Anti-corruption measures in large-scale construction projects

Abstract

Accepted by: Konstantinos Nikolopoulos

Corruption is affecting many developing countries, manifested often in construction projects. This study identifies the factors causing corruption and prioritizes anti-corruption measures in large-scale urban construction projects with empirical data from a developing country: Iran. The model consists of six main dimensions including 24 measures and was developed by systematically reviewing the literature as well as collecting primary data through interviewing experts. The model prioritizes the anti-corruption measures through an integrated method of the fuzzy best-worst method and fuzzy measurement alternatives and ranking according to the compromise solution. The field of corruption has a multi-faceted nature and inherent uncertainty, which necessitates this integrated approach for its analysis. The results reveal that the ‘lawlessness and deregulation in public construction projects’ and ‘top management and leader commitment’ are the most important causes of corruption. This study offers two main contributions. First, it develops a conceptual model to evaluate and prioritize anti-corruption measures; second, it generates practical solution for reducing corruption in municipal and urban management, thus enhancing the prospects of successful construction projects in developing countries.

1. Introduction

Various types of industrial projects contribute to the socio-economic growth of countries. Among these, construction projects are considerably important as they involve a large volume of capital and have a major role in a country’s socio-economic development (Banihashemi et al. 2017). Despite this significance, studies show that most projects, including urban construction projects, encounter various problems, such as operational delays, increased cost and reduced quality (Lee et al. 2019). These usually arise from factors such as inflation, legal difficulties, defects in design, corruption, financial limitations and poor project management (Sonuga et al. 2002). These problems could slow down the process of social and economic development of countries, increase poverty in society and reduce investment coming from domestic and foreign resources (Osei-Tutu et al. 2010).

Because massive financial resources allocated to large urban construction projects each year, such projects are targets for corruption. More specifically, countries lose a large amount of capital each year as a result of profiteering in large construction projects, although the capital could be spent on development plans. In some countries, the annual damages caused by corruption may amount to billions of dollars. Recent research conducted by Transparency International confirms that construction projects, despite their constructive and important function in the overall development of countries, involve a highest degree of corruption on a global scale. Similarly, the American Society of Civil Engineers estimates the annual corruption damage affecting the industry to be about $340 billion worldwide (Sohail & Cavill, 2008).

Corruption in construction projects can be defined as the misuse of power for personal gain (Le et al. 2014a). Corruption in these projects can take various forms, such as bribery, fraud, auction fraud, extortion, money laundering, collusion, embezzlement and nepotism. Such corrupt actions may occur in different phases of a project including identification, planning, finance, design, bidding, execution and maintenance (Stansbury, 2005; Sohail & Cavill, 2008; Jong et al. 2009; Le et al. 2014b; Shah & Alotaibi, 2017; Damoah et al. 2018).

Construction projects in developing countries face more severe challenges due to their increased exposure to corruption (Owusu et al. 2019). For example, Iran has a low Corruption Perceptions Index (CPI) according to the report published by Transparency International; it ranked 146th out of 180 countries in terms of corruption in 2019 (Transparency International, 2021). The Iranian Planning and Budget Organization reported that the budget of construction projects amounted to approximately 3.5 billion dollars in Iran in 2020 (Planning and Budget Organization, 2020). The global corruption rate in this industry is 10% (Azhar & Selph, 2011), which results in the loss of thousand billion tomans (thousand million dollars) annually from the country’s financial resources that could otherwise contribute to the country’s development, to the benefit of profiteers. Motivated by providing practical findings in order to enhance the situation of construction projects in the developing countries, the purpose of this study is to identify the factors causing corruption and to find and prioritize anti-corruption measures in large-scale urban construction projects in Shiraz, one of the major cities in Iran. In doing so, the study relies on multi-attribute decision-making (MADM) techniques in a fuzzy environment. Shiraz is a suitable case for this investigation because it has a large number of civil construction projects and a high turnover rate. In 2019, the city initiated 52 construction projects, which represented 71.1% of its total budget.

