Political Responses of Conditional Income Transfer Recipients: A Mechanism Approach

Abstract

This chapter studies whether participation in conditional income transfer programmes in Latin America generates observable political responses and whether these responses indicate improvements in the political inclusion of participants. It adopts a causal mechanism approach to study political responses to transfer receipt, seeking to cast light on the links existing between transfer receipt and political outputs and outcomes among recipients. A review of available literature helps identify three possible causal mechanisms: a support for redistribution mechanism; a bureaucratic mechanism encouraging political engagement; and a cognitive change mechanism. Analysis of empirical counterparts using attitudinal data from the AmericasBarometer for the period 2010–2019 confirms the relevance of these mechanisms for our understanding of political responses of conditional income transfers recipients.

1 Introduction

The emergence of social assistance, and particularly large-scale conditional income transfers, is the dominant factor in the recent dynamics of welfare institutions in low and middle-income countries (Barrientos 2013; Fiszbein and Schady 2009). A longstanding finding in the poverty literature is that disadvantaged groups face restrictions in their capacity to influence political decisions, which reinforces their disadvantage over time. Limited political inclusion is a key dimension of their disadvantage. This chapter studies whether participation in conditional income transfer programmes in Latin America generates observable political responses and whether these responses imply improvements in the political inclusion of participants.¹

Studies on welfare institutions in high-income countries suggest that social insurance institutions mobilise participants to support and protect their institutions (Baldwin 1990), although research on attitudes to the welfare state have mixed findings (Svallfors 2012; Laenen et al. 2020). Studies on social assistance, on the other hand, suggest that programme participation has net negative effects on political participation. Bruch et al. (2012) compare welfare programmes in the USA and find that “policy designs can have significant effects on civic and political engagement among the poor; the feedback effects of meanstested programs can be positive as well as negative; and such effects tend to be more positive when a policy’s authority structure reflects democratic rather than paternalist principles” (Bruch et al. 2012, 210).

In Latin America, the growth of large-scale social assistance this century raises the issue of whether it leads to political outcomes likely to result in the political inclusion of participants.² Within social assistance programmes, conditional income transfers³ have attracted particular attention due to the availability of impact evaluation and administrative data which make it possible to study this issue in a quasi-experimental context. Variation in programme implementation, a staggered territorial implementation for example, helps shape treatment and control groups. Comparisons across these groups support reliable identification of programme effects. Researchers have applied this methodology to the study of electoral effects stemming from conditional income transfer programme participation. Studies find higher electoral registration, voting, and political participation among participants compared to control groups (Nupia 2011; De la O 2013, 2015; Baez et al. 2012; Linos 2013).⁴ Another group of studies adopts a quasi-experimental approach to attitudinal data in order to identify political effects associated with participation in conditional income transfer programmes (Layton et al. 2017; Layton and Smith 2011, 2015; Zucco 2008, 2013).

In the context of assessing political inclusion, however, this black box approach has important limitations. With few exceptions, existing studies have focused on programme participation outputs such as registration and voting, but less attention has been paid to political outcomes such as political inclusion. Distinguishing between political outputs and political outcomes from programme participation is important. Electoral registration is an output of conditional income transfers when this is required for programme entry, while political inclusion is an outcome when it reflects greater political engagement among programme participants.⁵ This chapter adopts a causal mechanism approach to study political responses to transfer receipt. A mechanism approach considers how interventions come to have observed effects and in doing so it pays attention to outputs and outcomes. In the context of this chapter, a mechanism approach seeks to cast light on the causal links existing between transfer receipt and political inclusion outcomes.

The analysis in this chapter reviews the existing literature on political responses to conditional income transfer programme participation with a view to identify potential causal mechanisms which might shed light on the issue at hand. The analysis focuses primarily on recipients and on their responses to the receipt of the transfers.⁶ After identifying three causal mechanisms, they are assessed against attitudinal data from Latin America, from the AmericasBarometer (Latin American Public Opinion Project (LAPOP) 2020) for the period 2010–2019. Most Latin American countries have implemented large-scale conditional income transfers, justifying a regional focus.

