Introduction

Petty bribery is a serious problem that a wide range of citizens across the world experience first-hand while accessing public services. Recent public opinion figures show that nearly one in five citizens who access public services have to pay some form of bribe in Asia (Vrushi, 2020), and more than one in four in Africa (Pring & Vrushi, 2019) and Latin America (Vrushi & Pring, 2019). As a form of corruption, it involves an interpersonal exchange between a citizen and a public official by which the former grants a material benefit for the private advantage of the latter. Quite often, this type of bribery is extortionary, and it hinders access to basic public services, such as document issuance or health treatments (Rose & Peiffer, 2014). The negative consequences of bribery are many: compromising service quality (Hunt & Laszlo, 2012; Matsushima & Yamada, 2016), reinforcing social inequality (Justesen & Bjørnskov, 2014; Kaufmann et al, 2008; Rose & Peiffer, 2014), and interfering with citizens' subjective well-being (Sulemana et al., 2017). A particularly serious consequence of extortionary petty bribery is the extent to which it triggers a 'negative feedback mechanism,' causing citizens to trust institutions less and passively accept corruption (Lavallée et al., 2008; Morris & Klesner, 2010; Persson et al., 2013; Razafindrakoto & Roubaud, 2007; Seligson, 2002).

A variety of anti-corruption efforts aim to tackle bribery affecting access to public services, either by reforming public institutions or by empowering its victims (Hunt & Laszlo, 2012; Isbell, 2017; Rose & Peiffer, 2015; Rose, 2015; Chêne, 2019). One common strategy is to create bribery reporting tools enabling citizens to report incidents to authorities, such as by calling anonymous hotlines or using digital apps. Despite its popularity, successful cases of bribery reduction through such bribery reporting tools are anecdotal (e.g. Chêne, 2019; Deininger & Mpuga, 2005). Only a small number of digital bribery reporting tools flourish, and for a limited amount of time (Zinnbauer, 2015; Kukutschka, 2016) . In light of those limited effects, a seemingly more feasible and advocated goal of bribery reporting tools is signaling to citizens that public institutions are devoted to controlling corruption.

The concept of a "positive feedback mechanism" suggests that highlighting successful examples of addressed complaints sustains interest and engagement with these platforms. This strategy is often heralded as effective in empowering citizens and promoting efficacy in combating corruption more broadly (Zinnbauer, 2015: Kukutschka, 2016; Peixoto & Fox, 2016: Isbell, 2017). In that way, bribery reporting mechanisms, even if not directly decreasing the frequency of a specific form of corruption in the short term, could have a spillover effect by inspiring citizens to engage against corruption. This inspirational effect entails not just stimulating citizens to report bribery to authorities, but also to participate in other forms of corruption opposition such as joining anti-corruption organizations, contacting the media, and protesting against corrupt incumbents.

This paper tests the existence of a positive feedback mechanism: the proposition that advertising successful cases of institutional responsiveness not only persuades citizens of institutional commitment to fighting corruption but also inspires citizens themselves to take action. Although this inspiring effect seems theoretically self-evident, the academic literature empirically testing citizens' reactions to anti-corruption efforts—initiatives aimed at controlling corruption—is still incipient. Not much is known about what types of anti-corruption efforts result in more positive assessments of public institutions, with mixed results associated with different types of efforts (e.g. Ferraz & Finan, 2008; Bauhr & Grimes, 2014; Bobonis et al., 2016; Peiffer, 2020; Agerberg, 2021; Cheeseman & Peiffer, 2021; Cheeseman & Peiffer, 2022;).

This paper tests the positive feedback mechanism in the context of bribery reporting mechanisms using a sample of 12 countries in Africa. Bribery is not only a common problem in Africa, but citizens know it and believe strategies targeting bribery are the most valuable efforts against corruption.Footnote 1 Relatedly, a recent line of research postulates that when citizens perceive institutions to be committed to address corruption, they are more willing to uptake anti-corruption activities (Bauhr, 2017; Peiffer & Alvarez, 2016). This is believed to be particularly the case in contexts where corruption is widespread.

The empirical strategy relies on the existence of a survey question from the 6th round of Afrobarometer. It uses a subsample of individuals victimized by bribery and who reacted by reporting it to public authorities. The survey respondents' experiences vary as to what happened following the reported incident. According to their answer to a survey question, whether authorities took action against the bribe-asking officials or not. Individuals who answered that authorities took action are taken to have experienced institutional responsiveness to bribery reporting. To account for the first step of the "positive feedback mechanism" (concerning the persuasion effect), I test whether responsiveness is associated with more optimistic assessments about public institutions, measured using perceived anti-corruption commitment and institutional trust. To account for the second step of this mechanism (concerning the inspiration effect), I test whether institutional responsiveness is associated with a greater willingness to engage against corruption, measured using citizens' efficacy to fight corruption and willingness to mobilize when dissatisfied with government performance.

