The estimated reading time for this post is 37 Minutes

1.0 Abstract

This study empirically explores how an individual’s economic, social, and political capital affects the propensity to make bribe payments in exchange for public services. The empirical analyse conducted supports the main hypothesis in most Asian countries, that the poor bears the burden of bribery, with likelihood of paying bribes differing between different public services. This study then provides support for the social organisation model, to shows that the likelihood of bribery decrease with social capital and increases with political networks. These findings lead on to explore the effects of civil society and independent media on accountability and suggest come policies to counter the effect of bribery in Pakistan.

2.0 Introduction

When it comes to bribery, there are two main contrasting theoretical explanations that underpin the burden of bribery debate within the public sector. [1] Firstly, there is a “grease the wheel” hypothesis which states that bribery could have a positive impact to economic growth. Huntington (1968) proposes that, in a second-best country, bribery can help “grease” inefficient bureaucratic institutions through an investment hereby raising efficiency leading to increase in economic growth (Méon and Sekkat, 2003), therefore showing that, bribery is borne by rich individuals, with political connections, who can afford to “grease the wheels” and circumvent the bureaucratic red tape, especially in countries where there is a lack of accountability mechanisms. (Mbate,2016)

On the other hand, Hirschman (1970) proposes a competing “sand the wheels” hypothesis, within an ill functioning bureaucracy civil servants may cause unnecessary delays to generate income through bribery. In this case, the burden of bribery is borne by the poor as they are more reliant on public services due to income constraints (Mbate, 2016). The rich can afford an exit option and thus, can employ an alternative private service.

Using theories to determine who bears the burden of bribery is important as it can be used to implement anti-corruption reforms. However, most of the studies that test the validity of these theories use a unidimensional approach in economics, sociology, or political science. These studies were conducted in the late 70’s-80’s where reliable data was harder to collect and incorporate in to an econometric model. (Hunt and Laszlo, 2005)

Therefore, this paper aims to examine how economic, social, and political factors affect an individual’s likelihood to pay a bribe in exchange for public services like health, education, water security and permits in Pakistan. Given that neither of the previously discussed theories present a conceptual framework, this paper used Mbate (2016) work as a template to create a framework that incorporates all the determinants of bribe payments between bureaucrats and citizens, and extends it to illustrate the effect of accountability mechanism modifies the nature of this relationship.

3.0 Literature Review

Corruption in Pakistan has been extremely prevalent since its independence in 1947. Pakistan inherited a strong bureaucracy from British Raj, and to this day, no major change in the bureaucratic set up has happened (Shafqat, 1999). This had led to major speculation by the public that “corruption has seeped into higher echelons of bureaucracy” (Siddiqi, 2011). Data gathered by Mashru (2014), represented in Figure 1, shows that the perception of corrupt within bureaucracy is extremely high; with factors like Police showing perceived corruption levels of 90%.

  All of Them Most of them Some of them None of Them Don’t Know
Members of Parliament 12.7 31.2 28.3 8.2 19.6
Government Officials 12.4 39.4 38.4 2.4 7.4
Local Government 8.8 37.2 45.8 5.3 2.9
Police 10.2 55.5 24.3 7.9 2.1

Source: Corruption Perceptions Index 2016

Figure 1 – Perceptions of Corruption by the public for different Institutions in Pakistan

3.1 What causes Corruption

At a theoretical level, a vast quantity of literature dictates that there are three main pre-conditions for corruption to occur. Firstly, Klitgaard (1988) shows from the derivation of the principal agent model that there is a goal conflict between the citizens of the country, referred to as principals, and the bureaucracy, referred to as the public interest and agents. Those working in governmental and political positions feel that a manipulation or not fully disclosing information to the principals can lead to increase in wages for them. This information asymmetry leads to the incentive of corruption (Bardhan and Mookherjee, 2000).

Secondly, Jain (2001) explored the relationship between quantity of bribery and public service accepted by bureaucrats. He determined that rational bureaucrats engage in bribery if the benefits of being corrupt outweigh any costs, however Bardhan (1997) explains an interesting concept about corruption with and without theft. For example, assume that a £200 import license is required to import. A case of corruption with theft would meant that the individual pays less to the bureaucrat (£100) for the license and no money goes to the government. However, in the case of corruption without theft would mean the individual would pay a charge on top of the license fee (e.g. £220) and the additional would go to the bureaucrat (£20). In the instance of without theft, the bribery acts as a tax and hence acts as a burden to the rich individuals while the poor are totally excluded (Schleifer and Vishny, 1993). However, with theft, both the rich and the poor are better off bearing the burden of bribery.  

3.2 The Economist Argument

One of the most widely explored approaches within the economic world of bribery is economist argument, which argues that the burden of bribery is borne by rich individuals. Often referred to as “greasing of wheels” theory, it was pioneered by Leff (1964) and later developed by Rose-Ackerman (1978). The basic theoretical underpins, as explained by Lui (1985), that the rich individuals realise that they can manipulate bureaucrats to achieve a greater level of productivity which in turn allows for saved time and effort.

