University of Groningen
Faculty of Economics and Business
Thesis in International Business and Management (IB&M)
Topic:
Innovation and Corruption
A firm-level research on the relationship of innovation and corruption in
Portugal and Greece
Student: Georgios Chrysovalantis Dalampiras (s1941313)
g.c.d.dalampiras@student.rug.nl Supervisor: Gjalt De Jong,
g.de.jong@rug.nl
1
Groningen
2
Abstract
This paper investigates the relationship between the concepts of corruption and innovation by using data retrieved from BEEPS database, for 546 Greek and 505 Portuguese firms. Initially it defines corruption and innovation and the ways to measure them. The answer to the hypothesis is negative as no significant relationship has been found between firms that innovate and the
likelihood of bribery. Another finding of the study is that for this specific sample of the two countries, industry and ownership as firm characteristics are more significant determinants of bribery.
3
T
ABLE OF
C
ONTENTS
1. INTRODUCTION ... 5
2. THEORETICAL BACKGROUND AND HYPOTHESIS ... 9
2.1. Corruption ...9
2.2. Bribery and existing explanations ... 11
2.3. Measurement of corruption ... 13
2.4. Innovation ... 14
2.5. Hypothesis ... 18
3. DATA AND METHODOLOGY ... 21
3.1. Sample ... 21
3.2. Measures: Dependent variable ... 22
3.3. Measures: independent variable ... 23
3.4. Measures: Control variables ... 23
3.5. Econometric Model ... 25
3.6. Evaluation of method assumptions ... 26
4. EMPIRICAL RESULTS... 31
4.1 Descriptive statistics ... 31
4.2 Regression results ... 32
4.3 Robustness tests... 37
5. CONCLUSIONS ... 40
6. LIMITATIONS AND FURTHER RESEARCH ... 42
4
5
1.
I
NTRODUCTION
Corruption is considered as an immoral and harmful act for a society. It is usually defined as the abuse of a public officer’s power for private gain and it reflects a country’s sociopolitical, economic and legal institutions (Svensson, 2005). During the last decade research on corruption has witnessed an explosion of literature while it is a subject that attracts the attention of large financial institutions1. It is though corruption’s nature that does not allow us to estimate precisely its consequences due to the lack of extensive, objective and reliable data (Emerson, 2006). However, researchers of corruption have tried to develop models in an effort to evaluate the effect of corrupt officials’ acts on an economy (Rose-Ackerman, 1975; Reinikka and Svensson, 2004; Emerson, 2006).
In contrast to corruption, innovation can become the keystone of a country’s economic development. Innovation is part of the entrepreneurial initiatives that a firm takes as the latter attempt to introduce an invention or idea in a form of product (McDaniel, 2002). Given that the concept is considered as determinant of growth (Anokhin and Schulze, 2009), understanding it can probably explain the differentiated financial growth across countries (Ayyagari et al., 2010).
Even though academics have based their research on a de facto negative influence of corruption on a country’s economic development, very little research exists that tries to explain the relationship that the concepts of innovation and corruption have (Ayyagari et al., 2010; Anokhin and Schulze, 2009). So far, researchers have tried to explain the adverse effects of corruption on a country’s economic growth and development (Shleifer and Vishny, 1993; Ades and Di Tella, 1999; Svensson, 2003, 2005) but there is little that we know about the effects of corruption on the specific group of innovative firms.
1 The World Bank and the European Bank for Reconstruction and Development have invested on surveys in
6 Furthermore, the existing literature about innovation has focused on the analysis of the environment in which innovative companies operate (Baptista and Swann, 1998) or has examined the reasons that lead some firms to innovate more than others (Hanley and Perez, 2012). However, it is essential to realize if innovative firms need to pay bribes, so as to avoid the bureaucratic regulations and obstacles that arise from the incentives of the corrupted officials. Murphy et al. (1993), argue that innovators are more susceptible to public corruption than the companies that are already established due to the high and inelastic demand for state permits and licenses. If corruption can harm a country’s economic improvement subsequently can harm the entrepreneurial climate by discouraging innovative firms from introducing a pioneering produ ct or entering the market (Anokhin and Schulze, 2009).
Motivation of research
The investigation of the relationship between corruption and innovation is a fruitful field for research. The subject is very interesting for researchers as it includes two cont radictive sides of an economy, the negative one which is corruption and the positive one, innovation. Literature on corruption usually focuses on the examination of countries in a macroeconomic level and little has been written about the corruption that appears in specific firms or industries, like innovators (Ayyagari et al., 2010). Additionally, Veracierto (2008) argues that when many industries are victims of corrupt officials lower innovative activity appears, that leads to lower growth rates for the economy as a whole.
This master’s thesis has a main target to contribute to the literature that combines the two streams of research, corruption and innovation, with the intention of exploring the relationship of innovative firms and the bribery paid to the government officials. Following the aforementioned literature field, this research mainly shows interest to the effect of corruption on the firms that innovate. Consecutively, this research is based on the following questions.
Research Question:
7 As a result the main research question will be:
“What is the relationship between innovation and corruption?” The above main research question will have four sub-questions:
“What is corruption and how can it be measured?” “What is innovation and how can it be measured?”
“What is the relationship between innovation and corruption?”
“Is there evidence for the relationship of innovation and corruption in Portugal and Greece?”
More specifically, considering the company as the unit of analysis the goal of the thesis is to analyze whether companies pay bribes to officials so as to innovate. In order to fulfill the requirements of the research I hypothesize that the phenomenon of corruption has a positive relationship with innovation.
To answer the above mentioned research question and sub-questions, an extensive literature review has been written about the concept of corruption and its measures. Additionally, the phenomenon of bribery is reviewed and defined, since it is the form of corruption that the empirical part of the thesis also uses as a dependent variable. Moreover, the concept of innovation and its measures are explored by presenting the existing literature on innovation. The literature review also includes the principal agency theory that sets the necessary academic background which allows me to combine corruption and innovation so as to formulate the main hypothesis.
8 Regarding the selection of Greece and Portugal from the BEEPS database, it is worth noticing some relevant to the research characteristics. The two countries have almost the same population and GDP2. In terms of corruption, Transparency International ranks Greece on the 94th and Portugal on the 33rd position, among 176 countries, with corruption scores3 of 36 and 63 respectively. In terms of innovation, the Global Innovation Index4 ranks Portugal on the 35th position and Greece on the 66th, among 141 economies of the world (See table 27, Appendix C). Thus, at first glance the levels of corruption and innovation of these two countries can provide the necessary background for research.
