For the Record
On The Refinements on the Official Poverty Estimation Methodology,
The Sources of Differences of the Official Poverty Statistics and the
National Household Targeting System for Poverty Reduction Estimates,
and Other Official Poverty Statistics-Related Concerns
This article seeks to clarify some of the questions and misconceptions on the official poverty statistics released by the National Statistical Coordination Board (NSCB).
I. The Refinements on the Official Poverty Estimation Methodology.
We, official statisticians are guided by international guidelines, principles, and best practices in the performance of our duties. Most of these guidelines are adopted by the United Nations Statistical Commission (UNSC). Examples are the Fundamental Principles of Official Statistics (FPOS) 1 , the 1993/2008 System of National Accounts, and the Principles and Recommendations for Population and Housing Censuses Revision 2.
Principle No. 2 of the Fundamental Principles of Official Statistics adopted by the UNSC in April 1994 states “To retain trust in official statistics, the statistical agencies need to decide according to strictly professional considerations, including scientific principles and professional ethics, on the methods and procedures for the collection, processing, storage and presentation of statistical data”. Principle No. 3 on Accountability and Transparency states “To facilitate a correct interpretation of the data, the statistical agencies are to present information according to scientific standards on the sources, methods and procedures of the statistics”.
We would like to assure our public, our stakeholders, that we, in the NSCB abide by these principles.
Why Revise the Official Poverty Estimation Methodology?
One of the standard practices of official statisticians is the revision of methodologies2 towards better statistical measurement and the revision of estimates when new data support becomes available. Examples of applications of these practices are:
Treatment of Expenditures on Research and Development as part of Capital Formation and therefore included in the Gross Domestic Product instead of as Intermediate Inputs
(1993/2008 SNA, Paragraph 6.163/6.230 which revised the 1968 UNSNA, Paragraph 633) – effect is to increase the level of GDP, but not necessarily the growth rate of GDP.
Revision of Quarterly National Accounts estimates when complete/additional data for the reference quarter become available. Preliminary estimates for some sectors are generally based on incomplete data for the quarter ( for example, data for the first two months of the quarter instead of data for all the three months) – effect is either a downward or upward revision of both level and growth rate of GDP.
The revision of the poverty estimation methodology to improve the monitoring of poverty in the country is in line with this standard practice.
Since the first official methodology for poverty assessment in the Philippines was adopted in 1987, three major revisions have been made – in 1992 3, 2003/2005/2006 4, and 20115. The 1992 refinement addressed the possible overestimation of the nonfood component of the poverty threshold. The 2003/2005/2006 refinement addressed the demand for provincial level (versus regional level) disaggregation of poverty statistics. The 2011 refinement addressed the need to improve the comparability of the estimates across space and over time.
It may be recalled that when the methodology was changed in 1992, the NSCB was criticized for “statistical perfidy”, with the government accused of succeeding in reducing poverty by changing the definition of the poor. In 2003/2005/2006, for some reason, the criticisms were whittled down. When the 2011 refinement was released to the public on February 8, 2011, there was no violent reaction. Now, more than 9 months after its release, new and recycled criticisms have surfaced about the refined methodology.
In line with FPOS No. 3 on Accountability and Transparency, and as directed by the NSCB Executive Board when it approved the refined methodology on February 1, 2011, the NEDA Development Information Staff 6 and a Consultant 7 helped the NSCB Technical Staff craft a communication strategy to enhance public appreciation and understanding of the refined methodology. As a result, at least twenty-one presentations on the refinement and the estimates have been made by the NSCB Technical Staff in various fora 8 .
The 2011 Refinement
The 2011 Refinement of the Official Poverty Estimation Methodology was the result of serious and dedicated efforts of the Technical Committee on Poverty Statistics (TCPovStats)9 created by the NSCB Executive Board in 2003. It is composed of professionals/experts from the academic and research community, the data users, the data producers, and the private sector. It is chaired by Dr. Celia M. Reyes, Senior Research Fellow of the Philippine Institute for Development Studies (PIDS), and currently President of the Philippine Economic Society.