Interestingly, there is a vast amount of literature that pays support for this argument. Hunt and Laszlo (2012) use statistical tests to show that although the rich bear the burden, it also leads to economic growth within the field. They explore an example of a technological company in Uganda, who are believed to have used bribery to pass legislation on cashless payments. Although unexcepted, this has led to a decrease in black money in Uganda. Wiell (2010) also supports the theory showing that in advancing economies with inefficient government, a bribe will often lead to an increase in efficiency which leads to a greater level of productivity. With Rajak (2013)finding that richer individuals pay up to four times more in bribes that poorer individuals.

However, this level of bribery has some disadvantages. Firstly, Bardhan (2006) presents the argument the bureaucrats can reduce their efficiency and provide obstacles for business if they know that a level of bribes can be gained from them. Mutonyi (2002) explores the claim using Nigeria and concludes that a lot of time is wasted by rich individuals in coming up with a deal which offsets the level of efficiency gained from bribery. Finally, Hope (2014) shows that a lot of rich are unlikely to bear the burden as they form part of the bureaucracy or are politically elite, and thus can acquire information and solve in information asymmetry problem which drives corruption.

3.3 The Social Inequality Argument

A secondary argument approaches burden of bribery from the opposite direction, dictating that it borne by poor individuals. There are two main components to this argument, firstly, Smart (2008) shows the those within lower income bracket often face credit constraints which inhibited their abilities to access better private options. Using this knowledge, bureaucrats can manipulate poor individuals to pay a bribe to use public services (e.g. education, water supply).

Secondly, Shah (2006) has shown that poorer individual often have lower education status and suffer from bounded rationality, as they face high transaction costs which hinder the acquisition and processing of information. Piefer and Rose (2014) use this rationality to show the individuals can be manipulated by bureaucrats and charged a higher price knowing that poorer individuals are less mobile and unlikely to impost sanction on the corrupt bureaucrats.

Empirical studies by Justesen and Bjornskov (2014), concluded that those with low levels of purchasing power arising from lower incomes means that poor individuals rely heavily on public services, which acts as a signal for corrupt bureaucrats as an incentive to gain from bribery. Panel data exploring Asian countries found that poorer individuals are more than twice as likely to pay a bribe compared to rich individuals. (Khan and Griffin,1972)

3.4 The Social Organisation Argument

Whereas the previous arguments explore bribery as a monetary concept, Arrow (1972) postulates the depth of an individual’s social capital and inclusion. Social Capital is observed as an informal channel which both rich and poor can capitalise on to bypass bureaucratic procedures and legislation (Mbate, 2016). Empirical studies conducting by Putnam (1993), shows that countries with high social capital often finds that the level of honesty and thus, lowers the level of bribery.

Uribe (2014) expands the social capital can be distributed in to two sub categories; social and political networks. Social network, namely religious groups, lead to an increasing information flow, solving information asymmetry which lowera bureaucratic need and hence, a lower level of bribes paid. This in turn leads to an increase in political participation and allows for a greater level of transparency and accountability resulting in lower level of corruption in bureaucracy. (Olson, 1982)

On the other hand, political networks have an adverse effect on corruption due to increased promotion of self-interest motives of individuals (Portes, 1998). Kaufman and Wei (2000) examine the implication within African countries to show that politicians can sanction bureaucrats and impede their career advancement. Therefore, Arrow (1972) shows that bribes are greater for rich individuals as they can utilise political connection for gain.

3.5 The Behavioural Argument

Unlike the previously discussed approaches to burden of bribery, the behaviourist argument explores bribery through the lens of behavioural science (de Sardan, 1999). It assumes that individuals perceive other bureaucrats of being dishonest and hence their actions of bribery are simply building on the norm. Peiffer and Rose (2014) show that instances where public services are rendered in an honest manner, the level is bribery caused by bureaucrats is little to non-existent.

Although, when exploring Bangladesh, Bernheim and Rangel (2007)finds that this is rarely the case. Bribery is often seen as a valid method of additional income among bureaucrats, especially those working in low income governmental sectors (e.g. police). This is most likely because “short-term effect of being honest comparatively very high” which leads to the following proposition by bureaucrats: “well, of everyone else is corrupt, why shouldn’t I be corrupt?” (Persson et al., 2013)

Serra and Ryvkin (2012) develop this in to behavioural model to demonstrate that in instances where corruption is preserved to be higher, there is a lower level of sanctions and therefore greater level of corruption for the exchange of public services. Persson et al. (2013) expands the model, consisting of an individual and society as players, to concentrate on Kenya and Uganda. The results dictate that if both are honest, a pareto-efficent outcome is reached with both player gaining the best outcome as public services are optimally provided. However, even if one of the members of society is corrupt, not engaging in bribery is an irrational strategy, leaving one worse off. Therefore, the burden of bribery is conditionally based on perceptions of individuals and bureaucrats in society (Roy and Singer, 2006)

3.5 Accountability

Within corruption, accountability depends on political and legal institutions because they determine the probability of detection and sanctions (Lederman et. al.,2005).Claudio and Frederico (2011) argue that politicians have a higher level of accountability due to their election based on democratic system, and thus should provide public service in a relatively honest manner. This is further backed by Ferraz and Finan (2007)which the argument that their future in office is in the hands of the public and for them to maintain office in future terms, they must provide a greater level of utility, to the public, than past or future oppositions.