Outline
The rest of the thesis is as follows: The second section has as primary target to clarify the basic concepts used for the analysis referring to the existing literature with studies in the fields of corruption and innovation. Afterwards on this section, I formulate the hypothesis that is the subject of the quantitative analysis on firm level determinants of corruption. In the third se ction, techniques and the data are summarized while the fourth section includes the empirical results, concluding with the fifth and sixth section where the overall view of the thesis and the fields that can be further explored are presented.
2According to the International Monetary Fund report for the year 2012, Greece is on the 42nd position and Portugal
on the 45th position of GDP (measured in US dollars), available at http://www.imf.org
3
The Corruption Perceptions Index ranks countries and territories based on how corrupt their public sector is perceived to be. A country’s score indicates the perceived level of public sector corruption on a scale of 0 - 100, where 0 means that a country is perceived as highly corrupt and 100 means it is perceived as very clean. A country's rank indicates its position relative to the other countries included in the index.
Available at http://www.transparency.org/cpi2012/results
9
2.
T
HEORETICAL BACKGROUND AND
H
YPOTHESIS
2.1. CORRUPTION
In this part of the research, the definition and theoretical background used for the paper are presented. Literature on corruption has many different definitions of the term to offer, which have more or less the same meaning. However, it is proper to set the background in order to forward to the implication of the theory.
Rose-Ackerman (1975) conducted the first extensive studies about the impact of corruption on the economy by presenting a principal agent model in order to explain the incident of graft. After Rose-Ackerman many scholars have tried to implement either the same model or have presented new approaches in the specific field (Svensson, 2003; Blackburn et al., 2006; Tang et al., 2008). So far, the literature supports that corruption can become a major obstacle in a country’s way to growth and development while it can also have a severe effect on entrepreneurship and foreign investments (Zhahra et al, 2001; Emerson, 2006; Vivod, 2003; Campos and Lien, 1999). Firms are forced to pay bribes when they have to deal with public officers whose public position can affect their business operation (Svensson, 2003) whereas it can become an extra burden when referring to low-income countries (Foellmi and Oechlin, 2007). More specifically, the problem initiates at that certain point where the appointed public officer shows the willingness to participate in this illegal exchange, taking in advantage of the power that his position gives him.
Blackburn et al. (2005), support that there is overwhelming evidence of significant negative relationship between graft and economic growth, based on the previous literature. The authors state (Blackburn et al., 2005), that in the past corruption was faced even as “speed money” that can enhance the growth of an economy by circumventing the institutional rigidities which work against efficiency. In their research it is shown that the rent paid to officials as exchange to their illegal services, has a negative impact on investments and capital accumulation as these funds are not used for more productive activities.
10 2003; Palifka, 2006). The other side approaches corruption from a micro-level outlook and analyzes corruption on a firm level. In this kind of literature, bribery is considered a dependent, to the firm-level determinants, variable (De Jong et al, 2012; Svensson 2003; Gaviria, 2002).
Palifka (2006), has researched the way susceptible to corruption social and psychological factors can influence entrepreneurship since the funds paid for bribes are not allocated to more productive activities of companies. The author defines the bureaucratic economic corruption as: “the use of one’s influence as a public servant for economic gain”, (Palifka, 2006; page 6). It is similar to the definition of Anokhin and Schultze (2009; page 465) and Vinod (2003; page 873) that define it as “abuse of public power or authority for private benefit”. In another research, Foellmi and Oechslin (2007; page 95) refer to the non-collusive corruption as “corruption that imposes an additional burden on business activity” while Dreher and Gassebner (2007) state that corruption can also be considered as “the speed of money” which can reduce the slow-moving queues in the public offices.
From another aspect in the literature, Blackburn et al. (2005; page 2452) explain that “corruption arises from the incentives of public and private agents to conspire in the concealment of information from the government”. Taking in mind the above mentioned definitions, it is clear that a generally accepted definition does not exist since authors tend to define corruption according to the needs of their research.
For this research, the definition of corruption is used as it is presented from the Transparency International (from now on TI) anti-corruption plain language guide (2009). TI is an independent organization whose incentive is to enhance the battle against corruption in a worldwide level:
11 employees about corruption. TI emphasizes on the power of the public officials and the way they use their power so as to become an illegal part of the market, by gaining personal profits.
Even if the TI mainly focuses on the corruption that we meet at the public official level, it cannot be conclude that the private sector has no corrupted side. It is though important to notice that the impact of corruption on a bureaucratic level is much greater on entrepreneurship and investments. Additionally, McArthur and Teal (2002) argue that it is crucial to distinct the so called “local” corruption (firm level) from the economy wide or “global” corruption since as shown from Schleifer and Vishny (1993), the centralized corruption of bureaucracy delivers better results -in terms of growth- than in decentralized uncoordinated corruption.
The bureaucratic corruption is also engaged in my research in order to explain bribery. It makes its presence observable when innovators have to deal with public officials that have incentives to ask for higher rents when the probable magnitude of the profits is also high. The absence of law enforcement by public officials can create higher risks for innovators. Such as, innovators have to rely on the integrity of the officials who have to fulfill the terms of the legally signed contracts, even if they have already participated in a corrupted exchange (Anokhin and Schulze, 2009). Further, this is also a principal agency issue that will be explained in depth later, on the hypothesis section.
2.2. BRIBERY AND EXISTING EXPLANATIONS
Before explaining the model further implied in my research, it is also important t o clarify the term of bribery. Bribery is defined in different ways around the world but it is an illegal action in every country to bribe its own officials. Bribes are usually paid in secrecy and the officials are that get caught, are forced to resign from their public position (Nichols, 1999).
For the needs of this thesis the definition of World Bank also used also in De Jong et al. (2012) is employed, bribery is defined as “the offer of solicitation, promise or gift of undue pecuniary or other advantages whether made directly or through intermediaries, to (foreign) officials or to a third party which the aim of influencing the actions of a public official or the officials’ duties”.
12 rules. The corrupted official favors the demand of the bribe giver, but acts within the margins of the existing legislation. The second group is the against-the-rule benefits, where benefits are received from an official so as to take action without respecting the law (Oldenburg, 1987).
As Svensson (2005) notices, the supporters of “efficient corruption” claim that bribery is the factor that allows firms to get things done, within a bureaucratic environment that has applied bad and rigid laws. Lui (1985), supports that some systems are based on bribery so as to allocate the licenses and governmental contracts. This type of systems has as a result that typically efficient firms have the ability to pay the highest bribes and benefit from them.
As it can be easily derived from the literature, during the transaction of bribery at least two parties are involved. Different actors can be bribed according to the needs of a firm and the hierarchy level they have to reach, in order to succeed their purposes. Fisman and Svensson (2007), conducted a firm survey by using data about the bribes estimated to have been paid by Ugandan firms so as to identify the relationship of bribery, taxes and firm growth. Their findings showed that bribery and firm growth are negatively correlated.