The study primarily collects information about the topic through conducting a systematic review of the literature and through interviewing experts. Following that, the study identifies the causal factors of corruption and anti-corruption measures in large construction projects. Next, it draws on the fuzzy best–worst method (FBWM) to the weight the casual factors. BWM has several advantages over other traditional weight calculation methods such as the analytic hierarchy process (AHP) and analytical network process (ANP). These advantages include fewer pairwise comparisons, more consistent results and the capability to integrate better with other decision-making techniques (Ahmadi et al. 2017; Shojaei et al. 2017; Sofuoğlu, 2020). Moreover, the BWM has been shown to be effective compared to more recent weight calculation methods such as the full consistency method (FUCOM), as their difference in consistency is very negligible (Haqbin, 2022). This study chose the FBWM to calculate the weights of the corruption-causing factors over other methods such as FUCOM (Pamucar et al. 2018), ordinal priority approach (OPA) (Ataei et al. 2020) and level-based weight assessment (LBWA) (Žižović & Pamučar, 2019) for two main reasons. First, Mi et al. (2019) suggested that the appropriate multiple-criteria decision-making (MCDM) methods should be selected based on the problem structure. The hierarchical model of the corruption factors seems to be more compatible with BWM. Second, while other methods such as OPA have limited applications in the literature (Le & Nhieu, 2022), the accuracy of BWM results has been well validated by previous studies. The study then applies the fuzzy measurement alternatives and ranking according to the compromise solution (FMARCOS) method to rank the anti-corruption measures identified. The MARCOS method has several advantages over other MCDM methods. Although the MARCOS method is similar to the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) method in that the best alternative is located closest to the ideal solution and farthest from the anti-ideal solution (Bakır & Atalık, 2021), the MARCOS method can produce more reliable results because of the combination of the ratio approach and the reference point sorting approach (Deveci et al. 2021). Furthermore, Stanković et al. (2020) stated that the benefits of FMARCOS include developing the model by considering fuzzy reference points through the fuzzy ideal and fuzzy anti-ideal solution at the outset, calculating the degree of utility with respect to both set solutions with more accuracy and the capability to consider a large set of criteria and alternatives (which is the most compatible with the hierarchical model of this study).

Experts’ opinions are commonly used in decision-making techniques, but they are subject to uncertainty. However, the study employs MADM techniques in a fuzzy environment to address the uncertainty of the experts’ opinions (Bakır et al. 2021). Moreover, the combination of BWM and MARCOS in a fuzzy environment has produced consistent results in recent studies on various topics (Celik & Gul, 2021; Altay et al. 2023; Koohathongsumrit & Chankham, 2023).

Most of the previous studies on corruption in construction projects have used either quantitative or qualitative methods. However, few studies have adopted a mixed research framework that combines qualitative and quantitative methods in examining the topic under investigation. Moreover, this study can be regarded as novel because, to the best of the authors’ knowledge, it is the first attempt to rank anti-corruption measures based on a fuzzy MADM model in a developing country (Iran). This study also offers two major contributions: (a) the main contribution of this research is the conceptual model of corruption-causing factors and anti-corruption measures in a hierarchical structure, which enables the ranking of the measures using fuzzy MADM techniques, and (b) the findings also highlight the importance of public-sector managers and policy-makers (especially those involved in municipalities’ project management and urban management) to be aware of the sources of corruption and the measures they can implement to prevent corruption in such projects as much as possible. This preventive process could consequently enhance the prospects of successful construction projects in Iran and other developing countries worldwide.

2. Literature review

Many studies have explored corruption in construction projects. Zou (2006) reviewed commonly used measures to prevent corruption in the construction industry, offering ideas to improve them. Zou’s research revealed that current anti-corruption measures were more reactive than active and that it was necessary to improve the legal system and regulatory processes/strategies and to promote an ethical culture. Furthermore, measures such as regular and random inspections, severe punishment and prosecution of corrupt employees and a healthy construction culture were the only practical ways of ensuring effective and efficient performance. Tabish & Jha (2011a) sought to identify the factors that could increase the success of public construction projects while determining the relative importance of these factors in the overall performance of projects. Their findings identified four general factors were (a) awareness of and compliance with the rules and regulations; (b) pre-project planning and clarity of the scope of the project; (c) supervision and effective partnership among project partners; and (d) the full participation of the project owner. The results also showed that awareness of and compliance with the laws and regulations left more impact than the others factors on the overall performance of construction projects.

Gunduz & Onder (2013) explained the common types of fraud in the construction industry, as well as the reasons they would occur. They ultimately proposed methods to prevent fraud. The analysis of the data obtained from the survey confirmed that some measures reduced fraud and corruption in the industry. Such factors were governance, conducting internal controls, checking the background of employees before hiring them, conducting internal and external audits, training staff on fraud policies and procedures, developing ethical and behavioural regulations and having efficient systems for reporting corruption. Le et al. (2014b) conducted a systematic review, exploring corruption-related issues in construction management in engineering journals and the direction of future research investigating corruption in construction. The authors mentioned various forms of corruption, the destructive effects of corruption at various micro and macro levels and strategies to eliminate corruption.

