A System Approach To Modelling Corruption

Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

A Systems Approach to Modeling Corruption: Bridging the

Disciplinary Divide Between the Social and Technical Sciences


David Zelinka1, Bernard Amadei2

1
Mortenson Center in Engineering for Developing Communities, University of Colorado Boulder, [email protected]
2
Mortenson Center in Engineering for Developing Communities, University of Colorado Boulder, [email protected]

Abstract

Corruption is considered one of, if not the, largest issue hindering progress towards sustainable development and ultimately
the global eradication of poverty. Corruption decreases the amount of aid that reaches its intending destination for those in
need. It also reduces the efficiency of international aid thereby undermining its validity. Corruption depresses the incomes and
overall GDP of many countries in Africa, especially those reliant on aid with rich deposits of natural resources. This paper
focuses on the effects of corruption on natural resource governance. Specifically, the paper addresses the so called “natural
resource curse” as a guiding framework, which is an especially problematic paradox in African countries. The premise is that
an abundance of natural resources leads to more corruption, thereby to less economic growth, less democratic government,
and lower levels of development, as measured by the Human Development Index, than countries with less natural resources
endowments. A novel systems modeling methodology combining civil systems engineering, system dynamics, and political
science is presented to analyze the interactions of corruption with the environment in the context of the natural resource curse
in African countries. Until now, few systems approaches have been proposed to analyze different acts of corruption (rentier
effect, repression effect, and modernization effect) in any fashion, and the ones that do rely entirely on perception-based data,
namely the Transparency International’s Corruption Perception Index. The systems-based methodology presented herein is
compatible with the social sciences and directly utilizes peer-reviewed literature as a tool to build models. In doing so, we
advocate more collaboration and transdisciplinary research among technical scientists and engineers, system dynamicists, and
social scientists (especially political scientists to address corruption issues and complex issues more broadly). New system-
level understanding of corruption can be extracted from methods such as the one proposed and presented herein. To
demonstrate this capability, a qualitative system dynamic stock-and-flow diagram is created where each variable and
interaction is sourced to at least one peer-reviewed, social science article that can greatly increase model validation and
confidence in accuracy; it depicts the world as it is, not as the modelers thinks it is. The proposed methodology can effectively
aggregate the relevant data and literature into a high-level model, yielding significant insight and value to decision makers and
those working with complex social and political issues.
Keywords: corruption, natural resources, governance, qualitative data analysis, system dynamics

1. Introduction

The paper aims to bring social and political scientists together with engineers and system dynamicists. Throughout this paper,
we argue that these two groups have much to offer to each other. A major barrier to their collaboration, however, is that
“[s]ocial scientists and system dynamicists approach research from very different perspectives…[as] there appears also to be
a fundamentally different understanding of the way in which [causality] can be validly inferred” (Morris, 2001). These
differences can be remediated through hybridization of methodologies used by both groups – the more qualitative and societal
focus of the social sciences with the more quantitative and technical nature of engineering and system dynamics. The
transdisciplinary approach presented herein “can contribute an attractive alternative in the social sciences, broadening the
scope of research” (Morris, 2001). This unique approach is useful for all social issues but is especially designed for complex
systemic issues such as corruption.

The proposed methodology uses a systems-based approach, specifically the qualitative portion of system dynamics modeling.
It assumes that corruption is a behavior caused by structural deficiencies in the governance system. In the next section, we
begin with a discussion of corruption, its impacts, definitions, and its various interdisciplinary perspectives. In the third section,
the topic of corruption is bounded to natural resources governance (henceforth the Natural Resource Curse), since corruption
in and of itself is quite complicated and expansive. Five fundamental causal mechanisms will be discussed as they form the
main structures of the Natural Resource Curse (NRC) and the backbone of an eventual systems model, to be the topic of future
work. The results section shows a systems representation of the dynamics of corruption in the form of a stock-and-flow diagram
and provides a discussion of how to understand and read it. Lastly, we conclude with a discussion of the method and policy
recommendations.
2. Corruption

Corruption is a problem, plain and simple. It exists in various degrees and forms in all societies; it can happen at any time, in
any time-period, and has been present throughout history. It is endemic in all human systems, is a societal problem, and is a
global phenomenon that “transcends national boundaries and frontiers (Iyanda, 2012)”:

• 3.3% of the $30 trillion global economy is lost to bribes each year (Goel and Nelson 2010)
• 5-25% ($26-130 billion) of World Bank funds were lost or misused due to corrupt practices (Berkman et al., 2008)
with more recent estimates as high as a $1 trillion (Liotta and Miskel, 2012)
• Corruption also exacerbates income inequalities and poverty; the Gini coefficient, a measure of income inequality
ranging from 0 (perfect equality) to 1 (perfect inequality), worsens by 11 percentage points and income for the
bottom quartile of people falls by 4.7 points with a one standard deviation increase in corruption (Gupta et al., 2002)
• Developed countries are not saved from the economic impacts of corruption, where 1% ($170 billion) of the
European Union’s GDP is lost due to corrupt actions (Oberoi, 2014)
• India lost $7.7 billion a year (in 2000 $USD) between 1948-2000 in illegal capital flows (Oberoi, 2014)
• Corruption depresses the quality of Indian businesses by roughly 40% (Oberoi, 2014)
• 25 - 40% of any given African country’s GDP is lost from corrupt practices (Oberoi, 2014)
• A one standard deviation reduction in corruption improved annual investment in a country by 4% of GDP and
increase GDP’s annual growth by 0.5% (Oberoi, 2014)
• Corruption depresses the Kenyan economy by 12% of GDP (Liotta and Miskel, 2012), nearly as large as their entire
industry and manufacturing sector at 14% (“Economy of Kenya,” 2017)
• The average Kenyan income was 424 times smaller than what parliamentary members awarded themselves for their
salaries, a direct indicator of corruption (Liotta and Miskel, 2012)
2.1. What Is Corruption?