This chapter is organised as follows: Sect. 2 justifies the application of a causal mechanism approach and considers its main elements; Sect. 3 reviews the available literature on political responses to conditional income transfer receipt in the region; Sect. 4 identifies alternative causal mechanisms explaining political responses; Sect. 5 assesses the presence of these mechanisms using data from LAPOP; a final section concludes.

2 Why a Causal Mechanism Approach?

Interest in causal mechanisms as an approach to studying social relationships reflects a concern with the limitations of “black box” statistical/correlation approaches to explanation in the social sciences (Elster 1998; Hedström and Ylikoski 2010). A focus on causal mechanisms is motivated by the need to study the generative processes underpinning the social relationships of interest. Identifying the mechanisms at work helps trace causal links existing between initial conditions, behavioural responses, and particular outcomes.

Adapting Coleman’s boat (Coleman 1990) in Fig. 13.1, the dashed line represents the observed correlation, while the effect of the transfer on recipients and their eventual political response is represented by the solid arrows. The generating mechanism has three main elements (themselves commonly referred to as mechanisms in the literature). Transfer Programme stands for the policy implemented, which signals a change in resource distribution with a subsidy flowing to selected participants (S). Recipients are expected to react to transfers with specific behavioural change or action (A), in our particular context, increased electoral participation for example. Recipients’ behavioural change leads to a transformation of the social situation (T), in this case political inclusion.⁷ It is helpful to consider the generating process as consisting of inputs, outputs, and outcomes, matching the elements of the mechanism described above. The causal mechanism is therefore a representation of the generative causal links existing between inputs, outputs, and outcomes. This is often missing in statistical/correlation approaches or is only mentioned in passing by reference to simplistic behavioural models.

Fig. 13.1

Coleman’s boat adapted. (Source: Own presentation)

It should be acknowledged that the literature on causal mechanisms is variegated and a number of distinct approaches to their use are available (Hedström and Ylikoski 2010). Hedström and Ylikoski note that a

mechanism based explanation describes the causal processes selectively. It does not aim at an exhaustive account of all details but seeks to capture the crucial elements of the process by abstracting away the relevant details. The relevance of entities, their properties, and their interactions is determined by their ability to make a relevant difference to the outcome of interest. (Hedström and Ylikoski 2010, 53)

This applies directly to the analysis in this chapter.

Some approaches to causal mechanisms avoid the use of statistical analysis (Mayntz 2003; Morgan and Winship 2015; Mayntz 2020). But, as a matter of fact, the processes under study here are probabilistic in the aggregate, given the large number of influencing and contextual variables affecting observed heterogeneity in political responses (Elster 1998). Analysis of empirical counterparts will necessarily involve statistical analysis. The main objective of the analysis in this chapter is not restricted to finding out how the effects are feasible, nor solely to establishing the conditions under which the effects might be possible, rather it is to seek to establish whether the mechanisms are in actual fact present. This is in line with Morgan and Winship who “take the position that genuine casual depth must be secured by empirical analysis” (Morgan and Winship 2015, 346).

Applying a causal mechanism approach to the study of the political responses to conditional income transfers could make an important contribution. One feature of research on the expansion of social assistance programmes in low and middle-income countries has been the use of quasi-experimental methods to identify the effects of participation in transfer programmes. This applies particularly to conditional income transfer programmes (Barrientos and Villa 2015). Often, anti-poverty policies meet considerable political resistance from elites. Impact evaluation studies have provided a means of protecting anti-poverty policies by showing how effective they are (Levy 2006). Multilateral organisations and bilateral donors have supported impact evaluations in order to show aid effectiveness to distant electorates (Barrientos and Villa 2015). Impact evaluation studies identify a treatment and a control group and compare their outcomes before and after the implementation of the programme (Ravallion 2005). Normally evaluation data focus on the particular objectives of the conditional income transfer programmes, for example improvements in consumption and utilisation of basic services, health, and education.⁸