Deng et al. (2014) examined the main causes and origins of fraud in the construction industry. As the statistical analysis of the results demonstrated, chances of fraud in the industry were increased by some factors such as pressure/dissatisfaction experienced by workers and managers, frequent changes in projects, a lack of continuous monitoring of project sites due to their large size, long working hours, a lack of ethical codes and a lack of professional ethical training. Bowen et al. (2015) explored how customers, managers and construction professionals perceived corruption. According to the findings of this survey, corruption in this industry was a widespread phenomenon and was mostly manifested in such forms as conflicts of interest, fraud, collusion in the bidding process and bribery. Furthermore, government officials and project (sub)contractors were more likely to become involved in corruption in the tender process. Factors that facilitated corruption were a lack of transparency in contracts, private opening of tenders and the operational environment of the construction industry.

Responding to research gaps in terms of measuring corruption in construction projects, Shan et al. (2015) created a systematic model for measuring corruption. The model sought to improve the level of monitoring and evaluation of construction projects. In doing so, Shan et al. categorized 24 factors into five dimensions: immorality, unfairness, lack of transparency, procedural violation and contractual violation. Finally, they used the fuzzy set theory to quantify the overall level of corruption, trying to overcome problems associated with ambiguity, subjectivity and uncertainty in measuring corruption. Cerqueti & Coppier (2016) examined the interaction among firms, tax inspectors and politicians in a corrupt context by applying the game theory. They found that the compliance channel is more effective in a country with a low level of incentives. Zhang et al. (2017) examined the main causes of business-to-government (B2G) corruption. To accomplish this, they inspected the relative impact of this type of corruption on the bidding process to reduce and eliminate corruption in the construction industry. They divided the causes of corruption into six main dimensions, namely, flawed regulation systems, negative encouragement, a lack of professional ethics and codes of conduct, illegitimate gain, a lack of competitive and inequitable bidding practices and the impact of guanxi. Ameyaw et al. (2017) sought to investigate the prevalence rate of corruption and its various forms. Their findings indicated that corruption and immoral behaviour were more common among government officials, contractors and industry experts, and they could occur in various stages such as bid evaluation, bidding and contract execution. The most important issues that contributed to corruption were enhancing secrecy of public contracts due to political relations, excessive and reckless sole sourcing practiced in construction projects, construction companies’ refusal to address corruption in their mission statement and the simplicity of covering up corrupt activities in the operational environment in which construction projects take place. In another study, Cerqueti & Coppier (2018) explored the link between bureaucratic corruption and political corruption by using a theoretical game model. They discovered that political and bureaucratic corruption can coexist at a macro level. Furthermore, political and bureaucratic corruption are substitutes at the level of the firm because they rely on the capital of the firm. Owusu et al. (2020b) examined the effectiveness of anti-corruption measures in preventing the spread of corruption in the infrastructure procurement process in developing countries. The results of this evaluation showed that among the measures under investigation, probing measures, followed by management measures, were identified as the most effective anti-corruption measures in Ghana. Yap et al. (2020) explored issues such as the impact of corruption on project outcomes, the causes of corruption and the evaluation of anti-corruption measures. The analysis of the data obtained from copies of a questionnaire revealed that among the 18 causes of corruption, negative incentives, the nature of the construction industry and malfunctional monitoring systems were the three most important factors, respectively. Meanwhile, among the 11 anti-corruption measures identified, strict law enforcement, regulation and punishment, an honest construction culture and effective reporting channels displayed the highest ranks among preventive measures, respectively.

Furthermore, Opoku et al. (2022) carried out a qualitative research study to identify the causes and prevention strategies for corruption in Thailand’s construction industry. They interviewed 12 professionals and found that corruption is mainly caused by personal behaviour, red tape, conflicts of laws and organizational culture. To tackle corruption, the authors propose several measures such as improving organizational systems, decentralizing power, providing ethical training and fostering an ethical culture. Kiyabo (2022) conducted a thorough review of existing literature to gain deeper insights into corruption within construction projects. The finding of this study showed that corruption is widespread across the entire project lifecycle, from inception to completion, as well as infrastructure services associated with them. This pervasive corruption is influenced by multiple factors such as project characteristics, regulatory aspects and personal factors. To combat this issue effectively, the study suggests a range of interventions, including raising managerial and community awareness, implementing regulatory measures and collaborating among organizations. Soni & Smallwood (2023) discovered that bribery is prevalent in South Africa’s construction industry and inhibits whistle-blowing. Respondents concurred that corruption has negative effects on the industry’s economic growth, resulting in delays, poor workmanship and the use of substandard materials. Table 1 lists a summary of the articles reviewed in this study.

