Cash or In-kind Transfers? Evidence from a Randomised Controlled Trial in Delhi, India

TL;DR: In this article, a randomised intervention in Delhi, India, that provided unconditional cash transfers to a group of households as a replacement for the food security offered by a below-poverty-level card was examined.

Abstract: This article examines a randomised intervention in Delhi, India, that provided unconditional cash transfers to a group of households as a replacement for the food security offered by a below-poverty-level card. The experimental approach can differentiate beneficial effects due to either unconditional cash transfers or newly opened bank accounts. The unconditional cash transfer does not induce a decline in food security; rather, it provides opportunities for households to shift to other nutritious options in non-cereal product categories.

Summary

1. Introduction

• In recent years, cash transfer programmes have become increasingly popular in many developing countries.
• Moreover, cash-based welfare transfers have the virtue of being easier to administer than an in-kind welfare transfer, which is especially important in countries with weak institutional capacities to implement transfer programmes, such as in most sub-Saharan African countries and many Asian countries.
• The main reason many governments provide welfare transfers in-kind is their sense of paternalism (Currie & Gahvari, 2008).

2. Welfare Transfers In-Kind Versus in Cash

• In addition to the general advantages and disadvantages of cash transfers, different types could have differential effects.
• Arguing that the food subsidy is similar to an in-kind welfare transfer, the authors compare the impact of an in-kind transfer with that of a cash transfer.
• For a consumer 2, who previously consumed at X, the treatment improves the utility level.
• Theory cannot reveal – unless the authors know the exact preferences (or indifference map) – whether cereal consumption (that is those with curve 2) increases or decreases after the treatment.
• Vast literature emphasises the advantages of obtaining access to financial services (for example, savings, loans, insurance, credit, payments).

3. Experimental Design

• The unconditional cash transfer (CT) experiment started in mid-2010 in Raghubir Nagar (West Delhi).
• Successive NSS (National Sample Survey) data suggest a gradual decline in calorie intakes sourced from grains.
• The authors see a similar pattern in their target group where the calorie intake from cereals, pulses, edible oils, milk and sugar is 1583 Kcals a day on the average (calculated from their own survey).
• A fourth group, C3, comprised the 150 households that chose not to participate in the unconditional CT experiment.
• The authors also considered the households that originally did not want the intervention (C3 group); 22 per cent of this group requested the cash transfer in the midline survey.

4. Characteristics of Households in the Sample

• The authors experiment focuses on the sample of households that was willing to replace an in-kind welfare transfer (BPL card) with an unconditional cash transfer.
• Because the authors are observing self-selected households, self-selection bias should not be a problem, and internal validity should be guaranteed.
• It may be relevant to test if the randomised selection produced balanced groups.
• Even for this comparison, for nearly all the variables, the authors found no differences between households that self-selected for the CT and households that initially did not want a CT (group C3).
• The majority of these households included self-employed or regular wage earners; a minority relied on casual labour (defined as non-permanent labour, with no fixed employer or contract from any single employer) for their income.

5. Methodology

• The authors analysis seeks to determine the impact of the CT and the bank account.
• Because the randomised assignment process applies to a relatively small sample, some differences in unobservable characteristics may still exist.
• With their small sample, the authors include the control variables, measured in the baseline survey, to increase the precision of their estimates and control for any remaining differences across groups.
• Finally, as a robustness test, the authors estimated an analysis of covariance model and regressed the outcome variables (measured at the end-line) on the cash transfer dummies and the lagged outcome variable:.

6. Results

• In line with their previous explanation and Figure 1, their analyses focus on (1) the impact of the CT on cereal consumption and (2) the post-treatment adjustment in the consumption of non-cereal items.
• The main question thus is whether households buy more or less non-cereal food items or buy more ‘bad’ non-cereal items like alcohol.

6.1. Food Security

• To measure food security, the authors focus on the quantity of food consumed.
• The authors anticipate that the calorie value of the food consumed likely is affected by the substitution of cash transfers, and they use the ICMR (Indian Council of Medical Research) calorie conversion chart for this measure.
• From Figure 1, the authors already know that treated consumers may reduce their intake of cereals, that is, the items distributed through ration shops.
• The regression results indicate that this fear is probably unfounded.
• Those authors used NSSO 2004–2005 consumption data and compared BPL households with access to a ration card and those without such access.