The findings do not support the existence of a positive feedback mechanism. Although citizens who observe institutional responsiveness do appear convinced of public institutions' commitment to fighting corruption, the degree to which this leads to increased willingness to mobilize against corruption is not robust. What can be referred to as the persuasion stage of the positive feedback mechanism is often perceived as the most significant challenge, as citizens are not easily convinced that public institutions are genuinely combating corruption. Simultaneously, it is often assumed that, once citizens are persuaded that public institutions genuinely attempt to control corruption, inspiration to mobilize against corruption will automatically follow. The latter claim finds no support in this paper.

This paper contributes to the study of how citizens react to anti-corruption efforts, introducing and testing the concept of a positive feedback mechanism. This mechanism involves two stages: persuasion, where citizens view institutional performance in the realm of anti-corruption more positively, and inspiration, where citizens also become more inclined to oppose corruption. The focus is on a specific but underexplored anti-corruption effort – the responsiveness to bribery reporting by citizens.

Does institutional responsiveness to bribery reporting trigger a positive feedback mechanism? Insights from studies of citizens’ reactions to anti-corruption efforts

Institutional responsiveness refers to the capacity of an institution to react effectively, and appropriately to various situations, or inputs from its environment. This responsiveness can manifest in different ways, including addressing concerns, adapting to changes, or taking action based on feedback or information received. In the context of this paper, institutional responsiveness relates to how public institutions, such as government agencies or anti-corruption bodies, react when individuals report incidents of bribery.

Many anti-corruption practitioners believe that using bribery reporting tools can be a way to display a government's commitment to fighting corruption and hence inspire citizens to do the same (e.g. Isbell, 2017:15; Zinnbauer, 2015:13; Kukutschka, 2016:6)Footnote 2. However, they recognize it's challenging to persuade citizens that public institutions genuinely fight corruption. Despite the challenge, they believe that once convinced, citizens will naturally be motivated to join the fight against corruption.

The notion that citizens could be inspired by displays of institutional commitment to combat corruption is rooted in the perspective of corruption as a collective-action problem. Persson et al. (2013) suggest that in cases where corruption is systemic, it cannot be assumed that citizens will actively oppose corruption. In other words, corruption does not operate as a 'principal-agent problem,' where citizens act as 'principals' punishing 'agents' engaging in corruption. In contexts of widespread corruption, citizens might be hesitant to assume the role of principals who punish and resist corruption when given the opportunity. This hesitation stems from the anticipation that formal institutions will not serve as anti-corruption enforcers but rather as perpetrators of corruption. Consequently, from the citizens' perspective, yielding to corruption becomes less risky or costly than resisting it, even if they harbor moral opposition to corruption.

But an additional view claims that it is also possible that when public institutions perform in a manner contradicting these pessimistic expectations and, instead, fulfill their formal roles by enforcing anti-corruption norms, citizens are subsequently inspired to join the fight against corruption (Bauhr, 2017; Peiffer & Alvarez, 2016). This inspiration can manifest through various forms of resistance, such as protests, involvement in anti-corruption organizations, and contacting the media. The core idea is straightforward: if public institutions demonstrate commitment to fighting corruption, ordinary citizens—the primary victims of corruption—are likely to follow suit.

To clarify, I am referring to two sets of expectations. One set pertains to what citizens believe they are entitled to, such as a low-corruption society with effective institutionsFootnote 3. The other set relates to what citizens expect formal institutions to actually do. The latter does not involve what citizens feel they are entitled to but rather concerns their perceptions of how the world functions. The collective-action view of corruption emphasizes that, even if citizens believe they are entitled to a 'low-corruption society,' they hold pessimistic expectations regarding the actions of other citizens and public institutions. On the flip side, when public institutions demonstrate a commitment to anti-corruption—going against citizens' pessimistic expectations and fulfilling what they believe they are entitled to—it is likely to increase their willingness to engage in anti-corruption activities.

The positive feedback mechanism is theoretically reasonable in light of the collective-action approach to corruption (Bauhr, 2017; Peiffer & Alvarez, 2016; Persson et al., 2013). But while it would be important to test such expectations empirically, there is not much research into how citizens react to different types of anti-corruption efforts.