This review of the studies exploring corruption and anti-corruption measures in construction projects highlights the breadth and importance of topics and concerns in this area of research. The review also clarifies that numerous studies have tried to identify types of corruption, factors causing corruption and anti-corruption measures practiced worldwide in construction projects. However, the situation in Iran, as a developing country, has remained relatively unknown, apparently due to the high sensitivity of this issue in the country. In addition, currently there is no comprehensive model in the literature that could identify and process the causes of corruption, as well as anti-corruption measures, in construction projects. Given these considerations, the present study seeks to construct a model that identifies the causal factors of corruption, as well as anti-corruption measures, in large-scale urban construction projects in one of the major municipalities in Iran.

3. Methodology

3.1 Aims of the study

The purpose of this study was to identify the causes of corruption and to prioritize anti-corruption measures in large-scale urban construction projects through MADM techniques in a fuzzy environment in Shiraz, as a major city in Iran. Most of the previous studies on corruption in construction projects have used either quantitative or qualitative methods. However, few studies have adopted a mixed research framework that combines qualitative and quantitative methods in examining the topic under investigation. Moreover, this study is the first attempt to rank anti-corruption measures based on a fuzzy MADM model in a developing country (Iran).

3.2 Motivation for developing the methodology

In doing so, primarily the literature was reviewed and experts in this field were interviewed. Prisma tools were used to systematically analyze the literature. In addition, the interviews were transcribed and analyzed through qualitative content analysis. Next, the casual factors and anti-corruption measures were measured through content validity ratio (CVR). As result of this stage, a conceptual model was constructed that involved two major parts: the causes of corruption and anti-corruption measures to overcome such factors. Following that, the causes of corruption were weighted using the FBWM. Finally, the anti-corruption measures were prioritized using the FMARCOS method. Below, the analysis methods are further elaborated on. Figure 1 illustrates the process of conducting this research.

3.3 Content validity ratio

To ultimately confirm the factors causing corruption in large-scale urban construction projects, as well as the anti-corruption measures, the validity of the factors would have to be measured. Several methods could help to measure validity, although the CVR represents an extensively used method (Almanasreh et al. 2019). Developed by Lawshe (1975), the CVR calculates content validity based on expert opinions. In this study, the experts were asked to rate each question based on a scale (including items such as ‘Essential’, ‘Useful, But Not Necessary’ and ‘Not Necessary’). Then, according to the following formula, CVR was calculated, where N is the total number of experts and ne is the number of experts who chose the ‘Essential’ item (Lawshe, 1975).

Based on the number of experts who evaluated the questions, the minimum acceptable value for this index was determined. Factors that did not show an acceptable content validity rate were excluded.

3.4 Fuzzy best-worst method

The best-worst method is an MCDM technique proposed by Rezaei (2015) that is based on pairwise comparisons to obtain the weights of alternatives and criteria respective to various criteria (Shojaei et al. 2017). Guo & Zhao (2017) adapted the BWM to a fuzzy environment, solving the new model through several examples. The use of fuzzy numbers could help to overcome ambiguities in respondents’ opinions. Because the FBWM is a combination of the fuzzy set theory and the traditional BWM, it provides more reliable weights than the original BWM, and for this reason, it enhances the validity of decisions made via this technique. The FBWM is adopted in various subjects including sustainable supplier selection (Amiri et al. 2021; Bonab et al. 2023), sustainable urban development evaluation (Foroozesh et al. 2022) and site selection for renewable energy systems (Aghaloo et al. 2023). In addition, because the FBWM uses only five linguistic terms instead of a 9-point scale, it simplifies comparisons and reduces confusion for decision-makers. The FBWM steps for weighting the effective factors are as follows:

Step 1. Determining the set of decision criteria. In this step, the indicators are defined as {c1, c2,…, cn} for decision-making. In the present study, the causes of corruption in construction projects are decision-making criteria.

Step 2. At this stage, the best (most important and most desirable) and worst (least important) criteria are determined. The best criterion is Cb and the worst criterion is called Cw.

