Introduction

Benazir Income Support Programme is a female-focused welfare programme in Pakistan aimed at Poverty alleviation. Recently, it has been subsumed into the larger Ehsaas Welfare programme, and continues to provide unconditional cash transfers to millions of women across the country. The impact of the programme exists in multiple aspects. It affects factors like household consumption expenditure, levels of malnutrition in children, children’s literacy rates, among many others. What remains a relatively unexplored impact of the fund is its impact on politics, and specifically female voter turnout. This report will analyse the impact that increased rates of beneficiaries of the fund have on the political participation of women.


Literature Review

In response to concerns about nationwide chronic poverty, health and education issues, the Benazir Income Support Programme was launched by the Pakistani government as its Flagship Welfare policy (Mumtaz & Whiteford, 2017). The way BISP operates is that it firstly uses the Poverty Score Card method to assign a household a particular score, which was developed from the PSLM survey (Pakistan Social and Living Standards Measurement). This survey looks at household characteristics like literacy, household size, housing, and agricultural holdings. After a household had its information logged into the PSLM survey, a Proxy Means Test formula would be used to assign that household a score between 0 and 100. If a family receieved a score of 16.17 out of 100 or below, they would become eligible for BISP payments till the next reevaluation of their score which goes above the threshold (Mumtaz & Whiteford, 2017). Payments are made on a monthly basis, being the equivalent of sixty percent of the national poverty line income. In order to understand the scale of this programme, we must consider the fact that BISP coverage reached 22.6% of the entire Pakistani population by 2015. (Mumtaz & Whiteford, 2017)

Many previous studies have zeroed in on the impact that Benazir Income Support has on female empowerment. This is as the targetting of cash transfers to adult women in households was done on the assumption that women would become more empowered. This is important as women, while empowered, would make better decisions for the overall welfare of the household than their male counterparts. For example, one report found that women focused particularly on access to safe and clean drinking water, disproportionately more than men (Abdul Latif Jameel Poverty Action Lab,2006). This difference in preference stems from women’s predominant role as within the household and being more aware of household problems like child nutrition, water access and food consumption. (Abdul Latif Jameel Poverty Action Lab, 2006) The same study also found that women also voiced this support in the form of policy reforms and in the political realm. Following this line of reasoning, it is reasonable to assume that if BISP payments empowered women, they would vote in accordance to their public good preference (which would benefit household welfare). Thus, testing whether BISP results in female empowerment is crucial to understand whether the programme as a whole has the ability to alleviate other problems like child malnutrition and health issues.

I hypothesize the impact of this programme on female political empowerment to be positive. This is due to the fact that BISP has also been seen as a factor in the improvement of beneficiary women’s mobility. For example, beneficiary women have seen an 8% average mobility increase while analysing mobility to health centres, religious centres and friend’s homes (Tehmeena Iqbal et al., 2020). This is possibly due to the fact that women need to physically travel outside their homes to receieve these BISP payments. This may lead to a destigmatization of women leaving the home, and may result in increased mobility. This means that women should theoretically find it easier to travel to polling stations to cast their votes.

Another reason for this hypothesis is that the Oxford Policy Management Evaluation Report for BISP found that a possible reason for increased female empowerment was the requirement of making a Computerized National Identity Card (CNIC). Many women in Pakistan do not posess this ID card, making it impossible for them to vote in elections. Therefore, we can theorize that the creation of the CNIC for many deprived women might increase voter turnout. (Oxford Policy Management, 2016). However, whether women actually availed this newfound opportunity is dependent on factors like mobility, extent of political apathy, and socio-cultural constraints from husbands or other family/society members.

While the OPM report did find a positive impact on female political participation, this report differs in its methodology as well as time frame. The OPM report focuses on the years uptil 2016 (as it had multiple evaluation reports, with the final one being in 2016), while my data looks at data collected from and around the time period of the 2017 Census and 2018 General Elections. Secondly, the OPM’s sample comes from approximately 9000 households across 8 districts and 4 provinces. This report takes a different approach to the research question, as discussed below.