Furthermore, due to a lower level of proximity between citizens and bureaucrats the level of information asymmetry is low which in turn, lowers the cost as monitoring the behaviour of bureaucrats (Shah,2007). However, Manzetti (2014)shows that although accountability is sense as a function of electoral, Pakistan is characterized with actions like rigging, absence of term limits Naqash (2016), Crilley (2013).As a result, Young and Lewis (2013)argues that elections are clearly not ideal for enforcing accountability, and thus factors like independent media outlets can be used to solve collective action problems with perpetuate corruption.

Islam and Farmanullah (2015) explores both civil society movement and media as accountability mechanism, and finds that media leads to a greater decrease in bribery most likely because of the strength of independent media. Backed by the works of Khan and Khan (2004), independent media has already lead to a decrease in political corruption due to its close monitoring of political heavy weights in the country, however they argument later develops to show that independent media can be manipulated by the government to present a positive image to the people of Pakistan.

4.0 Method

Previous empirical models of exploring corruption only concentrate on either economic, political or social aspects to bribery. Therefore, it becomes evident that the uni-dimensional model used in previous studies is not adequate for this study. Hence, the conceptual framework used in this study is built from an adaptation of Khan (2007) which in turn, is uses the theoretical premise of Rose-Ackerman (1975,1996)and the political agent model developed by Hirschman (1970).

UB = u (w) + u (b,D) + p(w,b,ά, E) ∂VP

With works from Ibrahim (2009), Chêne (2008) showing that bureaucrat look to maximise their current and future levels of economic rent in Pakistan. This adaptation of Mbate (2016) work shows that the utility function depends on the bureaucrat’s wages, w, the quantity of bribe, b, which is dependent on factor “d”, which denotes economic, social and political factors. Finally, it also depends on a function that denotes the probability of holding office, ∂VP, which is denoted by wages, quantity of bribes and accountability mechanism, ά. This function is used to present hypothesis that will be tested later in this paper.

4.1 Empirical Model

To measure the burden of bribery, a baseline logistic econometric model is used that is further adapted to evaluate the validity of the hypothesises explained below. Unlike previous studies, this model aims to utilise prevailing economic, social, and political factors as shown below:

Yij = β0 + β1Zij + β2Sij + β3Pij + β4Bij + β5Xij + ŋi + εij

The model uses a binary dependant variable, bribe paid, which use the value of 1 to denote whether an individual, i, pays a bribe for public service, j, and 0 otherwise.  The equation incorporates factors like economic status of individual, Z, social factors (a binary variable for religious membership), S (political capital), P (cognitive effects) – binary variable to dictate trust regarding the corruption state of others – B. To account for any results occurring due to omitted variable bias, the variable X acts as a vector to incorporate individual characteristics controls (i.e. education, age, gender, and employment status). Finally, based on the works of Javaid (2010), the variable – ŋi – is used to account for time invariant factors like historical factors and ethnicity; and finally, ε, is the error term.

It is considering best practice to present results based on the naïve LPM model, as other specification (namely logistic regression) is simply a variant of it, and can be used to compare results using different techniques. In addition, it helps to show how severe a bias might be if you ignore the binary nature of the dependant variable. Therefore, because of the binary nature of the dependant variable, a logistic model based on the following specifications will be used, ensuring that unlike the LPM all coefficients will be between 0 and 1. This transforms LPM model to:

logπi1- πi=

α0 + α1jZij + α2jSij + α3jPij + α4jBij + α5jXij + ŋi + εij

Within the logistic model, πi =1, this equates that individuals paid a bribe and the value obtained from the results must be exponentiated to determine the odds ration relationship between the variable and bribery. This is used to test the following hypotheses.

Hypothesis 1: Poor Individuals have a greater frequency and likelihood of paying bribes, which differs among public services

Hypothesis 2: Public Service with exit option have a result in a higher level of bribery than those with no exit option

Hypothesis 3: Social Networks lead to a decrease in the likelihood of paying bribes, while Political Networks leads to an increase in the likelihood of paying bribes.

To explore the frequency of bribe payments, the baseline equation will be converted to an ordinal logistic model because the dependant variable has a natural ordering (0 = never, 1 = Once or Twice, 2 = A few times, 3 = Often). This will lead to the following equation:


δ (j) – (δ1jZij + δ2jSij + δ3jPij + δ4jBij + δ5jXij + ŋi + εij)

Hypothesis 4: Accountability measures, namely independent media, and civil society, decreases bureaucratic corruption.

A final adaptation of the model will be used to explore the accountability mechanisms effect on bribery. Jackson et. al. (2014)work is used as a template to formulate the equation; an interactive term between proxies for accountability and poverty is incorporated in to the logistic regression model. However, due to lack of accountability components included in the data set only voice can be used for enforcing accountability. Building on the components used by Mauro (1995) to represent voice (strength of civil society and free and independent media), the following equation can be derived:

logπi1- πi

= λ1Pov +

λ2Accountability +

λ3(Poverty)*(Accountability) +

λ4A1 + εi

Civil Society =

1, if its strong0, otherwise


Media =

1, if its strong0, otherwise

4.2.2. Poverty Index

Exploring previous literature, a variety of different factors are used as indicators for poverty. For example, Hunt and Laszlo (2012) use consumption when studying Uganda and Peru. However, it may not be the most efficient way as consumption may be a proxy for an individual’s choice rather than their poverty level. Although, Hunt (2007) uses income measures, studies from Khandker (1973) determine that respondents may be more likely to underestimate their wealth in survey data.

