Socio-Economic Survey Report

SURVEY 5
March 2010-July 2011

shiree

shiree Report
Panel surveys monitoring the changes in Socio-Economic of extreme poor households between March 2010 (survey 1) and July 2011 (survey 5)

March 2012

Executive Summary
1. Background: The six Scale Funds are working with a total of 82,850 extreme poor households. In March 2010, 64 households from each of the six Scale Funds were randomly selected for regular follow-up. This report provides information on changes in socio-economic and nutritional status of the same households studied in March 2010 and again a year later in March 2011. 2. Attrition: 329 households took part in all five surveys conducted in March, July and October 2010, March and July 2011. There was greater attrition in the urban sample (34%) than in the rural areas (11%). Information was collected on 1182 individuals of whom 683 were adults. 3. Male and female headed households and family size: In the total sample 38.6% of households were female headed and mean family size increased significantly from 3.3 in survey 1 to 3.6 in survey 5. Female headed households were smaller by, on average, 1.2 family members. 4. Schooling: Only 22.8% of heads of households had attended school significantly more so in male (30.2%) than female headed households (11.0%). Between surveys 1 and 5 school attendance increased significantly from 77.9% to 85.5% in school-aged children. Significantly more girls (92.2%) attended school than boys (79.1%). 5. Morbidity status: The health status of family members was determined on the day of the survey and over the previous 7 and 30 days. For all family members together cough and skin were the most common ailments on the day in survey 5, while fever and cough were the most common over the previous 7 days. Over the last 30 days one in three family members had a fever in survey 5 and 1 in 6 a cough in survey 5. . 6. Employment: Unemployment rose again in survey 5 but begging decreased to 7% in female headed households. Between surveys 2 and 4 self employment increased in male headed households and then stabilised. The number of days worked increased significantly between surveys 3 and 4 and then fell in survey 5. 7. Loans and savings: There was no consistent pattern in the number and amount of loans over the five surveys. In survey 1, 37% had cash savings and in survey 4 100% of households had cash savings and the mean amount increased significantly from 444 Taka (n=124, in survey 1) to 1103 Taka (n=336, in survey 4). 8. Income: In the total sample mean household income increased from 1700 Taka/month in survey 1 to 2850 Taka/month in survey 5. This increase was mainly due to the much higher income in the urban areas (over 2500 Taka/month higher). In rural areas the percentage of households below 22 Taka pppd income remained stable at about 70% for surveys 1 to 3 but fell to 50% in surveys 4 and 5; with a 26 Taka pppd threshold 80% of households were below the threshold in surveys 1 to 3 falling to 62% in survey 4 and 57.5% n survey 5. In urban areas the percentage below 26 Taka pppd fell from over 50% in survey 1 to under 30% in surveys 2 and 3, to 19% in survey 4 and to 16% in survey 5; the equivalent percentages for 30 Taka pppd were 60% in survey 1, i

33% in surveys 2 and 3, 21% in survey 4 and increased to 27.5% in survey 5. 9. Expenditure: Total household expenditure fell between surveys 1 and 2 (2129 and 1842 Taka/month, respectively) and then increased in surveys 3 and 4 (2296 and 2566 Taka/month, respectively followed by a slight fall in survey 5 to 2466 Taka/month. Male headed household expenditure was significantly greater than female headed by, on average, 872 Taka. Expenditure in urban areas was nearly double that found in the rural areas. There was a significant fall in the percentages below the 22 and 26 Taka pppd thresholds in the rural areas only in surveys 4 and 5, whereas the percentages below the urban threshold of 26 Taka pppd fell consistently from 25% in survey 1 to 9.1% in surveys 4 and 5. For the 30 Taka pppd threshold there was a fall from surveys 1 and 2 of about 41% to 20% in survey 3, 9.1% in survey 4 and an increase to 20.5% in survey 5. 10. Difference between income and expenditure: The difference between household income and expenditure (credit/debit balance) was calculated for each household and the overall mean changed from a debit in survey 1 to credit of +818 Taka/month in survey 5. When the average of the five surveys was calculated all NGO means were in credit except for SCF ( 182 Taka/month). 11. Household food intake and coping strategies: Food diversity improved in survey 5 and the mean number of foods consumed was 7.7 (maximum 13), up from 5.9 in survey 1. Mean food diversity also improved from 4.2 (maximum 7) in survey 1 to 5.3 in survey 5. DSK had the highest food diversity (5.6) and PAB and SCF the least (both 4.6). Food coping strategies showed significant improvement and mean coping strategies fell from 3.3 to 1.1 between surveys 1 and 5. 12. Social empowerment: More male headed households had a plan to improve their living conditions. Responsibilities within the household were more often shared between husband and wife in male headed households while in female headed households the wife was more likely to be responsible except for contraception and having children. About 1 in 8 women were not confident about talking to non-family males, taking small financial decisions or moving alone outside their locality. 13. Homestead garden: Just over 40% of rural households had a homestead garden but there was significant variation between NGOs (25% in UTTARAN and 81% in NETZ). The total value of the harvest averaged 302 Taka and very few households loaned to others.

ii

1. BACKGROUND
EEP/shiree (www.shiree.org) is a challenge fund supported by UKaid from the Department for International Development (DFID) in partnership with the Government of Bangladesh (GoB) to lift 1 million people out of extreme poverty by 2015. Harewelle International Ltd and PMTC Bangladesh Ltd manage the fund in consultation with EEP/shiree consortium partners including the Centre for Development Studies (CDS) at Bath University, the British Council and Unnayan Shamannay. EEP/shiree is one in DFIDs portfolio of projects designed to reduce extreme poverty and vulnerability in Bangladesh. The EEP/shiree Challenge Fund is worth 65 million British Pounds (around USD$130M) and is being disbursed over a period of 8 years (2008-2015). It is also referred to as shiree (the Bengali word for steps and an acronym for "Stimulating Household Improvements Resulting in Economic Empowerment") reflecting the aim of providing households ways out of extreme poverty. In order to monitor and evaluate socio-economic, empowerment and nutritional change, longitudinal (panel) surveys are being conducted (seasonally and annually) on randomly selected households. Besides these surveys, SHIREE is also supporting qualitative studies which will focus on key livelihood aspects of extreme poverty. The qualitative studies will provide rich longitudinal data which will be used with the surveys to gain more rounded insights into the choices and constraints facing extreme poor households. This report provides information on the changes in socio-demographic and economic characteristics of households (including household assets, income and expenditure and social empowerment) between March 2010 and July 2011 . shiree is working with 6 NGOs. 2 NGOs (CARE and PAB) are working in the far north-west of Bangladesh, NETZ in the north-west, DSK in two urban slums in Dhaka and SCF and UTTARAN in the south-west (Table 1). The total number of households that the 6 NGOs are working with is 82,850. Table 1 Location of the 6 NGOs and number of households NGO CARE DSK NETZ PAB SCF (UK) UTTARAN Location Gaibandha, Nilphamari, Rangpur, Lalmonirhat Dhaka slums Naogaon Gaibandha, Nilphamari, Rangpur, Lalmonirhat Khulna, Bagerhat Satkira, Khulna Number of Households 20,000 10,000 9,000 16,850 15,000 12,000