The literature has so far indicated two mainstreams on corruption and bribes paid, which mainly focus either on country level or firm level. Other studies also exist, that face bribery from a sociological and cultural perspective like Tsalikis and La Tour (1995) who explore the differences in the perception of two different cultures, the Greek and the American. Their research is based on hypothetical scenarios which present three different scenarios of a businessman’s bribe offer to a governmental official. In the same spirit, Rose-Ackerman (1999) investigated corruption from a more principled and moral perspective.
13 sheds light on the supply side of bribery has been created where researchers, as also the BEEPS’ data used for my research, investigate bribery on a firm level.
2.3. MEASUREMENT OF CORRUPTION
Studies that focus on firm level characteristics of bribery have increased while the challenge of retrieving the answers from the involved in bribery parties is quite demanding. It is important though to illuminate the way corruption can be measured. Normally, participants in bribery transactions have difficulties in answering the questions of the research conductor’s as they do not want to be found involved in cases of bribery scandals. So, researchers have to adapt their questionnaires in such way that they obtain the proper answers in order to forward to their analyses based on real data. Below, some characteristics are presented which according to previous studies are significantly related to corruption.
Svensson (2003), uses a data set with quantitative data about bribes paid in Ugandan firms. Having as a point of departure the question: “Who must pay bribes and how much?” Svensson’s findings show that firms usually have to pay bribes when they deal with officials who directly affect their business. Svensson supports that companies which operate in an economic environment which deals with exports and imports need to make usage of infrastructural services, so they are prone to bribery payments. Further in Svensson’s article (2003), we note that companies’ “ability to pay” bribes and refusal power explain the variation of bribes paid to officials, as also the amount of bribe paid is determined from a bargaining process that depends on their current and future profits. The author argues that the size of a firm, the range of products it trades and the exports it conducts increase their bribe negotiation power.
14 corruption. Finally, careful audits of specific projects conducted on a firm level can provide useful information about a project’s illegality.
As written in the above mentioned theoretical framework of Kaufmann et al. (2006), these can be the general guide lines in order to forward to real research. De Jong et al. (2012), focus on the supply side of bribery in order to analyze the relationship between bribery and performance in Vietnam. The hypothesis of the authors is that a non-monotonic relation between a firm’s performance and corruption exists. On the firm’s characteristics, the variables used are a firm’s age –since its foundation-, and the firm size which are both negatively related to bribery. The third variable used is ownership, which was found to be positively correlated to graft.
Reinikka and Svensson (2004), measure corruption by conducting public expenditure tracking surveys, on Uganda’s public spending on education. This alternative way of graft’s measurement allows the measurement of graft at the level of individual institutions and facilitates the study of corruption’s mechanisms. McArthur and Teal (2002), conducted their research by enacting the “extent of corruption” as determinant of firm performance in Africa. Their aim is was to test the relation of corruption and firm productivity, concluding that firms which operate in economies with pervasive bribes can be just one third productive in contrast with firms that operate in bribe-free economies.
In the same line of research, Gaviria (2002) links corruption and crime, showing that they have a negative effect on competitiveness and that they are positively related to the decrease of sales growth. He concludes that corruption and crime are strongly associated and cannot have any positive effects on an economy.
2.4. INNOVATION
15 Gaynor (2013), has focused on the terminology and explanation of innovation as it cannot be confused with ideas or inventions as ideas are born from the creativity of the individual. Unlike to ideas, proper and successful innovation requires the consideration of the idea from a team in such a way, with the intention that this idea can become a commercialized product which can be produced from a company.
Moreover, Johnson (2001; page 136) emphasizes that the clarification of the terms, innovation and entrepreneurship, is crucial as “language is part of the core of individual, business and organizational performance success”. He further supports that the confusion and ambiguity between the two terms can lead to lower levels of competitive output (Johnson, 2001). According to Johnson an organization can be characterized as innovative only when moves proactively away from “what is” to “what could be”. Innovation can take the following five forms, namely: R&D product development, new usage of established product, by the exploitation of the changes in the markets, the operational logistical innovation and business model innovation (Johnson, 2001).
Additionally, McDonald (2002) claims that innovation is only equated to entrepreneurship in the method used to produce and bring a new product or service to the market. According to McDonald (2002), in order for an entrepreneur to find a way of commercializing his new idea or creation has to combine it with one of the forms of innovation, namely: new products, new methods of production, new sources of raw material, new markets and/or new methods of organizing and industry. Wagner (2011) mentions that the ability of exporting companies to cope with their highly competitive industry sectors depends on their capability to commercialize high quality innovative products.
16 Based on a hypothesis that multinational companies have access to knowledge stocks of greater size, Criscuolo et al. (2004) find that globally-engaged firms innovate more than domestic firms. Even if they hire more researchers, it is the communication and daily attrition with the suppliers and customers of the globalized market that actually allows them to access a worldwide pool of information (Criscuolo et al., 2004). On the other hand, the increased levels of R&D investment from the private capital lead to a fall in the productivity of research (Lanjouw and Schankerman, 2004).
From another qualitative aspect on innovation, the influence of a company’s gender and age workforce composition can have an impact on its innovative performance. Pfeifer and Wagner (2012) find that firms with older employees present notably lower proportion of R&D expenditure in total revenues and R&D employment in total employment while there is evidence that companies with more female workers have a higher R&D activity. Furthermore, it is found that companies which turn their interest to exports are more likely to reveal innovative changes in their techniques of production, the year that they switch to exports (Hanley and Perez, 2012).
Having in mind the forms that innovation can take, it is needed to discover the ways it is so far measured in the academic literature. Rao (2007) writes that it is useful and vital to measure innovation as it is a way to assess the well-being of the rich economies while Rao et al. (2008) state that the quantitative measurement of innovations can help in the trading of this commodity via suitable derivative securities. Russo (2009), bases his research on the European Innovation Scoreboard and uses as equal–weighed indicators of innovation the R&D spending, the availability of venture capital and the number of students that have pursued degrees in science and engineering.
17 Lanjouw and Schankerman (1999), have used patent data for 100 U.S. manufacturing companies. The authors propose that innovation has two aspects, the quality of innovation and the value of innovation. Their aim is to create a model with earlier expectations of an already patented innovation as a latent variable by testing four indicators, namely: the number of patent claims, forward and backward citations and the family size. In a related research, Lanjouw and Schankerman (2004) conduct their research on the determinants of R&D productivity using for one more time a data panel of U.S firms and enact as factors the level of demand, the quality of patents and the technological exhaustion. Anokhin and Schulze (2009) measure the domestic innovation with the number of patent applications that were filled from the residents of the country yearly and the rate of realized innovation which is actually the rate of technological development realized in every country.
