Human Development Report

[Facilitator’s note: This query from UNDP Myanmar has been cross-posted on the PR-Net and HDRStats-Net. Please send your contributions to this query by 13 June 2007]

I version française I versión en español I

Dear Colleagues,

UNDP Myanmar is working to estimate a new poverty line for Myanmar. A household income and expenditure survey was conducted in 1997, on the basis of which the World Bank estimated Myanmar’s national poverty line in 1999. We are currently working to estimate an international poverty line for Myanmar using $1 a day and $2 a ay (in ppp dollars).

Therefore, we would appreciate if UNDP Country Offices, Regional Centres and HQ can share with us their experiences in estimating a poverty line, methodology, sample TORs, consultants and experts.

Best regards,

Hla Phyu Chit
UNDP Myanmar

---

Chers collègues,

Le PNUD Myanmar travaille à l’estimation d’une nouvelle ligne de pauvreté pour le Myanmar. Une enquête sur le revenu et les dépenses par foyer a été conduite en 1997 et ses résultats ont permis à la Banque Mondiale d’estimer la ligne de pauvreté nationale du Myanmar en 1999. Nous travaillons actuellement à l’estimation d’une ligne de pauvreté internationale pour le Myanmar utilisant 1$ par jour et 2$ par jour (ppa dollars).

Par conséquent, nous aimerions que des Bureaux de pays, des Centres Régionaux et le Siège du PNUD puissent nous faire part de leurs expériences en ce qui concerne l’estimation de lignes de pauvreté, la méthodologie, les termes de référence, des consultants et des experts.

Cordialement,

Hla Phyu Chit
PNUD Myanmar

---

Queridas y queridos colegas,

La Oficina del PNUD en Myanmar está trabajando para calcular una nueva línea de pobreza para Myanmar. Se condujo una encuesta de ingresos y gastos en hogares en 1997, sobre cuya base el Banco Mundial estimó en 1999 la línea de pobreza nacional. En estos momentos estamos trabajando para calcular una línea internacional de pobreza para Myanmar utilizando 1 dólar al día y 2 dólares al día (dólares ajustados en función de la paridad del poder adquisitivo).

Por lo tanto, nos gustaría pedir a las Oficinas de Pais del PNUD, Centros Regionales y unidades de la sede que compartan con nosotros sus experiencias a la hora de calcular líneas de pobreza, metodología, ejemplos de TORs, consultores y expertos.

Saludos,

Hla Phyu Chit
PNUD Myanmar

Dear Hla Phyu Chit,

Household surveys and estimation of poverty lines have been done in many countries, so you should not have too many problems finding related material. The $1 a day or $2 day are very telling figures for (international) advocacy (even in ppp), but are not always as relevant depending on the country.

If a poverty survey already has been done in 1997, I would think that comparability with the data from the survey is of the utmost importance for measuring progress towards the MDGs. So, why not use at least the definition used in 1997 in order to have comparable results over time? Also note that based on the same dataset, you can apply different definitions of the poverty line for the analysis. However, you should be very clear in explaining this in order to prevent confusion on percentages used.

I would like to share the experience of Congo where a household survey has been done in 2005 with results published in 2006. For more information you can visit http://www.cnsee.org/ (you can download it from the site). Annex 1 provides an explanation on how the poverty line was chosen in Congo based on the number of kCal per day – which subsequently can be transformed into a monetary value - and a non-food component (see extract below). I have also added an article on multidimensional poverty in Congo (in French).

One practical advice: please make sure that government, stakeholders and statisticians (both from the World Bank, consultants and other statistical offices) fully agree on the definition of the poverty line, the methods to calculate it and the weighting of the data. Two issues that have played in the Congo are 1) definition of strata and 2) analysis of data. The first is related to the definition of three main analysis groups: urban areas; semi-rural areas and rural areas. These groups have been used to define the strata and select households. At the end of the survey, politicians wanted to have data per department (province), however, it was very difficult to explain to them that in view of the selection method for the households, we could not provide statistically sound results per department. Therefore the needs of end-users should be very clear from the outset.

