NSCB - Technical Notes on the 2003 Poverty Estimates

Poverty statistics for 2003 were computed based on the poverty estimation methodology approved by the NSCB Executive Board in January 2003 with some modifications necessitated by the change in the design of the 2003 Family Income and Expenditure Survey (FIES), the main source of income and expenditure data used for poverty estimation. The FIES is a regular survey conducted by the National Statistics Office (NSO) every three years.

Prior to release, the 2003 poverty estimates and the methodology used to derive them underwent an open and rigorous review and evaluation through a series of meetings conducted by the Technical Committee on Poverty Statistics (TC-PovStat) from October 2004 to January 2005. The TC-PovStat has a multisectoral representation consisting of noted experts in the area of poverty measurement and is mandated to recommend improvements in the poverty estimation methodology, among other tasks.

In the interest of releasing reliable, accurate and consistent statistics, the NSCB Executive Board approved the recommendations of the TC-PovStat regarding the generation and release of the 2003 poverty estimates, which include, among others, some minor modifications on the methodology/practice previously approved/used. These are discussed in detail in Section B. Section A below provides an overview of the existing official methodology, while Section C cites the reasons for the revisions done on the 2000 estimates.

A. Existing official methodology (as approved on 15 January 2003)¹

The methodology starts with the computation of the food threshold, which refers to the minimum cost of the food items that satisfy nutritional requirements for economically necessary and socially desirable physical activities. The nutritional requirements are determined by the Food and Nutrition Research Institute (FNRI). Currently, these are based on 100 percent adequacy for the recommended dietary allowance (RDA) for protein and energy equivalent to an average of 2,000 kilocalories per capita, and 80 percent adequacy for the other nutrients. These nutritional requirements found in the 1989 RDA for Filipinos for Energy and Specific Nutrients² were computed by the FNRI based on the age and sex population structure of each region using the results of the 1990 Census of Population and Housing. The food threshold is also sometimes referred to as the subsistence threshold or the food poverty line.

The provincial food thresholds, which are computed separately for urban and rural areas, are determined by using regional menus (with separate menus for urban and rural areas) and provincial prices in accordance with NSCB Resolution No. 1, Series of 2003. For food items in the menu that come in different varieties, e.g., rice, which could be ordinary, special, NFA rice, etc., and hence, with different prices, the price of the cheapest variety available in the province is used in the computation of the provincial food threshold. For this reason, the price of special rice is not used since it will violate the minimum income/expenditure criterion. On the other hand, while NFA rice seems to be the most appropriate to use, it being the lowest-priced commodity, the issue on accessibility, especially to the poor arises. Thus taking into account these concerns on the use of special and NFA rice, the price of ordinary rice is used as a compromise between the high-priced special rice and the low-priced but possibly inaccessible NFA rice.

The expenditure pattern of households within the 10-percentile band around the provincial urban/rural food threshold as indicated by the ratio of food expenditure (FE) to total basic expenditure (TBE), or the FE/TBE ratio, is then used to derive the provincial urban/rural poverty threshold.

The poverty threshold or poverty line, also computed separately for urban and rural areas, refers to the cost of the basic food and non-food requirements. The basic non-food requirements cover the non-food expenditure items of the total basic expenditures. Thus, the poverty line may be viewed as the minimum income required to be able to buy the basic food requirements and other non-food basic needs.

The food and poverty thresholds are then used to estimate the subsistence and poverty incidence, respectively, using the income approach. The subsistence incidence refers to the proportion of families (or population) with per capita income less than the per capita food threshold to the total number of families (population). On the other hand, the poverty incidence refers to the proportion of families (or population) with per capita income less than the per capita poverty threshold to the total number of families (population).

A. summary of the features of the official existing methodology is given in Table 1.

Table 1: Features of the existing official methodology

Aspect	Feature
Menu	Region-based
Price	Provincial prices
Rice	Ordinary rice
FE/TBE ratio	Changing FE/TBE ratio (by province and by urban-rural disaggregation)
Disaggregation	National, regional, provincial
	Urban and rural