Summarizing this part, the discussion about innovation begins by distinguishing the concept from other terms such as ideas, inventions and entrepreneurship. As the literature review indicates, the previous are important factors that accompany innovation. From its qualitative side, innovation is based on group work (Gaynor, 2013). The group should have as a target to bring an idea to the stage of a commercialized product, by combining the idea with one of t he five forms of innovation.
The composition of the R&D team also matters since employees’ age and gender might differentiate a firm’s investment on R&D (Wagner, 2012). Firms that turn their interests to export, it is more likely that they have higher innovative activity (Hanley and Perez, 2012) and if they operate in a globalized market they have access to a larger pool of information that can enhance their innovative activity (Criscuolo et al., 2004).
18 this study uses as measure of innovation the amount of money spent on R&D, a variable that is transformed to a binary one.
2.5. HYPOTHESIS
In order to formulate the hypothesis usage of the principal agency theory is made As Waterman and Meier (1998) argue, principal agent models are often the basis of empirical and theoretical studies that relates bureaucracy to the elected officials. As the authors support, one side consists of the so called “buyer” of services who is the principal and the second side, which is the provider of services namely, “agent”. The agents take advantage of their position in order to obtain rent from the principals that are trying to overcome the obstacles set from the bureaucratic rules (Waterman and Meier, 1998).
The advantage of the agent is that he is in a position to have multiple principals that can demand bribe from. In reality, this is not always the way “things get done”. The “dyadic” model is the simplest form of bribe interaction but it is also possible that principals may have a “conflict” about the same service. Consider two principals having incentives on the same contract. This situation brings the agent into an advantageous position to increase its demands and gain from both sides while the agent himself might have to include other officials from higher position in the game.
De Jong et al. (2012), indicate that corruption in that sense can be either grand so it is associated to large amounts of bribery which end up in the hands of higher-level officials, or petty corruption that involves junior officials and smaller amounts of money paid. Furthermore, Emerson (2006) argues that models which involve a rent seeking behavior of a governmental official with a monopolist position are mainly implemented in early studies. Nowadays, literature has mainly focused on the causes and economic consequences of corrupt governments in developing countries (Emerson, 2006), in order to give explanation of the impact of competition to corruption.
19 present new products to the market might be discouraged from the imperfection of information about the markets. This creates an information asymmetry especially in cases of monopolies of the state owned companies of the sector (Palifka, 2006).
Moreover, a firm whose aim is to obtain a larger market share by developing new products or a newly established company in a competitive sector, need to face the officials who seek for personal gain. In that case, the firm which takes the decision to pursue an innovative opportunity depends on the added value that its initiative will capture in favor of the company (Anokhin and Schulze, 2009). However, the presence of corruption creates a big risk for innovators. The officials involved in the value chain maintain an opportunistic position against the profits derived from the innovation that the entrepreneurs are entitled to (Anokhin and Schulze, (2009).
Hence, principal agency theory5 offers the foundation for my study on the relation between innovation and corruption. Ayyagari et al (2010), examined 25.000 firms in 57 countries in order to test if bribery works as extra tax on innovation, especially in cases of small and young firms. Their findings show that firms with higher innovation rates are forced to pay more bribes to officials than non-innovating firms. Murphy et al (1993), point out that innovators are more exposed to the rent seeking behavior of the governmental officials as the characteristics of their projects create incentives for bribery. Particularly when they present needs for public services, licenses, new buildings, imports of new products or trademarks and the necessary inspections (fire, building, environmental) from different public services or participation in open competitions for public contracts; they will inevitably transact with public servants.
5A contradictive to agency theory is the cognitive evaluation theory which argues that a performance pay might
20 In order to achieve the purpose of this thesis, namely understanding whether the firm level characteristics play a role in corruption, I use the agency theory and in line with Ayyagari et al (2010) to hypothesize:
21
3.
D
ATA AND METHODOLOGY
This part of the study begins with the presentation of the survey and the sample used for the empirical test of the hypothesis. In the following sub-sections the usage of the independent and dependent variables are explained while afterwards the statistical model and the method of estimation is specified. Finally, the empirical assessment is presented in order to check if the methodological assumptions are satisfied. In the next section the empirical results are presented.
3.1. SAMPLE
For the estimations of the present thesis the data that have been used are available at the Business Environment and Enterprise Performance Survey (from now on BEEPS) database. BEEPS, is a joint initiative of the European Bank for Reconstruction and Development and the World Bank Group. The survey was first undertaken on behalf of the EBRD and World Bank in 1999 – 2000, when it was administered to approximately 4000 enterprises in 26 countries of Eastern Europe and Central Asia to assess the environment for private enterprise and business development.
The objective of the survey is to obtain feedback from enterprises in EBRD countries of operation from the state or the private sector. Its goal is to contribute in building a panel of enterprises’ data that will make it possible to track changes in the business environment over time. The survey examines the quality of the business environment as determined by a wide range of interactions between firms and the state. As such, it facilitates the research and serves as an input of the policy dialogue within the countries of Central and Eastern Europe. The survey has been so far implemented in four rounds within the years 1999, 2002, 2005 and 2009.
22 The data gathered for the two Mediterranean economies of Greece and Portugal come from the year 2004, year that the survey was conducted. The dataset used for the needs of the research include all the firms that participated in the survey, with 546 firms from Greece and 505 from Portugal.
3.2. MEASURES: DEPENDENT VARIABLE
The dependent variable used of the thesis is corruption. In the econometric model enacted, in order to test my hypothesis, bribery is tested as a measure of actual corruption.
EBRD has used different types of questions in order to facilitate the participants. It is expected that participants usually hesitate to answer directly to questions about the amount of money their firm actually pays in bribes or directly indicate which organizations and officials ask for an unofficial payment. Consecutively, they have created two different types of questions. The first type of questions measures bribery in an actual amount of money as a percentage of total annual sales and percentage of contract values paid in bribes. The other type of questions is based on a frequency table that indicates the frequency that a firm needs to pay money to un-officials of different public organizations and services in order to detour their duties. The frequency tables of BEEPS take values from never to always and don’t know: Never = 1, Seldom = 2, Sometimes = 3, Frequently = 4, Usually = 5, Always = 6, Don’t know = -9.
For the econometric model, as a measure of bribery I have constructed the dummy variable, Likelihood to bribe, which is the response to the question:
“On average, what percent of total annual sales do firm’s like yours typically pay in unofficial payments/gifts to public officials?”
23 The econometric model engaged to test the measure of corruption is the probit model as the dependent variable has been transformed into a binary one that reports 0 (zero) as no graft paid and 1 (one) as positive graft reported. This model allows the examination of the impact of a change by one unit of the independent variables on the likelihood that a company pays bribes. It has to be clarified that the goal of this research is not to examine the real magnitude of graft but to check which firms are prone to bribery, by making usage of certain firm characteristics. Consecutively, the model needs to include any size of bribery as it is needed to check the corrupted action by distinguishing it to graft and lack of graft.