Because there is no direct measure of poverty in the data-set, hence economic status is used as it encompasses a multi-dimensional assessment of poverty and deprivation. However, this statistic is presented as a categorical variable and therefore needs to be adapted to act as a proxy for poverty index.

Figure 1 – Distribution of poverty Index

The proxy is derived from a series of questions that the individuals answered based on their accessibility of public services over the last year. The respondent answered with a value between 0 and 4 – (0=Never,1=Just once or twice, 3=Several times, 4= Many/Often, 5 =Always) depending on where their family has gone without:

(1) Food,

(2) clean water for home use,

(3) medical care,

(4) fuel to cook food, and

(5) cash income.

The responses are aggregated using equal weighting. Figure 1 shows the distribution of the index, with the higher values indicating poor individuals with lack of necessities. A further pairwise correlation, Table 1, is conducted to show that the results are positive and statistically significantly correlated at the 10% level or lower.

  Fuel Water Medical Care Cooking Fuel Cash Income
Fuel 1        
Water 0.485** 1      
Medical Care 0.458* 0.492** 1    
Cooking Fuel 0.415** 0.426* 0.409** 1  
Cash Income 0.482** 0.305* 0.392** 0.352* 1

** Significance at 10%, * Significance at 5%

Table 1 – Pairwise Correlation (components of the poverty index)

4.3 Data

The hypotheses will be tested using the dataset gather from Transparency International 2016, a cross sectional individual survey that explores corruption and bribery on an individual survey from both poverty perception and actual bribes paid. The survey was conducting in the latter stages of 2016 using all seven major languages spoken in the Pakistan as to avoid any miss communication. Originally, it contained 20,000 individuals, however a small percentage of individuals were removed because of missing information.

This study employs an orthodox methodology, based on an adapted from previous studies conducted over several decades. The work has been approaches through a neoclassical perspective as it contains explicit hypotheses that are confirmed or rejected through quantitative research (Lynch and Walsh, 1998). Research is derived from epistemological position of positivism due to the application of scientific methods used to derive knowledge and an objectivist stance whereby economic phenomena exist independent of economic agents.

4.4 Robustness

It is important to ensure reliability and validity of results; the following limitations are considered before any results are obtained. Firstly, Knack and Azfar (2000)observes that in larger countries where a variety of languages are spoken, there is a greater chance for misinterpretation by respondents however, because the data was collected in a variety of languages the effect will be negligible and thus this method of bias can be ignored.

A second bias builds on from the first, in that respondents could misreport bribery if they believe that the data is being collected by the government in fear of being disadvantages in some manner (Sequeira, 2012). To circumvent this downward social desirability bias, both logistic regressions will be re-estimated after excluding the proportion of respondents that believed that the data was being collect/financed by the government. In this way, the results will be able to determine whether the findings relating to the hypothesis remains robust or not.

Finally, a concern about the econometric technique used in raised in the works of Husted (1999). If some individuals were never asked to pay bribes by bureaucrats, this could lead to the coefficient of poverty to be overestimated. Therefore, to address this issue, this paper employs the method presented by Andersen (2009), where there is a two-stage econometric model estimated; using a hurdle model first which would eliminate all zero value and ensure that the model contains positive counts only. Given that the dependant variable, bribe payments, there is a chance for over-dispersion because of the binary nature of the variable and hence, the results will be replicated using a negative binominal model to show the robustness of the data.

5.0 Results and Discussion

When exploring Pakistan, Dawn News (2017)shows that it is ranked 116 out of the 175 countries on the corruption scale, however, referring to Figure 1 shows that the perception of corrupt in Pakistan is extremely high. The survey shows that police are perceived to be the most corrupt with 65.7% of the individuals saying that most, if not all, of the force is corrupt. Even members of parliament, who have the greatest level of accountability for their actions are deemed to be corrupt by almost 44% of the population, mostly likely a reflection the exercise of discretionary power show in prior experiences. Only, a small perception of less than 9% perceived that there was a lack of corruption within the bureaucratic system of Pakistan, matching the results of TI (2015).

Using the quantiles established using the poverty index, provide actual statistics on a variety of public services and bribery accepted by bureaucrats. The first quantile represents the poorer individuals in society increasing to the 4th quartile, rich individuals. At a quick observation, a cumulative value of individuals in the 2nd and 3rd quantiles exceed that of either the richest or the poorest within society, showing that the burden of bribery isn’t always borne by other on either end of the social scale.

The results show that the police are paid the most bribes by the respondent with almost 43% admitting to it. In this instance, the burden of bribery is borne by richer individuals, however those in the middle quantiles face a similar level of discrimination, paying support to Jain (2001).

On the other hand, the burden of bribery for acquiring permits (e.g. fishing), are borne by poorer individuals. Bardhan and Mookherjee (2000)gained similar results when exploring African countries, however found that this is most likely because richer individuals have better political connection and thus, can side step permits especially in countries with greater levels of corruption, like Pakistan. Though, this can be case, Peifer and Rose (2014)attributed a variety of social capital incentives offered to bureaucrats instead of the conventional monetary units as explored in this study. His findings showed because it is difficult to attributed a physical monetary value on some social capitals, makes it harder to determine its value in studies. Per Paul (1992), in both police and permits, bureaucrats hold a monopolistic advantage and can charge a higher bribery rates as alternative options are not available to the public.