Step 3. Determining the preference of the best criterion compared to other criteria according to the linguistic terms in Table 2. The best-to-others vector is defined as follows:

where |${a}_{Bj}$| indicates the preference of the best criterion B over criterion j and it is obvious that

Step 4. Determining the preference of the worst criterion compared to other criteria according to the linguistic terms in Table 2. The others-to-worst vector is as follows:

where |${a}_{jW}$| indicates the preference of the criterion j over the worst criterion |$W$| and it is obvious that

Step 5. Find the optimal weights|$\left({w}_1^{\ast },{w}_2^{\ast },\dots, {w}_n^{\ast}\right)$|⁠. To determine the optimal weight of each criterion, the pairs |$\frac{W_B}{W_j}={a}_{BJ}$| and |$\frac{W_j}{W_W}={a}_{JW}$| were considered. To meet these conditions, a solution must be found to maximize |$\left|\frac{{\mathrm{W}}_{\mathrm{B}}}{{\mathrm{W}}_{\mathrm{j}}}-{\mathrm{a}}_{\mathrm{B}\mathrm{J}}\right|$| and |$\left|\frac{W_j}{W_W}-{a}_{JW}\right|$| for all js that have been minimized. It should be noted that W_B, W_j and W_W are triangular fuzzy numbers. The model can be formulated as follows:

By solving the above model, the optimal values (W₁^*,W₂^*, …,W_n^*) were obtained.

Step 6. Find the inconsistency rate of the FBWM. In the last step, after solving the model and extracting weight, the inconsistency rate is calculated using the equations mentioned in Guo & Zhao (2017).

3.5 FMARCOS method

The MARCOS method is a new MADM tool (Stević et al. 2020). This method regulates ranking based on the distance of alternatives from the ideal solution and the anti-ideal solution, in accordance with the criteria defined and their aggregation in a utility function (Stević et al. 2020). This method has many advantages over other MADM methods including the following ones: (a) it considers an anti-ideal solution and an ideal solution when the initial matrix is created; (b) it provides a closer determination of the utility degree in relation to both solutions; (c) it proposes a new way to determine utility functions and their aggregation; and (d) it makes it possible to consider a large set of criteria/alternatives while maintaining the stability of the method (Stević et al. 2020, p. 1). FMARCOS was used in different topics ranging from selecting the most appropriate equipment (Huskanović et al. 2023; Tešić et al. 2023) to choosing the best organizational structure (Khosravi et al. 2022). The FMARCOS method is conducted through the following steps (Stanković et al. 2020):

Step 1. Creating an initial fuzzy decision-making matrix. MCDM models include the definition of a set of n criteria and m alternatives. In this method, the linguistic terms for evaluating alternatives are defined in Table 3.

Step 2. Creating an extended initial fuzzy matrix. The extension is performed by determining the fuzzy anti-ideal |$\tilde{A}(AI)$| and fuzzy ideal |$\tilde{A}(ID)$| solution.

The fuzzy |$\tilde{A}(AI)$|is the worst alternative, while the fuzzy |$\tilde{A}(ID)$| is an alternative with the best performance. Depending on the type of criteria, |$\tilde{A}(AI)$| and |$\tilde{A}(ID)$| are defined by applying Equations (6) and (7):

B belongs to the maximization group of criteria, while C belongs to the minimization group of criteria.

Step 3. Creating a normalized fuzzy matrix |$\tilde{N}={\big[{\tilde{n}}_{ij}\big]}_{m\times n}$| obtained by applying Equations (8) and (9):

where elements |${x}_{ij}^l,{x}_{ij}^m,{x}_{ij}^u$| and |${x}_{id}^l,{x}_{id}^m,{x}_{id}^u$| represent the elements of the matrix |$\tilde{x}$|⁠.

Step 4. Computation of the weighted fuzzy matrix |$\tilde{\mathrm{V}}={\big[{\tilde{\ v}}_{ij}\big]}_{m\times n}$|⁠. Matrix |$\tilde{\mathrm{V}}$| is calculated by multiplying matrix |$\tilde{\mathrm{N}}$| with the fuzzy weight coefficients of the criterion |${\tilde{\mathrm{w}}}_{\mathrm{j}}$|⁠, Equation (10).

Step 5. Calculation of |${\tilde{\mathrm{s}}}_{\mathrm{i}}$| fuzzy matrix using the following Equation (11):

where |${\tilde{S}}_i\left({s}_i^l,{s}_i^m,{s}_i^u\right)$|represent the sum of the elements of the weighted fuzzy matrix |$\tilde{\mathrm{V}}$|⁠.

Step 6. Calculation of the utility degree of alternatives |${\tilde{\mathrm{K}}}_{\mathrm{i}}$| by applying Equations (12) and (13).

Step 7. Calculation of fuzzy matrix |${\tilde{\mathrm{T}}}_{\mathrm{i}}$|using Equation (14).

Then, it is necessary to determine a new fuzzy number |$\tilde{D}$| using Equation (15).

Next, it is necessary to de-fuzzify the number |$\tilde{D}$| using the expression |${df}_{crisp}=\frac{l+4m+u}{6}$| and obtaining the number |${df}_{crisp}$|⁠.