6.2. Change in Consumption Patterns: Impact of CT on Cereal and Non-Cereal Consumption

• The authors next check whether the households in their sample, in line with Shaw and Telidevara (2014), shift from cereals to non-cereals.
• The cereal group comprises items such as rice, wheat, maize, jawar, bajra and ragi; the noncereal category consists of other nutritious food such as pulses, milk, eggs, fish and meat, fruits and vegetables.
• In the second columns in Tables 5–7, the authors find differences in the per capita calories consumed as cereal between the C1 and T groups.
• Across all estimation methods, the consumption of other cereals also is positively affected by the shift to the cash transfer.
• (PCNoncerealexp; third column in Tables 5–7) are always negative and significant.

6.3. Change in Consumption of Alcohol or Non-Food Items

• The analysis so far suggests that unconditional CTs do not disrupt food security.
• Other arguments against these welfare methods indicate concerns that consumers might spend cash transfers on private ‘bad’ goods, such as alcohol, that can be detrimental to the individual’s health and have negative impacts on family welfare.
• To test these potential effects, the authors measure ‘per capita expenditure on alcohol’ .
• Yet the authors find no significant differences across groups (see column 5 in.

7. Conclusion

• This article has presented the results of a cash transfer (CT) experiment in Delhi.
• The pilot CT identified a group of households that would receive direct, unconditional cash transfers in lieu of their existing food security methods, which were based on a PDS.
• The authors study is a rigorous attempt to address this specific issue.
• Therefore, unconditional cash transfers do not appear to compromise food security, nor do they induce households to increase wasteful expenses.
• At the least, the results suggest that providing poor households with the opportunity to choose between an unconditional cash transfer and an in-kind welfare transfer increases their welfare.

Figures

Figure 1. Impact of a cash transfer.

Table 7. ANCOVA regression

Table 6. Fixed effects regression

Table 1. Sample for the CT experiment

Table 5. OLS regression

Table 2. Characteristics of dropouts and respondents

Table 3. Balancing test between groups

Full text

University of Groningen
Cash or In-kind Transfers? Evidence from a Randomised Controlled Trial in Delhi, India
Gangopadhyay, Shubhashis; Lensink, Robert; Yadav, Bhupesh
Published in:
Journal of Development Studies
DOI:
10.1080/00220388.2014.997219
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from
it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2015
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Gangopadhyay, S., Lensink, R., & Yadav, B. (2015). Cash or In-kind Transfers? Evidence from a
Randomised Controlled Trial in Delhi, India.
Journal of Development Studies
,
51
(6), 660-673.
https://doi.org/10.1080/00220388.2014.997219
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the
author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license.
More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne-
amendment.
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the
number of authors shown on this cover page is limited to 10 maximum.
Download date: 10-08-2022

Cash or In-kind Transfers? Evidence from a
Randomised Controlled Trial in Delhi, India
SHUBHASHIS GANGOPADHYAY*
,
**
,†,‡
, ROBERT LENSINK
‡,§
& BHUPESH YADAV*
*India Development Foundation, Gurgaon, India, **School of Humanities and Social Sciences, Shiv Nadar University, Greater
Noida India,
†
Gothenburg University, Gothenburg, Sweden,
‡
Faculty of Economics and Business, University of Groningen,
Groningen, The Netherlands,
§
Development Economics Group, Wageningen University, Wageningen, The Netherlands
(Final version received October 2014; Final version accepted October 2014)
A
BSTRACT This article examines a randomised intervention in Delhi, India, that provided unconditional cash
transfers to a group of households as a replacement for the food security offered by a below-poverty-level card.
The experimental approach can differentiate beneficial effects due to either unconditional cash transfers or newly
opened bank accounts. The unconditional cash transfer does not induce a decline in food security; rather, it
provides opportunities for households to shift to other nutritious options in non-cereal product categories.
1. Introduction
In recent years, cash transfer programmes have become increasingly popular in many developing
countries. Such a programme provides a modest, regular amount of money, as cash paid to an
individual or family, to supplement any income the person or household may earn. The positive
effects of the conditional cash transfer programme Oportunidades (formally PROGRESA) in Mexico
have prompted many policymakers to consider the feasibility and desirability of (conditional) cash
transfers as a mechanism for reducing poverty, improving health and increasing school enrolment.
Thus a growing number of governments are introduci ng transfer schemes for some sectors of their
populations, and several have expanded them to national levels. In addition, various multilateral and
bilateral donor agencies actively support experimental cash transfer schemes in various parts of the
world.
Existing research provides some evidence that cash transfers help reduce extreme poverty. Yet
many questions remain regarding how cash transfers work and what their ultimate effects are
(Gaarder, 2010). In particular, it is unclear whether cash transfers really improve health outcomes,
and only limited evidence sugg ests the impact of cash transfers on school achievements (Garcia &
Hill, 2010). Perhaps the most controversial issue relates to the issue of conditionality, that is, whether
cash transfers should be co nditional on households’ compliance with a set of conditions or if they
should be unconditional. Most research suggests that conditional cash transfers are more effective for
reducing poverty than unconditional cash transfers, yet most of these studies focus mainly on
improving school enrolment rates in Latin America (Baird, McIntosh, & Ozler, 2011). Preliminary
results reported by Haushofery and Shapiroz (2013) from a study in western Kenya between 2011 and
Correspondence Address: Robert Lensink, Faculty of Economics and Business, University of Groningen, Groningen,
The Netherlands. Email: b.w.lensink@rug.nl
An Online Appendix is available for this article which can be accessed via the online version of this journal available at
http://dx.d oi.o rg/10.1080/00220388.2014.997219
The Journal of Development Studies, 2015
Vol. 51, No. 6, 660–673, http://dx.doi.org/10.1080/00220388.2014.997219
© 2015 Taylor & Francis