Researchers have made use of evaluation data to study political effects of programme participation. This literature exploits information available from survey questionnaires and, where possible, matches evaluation and electoral data. Studies of conditional income transfer programmes in Latin America find higher electoral registration, voting, and political participation among participants (the treatment group) compared to non-participants (the control group) (Nupia 2011; De la O 2013, 2015; Baez et al. 2012). These findings have been confirmed with data collected independently (Schober 2019).

The findings from these studies are very informative and provide a precise measure of causal effects, but they shed limited light on how these effects are generated. For the purposes of assessing the capacity of conditional income transfers to generate political inclusion, information on the generating processes or causal mechanisms is essential.

3 Findings from the Latin American Literature

In this section we review the literature on political responses to participation in conditional income transfer programmes with the aim of extracting information and hypotheses on potential causal mechanisms. This literature is concerned with identifying the effects of programme participation on a range of political variables, but it seldom tracks the processes by which these effects come to be. Its interest is the “what” rather than the “how”. Causal mechanisms are often implicit in the postulated relationships, which the review will attempt to highlight.

Hunter and Sugiyama (2014) provide an excellent perspective on political responses to participation in conditional income transfer programmes based on extensive research in the Northeast of Brazil. Relying on interviews and focus group data they investigate whether Bolsa Família receipt enhances citizenship. Their main hypothesis is described as follows: “Social benefits, acquired through procedures that are judged to be reasonable, fair and transparent can only deepen poor people’s appreciation of their newfound political rights” (Hunter and Sugiyama 2014, 829).

They discuss two main pathways to improved citizenship, which will be interpreted for our purposes as causal mechanisms. First, transfer receipt enhances citizenship through social inclusion. Transfers signal to recipients that governments “recognise all members of the national community as worthy enough to have their basic needs met and life chances lifted” (Hunter and Sugiyama 2014, 830). Second, transfer receipt strengthens the agency of recipient households in as far as they “foster a sense of recognition, fairness, and rights [that] facilitate the exercise of ‘voice’, which helps citizens hold their government accountable for meeting basic needs” (Hunter and Sugiyama 2014, 830). Their research confirms the presence of both causal mechanisms. The strengthening of agency shows up among respondents who highlight greater economic independence associated with transfer receipt. Voice and accountability arise strongly in the context of recipients’ perceptions of the rules-based nature of the programme and their preparedness to engage in political activity to support the programme and resist attempts to dismantle it.

Schober (2019) is interested in the question whether conditional income transfer receipt leads to increasing political participation. Using a dedicated survey of a representative sample of adults in three municipalities in Mexico with high levels of poverty, he explores political participation outcomes and pathways, which will be interpreted as causal mechanisms. He then extends his analysis to other Latin American countries participating in LAPOP in 2014.⁹ His study contrasts political participation effects of conditional and unconditional income transfers.

His underlying theoretical framework is a rational choice model of political participation emphasising the role of benefits and costs. Transfers influence political participation by changing the balance of costs and benefits. Conditional income transfers raise the benefits of political participation because of the likelihood that participation will protect benefit receipt in the future. Additionally, he argues that conditions have positive net effects on political participation. On the one hand compliance with conditions will reduce the time available to transfer recipients to engage in political activity, say mothers taking over housework duties previously performed by their daughters. On the other hand, conditions involve greater engagement with other recipients and officials leading to improved organisational and communication skills. He describes these as “non-material resources”, which act to reduce the costs of political participation. For my purposes, the rational choice model can be dropped in order to focus attention on the observable causal mechanisms identified here.¹⁰