B. Modifications adopted for the 2003 poverty statistics

1. Adoption of a single regional menu for each province across years

Under the existing poverty estimation methodology, region-based menus have to be used for the computation of provincial food thresholds. However, the regional menus were prepared based on the 1988 regional configuration and many provinces have since been transferred to other regions. The question thus arose as to which regional menus to adopt for these provinces. In the interest of comparability of estimates across years, it was recommended that the same regional menu be adopted for 1997, 2000, and 2003. For instance, while Aurora has been transferred to Region III, the Region IV menu was still adopted in estimating the 2003 poverty statistics for the province since the Region IV menu has been used since 1997. Nonetheless, regional estimates of food/poverty threshold/incidence for 2003 (and also for the revised 2000 estimates) shall cover all provinces belonging to their respective regions in 2003, that is, using the 2003 regional configuration. Hence, the 2003 poverty incidence estimate for Region III includes the poor in Aurora, although the Region IV menu was used in computing the poverty threshold for this province.

A summary of the menus that were adopted for provinces that have shifted regions since 1988 is given in Table 2.

Table 2: Comparison of regional classification and menus used for food threshold estimation for provinces that have shifted regions since 1988

Province	Region in 1988	Regional menu adopted in the 1997, 2000 and 2003 poverty estimates	Present region
Aurora	4	4	3
Lanao del Norte	12	12	10
Saranggani	11 (as part of South Cotabato)	11	12
South Cotabato	11	11	12
Zamboanga Sibugay 1	9 (as part of Zamboanga del Sur)	9 (as part of Zamboanga del Sur)	9
City of Isabela 2	9 (as part of Basilan)	9 (as part of Basilan)	9
Compostela Valley 1	11 (as part of Davao Oriental)	11 (as part of Davao Oriental)	11
Cotabato City	12 (as part of Maguindanao)	12 (as part of Maguindanao)	12
Abra	1	1	CAR
Apayao	2	1	CAR
Benguet	1	1	CAR
Ifugao	2	1	CAR
Kalinga	2	1	CAR
Mt. Province	1	1	CAR
Lanao del Sur	12	12	ARMM
Maguindanao	12	12	ARMM
Sulu	9	12	ARMM
Tawi-tawi	9	12	ARMM
Basilan	9	9	ARMM
Agusan del Norte	10	10	CARAGA
Agusan del Sur	10	10	CARAGA
Surigao del Norte	10	10	CARAGA
Surigao del Sur	11	10	CARAGA

1 – Only 2003 estimates were generated for Zamboanga Sibugay and Compostela Valley since these are new provinces created under EO 36 and EO 103.
2 – Only 2003 estimates are available for Isabela City since separate estimate was generated only after its separation from its mother province.

2. Adoption of alternative estimation procedures for the five provinces without sample urban barangays

The new master sample design for the 2003 FIES did not use the urban-rural classification as an explicit stratification variable. This meant that the sample barangays would not necessarily include both urban and rural barangays, as what happened with five provinces, namely, Abra, Batanes, Marinduque, Siquijor, and Southern Leyte. Thus, separate estimates for urban and rural areas could not be generated for these five provinces and consequently, the urban-rural disaggregated regional poverty estimates (food/poverty thresholds and incidences) of the regions to which each of these five provinces are classified, i.e., CAR, Region II, Region IV-B, Region VII, and Region VIII, respectively, do not include the five provinces. In addition, the urban-rural disaggregated national estimate of poverty also excludes these five provinces. Estimates, therefore, on the five provinces are reflected only on the overall (that is, urban and rural areas combined) regional and national estimates.

Also, due to the absence of sample households coming from urban barangays, the weights (i.e., proportion of urban and rural food poor to the total food poor in the province) used to derive the overall provincial food threshold could no longer be applied for these five provinces. Computational exercises were done for the 2000 data in these provinces wherein provincial food thresholds were obtained using these weights and an alternative set of weights (proportion of urban and rural population to total population in the province as recommended by the NSCB). Overall, minimal differences were observed in the estimates resulting from the application of the two sets of weights. Thus, the TC-PovStat approved the weighting scheme recommended by the NSCB for the 2003 estimates.

As for the provincial poverty thresholds of these five provinces, these were derived directly from the overall provincial food thresholds (i.e., by dividing the provincial food threshold by the ratio of the food expenditure over total basic expenditure of each province) instead of getting the weighted averages of the provincial urban poverty thresholds and the provincial rural poverty thresholds, which was the case for provinces with samples coming from both the urban and rural areas.