2012 suggest instead that unconditional cash transfers have substantial positive effects in terms of
poverty reduction.
A closely related issue pertains to whether welfare programmes should offer in-kind transfers or
cash transfers. In many countries, welfare transfers are in-kind, granted in different ways, such as
provisions for health care, housing, childcare or food. In India for example, food security for the poor
is assured by the government through the Public Distribution System (PDS), which grants households
that fall below the poverty line (BPL) a card that they can use to buy rice, wheat, sugar and kerosene at
subsidised prices in so-called ration shops.
A cash-based welfare transfer has s ome obvious advantages: If ma rkets fail, (un conditional) cash
transfers do not limit households to a certain type of expenditure, so in principle, they should improve
social welfare for the society as a whole, more so than conditional or in-kind welfare transfers
(Cunha, 2014; Feirrera, 2009). Moreover, cash-based welfare transfers have the virtue of being easier
to administer than an in-kind welfare transfer, which is especially important in countries with weak
institutional capacities to implement transfer programmes, such as in most sub-Saharan Afr ican
countries and many Asian countries. In India, Dutta and Ramaswami (2001) caution about the
significant leakages from PDS. However, welfare transfers in the form of cash also may induce
moral hazards or expenditures on so-called temptation goods, such as alcoho l, tobacco and g ambling.
The main reason many governments provide welfare transfers in-kind is their sen se of paternalism
(Currie & Gahvari, 2008). Paternalistic governments prefer to induce increased consumption of
certain types of goods, so they provide in-kind transfers. Welfare transfers in-kind also can help
ensure that individual purchase decisions match both individual and societal preferences. If a
particular decision, such as to pursue an education, affects wider society, but the person makes
decisions solely on the basis of his or her individual preferences, underinvestment might occur at the
societal level.
This article seeks to extend the discussion of welfare transfers in-kind versus in cash. Dutta, Howes,
and Murgai (2010) compare two such Indian welfare programmes and find that the cash transfer
programme is more efficacious, in terms of progress and targeting. However, a local newspaper
(Cherian, 2013) also has challenged the replacement of the PDS with a cash-transfer scheme. The
ensuing public debate led to some requests for the judiciary to intervene, to stop any dilution of the
PDS with cash transfers. Proponents of cash transfers cited economic theory and empirical studies
demonstrating leakage and corruption in the PDS; opponents expressed fears that the money would be
misused to buy temptation goods. Yet neither side had access to systematic evidence about the
outcomes. This article examines, for the first time, the impact of cash transfers in lieu of PDS with
a randomised intervention in Delhi, India; the focal experiment provided cash transfers to a random
selection of BPL households.
The results re veal that replacing an in-kind we lfar e tr ansfe r with a welfare tr ans fer in cash does
not decrease food security. A s men tioned, a common criticism of cash transfers is that households
spend less on food and more on other things, which could lead to a loss of nutrition and harmful
long-term effects, especially among children. Our study provides no eviden ce along t hes e lines.
Rather, we find that cash transfers provide opportunities for househ olds to shift to other nutritious
options in non-ce real product catego ries, whereas the BPL card primarily subsidises cereal items,
such as rice and wheat. With the cash transfe r, BPL house hol ds can in crease the ir intake o f other
nutritious food items, such as pulses, milk, egg, fish, meat, f ruits a nd vegetables. I n line with Cunha
(2014), who examines t he Mexican food assistance programme Programa de A poyo Alimentario
(PAL), we also find no evidence that poor households use cas h transfers in socially detrimental
ways. These positive outcomes, coupled wi th the comparative ease of targeting and reduced leakage
through cash t ran sfers, should encour age governments to introduce such welfare programmes w hen
possible.
The remainder of this article is structured as follows: Section 2 contains a brief theoretical
discussion of welfare transfers in-kind versus welfare transfers in cash. After we explain the experi-
mental set-up in Section 3, we provide descriptive statistics in Section 4. Section 5 outlines our
methodology; Section 6 contains the main results. Finally, we conclude in Section 7.
Cash or in-kind transfers 661