Corruption is a broad term with many definitions spanning multiple disciplines, but a general definition that encompasses most
others is when someone does something without authorization for personal gain and usually at the expense of another.
Transparency International's (2016) definition is similar and the most widely used: “the abuse of entrusted power for private
gain.” Jancsics (2014) labels an agent as corrupt when they intentionally sacrifice their organization’s interest or betrays its
trust. Multiple definitions are compiled by Iyanda (2012) but when they are aggregated they sound familiar: (i) unlawful use
of official power to enrich oneself usually at the expense of others; (ii) when someone knows better but chooses the worse
option; and (iii) when someone acquires something to which they are not entitled. Taken together these definitions have
common components that determine when something is a corrupt act: (i) it is illegal, secret, or informal exchange of formally
allocated resources; (ii) at least one corrupt party is formally involved in the entity from which resources (broadly defined) are
extracted; and (iii) the corrupt act is against formal rules, laws, or policies (Jancsics, 2014). Beyond those mechanisms,
corruption requires three conditions, or an enabling environment, to arise and then persist: (i) discretionary power or authority
among relevant people; (ii) the incentives to be corrupt (i.e. incentives, utility, or gain to be made from being corrupt); and
(iii) weak institutions which make it easier for corruption to happen and continue (Aidt, 2003).

An act is considered corrupt so long as it adheres to the conditions and contains the components described above; Table 1 lists
and explains some common corrupt acts. Corruption is the prevalence of and damage caused by corrupt acts. In this work,
corruption is synonymous to rent-seeking, which is the desire and tendency for an entity to engage in corrupt acts for extracting
illegal rents. Rents are the benefits (financial, favors, or otherwise) acquired through the corrupt acts.

Table 1. Corrupt actions common in natural resource governance; adapted from Rose-Acker and Palifka (2016) and Iyanda
(2012)
Money, gifts, or favors given to a government or corporate official in exchange for breaking rules
Bribery to give preferential treatment, an advantage, or otherwise some kind of benefit to be received (see
extortion)
Demanding a bribe or favor from an official, where the official usually is at a position of lesser
Extortion
power (see bribery)

Exchange-of-
The exchange of one broken rule for another.
favors

Clientelism/
General favoritism to a specific group (see nepotism and cronyism)
patronage

Putting someone into a position of power based on the strength of social connections rather than on
Nepotism
merit or the legal process (see clientelism/patronage and cronyism)

Giving preferential treatment to one’s group (racial, ethnic, religious, political, or social) over other
Cronyism
groups (see clientelism/patronage and nepotism)

Embezzlement Theft from the employer (firm, government, or NGO) by the employee.

An autocratic state that is managed to maximize the personal wealth of the top leaders and is closely
Kleptocracy related to economic inequality; there are virtually no differences between the economic elites from
the political elites, with Russian oligarchs as a prime example.

Influence peddling Using one’s power of decision in government to extract bribes or favors from interest parties.

Avoiding tax can be legal or illegal. It can be a subset of bribery and is a mechanism reduces the
efficiency of revenue generating tools. Utilizing legal loopholes is a form of corruption in that the
Tax evasion
citizen is not paying all the taxes that they are supposed to due to structural inefficiencies or
problems in the revenue accumulation systems of a government.

Conflicts-of- Having a personal stake in the effect of the polices one decides; although not directly a corrupt act,
interest a conflict-of-interest can heavily influence corrupt practices.

2.2. Disciplinary Perspectives of Corruption

The definitions of corruption differ by discipline of which most come from the social sciences. For economists, people are
rational actors that seeks to maximize their profits (utility) and corruption is a mechanism to do so. Corrupt practices are also
manifestations of wage differentials between public servants and their more profitable private sector counterparts; governments
cannot, or will not, compete with private companies in the market when it comes to salaries of their employees (Ades and Di
Tella, 1999). When the costs of corruption are smaller than its benefits, corruption is more widespread and systemic (Andvig
et al., 2000; Heath et al., 2016) as is especially the case in the developing world. In general, economists are interested in the
impact of corruption on economic development (Torsello and Venard, 2016), thus they tend to view corruption mostly as a
problem to be solved as opposed to a phenomenon to be understood.

There are multiple theories about corruption in political science, which represents the most common discipline involved in
corruption research. One states that corruption is a way to overcome burdensome bureaucratic red tape to increase efficiency
(Andvig et al., 2000), or colloquially, corruption ‘greases the wheels.’ Similarly, poor institutional structure and quality can
be a catalyst for corruption, which causes inefficient bureaucracy. Weak institutions make corrupt acts easier and less risky.
Another states that corruption is a natural by-product of non-democratic (anocratic and autocratic) governance systems (Heath
et al., 2016). The institutional perspective, described by Gerring and Thacker (2004), further states that Unitarianism and
Parliamentarism governments exhibit lower corruption, since a single person does not hold disproportionately large amounts
of power as is the case in dictatorship and other autocratic governments.
A popular view among economic and political science scholars is that of the principle-agent, where the individual is
incentivized to act corrupt and intentionally does so; many theories stem from that (Jancsics, 2014). Economic and political
theories of corruption are highly-interrelated with one another, and are rarely, if at all, purely economic or political in nature.
Economic and political definitions also tend to focus on large government or corporate entities, which center on structural
(institutional and bureaucratic) explanations for corruption. This structural approach (collective-action problem) differs from
the principle-agent approach in that it stresses a more institutional, systemic approach and views corruption as a “social trap”
(Mungiu-Pippidi, 2013; Persson et al., 2013; Rothstein, 2011). Corruption is a manifestation of institutional quality, or lack
thereof, and differs from the principle-agent to a more collective-action problem (Rothstein, 2011). At this institutional-level
“the only thing worse than a society with a rigid, over-centralized dishonest bureaucracy, is one with a rigid, over-centralized,
honest bureaucracy” (Huntington, 2006). In international relations and in the context of the global political economy, closed
societies and economies suffer from high corruption because they are less reliant on the global economy removing leverage
that the international community might have (Gerring and Thacker, 2005). In sum, the economic and political science literature
takes two theoretical approaches: that of the principle-agent that operates at the micro-scale and the collective-action that
operates at the macro-scale (Jancsics, 2014; Mungiu-Pippidi, 2013; Rothstein, 2013).

Other fields, such as sociology, use more social and cultural perspectives to explain corruption, which explain some endemic
and societal-level corruption and operates primarily at the macro-scale. A sociological approach to understanding corruption
“derive[s] from exchange theory, which emphasizes the nature of the social relationships between [various] actors (Heath et
al., 2016).” More demographic theories involve gender, where higher levels of female empowerment and involvement in the
government and the general economy lead to lower corruption (Branisa and Ziegler, 2010). Coupling that idea with others
from political science, Branisa and Ziegler (2010) describe how social institutions that reduce gender inequality improve good
governance (reducing corruption).