Step 8. Determination of utility functions in relation to the ideal |$f\big({\tilde{K}}_i^{+}\big)$| and anti-ideal |$f\big({\tilde{K}}_i^{-}\big)$| solution by applying Equations (16) and (17).

After that, it is necessary to perform defuzzification for |$f\big({\tilde{K}}_i^{+}\big)$|,|$f\big({\tilde{K}}_i^{-}\big),{\tilde{K}}_i^{+}{\tilde{,K}}_i^{-}$| and apply the following step:

Step 9. Determination of the utility function of alternatives |$f\left({K}_i\right)$| by Equation (18).

Step 10. Ranking the alternatives based on the final values of utility functions. It is desirable that an alternative have the highest possible value of the utility function.

4. Results

4.1 Corruption factors and anti-corruption measures

To explore the literature on the topic, the study primarily searched the databases of Scopus, Web of Science, Emerald Insight, Science Direct, Sage and Google Scholar. As a result of the initial search procedure, 2059 articles were found, out of which 1139 ones were kept after the duplicate articles were removed. At the next stage, the articles were evaluated by considering their titles and abstracts. Following that, 518 articles were eliminated after their titles were inspected, whereas 135 articles were removed after their abstracts were examined. Out of the remaining 161 articles, 117 ones were removed following a full-text analysis. Ultimately, 44 articles were used to extract the factors and measures related to the research topic. Figure 2 depicts the search process conducted on the literature. Following a rigorous investigation of the full texts of the remaining articles, 57 factors were identified as the causes of corruption in large construction projects, while 36 anti-corruption measures were found that could be employed to overcome corruption in such projects.  Appendix A lists the factors and measures.

To select the experts (participants), the purposive sampling method was used. In this process, the sample was selected from five different groups including employers, contractors, consulting engineers, officials in supervisory organizations and university professors with more than 15 years of work experience. Following that, eight highly competent experts were selected who participated in structured interviews and elaborated on and further complemented the factors and measures extracted from the literature review. Next, by analyzing the transcribed versions of the interviews, several causes and measures were identified.  Appendix B shows the causes and measures extracted through the qualitative content analysis of the interviews.

In this research, industry and academic experts examined the content validity of the factors and measures extracted from the literature and the interviews. The data were gathered through copies of a questionnaire. Out of the copies submitted to the experts, 24 ones were properly completed and returned. In the light of the CVR calculations, the factors and measures that showed a value less than 0.37 were removed. After the experts determined the content validity ratios, six dimensions (organizational, psychological, project-related, legal, statutory and cultural) and 22 factors (causes) were identified (as described below). The final model of the research can be seen in Fig. 3.

4.2 Weighting the causes of corruption and prioritizing the measures

The six main dimensions and the 22 causes identified in the previous step were used to design an FBWM-based questionnaire. Copies of the questionnaire were completed by eight experts who had technical knowledge of the concerns addressed in this study including municipal planning, contract affairs, health planning and development, renovation and urban transformation. To explain further, each expert filled out seven questionnaires, one for calculating the weights of the main dimensions and six for determining the weights of sub-dimensions for each main dimension. First, a set of decision criteria was established based on the developed model in Fig. 3. Then, the best (most important and most desirable) and worst (least important) criteria were identified. Next, the optimal weights were calculated by constructing and solving the model using Lingo software. Finally, the model was solved and the inconsistency rate (Guo & Zhao, 2017) of the model was computed. Likewise, the models for all the dimensions and sub-dimensions were calculated for all the experts. After ensuring that all the models have the inconsistency rate lower than 0.1, the final weights were also determined using arithmetic mean of all the experts, which are presented in Table 4.

The study drew on the FMARCOS method to rank anti-corruption measures in large-scale urban construction projects managed by Shiraz Municipality. For this purpose, after calculating the CVR values, 24 anti-corruption measures were selected to be included in the prioritization process. The 24 anti-corruption measures, along with the 22 causes of corruption extracted in the previous stages, were used to design a questionnaire. Copies of this questionnaire were submitted to the same eight experts (who had participated in the previous stage). Following the steps of the FMARCOS method, the data were primarily collected in the form of fuzzy numbers, although it was then necessary to integrate experts’ opinions into a single matrix. This study used the arithmetic mean to integrate the experts’ opinions (see Mijajlović et al. 2020). First, an initial fuzzy decision-making matrix including n criteria and m alternatives was created for each expert using the linguistic terms for evaluating alternatives. Also, the aggregated initial fuzzy decision-making matrix was obtained using the arithmetic mean. Then, the extended initial fuzzy matrix was created by determining the fuzzy anti-ideal A ~(AI) and fuzzy ideal A ~(ID) solution. After that, the normalized fuzzy matrix was created. By following the rest of the steps, the final utility function of alternatives was determined. Table 5 shows the final utility function of alternatives and final ranking of the measures.