  1st 2nd 3rd 4th Paid Bribes Didn’t pay Respondents
Permits 25.7 11.6 10.2 8.5 2052 17791 19843
Water or Sanitation Services 8.3 7.2 14.7 10.4 6378 13465 19843
Treatment at Local Clinic 7.1 12.3 10.3 6.1 985 18858 19843
Police 7.8 13.4 10.6 21.7 8503 11340 19843
Placement in Primary School 10.2 8.8 6.5 3.2 1872 17971 19843

Source: Corruption Pakistan 2016

Table 2 – Disaggregation of individuals who paid a bribe (%), using quantiles defined by the Poverty Index

Building on Peifer and Rose (2014)analysis, it is important to remember the table above presents results conducted by (Shafqat, 1999)concentrating solely on monetary incentives accepted by bureaucrats, and to study concepts like the Social Organisation argument and the Behavioural argument, more econometrics based approach is used.

5.1 Determinants of paying bribes

The linear and logistic results presented in Tables 1 and 2, dictate the results of determinants of paying bribes.  The first column contains the dependant variable, which is bribe paid. A value of 1 indicates that the individual has paid a bribe for any of the public services. The dependant variable has been disaggregated to examine the effects on each different type of public service. The results incorporate results for all five of Pakistan’s provinces (Sindh, Punjab, Kashmir, Baluchistan, Sarhad) to control for unobserved heterogeneity.

Starting with poverty index, the LPM model shows that the results are statistically significant for all the public services. An interpretation of poverty coefficient shows that poorer individuals are more likely to pay a bribe to bureaucrats. For example, holding all other units constant, a one unit increase in poverty correlates to an increase in bribery by 1.8%. Factors like health, police and permits have the greatest increases of 2.1%,1.6% and 1.5% respectively, while education has the lowest unit increase at 0.3%. Data from table 2 shows that the logistic model displays a similar relationship with each unit of bribery leading to increase in the odds ratio by between 2.7% (education) up to 9.6% (health). These disaggregated findings pay support to the social inequality argument presented by (Shafqat, 1999), drawing the same analysis at Know (2009)who uses cross sectional analysis to explore in Bangladesh, they further develop their argument to show that health has a greater effect on bribes than factors like education.

Dependant Variable: Bribe Paid Bribe Index Permits Water Health Police Education
Poverty Index 0.018**


















Religious Group Member -0.025


















Voluntary Group Index 0.138


















Contact Local Councillor 0.039***


















Contact with MP 0.09


















Contact with Gov. Agency 0.008*


















Contact with Political Party -0.009


















Cognitive Effect (trust) -0.007


















Control Variables Yes Yes Yes Yes Yes Yes
N 19843 19843 19843 19843 19843 19843
R2 0.039 0.065 0.021 0.025 0.045 0.031
F Statistic 11.28 4.46 3.24 8.92 6.56 7.53

(t statistic in parenthesis. Significance at *10%, **5%, ***1%)

Table 3: Linear Probability Model

As both LPM and logistic model display a similar relationship, it can be assumed that there is no severe bias in logistic model. Khan and Griffin (1972) analysis of South Eastern Asian (India, Pakistan etc.), the poorer individuals in society tend to bear the burden of bribery which matches the results obtained. This in turn, provides enough evidence to accept the first hypothesis and draws the same conclusion as Shah (2006)that bribery has an “adverse distributional effect and does not constitute an elite problem in Asian countries”.

When exploring the effects of each public service, Bernheim and Rangel (2007)explains that services that have an exit option (i.e. a private alternative is available to richer individuals.) often require a greater level of bribe as bureaucrats must gain from the losses of richer individuals moving to private alternative. This statement is supported as a unit increase in poverty leads to an increase in bribes by 9.6% within the public health department. This result within the healthcare, is almost identical to the 9.4% value obtained by Paul (1992)in his Indian study, a country which shares very similar economic characteristics with Pakistan. However, Imran et. al. (2011) questionnaire finds that because the healthcare in Pakistan is largely private, majority of the individuals choose to not use public which could be a reason that bureaucrats demand a higher level of bribe for it.

However, when comparing the results of public services with exit options to those with no exit option, like police and permits, a similar level of increase in bribery can be seen in both the LPM and logistic models. Using police as an example, an increase in one unit poverty leads to increase of 1.6% while in the logistic regression, leads to an increase in bribery by 5.5%. This result supports the work of Crilley (2013), which conclude that factors like police should have lower positive effect on bribery compared to monopolized public services like water supply, which in this case is an increase of 3.1% in bribery for each unit of poverty index in the logistic model – providing support for the second hypothesis.

There are a couple of supporting argument, firstly it can be attributed to accountability.Agha et. al. (2005) explains that most of the bribes in police can be attributed to traffic police, within India – which has a similar socio-economic environment. Because of the content of the job, majority of the work is held in the “field” making it harder to report bureaucrats and decreases accountability, although within this framework it is often rich individuals that get targeted, as a greater level of bribe can be extracted from it.

Secondly, Shah (2006) approach is based on an adaptation of the behavioural argument, where it is believed that bureaucrats may accept a lower level of bribes for services where exit options are available. However, they must make sure the utility gained by the individual paying the bribe is greater than the utility of the private option, otherwise the individual may forego the bribery to the alternative option. Either way, there isn’t enough evidence to support the second hypothesis and therefore must be rejected as all data collected is statistically significant.