Research Design

This report uses a multivariate regression to test the relationship between Benazir Income Support and Female Voter Participation. The unit of analysis is the district level (Administrative zone level 2). The independent variable is the percentage of BISP beneficiaries compared to the total population in a district. The data for amount of beneficiaries was obtained from a data request from the programme itself. This was then divided by the total population of the district to represent a percentage, with population data being obtained from the 2017 census. The dependent variable is the amount of female voter turnout as a percentage of the total voter turnout of a district, with this data being obtained from the 2018 General Election voter data created by the Election Commission of Pakistan. The first control variable selected for this regression is literacy. This data was obtained on a household level from the PSLM 2018 (Pakistan Social and Living STandards Measurement), where the respondents answered two questions: “Can you read a sentence in any given language?”, and “Can you write a sentence in any language?”. If a respondent’s answer was yes, they were coded as 1, and if not, 2 (for both questions). If a respondent gave yes to both answers, I coded them as literate. Then I calculated the amount of literate people in a district as a percentage of the total respondents to find the literacy rate of each district. The second control is urban population as a percentage of the total population of a district, with both urban population and total population being taken from the 2017 census. The sample of this research is limited due to two reasons. Firstly, Azad Kashmir and Gilgit-Baltistan are administrative regions that do not contest general elections, thus voter data on their districts was not available. Secondly, there are some districts in Pakistan that are so remote that there is no census data on them, (Population, demographics, beneficiary %age, etc.) which is why they were excluded from the sample. This leaves 115 districts to be analysed, out of a total of 169 (including Gilgit-Baltistan and Azad Kashmir) Thus, this analysis can only be generalized to the four provinces: Punjab, Khyber Pakhtunkhwa, Sindh and Balochistan. To clarify, all measurements of urban population, welfare coverage, literacy and voter turnout are in percentages (ranging from 0-100).

data visualization


Note: For reference, these are where the provinces lie on the map of Pakistan. Azad Kashmir and Gilgit Baltistan are not included in the map analysis as they lack data, and as they do not contest general elections. Thus, they are not included in the regression as well (as said above). In addition, some districts in Balochistan, as well as a few in Sindh lack data as well. Hence, the white spaces in the map.



It is important to map these four variables as it allows us to discern certain provincial trends. For example, we can see that districts with a generally lower amount of female voter turnout tend to be near the Pakistan-Afghanistan border. This may be due to cultural conservatism, as well as safety issues, given that these districts lie in the province of Khyber Pakhtunkhwa (a region that has long been plagued by the terrorism of the Taliban).

Literacy, we can see, is highest in the districts of the North of the Punjab Province, while South Punjab and the rest of the provinces lag behind.

Urban Populations tend to be low throughout the country, given that Pakistan is a primarily rural nation.

Finally, the most important map to look at is of the geospatial distribution of beneficiaries as a percentage of district population. What this shows us is that districts with a greater amount of BISP support tend to lie in the province of Sindh. This, despite the fact that Balochistan is far poorer, shows us that Beneficiaries per capita of a District is not just a measure of poverty, but also of access to state infrastructure. Sindh has a higher population, as well as population density than Balochistan. Thus, it would be easier for poor people in Sindh to recieve BISP payments than in the more remote areas of Balochistan. This visualization reveals a key limitation in my research that I will discuss later on as well. In addition, this visualization also reveals some important policy and research implications, which will be discussed in the final section.


This barchart shows the mean BISP coverage of a district by Province. We can, again, clearly see that Sindh’s districts have greater BISP coverage than the other provinces’s districts (standing at an average of 6.5%, while the others range from 3.7 to 4.7%)



These plots are simply visual justifications for including Urban Population and Literacy as my control variables. Both controls seem to have an impact on my independent and dependent variables, and thus would act as confounding factors if left unaddressed by the regression.


Regression Analysis

The regression results show that the effects of BISP, and the Literacy control variable are both statistically significant. In addition, the interaction effects of BISP with literacy and BISP with urban population %age are statistically significant, yet close to zero. The effect of beneficiaries as a percentage of District population is 0.974, significant at the p<0.01 level. This means that for every one percent increase in BISP coverage in a district, there is a 0.974% increase in female voter turnout, holding literacy and urban population as a constant. For every increase in percentage of a District’s literacy rate, there is a 0.173% increase in female voter turnout, holding BISP coverage and urban population percentage as a percentage. The effect of urban population is both negligible and statistically insignificant. The adjusted R-squared is 0.293, which means that approximately 30% of the change in female voter turnout is explained by the percentage of beneficiaries in a district.

Both interaction terms are statistically significant as well. For every one percent increase in urban population percent, the effect of BISP welfare support on female voter turnout increases by 0.012%. Meaning in highly urban districts, the positive impact of BISP coverage on female voter turnout increases. This may be possibly due to close proximity to and greater numbers of polling stations, making it easier for women to vote. On the other hand, for every one percent increase in literacy, the effect of BISP welfare support on female voter turnout decreases by 0.023%. This means that in highly literate districts, the positive impact of BISP on female turnout actually decreases as women are already likely empowered due to being able to get an education.