In this paper, I distinguish information about corruption from exposure to anti-corruption efforts. Other than uncovering and broadcasting information about corruption, efforts can also be a way to display institutional commitment. Even without completely eradicating corruption, public institutions could exhibit a commitment to the advancement of an anti-corruption agenda. Thus, citizens may perceive that corruption in their surroundings is relatively pervasive and still think that there are public institutions seriously attempting to control this phenomenon. In this section, I review empirical evidence concerning citizens' reactions to three types of anti-corruption efforts: transparency initiatives, judicial convictions, and awareness-raising messages. I also take insights from studies focusing on perceptions of institutional commitment to fight corruption.

The first stage of the positive feedback mechanism entails that once citizens report bribery to authorities and authorities take action against it, then this will persuade citizens that the anti-corruption agenda is being taken seriously. Persuasion entails that citizens begin to evaluate institutional performance more positively. Although this assertion may seem tautological, as this section shows, empirical studies provide conflicting evidence about the types of initiatives that can persuade citizens that their government or other public institutions are earnestly fighting corruption.

Empirical evidence shows that many anti-corruption efforts have paradoxical consequences for how ordinary citizens view public institutions. That is, instances of “institutional responsiveness”, that should technically be regarded by citizens as positive developments, end up having negative consequences. For instance, transparency reforms, rather than fostering institutional trust and stimulating political involvement, are often associated with inflated corruption perceptions, distrust towards institutions, and less willingness to engage in political participation (Bauhr & Grimes, 2014; Costa, 2013; Worthy, 2010). Similarly, court convictions of former heads of state depress citizens’ institutional trust and willingness to vote and participate in demonstrations (Poertner & Zhang, 2023).

Particularly, awareness-raising messages, which consist of messages about the importance of fighting corruption or about the work done by anti-corruption institutions, have been found to be either ineffective (e.g. Peiffer, 2020; Walton & Peiffer, 2017), or worse, to undermine citizens' willingness to pay taxes (Cheeseman & Peiffer, 2021) or even to encourage bribery (Cheeseman & Peiffer, 2022). On the other hand, different message framings – which do not try to persuade citizens of their governments' achievements in fighting corruption – have been shown to elicit intended effects, increasing interpersonal trust, decreasing willingness to bribe, and harnessing more optimistic assessments about one's institutional context (Agerberg, 2021).

The main explanation for the unintended effects of transparency reforms, court convictions, and awareness-raising messages is priming. Citizens might paradoxically focus on how widespread corruption is instead of appreciating the efforts to control it.

However, other anti-corruption efforts have also been found not to lead to such unintended consequences. Anti-corruption efforts conveying clear and credible information about progress in the realm of anti-corruption seem to inspire citizens to engage against corruption. As field experiments set in Latin America suggest, exposing citizens to reports produced by anti-corruption audits incentivizes voting against corrupt incumbents (Bobonis et al., 2016; Ferraz & Finan, 2008: Gans-Morse et al., 2018). In these interventions, citizens are provided not only with high-quality information about corruption (summarized in audit reports), but may also become aware (through the media, for example) that certain public institutions not directly tainted with corruption are dedicated and committed to addressing that problem. In their seminal study about transparency reforms, Bauhr and Grimes (2014) posit that when citizens have access to information about corruption without having access to appropriate mechanisms to resist it, this leads to a sense of resignation. Thus, creating mechanisms that allow citizens to express themselves against bribery could have the opposite effect, as long as these mechanisms are not widely believed to be ineffective (Walton & Peiffer, 2017).

Thus, it may be that some types of anti-corruption efforts fail to persuade citizens of public institutions’ commitment to the anti-corruption agenda, whereas others are perceived as more convincing. When reporting bribery, citizens have direct experience about the effectiveness of that anti-corruption mechanism. Having first-hand experience of institutional responsiveness is arguably more persuasive than exposure to indirect information about institutional performance, such as that arising from transparency reforms and awareness-raising messages. Thus:

  • H.1. Citizens observing responsiveness evaluate public institutions more positively than citizens who report corruption to authorities but do not observe responsiveness.

In the previous stage of the positive feedback mechanism, I discussed to what extent responsiveness to bribery reporting is likely to be associated with positive evaluations of institutional performance. The first step thus relates to what citizens expect from public institutions and the extent to which they are perceived to fulfill those expectations. The second stage of such a mechanism refers to whether such positive evaluations translate into a greater likelihood of engaging in various forms of corruption opposition. It relates, thus, to how citizens see their role in this "positive feedback mechanism". This second expectation also derives from the notion of corruption as a collective-action problem, as it is related to the idea that an individual's behavior against corruption is contingent on what public institutions will do in the face of corruption.