Anthropology, which is underrepresented in the corruption literature, takes a different approach from other fields of social
sciences. Anthropologists try to avoid discrete definitions in their analyses, and instead rely on contextual morality as their
guiding principle in that “[d]ifferent societal contexts can have diverse and multiple moralities (Torsello and Venard, 2016).”
Anthropologists avoid definitions because in doing so, they argue, biases from categorizing and nuance of local culture is lost.
With multiple moralities, corruption is not viewed as something good or bad, it simply is; the specific local context determines
if it is a positive or negative factor. For example, the researcher might view petty corruption as a corrupt act in all cases, but
the local culture might view some petty corruption as a “practice that smoothens social exchange,” according to Torsello and
Venard (2016) or something that greases the wheels. Furthermore, what might “look like corruption to outside observers might
serve crucial social and symbolic functions from the inside or in a local context” (Jancsics, 2014; Torsello and Venard, 2016).

According to Torsello and Venard (2016), the anthropologist researcher focuses on the thought-processes, worldview, and
perceptions as non-static phenomenon in relation to social transformation, globalization, and development. Anthropologists
focus on the impact of culture generally more than the other social sciences but shares concepts with social norms that are
often discussed in the political science literature. To anthropologists, social norms exhibit a “dynamic nature of the morality
of corruption (Torsello and Venard, 2016),” which contrasts with the more static representation in political science.

Psychology, like anthropology, takes an individual-level perspective and might view political corruption as a person
succumbing to their intrinsic motivations or personal desires over the loyalty to their government job (Ryan and Deci, 2000).

These five disciplines – economics, political science, sociology, anthropology, and psychology – are part of the social sciences,
but they all attempt to understand corruption in different ways. In reality, it takes an amalgam of all disciplines (including ones
not discussed here) to being to form a more complete understanding of corruption, its influences, and its causes. Focusing on
just one area, ignores aspects of others including different scales – micro- and macro-scale. Anthropology as a singular field,
comes close to an approach (based on individual circumstances) to define and understand corruption. Taking parts from all
areas, the next section attempts to use a systems perspective on corruption.
2.3. Systems Definition

In many contexts, the definitions of corruption depend on societal norms, which are highly influenced by and structurally
ingrained within local, national, and ethnic culture. For example, some societies have lived with corruption for so long that it
has become normal for them, so what might seem corrupt for some societies will be business-as-usual for theirs. This is
especially pervasive in many African and Eastern countries that were former colonies of imperial powers (Bhattacharyya and
Hodler 2010; Goel and Nelson 2010). These various perspectives are of interest to anthropologists and sociologists but also
involve political and economic theories as well.

Many disciplines have commonalities in their approaches to understand corruption and model its effects. General issues of
political, governance, and societal structure and norms span most social sciences. In much the same way as poor government
structure can worsen corruption (political science), cultures with hierarchical and highly centralized religions (sociological
and anthropological) and economies (economic), and a general acceptance of power differentials are more likely to be corrupt
(Husted, 1999; Jancsics, 2014; Seleim and Bontis, 2009). Furthermore, strong personal relationships augment expectations
that favors will be reciprocated (Heath et al., 2016), a process that can be understood using social network analysis (Borgatti
et al., 2013).

Other societies might be so distrustful of their government that corrupt practices could be the only viable response to a corrupt
government. Other nations have societies that inherently function based on societal power differentials, like the caste system,
which causes endemic lack of trust among different socio-economic groups ingraining corruption into that society (Bjørnskov,
2003; Rothstein, 2013; Uslaner, 2013). In these groups, particularistic norms regulate behavior and could dominate
universalistic norms under which the society as a whole operates (Jancsics, 2014). Furthermore, social networks in general
strengthen interpersonal bonds and trust, thus reducing transaction costs associated with corrupt acts (leading to nepotism in
government corruption).

A common theme arises: corruption is a behavior that emanates from a system’s structure – the way a society is organized
(religion, culture, or otherwise) and the institutions of the government – as well as the duration those structures and institutions
have been around and have persisted. In general, a poorly structured (social, political, governance, economic, etc.) system will
generate poor behavior. Institutions can act as “opportunity structures to conduct corrupt practices” (Jancsics, 2014).
Corruption is not a universal concept and varies by culture or community. Socially, an act that is viewed as elite capture to
foreigners may not perceived as such within the community who view it as a leader earning their share. Nepotism is an example
of a corrupt act (Table 1), but the community might simply view it as doing favors (Seleim and Bontis, 2009), helping a friend,
or even a form of economic bartering. There are obviously a lot of overlaps among these themes because corruption is not
discrete, but rests on a multidimensional set of spectrums.

The focus of this work is on the structural causes to corruption. It is based on the concept of system sciences that the structure
of a system determines its behavior (Amadei, 2015; Benson and Marlin, 2017; Ford, 2010; Sterman, 2000; Ullah et al., 2012).
Furthermore, Rose-Acker and Palifka (2016) state that a “high level of corruption indicates that something is wrong with the
state’s underlying institutions…[signaling] a need for structural reform – not just more vigorous law enforcement.” A systems
approach to understanding corruption requires a systems-oriented definition of corruption: corruption is anything an actor
(human) does to enrich themselves usually at the expense of other actors by undermining the system effectiveness. A systemic
solution to corruption would be to restructure the governance system through structural reform to alter the negative behavior
(reduce corruption). In this way corruption is never directly addressed, rather indirect actions propagate throughout the system
eliciting lasting behavioral change and outcomes of that system (i.e. political officials can be dis-incentivized to act in a corrupt
manner by reducing the enabling environment to be corrupt).

This definition of corruption is broad enough to contain virtually all acts considered corrupt, as well as most definitions of
corruption regardless if they are economic, political, social, or environmental in nature. It is dynamic in that, if what a society
views as corrupt changes so does the definition, so long as it adheres to the general definition; an interesting approach used by
anthropologist (Torsello and Venard, 2016) discussed earlier.

The definition of corruption changes with the perspective of the observer or, in our case, the researcher or modeler who is
trying to understand a specific situation. But it is important to keep in mind that, as summed up by Oberoi (2014), corruption
is “what every society decides it is; it simply has to be internally consistent with formal institutions of a given society.”