The TCPovStats followed a deliberate process of reviewing the methodology that included 10 the Project on the Comprehensive Documentation and Analysis of Issues on the Official Poverty Estimation Methodology of the Philippines jointly implemented by the PIDS and the NSCB Technical Staff with funding from the UNDP, the creation of four Small Working Groups 11, and the conduct of a Workshop and a Consultative Forum before the TCPovStats came up with its final recommendations.
Thus, the refinement of the methodology was not done solely by the NSCB Technical Staff.
But the NSCB Technical Staff serves as the Secretariat of the TCPovStats. In the course of our work on poverty monitoring, we have presented technical papers and learned from the experiences of other official statisticians in various international fora such as the sessions of the United Nations Statistical Commission and the International Statistical Institute and various regional and international workshops. We also access the references/materials of the United Nations Rio de Janeiro City Group12 on Poverty Statistics13, which is a group of experts, primarily from national statistical agencies, working on the identification of indicators, methodologies and statistical sources on matters of poverty measurement, common procedures, and best practices. We would like to assure our stakeholders that when we attend an international gathering of statistical professionals, we bring home with us new knowledge and skills that we, after careful thought, apply in our work at the NSCB. We also invite our stakeholders to talk with and be enlightened by professionals who have had wide experiences in the actual compilation of official poverty statistics in other countries.
One of the important considerations in the refinement process was that the refined methodology should be implemented at the beginning of a new administration (or the end of the previous administration) to avoid any misappreciation of the motives behind the refinement. It was also agreed that the review of the methodology should be undertaken by the TCPovStats every ten years. Essentially, this implies that the methodology would be used over two administrations or for four FIES years ( 2009, 2012, 2015, and 2018 for the current methodology).
The TCPovStats recommended refinement was presented to and approved by the NSCB Executive Board on February 1, 2011 through NSCB Resolution No. 9 Series of 2011, Approving the Refinements in the Official Poverty Estimation Methodology
Highlights of the refinements of the official poverty estimation methodology presented during the Press Con are provided through the following link: http://www.nscb.gov.ph/poverty/2009/Presentation_LBersales.pdf. A Statistically Speaking article, entitled “Refinements in Measuring Poverty”, was also published to laymanize the refinements made in the official poverty estimation methodology. This article can be accessed through the following link: http://www.nscb.gov.ph/headlines/StatsSpeak/2011/021411_rav_joe_bbb.asp
In the official poverty estimation methodology, the “definition” of the poor has not changed: the poor are those whose income is not sufficient to pay for the basic (minimum) requirements. Basic requirements are those for food ( the food threshold) and for nonfood (the nonfood threshold). The food threshold and the nonfood threshold added together comprise the poverty threshold. The refinements are in terms of improving the way the requirements are quantified or given monetary values.
In quantifying the food requirements, the following criteria were used:
The food bundle must satisfy the nutrient requirements: 100% of the Recommended Energy and Nutrient Intakes (RENI) requirement for energy ( 2000 kilocalories on the average) and 80% of the RENI requirement for other nutrients.
The food items in the bundle must be locally available and are eaten in the area (in the province)
The food items in the bundle must be the cheapest available locally (in the province)
The food bundle must be visualizable ( can be cooked and eaten)
It must be stressed also that the requirements are based on an average per capita, using the population distribution of the country. This means that it does not specifically cater to requirements of babies or of senior citizens.
Also, the resulting “menu” is only an artifice to put monetary value to the food requirements. It is definitely not prescriptive.
Moreover, it may be mentioned that suggestions had been made in the past that the nutrient requirements be limited to the 100% adequacy for energy, i. e. to do away with the 80% adequacy for other nutrients. These suggestions were premised on the belief that the nutrient requirements were too stringent for poverty monitoring. However, during the series of consultative workshops conducted in 2000 and 2003 to discuss the then proposed refinements on the methodology, the participants generally did not consider the requirement of 100% adequacy for energy and 80% adequacy for other nutrients too stringent. The FNRI also pushed for the retention of these requirements 14. Thus, the food requirement of 100% adequacy for energy and 80% adequacy for other nutrients was retained.