It is indeed possible that, even if citizens are persuaded of institutional commitment to fight corruption, they still may not be inspired to personally do something about it. This phenomenon is documented by a qualitative study conducted in Sweden (Bauhr, 2012:79), where individuals exhibited high levels of institutional trust and a strong belief in the effectiveness of public institutions in controlling corruption. Consequently, they saw no compelling reason for civic engagement in anti-corruption efforts. While this passivity may be attributed to the relatively low corruption context in the Nordic country, it is also observed in other settings. In a survey experiment conducted in Tunisia and South Africa, individuals exposed to scenarios of judicial convictions against corruption showed an increased willingness to support institutions in the fight against corruption (Barbabela et al., 2022). However, there was no apparent inclination to change their attitudes toward bribery. These two studies suggest that even when individuals acknowledge government commitment to combat corruption, it does not automatically translate into increased personal willingness to actively engage in the fight against corruption. In essence, it can be that citizens do not think they have a personal role to play in the fight against corruption, instead believing that corruption is a problem to be addressed solely by public institutions.

On the other hand, cross-national research suggests that citizens can indeed be inspired to act by seeing public institutions address corruption, particularly in more corrupt settings. Comparing OECD and non-OECD countries, a study finds that in the latter – where corruption would be also more pervasive – perceived government commitment to address corruption correlates with various forms of engagement against corruption (Peiffer & Alvarez, 2016). Another study finds that, at high levels of perceived corruption, increased perceptions of government commitment go alongside engagement against corruption (Bauhr, 2017). Therefore, in regions of the world where bribery levels are relatively high, perceiving that public authorities are nevertheless acting to address this problem could be a powerful driver of citizens' mobilization against corruption. Thus:

  • H.2. Citizens observing responsiveness are more willing to mobilize against corruption than citizens who report corruption to authorities but do not observe responsiveness.

Empirical strategy and data

To test the two hypotheses comprising the notion of positive feedback, the analysis relies on data from the 6th round of the Afrobarometer public opinion survey, fielded between 2015 and 2016.

The analysis uses a subsample of the original survey that includes respondents who admit having paid a bribe and reporting it to authorities. Only a small fraction of survey respondents report bribery to authorities. Although this is true for all the 36 countries included in the 6th round of the Afrobarometer, bribery reporting varies a lot across countries. The sample only lists countries that asked the question about bribery reporting and also have more than 30 respondents answering such questions. This amounts to respondents across 12 of the original 36 countries, namely, Liberia, Egypt, Sao Tomé and Principe, Cameroon, Mozambique, Tanzania, Morocco, Algeria, Uganda, Kenya, Guinea, and Zambia. (See the appendix for additional information).

The hypotheses are tested using this subset of individuals, which includes people sharing a similar experience in terms of bribery victimization and reaction to those incidents, as I attempt to demonstrate when discussing Table 1 later in this section. To start, however, I explain what is the independent variable, as it shapes the overall structure of the analysis.

Table 1 Balance table: socioeconomic characteristics across responsiveness to bribery reporting

Independent variable

The independent variable is a binary variable coded using the statements following the question: “Which of the following happened the most recent time you reported a bribery incident to authorities? Authorities took action against the government officials involved.” Positive answers are coded as one and negative ones as zero.

Respondents are selected into the sample by reporting bribery, independently of what followed. 583 respondents claim that “authorities did not take action against the officials involved in bribery", against 252 who claim that they didFootnote 4. Figure 1 displays the survey structure: the questions are in boxes, and the number of observations is between brackets:

Fig. 1
figure 1

Sample selection using survey questions

Table 1 is a balance table showing that those survey respondents who have reported bribery to authorities and observed responsiveness share key sociodemographic characteristics referring to bribery engagement activities with those who did not observe responsiveness. The balance table includes several individual-level control variables. Firstly, the regressions include a continuous variable for age, ranging from 18 to 88, and a binary indicator for female respondents, as male and younger citizens are more prone to be involved in bribery accessing services in Africa (Justesen & Bjørnskov, 2014). As usual, the analysis also includes education, which is an ordinal variable indicating whether respondents have low levels of education (no formal schooling up until completed primary schooling), some education (some or completed high school), or high levels of education (post-high school education), as education is associated with anti-corruption engagement (Bazurli & Portos, 2021; Walton & Peiffer, 2017). To account for possible differences of moral values, the analysis includes a dummy variable indicating that the respondent engages in religious practices like prayer, reading a book, or attending a service (Walton & Peiffer, 2017). As individuals living in urban areas are more prone to pay bribes due to a concentration of public services in these areas, the models include a dummy variable indicating that the respondent lives in an urban or peri-urban area (Justesen & Bjørnskov, 2014). There are two objective indicators related to one's socioeconomic status. To account for poverty, the usual index measuring access to food, water, medical care, cooking fuel, and cash income is included (Mattes et al., 2002). Additionally, the models include a dummy indicating that the respondents have a full-time job (Justesen & Bjørnskov, 2014). Lastly, there is a binary indicator showing that the respondent is a member of an association, as this is a relevant predictor of access to service and because belonging to relevant networks may be a way to secure responsiveness to one's claims (Justesen & Bjørnskov, 2014; Peiffer & Rose, 2018; Walton & Peiffer, 2017).