The second is related to the weighing of data and subsequent analysis. Although I do not have all the details, in essence Afristat and the World Bank disagreed on how to analyze the data. Unfortunately this only came evident after the report had been officially published, causing concerns about the validity of data. Therefore I would advice you to keep a very close eye on the statistical discussions and making sure that everybody is on board.

I have some contacts at Afristat with expertise in Poverty Household surveys, but they are all francophone. Please let me know if it is useful to send you their contact details.

Kind regards,

Micha van Lin
UNDP Congo
Brazzaville, République du Congo

[Facilitator’s note: This query from UNDP Myanmar has been cross-posted on PRNet and HDRStats-Net. Further contributions are welcome.]

Dear Hla Phyu Chit,

The $1 per day poverty line is appealing for advocacy purposes because it gives an indication of how much deprivation is out there in the world. Its use at the national level is very limited because the nature and characteristics of poverty differ from country to country. It would be more pertinent to establish a poverty line that allows for monitoring of poverty over time and space. In this regard, it may be useful to build on the work of the World Bank.

For a developing country like Myanmar, I would like to believe that the World Bank used the absolute poverty line, which is linked to a certain level of welfare within the context of Myanmar.

A common  problem with  poverty lines is the assumption that household resources are equitably distributed among members, which is not the case in real life situation. Complementing household expenditure survey data with other source of data and use of other poverty indicators are useful for gauging improvement in peoples well-being. 


Whatever approach you decide to use at the end of the day, Micha's point on stakeholder consultation is key.  A poverty line is useful only if it is acceptable to the people. Therefore ensuring that the process is as transparent as possible and that the consumption basket correspond to the consumption patterns of a majority of the population is very an important point tobear in mind.

For more reading on poverty lines refer to Martin Ravallion' s Poverty lines in Theory and Practice -  click here.

I hope this is useful,

Amie




-------------------------
Amie Gaye
Policy Specialist
HDRO
304 East 45th Strret
FF12102
Ny 10017

Edited: 06/04/2007 at 12:06 PM by HDR - Measurement Sub-Network Moderator

Dear colleagues in Myanmar (and elsewhere),

In response to your request allow me to share with you the methodology we used in the 2003 Poverty Assessment in Zimbabwe. I have also attached the CV of a highly qualified and experienced consultant we used to assist the national team of experts in the derivation of poverty lines. While I understand the query is mostly focusing on estimating an international poverty line, I believe it may be useful to first compare different methodologies for determining your national poverty line(s). Your first step anyway would be to come up with a poverty line expressed in the national currency which is most relevant if you want to use it as a basis for national or local policy decisions aimed at poverty reduction.

Expenditure-based approach 

Let me start by mentioning that the main objective of 2003 Poverty Assessment Study Survey (PASS) in Zimbabwe was to measure both income poverty and human poverty. The survey itself was carried out by a national project team in the Ministry of Public Service, Labour and Social Welfare. Core funding and technical assistance was provided mainly by UNDP and in the end stage also by the World Bank. Additional contributions came from UNICEF, UNFPA and UNIFEM.  

In terms of methodology, consumption expenditure was used as a proxy for income, mainly because household incomes were considered underreported to a large extent, hence unreliable. A national food basket was derived by using consumption expenditure data based on actual consumption patterns of the 40% poorest households (2nd to 5th deciles) at the time of the survey. Expenditures were taken as reported by households since this was considered more accurate than using the price data collected from local shops. The latter commodity price series were incomplete and unreliable as official Government-determined prices were collected in a situation whereby a substantial number of food items were in short supply or even unavailable in the shops during the survey. In practice, households therefore had to purchase most commodities on the informal (black) market at much higher prices.

This situation was particular to Zimbabwe at the time (and still is), but it goes to show that you always have to take the country-specific context into account when making choices in the methodology to use. In terms of methodology, the food items from household purchases, own production, and public and private transfers to households were weighted by expenditure shares and quantities. The first 30 items with the highest weighted expenditure formed the food basket. The expenditure-based approach adopted for calculation of the food basket recognises the preferences i.e. actual eating habits of the poor and was therefore preferred over selecting a ‘least-cost food basket’.