The study finds that conditional transfer recipients are more likely to engage in a greater number of political activities (voting, contacting public officials, community activism, and civil society engagement) compared with unconditional transfer recipients in the sample who are only likely to engage in a single activity (voting).¹¹ Looking further into possible pathways, the study finds that conditional transfer “participation is positively associated with each measure of civic skills, whereas UCT [unconditional income transfer, AB] participation is only positively associated with attending meetings. In terms of the substantive effect, it is estimated that CCTs [conditional income transfers, AB] lead to a 59 percent increase in the exercising of civic skills for the average respondent, while UCTs only lead to an increase of 29 percent” (Schober 2019, 598–99). He does not rule out the possibility of alternative pathways or causal mechanisms but finds no supporting evidence for alternative causal mechanisms put forward in the literature.¹²

Analysis of LAPOP cross-country data for 2014 confirms that for conditional income transfers, “participation is positively and significantly associated with several modes of political participation, including campaign activism, contacting public officials, civil society engagement, and protest” (Schober 2019, 602).

Layton and Smith (2011) study the electoral outcomes from transfer receipt using 2012 LAPOP data for twenty-three countries. Like Schober, they start with a basic rational choice model in which transfers influence the balance of costs and benefits. As they put it, “by altering beneficiaries’ pocketbook calculations and directly linking recipients’ well-being with state actors and policies, social assistance has the obvious potential to alter voting behaviour” (Layton and Smith 2011, 855). They pay particular attention to programme conditions as a source of political effects.

The paper correlates conditional income receipt with voter choice and turnout preferences. Four of the institutional conditions influencing preferences are modelled directly: enforced compulsory voting, programme conditionality, ideological leftist president, and programme independence. In terms of causal mechanisms, meaning how these effects are generated, the paper only provides a basic discussion. It makes reference to the possibility that: “Familiarization with state bureaucracies could also reduce the psychological or cognitive costs of turning out at the polls for marginalized families with very few prior positive experiences with the state” (Layton and Smith 2011, 859). It also suggests that “benefits may increase psychological attachment to the state and national politics … Once they discover that the state can directly meet their needs, assistance recipients gain a new stake in political contests” (Layton and Smith 2011, 859).

Layton et al. (2017) examine whether the observed political effects of participation in Bolsa Família constitute reciprocal support for incumbents (transfers for votes) or reflect broader positive attitudes to state legitimacy. The study relies on LAPOP data for Brazil for the period 2007–2014. Methodologically, the study compares attitudes among Bolsa Família recipients with a control group constructed using propensity score matching techniques.¹³ The authors adopt a “thick” notion of state legitimacy requiring “attachment to a political object for its own sake … independent of system outputs” (Layton et al. 2017, 102), also referred to as diffuse forms of legitimacy. Their main finding is that “the immediate political consequences of CCTs are not limited to positive electoral results from incumbents …. [but] we find little support for the hypothesis that Bolsa Família is associated with diffuse dimensions of legitimacy … Brazil’s CCT recipients are not any less supportive of diffuse elements of the political regime than nonrecipients” (Layton et al. 2017, 113).

The operationalisation of state legitimacy in the context of the LAPOP is of particular interest for our purposes. An index of political legitimacy combines six dimensions: support for incumbent; support for local government; support for core political institutions; support for core regime principles; regime economic performance; and sense of political community. This is a complex index with some dimensions themselves an index of specific survey responses. Placed against a causal mechanism framework, the legitimacy diffusion index has the disadvantage of mixing inputs, outputs, and outcomes. For example, incumbent support is measured additively in terms of the perceived efficacy of incumbents on three issues: whether they combat corruption, improve citizen security, or manage the economy well. Support for core political institutions is operationalised through an additive index of eight survey responses, including trust in state institutions, respect for political institutions, and support for the political system. This dimension is closest to describing an outcome measure of a causal mechanism. Their estimation finds that “across all survey years, Bolsa Família recipients report significantly higher levels of support for core political institutions, trust in local government, and support for political actors than their matched nonrecipient peers” (Layton et al. 2017, 109).