The differences between respondents who have observed responsiveness and those who did not are not statistically significant, as indicated by the fact that all p-values of country-fixed effects regressions on the control variables are over 0.05 (see Table 1).

The fact that the sample is balanced indicates that the values of the dependent variable are displayed by similar individuals (according to observable socioeconomic characteristics) living in different countries. Previous studies widely document that citizens who experience bribery display systematic differences concerning citizens who do not (Justesen & Bjørnskov, 2014; Mbate, 2018; Peiffer & Rose, 2018). The most common approach to address these differences is to simply control for individual-level characteristics in a regression, which is also done in this paper. However, large between-group differences may bias regression estimates (e.g. Kim & Palmer, 2008; Rubin, 2001). For this reason, methods such as propensity score matching seek to re-balance the dataset to compare cases where the observed covariates are as similar as possible (Ho et al., 2007). In the sample analyzed in this paper, the balance concerning individual-level observable socioeconomic characteristics did not have to be achieved through matching. It is a natural feature of the data and it may have been produced by the fact that similar individuals across different countries are self-selected into reporting bribery, which is arguably a rare practice.

The country-specific balance tables corroborate that there is no systematic observable difference between citizens claiming that authorities did not act and those who claim they did (see appendix). At most, in some cases, there is an imbalance concerning one or two indicators, but even in that case, the imbalance goes in different directions in different countries, showing that these observable differences are not systematic. In any case, the main estimation strategies include country-fixed effects.

Dependent variables

Two sets of dependent variables account for the positive feedback mechanism: the persuasion stage concerns citizens' assessments of public institutions, and the inspiration stage concerns citizens’ willingness to mobilize against corruption. Table 2 shows the descriptive statistics for all dependent variables.

Table 2 Descriptive statistics of dependent variables in the positive feedback mechanism

Both steps rely on more than one indicator. In step 1 (concerning the persuasion effect), there are two indicators: perceptions of anti-corruption institutional commitment and institutional trust. A distinction is made between them to account for the possibility that citizens may acknowledge public institutions have not completely `solved' corruption but still believe they are committed to addressing it. Perceptions of anti-corruption commitment refer to an item asking citizens "How well or badly would you say the current government is handling the following matters, or haven't you heard enough to say: Fighting corruption in government?" Original answers are 1 (very badly); 2 (badly); 3 (well); and 4 (very well), but as the goal is to assess whether individuals evaluate such a performance in a positive light or not, the variable is recoded as a binary variable (1 = very well/well, 0 = very badly/badly). Institutional trust is an index combining answers about trust in the president, local government, and courts, each of which originally ranges from 1 (no trust at all) to 4 (a lot of trust). Trust in other institutions is not included in the index as the goal is to derive a measure that is not sensitive to political preferences (Bauhr & Grimes, 2014:299). The index is calculated using polychoric correlations and factor analysis, where higher levels indicate increased levels of institutional trust.

In step 2 (concerning the inspiration effects), citizens` willingness to mobilize against corruption is operationalized by combining of two indicators. To account for citizens' efficacy (Bauhr, 2017; Kostadinova & Kmetty, 2019; Pavão, 2019; Peiffer & Alvarez, 2016) there is a binary indicator using the question "Please tell me whether you agree or disagree with the following statement: ordinary people can make a difference in the fight against corruption?" Original answers range from 1 (strongly disagree) to 5 (strongly agree), but the variable is recoded as binary (1 = any level of agreement with the statement, 0 = otherwise)Footnote 5. The other is a mobilization index using four indicators to account for past experiences as well as reported intention to engage in forms of political participation to express dissatisfaction with government performance, similarly to previous studies (Bauhr & Grimes, 2014; Bazurli & Portos, 2021; Agerberg, 2019). The index uses the following question: "Here is a list of actions that people sometimes take as citizens when they are dissatisfied with government performance. For each of these, please tell me whether you, personally, have done any of these things during the past year. If not, would you do this if you had the chance". The four items are (i) Join others in your community to request action from the government; (ii) Contact the media, like calling a radio program or writing a letter to a newspaper; (iii) Contacted a government official to ask for help or make a complaint; (iv) Participated in a demonstration or protest march. The index is also calculated using polychoric correlations and principal component analysis, with higher values of factor loadings representing increased frequency of participation. This is an appropriate indicator to assess the existence of spillover effects as item (iii) “contacting officials for help or make a complaint” is arguably close to what reporting bribery to authorities entails, whereas the others represent different ways to resist corruption.