2. AIMS OF THE SEASONAL SURVEYS

Through the annual surveys the project aims to determine:(a) household seasonal change in socio-economic and empowerment status as a result of the shiree programme (b) intra-individual (primarily mother and <5 year old children) change in nutritional status (c) differences in nutritional, socio-economic status and empowerment between new and old recruits within the same NGO, and in the longer term (d) differences in nutrition, socio-economic status and empowerment between participants from different NGOs (e) differences between rural and urban cohorts

3. STUDY DESIGN
A longitudinal (panel) study design is being used (Figure 1) in which 384 households, 64 households from each NGO, were randomly recruited in March 2010 and a further 128 households were recruited in March 2011 (64 urban households and 64 rural households from NETZ). The longitudinal design will examine (a) within subject changes (the yellow lines) (b) between cohort comparison of old and new cohort (purple lines) (c) recruitment homogeneity (red line) and (d) by year 3 for differences between NGOs.
Figure 1 Study Design Baseline 2010 2011 2012 2013 2014 2015

384

345 128

310 115
425

279 104
383

251 94
345 311

226 85

Total 384

473

In March 2010, 64 representative households were selected from each of the 6 NGOs on the basis of the variables provided by the NGOs, usually the reported monthly income, educational level of the head of household, presence of under five year old in the household, age of the household head, household size and sex of household head. A representative back-up list was also generated in case households were absent on the day of the survey. A similar exercise was undertaken in the selection of the additional 128 households in March 2011.

4. FIELD WORK

The survey was completed in 27 days commencing on 6th July 2011 and finished on 1 August 2011. A total of 24 people were involved in conducting the survey comprising 1 Researcher from Cambridge University, 1 Bengali Young Professional, 2 Programme Managers, 1 Data Manager, 6 Research Officers (but 1 absent later because of a health problem), 6 Research Assistants, 4 Enumerators and 4 Measurers. A flexible survey team structure was used. Mainly 2 sub-teams were used, each team comprising 7 members (5 enumerators who were responsible for the questionnaire with supervisors (Researcher, Young Professional, Research Officer, Programme Manager or Data Manager) to supervise the questionnaire data collection. In one day 16 households were visited by each team (32 households in total), hence it took 2 days usually to survey each NGO, except NETZ (3 days) and DSK (4 days) so as to complete an additional 32 households. The timetable allowed for some slippage as well as movement from 1 NGO to the next. A trained Bengali enumerator asked a series of pre-tested questions to the head of household (or if the male head was absent, his spouse). The structured questionnaire covered 9 key areas:a. socio-demographic characteristics b. disability, chronic illness and health status of all household members c. cash loans d. household income and expenditure e. household food intake and food security f. gender and empowerment issues The interview usually lasted between 45 and 60 minutes.

5. RESULTS 5.1 BASIC SOCIO-DEMOGRAPHIC CHARACTERISTICS OF THE SAMPLE

In total 329 household participated in the five surveys (March 2010, July 2010, October 2010, March 2011 and July 2011, called surveys 1 to 5, respectively) from the initial sample of 384 households. There was significant differential attrition across NGOs (Table 2, p=0.02) with greatest loss in DSK (34%) and least in UTTARAN, (8%) but there was no significant difference in attrition rate between the five rural NGOs. Information was collected on 1182 individuals, 683 adults, 339 children five to fifteen years old and 160 children under 5 years of age. Just under 40% of households had a female head (38.6%) compared with 10.2% nationally (Household Income and Expenditure Survey, HIES, 2005) but there was highly significant variation between NGOs (p<0.001, Table 2) with most female headed households in DSK and NETZ and least in CARE. Female heads were primarily widowed (60%) or divorced/abandoned (23%) and only 16.5% were married while nearly all male heads were married (97.0%). Table 2 Attrition (%) between surveys 1 and 5 by NGO and Female headed households (%) by NGO in survey 4 NGO CARE DSK NETZ PAB SCF UTTARAN Total Rural Total Attrition (%) 14 34 10 10 14 8 11 14 Female headed households (%) 14.5 59.1 53.4 25.9 45.9 37.3 35.4 38.6

Repeated measures analysis of variance was used to examine the change in overall family size over the five surveys and as Figure 2 shows there was a small, but highly significant, increase in family size from survey 1 to survey 5 (mean family size, 3.3 in survey 1 and 3.6 survey 5 (p<0.001) with male headed households having, on average 1.2 more family members (4.0 versus 2.8, p<0.001).

Figure 2 Changes in mean family size over the five surveys

5.2 SCHOOLING
Only 22.8% of heads of households had attended school compared with 49% nationally and male heads were more likely to attend school than female heads (30.2% and 11.0%, respectively, p<0.001). Of all adults about 30% had attended school more so in male (34%) than female headed households (25%, p<0.001). There was a significant increase in school attendance between survey 1 and survey 5 in children 5 to 15 years of age. In survey 1, 77.9% of children attended school increasing to 85.5% in survey 5 (p<0.001). In survey 5 significantly more girls attended school than boys (92.2% versus 79.1%, respectively).

5.3 MORBIDITY STATUS

The reported prevalence of morbidity was obtained at each survey. For household heads the main findings were of no significant changes in diarrhoea over the five surveys on the day of the survey and the previous 7 days but a significant fall in the last 30 days (Table 3). Fever prevalence tended to fall between July 2010 and 2011. Cough showed an inconsistent trend but fell between surveys 1 and 5. Eye infection and passing of worms both fell sharply between surveys 1 and 5. On the day of the July 2011 survey fever, cough and skin infections were the most common ailments in family members (Table 4). Over the previous 7 and 30 days fever and cough were the most common ailments (Tables 5 and 6). For all family members together (Table 7) cough and skin were the most common ailments on the day in survey 5, while fever and cough were the most common over the previous 7 days. Over the last 30 days one in three family members had a fever in survey 5 and 1 in 6 a cough in survey 5.

