1 Introduction

Although the word ‘corruption’ (or corrupt practices) is commonly used to refer to illegal transactions in return for receiving favours, it is difficult to define exactly what corruption is in practice. Experts have provided a variety of definitions of corruption, but none can be considered as universally accepted.1 For instance, an interesting example is provided by Ray (2006) who observed that offering gifts in traditional Eastern cultures is considered a custom. The same is regarded as corruption in the Western world. Due to the difficulty in adopting a universally acceptable definition of corruption, the United Nations Convention against Corruption (UNCAC) defines specific acts of corruption through a series of Articles (legal articles).2

Irrespective of how I define it, corruption is considered a major issue for the global economy, especially for developing countries, as it results in the loss of potential resources. It also limits a country’s ability to meet the United Nations 2030 Agenda for Sustainable Development and the underlying Sustainable Development Goals (SDGs). Corrupt practices have taken a substantial amount of money that could have been invested into governance, social protection and transitioning to a green economy. These are the priority areas the SDG Push had designed to prevent 169 million people from being driven into extreme poverty. UNDP (2022) observed that, for African countries with high levels of corruption, governments spend 25% less on health and 58% less on education. Corruption is also believed to hinder the inflow of foreign direct investment (FDI), especially in developing countries with limited avenues for domestic investment.

With this background, the aim of this chapter is to discuss and investigate the cost and impact of corruption with a specific focus on FDI. Besides providing conceptual details and theoretical discussion, the chapter also presents new empirical evidence on the impact of corruption on economic growth and investment using data from a sample of 40 countries (the 20 least corrupt and the 20 most corrupt). Although empirical findings suggest that corruption does impact growth and FDI, the relationship is non-linear. While empirical evidence supports a ‘grabbing hand’ view for the sample of 20 least corrupt countries, a ‘helping hand’ view is also evident for the 20 most corrupt countries. The chapter is organised as follows. Section 2.2 discusses the economic cost of corruption; Sect. 2.3 reviews the empirical literature on corruption, while some data presentation and an econometric framework for empirical analysis are provided in Sect. 2.4. Section 2.5 discusses the results and Sect. 2.6 draws conclusions.

2 The Economic Cost of Corruption

As stated above, corruption takes away potential resources and could have a substantial cost to the economy. Some estimates suggest that the economic cost of corruption is extremely high and is more than 5% of global GDP.3 Johnsøn and Taxel (2015) report that, globally, around USD1 trillion is paid in bribes each year. The report also claimed that 20 to 40% (equivalent to USD20 to USD40 billion) is stolen each year from official development assistance thanks to high-level corruption within the public budgets of developing countries and subsequently hidden overseas.4

It is believed that the risk and amount of corrupt practices in public sector spending would have increased during crisis periods such as the COVID-19 pandemic. Potential practices include misappropriation of allocated funds as part of relief measures, the delivery and distribution of vaccines and the procurement of medical equipment such as personal protective equipment, swabs and ventilators.5 Cadrado (2022) provides some evidence of corruption during COVID-19 in different countries. The main sources of corruption include the sales of falsified COVID test certificates (South Africa, Lesotho, the UK, France and Spain); bribes and queue-jumping to access COVID-19 vaccines (Lebanon, Malaysia, the Philippines, Peru, Argentina, Spain, Poland, Canada and Ecuador); profiteering through artificially creating an oxygen crisis, hefty fees for hospital bed access, as well as artificial scarcity for such beds (India and Peru); and a black market for vaccines and the use of fake vaccines (Venezuela and Iran).

Scholars have provided detailed estimates of the cost of corruption in different countries (see Table 2.1). Although the authenticity of these estimates is dependent on the data and methodologies used, the substantial cost of corruption, particularly in developing countries, means that the issue remains important and requires further investigation. The estimates highlight some extreme cases of how corrupt practices could impact the GDP or the national budget of an economy. For instance, Reinikka and Svensson (2004) provide estimates that losses due to corruption in public spending of educational funds intended to cover schools’ non-wage payments in Uganda amounted to 87% of the total allocated funds. Olken (2007) reports that Indonesia lost 24% of the cost of the building of rural roads (funded through a national government programme) due to corruption. Khawaja and Mian (2005) found that politically connected loans in Pakistan led to a loss of GDP in the range of 0.3 to 1.9%. Anwar and Khalid (2006) observed that estimated public sector losses due to corruption in the same country accounted for about 10% of GDP. Niehaus and Sukhtankar (2013) estimate that in India 79% of labour expenditures have been lost from the ‘wages on the National Rural Employee Guarantee Scheme’. According to Gorodnichenko and Peter (2007), bribes received by public sector employees in Ukraine amount to 1% of GDP. Javaid (2010) estimated that, as a result of corruption in developmental projects, particularly in public sector procurement, Pakistan loses PKR200 billion (USD2.35 billion; at 2010 exchange rate) to the economy every year. Unfortunately, this is just the tip of the iceberg in the context of a ‘corruption pandemic’, particularly in developing countries, and implies a huge cost to the global economy.6

Table 2.1 Economic cost of corruption

Mo (2001) used data covering a large sample of 54 countries over the period 1960 to 1985 and found that for every 1% increase in corruption level, economic growth declines by 0.72%. Although his findings suggest that corruption impacts the level of human capital and the share of private investment, the most important channel through which corruption affects economic growth is political instability, which can account for around 53% of the total effect. Hopkins and Rodriguez-Pose (2007) differentiates between government interventions which could be linked to corruption or non-corruption. They found that in countries where private business activities are lightly regulated, a high level of public spending is linked to low levels of corruption.