Estimation strategy and limitations

The main estimation strategy of Hypotheses 1 and 2 varies depending on the level of measurement of the dependent variables. The models using binary dependent variables are fitted using logistic regressions, and those with indices use OLS regressions and HC3 heteroscedasticity robust standard errors. Additionally, in the next section, instead of logit regression coefficients (which are reported in the appendix), predicted probabilities are displayed to enable a more intuitive and substantive interpretation. In terms of predicted probabilities, the sign of the shifts indicates the direction of association between the independent and dependent variable and its size indicates how big the changes in the likelihood of observing the condition represented by the dichotomous dependent variable are. Other than the individual-level controls, country-fixed effects are included to account for observed and unobserved country-level differences. The models do not use survey weights, as the aim of the test is not representativeness at the national level but rather to assess differences across two groups of survey respondents with different experiences in terms of institutional responsiveness.

One important limitation of the empirical strategy is that, as this is not an experimental study, one cannot rule out that the individuals in the sample differ in some unobserved dimension. The tests conducted in this paper rely on observational data, and there is no way to ascertain that an unobserved factor other than the experience of seeing authorities take action after having reported bribery is what drives different assessments across these two groups of individuals.

The empirical strategy can also not rule out the possibility of reverse causality. Although the theorized direction of the positive feedback mechanism entails that institutional responsiveness influences evaluations of institutional performance and then influences willingness to engage against corruption, it may be that willingness to engage against corruption precedes positive evaluations of institutional performance (that is, H2 could precede H1).

Relatedly, it may be that individuals' answers about what happened after they reported a bribe to authorities are not grounded in an experience of responsiveness, but instead stem from high levels of institutional trust. The independent variable (observing authorities take action) is arguably very close to the dependent variables, particularly those consisting of step 1, concerning evaluations of institutional performance. In that sense, trusting individuals would simply assume that once they report corruption to authorities, authorities would do something about it. Nevertheless, even if that was the case, the step 2 results would still be indicative of these high-trusting individuals' willingness to engage against corruption, which is arguably still informative.

Results

In this section, I test the claim that responsiveness to bribery reporting triggers a positive feedback mechanismFootnote 6. I start by testing Hypothesis 1, concerning the persuasion effect, which posits a positive association between responsiveness to bribery reporting and more optimistic assessments about public institutions, in terms of how individuals evaluate institutional performance and trust institutions.

Table 3 shows results from four regressions concerning Hypothesis 1. Model 1 uses the dichotomous variable indicating optimistic assessments of institutional anti-corruption commitment as a dependent variable and shows the predicted probabilities derived from the logit model. Models 2 to 4 show the regression coefficients of OLS regressions with robust standard errors and the institutional trust index as a dependent variable.

Table 3 H.1—Persuasion: regression results

Model 1, which includes individual-level controls and country-fixed effects, shows that those observing authorities taking action after reporting an incident of bribery are roughly 8% more likely to believe the authorities are doing a good job or a very good job in the fight against corruption than those who report bribery but see no action to follow. These results are significant at the 95% level and go in line with H1. The size of the effects in model 1 is relatively substantial and in line with those eventually reported in awareness-raising message studies (e.g. Agerberg, 2021; Cheeseman & Peiffer, 2022) – which, when they are not null results, tend to range from 4 to 10% in terms of predicted probabilities shifts concerning attitudes towards institutions and corruption -.

Models 2 to 4 show the results of OLS regressions with institutional trust as a dependent variable. Model 2 reports the results of a bivariate regression, model 3 includes individual-level controls and model 4 includes in addition country fixed effects. The progressive inclusion of controls shows that the effects from the bivariate regression are not robust. The model 2 estimates suggest that individuals who observe authorities taking action after they report bribery display levels of institutional trust which tend to be on average 4 points higher than those who report bribery but see no action to follow. These results are relatively small, as the dependent variable ranges from 0 to 100, but still significant at the 95% level.

Other than being weak, this effect is not robust to the inclusion of the same controls as in model 1. Including individual-level controls in model 3 decreases the effect slightly (from 4.1 to 3.4 points), and the results are no longer statistically significant. This does not change with the inclusion of country fixed effects, and the coefficient becomes even smaller (roughly 2 points in a 0–100 index). Coupled with the small effect in model 4, the lack of statistical significance means that the chances that individuals observing bribery responsiveness display the same levels of institutional trust as those who do not observe responsiveness are relatively large.