The review of this literature shows that most studies reference causal mechanisms, even if only implicitly. Where referenced, causal mechanisms are mainly accessories to the estimation of correlates and their interpretation. The review drew attention to four potential mechanisms structuring recipients’ political responses to transfer receipt: (i) reciprocal support for incumbent mechanism (or votes for transfers); (ii) a support for redistribution mechanism encouraging electoral participation in support of incumbent redistributive policies; (iii) a bureaucratic mechanism encouraging political engagement by transfer recipients through the implementation features of the transfer programmes, for example birth and electoral registration, compliance with programme conditions, interactions with programme agencies; and (iv) a cognitive change mechanism in which transfers facilitate a cognitive change in recipients as regards their societal and political inclusion.¹⁴

4 Causal Mechanisms Structuring Political Responses to Transfer Receipt

This section discusses in more detail the four potential mechanisms identified in the previous section. In fact, applying the causal mechanism approach rules out reciprocal support for incumbents as a causal mechanism. The discussion starts with this point.

4.1 Reciprocal Support for Incumbents Is not a Causal Mechanism

The literature on political responses to transfer receipt often finds relatively higher support for the incumbent among participants in conditional income transfer programmes (when compared to non-participants) (De la O 2015; Zucco 2013; Díaz-Cayeros et al. 2016; Baez et al. 2012). From the perspective of political coalitions, this finding appears to confirm the view that conditional income transfers are primarily an electoral strategy open to policymakers. One advantage of approaching political participation from the perspective of recipients is to raise alternative interpretations for this finding.¹⁵ A pocketbook approach to electoral politics views politicians as entrepreneurs placing a “retail offer” of policies to voters. Voters rank the policy portfolios in line with their pocketbook and vote accordingly. In the context of social assistance transfers, participants can be seen to reward incumbents that provide them with public transfers. A condition for this to work is that transfer recipients associate the transfer with a particular incumbent. This condition is easily met in conditional income transfers. An alternative interpretation is that recipients vote for incumbents in order to protect the programmes that provide benefits to them. They are concerned that if an opposition candidate is elected, they will drop or reform the transfer programme. This was mentioned in Hunter and Sugiyama’s research (Hunter and Sugiyama 2014). Electoral competition in the context of re-election motivates political leaders to search for additional votes.

The postulated relations do not amount to a mechanism. This follows from the description of causality in directed graphs. Causality requires, inter alia, that directed graphs are acyclic. In directed acyclic graphs, “no directed path [the direction of the arrows] emanating from a causal variable also terminates at the same variable … ruling out representations of simultaneous causation and feedback loops”, see Morgan and Winship (2015, 80). Figure 13.2 depicts the links. Arrows show causal links. They indicate that recipients vote for incumbents in order to preserve the transfer programme or as a reward for the benefits, while politicians run the programme as a means of eliciting electoral support from participants. Note the circular nature of the relationships implying that no causal direction could be identified by relating transfer receipt and higher incumbent support after the programme has started running. The transformation component that is present in Fig. 13.1 is missing in Fig. 13.2 as the original social situation is reinforced by recipients’ actions in response to the transfer.

Fig. 13.2

Reciprocal support for incumbent. (Source: Own presentation)

4.2 Support for Redistribution Mechanism

A causal mechanism can be proposed when recipients vote for the incumbent because they acknowledge, qua citizens, that the incumbent is committed to poverty reduction and social justice. This mechanism entails several conditions: that recipients can identify the relevant politicians or coalitions reasonably well and can be reassured that voting for incumbents does not generate negative trade-offs with other policy fields, say economic policy for example. Ideological or partisan loyalties with the governing coalition could help strengthen this mechanism.

