1 Assessment of district performance in making progress towards MDGs in Bangladesh
Assessment of district performance in making progress towards MDGs in Bangladesh
Carel de Rooy1 and Siping Wang2 Overview Properly planned and conducted household surveys are the most reliable mechanism to assess progress regarding achievement of Millennium Development Goals (MDGs) in countries where the routine availability of development outcome or impact information is not readily available. UNICEF has been supporting the Government of Bangladesh (GoB) since the year 2000 to undertake such surveys. Their results have been made available to GoB as well as Development Partner institutions and organizations for planning, and prioritizing investment and action. A sizeable share of data that has enabled the understanding of progress regarding achievement of MDGs has come from these surveys. Since the year 2000 three such surveys’ data have been made available (2000, 2003 and 2006)3. All these were undertaken with a household sample size of approximately 60,000 and had a geographical resolution to the district level, allowing the comparison between the 64 districts of Bangladesh in each of these years. Approximately 20 indicators covering most MDGs were used in each survey. Unfortunately somewhat different indicators were used every time allowing for comparison over time of only approximately one third of the indicators used. This brief paper seeks to make a trend analysis over the 2000 to 2006 period using eight indicators that could be compared over time. Data for the analysis The Child Risk Measure (CRM) is a composite index. It comprises 8 indicators which have data by district for the years 2000, 2003 and 2006. The 8 indicators are:
• Infant mortality rate (IMR), • Proportion of births not attended by skilled health personnel, • Proportion of children 6-59 months without supplementation of vitamin A, • Proportion of households without consuming iodized salt, • Proportion of households without access to an improved water source, • Proportion of households without access to an adequate sanitation facility, • Proportion of primary school age children not attending school, and; • Proportion of children under 5 without a birth registration.
1 UNICEF Representative, Bangladesh 2 Chief Planning, Monitoring and Evaluation Section, UNICEF Bangladesh 3 BBS and UNICEF, 2000: "Progotir Pathey 2000: Achieving the Goals for Children in Bangladesh", Dhaka, Bangladesh; BBS and UNICEF, 2003: "Progotir Pathey 2003: on the road to progress", Dhaka, Bangladesh BBS and UNICEF, 2007: "Bangladesh Multiple Indicator Cluster Survey, 2006, Final Report, Dhaka, Bangladesh, BBS and UNICEF
2 Assessment of district performance in making progress towards MDGs in Bangladesh
Source of data Except for IMR, data for 7 indicators are from the MICS 2000, MICS 2003, and MICS 2006. The data for IMR are from the Bangladesh Annual Vital Registration Sample Survey. Methodology of computation The index of each indicator for each district is calculated as a relative deviation from the national average. A district with a negative value means that it has a comparatively lower risk than a district with a positive value. The CRM is the weighted average of the index of each of the 8 indicators. Each index is the standard deviation of a given district value from the national average. The weight is determined based on a conceptual framework (see the figure on the next page). The IMR is given a weight of 4. The proportion of births not attended by skilled health personnel and proportion of children 6-59 months without supplementation of vitamin A are given a weight of 3. A weight of 2 is allocated to the following indicators: proportion of households without consuming iodized salt; proportion of households without access to an improved water source; proportion of households without access to an adequate sanitation facility; proportion of primary school age children not attending school. The proportion of children under 5 without a birth registration is given a weight of 1. CRM maps For the years 2000, 2003 and 2006 a color code was given to child risk related ranges for each index. Red was allocated to represent districts where children are most at risk, here represented by the value of a given index for districts being higher than the upper limit of the standard deviation from the national average of that index. Blue was used to represent districts where children are exposed to the relatively lowest risks, represented by index values lower than the lower limit of standard deviation from the national average of that index. Yellow represents districts where children are exposed to relatively medium risks and is characteristic of index values between the lower limit and upper limits described above. Maps depicting CRM trends Additionally, color coding has been used to show evolution or changes over time in the CRM ranking. This was done by showing the difference between data from 2003 and 2000; 2006 and 2003, as well as 2006 and 2000. Red represents a drop in CRM ranking. Pink depicts no progress, at high risk in CRM rank. Yellow represents no progress at medium risk in CRM rank. Blue shows districts that sustained a low risk CRM rank. Brown identifies those districts that have evolved from high to medium risk in CRM rank. Green shows progress from medium to low risk in CRM rank. Rapid Assessment of District Performance Determinants For further analysis UNICEF sought to understand through a rapid assessment (see Annex 1), the most important positive determinants that might explain what makes districts such as Jhenaidah, Munshigani, Meherpur, Dhaka, Narail and Khulna perform well, so that these can be supported, promoted and enhanced elsewhere. Likewise it sought to be equally important to acquire an understanding of the negative determinants that should be overcome, avoided and neutralized to allow districts such as Bandarban, Cox’s Bazaar, Sherpur, Rangamati and Jamalpur make accelerated progress towards MDGs.