3.3. MEASURES: INDEPENDENT VARIABLE
InnovationInnovation is the independent variable of the model which is connected to the hypothesis. Innovation (as mentioned in sub-section 2.3) can be measured in different ways. Within the BEEPS a set of options was given to the participants in order to explain how innovation is executed within their firms. I have chosen to include as a measure of innovation the amount of money spent on Research and Development. The amount of money spent on R&D has been transformed into a binary variable that takes the value 1 if a firm invests money on R&D and 0 otherwise (De Rosa et al., 2010; Ayyagari et al., 2010). As stated in the hypothesis, the assumption is that innovation has a positive effect on the likelihood of corruption.
3.4. MEASURES: CONTROL VARIABLES
In the econometric model four control variables are used, namely: age, size, ownership and industry.
Age
24 newly established company that counts 2 years of life than on an a firm that is operating for more than 100 years. Therefore, we examine the impact of a firm’s inveteracy as a percent.
Size
The second control variable is a firm’s size. The number of employees is the indicator for the firm size in the BEEPS database (Ayyagari et al., 2010). The variable for size takes the value 1 (one) if a firm is small, 2 (two) if it is a medium size firm and 3 (three) if it is a large firm. As it was already known from the BEEPS explanatory document, both Greece and Portugal have a high concentration of small firms in the sample, 81% and 77% respectively.
Table 1. Explanatory table for the ordinary variable of size
Size Number of Employees
Small Firm 2-49
Medium Firm 50-249
Large Firm 250-9.999
Ownership
The third control variable is a firm’s ownership structure. In order to check where the sample is concentrated, a frequency test was conducted (See frequency tables 11 and 13, Appendix A). After conducting a frequency test for both countries, two (2) dummies were created according to the characteristics and the concentration of the initial variables (De Rosa et al, 2010; Seker and Yang, 2012).
25 Industry
The fourth control variable is the industry in which the firm operates in. For this category, a frequency test was again conducted so as to check where the sample for industry is concentrated. The samples of both countries have shown again similarities. The logic behind the formation of the new dummies is that they are categorized according to their nature and afterwards according to their concentration. The initial variables of the sample were transformed into four dummies (See tables 12 and 14, Appendix A).
Consecutively the industrial sector is represented with 2 (two) dummies. The first dummy takes the value 1 if a company belongs to the branch of mining or construction otherwise 0, and the second one takes the value 1 if belongs in the branch of manufacturing otherwise 0.
The sector of services is also represented from 2 (two) dummies. Such as, the third dummy takes the value 1 if it belongs in transportation services otherwise 0. The fourth dummy takes the value 1 if it belongs in retail and repair services, otherwise 0.
The base scenario (Hill, 2008) when all 4 (four) dummies take 0 values indicates representatives responding that their firm operates in any other kind of services namely: Motion picture and video activities, radio and television activities, other entertainment activities, news agency activities, washing and dry cleaning, hairdressing, funeral and related activities and other service activities.
3.5. ECONOMETRIC MODEL
The econometric model employed for testing the hypothesis has taken the following form:
26
Table 2. Indicative table of the variables used in the econometric model Type
Name
Independent Dependent Control Variable Dummy Comments Likelihood to bribe x x Transformed from a percentage of annual sales to dummy Log_age x x Logarithm of a company’s inveteracy size x x Measured in real number of employees Single x x x Single proprietorship or corporation privately held Partner
x x x Dummy for Partnership
or Cooperative firm R&D
x x x
Transformed from actual amount of money to dummy
Mining
x x x
Industrial dummy for Mining and
Constructions Manufacturing
x x x Industrial dummy for
Manufacturing firms Transport
x x x Service dummy for
transportation firms RetailRepair
x x Service dummy Retail
and Repair fims
3.6. EVALUATION OF METHOD ASSUMPTIONS
In order to test if the econometric model of the research provides the best linear unbiased estimates, three critical tests were conducted, namely: heteroscedasticity, normality of the residuals and the variables and multicollinearity of the explanatory variables. A probit model is used in order to run the regression of the econometric model since the dependent variable, likelihood of bribery, is a binary one.
27 Greece and Portugal can be found in tables 25 and 26, Appendix B. Below, the aforementioned tests are reported.
3.6.1.HETEROSCEDASTICITY
It exists when there is not a constant variance among the observations of a sample and when it is observable, the standard errors of the estimates are biased and as a result the uncertainty increases in the regression model. The plots of the residuals were examined by the Breusch-Pagan-Godfrey test which was implemented under the purpose of checking for heteroscedasticity. In this test the null hypothesis suggests that the error variances are constant and its purpose is to detect the presence of heteroscedasticity. After the linear regression has been run, the error term is squared and used as independent variable. Then the later variable is regressed on all independent variables in order to find out whether the p-value is less than the chosen level of significance. If it is less than 0.05 we are provided with statistical evidence against homoscedasticity, otherwise we cannot reject the null hypothesis. The problem with heteroscedasticity is that more weight is given to observations with potentially larger error terms. In that case observations furthest away from the true regression line provide us with the least information about the true regression line. In case heteroscedasticity is detected, there are methodological ways to correct for the weight of larger error variances to get estimates with the smallest sum of squared errors. Both countries present a p<0.05 (for 95% confidence interval). The closer to zero the p-value is, the stronger the justification of the rejection of the Ho hypothesis of homoscedasticity is (See Table 3). According to table 3, in both countries heteroscedasticity seems to be an issue.
Table 3. Heteroscedasticity Ho:
Constant variance
Variable: Bribery Greece Portugal
Breusch - Pagan - Godfrey : Test for Heteroscedasticity
Chi2(9) =17.92431 Prob>Chi2 = 0.0350
28
3.6.2.MULTICOLLINEARITY
Multicollinearity occurs when there is high correlation between the explanatory variables. The assumption is that the independent variables are not perfectly correlated such that “the values of xik are not exact linear functions of other explanatory variables.” (Hill et al., 2009: 154). If this assumption is violated, variables are said to be collinear, which makes it difficult to isolate the relationship between variables. OLS estimates will not be biased and still be the best linear unbiased estimates. However, we would get relatively imprecise information about our unknown parameters due to large standard errors. Difficulties might occur in the prediction of the true parameters, an issue which might even become more problematic when there is little variation in the explanatory variables. This latter instance could in fact be present as some of the sample variables originate from categorical data. To test for the presence of multicollinearity the variance inflation factor (VIF) is calculated, which is an index estimating how much the variance of an estimated regression coefficient is inflated due to collinearity. The VIF have values ranging from 1.17 to 5.90 for Greece which is below the cut off figure of 10, as this is recommended by Hill (2009), (See table 4; table 5). In Portugal, both ownership dummies present high values of multicollinearity close to 10, 8.97 for the single ownership dummy and 9.13 for the partnership/cooperative ownership dummy.