Figure 13.3 shows the causal links. The dashed line describes the finding from correlation studies on the presence of a transfer programme effect on support for incumbents. The figure shows in more detail the main elements of the causal mechanism at work. Recipients are shown to vote for incumbents supporting transfer programmes because of their commitment to the reduction of poverty and inequality. Isolating the links existing between recipients, incumbent support, and support for redistributive policies now shows acyclic causal links, compared to the cyclical relationships in Fig. 13.2. Note that the support for redistribution mechanism rejects the proposition that recipients vote for incumbents solely because of the transfer, as in the reciprocal support hypothesis discussed above. Recipients are unlikely to vote for politicians who might run conditional income transfer programmes as a pure electoral strategy but who otherwise fail to support wider redistribution policies.

Fig. 13.3

Support for redistribution mechanism. (Source: Own presentation)

4.3 Bureaucratic Mechanism Encouraging Political Engagement

The review of the literature on political responses to conditional transfer programme participation suggests that, in the context of restricted citizenship for disadvantaged groups well established in the literature, conditional income transfers work to raise political engagement and extend citizenship (Hunter and Sugiyama 2014). The predicted outcome is increased support for political institutions among low income groups.

Political engagement associated with programme participation might work through alternative channels. A bureaucratic channel operates through the requirements for birth and electoral registration embedded in the implementation of the transfer programmes (Baez et al. 2012; De la O 2015; Hunter and Sugiyama 2018). In particular, electoral registration is expected to generate improvements in turnout and engagement in associated electoral activities. Figure 13.4 depicts the set of relationships posited by this mechanism.

Fig. 13.4

Bureaucratic mechanism encouraging political engagement. (Source: Own presentation)

5 Cognitive Change Mechanism

A different channel identified in the literature works through cognitive changes associated with programme participation. This is related to recipients’ engagement with programme agencies and local politicians involved in the implementation of the programme (Schober 2019). An ideational channel works through the fact that transfers strengthen perceptions among recipients that government policy acknowledges their disadvantage and is committed to addressing it. In so doing recipients are recognised as full members of a polity, as citizens, with associated rights (Hunter and Sugiyama 2014). An outcome of this mechanism is that recipients are likely to show support for political institutions (compared to non-recipients). Figure 13.5 depicts the relationships involved in this mechanism.

Fig. 13.5

Cognitive change mechanism. (Source: Own presentation)

In the next section, the three mechanisms will be assessed with attitudinal data from Latin America. Before doing so, it will be helpful to underline the main contribution of the causal mechanism approach. Applying a mechanism perspective helps to clarify the causal links existing between conditional income transfer programmes, behavioural responses, and outcomes. The discussion in this section points to three important contributions.

First, while acknowledging the valuable insights from the existing literature on the political effects of large-scale transfer programmes, an explicit focus on causal mechanisms rectifies two weaknesses in this literature: (i) its one-sided focus on policymakers; and (ii) its inattention to the transformational components of social assistance. Shifting the focus onto transfer recipients and paying attention to outcomes extends the causal links in productive ways. The existing literature is perhaps too focused on correlating inputs and outputs, much less so on outcomes.

Second, applying a causal mechanism framework shows that the most common finding in the literature, recipients’ reciprocal electoral support for incumbents, does not in itself support causal inference. A causal mechanism approach shows that the transformational component is missing.

Third, paying attention to how the transfers may influence recipients’ political engagement is essential to our understanding of the broader impact of conditional income transfers on the (political) inclusion of disadvantaged groups.