Table 3 Morbidity status (%) of head of household over the five surveys
Condition 1 1.5 7.4 21.7 8.6 20.8 14.9 2 3.3 13.5 9.9 11.4 4.2 1.2 Day of survey Survey 3 4 5 4.5 2.1 2.2 11.4 7.4 7.3 18.9 13.1 10.7 19.8 12.5 8.3 4.2 4.2 1.8 3.6 0.3 0.6 1 to 5 p ns ns <0.001 0.001 <0.001 <0.001 2&5 p ns 0.010 ns ns ns ns Previous 7 days Survey 3 4 5 12.6 12.5 7.2 25.7 16.1 17.1 27.2 18.2 16.2 20.7 12.5 9.2 5.7 5.4 2.4 7.5 1.5 1.8 1 to 5 p ns ns 0.001 0.001 <0.001 <0.001 2&5 p ns 0.015 ns ns ns 0.034 Previous 30 days Survey 3 4 5 23.7 21.4 16.2 47.6 36.6 35.8 37.1 28.9 23.2 21.6 12.8 10.1 15.9 6.3 3.4 10.8 4.2 4.3 1 to 5 p <0.001 0.001 0.002 <0.001 <0.001 <0.001 2&5 p <0.001 <0.001 ns ns ns 0.001

Diarrhoea Fever Cough Skin infection Eye infection Passed worms

1 8.9 23.5 26.8 8.6 22.6 17.6

2 10.2 24.9 15.9 13.5 4.2 4.8

1 19.6 43.5 38.1 8.9 23.2 21.1

2 21.3 51.5 28.4 13.5 6.0 11.4

Table 4 Morbidity status (%) of all family members on the day of the study over the five surveys
Condition 1 2.0 7.7 16.0 7.0 16.9 14.7 2 3.0 11.0 8.4 8.8 3.0 1.2 Survey 3 3.2 11.4 16.8 16.1 3.3 2.6 All adults 4 2.0 7.7 11.3 9.0 3.5 0.3 5 1.7 6.8 8.9 8.0 1.7 0.6 1 to 5 p ns ns <0.001 <0.001 <0.001 <0.001 2&5 p ns 0.012 ns ns ns ns 1 0.6 8.6 9.3 3.7 0.3 19.4 2 0.3 6.5 6.2 2.2 0 1.5 5-15 year old children Survey 1 to 5 3 4 5 p 2.2 0.9 1.0 ns* 4.7 3.7 2.9 ns 8.2 4.0 1.9 0.008 4.7 5.6 3.8 ns 0.3 0.6 0.3 ns* 2.8 0 0.3 <0.001 2&5 p ns 0.031 0.006 ns ns* ns < 5 year old children Survey 1 to 5 3 4 5 p 7.0 3.3 2.0 ns 7.0 13.0 9.2 ns 14.1 12.4 10.5 ns 10.6 5.3 9.9 ns 0 1.3 0.7 ns* 3.5 1.3 0 <0.001 2&5 p ns ns ns ns ns* ns*

Diarrhoea Fever Cough Skin infection Eye infection Passed worms

1 3.6 10.8 16.5 5.0 1.4 19.4

2 6.0 9.7 9.0 5.2 1.5 1.5

*Exact test, **Yates correction

Table 5 Morbidity status (%) of all family members in the previous 7 days over the five surveys
Condition 1 8.2 19.8 20.1 7.0 18.3 18.6 2 8.2 21.5 11.9 10.2 3.3 5.0 Survey 3 10.3 24.5 24.1 16.2 5.5 7.0 All adults 4 9.2 18.8 15.8 9.0 4.5 1.5 5 6.0 14.2 13.1 8.3 2.3 1.2 1 to 5 p ns <0.001 <0.001 <0.001 <0.001 <0.001 2&5 p ns 0.001 ns ns ns <0.001 1 3.1 17.6 13.0 4.0 0.6 21.3 2 7.2 18.5 4.6 2.2 0 3.7 5-15 year old children Survey 1 to 5 3 4 5 p 8.8 7.2 2.2 0.012 13.2 12.1 10.8 0.018 8.5 7.2 5.4 0.012 5.0 5.6 4.4 ns 3.1 1.3 0.3 0.015* 6.6 1.6 1.6 <0.001 2&5 p 0.003 0.006 ns ns ns* ns < 5 year old children Survey 3 4 5 9.6 5.8 5.3 26.7 24.0 25.0 18.5 20.9 16.4 10.4 6.5 10.5 3.7 1.3 2.0 12.6 6.5 3.9 1 to 5 p ns ns ns ns ns 0.005 2&5 p 0.042 ns ns ns ns** 0.007

Diarrhoea Fever Cough Skin infection Eye infection Passed worms

1 10.8 25.9 25.9 4.3 2.2 19.4

2 6.3 25.9 14.9 5.9 3.7 12.7

*Exact test, **Yates correction

Table 6 Morbidity status (%) of all family members in the previous 30 days over the five surveys
Condition 1 16.0 38.9 29.7 7.1 18.6 20.3 2 17.5 44.0 23.3 10.5 5.0 10.8 Survey 3 20.3 44.4 33.5 17.1 14.4 9.1 All adults 4 17.6 32.5 24.2 9.2 5.1 3.6 5 13.8 28.8 18.9 9.2 2.9 2.9 1 to 5 p <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 2&5 p <0.001 <0.001 ns ns ns <0.001 1 9.0 32.4 19.4 4.0 1.2 22.5 2 11.1 43.0 20.1 2.2 0.3 10.8 5-15 year old children Survey 3 4 5 14.5 9.1 5.7 31.8 20.9 24.8 20.4 11.8 10.2 6.3 5.6 4.8 13.8 1.9 1.3 8.8 4.4 4.4 1 to 5 p ns <0.001 0.003 ns <0.001 <0.001 2&5 p 0.015 <0.001 <0.001 ns ns 0.003 < 5 year old children Survey 3 4 5 9.6 5.8 13.8 40.0 41.6 46.7 33.6 27.4 27.6 11.9 6.5 11.8 8.1 1.3 2.0 19.3 12.4 12.5 1 to 5 p ns 0.015 ns <0.001 ns ns 2&5 p ns 0.005 ns ns ns** 0.039

Diarrhoea Fever Cough Skin infection Eye infection Passed worms

1 10.8 48.2 38.1 5.8 2.2 23.0

2 11.9 63.4 33.6 6.0 4.5 21.6

*Exact test, **Yates correction

Table 7 Morbidity status (%) of all family members together on the day, previous 7 and 30 days over the five surveys
Condition 1 1.8 8.4 14.1 5.8 10.2 16.6 2 2.6 9.5 7.8 6.5 2.0 1.3 Day of survey Survey 3 4 5 3.4 1.8 1.5 8.9 7.3 6.0 14.0 9.4 7.2 12.1 7.6 7.1 2.1 2.4 1.2 2.8 0.4 0.4 1 to 5 p ns ns <0.001 <0.001 <0.001 <0.001 2&5 p ns 0.003 ns ns ns 0.024 Previous 7 days Survey 3 4 5 9.8 8.2 4.8 21.6 17.5 14.7 19.0 15.8 11.4 12.3 7.7 7.5 4.6 2.6 1.7 7.5 2.2 1.7 1 to 5 p 0.006 <0.001 <0.001 <0.001 <0.001 <0.001 2&5 p 0.001 <0.001 ns ns ns <0.001 Previous 30 days Survey 3 4 5 15.9 13.6 11.5 40.3 30.4 30.1 29.8 21.1 17.6 13.4 7.8 8.3 13.5 3.7 2.3 10.2 5.0 4.7 1 to 5 p 0.035 <0.001 <0.001 <0.001 <0.001 <0.001 2&5 p 0.017 <0.001 0.001 ns ns <0.001

