Multidimensional poverty

The Planning Commission has recently published a report on Multidimensional Poverty in collaboration with the Oxford Poverty and Human Development Initiative and the UNDP in Pakistan. The primary source of data is the Pakistan Social and Living Standards Measurement Survey (PSLSMS) carried out periodically by the PBS.
The transition from income or basic needs approach to measurement of poverty on a multidimensional basis is an important step forward. It provides a comprehensive view of the sources of poverty and enables the development of a strategic framework of policies and programmes for alleviating poverty. The report also identifies the incidence of poverty at a regionally disaggregated level down to the districts of Pakistan.
The principal findings of the report are discussed below.
Incidence of poverty has apparently shown a strong declining trend
According to the report, multidimensional poverty has declined from a headcount of 55.2% of the population in 2004-05 to 38.8% in 2014-15. Despite fluctuations in the growth rate of the economy and variation in the degree of inclusiveness of the growth process, Pakistan has apparently been successful in bringing down poverty on a continuing basis, especially during the tenure of the last PPP government.
The fall in poverty is visible in each Province and in the urban and rural areas of the country. At the district level, only 11 districts out of the total of 115 districts in the country have witnessed an increase in the incidence of poverty.
Contradiction with other findings
However, the decline reported in incidence of multidimensional poverty in Pakistan is not consistent with other findings. The global Human Development Report of the UNDP gives estimates of multidimensional poverty in each country. According to the latest report of 2015, multidimensional poverty in Pakistan has increased from a headcount of 43.5% of the population in 2006-07 to 45.6% in 2012-13.However, it is reassuring to note that the poverty headcount ratio is lower in Pakistan than in India or Bangladesh.
The Social Policy and Development Centre (SPDC) has also undertaken research earlier on multidimensional poverty. Sophisticated statistical techniques have been used for measurement of poverty. The key finding is that the incidence of multidimensional poverty fell from 49.4% of the population in 2005 to 47.4% in 2009. It has, however, risen to 48.2% by 2011. What explains these contradictory findings?
Problems with choice of indicators
The primary reason for the difference in the findings on the trend of poverty in Pakistan is in the choice of indicators. The Planning Commission report uses 15 indicators to measure poverty. Three indicators relate to the level and access to education, four to health and eight to the standard of living.
The Global HDR of UNDP uses ten indicators. Two indicators relate to education, two to health and six to the standard of living. The big difference between the two approaches is in the health indicators. In its latest estimate for Pakistan, UNDP focuses on the incidence of malnutrition among women and children in a household and on the level of child mortality. The Planning Commission looks at access to health facilities, immunisation, ante-natal care and assisted delivery. Malnutrition is not included in the indicators, as the PSLSMS does not contain this information.
One-third of the combined weight in measurement of multidimensional poverty is attached to health indicators. The incidence of malnutrition among children has been increasing sharply in Pakistan according to the Nutrition Surveys carried out periodically. The number of children either stunted or wasted has risen rapidly. This is attributed primarily to the decline in food security of the people, especially in terms of affordability.
The exclusion of malnutrition in the Planning Commission set of poverty indicators introduces a significant bias in the findings. The rise in malnutrition clearly has had negative implications on the trend of poverty in the country, which is not captured in the estimates by the Planning Commission.
The SPDC study also does not include malnutrition as an indicator. However, the status of employment; either employed or unemployed, of the head of the household is incorporated in the indicators on the standard of living. This is an appropriate indicator of likely incidence of poverty in a household.
The unemployment rate in Pakistan has shown a fluctuating trend. It was relatively low up to 2009 and since then has been rising. It is not surprising that SPDC concludes that multidimensional poverty fell from 2005 to 2009 and has increased thereafter from 2009 to 2011. The Planning Commission report also does not focus on unemployment.
Overall, it appears that the choice of indicators by the Planning Commission for measuring multidimensional poverty understates not only the level of poverty but also tends to create a bias towards a declining trend, when the underlying reality may be rising or, more or less, unchanged incidence of poverty in Pakistan, especially in line with the decline in the last seven years in the growth rate of the economy.