Table 4. Test for Multicollinearity-GREECE
Variable Uncentered VIF Centered VIF
Log Age 2.96 1.23 Size 8.16 1.51 Single - ownership 43.17 5.90 Partnership/cooperative - ownership 6.62 5.89 Mining/construction - industry 1.42 1.26 Manufacturing - industry 1.74 1.43 Transport - services 1.27 1.17 Retail/repair - services 2.08 1.41
Research and Development
1.46 1.36
Mean VIF
29
Table 5. Test for Multicollinearity–Portugal
Variable Uncentered VIF Centered VIF
Log Age 2.37 1.32 Size 8.91 1.79 Single - ownership 14.38 8.97 Partnership/cooperative - ownership 22.51 9.13 Mining/construction - industry 1.46 1.29 Manufacturing – industry 2.53 1.82 Transport - services 1.24 1.18 Retail/repair - services 2.19 1.54
Research and Development
1.64 1.40
Mean VIF 3.16
VIF :MULTICOLLINEARITY TEST
3.6.3.NORMALITY
In order to test the normality of the residuals and the variables, it is needed to check the values of Skewness and Kurtosis. Skewness involves the symmetry of the distribution. When Skewness is normal then it involves a perfectly symmetric distribution. A distribution that is positively skewed has the scores clustered to the left side, with a tail that is extended to the right while a negatively skewed distribution has scores clustered to the right and a tail extended to the left. Kurtosis deals with the peak of the distribution. A normal Kurtosis has a bell-shaped distribution and neither too peaked or too flat. A positive Kurtosis is indicated by a peak while a negative Kurtosis is indicated when a flat distribution appears.
30
Table 6. Normality test for residuals
Country Skewness Kurtosis
Greece 1.17 2.72
Portugal 1.68 4.27
Table 7. Normality test for all variables – Greece and Portugal
Portugal Greece
Variable Skewness Kurtosis Skewness Kurtosis
31
4.
E
MPIRICAL RESULTS
4.1 DESCRIPTIVE STATISTICS
The data that were collected from the BEEPS database concern 1051 observations in total, for the year 2004. The data did not contain a significant number of missing observations, for the variables used in this study. Missing values have appeared only in the variable of Research and Development. The reason is that when firms did not dedicate funds for the R&D department6, simply answer a zero (0) amount of funds. In cases of no profits, it was not possible to spend money on R&D and this might have caused the presence of missing values in the sample.
In terms of bribery, 118 Greek and 73 Portuguese firms have reported bribery, as a percentage of their total annual sales. As explained earlier (part 3.2 and 3.3), the data contained specific numbers of funds that were spent on R&D and specific percentages of total annual sales paid in bribes but due to the different needs of this research, they were transformed into dummy variables.
The tables of the descriptive statistics of the variables can be seen in the Tables 17 and 18 of the Appendix B. Moreover, the Pearson correlation matrix presents no significant correlation among the variables of the econometric model with the exception of the two ownership dummies in Portugal (See tables 19 and 20, Appendix B).
Table 8. Indicative table of firms that reported bribe and R&D expenditures
Country Reported Bribe Firms that reported funds
spent on R&D
Greece 21,6% 483 / 546
Portugal 14.4% 412 / 505
The percentages for the firms that reported bribes paid to officials result by dividing the number of affirmative answers with the number of the firms that were interviewed. As it can be seen in
6 We have to keep in mind that the firms which participate have a very large number of single owned firms.
32 table 8, for the year 2004 Greece has a higher percentage of firms that paid bribes than in Portugal.
4.2 REGRESSION RESULTS
The main test of the model was conducted separately for Greece and Portugal. The high multicollinearity and correlation that appeared for the dummies of ownership in the sample of Portugal did not allow the usage of the dummy variables of ownership (See table 5; table 19, Appendix B). Such as, the control variable of ownership is not included in the regression for the Portuguese sample.
Greek firms
Table 9 below, provides a summary of the probit regression results for the 546 Greek firms. The first model estimates the effect that the control variables have on the likelihood of bribery and includes the independent variable of R&D. The second model excludes the independent variable of R&D.
Hypothesis 1 predicts that innovation has a positive effect on the likelihood of corruption. Model 1 shows that this hypothesis is rejected due to the fact that the coefficient of R&D is very low (β= 0.19) and not significant (p>0.1), (See table 9).
33
Table 9. Regression results of the impact of innovation (R&D) on the likelihood of bribery: Greek (N=546) firms
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Portuguese firms
As it is already mentioned, the high correlation and multicollinearity of the dummy variables which indicate the ownership status of the Portuguese firms had to be excluded, as the regression could not produce any results (See table 19, Appendix B; table 5). Table 10 consists of the first model that estimates the effect of the control variables on the likelihood of bribery as also the
Variable Model 1 Model 2
34 dummy variable of R&D. The second model excludes the R&D variable in order to check the differentiation of the results.
35
Table 10. Regression results of the impact of innovation (R&D) on the likelihood of bribery: Portuguese (N=505) firms
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Variable Model 1 Model 2
36 Likelihood ratio test
The main objective of the econometric model is to examine if R&D expenditures influence the probability of bribery. After estimating a model with R&D and one without it, we need to test how much this variable increases the predictive performance of the model. For this purpose, the Likelihood Ratio test is implemented (Verbeek, 2012). Unrestricted is the model that includes the R&D variable. Under the null hypothesis of no improvement, the statistic LR=2* (Logunrestricted –Logrestricted) where the first value is the log likelihood of the model 1 while the second is the log likelihood of the model 2, the one without the R &D variable and follows chi-squared distribution with 1 degree of freedom (1 df as we examine one variable only). Calculating the statistic:
H0: The model is not improved by inclusion of the variable.
H1: The model is improved by inclusion of the variable
Greece: LR= 2[-247.1095 – (-271.1727)] = 2*24.0632 = 48.1264
Portugal: LR= 2[-166.5907- (-198.0252)] = 2*31.4345 = 62.869
LR is higher than 3.84 for both countries, thus the null hypothesis of no improvement in the model is rejected, at the 95% confidence level showing that the inclusion of R&D is justified (Verbeek, 2012).