Diarrhoea Fever Cough Skin infection Eye infection Passed worms

1 7.0 19.9 18.8 5.8 11.2 19.5

2 8.4 21.1 10.1 7.3 2.4 5.6

1 13.3 38.2 27.8 6.0 11.6 21.3

2 15.0 46.1 23.6 7.5 3.6 12.0

5.4 EMPLOYMENT
There were significant changes in the main occupation of male headed (but not female headed) households across all five surveys but not between surveys 2 and 5 (Table 8). Unemployment increased in both male and female headed households while begging fell to 7% in female headed households. Information on self employment was only ascertained in surveys 2 to 5. In male headed households there was an increase in self employment in the total sample from surveys 2 to 4 and close to 50% of male headed householders were self employed in surveys 4 and 5 although there was considerable variation between NGOs (Table 9). In female headed households there was an inconsistent pattern of self employment over the surveys. Information on the number of days worked in the last 7, 14 and 30 days and hours worked in the last 7 days was only collected in surveys 3 to 5. There was a highly significant increase in the days and hours worked between surveys 3 and 4 and a fall between surveys 4 and 5. Male heads worked less days than female heads (mean of female heads set to 0,Table 10). The self employed worked more days. Questions on advanced sale of labour were only asked in surveys 3 to 5 (Table 11). The analyses showed that the percentage paid in advance fell from 3% to under 1% but rose again in survey 5.

Table 8 Main occupation (%) of head of households over the five surveys
Occupation 1 5.4 35.1 20.0 0.5 16.1 4.4 4.9 0 1.5 8.8 3.4 0 2 2.5 28.8 19.2 1.0 19.2 3.0 9.6 0.5 2.0 11.1 3.0 0 Survey 3 1.0 27.5 12.0 1.0 18.5 6.0 6.5 1.0 6.0 16.5 3.0 1.0 Male 4 0 28.7 12.2 0.5 19.1 5.3 5.9 1.1 6.4 17.6 2.1 1.1 5 2.6 29.5 11.6 0.5 16.3 6.8 7.4 0.5 6.3 17.4 1.1 0 1 to 5 p 0.017 2&5 P ns 1 6.1 15.3 9.2 31.3 0 0.8 2.3 0 1.5 11.5 16.8 5.3 2 3.2 13.7 12.9 25.0 0 0.8 4.0 2.4 2.4 12.9 12.1 10.5 Survey 3 2.2 19.1 9.6 22.1 0 1.5 2.9 2.9 2.9 16.2 11.8 8.8 Female 4 0 20.0 10.0 23.8 0 1.5 3.1 3.1 2.3 16.2 11.5 8.5 5 4.7 13.3 10.2 19.5 0 1.6 1.6 6.3 7.0 14.8 7.0 14.1 1 to 5 p ns 2&5 p ns

Unemployment Agricultural day labourer Other day labourer Domestic maid Rickshaw Skilled labour Fishing/aquaculture Livestock Cottage/garment Petty trade Begging/scavenging Housework

NB: all tests were exact tests Table 9 Self-employed heads of households (%) by NGO over surveys 2 to 5
NGO 2 36.2 52.6 3.8 38.1 61.3 40.5 38.6 Male Survey 3 4 40.4 50.0 61.1 52.6 11.5 23.1 31.0 30.0 64.5 64.3 48.6 66.7 41.8 47.7 Female Survey 3 4 37.5 37.5 69.0 64.3 60.6 36.4 43.8 25.0 68.0 65.2 54.5 28.6 59.4 45.0 Total Survey 3 4 40.0 48.1 66.0 59.6 39.0 30.5 34.5 28.6 66.1 64.7 50.8 52.6 48.8 46.6

CARE DSK NETZ PAB SCF UTTARAN Total

5 53.2 60.0 13.0 27.5 71.4 63.9 48.1

p ns ns ns ns ns ns ns

2 50.0 69.0 51.5 25.0 60.0 27.3 49.6

5 75.0 57.1 55.9 43.8 66.7 35.0 54.0

p ns ns ns ns ns ns ns

2 38.2 62.5 30.5 34.5 60.7 35.6 43.0

5 56.4 58.1 38.9 32.1 69.4 53.6 50.6

p ns ns ns ns ns ns ns

Table 10 Mean number of days and hours worked by head of household, urban-rural and type of employment in surveys 3 to 5
Number of days worked In the last 7 days In the last 14 days In the last 30 days Hours worked in the last 7 days 3 3.51 7.05 14.45 5.34 Overall 4 5 4.34 3.66 8.65 7.67 18.30 16.42 6.31 5.36 p <0.001 <0.001 <0.001 <0.001 3 -0.59 -1.52 -2.36 +0.74 Male* 4 -0.19 -0.37 -0.41 +1.02 5 -0.61 -1.21 -1.65 +1.16 p 0.035 0.011 ns 0.001 3 +1.56 +3.32 +5.39 +2.17 Urban** 4 +0.65 +8.46 +2.46 +0.02 5 +0.14 +7.53 +2.48 +0.55 p 0.012 0.001 0.004 0.035

* in these analyses the reference category (set to a mean of 0) were female headed households ** in these analyses the reference category (set to a mean of 0) were rural households
3 +1.87 +3.73 +7.40 p <0.001 <0.001 <0.001 Self* 4 p +1.19 <0.001 +2.09 <0.001 +4.70 <0.001 5 +1.81 +3.52 +8.04 p <0.001 <0.001 <0.001

In the last 7 days In the last 14 days In the last 30 days

Hours worked in the last 7 days +1.25 0.006 +0.04 ns +0.35 ns *Independent sample t-test was performed to compare mean working day (or hours) between self vs non-self employment.

Table 11 Advanced sale of labour (%) in surveys 3 to 5

Advanced sale of labour % Range (days) *Exact test 3 3.0 0-7 Last 7 days 4 5 0.6 0.9 0-7 0-7 Last 14 days 4 5 0.9 1.2 0-14 0-10 Last 30 days 4 5 0.9 1.5 1-19 0-10 Last 3 months 4 5 0.3 0.9 0-2 0-15

p 0.026

3 3.0 0-14

p 0.044

3 3.0 1-30

p ns

3 1.5 0-18

p ns*

10

5.5 CASH LOANS AND SAVINGS

5.5.1 Cash Loans Five sources of cash loan were identified (i) free informal (ii) informal loans with interest (iii) interest loans from shomiti (iv) interest loans from microfinance institutions and (v) interest loans from bank or Government of Bangladesh. There was no consistent pattern as to the number or amount borrowed over the 5 surveys (Table 12). The number of microfinance loans was highest in surveys 4 and 5. Table 12 Number of loans, average amount of loan over the 4 surveys
Survey Type 1 Free informal Interest informal Shomiti Microfinance Bank Total 2 Free informal Interest informal Shomiti Microfinance Bank Total Free informal Interest informal Shomiti Microfinance Bank Total Free informal Interest informal Shomiti Microfinance Bank Total Free informal Interest informal Shomiti Microfinance Bank Total Number Household Mean/loan Mean/household 94 59 1695 2700 94 49 2447 4694 8 8 3258 3258 17 15 3902 4422 8 7 5667 6476 221 138 2373 3818 102 36 8 28 1 175 130 84 9 21 1 245 90 79 11 33 1 214 112 44 2 32 3 193 77 34 8 25 1 145 87 57 8 21 1 174 65 57 9 31 1 161 78 31 2 31 3 145 1553 2826 2975 3929 7770 2296 1189 2492 3364 3716 16700 1996 1972 4035 2510 4692 12000 3227 1498 3583 5750 4967 5874 2660 2058 2992 2975 4401 7770 2771 1777 3674 3364 3716 16700 2810 2730 5795 3068 4994 12000 4290 2150 5085 5750 5128 5874 3541