Additionally, it can be seen from both the LPM and logistic regressions that the effect of political and social capital differ greatly.  The results support the social network argument by firstly, displaying a negative link with political networks, the main component present in the dataset for social capital (Kaufman and Wei, 2000). The logistic model shows that being a member of a religious group is associated with a decrease of 5.4% in bribes, assuming all other factors are kept constant. This finding is statistically significant at the 10% interval providing support to the works of Triesman (2000), which propose that an access to social organisations, like religious groups, is fundamental in “third part enforcement on officials in the public sector.”

However, North and Orman (2013)shows religion may not be the best component for determining the effects of social organisations because religion can be manipulated by religious scholars and bureaucrats to increase bribes.Paldam-Kyklos (2001) empirical studies on religions, treat religion as a cultural index and show that it has a positive effect on corruptions. However, when his study later develops based on different religion, his findings show conclusive support for a decrease in bribery within religiously dominated countries, like Pakistan.

Dependant Variable: Bribe Paid Bribe Index Permits Water Health Police Education
Poverty 0.041**


















Religious Group Member -0.056**


















Voluntary Group Member 0.181**


















Contact Local Councillor 0.040*


















Contact with MP 0.153*


















Contact with Gov. Agency 0.670


















Contact with Political Party -0.039


















Cognitive Effect (trust) -0.552


















Employment 0.670


















Education 0.042


















Religion 0.063


















Gender -0.420


















Age -0.055


















Urban -0.390


















Constant 0.059**


















N 19843 19843 19843 19843 19843 19843
Pseudo R2 0.029 0.058 0.037 0.041 0.025 0.071

(t statistic in parenthesis. Significance at *10%, **5%, ***1%)

Table 4: Binary Logistic Model

By contrast, membership in voluntary societies increases the likelihood of paying a bribe. The results support the social organisation argument, by showing there is an increased likelihood for paying bribes: 14.8% for water, 21.4% for health, 6.0% for police and 10.1% for education all within a reasonable significance interval. A similar positive and statistically significant relationship can be seen with political components in both LPM and logistic regression. When looking at contact with local MP for access to public services, increases the odds of bribery by 16.5%, with permits leading to an increase by 109%. These findings match the studies of Knack and Keefer (1997), that show pay strong support to the social organisation argument and hypothesis 3.

Finally, it is important to explore the how cognitive factors influence the probability of paying a bribe. In both LPM and logistic model, a negative value can be attributed to all public service, although it is not statistically significant with any reasonable confidence level. This result pays support to the finds of Uribe (2014), who show that social norms and ideologies are endogenous and constantly change over time. This could then explain the low explanatory power of cognitive factors as determinants of bribery across the different public service.

5.2 Determinants of the frequency of bribe payments

Consistent with the finds of the earlier LPM and logistic model, poor individuals have a greater likelihood of paying bribes as well as paying more frequently. Once again, the coefficient for poverty is positive and is statistically significant at the 5% level, even after the inclusion of several control variables and local states fixed effects.

The odds of paying bribes for poor individuals is multiplied by a factor of 11% (e0.105), which is equivalent to saying that, controlling for other explanatory variable, a 1 unit increase in the poverty index can be associated with a 11% increase in the odds of giving a response that indicates higher frequency in paying bribes in exchange for permits. The empirical results further show that despite an even distribution of the frequency of bribe payment across the different public services, the magnitude is stronger for health (13.5%), police (37.5%) and education (13.3%). These results support the descriptive statistics, with police work accepting bribes with greater frequency than other services as well as showing that water and education have a very similar frequency in both models.

Dependant Variable: Bribe Paid Permits Water Health Police Education  
Poverty 0.105**















Religious Group Member -0.004















Voluntary Group Member 0.025















Contact Local Councillor 0.189















Contact with MP 0.024















Contact with Gov. Agency -0.049















Contact with Political Party 0.039















Control Variable Yes Yes Yes Yes Yes
N 19843 19843 19843 19843 19843
Pseudo R2 0.059 0.036 0.038 0.021 0.052

 Table 5 – Ordinal Logistic Model

  Often Few Times Once or Twice Never
Permits 7.5 11.2 28.4 17
Water or Sanitation Services 3.2 8.5 5.7 22
Treatment at clinic 4.6 11 13.1 14
Police 2.2 21.2 12.6 12
Placement in Primary School 4.2 7.9 12.2 24

Table 6: Percentage of individuals who paid bribes disaggregated by frequency of payments

These finds support that of Shah(2007), who proposed that “corruption hits people when they’re down”. Ferraz and Finan (2007)exploration of bribery within police, finds that those working away from the office often gain more in bribes as there is a lower level of accountability and individuals gain the utility of not acquiring a larger fine or points. He further shows, which is supported by the findings above, that the burden of bribery falls disproportionately with the poor and within public services which exhibit costly exit options like education and health. These findings provide further enough support for hypothesis 1.

5.3 Role of Accountability Mechanisms

Earlier observations from data has shown that poor burden the bribery and therefore it is important for Pakistan to cushion the poor from bearing the burden by altering bureaucrat’s opportunistic behaviour and increasing levels of political awareness by tackling accountability mechanisms.