4.2.1. Validation of results and sensitivity analysis

To evaluate the results obtained using FMARCOS, the solutions will be ranked using three other methods: fuzzy simple additive weighting (FSAW) (Roszkowska & Kacprzak, 2016), FTOPSIS (Patil & Kant, 2014) and fuzzy multi-attributive border approximation area comparison (FMABAC) (Bozanic et al. 2018). The ranking results of using these methods are shown in Fig. 4. As can be seen, there is very little difference between the rankings. Indeed, it has been shown that the results using the FMARCOS do not deviate from the results obtained using other fuzzy methods. To verify the results, Spearman’s correlation coefficient (SCC) (Bozanic et al. 2022) was applied. The SCC values are given in Table 6. The SCC values range from 0.957 to 1. This indicates a very high rank correlation value. Therefore, it can be concluded that the results of the FMARCOS method are satisfactory, and the robustness of the presented method has been demonstrated.

5. Discussion

This study explored corruption in large-scale urban construction projects, proposing a model that processed both causes of corruption and anti-corruption measures. The elements constituting the model, namely, the causes and the measures, were extracted from the literature on the topic and from the opinions of experts who participated in this research. The elements were finally confirmed through CVR calculations. The model was composed of two major parts; the first part addressed the causal factors of corruption in large-scale urban construction projects, and the second part was concerned with anti-corruption measures in such projects.

In the literature on this topic, some studies have proposed models. One of the important ones was constructed by Owusu et al. (2017), who included ‘organizational’, ‘psychological’, ‘project-related’, ‘legal’ and ‘statutory’ dimensions and identified the causes of corruption in different cultures. The present study tried to further complement the model of Owusu et al. (2017), by introducing a ‘cultural’ dimension to their original model. The addition of the ‘cultural’ dimension represents one of the major contributions of this study.

In addition to expanding the dimensions, the study further explored the causes of corruption and anti-corruption measures, by conducting interviews with eight experts active in this field and by rigorously reviewing the literature. The validity of all causes and measures identified in the study was confirmed via the CVR. From a methodological perspective, this study relied on the FBWM and the FMARCOS method to rank the causes and measures, which can further highlight the novelty of this research and its difference from other investigations dealing with construction projects.

A review and examination of the previous studies and researches in the field of corruption and anti-corruption measures in construction projects reveals that most of the researches conducted so far have relied on quantitative or qualitative methods.

Some research that addressed corruption in construction projects primarily employed quantitative and statistical methodologies, such as Tabish & Jha (2011b), Gunduz & Onder (2013), Deng et al. (2014) and Ameyaw et al. (2017). Others, such as Le et al. (2014b) and Owusu et al. (2017), reviewed corruption in projects. Some also conducted qualitative content analysis of corruption and its contexts, such as Brown & Loosemore (2015) and Chen & Wang (2017). Some research also applied mixed qualitative and statistical methods to analyze the findings, such as Arewa & Farrell (2015) and Zhang et al. (2017). Moreover, data mining and fuzzy methods were used in the research of Yu et al. (2019), Owusu et al. (2020d) and Owusu et al. (2020c). Based on this, the field of multi-attribute decision-making in the fuzzy environment, which was employed in the present study, creates a novel approach in the analysis of the field of corruption and solutions to cope with it in this scientific field.

The results of this study indicated that, among the 22 causes in the proposed model, ‘lawlessness and deregulation in public construction projects’, ‘lack of accurate, genuine, and rigorous supervision’ and ‘structural and organizational malfunctions’ were among the most important causes of corruption in large-scale urban municipal projects.

This observation was in line with the findings of Yap et al. (2020), Owusu et al. (2020b), Owusu et al. (2019), Saim et al. (2019) and Tabish & Jha (2018). On the other hand, factors such as ‘failure to estimate or totally miscalculate cost before launching projects’, ‘ignoring auditing culture’, ‘failure to disclose or report cases of corruption’ and ‘failure to implement the standards of project management’ were not considered very important in this study although they were highly stressed by Owusu et al. (2020a), Owusu et al. (2017), Ameyaw et al. (2017) and Arewa & Farrell (2015). This difference could be attributed to cultural differences that influenced the environments in which the studies were conducted.