11

5.5.2 Cash Savings The respondents were asked about the extent of their cash savings (Figure 3) and a repeated measures analysis of variance showed that for the total sample the amount of savings increased significantly from surveys 1 to 5 (p<0.001) up from a mean of 410 Taka/household to 2150 Taka/household and by July 2011 85% of all households had some cash savings. There was no significant difference in mean cash savings between male and female headed households. Figure 3 Mean cash savings by survey

There was significant variation between the six NGOs (Figure 4) as well as the 5 rural NGOs in the amount of cash savings and at survey 5 DSK and UTTARAN households had, on average, the highest savings, while PAB had the least savings. When only those households with savings were analysed, the mean savings fell from 1114 Taka in survey 1, to 747 and 981 Taka in surveys 2 and 3 respectively and then increased to 1384 and 2527 Taka in surveys 4 and 5 respectively. Figure 4 NGO cash savings by survey

12

5.6 HOUSEHOLD INCOME

Repeated measures analysis of variance was used to examine the changes in income (based on HIES criteria) over the five surveys by both head of household and by NGO. Overall the mean income increased from about 1700 in survey 1 to 2850 in survey 5 although the main increase was between surveys 3 and 4 (up from 1989 to 2791 Taka, respectively). As can be seen in Figure 5 there was a higher mean income in male headed households by about 1000 Taka (average of the five surveys) but the gap in income fell from surveys 1 to 3 and then increased in surveys 4 and 5. Figure 5 Mean income by head of household by survey

Part of the increase in mean income was due to the much higher income in the urban area (average of the five surveys, 4563 Taka/month) compared with rural areas (average of the five surveys, 2013 Taka/month, Figure 6). Figure 6 Mean income by NGO by survey

13

The analyses were repeated for the five rural NGOs and there was significant variation (p<0.001) between the NGOs with overall higher means for CARE and UTTARAN. Most of the changes occurred in surveys 4 and 5. The mean per capita income increased over the five surveys particularly between surveys 3 and 4. The mean per capita income for the total sample was not significantly different between male and female headed households (23.7 and 20.9 Taka pppd, respectively, average of the five surveys, p ns, Figure 7). Figure 7 Mean income pppd by head of household over the five surveys

The mean Taka pppd in the urban areas was more than double that in the rural areas (40.8 versus 17.4 Taka pppd, Figure 8). There was an upward trend in per capita income in the urban area up to survey 4 while in the rural areas there was a small fall over the first three surveys and only in surveys 4 and 5 did the mean per capita income significantly improve. In the rural areas CARE had significantly the highest overall mean (average of the 5 surveys) but there was highly significant heterogeneity in the pattern of means between NGOs (p<0.001, i.e. non-parallel lines). Figure 8 Mean income pppd by NGO over the five surveys

14

Based on 22 Taka pppd (2007 prices) the percentage of rural households below the threshold remained stable for surveys 1 to 3 at about 70% (Table 13 p ns), but there was a significant fall in survey 4 to just under 50% (p<0.001) and a further slight fall in survey 5. With a 26 Taka pppd (2009 prices) threshold about 80% of households were below the threshold for surveys 1 to 3 falling to 62% in survey 4 (p<0.001) and then to 57.5% in survey 5. In the urban areas over 50% of households were below the 26 Taka pppd (2007 prices) in survey 1 falling to under 30% in surveys 2 and 3, to 18.2% by survey 4 (p<0.001) and then 15.9% in survey 5. With a 30 Taka pppd threshold, 65.9% were below the threshold in survey 1 falling to over 30% in surveys 2 and 3, just over 20% in survey 4 but increasing to 27.3% in survey 5. Table 13 Percentage of households below per capita cash income (Taka pppd) thresholds over the five surveys Survey 1 2 3 4 5 <22 pppd 69.1 69.8 77.5 49.8 48.1 Rural <26 pppd 78.6 80.4 83.5 62.5 57.5 Urban <30 pppd 65.9 36.4 31.8 20.5 27.3

<26 pppd 54.5 29.5 29.5 18.2 15.9

In-kind income increased significantly especially between survey 2 and survey 3 (331 and 497 Taka per month respectively, p=0.025) and rose to 560 Taka in survey 5. Female headed households had significantly greater in-kind income than male headed households based on the average of the five surveys (523 versus 349, respectively, p<0.001) however the gap between female and male headed households in-kind income closed in survey 5 (Figure 9). Figure 9 Mean in-kind income by head of household over the 5 surveys

15

When the average of the five surveys was calculated DSK had significantly higher mean cash income and total income. Among the rural NGOs CARE had significantly higher cash and total mean income than NETZ, PAB and SCF/ There was no significant heterogeneity between NGOs in in-kind income (the mean of the five surveys was 416 Taka, Table 14). Table 14 Mean income (Taka) based on the average of the five surveys by NGO NGO CARE DSK NETZ PAB SCF UTTARAN Total Rural Total p (Rural) p (Total) Cash income/month 2651 4563 1442 1878 1762 2333 2011 2342 <0.001 <0.001 In-kind income/month 422 537 441 378 326 415 397 416 n.s. n.s. Total income/month 3073 5092 1882 2256 2090 2749 2409 2756 <0.001 <0.001

The percentage that in-kind income contributed to total income in the total sample rose from 18% in survey 1 to 22% in survey 3 and then fell back to 19% and 18% in survey 4 and 5 respectively. DSK had significantly the lowest percentage throughout the five surveys (overall 8%) while NETZ had the highest (overall 28%, Table 15). Table 15 In-kind income as a percentage of total income by NGO over the five surveys NGO CARE DSK NETZ PAB SCF UTTARAN Total Rural 1 17 8 17 16 24 22 17 19 2 16 5 32 23 11 22 18 18 Survey 3 19 9 34 29 19 19 22 22

