In the developing world, NGOs frequently step in to fill gaps left by limited public spending on social assistance, health, and education. Yet, it remains uncertain whether they truly move the needle on development. While numerous studies evaluate individual NGO projects, they primarily focus on project implementation by NGOs versus governments (Henderson & Lee, 2015; Fitch-Fleischmann & Kresch, 2021; Mo et al., 2020; Cameron et al., 2019; Vivalt, 2020) -- leaving an additional critical question unanswered: does having more NGOs in a region lead to meaningful improvements in development?
In my job market paper, co-authored with Attique Ur Rehman, we tackle this question. Specifically, we ask: How does a reduction in NGO presence affect human capital outcomes? Are government services able to fill the gap? And what are the impacts on community-level outcomes, such as aggregate economic activity and social stability?
These are not easy questions to answer. NGOs often focus their efforts on high-need areas rather than being randomly distributed. Consequently, a naïve comparison of areas with more vs. fewer NGOs would likely produce downward-biased estimates.
A natural experiment: FATF compliance and NGO deregistration in Punjab
To address these selection issues, we leverage a unique natural experiment in Pakistan's Punjab province, where an unexpected government regulation led to a 25% decrease in the number of NGOs. The regulation was introduced as part of Pakistan’s compliance with the Financial Action Task Force (FATF) action plan to remove the country from its “grey list.” The FATF is an intergovernmental organization established to counter terrorist financing and money laundering. Countries identified as having weak financial safeguards are placed on the grey list, signaling heightened scrutiny. If a country fails to make improvements, it may face blacklisting and economic sanctions.
In 2018, the FATF placed Pakistan on its grey list. A key issue was the potential misuse of the non-profit sector by terrorist or extremist organizations, prompting increased scrutiny of NGOs. In response, the Pakistani government initiated a comprehensive review of the non-profit sector, requiring all local NGOs to submit financial records within a two-week deadline announced via newspaper. Anecdotal evidence suggests that many NGOs either missed this announcement or lacked the necessary documentation. Consequently, around 1,300 NGOs -- about 25% of all NGOs in Punjab -- failed to comply and were officially deregistered by mid-2019.
Data and methodology: Mapping the impact of NGO deregistration
This policy shift affected some areas more than others, with NGO deregistration varying across 144 areas (tehsils) in Punjab. We leverage this quasi-random variation in deregistration -- and, by extension, NGO presence -- to estimate the impact of NGOs on development outcomes. Using a generalized difference-in-differences approach, we compare tehsils more affected by the shock with those less affected, both before and after the shock.1
We digitize proprietary administrative data on both registered and deregistered NGOs in Punjab, geocode NGO locations to determine their tehsil of operation, and calculate the proportion of NGOs deregistered in each tehsil. The NGOs we study are small and primarily local, with operations typically concentrated within a single tehsil. This localized nature of NGO activity makes tehsil-level variation a meaningful measure of deregistration intensity. Using this continuous measure of the deregistration intensity, we categorize tehsils into "more-affected" (those with a proportion of deregistered NGOs higher than the median) and "less-affected" groups.
To support the validity of our identifying assumption, we show that key outcomes display parallel pre-trends. Moreover, we show that NGOs are not disproportionately deregistered in tehsils with specific baseline characteristics, such as poorer regions or areas led by opposition political leaders. Additionally, “more-affected” and “less-affected” tehsils exhibit similar baseline levels of information access, NGO presence, and remoteness.
For our outcomes, we draw from multiple data sources, including nationally representative household surveys, satellite data, annual statistical reports, and geo-coded conflict data, to answer the following research questions:
1. How does a decrease in NGO activity affect the likelihood of attending NGO-run schools, vaccination rates, and access to prenatal care?
2. To what extent do government services compensate for gaps in essential service delivery when NGO activity declines? Specifically, we examine whether children transition from NGO-run schools to government-run schools and how government healthcare provision adjusts when small health centers, often run by NGOs, decrease in number.
3. What are the broader community-level impacts of reduced NGO activity on aggregate economic activity, social stability, and violence against women?