3 Assessment of district performance in making progress towards MDGs in Bangladesh
The rapid assessment, undertaken by UNICEF field staff over a period of a three days, revealed that for several indicators more favourable results emerged for the “Low Performance Districts” (>10% difference):
• Pupil-teacher ratio • Population-doctor ratio • Population-health personnel ratio • % population affected by major natural disasters • Frequency of turn-over in key district level posts • Per-capita expenditure of the MoHFW • Average NGOs per district
Other indicators did not show any relevant difference between “Low Performance Districts” and “High Performance Districts” (<10% difference):
• Average population per district • Frequency of turn-over in key UNO level posts • Frequency of turn-over in key Upazila level posts • INGOs • Bi-lateral donors • Multi-lateral donors
The only indicators found that might explain the difference in performance between “Low Performance Districts” and “High Performance Districts” were poverty and geographic isolation:
• % of Poor population (20% higher in low performance districts)4 • % of Unions not seasonally accessible (two and one half more disfavourable for the
low performance districts) Conclusion When comparing data from 2000 with that of 2006, twelve districts substantially declined while nine districts improved in their CRM ranking. Of the 9 originally classified in 2000 as relatively high-risk districts 4 actually evolved into the medium risk category. In contrast seven districts out of the 13 originally classified as relatively low-risk dropped into the medium-risk category. Roughly 20% of the districts in the medium-risk category moved either into the high-risk or low-risk categories. Out of 15 variables assessed to attempt to explain the difference in performance between the two categories of districts, only two variables emerged: poverty and geographic isolation.
4 World Bank, Bangladesh Bureau of Statistics and World Food Programme, 2009 : "Updating Poverty Maps of Bangladesh: Key Findings, 2005", Dhaka, Bangladesh
4 Assessment of district performance in making progress towards MDGs in Bangladesh
The current drive of the MoHFW to promote the establishment of community clinics seems very well placed in the context of the above related findings. If these clinics are:
• Well equipped; • Provided with a regular supply of high quality medicines; • Sustainably resourced with qualified human resources; • Targeted upon the areas of the country that are seasonally not accessible and; • Focused upon issues where inequity of access or outcome are greatest .......
this strategy could have an impact upon under-five, new-born and maternal mortality reduction. A recent study published in The Lancet5 implicitly suggested that economic determinants have a lot of weight in explaining health outcomes in Bangladesh. This also implies that rapid economic growth tends to shadow other determinants of health (and possibly other development) outcomes. The above mentioned rapid assessment substantiates this finding. In 2009 UNICEF supported the GoB through its Bangladesh Bureau of Statistics to conduct yet another MICS. This time however 300,000 households were surveyed allowing for an enhanced geographic resolution down to the upazila (sub-district) level. The results of this survey will be launched in November 2009. They are potentially an important baseline for the current government, both its administration and the country’s elected officials. Similar surveys – with geographic resolution at sub-district level - will be conducted in 2012 and 2015 allowing the government to continuously assess progress towards the achievement of MDGs. Subsequent ranking of districts will be made possible to recognize, acknowledge and better understand those that have made most progress. Most importantly, this approach will be replicated for sub-districts as well. This will, with other management tools emerging for fine-tuning of social sector investment decisions6, facilitate the prioritization of investment decisions by the Government of Bangladesh and its development partners alike. It will also allow for the undertaking of remedial action for the least performing upazilas and districts so that MDGs can be achieved with equity. Finally, it is interesting to note that the rapid assessment suggests that over one third of the population in both categories of districts assessed has been affected by major natural disasters. This finding, although likely to be an over-estimation and therefore it requires substantiation, however calls for a much more proactive approach to address emergencies. Instead of being reactive the government, with support of its development partners, should enhance its investment in resilience building to minimize impact of natural disasters and allow communities and families to rapidly bounce back to normalcy one the critical phase of natural disasters has passed.