The reason for the meager effects concerning institutional trust resonates with the claim that these attitudes are relatively resistant to change, particularly positive change, as previous studies usually report negative (unintended) effects in that regard (e.g. Bauhr & Grimes, 2014). Together with the results in Model 1, the Model 4 results indicate that even in situations where citizens evaluate anti-corruption performance more optimistically, harnessing institutional trust may be a more challenging and time-consuming endeavor (Morris & Klesner, 2010; Sharafutdinova, 2010).

In the appendix, I explore the robustness of the results from Table 3. I conduct ordered probit regressions with the original ordinal version of the variable measuring perceptions of institutional anti-corruption commitment. I also test the effects of responsiveness on institutional trust using different coding strategies for the dependent variable. Additionally, I fit multilevel models, using a generalized linear mixed-effect model with the perception of commitment dichotomous variable as the dependent variable and linear mixed-effects models with the institutional trust index as the dependent variable. These alternative estimation strategies largely confirm the trends in the main analysis.

I proceed to test the following stage in the positive feedback mechanism. Table 4 shows the results concerning Hypothesis 2, concerning the inspiration effect and the positive association between responsiveness to bribery reporting and a generalized willingness to engage against corruption. Model 1 shows the results of logistic regression with a binary indicator of a belief that ordinary citizens can make a difference in the fight against corruption. Once again, I report predicted probability shifts to make interpretation more intuitive. Model 2 shows the results of an OLS regression with robust standard errors and the mobilization index as a dependent variable. Both models include individual-level controls and country-fixed effects. The bivariate regression and the regression including only individual-level controls are reported in the appendix, and the results are virtually the same.

Table 4 H.2—Inspiration: regression results

In both models in Table 4, the effect size is relatively small, and the results are not statistically significant. Model 1 indicates that those observing authorities taking action after reporting an incident of bribery are roughly 5% less likely to believe that ordinary citizens can make a difference in the fight against corruption than those who report bribery but see no action to follow. This effect even works counter to the prediction in hypothesis 2. However, the results are not significant at the 95% or even the 90% level (the p-value is 0.182). Model 2 estimates suggest that individuals who observe authorities taking action after they report bribery display levels of mobilization which tend to be on average 1.5 points higher than those who report bribery but see no action to follow. These results are not only small in light of the 0 to 100 range of the dependent variable, but also not significant, even at the 90% level (the p-value is 0.345).

In the appendix, I report the results of an ordered probit regression, with the original ordinal version of efficacy as a dependent variable. The results go in line with those in model 1, Table 4: the responsiveness coefficient is still negative but also statistically significant at the 95% level. The tau cuts indicate that this is mostly due to a shift between people who “agree” and “strongly agree” with the statement that “ordinary citizens can make a difference in the fight against corruption”.

Concerning the model 2 results in Table 4, 4 also analyze in the appendix the four individual variables that comprise the mobilization index. I report the results of logit regressions in which the dichotomous dependent variables indicate whether people have taken part in any of the forms of political participation but also whether one would do it if they had the chance. I also report the results of ordered probit regressions using the original ordinal version of these variables (which has five categories). The results are not statistically significant, but the size of the logit and ordered probit coefficients is larger concerning "contacting officials to ask for help or make complaints", arguably the form of participation that is closer to reporting corruption to authorities. Although a positive association between responsiveness and such variable is theoretically an easy test, even in this case it is not possible to reject the null hypothesis concerning H2. Lastly, I also report results using a multilevel model with country random effects, and the results go in line with those from Table 4.

A current discussion in the empirical literature argues that, in corrupt contexts, displays of institutional commitment to fight corruption inspire civic engagement (see Peiffer & Alvarez, 2016). Considering the relatively high levels of bribery in the 12 countries represented in the dataset, it would be expected that responsiveness to bribery reporting is positively associated with citizens' efficacy against corruption as well as with reported levels of mobilization. However, this inspiration effect is not supported by the results.

The results also do not support the conclusion that individuals are less likely to mobilize when they perceive public institutions to be fighting corruption accordingly (see Bauhr, 2012). It could be that other forms of mobilization, other than those studied in this paper,—such as voting for clean candidates or signing petitions against corruption, for instance—are affected by displays of institutional commitment. It is also important to reiterate that the sample is comprised of individuals who have already taken part in a form of mobilization, as they have reported bribery to authorities.