The important features of the refinements are as follows:
Estimation of the Food Threshold
The use of a national food bundle as starting point – to enhance comparability across space
The national food bundle was converted to a provincial food bundle (the old methodology used a regional bundle) using the four criteria listed above. A summary of the nutritional adequacy of the provincial food bundles is shown in Annex 4, which indicates that all the provincial food bundles satisfy 100 % adequacy for energy and 80% adequacy for other nutrients. Annex 4 also shows an example where the use of a provincial food bundle (Bataan) instead of a regional food bundle (Region III) results in a cheaper cost without sacrificing the nutritional adequacy of the food bundle.
To further enhance comparability across space and consistency of the food bundles, they were subjected to the Test of Revealed Preferences16 as recommended by Dr. Martin Ravallion, an eminent poverty expert.
Estimation of the Poverty Threshold
As is done in most countries, the nonfood requirement is indirectly estimated. In fact, we do not know of any country where the official poverty statistics directly estimate the nonfood requirement.
To enhance comparability across time, the ratio of food expenditures to total basic expenditures is held constant, estimated using four FIES years (2000, 2003, 2006, and 2009)
On the average, the food thresholds and the poverty thresholds are higher by 8% and 13%, respectively, using the old methodology as compared to the refined methodology. Consequently, the subsistence incidence and the poverty incidence as well as the count of the poor using the old methodology are generally higher compared to those obtained using the refined methodology
However, the trends of the subsistence incidence and the poverty incidence are the same, whether using the old or the refined methodology.
The official poverty statistics of the NSCB placed the number of poor families in 2009 at 3,855,730 which is 1.4 million families or 26.6% lower compared to the DSWD NHTSPR estimate of 5,255,118 in 2009-2010. The sources of these differences are :
The NSCB used the 2009 Poverty Threshold obtained from the refined methodology which came up with a national annual per capital poverty threshold of PhP 16,841. The DSWD used the 2006 Poverty Threshold obtained from the old official methodology of PhP 15,057. Note that the DSWD threshold is lower than the NSCB threshold but the DSWD count of the poor is higher than the NSCB count. If the DSWD used the threshold for 2006 using the refined methodology, it should have been PhP 13, 348 instead of PhP 15,057 and the DSWD count would have been lower than 5.26 million.
Reference Period – The NSCB estimate used calendar year 2009 ( based on the FIES) while the DSWD used March 2009 – January 2010.
III. Other Official Poverty Statistics-Related Concerns
The NSCB/PSS, in our continuing effort to be more relevant to our stakeholders, is committed to move forward and address other pressing and valid issues on poverty statistics with the help of our development partners (the World Bank, UNDP, among others). These efforts are geared towards the following:
With the availability of the microdata of the 2007 PopCen and the merged 2009 FIES-LFS in the 2nd semester of 2011, the NSCB has started the generation of 2009 city/municipal poverty estimates. This is scheduled to be released in March 2012.
In the generation of the 2006 and 2009 poverty statistics, FIES microdata were made available 13 months after the reference year. To decrease the time lag of the release of the results of the FIES 19, which is the source of income and expenditure data for the generation of official poverty statistics, the NSO has undertaken a study on the possibility of conducting the FIES in one round only (instead of two).
For the 2009 estimates, the NSCB released the official poverty statistics less than a month after the provision of the required FIES microdata. If plans on the conduct of the FIES with only one visit push through, and/or if the FIES microdata will be released by the NSO earlier than its current schedule, the NSCB will be able to release the poverty statistics earlier.
An integrated review of selected surveys is being undertaken to explore the possibility of generating annual poverty statistics, among others. This is included in the directions of the Technical Committees on Poverty Statistics and Survey Design for 2012.
Finally, we would like to clarify some points raised by Dr. Mahar Mangahas during his presentations in the 1st Monitoring and Evaluation Network Forum organized by the National Economic and Development Authority (NEDA) and the United Nations Children’s Fund (UNICEF) – Office in Manila on 7 November 2011 in Crowne Plaza Manila Galleria in Quezon City and the Meeting of the House Committee on Poverty Alleviation on 29 November 2011. Some points raised in his presentation were published in his article in the Philippine Daily Inquirer on 2 December 2011.