2.4. Scales and dimensions of corruption

Corruption can operate at any level, from petty corruption between common citizens to grand corruption that occurs among
the political and economic elite. Grand corruption involves a small number of powerful actors dealing with massive quantities
of money and can be political or corporate in nature (Rose-Ackerman and Palifka, 2016). It is often between the two, especially
in the case of natural resources and large infrastructure projects. This form of corruption is out of reach to most citizens, who
can only really be guilty of petty corruption (Rose-Ackerman and Palifka, 2016). This street-level corruption is involved with
everyday activities of citizens (Iyanda, 2012). As a result, grand corruption is more pervasive in national governments and
other locations in the economy that deal with commanding and authoritative figures operating in high-power centers involving
large sums of money (Iyanda, 2012; Rose-Ackerman and Palifka, 2016) such as in the military and national defense (Gupta et
al., 2001), natural resources (Bulte and Damania, 2008; Busse and Gröning, 2012; Corrigan, 2014; Dimant and Tosato, 2017;
Kolstad and Søreide, 2009), and the infrastructure sectors (Tanzi, 1998).

Business corruption (petty corruption’s private sector counterpart) occurs with small- to medium-sized businesses who interact
with average citizens, while corporate corruption (grand corruption’s corporate counterpart) involves large multinational
companies interacting with governments. In contrast, businesses and corporations might also act corrupt to avoid regulations.
For example, Mitchell (1994) describes a situation when oil-tankers would dump their waste oil out in the ocean before entering
port so authorities would not detect and fine them. This compliance problem is a form of corruption when business act in ways
that prioritize company gain over societal good; this views corruption as a collective-action problem as opposed to the
principle-agent problem (Bauhr, 2017; Mungiu-Pippidi, 2013; Persson et al., 2013; Rothstein, 2011). Eventually, the structure
of the environmental regulations governing how tankers disposed of their waste oil was successfully reformed to disincentivize
this negative behavior (Mitchell, 1994).

3. The Natural Resource Curse

This section is a truncated version of a more extensive literature review, which was used to directly construct the stock-and-
flow diagrams shown later in Figures 2-6. The methodology to convert this literature review into a systems model is explained
in the following section.

The natural resource curse is a phenomenon in which countries with large quantities of natural resources and mineral wealth
counterintuitively suffer from lower economic and social development (Bhattacharyya and Hodler, 2010; Busse and Gröning,
2012; Corrigan, 2014; Frankel, 2010; Kolstad and Søreide, 2009; Teksoz and Kalcheva, 2016). It occurs throughout the Global
South but is especially pervasive in many African and some Middle Eastern countries: all of the 12 oil exporting countries in
Africa are in the bottom half of the United Nations Human Development Index (Diamond and Mosbacher, 2013).

The guiding principle in the natural resource governance literature is to understand “why some countries succumb to the
resource curse while others seem to benefit economically from their natural resources (Corrigan, 2014).” The most common
reason proposed in the literature involves poor institutional quality and governance. That is only part of the picture, but there
are a few specific, well-documented causal pathways that connect natural resources with governance, namely the rentier,
repression, and modernization effects (Bhattacharyya and Hodler, 2010; Busse and Gröning, 2012; Corrigan, 2014; Kolstad
and Wiig, 2009).
Each pathway incorporates different corrupt acts (see Table 1), in different combinations, and at different points along the
chain, while at the same time intersecting and influencing one another. The rentier effect (Figure 1, Figure 2, and Figure 3) is
directly influenced by the amount of revenues (rents) the government is able to extract from natural resources (with oil being
the prime example) and occurs when those revenues are substantial in relationship to other sources of income. Substantial in
this context refers to when resources revenues are so large that the government needs not rely on other sources of income. The
effect comes from what the national government does with those revenues (Teksoz and Kalcheva, 2016). The rentier effect
can be further delineated into the taxation, group formation, and spending effects. The resource revenues could be used to
reduce the need to tax the population (taxation effect in Figure 3) making them happier with the government, which in turns
reduces their desire to hold the government accountable; unrest and general unhappiness with the government accumulates at
a slower rate (Busse and Gröning, 2012; Corrigan, 2014) increasing the tolerance of corruption. This money can also be used
for development projects to appease or distract the population. Along the same lines, Busse and Gröning (2012) describe a
group formation effect (Figure 2) which occurs when the government uses the rents to prevent the formation and operation of
social and special interest groups who would otherwise influence the government to reduce corruption. The same effect can
happen when those revenues are spent on cronyism (broadly favoritism), which Busse and Gröning (2012) refer to as the
spending effect (Figure 1). These tend to resemble kleptocracy in extreme cases, such as what is occurring with Russia’s
current political trajectory. Either way, through the reduction in taxes and the increase in general favoritism, accountability is
reduced because the population sees more money and will turn the other way when the government or political officials
participate in corrupt acts. This favoritism leads to more natural resource economic policies accelerating their extraction. In
the short- to medium-term, revenues increase feeding back into the cycle. Over time, the amount of resources diminishes with
little benefit to show for the steep payment (Corrigan, 2014).

The repression effect (Figure 4) is when the government uses the revenues to suppress opposition, as opposed to gaining favor
among the populace. It is a more extreme version of the group formation effect and can be violent or militaristic. The rents
again come from the significant natural resource wealth endowed within a country’s borders, but also can come from individual
government officials, who receive illegal income from corrupt acts like bribery (Busse and Gröning, 2012; Kolstad and Wiig,
2009). The government might spend more on the military further entwining the political and military spheres in a country.
When that happens, it is easier for the military to oppress or terrorize the people. In extreme cases, the repression effect can
take the form of violent suppression of protests or attempted revolution. A salient example is the Arab Spring. In short, the
repression effect “could impede aspirations among the population for more democracy or better institutions and government
services (Busse and Gröning, 2012).”

Busse and Gröning (2012), Corrigan (2014), and Kolstad and Søreide (2009) all describe a modernization effect (Figure 5)
when the government actively delays modernizing and industrializing their country because such a society will have alternative
sources of economic and political power with which to oppose the government. It some cases, it can be viewed as a passive
and indirect version of the repression effect. Delaying modernization reduces the enabling environment for protests to form
and revolts to succeed. An economy based on natural resources and agricultural has little need for a large skilled labor force,
and education positively correlates with demand for political reforms. The government actively comes in the way of social
and cultural change in fear of losing power. Many countries in the Middle East (Frankel, 2010), namely Iran and Saudi Arabia,
have been especially susceptible to this form of corruption.