McFadden R-squared
A goodness-of-fit measure indicates the predictive performance of a model. In contrast to the linear regression where there is a single such measure (R2), in qualitative models the predictive accuracy is examined either in terms of the fit between the calculated probabilities and observed frequencies or how well the model predicts observed responses (Verbeek, 2012). Usually, the McFadden R2 is used to examine how well a model with variables (unrestricted) performs compared to a model with only a constant (very restricted). The McFadden R2 which is displayed in the table is calculated by the formula: McFadden R2=1- logLunrestricted – logrestricted.
37
4.3 ROBUSTNESS TESTS
Robustness tests are frequently used so as to ensure that the predictions of the model are not biased in any way. They are conducted in order to examine the behavior of the regression coefficients when the econometric model specification is modified. The model has been initially run against three different measures of bribery and then against three different measures of innovation.
The first measure of bribery enacted, is the percentage of total annual sales that have been paid for bribes, primary form of the dummy variable of bribery of the main model, (see section 3.2). Hypothesis 1 is not confirmed as the coefficient of R&D is (β=0.10; p>0.10) for the Greek firms and (β=-0.03; p>0.1) for the Portuguese firms. The only significant predictor of bribery is the dummy variable for retail services of Portugal (β=-0.25; p<0.1), (see table 28; table 29, Appendix C).
As second measure of bribery, the percentage of the contract value that firms paid so as to secure a contract has been used. The OLS regression results show that R&D variable is not a significant predictor of bribery for the two countries (p>0.10). The partnership dummy (β=-2.66; p<0.05), mining dummy (β=0.98; p<0.1) and logarithm of age (β=0.02; p<0.1) for the Greek model and the logarithm of age (β=0.03; p<0.01) and retail dummy (β=0.03; p<0.01) are significant predictors of the likelihood of bribery (see table 30; table 31, Appendix C). It is worth noticing that only 78 of the 505 Portuguese firms have given an answer, so this fact might have influenced the results.
38 The ordinal probit model results are available in Tables 32 and 33 whilst they are similar to the main model. R&D dummies for Greece and Portugal are not significant (β=-0.2; p>0.1) and (β=0.19; p>0.10) respectively. Firm size has a statistically significant influence on the likelihood of bribery for both countries (p<0.05). The variables of ownership, mining and manufacturing are significant predictors of bribery for the Greek firms, all with values (p<0.05). The model of Portugal has shown again that retail services dummy is a significant predictor of bribery (β=-0.31; p<0.1).
In order to better analyze the implications of the model, I have further run the model against three different measures of innovation (see table 34; table 35, Appendix C). The first measure of innovation is the funds spent from firms for new buildings, machinery and equipment in thousand “000” euro. Τhe coefficient of innovation variable for Greece is β=0.19 with (p>0.1) and has been found marginally statistically significant for Portugal (p<0.1). For both countries transportation dummy is a significant predictor of bribery (p<0.1). The control variables of partnership (p<0.05), single ownership (p<0.1), manufacturing (p<0.1) and mining (p<0.05) are significant predictors of bribery for the Greek firms as well as size and retail variables (p<0.01) for the Portuguese sample.
Secondly, the model has been tested against the measure of innovation in thousand “000” euro invested on R&D and a summary can be found in Tables 36 and 37. The R&D variable for Greece is not a significant predictor of bribery (p>0.1) and has been found marginally statistically significant for Portugal (p<0.1). Moreover, the variables of transportation (p<0.1) and size are significant predictors of graft for both countries. Partnership dummy (β=1.24; p<0.1), mining (p<0.01) for the Greek model and retail-repair (p<0.01) for Portugal are statistically significant.
40
5.
C
ONCLUSIONS
This paper concerns a Master’s thesis for a study about the concept of innovation and its relation to corruption. The aim of the thesis was to understand the relationship between innovation and corruption as these two concepts seem to have differentiated impact on an economy. The existing literature on corruption supports that the latter can become an obstacle of a country’s economic development as long illegality between firms and government officials exist. Contrarily, literature on innovation supports that the concept can become a source of economic growth.
In order to answer the main research question and the sub-questions, I initially conducted a literature review that specifies and defines corruption and innovation. The phenomenon of corruption is defined as the abuse of a public officer’s position with the incentive of private gain. Moreover, the concept of innovation was distinguished from ideas and inventions as it includes the necessary team work that can transform an idea to a commercialized product. Further, the measures of innovation and corruption have been presented so as to facilitate the empirical part of the thesis. As a result of the literature review about the two concepts, bribery was used as measure of corruption and R&D expenditures as measure of innovation.
Thereafter, the research included two empirical studies for the countries of Greece and Portugal. The sample of analysis was retrieved from the BEEPS survey of the year 2004 and created a database of 546 firms for Greece and 505 for Portugal. Both countries present high indicators of corruption while Portugal’s innovative activity is higher than Greece’s. It has to be specified that the conclusions refer only to the specific sample of the 1051 firms that were examined so they cannot be generalized for all the firms that operate in these countries. The examination of the relationship of corruption and innovation in the two countries has come to the following conclusions.
41 countries has a greater effect on the incidence of bribery, than other firm characteristics. Additionally, both single and cooperative ownership variable are found to be significant for the Greek sample. This might be a sign of greater influence of a firm’s ownership structure to the likelihood of bribery. If we also consider that the majority of the Greek firms participating in the BEEPS survey are privately held or single owned, then they might be more exposed to a possible case of bribery.
42
6.
L
IMITATIONS AND FURTHER RESEARCH
When investigating a specific subject some limitations are supposed to be risen. Consecutively, my research is not without limitations. Within side, innovation might not be an important firm level characteristic that triggers the phenomenon of corruption in these two countries. Indeed, if the sociopolitical background of a country’s encourages bribery, then the government officials might keep pursuing bribes arbitrarily despite the relevant firm characteristics.
Another limitation is the simplicity of the econometric model used as there is only one linear effect. Such as, innovation might be a moderator or the relation of corruption and innovation might not be linear making any kind of non-monotonic relation possible. For instance, we cannot exclude the possibility that while an increase of innovation is accompanied by an increase of bribery, a threshold might exist, above which a further increase of innovation might lead to the opposite effect. If a U shaped curve could more appropriately depict the relation between the two variables, this could be a reason for misspecification. Violation of monotonicity was not possible to be tested by the present model and it could be an interesting avenue for further research.
Additionally the change in the significance of the control variables could be an evidence of inconsistency of the model. After the exclusion of innovation, the manufacturing variable turns out to receive even stronger weight in predicting the bribery outcome.
Furthermore, the study uses cross sectional data and not panel data, so it is difficult to check the model for causality as the collected data come only from one year. Finally, one more limitation is the countries analyzed as the similarities between them might have biased the results. Therefore, an implication of this econometric model to a set of different countries might change t he findings so this could be also motivation for further research.