2. Welfare Transfers In-Kind Versus in Cash
In addition to the general advantages and disadvantages of cash transfers, different types could have
differential effects. For example, some cash transfers are targeted on the basis of income, usually
intended for ‘the poor’ only; other schemes provide cash to all, regardless of their recorded income or
activity status. Both systems have drawbacks.
1
Because the different types of cash transfers are beyond
the scope of this investigation, we do not offer an overall survey of cash transfers here. Instead, we
study the effects of substituting a cash transfer for a food subsidy, in this case, the BPL card. Arguing
that the food subsidy is similar to an in-kind welfare transfer, we compare the impact of an in-kind
transfer with that of a cash transfer.
In our experiment (see Figure 1), we assume that the subsidy, or in-kind transfer, refers to cereals.
For all other commodities and services, no subsidy (or in-kind welfare transfer) is available. The
horizontal axis in Figure 1 measures the consumption of cereals, and the vertical axis measures the
consumption of all other commodities and services (composite commodity, money, numeraire good).
The figure compares the post-transfer situations for in-kind and cash transfers. For the BPL card (that
is, in-kind transfer), the consumer starts with A units of money; by buying cereals, she travels along the
kinked budget line ABC. For the first few units of cereals, she pays a lower price, as given by the slope
of the segment AB. This low price is a subsidised price for people entitled to buy from the ration shops.
However, after the consumer has bought an amount b of cereal, she is not entitled to this low price
anymore; if she wants to consume more cereal, she must pay the higher market price. Thus in Figure 1,
segment BC is steeper than AB. The changing price of cereals beyond b produces the kink at B.
Money
(non-cereal)
Cereal
A
D
C
B
E
O
1
2
X
Y
3
P
b
Figure 1. Impact of a cash transfer.
662 S. Gangopadhyay et al.

In our experiment, when the BPL card is replaced by an unconditional cash transfer, the consumer’s
budget line changes from ABC and starts to look more like a conventional, textbook version, EBC.
Removing the food subsidy and providing a cash transfer instead moves the consumer’s budget line to
EBC. This change to the budget line enables the consumer to purchase a basket B both before and after
the treatment. Unconditional cash is weakly preferred to the in-kind transfer. Moreover, the difference
between what the consumer bought before the treatment and what she purchases after the treatment
depends crucially on the initial purchase (that is, whether the consumer was to the left or right of B).
Suppose that the consumer’s indifference curve (iso-utility line) is 3, which is tangential to the
segment BC at P and is the highest such curve that the consumer can reach and still appear on the
budget line ABC. When we change her budget line to EBC, 3 continues to be her highest attainable
indifference curve (that is, greatest utility level), so she stays at P. That is, she remains where she was
before the treatment. However, for a consumer 2, who previously consumed at X, the treatment
improves the utility level. This consumer moves to a new point of tangency, between the new budget
line and a higher utility level at a point such as D. The only additional consideration is that it is entirely
possible for someone to be at B when the budget line is ABC and continue there, even when it changes
to 1. In that case, the curve is tangential to EBC at B.
2
From our representation of point Y, it may appear that cereal consumption in Y is the same as that of
X. However, theory cannot reveal – unless we know the exa ct preferences (or indifference map) –
whether cereal consumption (that is those with curve 2) increases or decreases after the treatment.
Rather, it simply suggests that 3 (that is, those with curve 3) does not change cereal consumption and 1
definitely decreases it.
Beyond this initial investigation of the data generated by the experiment, we also address the post-
treatment shift in the consumption of non-cereal by consumer types 1 and 2. Do they buy more or less
of non-cereal food items, or do they buy more ‘bad’ non-cereal items, such as alcohol and gambling?
Finally, we note that in our experiment, the cash transfer went automatically into a bank account.
Opening a bank account could have beneficial effects by itself, beyond the individual welfare-
increasing effects of the cash transfer, by initiating a process of ensuring access to financial services
among the poor. Vast literature emphasises the advantages of obtaining access to financial services (for
example, savings, loans, insurance, credit, payments). By providing these financial services, the
financial sector can effectively help the poor escape poverty, such as when loans increase their level
of consumption and minimise risk. Therefore, we ensure that our experiments differentiate b etween the
effects of the cash transfer and the outcomes of opening a bank account. Before presenting these
effects, we explain our experimental design.
3. Experimental Design
The unconditional cash transfer (CT) experiment started in mid-2010 in Raghubir Nagar (West Delhi).
Delhi is a large city, with a food supply infrastructure that is much better organised than in many cities
in India. This location was purposeful, because in our effort to observe changes in household
behaviour, we did not want households to feel restricted in their access to alternative food supply
sources. Conducting the experiment in Delhi ensured that respondents had easy access to private
stores, in addition to ration shops.
On the other hand, the population here are representative of the general patterns regarding nutrition
among the poor. For example, successive NSS (National Sample Survey) data suggest a gradual
decline in calorie intakes sourced from grains.
3
The standard set for urban India was 2200 Kcals a day
per adult. Almost all of it was supposed to come from cereals and grains. However, with time, various
new food items have allowed people to substitute away from grain and consume more of fruits,
vegetables, milk, etc. We see a similar pattern in our target group where the calorie intake from cereals,
pulses, edible oils, milk and sugar is 1583 Kcals a day on the average (calculated from our own
survey). The rest of the calorie intake comes from vegetables, fruits, meat and eggs (items for which
quantities are difficult to get).
Cash or in-kind transfers 663