Conclusion

The paper investigates whether institutional responsiveness to bribery reporting triggers a positive feedback mechanism. It does so by relying on a homogenous sample in terms of observable socioeconomic characteristics of individuals living in 12 African countries, which had experiences with bribery and reported it to authorities. The positive feedback mechanism outlined in this paper entails that citizens are persuaded that public institutions fight corruption, which in turn inspires them to engage in various forms of corruption opposition. The persuasion stage of the positive feedback mechanism—the positive association between observing responsiveness and positive assessments of public institutions—is only partly supported by the evidence. The results concerning the inspiration stage of this mechanism – observing responsiveness leads to an increased willingness to engage against corruption – are much weaker and inconclusive.

The case analyzed in the paper was prone to support the existence of a positive feedback mechanism for three reasons. First, because the individuals in the sample have direct exposure to an anti-corruption effort. Second, the operationalization of the independent variable is closely associated with the dependent variables. Third, the sample was taken from a context where corruption in the form of bribery is relatively serious, where displays of public institutional commitment are argued to be able to make citizens more engaged against corruption (Bauhr, 2017; Peiffer & Alvarez, 2016).

Given that this was a relatively easy test for the existence of a positive feedback mechanism, the null results reported here are meaningful. Null results are common in studies investigating how ordinary citizens react to anti-corruption efforts (e.g. Chong et al., 2015; Peiffer, 2020) . The null results concerning institutional trust are perhaps less surprising, given that harnessing this type of attitude is more challenging than corroding it (Morris & Klesner, 2010; Sharafutinova, 2010). However, the absence of a positive association between responsiveness and willingness to mobilize (operationalized using citizens' efficacy and the mobilization index) is particularly important, especially given the fact that responsiveness was indeed associated with more optimistic assessments about anti-corruption commitment. That is, the individuals in the sample seem to have been persuaded to a certain extent by their governments' commitment to fight corruption and still they do not display increased levels of efficacy and neither more willingness to mobilize in the form of joining others, contacting the media, contacting public officials or protesting.

This paper contributes to the emerging empirical literature documenting citizens' reactions to anti-corruption efforts. Previous studies have investigated citizens' reactions to anti-corruption efforts in the form of transparency reforms (e.g. Bauhr & Grimes, 2014; Costa, 2013; Worthy, 2010), anti-corruption messages (Peiffer, 2020; Cheeseman & Peiffer, 2021; Agerberg, 2021; Cheeseman & Peiffer, 2022), judicial convictions (e.g. Barbabela et al. 2022), or simply using perceptions of anti-corruption performance (Bauhr, 2017; Peiffer & Alvarez, 2016). This study expands this literature by investigating citizens' reactions to another type of anti-corruption effort: responsiveness to bribery reporting. Additionally, this study tests a popular claim based on the notion of corruption as a collective-action problem (Peiffer & Alvarez, 2016; Persson et al., 2013): that once citizens are persuaded that public institutions are earnestly fighting corruption, they would be more inspired to engage against corruption themselves. It does so by articulating the notion of a positive feedback mechanism consisting of two steps: one referring to persuasion about institutional commitment to fight corruption, and the other to inspiration to engage against corruption.

As the sample in this study only comprises relatively few individuals in relatively few countries, further research is warranted to determine to what extent the findings reported in the previous section are generalizable. The sample also comprises citizens who have been personally victimized by corruption, which on the one hand implies they may be more pessimistic than the average citizen (Cheeseman & Peiffer, 2021). At the same time, these individuals are also among the very few who report corruption to authorities in their countries, which can be indicative of their greater propensity to resist corruption (Persson et al., 2013). An experimental design applied to a more representative group of citizens in more countries in Africa or a different region, such as Latin America or Eastern Europe, could investigate whether responsiveness to bribery reporting is particularly unfit to stimulate citizens' mobilization because it concerns a petty type of corruption. Additionally, it could be that responsiveness to bribery reporting inspires other types of mobilization strategies not investigated in this study, such as voting for clean candidates or refraining from paying bribes instead of reporting them to authorities.

In terms of policy contribution, this paper fails to support the common recommendation that authorities create bribery reporting channels and publicize these cases as a strategy to mobilize citizens to act against corruption. These channels may have other goals that justify their existence, they may be normatively defensible, and they may occasionally lead to the uncovering of very serious corrupt schemes. These are all highly important and laudable goals. However, given the evidence available until now, it is far from clear—at least from the results presented in this paper—that bribery reporting tools serve as an effective strategy to generate widespread engagement against corruption amongst citizens. More research into this topic is necessary before these policy recommendations with the aim of generally mobilizing citizens are more widely reproduced.