Can conditional cash transfers reduce vulnerability to climate change?

ABSTRACT

In the last decades several Conditional Cash Transfers (CCTs) have been implemented to interrupt the intergenerational transmission of poverty. More recently, they have also been used as a policy response to provide support to households hit by the COVID crisis. CCTs are being deployed as a development tool to reduce local communities’ vulnerability, increasing their resilience and capacity to adapt. In this context, this paper investigates if CCTs can reduce vulnerability to climate change by increasing the adaptative capacity of local communities. Indeed, while some literature has found that CCTs can reduce specific dimensions of vulnerability to socio-economic shocks, we argue that more research is necessary to understand their impact on vulnerability to climate change. Empirically, we use a panel of household survey data collected in Colombia in 2002 and 2006. We employ a difference-in-differences approach to analyse the impact of the CCT programme Familia en Acción (FA) CCTs on rural households. We consider seven components of vulnerability linked to climate change aggregated in an index: wealth, health, access to information, access to basic facilities, financial vulnerability, resilience to natural disasters, and nutrition. Our results differ from previous research and indicate that CCTs provided by the FA program do not decrease vulnerability to climate change. This finding suggests that CCTs are not a universal solution to reducing vulnerability to climate change and may be ineffective in specific contexts.

Key policy insights

• The literature often assumes that CCTs can have a mitigating effect on vulnerability to climate change. Our research shows that in the context of Colombia, this assumption is incorrect.

• The idea that providing monthly payments to households can reduce vulnerability to climate change should be considered with caution, given limited empirical evidence.

• Policy makers and scholars should consider the multidimensional nature of vulnerability and design targeted interventions and avoid considering CCTs as a panacea to all aspects of vulnerability to climate change.

• Our models show that the FA programme failed to have a significant impact in reducing vulnerability to climate change using several different specifications. We show that CCTs do not have an impact on numerous essential components such as wealth, health, access to infrastructure, information and knowledge, financial coping mechanisms, food consumption, and exposure to natural disasters.

KEYWORDS:

• Conditional cash transfer
• vulnerability
• Familia en Accìon
• climate change
• Colombia

Acknowledgments

The authors are grateful to Daniele Malerba for generously sharing the data on the Familia en Acción project with us and to the reviewers and editors of Climate Policy for their helpful and constructive comments. The authors remain responsible for any shortcomings.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Disclaimers

The views and opinions expressed in this article are those of the authors and do not necessarily reflect the views or positions of any entities and institution they represent or are affiliated to.

Notes

1 For the purposes of this paper, the term “vulnerability” refers to vulnerability to climate change.

2 s Note that our definition departs from those of previous studies in that it does not consider environmental factors of shock or stress as part of vulnerability. This is because it would be unreasonable to argue that CCTs could have an impact on these external stressors.

3 SISBEN is an indicator of economic well-being that is used in Colombia for targeting welfare programmes.

4 Eligible villages had to be <100.000 inhabitants as of 1999.

5 The data was later withdrawn from public access from the government website. Our analysis is based on the replication dataset provided from Malerba (2020). Without his support this paper would not have been possible.

6 Due to the low rate of mobility and the efforts made to follow moving households, the attrition rate was only 17%. Also, note that the surveys were conducted in 122 municipalities divided between treatment and control groups. Our final data contains households from 109 of these municipalities for which survey respondents were consistently available across the two waves.

7 Given that the control group is matched, we are able use an appropriate subset of control households for each group of treated ones.

8 Another issue is practical. Data for 2003 was no longer available to the authors of this paper.

9 Note that each dimension uses comparable indicators. These are either consist of a series of dummies, indicating whether the household has a particular attribute, or by expenditures that are inherently comparable. Note that to reduce the effect of outliers we take the natural log of all expenditure measures. For the ordinal variables that indicate natural disaster, we standardised the variables to account for different averages and standard deviation, then we added them together and rescaled to be bounded between 0 and 1.

10 Dimensions with only one indicator are simply standardised between 0 and 1.

11 Note that even the best performing household have a relatively high-level vulnerability given that, if a household performed perfectly in all the components (with a score of 1), the range of the FAVI index would be between 0 and 7.

12 Note that the control groups for the CT and ET households are drawn from municipalities matched by Attanasio et al. (2004) to have similar population size and an index of quality of life index, hence making our identification strategy more robust.

13 Note that this coefficient, while significant, is positive, suggesting that CCTs increase vulnerability. However, for the reason described in the main text, we believe that this finding should be interpreted as a sign of upward bias experienced in the early treated group rather than as an indication of a negative correlation between CCTs, and the health component of vulnerability.