4 14 9 25 24 22 18 19 19

5 18 11 31 17 18 14 18 20

16

5.7 HOUSEHOLD EXPENDITURE

Male headed food expenditure was significantly higher than female headed expenditure, on average, over the five surveys (overall 1952 and 1194 Taka, respectively). Food expenditure did not show a consistent pattern over the five surveys but was highest in surveys 1 and 4 (Figure 10). There was significant variation between NGOs with the highest spending in DSK across all surveys (Figure 11). When analyses were restricted to rural NGOs the average food expenditure over the five surveys by CARE households was significantly higher than the other rural NGOs although in survey 5 UTTARAN households, on average, spent slightly more than CARE households. On average, rural male headed households spent over twice as much on food each month than rural female households (1886 versus 961 Taka, respectively, p<0.001). Food per capita expenditure did not show a consistent pattern over the five surveys and was highest in surveys 1 and 4 (Figure 12). Even after correcting for household size the urban area had the highest mean food expenditure and male headed households spent significantly more on food, on average, than female headed households (17.2 and 15.3 Taka pppd, respectively, Figure 12). The pattern of food per capita expenditure varied significantly (p<0.001) by NGO across surveys (Figure 13). When the rural NGOs were analysed separately there were significant differences in overall means (average of the five surveys) with NETZ significantly the lowest per capita spending on food. Male headed households spent significantly more than female headed households (16.8 versus 13.0 Taka pppd, p<0.001). Household expenditure over the previous month, on average, fell between surveys 1 and 2 but increased thereafter in female headed households while in male headed households there was a small fall between surveys 4 and 5 (from 878 to 831 Taka). Overall male and female headed households had very similar spending on household items (Figure 14). Urban expenditure was far higher (1950 Taka/month), on average, than rural expenditure (between 300 and 600 Taka/month, Figure 15). When the rural areas were analysed separately, there was a just significant difference between overall means (p=0.022) mainly due to the difference between NETZ and CARE means. Rural male headed households spent more, on average, (531Taka) than female headed households (346 Taka, p=0.001). Household per capita expenditure did not vary significantly over the five surveys. Female headed households spent more, on average, per capita than male headed households (9.6 versus 6.1 Taka pppd, p=0.029, Figure 16). Urban households spent nearly three times as much per capita than rural households (Figure 17). There were no significant differences between the overall rural means (Figure 17). Work related expenditure did not differ overall between male and female headed households but there was significant variation between surveys with an increase in expenditure up to survey 4 in male headed households and up to survey 3 in female headed households (Figure 18). There was considerably more spent in the urban areas, on average, than in the rural areas (mean 236

17

versus 47 Taka, respectively, Figure 19). Per capita there was a just significant higher mean in male headed households (0.53 versus 0.25, respectively, p=0.037). The mean per capita spending increased in male headed households from surveys 1 to 4 while in female headed households the mean increased from survey 1 to 3 (Figure 20). Urban households spent more per capita than rural households (Figure 21). Total expenditure showed a fall between surveys 1 and 2 (2129 and 1842 Taka, respectively) and then an increase in surveys 3 and 4 (2296 and 2566 Taka) and slight fall in survey 5 to 2466 Taka/month. The overall percentage increase in expenditure between the baseline survey and survey 5 was 16% (p<0.001). Male headed household expenditure was significantly greater than female headed by, on average, 872 Taka (2694 versus 1822 Taka, respectively, p<0.001) and the gap appeared to be stabilising (Figure 22). Expenditure in urban areas was nearly double that found in the five rural areas (Figure 23). The rural analyses indicated that there were significant differences between NGOs with least overall expenditure in NETZ (1404 Taka/month) and greatest in CARE (2678 Taka/month). The six NGOs differ in their approach to asset transfer, target groups and working areas and more details can be obtained from www.shiree.org. Mean expenditure in rural male headed households was almost double that in rural female households (2476 versus 1282 Taka, respectively, p<0.001) but this difference did not take into account household size (see next paragraph). The pattern of expenditure also varied significantly between NGOs over the five surveys. Total per capita expenditure did not vary significantly over the five surveys with expenditure varying between 20.7 and 25.8 Taka pppd, nor where there any significant differences between male and female headed households (Figure 24). Overall the urban areas had greatest expenditure (Figure 25). The rural analyses indicated no significant differences in overall means, by head of household but per capita expenditure was highest in SCF (23.7) and CARE (23.3) and least in NETZ (16.5, p<0.001). Based on the 22 Taka pppd there was a significant fall in the numbers below the poverty thresholds in rural areas in survey 4 to 50% and no further change in survey 5 (Table 16). With the higher threshold of 26 Taka pppd the main fall was in survey 4 to 65.6% with no further change in survey 5. In the urban areas using the 26 and 30 Taka pppd thresholds there were steady falls from surveys 1 to 4 and in survey 4 only just over 10% of households were below the thresholds. In survey 5 there was an increase in those below the 30 Taka pppd to 20%.

18

Table 16 Percentage of household below per capita expenditure (Taka pppd) thresholds over the five surveys Survey <22 Taka pppd 2 3 4 68.5 56.3 40.0 89.7 75.9 60.3 69.0 63.8 60.3 77.4 52.8 47.3 73.7 64.2 40.7 75.7 63.0 49.8 Survey <26 Taka pppd 2 3 4 83.3 79.2 58.2 98.3 84.5 77.6 81.0 75.9 75.9 83.0 71.7 63.6 87.6 79.2 52.5 86.8 78.1 65.6 <30 Taka pppd 41.0 20.0 9.1

Rural CARE NETZ PAB SCF UTTARAN Total Rural Urban DSK

1 54.5 74.1 43.1 76.4 66.1 62.6

5 50.9 75.9 51.7 45.5 27.6 50.4

1 69.1 84.5 53.4 83.6 79.7 74.0

5 65.5 84.5 67.2 56.4 53.4 65.5

<26 Taka pppd 25.0 23.1 15.0 9.1

9.1 40.9

20.5

19

Figure 10 Mean food expenditure by head of household over the five surveys

Figure 11 Mean food expenditure by NGO over the five surveys

20

Figure 12 Mean food expenditure pppd by head of household over the five surveys

Figure 13 Mean food expenditure pppd by NGO over the five surveys

21

Figure 14 Mean household expenditure by head of household over the five surveys

Figure 15 Mean household expenditure NGO over the five surveys

22

Figure 16 Mean household expenditure pppd by head of five household over the five surveys

Figure 17 Mean household expenditure pppd by NGO over the surveys

23

Figure 18 Mean work-related expenditure by head of household over the five surveys

Figure 19 Mean work-related expenditure by NGO over the five surveys

24

Figure 20 Mean work-related expenditure pppd by head of household over the five surveys

Figure 21 Mean work-related expenditure pppd by NGO over the five surveys

25

Figure 22 Mean total expenditure by head of household over the five surveys

Figure 23 Mean total expenditure by NGO over the five surveys

26

Figure 24 Mean total expenditure pppd by head of household over the five surveys

Figure 25 Mean total expenditure pppd by NGO over the five surveys

27

5.8 DIFFERENCE BETWEEN HOUSEHOLD INCOME AND EXPENDITURE

The difference between household income and expenditure based on HIES criteria of income minus expenditure (positive sign indicates credit and negative sign debit) was calculated for each household at each survey. Repeated measures analysis of variance was used to examine the pattern of credit/debit over the five surveys and on average households went from a debit in survey 1 to increasing credit reaching +818 Taka in survey 5. There was no significant difference between mean credit of male and female headed households (Figure 26) although there was a fall in female headed households credit of nearly 300 Taka between surveys 4 and 5. When the average of the five surveys was calculated all NGOs were in credit (range +19 Taka/month to +546 Taka/month) except for SCF (-182 Taka/month, Figure 27).