1. “My comments consisted, firstly, of affirmation of my October 5 memorandum to Rep. Mendoza (see my Inquirer column "No meat allowed for the poor," October 8, 2011), showing how NSCB downgraded the quality of its food-poverty menu: no more milk for children; no more pork adobo (actually, no meat at all); only one ulam, not two, for lunch; rice and vegetables should be boiled, not fried nor guisado; only one banana now, not two; no more margarine for the pandesal.”
Apparently, NSCB's aim is to adjust the poverty boundary to the least cost of providing nutritional requirements that it can discover. To me, this treats the poor not as people entitled to minimum satisfaction from food, but like work animals to be efficiently fed to maintain their productivity.”
Thus, the provincial bundles developed are based on evidence - empirical results that have been validated through surveys conducted by the NSO and BAS.
In 2005, interim modifications in the official estimation methodology for the 2003 poverty statistics were provided by NSCB Resolution No. 5, Series of 200525. During this time, the provincial food bundles were still being developed and subjected to the Test of Revealed Preference to ensure comparability across space. Hence, the 2005 Resolution still provided for the use of a regional food bundle – a single regional food bundle to be used across years for each province within the region. Further, the 2005 Resolution addressed the estimation of poverty statistics for the five provinces with no sample urban barangays26 (under the 2003 master sample used by the NSO). The NSCB then released the 2000 and 2003 national and regional poverty statistics.
In 2006, NSCB Resolution No. 11, Series of 200627 addressed concerns on the release of provincial-level poverty statistics, in view of the design limitations of the 2003 FIES which follows the 2003 master sample. Official provincial poverty statistics for 2000 and 2003 were released by the NSCB Technical Staff in accordance with the 2006 Resolution.
2. “I pointed out that the NSCB's failure to do specific research on non-food needs makes its poverty line totally insensitive to price changes in other basics like water, electricity, shelter, transportation etc. (see my "The poor don't live by bread alone," March 5, 2011).”
The NSCB Technical Staff has recognized long time ago that the indirect estimation of the nonfood requirement essentially assumed the same rate of price changes in the food and the nonfood requirement. This is of course not generally true. One alternative is to directly estimate the nonfood requirement. As earlier mentioned, this is not done by official statisticians.
The issue on the direct estimation of non-food component was reviewed and deliberated by the TC-PovStat and its Small Working Group on the Estimation of Poverty Threshold, chaired by Dr. Arsenio S. Balisacan. The Committee and the SWG felt that unlike for food where the experts of the FNRI could be tapped to define the standard nutrient requirements, there would be difficulty in setting non-food standards on education, health, housing, fuel, transportation cost, etc. without being arbitrary and which will be generally acceptable. Dr. Balisacan et. al.30 cited as example the type of materials used for a dwelling unit such as crude sawali walls, which would seem decent in poor areas, whereas the more expensive lawanit or plywood would be judged adequate in the more progressive areas. Setting non-food standard is a big challenge, not only to the Philippines but also to most countries worldwide. Dr. Martin Ravallion31, an eminent poverty expert, explained that one might write down a bundle of non-food goods; but it is unclear whether a fixed bundle of non-food goods would gain wide acceptance, or maintain its relevance over time, such as with rising average standards of living. He added that of all the data that go into measuring poverty, setting the non-food component of the poverty line is probably the most contentious. Hence, it is common practice among countries to use indirect estimation for the non-food component. Aside from the Philippines, the following countries also use the indirect method for measuring the non-food component of the poverty threshold: Bangladesh, Thailand, Jordan, Oman, etc. The TC PovStat thus recommended to retain this approach in the refined methodology.
3. “It showed NSCB's own slides depicting the Philippines as fourth-poorest after Laos, Myanmar and Cambodia with its new poverty line, but only second-poorest with its old poverty line. Thus NSCB fosters complacency by diverting attention to an ASEAN comparison. (At the hearing, Rep. Rufus Rodriguez remarked, correctly, that what NSCB has done is simply window-dressing.)”
4. “Another original NSCB slide went anti-worker by proclaiming, using its new poverty line, that a minimum-wage worker in Metro Manila can now support a family of six.”
5. “At the NEDA-UNICEF forum, I called for rejection of the NSCB"refinement" because it reduced the evidence on poverty, which is against the principle of evidence-based policy making.”
