) and ‘level 1 prosperous families’ (keluarga sejahtera 1 or KS-1)

for economic reasons. These two additional classifi cations are referred to

as KPS ALEK and KS-1 ALEK.14 The indicators used by village-based,

family planning cadres to determine which families fall into these categories

are as follows: any family will be classifi ed as KPS ALEK if it fails to meet

any one of the following criteria:

• All family members are usually able to eat at least twice a day;

• All family members have different sets of clothing for home, for work

or school, and for formal occasions;

• The largest section of the fl oor of the family home is not made of

earth; and

• Sick children are able to receive modern medical attention and women

of fertile age are able to access family planning services.

Any family will be classifi ed as KS-1 ALEK if it fails to meet any one

of the following criteria:

• At least once a week the family is able to eat meat, fi sh or eggs;

• Every family member has obtained at least one new set of clothes

during the past year;

• There is at least 8 m2 of fl oor space in the family home for every

member of the household; and

• All children between seven and 15 years of age are presently attending

school.

As we shall see in our account of specifi c programs, too close a dependence

upon this classifi cation has created some serious problems. However, in the

diffi cult circumstance of mid-1998 when many of the safety net programs

were being conceived in haste because of the prevailing sense of crisis –

fueled by some alarming reports in both the domestic and international press

that later proved to be unnecessarily alarmist and exaggerated – program

planners really had no other option but to fall back on the BKKBN data

as there was really no alternative listing at the village level of poverty or

welfare status.

By contrast, the National Socio-Economic Survey (SUSENAS) data on

consumption and expenditure, which has been widely used by both BPS and

independent scholars to calculate poverty rates and poverty lines, is only

available as part of a survey of a selected sample of the entire population.

The SUSENAS is carried out on an annual basis (usually in February

every year), and consists of a Core survey that is administered to over

200 000 households, and which is estimated to cover more than 800 000

individuals across the entire country. It collects data on a broad range of

socio-economic indicators, including education, health and employment as

well as consumption and expenditure. From this sample group, a subset of

65 000 households are surveyed on a certain topic or set of issues in much

greater detail. This is referred to as the SUSENAS Module. Once every three

years, as in 1999, the SUSENAS Module collects detailed information about

consumption and expenditure, and this is widely regarded as the best data

source for calculating poverty incidence as measured by consumption.

In addition, from time to time BPS uses the SUSENAS to collect data

on other matters in what are referred to as Special Modules. In 1999 the

SUSENAS included a Special Module on the government’s Social Safety

Net program. Consequently, the 1999 SUSENAS – both the Core survey

and the Special Module – has been an invaluable tool of analysis for

those observers who have monitored and analyzed the effectiveness of the

targeting and program performance. We consider the results of these studies

below. However, it is important to bear in mind some important limitations

of this data source.

The 1999 Special Module consisted of a set of special survey items

designed to test each respondent’s awareness of and participation in the

Social Safety Net programs (Badan Pusat Statistik, 1999). Yet some of

the questions were framed in a deliberately general manner, avoiding, for

example, reference to the offi cial names of specifi c programs that were not

likely to be widely known by village respondents. Yet there was still the

possibility of confusion about which actual programs or poverty alleviation

measures were being surveyed.

Another issue of particular importance is the precise period of time

being covered by the 1999 SUSENAS Special Module. Respondents were

surveyed in February about their knowledge of and participation in the

Social Safety Net programs during the preceding six months (that is from

August–September 1998). However, many programs did not properly get

underway until well after July 1998, and the actual implementation of some

measures, and the delivery of assistance to benefi ciaries, did not really

begin until early 1999. This means that since only the fi rst period of the

implementation of the safety net programs is covered by the survey, any

conclusions that extrapolate to the period after that date should be treated

with caution. No further questions about participation in these programs

were included within the annual SUSENAS until the 2002 round when

a further limited series of items were included as part of the SUSENAS

Core. This has enabled some comparison of performance across a wider

time frame and has been used in at least one of the most recent analytical

studies (Sparrow, 2003a).

Another data source, the 100 Village Survey conducted by the Central

Board of Statistics and UNICEF, also contained some questions about

Social Safety Net program participation in its October 1998 round and

several of the earliest studies assessing these programs were based on an

analysis of the results (Cameron, 2001 and Suryahadi et al., 1999). However,

this survey of 12 000 households in 100 relatively poor rural villages located

in 10 districts and 8 provinces was far more limited in its scope. Moreover,

the original selection of locations had not been designed to be statistically

representative of the entire country. As a result, the fi ndings from these

studies, although they gave some indication of important trends, are not as

reliable as the later studies that were based on the full 1999 SUSENAS.