4. Methods and Results

“One might say that the ultimate goal of any science is to establish causal relations (Denis, 2016),” where at its core causality
is the relationship shared by two or more variables. Science attempts to simplify complex systems (Sterman, 2000) to identify
direct causality (that is causal relationships where one variable directly influences another without intermediaries) of specific
structures within those systems. Science (social, natural, physical, etc.) utilizes many tools to analyze causality, and it is the
knowledge produced through these tools that will be the focus of this analysis. Explicitly, this work explores how to put
together all those simple disjointed pieces produced from the (social) scientific research in a way that is valid and coherent.
The proposed methodology is a hybrid of qualitative data analysis (QDA) and systems dynamics. The technique, called
qualitative structural data analysis (QDSA), is applied to the relevant social science corruption literature to form a deep
understanding of the Natural Resource Curse in the form of a stock-and-flow diagram, which has the potential to yield deeper
insights than current methodologies in the social sciences. This method can be used for many social issues whose behavior is
determined or governed by a complex system, especially those “whose main actors are directly and strongly influenced by
complex human socio-economic-political interactions but are themselves not humans,” which we refer to as anthropocentric
systems (Zelinka and Amadei, 2018a).

Through a detailed literature review, it has become apparent to the authors and many corruption scholars over recent years that
corruption negatively affects multiple sectors through many channels and causal mechanisms; simply put, corruption is a
complex issue. Most articles go through individual mechanisms or attempt to relate corruption by only a few dependent
variables. Few articles look at the interactions among the multiple mechanisms at work, and fewer still analyze the various
upstream and downstream causes and effects of systemic corruption. In short, the vast majority of the existing work looks at
corruption through disciplinary silos, a narrow focus, and generally without much consideration to account for complexity. A
systems-based approach is necessary to understand and effectively model these kinds of complex social issues.

4.1. Interdisciplinary Collaboration

Corruption is mostly the focus of social sciences with most of the literature emanating specifically from political science,
economics, and to a lesser extent all other social sciences (Jancsics, 2014). Although many of the issues regarding corruption
overlap many fields within the social sciences, there is relatively little collaborative work among them (Jancsics, 2014; Torsello
and Venard, 2016) and much less so with the technical sciences and engineering. For example, only a few articles use system
dynamics to analyze corruption (see Méndez-Giraldo et al., 2017; Soto Torres et al., 2007; Ullah et al., 2012; Ullah and
Arthanari, 2011). Jancsics (2014), a sociologist, notices this deficiency in the corruption literature and the social sciences in
general; he describes the “lack of interdisciplinary communication about corruption, such that models developed by different
academic disciplines are often isolated from each other.” He discusses three types of corruption studies and approaches: a
micro-level perspective, a micro-level perspective, and a relational approach focusing on social interactions and networks that
bridges the other two. Each social science discipline generally operates at different scales (psychology versus sociology,
international relationships versus development studies, micro- versus macro-economics, etc.), possibly explaining some of the
lack of collaboration – it is more difficult to work together when people inherently do not perform research at the same scale.
The last approach can be used to bridge the disciplinary divides (Jancsics, 2014) by providing a method compatible with
system dynamics.

Collaboration between the disciplines is not a new idea but has received little attention (Morris, 2001). System sciences, we
argue, can bring together various fragmented fields through a transdisciplinary and collaborative approach. They can blend
crucial elements of the social sciences with that of the technical science and engineering through an hybrid qualitative and
quantitative methodology (see Zelinka and Amadei, 2018b). Social science methodologies, generally, tend to emphasize
individual relationships and avoid the big picture, although they are proficient at detailing individual factors and nuance to
identify fundamental and underlying mechanisms through which corruption affects the natural resource sector, or vice versa;
but it can be subject to tunnel vision. Engineering methods can fill in data gaps and present a set of quantitative tools and
modeling techniques with which social scientists might not often use in their work. Engineering, generally, is good at
understanding the big picture, but misses the nuance when understanding social issues.

The weaknesses in the social sciences, engineering, and system dynamics can be overcome, at least partially, through
acceptance and collaboration. Indeed, a collaborative system is greater than the sum of its parts; it shows synergistic properties.
When groups work together, individual parts can be thoroughly analyzed and understood, ideally, but can also be combined
to provide a relatively accurate and valid representation of the issue at hand. Transdisciplinary collaboration can provide high-
resolution details at a large-scale to see how the many mechanisms interact with one another, explore reverse causality, and
detect indirect influences and effects. Perhaps most importantly, a system approach can make complex and ill-defined social
issues understandable to a non-social-scientist, namely policy and decision makers. System dynamics is a computer-aided
modeling approach used in the technical sciences and engineering that can function as a tool to model and understand ill-
defined social issues better.

4.2. System dynamics modeling

This paper will forgo a detailed description of system dynamics modeling as it is covered in the authors’ other paper in this
conference (see Zelinka and Amadei, 2018b). Systems dynamics is a relatively recent branch of systems science that originated
with the work of Dr. Jay Forrester at the Massachusetts Institute of Technology in the 1950s and 1960s. It represents a
milestone in the overall evolution in the application of systems thinking and the development of tools to address complex
issues in a wide range of disciplines such as engineering, business, economics, health, planning, management, etc. A simple
way to see system dynamics is that it studies “how systems change over time” (Ford, 2010). The System Dynamics Society
(2016) defines system dynamics as a computer-aided approach to policy analysis and design. It applies to dynamic problems
arising in complex social, managerial, economic, or ecological systems — literally any dynamic systems characterized by
interdependence, mutual interaction, feedback, and causality.

Since its inception, system dynamics has accrued popularity in various STEM and some social science fields, including topics
such as corruption (Méndez-Giraldo et al., 2017; Soto Torres et al., 2007; Ullah, 2012), urbanization and demographic process
(Zelinka et al., 2018), the Sustainable Development Goals (Moffatt and Hanley, 2001; Nilsson et al., 2017; United Nations
Economic and Social Council et al., 2015; Zelinka and Amadei, 2018a, 2018c, 2017; Zhang et al., 2016), health and infectious
diseases (Ghaffarzadegan et al., 2010; Pedamallu et al., 2012), education (Pedamallu et al., 2010); sustainable community
development (Amadei, 2015), environmental and ecological process (Ford, 2010), and business and management science
(Sterman, 2000). In general, systems dynamics has unique characteristics that warrant its use in modeling complexity in social
and governance systems (Zelinka and Amadei, 2018b).