6. “Sensibly, the [DSWD] has decided that the NHTS-PR will stick to its present target list of 5.2 million households, and not let its programming be roadblocked by NSCB's reduction of the poverty line.”
We have long recognized that while the 2009 official count of the magnitude of poor families is 3.9 million, the 5.2 million identified by the NHTS-PR includes vulnerable families/population with income slightly above the poverty threshold.
Based on special computations made by the NSCB, these are families whose 2009 incomes are approximately not more than 17 percent above the poverty threshold. To illustrate, the daily income of these vulnerable families with five members did not exceed PhP 270 while the official daily poverty threshold for a family of five in 2009 is PhP 231… just a difference of PhP 39.
In fact, we believe that the government must continue to help not only the poor but also the vulnerable families – those whose income may be above the poverty threshold but not much so and therefore can easily fall back to poverty in times of crises.
Finally, one can examine the self-rated poverty statistics of SWS and note the following33:
1. The self-rated poverty threshold fluctuates wildly:
2. The self-rated poverty threshold did not change for a long period of time:
3. Similarly, the self-rated poverty incidence for the Philippines fluctuates wildly from 74% in July 1985 to 43% in March 1987 to 66% in September 1988.
Is it believable/possible/valid that over a short time, such measures as poverty threshold and poverty incidence could fluctuate as wildly as the SWS data show? Or that the poverty threshold does not rise with inflation? As we have always said, the SWS type of data have their own uses; they can be very useful to politicians; but what kind of policies can NAPC or DSWD or DA or DepEd craft using these statistics?
For inquiries, please contact Dir. Jessamyn O. Encarnacion or Ms. Mildred B. Addawe of the NSCB Social Statistics Office and NSCB Social Sectors B Division at telephone numbered (632) 896-1730 / (632) 896-7981 or through the e-mail addresses firstname.lastname@example.org/ email@example.com.
1 The Fundamental Principles of Official Statistics is currently being reviewed by a Friends of the Chair (FOC) group headed by Katherine Wallman of the United States of America. Dr. Romulo A. Virola, Secretary General of the National Statistical Coordination Board of the Philippines is a member of the FOC. The other members are Jil Matheson, UK; Enrico Giovannini, Italy; Sibylle von Oppeln-Bronikowski, Germany, Jorge Bustamante, Colombia; Pali Lehohla, South Africa; Brian Pink, Australia; Eduardo Sojo Garza-Aldape, Mexico; Oscar E. Mangula, Tanzania; Pablo Tactuk, Dominican Republic; Olav Ljones, Norway; Marie Bohata, Eurostat; and Lidia Bratanova, UN Economic Commission for Europe.
3 http://www.nscb.gov.ph/resolutions/1992/10.asp, which was used for the computation of the following: 1) 1991 official poverty statistics; 2) revised 195 and 1988 national and regional poverty statistics.
4 NSCB Resolution No. 1, Series of 2003 - used for the computation of the first-ever official provincial poverty statistics covering the years 1997 and 2000. NSCB Resolution No. 5, Series of 2005 - used for the computation of the 2003 national and regional poverty statistics as well as back estimates for 2000. NSCB Resolution No. 11, Series of 2006 - used for the computation of the following: 1) 2003 provincial poverty statistics; 2) revised 2003 national and regional estimates; and 3) revised 2000 national, regional, and provincial poverty statistics
5 http://www.nscb.gov.ph/resolutions/2011/9.asp , which was used for the computation of the following: 1) 2009 provincial poverty statistics; 2) revised 2003 and 2006 provincial poverty statistics; and 3) 1991 regional poverty statistics, for purposes of monitoring the MDGs.
6 Under Director Victoria V. Quimbo
7 Ms. Maria Mercedes F. Robles, Managing Director, Corporate Communications of the University of Asia and the Pacific
11 SWG on the Estimation of Food Threshold; SWG on the Estimation of Poverty Threshold; SWG on the Estimation of Incidence and Other Measures of Poverty; and SWG on the Improvement of the Relevance of Official Poverty Statistics.