The consequences of NGO deregistration on human capital
We find that human capital takes a hit. Figure 1 illustrates these impacts.
In Figure 1a, we observe that children aged 5 to 16 years (N = 201,541) in more-affected tehsils are less likely to attend NGO-run schools relative to children in less-affected tehsils; a year after deregistration, the likelihood of NGO school attendance drops by 67% in more-affected tehsils compared to less-affected tehsils. This effect is more pronounced for girls than for boys, with an 80% decrease for girls versus a 58% decrease for boys. The larger impact on girls makes sense: NGOs often prioritize removing barriers to girls’ education (Evans et al., 2021). When NGOs shut down, the support diminishes, raising the cost of educating girls and resulting in a sharper drop in their school attendance.
Figure 1: Impact of decrease in NGO activity on human capital outcomes
Figure 1b highlights the impact on healthcare. Rural women in more-affected tehsils are 10.5% less likely to receive prenatal care (i.e., whether the woman attended any medical consultations during her pregnancy) relative to rural women in less-affected tehsils, even a year after NGOs exited (N = 18,346). This is intuitive, as NGOs often fill critical healthcare gaps in rural areas where services are already limited. When NGO support is removed, access to prenatal care declines, which can have long-term consequences for the health of mothers and their children. We also find that reduced NGO presence decreases immunization rates for girls, particularly in rural areas.
Overall, our findings highlight how NGOs play a vital role in bridging gender gaps in healthcare and education.
Government steps up, but not enough
We find that while the government steps up, its efforts fall short. Students do switch from NGO-run schools to government-run schools at the primary and middle levels, but we do not find evidence of this substitution at the secondary schooling level. In healthcare, we observe adjustments in service provision. Following a decrease in the number of small health centers (i.e., those with fewer than 10 beds), often managed by NGOs, the government responds with a 15% increase in government-run health centers in areas most affected by the NGO exit relative to the less-affected areas. However, this increase is short-lived, with the government unable to sustain these efforts a year after the drop in NGO presence.
The ripple effects: Economic and social consequences
First, we observe a decline in aggregate economic activity in tehsils with higher rates of NGO deregistration. Using night lights intensity as a proxy for economic activity, we identify a significant downturn in areas more affected by NGO exits compared to those less affected, as illustrated in Figure 2 (N = 144 tehsils). Five years after deregistration, a one standard deviation (SD) decrease in NGO activity leads to a 0.09 SD decline in night lights intensity, corresponding to an 8% reduction in economic activity.
Figure 2: Impact of a decrease in NGO activity on night lights intensity
We find that social stability deteriorates more significantly in areas heavily affected by NGO deregistration compared to less-affected areas, with a 23% increase in demonstrations and riots observed five years after deregistration (N = 144 tehsils). Furthermore, violence against women rises disproportionately in more-affected areas, as evidenced by a sharp increase in shelter admissions for victims of extreme violence, particularly during and immediately after the COVID-19 lockdown. For instance, in October 2020, the number of victims of extreme violence in more-affected areas is 17% higher relative to less-affected areas. Although COVID-19 disrupted all areas similarly, those with fewer NGOs experienced significantly higher rates of violence against women. This highlights the vital role NGOs play in providing essential support and maintaining social stability during crises.
Policy implications and future work
Our findings show that NGOs truly move the needle on development. Policymakers can amplify this impact by fostering stronger cooperation between NGOs and the government. Implementing fair accountability measures would also allow NGOs to operate freely without restrictive regulations, maximizing their positive influence on development outcomes.
Additionally, fostering networks among NGOs could strengthen their collective impact. My future research will explore whether building connections between NGOs enables them to deliver services more effectively and address local development needs better than working in isolation.
Sarah Shaukat is a PhD candidate at Tufts University.
1 The analysis timeframe varies by outcome: yearly for human capital and nighttime lights (2019 “partially treated,” pre-post dummy for 2018), six-monthly for demonstrations and riots (pre-post dummy for early 2019), and monthly for violence against women (pre-post dummy for June 2019).
Join the Conversation