4.3. Qualitative Structural Data Analysis

This section is a brief overview of qualitative structural data analysis (QSDA), which codes and then translates text from
peer-reviewed articles and reports from reliable sources into causal-loop diagrams (CLDs), stock-and-flow diagrams, and
eventually full system dynamics models. Corruption is an endemic social issue that has been around since the dawn of human
civilization, but there exist few reliable methods to understand its systemic causes, effects, and impacts. Corruption will
provide a proof-of-concept case study of QSDA to understand and then model complex social issues commonly studied within
the social sciences. The methodology proposed herein is in its early stages, and as such it will undergo many future iterations.
Even in its current state, it can be used to provide an informative and reliable representation of the Natural Resource Curse
mentioned earlier in this paper (Section 3). A stock-and-flow diagram depicting the causes and consequences of corruption in
the natural resource sector focuses on the five significant mechanisms detailed in the literature and discussed earlier. Each
variable, type of variable (direct versus indirect and moderating versus mediating), relationships among variables, and
directions of those relationships (positive or negative) can be captured in QSDA. Eventually, a system dynamics (SD) model
can be constructed that encompasses the entire literature synthesis and all the articles analyzed within. The main difference
between a stock-and-flow diagram and a full system dynamics model is the inclusion in the latter of initial conditions and
equations, which will be left to future work.

QSDA is heavily influenced by the qualitative data analysis (QDA) which “is the range of processes and procedures whereby
we move from the qualitative data that have been collected into some form of explanation, understanding or interpretation of
the people and situations we are investigating” (Gibbs and Taylor, 2010). NVivo 11 by QSR International was the QDA
software utilized for our QSDA, and although, not originally designed for addressing such needs, NVivo proved to be sufficient
for the task at hand. Nominally, the coding mechanisms between the two analytical methods are the same, but instead of being
used to interpret and understand qualitative information like in QDA (Bernard and Ryan, 2010), QDSA is used to structurally
identify and (reliably) source variables, relationships, and more complicated causal mechanisms. The variables become the
convertors, stocks, and flows that constitute CLDs, stock-and-flow diagrams, and system dynamics models; the relationships
become the interconnections; and the causal mechanisms – the rentier, repression, and modernization effect –become large
structures around which the rest of the model is built.

In any system dynamics model, there are three types of variables: flows, stocks, and convertors. These basic components
appear in Figures 2-6. Stocks, represented by rectangular boxes, are net accumulations of something at one point of time and
represent state variables that define the current state of the systems. Flows appear as pipelines with a faucet controlling the
flow and refer to processes that cause change over time (dynamic) of information or materials; they are control variables that
create change in the systems and affect stocks. Convertors, shown as a name or a circle-and-name, transform (convert)
information from one stock-and-flow path to another. For example, greenhouse gases (GHG) are emitted per year (flow) and
accumulate in the atmosphere (stock). Each person uses electricity that in turn emit GHGs; electricity and GHGs are two
stocks-and-flow paths connected by at least one convertor. The variables are called nodes in NVivo. While manually coding
the literature, each variable type was noted and when the stock-and-flow diagram was constructed, the variables were assigned
their appropriate variable type.

NVivo uses the term relationships to signify the interconnections between two variables. Interconnections can either negatively
or positively affect another variable. Once the five larger causal mechanisms, discussed in the NRC section above, were
classified, the necessary variables were identified and connected as needed with relationships. Cases in NVivo represent units
of observation, which in QSDA are causal mechanisms and other important structures. They are composed of individual nodes,
relationships, or groups of them. Cases can group multiple sources that relate to the same structure or mechanism and help to
organize them for when it comes time to construct the stock-and-flow diagram and then the system dynamics model.

Once all these variables, interconnections, and causal mechanisms were identified and properly coded from the literature in
NVivo, the information was, as objectively as possible, incorporated into STELLA (Systems Thinking Experiential Learning
Laboratory with Animation) Professional (version 1.6), a system dynamics modeling software by isee systems, Inc. The
components of the QSDA were added into a single stock-and-flow diagram that encompassed all five major causal mechanisms
identified in the literature review. For ease of viewing, each mechanism was given its own figure and highlighted in blue as
shown in Figures 2-6.

5. Discussion and Conclusion

Because the literature review above is directly based on the same QSDA as the figures below, a detailed explanation and
walkthrough of the stock-and-flow diagram is not required. Each variable, interconnection, and structure depicted in the stock-
and-flow diagram can be sourced to specific articles and locations within those articles. All articles used in our analysis were
themselves based on many peer-review articles. Hence, the QSDA effectively aggregates very many articles, thus acting as a
form of a systems-based meta-analysis. Since all the articles used in this analysis are peer-reviewed, the assumption is that any
models based on them exhibit a high level of validity. The methodology presented herein is designed to reliably connect the
model with the articles on which it is based.

It is the goal of this work to show that social scientists, engineers, and system dynamicists have much to offer each other and
that collaboration among them is crucial to understand and analyze complex issues. QSDA provides proof-of-concept of a
transdisciplinary method through which these seeming conflicting groups might regularly form collaborative research groups.
The authors see a day soon when most articles are published under the auspices of multiple fields, and systems thinking
prevails over linear thinking. This work is meant to advocate for that paradigm shift and provide a means to do so.
Figure 1. Rentier effect (spending effect)
Figure 2. Rentier effect (group formation effect)
Figure 3. Rentier effect (taxation effect)
Figure 4. Repression effect
Figure 5. Modernization effect
References

Ades, A., Di Tella, R., 1999. Rents, Competition, and Corruption. Am. Econ. Rev. 89, 982–993.
https://doi.org/10.1257/aer.89.4.982

Aidt, T.S., 2003. Economic analysis of corruption: A survey. Econ. J. 113, 632–652. https://doi.org/10.1046/j.0013-
0133.2003.00171.x

Amadei, B., 2015. A Systems Approach to Modeling Community Development Projects. Momentum Press Engineering.

Andvig, J., Fjeldstad, O., Søreide, T., 2000. Research on Corruption: A policy oriented survey, Chr. Michelsen Institute.
Bergen, Norway.

Bauhr, M., 2017. Need or Greed? Conditions for Collective Action against Corruption. Governance 30.
https://doi.org/10.1111/gove.12232

Benson, T., Marlin, S., 2017. Systems Thinking Overview [WWW Document]. Waters Found. URL
http://watersfoundation.org/systems-thinking/overview/ (accessed 2.2.17).