Moreover, among the 24 anti-corruption measures, ‘top management and leader commitment’, ‘improving transparency mechanisms’ and ‘developing and implementing professional guidelines/standards’ showed the highest ranks, respectively. These findings were consistent with the observations of studies conducted by Yap et al. (2020), Owusu et al. (2020d), Owusu et al. (2020b), Owusu et al. (2020a), Owusu et al. (2019), Tabish & Jha (2018) and Chen & Wang (2017). Meanwhile, ‘Determine the minimum and maximum required budget’, ‘improving working conditions and raising employees’ subsistence rates’, ‘selecting contractors based on specific criteria’ and ‘employing a quality management system in different project stages’ were among the least important factors observed in this study. They were, however, more stressed in the studies conducted by Yap et al. (2020), Owusu et al. (2020b), Adamu et al. (2018), Rizk et al. (2018), Tabish & Jha (2018), Le et al. (2014a), Deng et al. (2014), Gunduz & Onder (2013), Ma & Xu (2009), Kenny (2009), Sohail & Cavill (2008) and Lester (1999). This difference in observations, too, could be explained by considering different cultural conditions under which the studies were conducted.

5.1 Managerial implications

The results of present study revealed that the lawlessness and deregulation in public construction projects is the most important cause of corruption, and top management and leader commitment is the most significant anti-corruption measure in large urban construction projects in Iran. Considering that, this paper provides the best anti-corruption measures in construction projects, it gives the managers and policy-makers of government sectors, especially in the fields of project management and urban management of municipalities, the opportunity to improve administrative health as much as possible by knowing these measures, followed by preventing the occurrence of corruption in these projects and finally increasing the chances of success of the projects. Among the measures presented in this research, top management and leader commitment, improving transparency mechanisms, developing and implementing professional guidelines/standards, developing and implementing rules and regulations and preventing institutional corruption and implementing administrative reforms along with paying attention to cultural dimensions have the greatest impact in overcoming causes of corruption.

6. Conclusion

The purpose of this study was to identify the causes of corruption and to prioritize anti-corruption measures in large-scale urban construction projects undertaken by Shiraz Municipality. In doing so, the study drew on MADM techniques in a fuzzy environment. This research makes a significant contribution by developing a conceptual model of causal factors of corruption and anti-corruption measures in a hierarchical structure, which enables the easy ranking of the measures using fuzzy MADM techniques. Another innovative aspect of this study is to raise the awareness of public-sector managers and policy-makers (especially those engaged in municipalities’ project management and urban planning) about the sources of corruption and the strategies they can adopt to prevent corruption in such projects as much as possible. This preventive process could consequently enhance the chances of success in construction projects in developing countries worldwide. The results of this study could help managers and policy-makers in the public sector (especially in the areas of project management and urban management of municipalities) to raise their awareness of anti-corruption measures so that they can increase the health of their projects, prevent corruption in them and ultimately increase the chances of project success. Naturally, a research investigation inevitably involves some limitations and problems. This study also faced some limitations, the most significant of which was the COVID-19 crisis. In particular, some of the data required for this research were obtained through in-person/online interviews with experts and through questionnaires, and the ongoing pandemic posed challenges for the data collection phase and thus postponed the completion of the research project. This pandemic created a lot of apprehension among the experts in completing the questionnaires and especially the face-to-face interviews. Although the results of this research are noteworthy, they reflect the socio-cultural environment of a developing country, and thus care should be taken when applying the results in another potentially different context. In other words, since the issue of corruption has social and cultural roots that are pertinent to the studied society, generalization of the results should be conducted with more prudence. Also, limited access to experts due to the nature of the issue of corruption is another limitation of this research that the researchers faced. The investigations that have been conducted indicated that numerous studies have been carried out in the field of identifying types of corruption, factors influencing it and ways to enhance administrative health in construction projects in countries around the world, but in Iran, despite its great importance, this issue has been neglected due to its high sensitivity. These sensitivities resulted in a limited number of experts cooperating with the current study. Future investigations can utilize the model constructed in this study in the case of the large-scale construction projects in other communities. Furthermore, the validity of the causes of corruption and the anti-corruption measures could be measured through other methods. Similarly, other decision-making approaches, such as the rough set theory, gray theory and fuzzy spaces, can be used to analyze corruption in construction projects and to prioritize anti-corruption measures. Moreover, the validity and robustness of the results should be assessed by conducting simulation in weighting methods. The findings of such studies could be compared with those of the present research. Another recommendation is to develop similar models to the ones presented in this research in other public and private sector organizations as well as family businesses and compare their outcomes.

Funding

The present study did not receive any specific grant from funding agencies in the public, commercial or non-profit sectors.

Conflict of interest

The authors declare that there were no conflicting interests.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

References

Appendix A

See Tables A1 and A2.

Appendix B

See Table B1.