12 City groups are created by the United Nations to work on selected statistical methods. The City Group is usually named after the city where the first meeting was held. The UN recognizes that these informal consultation groups are an innovative way to improve and speed up the international standards development process. Some other existing City Groups are the London Group on Environmental Accounting, Washington Group on Disability Statistics, and Delhi Group on Informal Sector Statistics. (Reference: http://unstats.un.org/unsd/methods/citygroup/index.htm)
14 As cited by Dr. Jocelyn Juguan during her presentation of the Highlights of the SWG on the Nutritional Requirements for the Estimation of Food Poverty Line, on 13 April 2010, according to Dr. Corazon Barba, former FNRI Director, “food intakes below the RDA for energy or iron may compromise an individual’s capacity for production or work productivity and school performance and productivity among schoolchildren”.
19 Prof. Solita “Mareng Winnie” Collas-Monsod suggested during the Consultative Workshop on the Formulation of the Philippine Statistical Development Program (PSDP) 2011-2017 on 12 December 2011 to consider a reference period other than January to December (e.g., October to September) to ensure that the Philippine Statistical System (PSS) is able to release official poverty statistics before the elections and therefore provide guidance to voters during the elections.
20 The efforts to refine the poverty methodology benefited from the expertise of Dr. Rodolfo Florentino and Dr. Corazon Barba, both retired Directors of the FNRI, Dr. Regina Pedro, former FNRI Division Chief and Dr. Jocelyn Juguan, currently Senior Science Research Specialist of the FNRI.
21 Castro, Lina V., Addawe, Mildred B., and Agtarap, Kristine Faith S. Assessing Poverty Lines by Revealed Preference. 10th National Convention on Statistics. National Statistical Coordination Board. October 2010. (http://www.nscb.gov.ph/ncs/10thNCS/papers/invited%20papers/ips-01/ips01-03.pdf) For urban areas, adjustment/revision of food bundles resulted in 76 out of the 80 food bundles completely passing the consistency test. The five failed tests came from the menus of Camarines Norte (1), South Cotabato (2), Basilan (1), and Lanao del Sur (1). With a target index of 1.0 or greater to pass the Test of Revealed Preference, indexes for these five provinces were less than 1.0 but at least 0.9862. For rural areas, adjustment/revision of menus resulted in 61 out of the 80 menus completely passing the consistency test. It must be stressed, however, that the 32 failed tests have indexes greater than 0.98, except for the use of Cebu food bundle for Basilan that resulted in an index of 0.9652 which is still greater than 0.95.
22 Food and Nutrition Research Institute and National Statistical Coordination Board. Development of Provincial Low-Cost Nutritionally Adequate Menus Terminal Report. October 2001.
23 The Recommended Energy and Nutrient Intake (RENI) used to be referred to as RDA.
26 Abra, Batanes, Marinduque, Siquijor, and Southern Leyte.
29 Virola, Romulo A., Asra, Abuzar and David, Isidoro P. 1997.Poverty Assessment in the Philippines and Indonesia: A Methodological Comparison. Number 44, Vol. XXIV Journal of Philippine Development.
30 Balisacan, A., R. Alonzo, T. Monsod, G. Ducanes, and J. Esguerra (1998), Conceptual Framework for the Development of an Integrated Poverty Monitoring and Indicator System. This paper was prepared in support of the project Strengthening Institutional Mechanisms for the Convergence of Poverty Alleviation Efforts.
31 Ravallion, Martin (1998), "Poverty Lines in Theory and Practice," Living Standards Measurement Study Working Paper 133, World Bank, Washington DC.
32 Minimum wage earner in Caraga and ARMM employed in the agriculture and non-agriculture sector. For Caraga: Rate is effective 20 June 2008, per Wage order No. 9 (Source of data: http://www.nwpc.dole.gov.ph/pages/statistics/stat_wage%20rates1989-present_non-agri.html ) For ARMM: Rate is effective 29 June 2008, per Wage order No. 11 (Source of data: http://www.nwpc.dole.gov.ph/pages/statistics/stat_wage%20rates1989-present_non-agri.html ) For this exercise, number of working days considered in a month is 26. Gross family income was used.
34 Used CPI for all items for areas outside Metro Manila
Posted: 13 December 2010.