5 “Effect of the Integrated Management of Childhood Illness strategy on childhood mortality and nutrition in a rural area in Bangladesh: a cluster randomized trial” Shams E Arifeen et al. August 2009. 6 World Bank, WHO, UNFPA and UNICEF supported “Marginal Budgeting for Bottlenecks (MBB)”
5 Assessment of district performance in making progress towards MDGs in Bangladesh
Conceptual Framework of the Child Risk Measure The methodology of computation of the child risk measure (CRM), a composite index, is: CRM = 4 * RD1 + 3 * (RD2 + RD3) + 2 * (RD4 + RD5 + RD6 + RD7) + RD8 where RDij is the relative deviation of the indicator Rij, RDij = (Rij – ARi) / SDi, i = 1, 2, … 8; j = 1, 2, … 64 Rij is the indicator i for the district j;
• R1j: Infant mortality rate (IMR), • R2j: Proportion of births not attended by skilled health personnel, • R3j: Proportion of children 6-59 months without supplementation of vitamin A, • R4j: Proportion of households without consuming iodized salt, • R5j: Proportion of households without access to an improved water source, • R6j: Proportion of households without access to an adequate sanitation facility, • R7j: Proportion of primary school age children not attending school, and; • R8j: Proportion of children under 5 without a birth registration.
64
ARi is the average of 64 districts for the indicator i, ARij = ∑ (Rij / 64) j=1 64
SDi = standard deviation of districts for the indicator i. SDi = {∑[(Rij – ARi)2 / (64-1)] }1/2 j=1
Infant Mortality
Households with no access to an
improved drinking water source
Households with no access to adequate sanitation facility
Households not using iodized
salt
School-age children not
attending school
Births not registered
Manifestation
Immediate factor
Underlying factors
Process factor
Births not attended by skilled health personnel
Vitamin A not supplemented to
children 6-59 months
6 Assessment of district performance in making progress towards MDGs in Bangladesh
Annual Child Risk Measure (CRM) Figure 1. CRM 2000 Figure 2. CRM 2003
Figure 3. CRM 2006
Index -23.3 - -10.3
-10.2 - 10.2
10.3 - 24.3
Index -18.4 - -9.5 -9.4 - 9.3 9.4.- 18.0
Index -15.4 - -9.6 -9.5 - 9.5 9.6 - 24.3
7 Assessment of district performance in making progress towards MDGs in Bangladesh
Trends in Child Risk Measure (CRM)
Figure 4. Change from 2000 to 2003 Figure 5. Change from 2003 to 2006
Figure 6. Change from 2000 to 2006
Code Declined in CRM ranking No progress, static at high risk of CRM rank
No progress, static at medium risk of CRM rank
Sustained at low risk of CRM rank
Improved from high to medium risk of CRM rank
Improved from medium to low risk of CRM rank
8 Assessment of district performance in making progress towards MDGs in Bangladesh
Distribution of Districts According to the Child Risk Measure
2006
-25
-20
-15
-10
-5
0
5
10
15
20
25
10 districts, 16% 10 districts, 16%
44 Districts, 68%
Low riskMedium riskHigh risk
9 Assessment of district performance in making progress towards MDGs in Bangladesh