The remaining part of the comments focus on the incidence of multidimensional poverty in 2014-15 and not on the trend since 2004-05.
Big difference between urban and rural poverty
According to the Planning Commission report, the incidence of multidimensional poverty in 2014-15 is only 9.4% in urban areas as compared to 54.6% in rural areas. This difference is much larger than that observed either in the case of income poverty or basic needs poverty. This is due to the differential access to basic health, education and economic services.
However, there are reasons why the incidence of multidimensional poverty may be understated in urban areas, given the approach adopted for measurement. First, there is the non-inclusion of an employment indicator. The unemployment rate in urban areas in 2014-15 was substantially higher at 8% as compared to 5% in rural areas.
Second, the issue is not only of access to a service but also to the quality of service provided. For example, over 85% of the households in the primate city of Pakistan, Karachi, have access to tap water. But, unfortunately, little water flows out of the taps, especially in lower income neighbourhoods. Similarly, the quality of service provided in government hospitals in district or tehsil headquarters is poor, with an acute shortage of beds and medicines. Consequently, lower income urban households have to seek the services of low quality private practitioners. Further, the level of power load shedding varies substantially among locations.
Therefore, the PSLSMS questionnaire needs to be expanded to also capture the quality of service provided and employment status, with the multidimensional poverty indicators being refined accordingly. This is likely to raise the level of urban multidimensional poverty and narrow the gap with rural poverty.
AJ&K has the lowest multidimensional poverty
An extremely surprising finding is that among the different regions, Azad Jammu and Kashmir has the lowest incidence of multidimensional poverty. It was 24.9 % as compared to the national average of 38.8%%. This is not consistent with the fact that unemployment rate in AJ&K is extremely high at over 14%, more than twice the national average in 2012-13. The next lowest incidence is in Punjab, followed by Sindh, Khyber Pakhtunkhwa and Balochistan. The extremely high incidence of poverty in FATA may be one of the root causes of the rise in extremism and terrorism in this part of country.
Lack of education contributes most to poverty
The combined contribution of education indicators to multidimensional poverty in Pakistan is the highest at 42.8% in 2014-15. This is not surprising given the relatively low literacy and primary enrolment rates, especially in the rural areas. Next in importance is the standard of living with a contribution of 31.6%. The smallest contribution is by the health related factors of 25.6%.
However, according to the global UNDP human development report the impact of health indicators is higher at 32.3% in 2012-13. This highlights again the importance of including a malnutrition indicator in the measurement of multidimensional poverty.
Intra-regional variation in poverty is most pronounced in Sindh
The difference in poverty among districts is most pronounced in Sindh, as measured by the difference in poverty incidence between the best and worst districts, respectively. Balochistan comes next in this indicator, followed by Punjab and Khyber Pakhtunkhwa. However, a more sophisticated measure of variation in poverty across districts in a province is required.
Also, given the small sample size on average at the district level in the PSLSMS, the reliability of estimates is low. For example, the district of Awaran in Balochistan had a head count of poverty of 84.3% in 2008-09. This fell sharply in two years to 58.9% in 2010-11, but rose to 93% in the next two years. Such large fluctuations in such short spans of time are unlikely. Further, changes in district boundaries by creation of new districts have to be allowed for.
The fall in poverty is not consistent with peoples' perceptions
The PSLSMS also includes questions on perceptions of respondent households about their economic situation. In each survey, over the period 2008-09 to 2014-15, almost 80% of the households have reported that their condition has either remained unchanged or worsened. Clearly, the response should have been more positive if poverty was falling rapidly.
Recommendations
The Planning Commission rightly recommends that the multidimensional poverty index (MPI) should be used in future to complement existing official measures of poverty. Also, the MPI may be used as a basis for resource allocation, especially at the district level.
The recommendation that MPI variables be included in the next Census questionnaire may not be feasible. However, the PSLSMS survey should be expanded both in terms of sample size and by inclusion of measures of employment and nutrition status. It is also necessary to promote greater involvement of local research institutions in this important exercise.
(The writer is Professor Emeritus and former Deputy Chairman of the Planning Commission)