Impact of BISP on Female Voter Turnout
Dependent variable:
Female Turnout (%age)
bpc 0.974***
(0.241)
LITERACY 0.173***
(0.030)
urbanpercent -0.024
(0.022)
bpc:LITERACY -0.023***
(0.006)
bpc:urbanpercent 0.012**
(0.005)
Constant 36.091***
(1.310)
Observations 115
R2 0.324
Adjusted R2 0.293
Residual Std. Error 1.730 (df = 109)
F Statistic 10.463*** (df = 5; 109)
Note: p<0.1; p<0.05; p<0.01


Limitations


The map of geographical distibution of BISP coverage reveals a flaw in the variables chosen for this project that could not be fixed due to time restraints. Seeing that Sindh had greater BISP coverage in general than Balochistan, we can evidently see that access to state infrastructure is an important factor in access to BISP payments, as well as polling stations. While Urban Population can potentially act as an indicator for ability to access state infrastructure, it is now evident that population density would have been more appropriate for this project. This is as rural populations in the four provinces are put in the same category. However, a person in rural Balochistan (due to its difficult terrain and low population density) is far more removed from state infrastructure than someone in (for example) rural Sindh. Thus, population density would have been a better indicator.

Another limitation was in regards to research design. It would be better to view the impact of BISP coverage on voter turnout by adding a time dimension to the analysis. For example, obtaining data from two points in time and seeing if the regression coefficient for BISP coverage changed over time. However, this was not possible due to a lack of data as much of it comes out of the census. With the issue being that the last census was in 2017, with the one before that in 1998, and that BISP was launched in 2008. Thus, the 2017 census is the only point in time up till the present where all my variables (urban and total population, welfare coverage, voter turnout, and literacy) were available.

Finally, we cannot say that the relationship between BISP coverage and female voter turnout is causal. This is due to the fact that there are numerous other confounding variables that are probably unaccounted for. such as population density. Therefore, while both are correlated, we cannot say that a causal link exists.


Conclusion (Research and Policy Implications)

While this project tells us that BISP Welfare Support has a positive impact on female voter turnout, we cannot ascertain whether women are disproportionately voting for one particular party over another. However, we can make inferences from political dynamics in Pakistan, as well from the geospatial distribution of BISP payments. Firstly, it is important to note that Sindh, whose districts were seen to have a greater percentage of BISP beneficiaries, is also an electoral stronghold for the Pakistan People’s Party (PPP). This is important as BISP (Benazir Income Support Programme), was named after former Prime Minister Benazir Bhutto, who was assassinated in a terrorist plot and was also from the PPP’s Bhutto Dynasty. Therefore, the greater incidence of BISP payments in Sindh is important to analyse as BISP is used as electoral rhetoric by the PPP, especially as some people (especially in rural areas) are under the misconception that BISP payments come from Benazir Bhutto’s own wealth rather than the government (First Post Survey Activities Report, 2013). Therefore, are women under BISP welfare more likely to vote for the PPP? That is a potential research implication.

For policy implications, these findings may strengthen the support for BISP, especially from non-PPP parties. As poverty reduction is not just the only potential outcome for this programme, but also political empowerment, as well as better health and educational outcomes (as researched by other academics). This is especially important, given BISP’s recent renaming to Ehsaas-Kafaalat reduces its ties to the PPP and its ability to be used as electoral rhetoric by the party. This could potentially increase multipartisan support for the programme, and impact whether it continues to expand (whether in terms of payments or coverage of the population).


Bibliography

Ain’t No Stopping Us Now - Women as Policy Makers (2006). Abdul Latif Jameel Poverty action lab. Available at: https://www.povertyactionlab.org/sites/default/files/publication/43_Policy_Briefcase_1.pdf (Accessed: December 18, 2022).

Benazir Income Support Programme: Final Impact Evaluation Report (2016). Oxford Policy Management. Available at: https://www.opml.co.uk/files/Publications/7328-evaluating-pakistans-flagship-social-protection-programme-bisp/bisp-final-impact-evaluation-report.pdf?noredirect=1.

First Post Survey Activities Report (final) - bisp.gov.pk (no date). Available at: https://www.bisp.gov.pk/SiteImage/Misc/files/Cluster-A-First-Post-Survey-Activities-Report-Final(1).pdf (Accessed: December 18, 2022).

Haider, K. (2019) “Challenges to the electoral politics of PPP in Sindh in 21st Century,” Pakistan Social Sciences Review, 3(II), p. 647. Available at: https://doi.org/10.35484/pssr.2019(3-2)50.

Iqbal, T., Haq Padda, I.ul and Farooq, S. (2020) “SUSTAINABLE IMPACTS OF SOCIAL SAFETY NETS: The Case of BISP in Pakistan,” Pakistan Journal of Applied Economics, Vol.30 (No.2).

Mumtaz, Z. and Whiteford, P. (2017) “Social Safety Nets in the development of a welfare system in Pakistan: An analysis of the benazir income support programme,” Asia Pacific Journal of Public Administration, 39(1), pp. 16–38. Available at: https://doi.org/10.1080/23276665.2017.1290902.