Berkman, S., Boswell, N.Z., Brüner, F.H., Gough, M., McCormick, J.T., Pedersen, P.E., Ugaz, J., Zimmermann, S., 2008. The
fight against corruption: international organizations at a cross-roads. J. Financ. Crime 15, 124–154.
https://doi.org/10.1108/13590790810866863

Bernard, H.R., Ryan, G.W., 2010. Analyzing qualitative data: systematic approaches. SAGE.

Bhattacharyya, S., Hodler, R., 2010. Natural resources, democracy and corruption. Eur. Econ. Rev. 54, 608–621.
https://doi.org/10.1016/j.euroecorev.2009.10.004

Bjørnskov, C., 2003. Corruption and social capital. Work. Pap. 1–27. https://doi.org/10.4467/20843968ZP.13.021.1421

Borgatti, S.P., Everett, M.G., Johnson, J.C., 2013. Analyzing Social Networks. SAGE Publications.

Branisa, B., Ziegler, M., 2010. Reexamining the link between gender and corruption: The role of social institutions (No. 24),
Poverty, Equity and Growth - Discussion Papers.

Bulte, E., Damania, R., 2008. Resources for Sale: Corruption, Democracy and the Natural Resource Curse. B. E. J. Econom.
Anal. Policy 8, Article 5.

Busse, M., Gröning, S., 2012. The resource curse revisited: Governance and natural resources. Public Choice 154, 1–20.
https://doi.org/10.1007/s11127-011-9804-0

Corrigan, C.C., 2014. Breaking the resource curse: Transparency in the natural resource sector and the extractive industries
transparency initiative. Resour. Policy 41, 17–30. https://doi.org/10.1016/j.resourpol.2013.10.003

Denis, D.J., 2016. Applied Univariate, Bivariate, and Multivariate Statistics. John Wiley & Sons, Ltd.

Diamond, L., Mosbacher, J., 2013. Africa’s Coming Resource Curse—and How to Avoid It. Foreign Aff.

Dimant, E., Tosato, G., 2017. Causes and Effects of Corruption: What has past decade’s empirical research taught us? A
survey. J. Econ. Surv. 0, 1–22. https://doi.org/10.1111/joes.12198

Economy of Kenya [WWW Document], 2017. . Wikipedia.


https://doi.org/https://en.wikipedia.org/wiki/Economy_of_Kenya#Industries

Ford, A., 2010. Modeling the Environment, 2nd ed. Island Press.

Frankel, J.A., 2010. The Natural Resource Curse: A Survey. SSRN Electron. J., NBER Working Paper 17–57.
https://doi.org/10.2139/ssrn.1565588
Gerring, J., Thacker, S.C., 2005. Do Neoliberal Policies Deter Political Corruption? Int. Organ. 59, 233–54.
https://doi.org/10.1017/S0020818305050083

Gerring, J., Thacker, S.C., 2004. Political Institutions and Corruption: The Role of Unitarism and Parliamentarism. Br. J. Polit.
Sci. 34, 295–330. https://doi.org/10.1017/S0007123404000067

Ghaffarzadegan, N., Lyneis, J., Richardson, G.P., 2010. How small system dynamics models can help the public policy
process. Syst. Dyn. Rev. 8, n/a-n/a. https://doi.org/10.1002/sdr.442

Gibbs, G.R., Taylor, C., 2010. What is Qualitative Data Analysis? (QDA) [WWW Document]. Online QDA. URL
http://onlineqda.hud.ac.uk/Intro_QDA/what_is_qda.php (accessed 3.6.18).

Goel, R.K., Nelson, M.A., 2010. Causes of corruption: History, geography and government. J. Policy Model. 32, 433–447.
https://doi.org/10.1016/j.jpolmod.2010.05.004

Gupta, S., Davoodi, H., Alonso-Terme, R., 2002. Does corruption affect income inequality and poverty? Econ. Gov. 3, 23–
45. https://doi.org/10.1007/s101010100039

Gupta, S., De Mello, L., Sharan, R., 2001. Corruption and military spending. Eur. J. Polit. Econ. 17, 749–777.
https://doi.org/10.1016/S0176-2680(01)00054-4

Heath, A.F., Richards, L., de Graaf, N.D., 2016. Explaining Corruption in the Developed World: The Potential of Sociological
Approaches. Annu. Rev. Sociol. 42, 51–79. https://doi.org/10.1146/annurev-soc-081715-074213

Huntington, S.P., 2006. Political Order in Changing Societies, in: Political Corruption: A Handbook. Yale University Press,
New Haven, Connecticut, pp. 59–71.

Husted, B.W., 1999. Wealth, Culture, and Corruption. J. Int. Bus. Stud. 30, 339–359.
https://doi.org/10.1057/palgrave.jibs.8490073

Iyanda, D.O., 2012. Corruption: Definitions, Theories, and Concepts. Arab. J. Bus. Manag. Rev. 2, 37–46.

Jancsics, D., 2014. Interdisciplinary perspectives on corruption. Sociol. Compass 8, 358–372.


https://doi.org/10.1111/soc4.12146

Kolstad, I., Søreide, T., 2009. Corruption in natural resource management: Implications for policy makers. Resour. Policy 34,
214–226. https://doi.org/10.1016/j.resourpol.2009.05.001

Kolstad, I., Wiig, A., 2009. Is Transparency the Key to Reducing Corruption in Resource-Rich Countries? World Dev. 37,
521–532. https://doi.org/10.1016/j.worlddev.2008.07.002

Liotta, P.H., Miskel, J.F., 2012. The Real Population Bomb: Megacities, Global Security, and the Map of the Future, 1st ed.
Potomac Books, Washington, DC.

Méndez-Giraldo, G.A., López-Santana, E.R., Franco, C., 2017. A Hybrid System Dynamics and Fuzzy Logic Approach to
Social Problem of Corruption in Colombia, in: 4th Workshop on Engineering Applications. Springer International
Publishing, Cartagena, Colombia, pp. 250–262. https://doi.org/10.1007/978-3-319-66963-2_23

Mitchell, R.B., 1994. Regime Design Matters: Intentional Oil Pollution and Treaty Compliance. Int. Organ. 48, 425–458.