Frequently asked questions

Q1. What have the authors contributed in "Cash or in-kind transfers? evidence from a randomised controlled trial in delhi, india" ?

This article examines a randomised intervention in Delhi, India, that provided unconditional cash transfers to a group of households as a replacement for the food security offered by a below-poverty-level card. The unconditional cash transfer does not induce a decline in food security ; rather, it provides opportunities for households to shift to other nutritious options in non-cereal product categories. 

Q2. What was the effect of the CT intervention on households?

Their CT intervention included an exit option for all recipients after six months, which was important because the CT replaced a public programme to which households already had access. 

Q3. What are the main reasons why governments provide welfare transfers in-kind?

welfare transfers in the form of cash also may induce moral hazards or expenditures on so-called temptation goods, such as alcohol, tobacco and gambling. 

Q4. How many BPL families did they choose to participate in the experiment?

Of the 690 BPL families introduced to the experiment through the awareness programme (as of 1 September 2010), 362 chose to participate, 261 decided not to participate and 67 expressed indecision. 

Q5. What is the main reason why Indian welfare transfers are in-kind?

If markets fail, (unconditional) cash transfers do not limit households to a certain type of expenditure, so in principle, they should improve social welfare for the society as a whole, more so than conditional or in-kind welfare transfers (Cunha, 2014; Feirrera, 2009). 

Q6. What is the main reason why Indian governments provide welfare transfers in-kind?

cash-based welfare transfers have the virtue of being easier to administer than an in-kind welfare transfer, which is especially important in countries with weak institutional capacities to implement transfer programmes, such as in most sub-Saharan African countries and many Asian countries. 

Q7. How many households were selected to participate in the study?

Through random selection (conducted on 5 September 2010), the authors identified 350 of the 362 willing households to receive cash transfers; of the 261 non-participating households, the authors randomly selected 150 households to investigate. 

Q8. What is the main reason why governments provide welfare transfers in-kind?

Welfare transfers in-kind also can help ensure that individual purchase decisions match both individual and societal preferences. 

Q9. Why is it difficult to gain a clear idea of the quantities consumed?

Because the non-cereal segment contains such heterogeneous products, it is difficult to gain a clear idea of the quantities consumed. 

Q10. Why do the authors include the control variables in the baseline survey?

With their small sample, the authors include the control variables, measured in the baseline survey, to increase the precision of their estimates and control for any remaining differences across groups. 

Q11. Why did households drop out of the study?

the authors gave them the option to go back to the PDS; the midline survey results indicated that only four households that received the CT (4%) did not want to continue, so theydropped out of the analysis. 

Q12. What is the effect of opening a bank account?

Opening a bank account could have beneficial effects by itself, beyond the individual welfareincreasing effects of the cash transfer, by initiating a process of ensuring access to financial services among the poor.