Complete Data: Child Risk Measure Ranking Annual Data Trend Data Districts 2000 2003 2006 2003-2000 2006-2003 2006-2000 Cox''s Bazar 18.0 21.1 17.7 Bandarban 17.4 20.6 13.5 Sherpur 16.5 2.7 22.1 Rangamati 15.4 -4.6 18.1 Jamalpur 14.7 13.7 10.8 Brahmanbaria 14.5 9.4 4.6 Panchgarh 12.4 -0.5 -6.2 Sunamganj 11.9 24.3 7.1 Noakhali 10.8 6.8 5.0 Sylhet 9.2 -12.5 -0.8 Sirajganj 9.2 4.9 11.8 Netrokona 9.0 7.8 16.3 Kishoreganj 8.5 13.3 14.9 Khagrachhari 8.5 1.7 7.5 Bhola 8.0 2.8 7.1 Nilphamari 7.3 11.0 -1.7 Rangpur 7.2 7.5 -0.7 Sariatpur 6.1 10.8 6.7 Joypurhat 6.0 -12.8 -1.3 Kurigram 5.6 6.8 0.3 Baherhat 5.4 0.6 -2.0 Naogaon 5.4 -5.0 1.9 Pabna 5.3 6.2 5.0 Thakurgaon 4.6 12.2 7.0 Habiganj 3.7 10.5 16.8 Gopalganj 2.0 -6.7 -0.6 Manikganj 1.9 -11.0 1.6 Chittagong 0.9 -10.8 -1.8 Madaripur 0.6 0.9 -0.2 Moulvi Bazar 0.1 7.2 -6.0 Bogra -0.1 -6.1 -3.9 Narsingdi -0.2 3.2 -6.9 Mymensingh -0.5 9.2 17.1 Patuakali -0.6 6.7 6.5 Barisal -0.7 0.2 -13.9 Laxmipur -1.5 7.7 -1.4 Chandpur -2.0 -15.4 -6.7 Gaibandha -2.1 3.1 6.1 Nawabganj -2.3 -2.5 8.3 Lalmonirhat -4.1 4.7 2.0 Rajbari -4.4 -4.4 2.4 Comilla -4.7 -9.9 -11.5 Pirojpur -4.9 -9.6 -2.2 Faridpur -5.1 -6.5 -2.7 Satkhira -6.0 -2.5 5.3 Dinajpur -6.0 -3.1 3.2 Feni -6.2 -2.9 -16.7 Chaudanga -6.9 -5.9 -13.5 Natore -7.3 -4.4 -2.9 Tangail -8.5 4.6 7.4 Rajshahi -9.0 -11.3 -7.3 Jessore -10.0 -14.0 -9.0 Kushtia -10.2 -10.5 -9.5 Jhenaidah -10.2 -9.1 -13.9 Narayanganj -10.3 -6.9 -6.1 Meherpur -10.3 -13.0 -23.3 Gazipur -10.4 1.9 -3.1 Munshiganj -10.7 -14.6 -22.7 Dhaka -11.1 -15.1 -18.6 Khulna -14.0 -13.4 -13.3 Narail -14.4 -1.3 -12.9 Barguna -17.3 0.0 -6.4 Magura -17.9 -1.2 -8.6 Jhalkathi -18.4 3.3 3.9 Average 0.0 0.0 0.0 Stdev 9.3 9.5 10.2 Lower limit -9.4 -9.5 -10.2 Upper limit 9.3 9.5 10.2
Declined in CRM ranking
No progress, static at high risk of CRM rank
No progress, static at medium risk of CRM rank
Sustained at low risk of CRM rank
Improved from high to medium risk of CRM rank
Improved from medium to low risk of CRM rank
High Risk in CRM ranking
Medium Risk in CRM ranking
Low Risk in CRM ranking
10 Assessment of district performance in making progress towards MDGs in Bangladesh
Acknowledgements All UNICEF sections: Health & Nutrition, Water and Environmental Sanitation, Child Protection, Education, Planning, Monitoring & Evaluation and Field Operations participated in development of this brief paper. Field Operations were instrumental in very rapidly collecting data for the Rapid Assessment of District Performance Indicators. Annex 1. Questionnaire for Rapid Assessment of District Performance Determinants 1. Basic Information:
Indicator Total Total population of the district (2006) (source: DC/CS Office) Number of Unions Number of Unions seasonally not accessible by road or boat Number of Primary Schools (include both Government and registered non- Government schools)
Total number of primary school teachers (2000/2003/2006, absolute number corresponding to year )
Total number of primary school students (2000/2003/2006, absolute number corresponding to year )
Number of Health Facilities (hospitals and clinics, meaning: District Hospital, UHCs, H&FWCs, Health Sub-centres)
Total number of doctors (2000/2003/2006, absolute number corresponding to year)
Total number of medical assistants (2000/2003/2006, absolute number corresponding to year)
Total number of nurses (2000/2003/2006, absolute number corresponding to year)
Number of population affected by major natural disasters from 2000 to 2006 Names of major natural disasters from 2000 to 2006 ( Cyclone or Flood year wise )
2. Governance 2.1 Frequency of turn-over of key government officials at district level
Key District Posts Number of persons on the post from 2000 to 2006