3. Release of only the national and regional-level estimates for 2003

Provincial poverty estimates were computed based on the existing methodology and these were presented to the TC-PovStat in January 2005. It was observed that 12 provinces experienced declines in poverty incidence ranging from 10 to 20 percentage points, while 2 provinces had their poverty incidence falling by more than 20 percentage points. On the other hand, 7 provinces were found to exhibit an increase in poverty incidence by more than 10 percentage points. Thus, a thorough validation needs to be done to ensure the reliability of the provincial estimates. For this reason, upon the recommendation of the TC-PovStat, the NSCB Executive Board approved the modifications on the methodology and the release of only the national- and regional-level poverty estimates for the meantime to address the immediate demand for the data. Final estimates with provincial-level disaggregation are targeted for release in the first quarter of 2006.

Tables 3.a and 3.b show the coefficients of variation (CVs) of the regional poverty incidence estimates. The CV is a widely used measure of precision of estimates. Some theoreticians consider 10 percent as an acceptable level for the CV, but among official statisticians, no agreement has been reached on the issue and enlightened users have used statistics with CVs higher than 10 percent. The corresponding margins of error of the estimates are presented in Tables 3.c and 3.d.

Table 3.a: CVs of regional poverty incidence estimates, 2003

Region	CV (%)	Region	CV (%)
PHILIPPINES	1.3	Region VII	5.6
NCR	9.3	Region VIII	4.6
Region I	6	Region IX	4.6
Region II	5.8	Region X	5.3
Region III	6.5	Region XI	5.9
Region IV-A	6	Region XII	5.4
Region IV-B	4.7	CAR	6.9
Region V	3.7	ARMM	5.4
Region VI	4.2	Caraga	4.3

Table 3.b: Distribution of CVs of regional poverty incidence estimates, 2003

CV (%)	No. of Regions
0 – 2	0
>2 – 4	1
>4 – 6	13
>6 – 8	2
>8 – 10	1
Total	17

Table 3.c: Margins of error (MEs) of regional poverty incidence estimates, 2003

Region	ME	Region	ME
PHILIPPINES	0.65	Region VII	2.58
NCR	0.9	Region VIII	3.24
Region I	2.84	Region IX	3.98
Region II	2.2	Region X	3.96
Region III	1.75	Region XI	3.24
Region IV-A	1.73	Region XII	3.4
Region IV-B	3.66	CAR	3.35
Region V	2.91	ARMM	4.83
Region VI	2.57	Caraga	3.95

Table 3.d: Distribution of MEs of regional poverty incidence estimates, 2003

ME	No. of Regions
0 – 1.0	1
>1.0 – 2.0	2
>2.0 – 3.0	5
>3.0 – 4.0	8
>4.0	1
Total	17

C. Revision of the 2000 poverty estimates

Revision of the previously released 2000 poverty estimates was done to adopt the revised/improved methodology for estimating missing prices of pork liver and chicken liver, the actual prices of which are available only for 1994. Before, prices of these food items were estimated by inflating 1994 prices using the CPI for meat. The methodology was revised such that current year’s prices are estimated by using the growth rate (from 1994 to current year) of pork meat and chicken meat to inflate the 1994 prices of pork liver and chicken liver, respectively. Minor revisions were also done to correct for computational errors (e.g., the weights used for Masbate, wrong pick-up of prices of certain food items like rice, that is, special rice instead of ordinary rice, etc.).

Table 4 shows the coverage of each region following the current regional configuration.

Table 4: Regional configuration used in the revised 2000 and 2003 poverty estimates

¹The details of the existing poverty estimation methodology are presented in the technical notes released in July 2003 concerning the official poverty statistics (http://www.nscb.gov.ph/technotes/poverty_tech.asp).

²The dietary standards in the RDA have been revised with the issuance of the 2002 RENI or Recommended Energy and Nutrient Intakes, also prepared by the FNRI.

³The use of a menu in the derivation of the official poverty line is sometimes misunderstood. The menu is just an artifice to derive the cost of the nutritional requirements. It makes it easier to appreciate the fact that the “food bundle” that satisfies the nutritional requirements is in fact, palatable to and/or fit for consumption by human beings. It is not meant to be prescriptive, but the “food bundle” is chosen to satisfy criteria such as cheap, locally available, and if possible, typical of the local eating practices.


E-mail the webmaster	Terms of Use	Home • Top of Page
	1997-2008, National Statistical Coordination Board Makati City, Philippines