The table above show the effect of civil society movement which is both statistically significant and negative. In the instance when civil society is weak (i.e. equals zero), poor individuals are 12.2% likely to pay a bribe for public service, however when bureaucrats and public officials are made accountable off immorally accepting bribes (i.e. civil society equals 1), the likelihood of bribes halves to 6.6% (e (0.115-0.021)). These results are replicated when using media as a proxy of accountability instead of civil society. Figure 7 shows that there is a decrease in likelihood of bribes by 6.6% when civil society movement are used as an accountability mechanism. These findings support a plethora of literature in corruption, all the way from Gregory (1995)to Manzetti (2014) helping provide support for the final hypothesis.

  Bribe Index Permits Water Health Police Education
Poverty 0.115**


















Civil Society 0.019


















Poverty*Civil Society -0.021**


















Control Variables Yes Yes Yes Yes Yes Yes
N 19000 19000 19000 19000 19000 19000
Pseudo R2 0.019 0.016 0.012 0.021 0.017 0.012

Z statistic in parenthesis. Robust standard errors used. Significant *10%, **5%, ***1%

Table 7 – Binary Logistic Model: Role of Civil Society Movements

However, studies conducting the difference between the two accountability measures displayed slightly different results (Clark et. al., 1994).In their study of Asian countries, civil society had a greater effect on accountability than media but Campante (2014)suggests that in Pakistan media is a formidable method of accountability, backed with examples like it function in removing Prime Minister after the Panama Tax Evasion incident (The News,2017).  

  Bribe Index Permits Water Health Police Education
Poverty 0.161


















Media 0.157


















Poverty*Media -0.021


















Control Variables Yes Yes Yes Yes Yes Yes
N 19000 19000 19000 19000 19000 19000
Pseudo R2 0.021 0.018 0.024 0.009 0.008 0.017

Z statistic in parenthesis. Robust standard errors used. Significant *10%, **5%, ***1%

Table 8 – Binary Logistic Model: Role of Independent Media Movements

5.3.1 Policy Recommendations

Most importantly, the results show that an accountability measured clearly influence the decreasing of burden of bribery on poor individuals. Huther and Shah (2000)proposes that, firstly, that policies the tackle both corruption and decrease poverty should be implemented, which is currently lacking in Pakistan. Secondly, increasing exit options for poor individuals could be vital in decreasing burden on poor individuals, which can be achieved by subsiding private options for individuals with income constraints. This method will not only increase allow for a lower likelihood of paying bribes but could lead to a better service for the individual. Vannucci (2009)

Furthermore, supported by the works of Tulchan and Espach (2000),the government should promote membership in religious and community which will lead to an increase in solutions for information asymmetry and collective action problems which perpetuate corruption. Finally, building and incorporating an infrastructure to increase accountability with immoral actions of bribery are vital. Although, this study only explores two methods due to the lack of availability of reliable data, there are a vast amount of methods than can lead to a greater level of accountability and thus, make public services more accessible at a lower cost.

5.4 Limitations and Future Research

Although this study aims to be as conclusive and rigorous as possible, several caveats remain. Firstly, although data suggests that poorer individuals are more likely to pay, the magnitude of the payment is never discussed. This could mean the richer individuals pay a larger quantity and overall a greater amount although poorer individuals may have a greater likelihood. However, the dataset doesn’t provide quantitative data on the quantity of each bribe.

Secondly, an area that isn’t address within the dataset or any of the other studies is whether bribe is paid due to a demand from bureaucrats or whether its individuals that drive the bribery. And finally, the third issue relates to reverse causality. For example, the data collected shows that in Pakistan the burden of bribery is borne by poor individuals to access public services, however it is difficult to determine whether the poor individual might poorer to begin with and could perpetuate bribery in exchange for public services.

Finally, due to the trade-off between quantitative and qualitative techniques, the analysis doesn’t fully capture the underlying processes and mechanisms which account for variations in public service provision and accountability between better and worse performing local counties. Therefore, while there are concerns with the results are fully acknowledged, this study aims to offer vital insight in to the dynamics of bureaucratic bribery in Pakistan.

6.0 Conclusion

This following three main findings can be concluded from his study. Contrary on earlier believes, it can be concluded that in Pakistan bribery is a function of an individual’s income. The results show that poor individuals are 4.3% more likely to pay bribes to access public services. However, the public service has a systematic effect on the quantity of the bribe, with service that have exit options often leading to a lower level of bureaucratic corruption. Finally, the study shows how independent media and civil society have a significant effect on decreasing the level of bribery.

The following provide support to a plethora of papers covering the corruption debate around the globe. Starting with the implications of political and social capital, the current literature sits between the proportion that such networks preserves incentives through elite capture or lower the incentive of bribery by promoting trust and norms of civic cooperation which constrain bureaucrat’s behaviour (Knack and Keefer, 1997). Consistent with the theoretical proposition of Olsen (1982), the empirical results show that political capital increases the likelihood of bribery however social capital, in form of religious organisations, are associated with a lower likelihood to bribe.

Secondly, this study contributes to the literature of the determinants of bribe payments in public services provision. Majority of the literature uses data based on perception indexes, whereas this study uses an individual survey which mitigates the problem the problem of perception and cognitive biases such as bandwagon and halo effects. (Treisman, 2007). Finally, it adds value to literature on behavioural economics within corruption, where there is a lack of exploitation on Asian countries, namely Pakistan, although Pakistan is one of the largest countries in the world.