Food Poverty Line and Total Consumption Poverty Line

At national, provincial and district level a Food Poverty Line (FPL) was derived based on the food basket taking into account the WHO/FAO recommended minimum nutritional requirements of 2,100 kilocalories per person per day for an average active person. Subsequently, a Total Consumption Poverty Line (TCPL) was calculated using the so-called traditional method of scaling up, i.e. determining the average level of total expenditure of households whose food expenditure was equal to the FPL. The traditional method was used, because it is likely to include the basic non-food needs of the poor. While this method can be argued to commit a higher error of inclusion of some non-basic items, from a planning point of view it was considered preferable to slightly over-estimate than to under-estimate requirements.

For a more detailed description of the methodology please see the attached doc.

Hope this is useful.

Best regards,

Alwin Nijholt
Trust Fund Manager
Joint M&E Programme
UNDP Malawi
Tel:  + 265 (0) 1 773 500 ext 246
Fax: + 265 (0) 1 773 637

Dear Colleagues,
 
Without any irony, I am a great fan of international comparability. Alwin's description of the methodology highlights the difficulty to achieve this for one single poverty indicator based on income measured in dollars. You are all aware of the weaknesses of this indicator, and yet I bet that most of you would insist to keep it.
 
However, it is worthwhile to look at the other extreme: Batteries of relatively weak indicators that yield a) a lot of detail and b) overall indices that are relatively reliable because the weaknesses of individual indicators compensate each other.
 
Here is one example for that technique that I consider very advanced:
 
The Tanzania Districts Dashboard shows how Poverty and Human Development Report 2005 indicators could be used to assess the overall performance of districts.
The Regional Development Index is composed of 3 sub-indices (in brackets: no. of indicators): Economy (7), Social services (20), Environment (6). The database contains 33 indicators for 120 districts.

To see the full story, right-click twice into "Regional Development Index" in the lower left pie chart, then left-click in any of the white indicator legends on the right.
 
Now the question to you is obvious: If you have a given amount of human and financial resources dedicated to "development statistics", how would you distribute it over a) a precise poverty line and b) a not so precise but detailed indicator system?
 
Best regards,
Jochen

European Commission Joint Research Centre
Italy


This mail is strictly personal and does not in any way affect the European Commission
Visit the Dashboard of Sustainability

-------------------------
[L=http://esl.jrc.it/dc/mdg_unsd/examples.htm]MDG Dashboard[/L]: a user-friendly look at the MDG indicators

Dear Colleagues,
 
I would like to contribute to this discussion by proposing an interesting methodology  which can be used for your poverty assessment. It is aimed at measuring the level of satisfaction of basic needs.  This methodology can be used  for national or local purposes and it is affordable. You may also appreciate the comparaisons made of number of  methodologies/ instruments used for measuring poverty.

The paper is in French and was presented at the AfrEA (Africa Evaluation Nework) workshop held in Niamey - Niger in January 2007.

Hope it will help. I would also appreciate any comments on the paper.
Thanks and best regards.

Theo. 

UNDP Niger

-------------------------
Theodore Mpatswenumugabo
Senior Economist
UNDP Niamey/Niger

[Facilitator’s note: This query from UNDP Myanmar has been cross-posted on PRNet and HDRStats-Net. Please find below responses from Bob Baulch, University of Sussex, David Satterthwaite, International Institute for Environment and Development and Serdar Bayriyev, UNDP Turkmenistan. Further contributions are welcome.]

Bob Baulch, University of Sussex,wrote:

Dear All

I very much agree with Jochen's comments. One additional factor to consider in the case of Myanmar, and some other (usually transition or post-conflict) countries, will be whether adequate purchasing power parity (PPP) factors exist for the country in the World Bank's ICP (formerly the Penn World) tables. As I've learned when working on poverty issues in Vietnam, non-existent or problematic PPPs conversion factors can fundamentally undermine the value of the WB's $1 and $2/day international poverty lines (which are, of course, expressed in PPP terms).

