Issues

Comments

1. Too stringent menu (80% nutritional adequacy for other vitamins and minerals)

The Philippines need not worry on this issue

2. Use of a cheaper commodity with the same caloric requirements (e.g., example on the 2 sets of menu using cassava and rice)

For as long as the price data show that cassava is cheaper in one region compared to rice, it will be all right to use/include this in the bundle

3. Obtaining the consumption pattern of a reference group, nationally (say, families within the 25-35% bandwidth)

Families from some of the “richer” regions might not fall within the bandwidth. Bandwidth may be adjusted

In the official methodology, items in the menu are chosen not just because they are eaten in the area but more of they are cheap/low cost

Iteration and validation processes should be done by the NSCB, whenever the poverty incidence generated does not fall within the bandwidth initially set

4. Tests of revealed preferences

Issue on comparability across space, specifically on the menus of the “poorer” and “richer” regions

In the rapid appraisal during the forum, there seemed to be no preference for either of the menus (Bicol region, which is representative of the “poorer” region, and the National Capital region (NCR), which is representative of the “richer” region)

One participant indicated her preference of the Bicol menu over NCR while Dr. Ravallion preferred NCR’s menu to Bicol. The other participants were ambivalent on this issue

Suggestion to do revealed preference test

On the Issue on Comparability Over Time

5. Options in addressing this issue

1. Taking the poverty line in one reference period and update it using price index

2. Recalculating the poverty lines, which is more robust to survey comparability

Option no. 1 is almost similar to what is being done in the official methodology

6. Bank’s methodology

Computed by getting the consumption behavior of those who can either just attain or just afford the food poverty line

Similar with the official methodology since it is getting the expenditure pattern of the families within the +/-10 percentile of the food threshold

On income versus expenditure

7. Considerations

Quality of income and expenditure data from the Family Income and Expenditures Survey (FIES)

• Income: 220 data cells
• Expenditure: 1100 data cells

Possible problem in capturing income data from self-employment activities, informal sector, etc.

Problems in distinguishing and/or capturing consumption from expenditure data

8. Dr. Ravallion’s general comment on the issues on the Philippines ’ official poverty estimation methodology

There is quite a narrow range of issues in the official poverty estimation methodology in the Philippines compared to other countries. Dr. Ravallion added that there is no one perfect method in estimating poverty

- on the question on which is more important: changes in rankings of provinces or the trends in poverty incidence

Would depend on the needs of the government’s intervention programs and projects

10. Validation process

In the case of Indonesia , consistency of the results using the official and the Bank’s methodologies were checked using Spearman’s rank correlation coefficients

Philippines can do similar exercises on correlations to check the robustness of the poverty estimates (including the rankings)

11. Variation of food shares (FE/TBE ratio) across regions

Suggestion to review data

12. FIES sample size

41,000 total sample families are relatively better compared to other countries in the Region