6 Interrogating LAPOP Data

The causal mechanisms identified in the preceding section will be examined empirically using data from the AmericasBarometer (Latin American Public Opinion Project (LAPOP) 2020). LAPOP carries out regular opinion surveys across all the countries in the Americas. LAPOP remains the only regionally consistent dataset that identifies social assistance respondents. An additional advantage of LAPOP surveys for our purposes is their extensive coverage of political attitudes. During the 2000s, selected country surveys asked respondents whether they received any support from conditional income transfers. See Table 13.1 for a listing of the surveys including identifiers for conditional income transfer participation. Identification of conditional income transfer recipients is very reliable because respondents were presented with an explicit reference to the main transfer programme. It asked: Now referring specifically to [a name of the country’s flagship programme followed] are you or someone in your household a beneficiary from this programme? This question was applied in countries with at least one large-scale conditional income transfer programme but only in selected years for selected countries. In some cases, country samples were divided in two and this question was included only for one half of the sample.¹⁶

Table 13.1 Conditional income transfer recipients in LAPOP by wave and country

The empirical work reported below interrogates pooled data for the period 2000–2019. It does not consider the extent to which particular country conditions influence political responses. This is an important issue which will be the subject of further work, but the focus here is on identifying core causal mechanisms. It is important to keep in mind that LAPOP collects a similar number of respondents from each of the countries covered, say 1000 interviews from Brazil and 1000 interviews from Paraguay, so that the pooled data is not representative of the population in Latin America taken as a whole.¹⁷ The analysis will compare political responses and outcomes between conditional income transfer participants and non-participants. This is justified because the objective is to examine whether participation in conditional income transfer programmes improves the political inclusion of disadvantaged groups.¹⁸

Table 13.2 reports on the odds ratios of the particular variables of interest from a logistic regression of the pooled data. The three causal mechanisms identified in the previous section are tested separately. The table reports on the main variables of interest. The controls include a wealth index, and country and year dummies.

Table 13.2 Odds ratios using LAPOP 2010–2019 data

Beginning with the support for redistribution mechanism, results are reported in column (1) in the table. This causal mechanism links transfer receipt to voting and then to preferences for redistribution policies: transfer recipients are more likely to vote for incumbents in order to support redistributive policies. Transfers are the input, voting for incumbents is the output, and support for redistributive policies is the outcome. The latter reflects the transformational component. The empirical counterparts are as follows. Recipients are directly identified in the survey. The survey asked respondents whether, if presidential elections were to be held next week, they would be likely to abstain from voting, spoil their vote, vote for the incumbent or vote for the opposition. To simplify matters the first two options are combined and are the reference category in the regression (referred to below as “not voting”). The surveys also asked respondents whether they agree with the statement: The state should implement strong policies to reduce income inequality between rich and poor by selecting a value on a scale from one to seven, where one is “strongly disagree” and seven is “strongly agree”. A “supports redistribution” binary variable was generated by combining categories six and seven, those most strongly agreeing with the statement.

As can be seen from the first column of Table 13.2, conditional transfer recipients are 1.08 times more likely to support redistribution than non-recipients, while those planning to vote for the incumbent were 1.26 times more likely to support redistribution than those not voting. Figure 13.6 shows linear combinations of the odds ratio of the variables of interest. As can be seen there, transfer recipients planning to vote for the incumbent are 1.36 times more likely to support redistribution than non-recipients not planning on voting. This is consistent with the support for redistribution mechanism, indicating that transfers encourage recipients to register and vote, and to vote for incumbents with the objective of supporting redistribution policies.

Fig. 13.6

Odds ratios linear combinations of variable estimates. (Source: Own elaboration)

Turning to the bureaucratic mechanism encouraging political engagement, the results can be found in column (2) of Table 13.2. This mechanism links transfer receipt to bureaucratic requirements associated with participation, electoral registration in particular, themselves leading to greater political engagement and support for the political system. To capture support for the political system the analysis relies on a binary transformation of responses to the question: On a scale of one to seven, where one is “none” and seven is “a lot”, to what extent you believe the political system should be supported? The binary transformation coded support for the political system by combining responses six and seven on the scale. When examined, LAPOP data show no significant difference in the rates of registration among recipients and non-recipients. If participation in conditional income transfer programmes required voter registration at the start of programme participation, this effect is bound to decline after the implementation of the programme as it applies to new entrants only.¹⁹ Registration is not included as a separate variable in the regression analysis but voting for the incumbent or the opposition can be used as a proxy.