44
R
EFERENCES
Ades, A. and DiTella, R., (1999), “Rents, Competition and Corruption”, American Economic Review, Vol.89, pp. 982-993;
Anokhin, S. and Schulze, W.S., (2009), “Entrepreneurship, Innovation and Corruption”, Journal of Business Venturing, Vol.24, pp. 465-476;
Ayyagari, M., Demirguc-Kunt, A. and Maksimovic, V., (2010), “Are Innovating Firms Victims or Perpetrators? Tax Evation, Bribe Payments, and the Role of External Finance in Developing Countries”, Policy Research Working Paper, pp. 1-34
Baptista, R. and Swann, P., (1998), “Do firms in clusters innovate more?”, Research Policy, Vol. 27, pp. 525-540;
Blackburn, K., Bose, N. and Haque Emranul, M., (2006), “ The Incidence and Persistence of Corruption in Economic Development”, Vol. 30, pp. 2447–2467;
Campos, E.D. and Lien, D., (1999), “The Impact of Corruption on Investment: Predictability matters”, World Development, Vol. 27, pp. 1059-1067;
Criscuolo, C., Haskel, J.E. and Slaughter, M.J., (2004), “Why Are Some Firms More Innovative? Knowledge Inputs, Knowledge Stocks, and the Role of Global Engagement”, pp. 1-36;
De Jong, G., Tu, A. P. and Van Ees, H., (2012), “Which Entrepreneurs Bribe and What Do They Get From It? Exploratory Evidence from Vietnam”, Entrepreneurship Theory and Practice, Vol. 36, pp. 323-345;
45 Dreher, A. and Gassebner, M., (2007), “Greasing the Wheels of Entrepreneurship? The Impact of Regulations and Corruption on Firm Entry”, KOF Swiss Economic Institute, KOF Woriking Paper No.166 and CSEifo Working Paper No. 2013;
Emerson, P. M., (2006), “Corruption, Competition and Democracy”, Journal of Development Economics, Vol. 81, pp. 193-212;
Fisman, R. and Svensson, J., (2007), “Are Corruption and Taxation Really Harmful to growth? Firm Level Evidence”, Journal of Development Economics, Vol.83, pp.63-75; Foellmi, R. and Oechslin, M., (2007), “Who Gains from non-collusive Corruption”,
Journal of Development Economics, Vol. 82, pp. 95-119;
Gaviria, A., (2002), “Assessing the Effects of Corruption and Crime on Firm Performance: Evidence from Latin America”, Emerging Markets Review, Vol. 3, pp. 245-268;
Gaynor, G.H., (2013), “Innovation; Developing the Visual Prototype”, IEEE Engineering Management Review, Vol.41, pp.5-6;
Hanley, A. and Perez, J.M., (2012), “Are newly exporting firms more innovative? Findings from matched Spanish innovators”, Economic Letters, Vol. 116, pp. 217-220; Hanley, A. and Perez, J.M., (2012), “Are Newly Exporting Firms More Innovative?
Findings from Matched Spanish Innovators”, Economic Letters, Vol. 116, pp. 217-220 Haque, N.U., (2007), “Entrepreneurship in Pakistan”, Pakistan Institute of Development
Economics, Vol. 29, pp. 1-54;
Hill, R. C., Griffiths, W. E. and Lim, G. C., (2008), “Principles of econometrics”, Third Edition, Danvers: MA Wiley;
46 Kaufmann, D., Kraay, A. and Mastruzzi, M., (2006), “Measuring Corruption: Myths and
Realities”, Development Outreach, pp.1-7, available at World Bank.org;
Kunz, A.H. and Pfaff D., (2002), “Agency Theory, Performance Evaluation, and the Hypothetical Construct of Intrinsic Motivation”, Accounting, Organizations and Society, Vol.27, pp. 275-295
Lanjouw, J.O. and Schankerman, M., (1999), “The Quality of Ideas: Measuring Innovation with Multiple Indicators”, NBER Working Paper Series, pp.1-36, Available at: http://www.nber.org/papers/w7345;
Lanjouw, J.O. and Schankerman, M., (2004), “Patent Quality and Research productivity: Measuring Innovation with Multiple Indicators”, The Economic Journal, Vol. 114, pp. 441-465
Lui, F.T., (1985), “An Equilibrium Queuing Model of Bribery”, Journal of Political Economy, Vol.93, pp. 760-781
McArthur, J. and Francis, T., (2002), “Corruption and Firm Performance in Africa”, The Centre for the Study of African Economies, Working Paper Series: 169, pp.1-24;
McDaniel, B.A., (2002), “Entrepreneship and Innovation: An Economic Approach”, Book Reviews / The Social Sciance Journal, Vo.40, pp. 343-356;
Nichols, P.M., (1999), ‘’Regulating Transnational Bribery in Times of Globalization and
Fragmentation’’, Yale Journal of International Law, Vol.257, pp. 257-303
Oldenburg, P., (1987), ‘’Middleman in Third-World Corruption: Implication of an Indian Case’’, World Politics, Vol. 39, pp. 508-523
Palifka, B.J., (2006), “Corruption and Entrepreneurship in Brazil”, paper presented at the 150 mile conference Edinburg, Texas April 22, 2006
47 Rao, B.C., (2007), “Economic Recognition of Innovation”, Paper presented at the Singapore Economic Review Conference (SERC), Singapore, Available at: http://worldbenefit.case.edu/research/paperseries/?p=128;
Rao, B.C., (2010), “How to Measure Innovation”, The Magazine of Economic Affairs, Vol. 53, pp. 109-125
Rao, B.C., Sundararajan, A. and Rao, S.C., (2008), “Trading of Innovations to Combat Globalization and Climate Change”, Challenge, Vol. 51, pp.70–84
Reinikka, R. and Svensson, J., (2004), “Local Capture: Evidence from a Central Government Transfer Program in Uganda”, The Quarterly Journal of Economics, Vol.119, pp. 679-705;
Rose-Ackerman, S., (1975), “The Economics of Corruption”, Journal of Public Economics, Vol.4, pp. 187–203;
Rose-Ackerman, S., (1999), “Corruption and Government”, Cambridge: Press Syndicate University of Cambridge;
Russo, G., (2009), “NatureJobs”, Nature,Vol. 457, pp. 357-386 ;
Seker, M. and Yang, J.S., (2012), “How Bribery Distorts Firm Growth: Differences by Firm Attributes”, Policy Research Working Paper 6046, pp. 1-40;
Shleifer, A. and Vishny, R.W., (1993), “Corruption”, The quarterly Journal of Economics, Vol. 108, No.3, 599-617;
Svensson, J. (2003), “Who Must Pay Bribes and How Much? Evidence from a Cross Section of Firms”, The Quarterly Journal of Economics, Vol.118, pp. 207-230;