Moffatt, I., Hanley, N., 2001. Modelling sustainable development: Systems dynamic and input-ouput approaches. Environ.
Model. Softw. 16, 545–557. https://doi.org/10.1016/S1364-8152(01)00024-X

Morris, D.R., 2001. Causal Inference and System Dynamics in Social Science Research: A Commentary with Example, in:
Proceedings of the 19th International Conference of the System Dynamics Society.
Mungiu-Pippidi, A., 2013. Controlling Corruption Through Collective Action. J. Democr. 24, 101–115.
https://doi.org/10.1353/jod.2013.0020

Nilsson, M., Griggs, D., Visbeck, M., Ringler, C., Mcollum, D., Mollier, L., Seyler, F., Chotte, J.-L., Howden-Chapman, P.,
Siri, J., Chisholm, E., Chapman, R., Doll, C.N.H., Capon, A., Echeverri, L.G., Riahi, K., Parkinson, S., Schmidt, S.,
Neumann, B., Waweru, Y., Durussel, C., Unger, S., 2017. A Guide to SDG Interactions: From Science to
Implementation. International Council for Science, Paris, France. https://doi.org/10.24948/2017.01 ICSU

Oberoi, R., 2014. Mapping the Matrix of Corruption: Tracking the Empirical Evidences and Tailoring Responses. J. Asian
Afr. Stud. 49, 187–214. https://doi.org/10.1177/0021909613479319

Pedamallu, C., Ozdamar, L., Ganesh, L., Weber, G.-W., Kropat, E., 2010. A System Dynamics Model for Improving Primary
Education Enrollment in a Developing Country. Organizacija 43, 90–101. https://doi.org/10.2478/v10051-010-0010-5

Pedamallu, C.S., Ozdamar, L., Kropat, E., Weber, G.-W.W., 2012. A system dynamics model for intentional transmission of
HIV/AIDS using cross impact analysis. Cent. Eur. J. Oper. Res. 20, 319–336. https://doi.org/10.1007/s10100-010-0183-
2

Persson, A., Rothstein, B., Teorell, J., 2013. Why Anticorruption Reforms Fail-Systemic Corruption as a Collective Action
Problem. Governance 26, 449–471. https://doi.org/10.1111/j.1468-0491.2012.01604.x

Rose-Ackerman, S., Palifka, B.J., 2016. Corruption and Government: Causes, Consequences, and Reform, 2nd ed. Cambridge
University Press, New York, NY.

Rothstein, B., 2013. Corruption and Social Trust: Why the Fish Rots from the Head Down. Soc. Res. (New. York). 80, 1009–
1033. https://doi.org/10.1353/sor.2013.0040

Rothstein, B., 2011. Corruption and Risks - Anti-corruption: the indirect “big bang” approach. Rev. Int. Polit. Econ. 18, 228–
250. https://doi.org/10.1080/09692291003607834

Ryan, R.M., Deci, E.L., 2000. Self-determination theory and the facilitation of intrinsic motivation, social development, and
well-being. Am. Psychol. 55, 68–78. https://doi.org/10.1037/0003-066X.55.1.68

Seleim, A., Bontis, N., 2009. The relationship between culture and corruption: a cross‐national study. J. Intellect. Cap. 10,
165–184. https://doi.org/10.1108/14691930910922978

Soto Torres, M.D., Fernandez-Lechon, R., Fernandez Soto, P., 2007. A system dynamics model about public corruption: the
influence of bribes on economic growth. Proc. 2007 Int. Conf. Syst. Dyn. Soc.

Sterman, J.D., 2000. Business Dynamics: Systems Thinking and Modeling for a Complex World. McGraw-Hill Higher
Education.

System Dynamics Society, 2016. Introduction to System Dynamics [WWW Document]. Syst. Dyn. Soc. URL
http://www.systemdynamics.org/what-is-s/ (accessed 7.4.17).

Tanzi, V., 1998. Corruption around the world: Causes, consequences, scope and cures (No. WP/96/63), IMF Working Paper.

Teksoz, U., Kalcheva, K., 2016. Institutional differences across resource-based economies (No. 2016/63), WIDER Working
Paper,.

Torsello, D., Venard, B., 2016. The Anthropology of Corruption. J. Manag. Inq. 25, 34–54.
https://doi.org/10.1177/1056492615579081

Transparency International, 2016. What is Corruption? [WWW Document]. URL https://www.transparency.org/what-is-


corruption (accessed 10.18.17).
Ullah, M.A., 2012. Enhancing the Understanding of Corruption through System Dynamics Modelling: A Case Study Analysis
of Pakistan. University of Auckland.

Ullah, M.A., Arthanari, T., 2011. Using a Qualitative System Dynamics Approach to Investigate Perceptions of Corruption.
Int. Conf. Syst. Dyn. Soc.

Ullah, M.A., Arthanari, T., Li, A., 2012. Enhancing the understanding of corruption through system dynamics modelling. Proc.
30th Int. Conf. Syst. Dyn. Soc. 1–30.

United Nations Economic and Social Council, United Nations, United Nations Economic and Social Council, 2015. A Nexus
Approach For The SDGs. New York.

Uslaner, E.M., 2013. Trust and corruption revisited: how and why trust and corruption shape each other. Qual. Quant. 47,
3603–3608. https://doi.org/10.1007/s11135-012-9742-z

Zelinka, D., Amadei, B., 2018a. A Systems Approach for Modeling Interactions Among the Sustainable Development Goals
Part 1: Cross-Impact Network Analysis (pending review). Int. J. Syst. Dyn. Appl.

Zelinka, D., Amadei, B., 2018b. Analyzing sustainable development: systems-based methods and modeling tools for the
Sustainable Development Goals, in: 24th International Sustainability Development Research Society Conference.
Messina, Italy.

Zelinka, D., Amadei, B., 2018c. A Systems Approach for Modeling Interactions Among the Sustainable Development Goals
Part 2: System Dynamics (pending review). Int. J. Syst. Dyn. Appl.

Zelinka, D., Amadei, B., 2017. A Methodology to Model the Integrated Nature of the Sustainable Development Goals:
Importance for Engineering Education, in: 2017 ASEE Annual Conference & Exposition. American Society for
Engineering Education, Columbus, Ohio.

Zelinka, D., Tarbert, M., Amadei, B., 2018. Dynamically modeling the interconnections between demographics and the water-
energy-food-climate nexus, in: Nexus Conference 2018: Water, Food, Energy, Climate. Chapel Hill, NC.

Zhang, Q., Prouty, C., Zimmerman, J.B., Mihelcic, J.R., 2016. More than Target 6.3: A Systems Approach to Rethinking
Sustainable Development Goals in a Resource-Scarce World. Engineering 2, 481–489.
https://doi.org/10.1016/J.ENG.2016.04.010

You might also like