A couple of years ago, the ADB's Research Department came up with an interesting alternative to the WB's $1/day metric, which I don't think has received another attention. This is to examine the basis of how the national poverty lines in different countries are estimated, and then to compare these if the methodologies used for constructing them is similar. In many countries in Asia, for example, the Cost of Basic Needs (CBN) approach, with a daily per capita calorie requirement of 2100 Kcals/person/day plus a modest allowance for non-food expenditures, is used to estimate the national poverty line. (In India, and 1 or 2 more Asian countries, a slightly higher daily calorie requirement is used- -but the same CBN method is employed to estimating national poverty lines). This approach could provide an alternative for making international comparisons across groups of developing countries, in particular in Asia, although comparisons with other countries (in particular those, such as in Europe where relative poverty lines based on half or 40% of median per captia income are used) would not be possible. See the first chapter in the ADB's 2004 Key Indicators publication for more on this. http://www.adb.org/Documents/Books/Key_Indicators/2004/default.asp

Last year, Don Sillers from USAID put together a very useful, non-technical, overview on 'National and International Poverty Lines'. The two background papers by Chen and Ravallion on the World Bank's Povcalnet website (which does not include Myanmar) are also worth reading (http://iresearch.worldbank.org/PovcalNet/jsp/index.jsp)

Hope this helps a bit.

Bob Baulch
Fellow, Institute of Development Studies
University of Sussex, Brighton, UK

---

David Satterthwaite, International Institute for Environment and Development, wrote:

How robust (or even tested empirically) is the assumption that a household that is able to meet all its food requirements is very likely to be able to meet basic non-food requirements? Having worked in informal settlements in many different cities over the last 30 years and seeing the high prices paid for poor quality accommodation, water, use of toilets, health care, keeping kids at school, transport..... it is likely that large sections of the population that can afford sufficient food still lack the income to allow them to allow many non-food deprivations. has anyone who has done a large representative household survey in a city ever gone to a group of households identified as not being poor because they spend enough on food to avoid food poverty to see if they also avoid other deprivations?

David Satterthwaite
International Institute for Environment and Development
U.K.

---

Serdar Bayriyev, UNDP Turkmenistan, wrote:

Dear Hla Phyu Chit:

With reference to your query, I would strongly recommend to review a wealth of materials available on the World Bank Poverty Net: click here

Here are some of the Poverty Net resources related to poverty line measurement:

1. Poverty Manual - click here

2. Poverty Lines - click here

3. Poverty levels and trends - click here

4. Poverty comparisons: A Guide to Concepts and Methods - click here

5. Poverty Lines in Theory and Practice - click here

6. Poverty measurement and analysis - click here

Best regards,

Serdar Bayriyev
UNDP Turkmenistan

[Facilitator’s note: This query from UNDP Myanmar has been cross-posted on PRNet and HDRStats-Net.]

Dear Colleagues in Myanmar,

I found the work "The Compendium of best practices in poverty measurement (CBPPM)" undertaken by the Rio group very useful for a new poverty assessment, in particular if your goal is to estimate internationally comparable poverty level. Hope you will also find this study useful.

The Compendium is available on-line as a PDF file at: http://www.ibge.gov.br/poverty/pdf/rio_group_compendium.pdf

PAHO message of September 2006: "This Compendium sums up the result of a long-term effort in compiling and organizing poverty measurement practices around the world. The work of the Rio Group has been supported by the participation of 22 countries and 18 regional or international institutions, which have been represented at one or more of the seven meetings that took place between 1997 and 2004”.

The Expert Group on Poverty Statistics was created in 1996 by the United Nations Statistical Commission, with the purpose of identifying the best practices in poverty measurement and making them available for the international community, contributing to the progress towards a higher standardization of processes. It is chaired by the IBGE, and ECLAC serves as its technical secretariat.

The Compendium is organized according to the three major approaches for poverty measurement: poverty lines, unmet basic needs or deprivation indicators, and the integration of both approaches. It has an introductory chapter written by Professor Peter Townsend, and its final chapter covers topics such as the link between poverty measurement and public policy, the conditions for the international comparability of results, and possible strategies for the improvement of information sources.