Remarks (if any)
Deputy Commissioner Civil Surgeon District Primary Education Officer Executive Engineer DPHE Deputy Director Local Government Deputy Director Social Service Total
11 Assessment of district performance in making progress towards MDGs in Bangladesh
2.2 Frequency of turn-over of key government officials at upazila level No of post in the district Number of
persons on the posts
from 2000 to 2006
Estimated number of months that posts
remained vacant between 2000 to
2006
Key Upazila Posts
Total no. of posts in the
district
Currently occupied
Remarks(if any)
Upazila Nirbahi Officers Upazila Health and Family Planning Officers
Upazila Primary Education Officers
Sub Assistant Engineer, DPHE
Upazila Social Service Officers
Total 3. Financial resources
3.1 Annual allocations in thousands of Taka:
Sectors 2000 2001 2002 2003 2004 2005 2006 Health & Nutrition
Education WATSAN Social Welfares Local Government
3.2 Annual expenditure in thousands of Taka
Sectors 2000 2001 2002 2003 2004 2005 2006 Health & Nutrition
Education WATSAN Social Welfares Local Government 4. Development partners
Names of partners Local NGOs International NGOs
Bilateral donors
Multi-lateral donors
12 Assessment of district performance in making progress towards MDGs in Bangladesh
Annex 2. Outcome of the Rapid Assessment of District Performance Determinants
% of population
to total population
Total population
% of Population
living under
national poverty
line
% of unions seasonally
not accessible
Pupil-teacher
ratio
Population doctor ratio
Population-health
personnel ratio
% of population affected by
major natural
disasters
Frequency of turn-
over in DC post
Frequency of turn-
over in the district key
posts
Frequency of turn-over
in UNO posts
Frequency of turn-over in Upazila key posts
Per capita expenditure of MoHFW LNGOs INGPs
Bilateral donors
Multi-lateral donors
2006 2006 2005 2000-2006
2000-2006 2000-2006 2000-2006 2000-2005
Cox''s Bazar 1.6 2,257,809 52 63 109 24,541 12,117 1.1 6 6 5 3 83.94 9 3 0 5 Bandarban 0.2 318,616 65 41 52 7,586 3,402 1.4 6 5 4 3 294.32 17 3 2 7 Jamalpur 1.6 2,234,166 54 88 77 26,597 9,009 8.4 10 7 7 4 90.43 773 4 2 9 Habiganj 1.3 1,880,380 47 39 68 22,385 11,083 8.5 6 5 5 3 91.59 4 1 1 0 Barguna 0.7 996,986 61 21 42 24,719 8,497 0.0 9 3 4 2 115.36 7 1 1 3 Panchgarh 0.7 948,572 56 0 157 27,102 10,236 0.0 7 7 6 3 105.05 9 1 0 0 Sunamganj 1.6 2,305,939 49 56 107 37,597 18,062 8.5 5 7 5 2 77.55 8 3 1 0 Barisal 2.0 2,855,780 60 26 63 36,302 12,692 4.7 6 5 5 3 254.44 9 1 2 3 Munshiganj 1.0 1,463,010 19 0 76 14,630 6,989 7.7 7 8 8 3 109.76 0 1 0 7 Narail 0.6 792,335 45 0 46 25,559 8,225 1.9 7 6 9 5 111.35 307 2 1 8 Low performance for children High performance for children
Average population
per district
% Poor Population
% of unions
seasonally not
accessible
Pupil-teacher
ratio
Population-Doctor ratio
Population-health
personnel ratio
% Population
affected by major natural
disasters
Frequency of turn-
over in DC post
Frequency of turn-
over in the district
key posts
Frequency of turn-over in UNO posts
Frequency of turn-over in Upazila
key posts
Per capita expenditure of MoHFW
Average LNGOs
per district INGPs
Bilateral donors
Multi-lateral donors
1,537,591 56 55 71 22,458 9,433 32 7 30 5 3 168.91 162 2 1 5 1,673,127 46 22 79 27,339 11,139 38 6 39 6 3 131.63 67 2 1 4