The estimates in Table 13.2 show that transfer recipients are 1.4 times more likely to support the political system than non-recipients. Respondents planning to vote for the incumbent are 1.2 times more likely to support the political system than respondents not voting. The estimates for respondents planning to vote for the opposition are lower but positive and significant. Figure 13.6 shows the linear combinations of the odds ratio of a transfer recipient who plans to vote for the incumbent. It shows that recipients are 1.7 times more likely to support redistribution than a non-recipient not voting. This result is consistent with the bureaucratic mechanism encouraging political engagement.

Column (3) of Table 13.2 reports on the cognitive change mechanism. This mechanism stresses cognitive change associated with transfer recipients’ awareness of their fuller membership in their polity, leading to the same transformational element as the previous mechanism, namely a strengthening of their support for the political system. To capture this particular mechanism, the analysis relies on a survey question asking respondents whether they agree with the following statement: Political leaders in this country are interested in what people like you think, again on a Likert scale with one denoting “strongly disagree” and seven denoting “strongly agree”.

The odds ratios reported in column (3) in the table indicate that transfer recipients are 1.2 times more likely to support the political system. Respondents agreeing that political leaders listen to people like them are 1.7 times more likely to support the political system. Figure 13.6 shows that transfer recipients who believe the government listens to people like them are 2.1 times more likely to support the political system. Again, these findings are in line with the cognitive change mechanism proposed above.

To sum up, the analysis compared political responses and outcomes for conditional income transfer recipients and non-recipients. Analysis of the LAPOP 2010–2019 data fails to reject the presence of all three causal mechanisms proposed in the previous section.

7 Conclusions

The chapter tackled the issue whether conditional income transfer receipt generates political responses associated with improvements in political inclusion. This is central to an assessment of the emerging new welfare institutions in low and middle-income countries. Political inclusion is key to sustainable and effective poverty reduction.

The chapter is related to the wider literature examining the political implications of conditional income transfer programmes. It contributes to this literature in several ways. First, a causal mechanism approach opens up the black box of quasi-experimental evaluation studies, paying attention to the “how” as a complement to the “what” and the “why”. Second, and this is perhaps another way to make the first point, a causal mechanism approach clarifies the linkages between interventions, responses, and outcome dimensions of welfare institutions. An assessment of whether participation in conditional income transfer programmes improves political inclusion requires that we pay attention to political outcomes (redistribution, strength of the political system), not just to actions or outputs (voting, registration). A causal mechanism approach also offers essential clues as to transformational components of social assistance. The focus on participants in transfer programmes is essential in this context. Third, the analysis of LAPOP data confirms that the mechanisms proposed in the literature can be productively studied with empirical counterparts.

The analysis in the chapter identifies three main causal mechanisms linking conditional income transfer receipt to political inclusion. They include a support for redistribution mechanism, a bureaucratic mechanism encouraging political engagement and a cognitive change mechanism. Analysis of empirical counterparts in the LAPOP data provides some support for their relevance to our understanding of political responses among participants.

The discussion raises several issues for further research. The approach and findings in this chapter aimed at explanation as opposed to prediction. As noted above, pooled LAPOP data is not representative for Latin America as a whole. The empirical results are indicative of the direction of causal links, not necessarily of their particular weight. The causal mechanism approach does not provide clear guidance on how to assess the relative relevance or complementarity of competing mechanisms. Further research will need to be undertaken in order to clarify under what conditions causal mechanisms can be strong or weak, for example by examining causal mechanisms across political regime conditions at country level. The same applies to the need to study changes in political regimes over time. Finally, there is only scant literature that considers the role of programme design in strengthening or weakening causal mechanisms and outcomes (Schober 2019; Bruch et al. 2012). Future research tackling comparisons across social assistance programmes might prove to be informative.