Andersen, L. (2006). The Authoritarian Executive? Horizontal and Vertical Accountability in Nicaragua. Latin American Politics and Society, 48(2), 141-169.

Andersen, T. B. (2009). E-Government as an anti-corrupt strategy. Information Economics and Policy, 21(3), 201-210.

Bardhan, P., & Mookherjee, D. (2000). Capture and Governance at Local and National Levels. American Economic Review, 90(2).

Bernheim, B., & Rangel, A. (2007). Behavioural Public Economics: Welfare and Policy Analysis with Non-standard Decision-Makers. Princeton University Press.

Bird, R. (2008). Tax Effort in Developing Countries and High Income Countries: The Impact of Corruption, Voice and Accountability. Economic Analysis and Policy, 38(1), 55-71.

Chang, E., & Chu, Y.-H. (2006). Corruption and Trust: Exceptionalism in Asian Democracies. Journal of Politics, 68(2).

Claudio, F., & Frederico, F. (2011). Electoral Accountability and Corruption: Evidence from the Audits of Local Governments. The American Economic Review, 101(38), 1274-1311.

Crilly, R. (2013, May 11). Nawaz Sharif claims victory in Pakistan election. Retrieved April 10, 2017, from Telegraph:

Dawn News. (2017, January). Dawn. Retrieved April 16, 2017, from Pakistan improves slightly to 116 rank on global corruption index:

Ferraz, C., & Finan, F. (2007). Electoral Accountability and Corruption in Local Governments: Evidence from Audit Reports. IZA Discussion Paper No. 2843.

Goetz, A., & Jenkins, R. (2004). Voice and Accountability in Service Delivery. World Bank Institute, Development Outreach. Washington: The World Bank.

Griffin, K., & Khan, A. (1972). Growth and Inequality in Pakistan. Oxford: The Macmillian Press Ltd.

Hirschman, A. (1970). Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States. Harvard University Press.

Hunt, J., & Laszlo, S. (2005). Bribery: Who pays, who refuses, what are the payoffs. National Bureau of Economic Research.

Huntington, S. (1968). Political Order in Changing Socities (7th ed.). United States of America, Baskerville: Yale University Press.

Husted, B. (1999). Wealth, Culture, and Corruption. Journal of International Business Studies, 30(2), 339-359.

Jain, A. (2011). Corruption: A review. Journal of Economic Surveys, 15(1), 71-121.

Khandker, R. (1973). Distribution of Income and Wealth in Pakistan. Pakistan Economic and Social Review, 11(1), 1-39.

Klitgaard, R. (1988). Controlling Corruption. California, United States of America: University of California Press.

Knack, S., & Azfar, O. (2000). Are Larger Countries Really More Corrupt? Policy Research Working Paper.

Lederman, D., Loayza, N., & Soares, R. (2005). Accountability and Corruption: Political Institutions Matter. Economics and Politics, 17(1), 1-35.

Leff, N. (1964). Economic Development Through Bureaucratic Corruption. American Behavioral Scientist, 8(3), 8-14.

Lewis, W., & Young, T. (2013). The Politics of Accountability. Teacher Education Policy, 27(2), 190-216.

Manzetti, L. (2014). Accountability and Corruption in Argentina During the Kirchners’ Era. Latin American Research Review, 2, 173-195.

Mashru, R. (2014, January 08). Pakistan: As Corrupt As Ever? Retrieved from The Diplomat:

Meon, P.-G., & Sekkat, K. (2003). Does Corruption grease or sand the wheels of growth. Public Choice, 122, 69-97.

Naqash, T. (2016, Feb 25). Imran accuses Zardari, Sharif of election rigging. Retrieved April 20, 2017, from Dawn News:

Quah, J. (1999). Corruption in Asian Countries: Can It Be Minimized? Public Administration Review, 59(6), 483-494.

Rajak, J. S. (2013). Corruption in India: Nature,Causes, Consquences and Cure. Journal of Humanities and Social Science, 18(5), 20-24.

Roy, A., & Singer, A. (2006). Reducing Corruption in International Business: Behavioural, Managerial and Political Approaches. Journal of Economic and Social Policy, 10(2).

Shafqat, S. (1999). Pakistani Bureaucracy: Crisis of Governance and Prospects of Reform. The Pakistan Development Review, 995-1017.

Shah, A. (2007). Performance Accountability and Combating Corruption. World Bank Group.

Siddiqi, M. (2011, April 25). A guide to the CSS: Want to change Pakistan? Try joining the government as an honest bureaucrat. Retrieved from The Express Tribune:

#dissertation #thesis #essay #assignment #phd #phdstudent #phdlife #essaywriting #research #phdjourney #dissertationlife #gradschool #assignments #phdproblems #academicwriting #assignmenthelp #thesiswriting #gradstudent #dissertationproblems #university #academiclife #gradschoolproblems #academia #researchpaper #gradschoollife #graduateschool #college #doctoralstudent #dissertationdone #writing

We can help you complete this assignment or another one similar to this. Just hit “Order Now” to get started!

error: Content is protected !!
Directly chat?
Do you need any help from us?
Thankyou for visiting our website. We can help you to place your order via the order system. Just send the instructions including attachments to our WhatsApp Live chat.
Thank you!