28

Figure 26 Mean net income by head of household over the five surveys

Figure 27 Mean net income by NGOs over the five surveys

29

5.9 HOUSEHOLD FOOD INTAKE

The households were asked how often family members had eaten 13 food items in the 7 days prior to the study (Table 17). Rice was eaten by nearly all households in all five surveys and potatoes were consumed by most households in July 2011. Eggs and poultry consumption increased significantly in surveys 4 and 5 and fruit and green vegetables in survey 5. The extent of household food diversity was determined in two ways (a) based on the mean of the number of foods eaten (maximum 13) and (b) based on the 7 food groups (grains, roots and tubers, legumes and nuts, dairy products , flesh foods, eggs, vitamin A rich fruits and vegetables and other fruit and vegetables) as defined by WHO and UNICEF. Consumption of any amount of food from each food group is sufficient to count i.e. there is no minimum quantity. In the total sample the mean number of foods consumed in the last 7 days increased significantly from 5.9 in survey 1 to 7.1 and 7.0 in surveys 2 and 3, respectively before falling to 6.7 in survey 4 and increasing to 7.7 in survey 5. There was no significant difference between male and female headed households (Figure 28) but there was highly significant difference (p<0.001) between NGOs with DSK having the highest mean (8.0) and NETZ the lowest (6.4, Figure 29). Repeated measures analysis of variance just with the rural NGOs revealed no significant difference in the average number of foods consumed over the five surveys between the rural NGOs. Food diversity also varied by surveys with least diversity in survey 1 (4.2) increasing to 5.0 in survey 2 and falling to 4.9 and 4.8 in surveys 3 and 4 respectively and rising to 5.3 in survey 5. There was no significant difference between male and female headed household means (Figure 29). DSK had the highest mean diversity (5.6) and PAB and SCF the least (4.6, p<0.001, Figure 30). There was no significant heterogeneity between rural means.

30

Table 17 Number of days (%) in the last week that household members consumed foodstuffs
Food type 1 Rice 0 1 2 3+ Flour 0 1 2 3+ Pulse 0 1 2 3+ Potato 0 1 2 3+ Green vegetables 0 1 2 3+ Other vegetables 0 1 2 3+ Fruits 0 1 2 3+ Milk 0 1 2 3+ Eggs 0 1 2 3+ Fresh fish 0 1 2 3+ Dried fish 0 1 2 3+ Poultry 0 1 2 3+ Meat 0 1 2 3+ Mean foods eaten Mean food diversity 0 0 0 100 72.3 10.6 8.8 8.2 61.1 24.0 9.4 5.5 1.5 1,2 5.5 91.8 18.8 16.1 30.7 34.3 5.2 3.6 23.1 68.1 91.8 5.5 1.5 1.2 92.1 4.9 0.9 2.1 70.5 22.5 3.6 3.3 37.7 34.0 17.9 10.3 73.9 10.0 8.8 7.3 95.4 3.0 0.6 0.9 90.3 7.6 1.5 0.6 5.9 4.2 2 0 0 0.3 99.7 64.1 16.6 11.7 7.7 37.7 32.8 21.5 8.0 2.8 2.8 9.8 84.7 7.1 11.0 27.6 54.3 5.2 6.4 22.1 66.3 57.7 27.0 7.7 7.7 85.0 7.4 4.0 3.7 54.6 30.1 11.0 4.3 20.2 34.7 21.5 23.6 80.4 9.8 4.9 4.9 92.6 6.4 0.3 0.6 92.6 5.2 0.6 1.5 7.1 5.0 Survey 3 0.9 0.3 0 98.9 66.9 15.5 10.9 6.7 35.9 26.1 24.3 13.7 8.8 7.6 13.7 69.8 6.4 14.3 28.0 51.2 16.8 8.5 20.1 54.6 54.3 16.8 16.2 12.8 86.3 4.6 4.0 5.2 56.7 21.6 16.2 5.5 10.4 24.1 28.0 37.5 80.5 7.0 4.3 8.2 91.5 7.6 0.6 0.3 97.3 1.5 0.3 0.9 7.0 4.9 4 0 0 0 100 76.9 7.6 7.0 8.5 55.0 22.8 14.6 7.6 0.9 0.3 0.3 98.5 14.0 22.5 31.6 31.9 9.1 10.0 19.5 61.4 72.6 9.4 11.2 6.7 83.9 8.5 1.8 5.8 40.7 24.3 18.8 16.1 23.7 26.7 21.6 28.0 78.7 8.2 6.4 6.7 83.6 11.9 3.6 0.9 91.8 7.0 0.9 0.3 6.7 4.8 5 0.6 0.6 0 98.8 <0.001 67.8 10.6 11.2 10.3 <0.001 35.9 21.6 22.2 20.4 <0.001 2.4 1.8 6.4 89.4 <0.001 4.3 10.6 30.7 54.4 <0.001 5.8 10.9 17.6 65.7 <0.001 33.7 24.3 20.4 21.6 0.001 76.3 8.5 4.0 11.2 <0.001 38.0 28.6 17.0 16.4 <0.001 17.0 22.8 20.7 39.5 ns 76.3 5.5 6.7 11.6 <0.001 83.6 51.2 4.0 1.2 <0.001 87.8 9.1 2.1 0.9 7.7 5.3 p -

<0.001 <0.001

31

Figure 27 Mean number of foods consumed by Head of household over the five surveys

Figure 28 Mean number of foods consumed by NGO over the five surveys

32

Figure 29 Mean food diversity by head of household over the five surveys

Figure 30 Mean food diversity by NGO over the five surveys

33

5.10 HOUSEHOLD FOOD SECURITY

The households were asked about the coping strategies they used as a result of financial hardship in the seven days prior to the survey with a pre-coded list of 10 food strategies (Table 18). There were significant improvements in 9 of the 10 strategies. For example households reported eating smaller portions of food between the baseline study (March 2010) and July 2011 fell from 83.3% to 31.7%; eating less than 3 meals a day (down from 67.2% to 20.6%), eating food of less quality (down from 63.9% to 15.8%) and less adults ate no food in the previous 24 hours (down from 7.0% to 0.3%). Borrowing money to buy food fell from 19.1% to 3.6% and buying food on credit fell from 28.6% to 10.6%. There was significant improvement (reduction) in food coping strategies with a fall in mean from 3.3 to 1.1 between surveys 1 and 5. There was no significant difference in mean coping strategies between male and female headed households (Figure 31) but there were significant differences between NGOs and DSK had the best mean food coping strategy (1.3) and PAB the worst (3.7, p<0.001, Figure 32).