The Compendium focused on good practices that had the following characteristics:

(a)     Have a clear definition of the relevant standard and its units of measurement.(b)     Are based on an existing source of information that meets minimum quality standards.
(c)     Have been applied more than once, ideally for the same country or region. If, however, calculations have been carried out only once but for many countries, they may still qualify.
(d)     Produce information that is a useful input for public policies or that is related to aspects falling within the purview of those policies."

Regards,

D. Nergui
UNDP Mongolia

[Facilitator’s note: This query from UNDP Myanmar has been cross-posted on PRNet and HDRStats-Net.]

Hi Colleagues in Myanmar,

Apologies for this late contribution to your question on poverty lines but I have been on mission in Solomon Islands calculating national poverty lines from a recent household survey, so your question is very topical.

It is not clear from your original question whether you intend to “start-from-scratch” to estimate a new poverty line or whether you simply wish to update the earlier World Bank work to the present. If the former then you would really need to have a new household income and expenditure survey and this would require consideration of all the methodological issues raised by other contributors. It would also require considerable time and effort to design and conduct the household survey, and then to process and analyse the data.

On the other hand if you wish to update the 1997 WB data then it seems to me that you have a similar problem in your task to that which we have in the Pacific: no country-specific PPP indicator. Neither Myanmar nor Solomon Islands (or any other Pacific Island countries) are covered by the World Bank’s ICP programme which is the “official” calculator of PPP conversion factors. Thus it is not possible to calculate fully comparable US$1 or US$2 per day poverty indicators for Myanmar. That is essentially why there is no current PPP US$1 per day MDG 1 poverty indicator for Myanmar, (or for the Pacific Island countries either for that matter).

There are two alternatives however: Firstly it would be possible to use the CIA estimate of the PPP to get a reasonable estimate of the US$1 per day, the CIA PPP factors are generally similar but are not always the same as those used by WB; for example the WB PPP conversion factor for Philippines is 12.8 for 2005 while the CIA factor for 2006 is about 13.3. (This means WB estimates that the 2005 average PPP exchange rate to the US$ was Ps12.8 compared with the official market rate of around PS55.09, while CIA puts the 2006 figure at Ps13.3 compared to market are of 51.2). Does UNDP have a policy on using any information from the CIA, I wonder?

Anyway, if you wanted to pursue this avenue, for Myanmar the CIA data suggests that the 2005 PPP conversion factor was 0.663 (i.e. the PPP exchange rate is 0.663 compared with the official rate of 5.761 - as quoted by CIA). I am not familiar with the exchange rate situation in Myanmar so you will have to be the judge of the accuracy of the CIA official market rate. Using this conversion factor would enable you to adjust the 1997 national poverty line figures to a PPP value and/or to the current year.

To do this you would need firstly to extrapolate the 1997 national poverty line and individual household incomes (assuming that a) you wished to estimate the 2006 poverty indicator, and b) you have access to the survey data or a detailed breakdown of income by deciles) by the Myanmar CPI (broken into food and non-food components if possible depending on how the 1997 PL was calculated); this further assumes that the pattern of expenditure will not have changed markedly over the period.

The national poverty lines and individual incomes are then adjusted by the PPP factor to bring the PL to the adjusted PPP value at the new date. By comparing the adjusted US$1 (actually US$1.08 in 1993 prices and around US$1.50 now) and US$2 per day PPP equivalents you could estimate the current poverty incidence.

The second alternative, and one that we are looking at for Pacific Island countries is to take the national poverty lines and compare them with a neighbouring country’s poverty lines where the neighbour is in the ICP. Since most national poverty lines are based on a broad basket of goods to give a food energy intake of around 2100 kcal per day this can become a “numeraire” and the link between the ICP country P/L and the non-ICP P/L. taking the ratio between the Myanmar P/L and say Vietnam, Cambodia or Laos would thus enable you to make an estimate of the likely PPP factor for Myanmar and the US$1 poverty indicator.

Hope this is helpful.

David

David Abbott
Pacific Regional Macro Economic & Poverty Reduction Advisor, UNDP 
Suva, Fiji Islands
http://regionalcentrepacific.undp.org.fj