34

Table 18 Food coping strategies over the five surveys

Food coping strategy 1 Eat smaller portion 0 days 1 day 2 days 3+ days Eat < 3 times a day 0 days 1 day 2 days 3+ days Eat food of less quality 0 days 1 day 2 days 3+ days Eat gathered food 0 days 1 day 2 days 3+ days Eat no food in 24 hours adult 0 days 1 day 2 days 3+ days Eat no food in 24 hours child 0 days 1 day 2 days 3+ days Borrow money to buy food 0 days 1 day 2 days 3+ days Bought food on credit 0 days 1 day 2 days 3+ days Send family member elsewhere for food 0 days 1 day 2 days 3+ days Give more food to earning household members 0 days 1 day 2 days 3+ days Mean food coping 16.7 8.8 28.0 46.5 32.8 4.0 16.7 46.5 37.1 21.6 21.0 20.4 79.9 7.4 7.9 2.7 93.0 6.1 0.9 99.7 0.3 2 20.2 13.1 33.0 33.6 27.8 4.9 16.8 50.5 49.8 20.5 18.7 11.0 50.2 15.3 15.3 19.3 97.9 1.5 0.6 99.4 0.3 0.3 Survey 3 31.8 7.0 21.1 40.1 48.9 6.1 11.9 33.0 52.9 9.8 15.9 21.4 58.7 11.9 14.1 15.3 97.2 2.1 0.6 99.7 0.3 p 4 48.0 7.9 19.1 24.9 60.2 3.3 13.4 23.1 67.5 10.6 13.1 8.8 80.9 9.4 7.3 2.4 98.2 1.8 100 5 <0.001 68.3 5.2 10.9 14.6 <0.001 78.4 4.3 5.2 12.2 <0.001 84.2 3.3 7.0 5.5 <0.001 90.0 2.7 4.9 2.4 0.001 99.7 0.3

ns 100

0.001 80.9 11.2 6.1 1.8 71.4 11.2 11.2 6.1 80.1 11.0 7.6 1.2 63.9 15.9 11.0 9.2 82.3 10.1 4.6 3.1 69.4 11.6 9.8 9.2 90.6 5.8 3.3 0.3 78.1 9.1 8.8 4.0 96.4 2.4 0.9 0.3 0.001 89.4 5.2 2.7 2.7 <0.05 82.7 5.2 5.5 6.7 88.7 4.3 3.1 4.0 84.7 3.1 6.4 5.8 86.9 1.5 6.7 4.9 94.8 1.2 1.2 2.7 0.001 65.3 3.0 7.0 24.6 3.3 62.4 2.1 7.3 28.1 3.5 63.0 0.6 8.3 28.1 2.9 67.2 2.4 30.4 2.2 86.6 0.3 3.3 9.7 1.1

<0.001

35

Figure 31 Mean food coping strategy by head of Household over the five surveys

Figure 32 Mean food coping strategy by NGO over the five surveys

36

5.11 SOCIAL EMPOWERMENT

Questions were put separately to the male and female heads of household and to female spouses (Tables 19 to 22). More male headed households had a plan to improve their living conditions as well as feeling they could implement the plan. More female headed households consulted a relative when the had money problems or other difficulties. Fewer than 20% of household heads consulted an influential person in the last 3 months. Responsibility in the household was more often shared between husband and wife in male headed households. In female headed households the wife was more likely to be responsible except for contraception and having children which was a joint decision of husband and wife. Treating boys and girls more equally was more common in male than female headed households About 1 in 8 women were not confident about talking to non-family males, taking small financial decisions or moving alone outside their locality (Table 22).

37

Table 19 Social empowerment

Question Do you have a plan about improving your living condition? Yes answer (%) If yes, do you feel able to implement the plan? Yes answer (%) Where do you go for help when you face money problems or other difficulties? Neighbour Relative In the last 3 months has your household ever contacted any of the following UP chairman, members of village shalish, political leaders or other influential people? Table 20 Responsibility in the household Question Who decides on the use of household money earned? Who decides on the use of household cash savings? Who decides on the taking of a loan? Who decides on spending money on education for your children? Who decides on spending money for health care of family members? Head Male Female Male Female Male Female Male Female Male Female Male Female Male Female Male Female Husband only 29.0 12.0 25.2 11.1 26.7 11.4 12.6 8.5 18.3 11.4 8.2 4.0 9.0 0.0 13.5 9.2 Wife only Both Other 7.6 61.1 2.3 59.8 13.7 14.5 6.9 65.6 2.3 60.7 14.5 13.7 7.6 62.6 3.0 59.6 14.9 14.0 5.8 80.6 1.0 63.4 18.3 9.9 4.6 71.8 5.4 61.9 14.3 12.4 9.3 79.4 3.1 20.0 52.0 8.0 3.0 87.0 1.0 23.8 71.4 4.8 2.7 82.0 1.8 52.6 25.0 13.2 Male 93.1 77.1 28.2 40.4 16.8 Head Female 85.6 62.4 24.8 57.4 19.2

Who usually decides on the use of contraception? Who usually decides on when to have children? Who usually decides on the marriage of your children (age and partner?)

38

Table 21 Importance of boys and girls Question

Is food more important for boys or for girls? Are sons more important than daughters? Is education more important for boys or for girls?

Head Male Female Male Female Male Female

Boys 17.6 26.4 27.5 34.4 19.8 24.8

Girls 14.5 14.4 12.2 20.8 9.9 16.0

Both equally important 67.9 59.2 60.3 44.8 70.2 59.2

Table 22 Womens confidence Question

Do you feel confident in talking to men who are not members of your family? Do you feel confident in taking small financial decisions alone (e.g. buying a sari)? Do you feel frightened of moving alone outside your village or urban area?

Confident Uncertain 80.5 7.5 76.9 11.4 81.2 6.2

Not confident 12.0 11.7 12.7

39

5.12 HOMESTEAD GARDEN

Only rural households had a homestead garden and the percentage with a garden ranged from 25.5% (UTTARAN) to 81.0% (NETZ) with an average of 42.1% (Table 23) across the five rural NGOs. Between 12% and 57% had grown produce on their homestead garden in the last 3 months with up to 5 different products being grown. Table 23 Percentage of rural households having a homestead garden NGO CARE NETZ PAB SCF UTTARAN Total % with homestead garden 29.1 81.0 37.9 36.4 25.4 42.1 p <0.001

The total value of the harvest averaged 302 Taka (Table 24), and only 9 households reported selling any produce, while food consumed by the household from the garden was worth on average 178 Taka. Only 8 households loaned food to others, worth on average 46 Taka and 18 household reported some costs associated with their garden (mean 148 Taka). Table 24 Value of the produce from the homestead garden
NGO CARE NETZ PAB SCF UTTARAN Total Value 294 261 271 399 282 302 p ns Sales 7 157 103 500 155 p ns Consumed p value 196 ns 166 151 205 182 178 Loaned p Value 20 ns 28 20 110 20 46 Costs of p garden 307 ns 110 300 111 12 148

40

shiree House 5, Road 10, Baridhara Dhaka 1212, Bangladesh Phone: 88 02 8822758, 88 02 9892425 E-mail: info@shiree.org

www.shiree.org