Mauro et al. (2019) observed that natural resources (especially oil and mining), state-owned enterprises (energy, utilities and transportation) and public sector spending on purchases of goods and services by the government (especially during crises, as stated above in this section) are the main sectors bogged down by corrupt practices. The latest Investment Dispute Settlement Navigator report by UNTAD shows that 343 of the total 1257 long-term investment disputes are still pending and date back to projects initiated in 1981 UNCTAD (2022).7 Their estimates suggest that public sector procurement accounts for 13 % and 36% of GDP in OECD countries and advanced economies, respectively. They also claim that corrupt practices are relatively difficult in the education and health sectors. This is perhaps the reason why more corrupt countries have a small proportion of GDP spent on education and health; Fig. 2.1 shows that spending on education and health is lower in relatively more corrupt countries. Mauro et al. (2019) show that public spending (as a percentage of total spending) on health and education, on average, in low-income countries is about 10% less in high corruption countries as compared to low corruption countries. The gap is about 4% in emerging market economies and only 2% in advanced economies.8 UNDP (2022) also reported that more than half of the corruption cases are related to public spending.

3 Review of the Empirical Literature on Corruption

Corruption is a major problem in many developing countries. During the 1980s and 1990s,9 some Asian countries implemented a series of liberalisation and economic reform policies. These policies helped to improve trade share and attract foreign investment, thus achieving high growth targets. This inspired many other developing countries to follow suit. However, negotiations on foreign investment agreements also gave incentives for kickbacks and illegal commissions.10 Such corrupt practices not only increase the cost of doing business but also discourage genuine investors.11 Many experts have focussed on investigating how corruption impacts economic growth. Although most of the published work and commentaries support the view that corruption hampers economics, Ang (2020) argues that this view is ‘over-simplistic’.12 Interest in this area of research increased after 1995 when Transparency International started publishing rankings and absolute values of corruption perception for each country. Although there is a rich empirical literature on the issue of how corruption impacts growth, this review is restricted to the corruption–investment relationship which is the main focus of this chapter.

3.1 The Corruption and Investment Debate

The theoretical literature differentiates between the positive and negative effects of corruption on investment. Accordingly, scholars have provided contrasting views on corruption. The view of the ‘grabbing hand’ refers to corrupt practices which create uncertainty and substantially increase the cost of foreign funds, thus negatively affecting investment flows into a host country. The quality of the institutional environment is considered one of the main determinants of corruption. A weak institutional environment and regulatory structure provide incentives for more corrupt practices, or as it is termed, the ‘grabbing hand’. Conversely, a ‘helping hand’, a bribing mechanism, helps to facilitate transactions and procedures to set up and start businesses, thus encouraging foreign investment.13

An interesting theoretical paper by Abosti (2016) observed that corruption is inevitable in developing countries. Using firm optimization theory, he claims corruption has a positive impact on FDI at a high level of institutional quality and a negative impact with a low level of institutional quality. He refers to this threshold as a ‘corruption tolerable level of investment’.14

The empirical literature is inconclusive on whether corrupt practices decrease or increase foreign investment. Most scholars studying the effects of corruption on FDI support the ‘grabbing hand’ view of the phenomenon.15 Some scholars support the ‘helping hand’ view,16 though these empirical findings have been made by only a few and are inconclusive.17

Mauro (1995) is perhaps one of the first studies investigating the impact of corruption on investment using survey data. Using a sample of 67 countries, he finds that corruption impacts negatively on the ratio of investment to GDP. Campos et al. (1999) argues that both the level of corruption and the nature of corruption are important in determining the impact of corruption on investment. Corrupt regimes that are more predictable have less negative impact on investment than those that are less predictable. Wei (2000) investigated the effect of taxation and corruption on FDI from 14 source countries to 45 host countries. His empirical findings support the view that an increase in corruption reduced inward FDI. Vinod (2003) used data for 14 Asian countries and found that FDI is low in corrupt countries. Zhao et al. (2003) used panel data for 40 countries over a seven-year period. The results suggest that high corruption and low transparency negatively impact the inflow of FDI to host countries. Felipe and Travares (2004) used a broad cross-section of countries over the period 1970 to 1994 and found that a lower corruption level is associated with a high level of FDI.

Freckleton et al. (2012) found a similar result using a panel of 42 developing and 28 developed countries (1998–2008), namely that a low level of corruption leads to a positive impact of FDI on growth. Ayadi et al. (2014) investigated the relationship between the degree of transparency and the level of FDI inflows for 13 sub-Saharan African countries (1998–2008) and found a positive association between the two. Alemu (2012) used a sample of Asian economies and found that corruption hinders inflow of FDI in those countries. More specifically he found that a decrease in the level of corruption by 1% raises the inward FDI by 9.1%. Yahyaoui (2023) used data from a sample of African economies for the period 1996 to 2016 and determined that corruption mitigates the effect of FDI on economic growth. Azam and Ahmad (2013) found that the level of FDI inflow is influenced by the level of corruption, market size and inflation. Hakimi and Hamdi (2017) investigated the effects of corruption on investment and growth in 15 MENA countries. Using data over the period 1985 to 2013 and employing the panel vector error correction model (PVECM) they found that corruption negatively affects the inflow of FDI, thereby impacting growth in the sample countries.

Gründler and Potrafke (2019) used data for 175 countries over the period 2012–2018 and found that ‘the cumulative long-run effect of corruption on growth is that real per capita GDP decreased by around 17% when the reversed CPI increased by one standard deviation’.

In a very interesting paper, Ledyaeva et al. (2013) used a large firm-level panel dataset over the period 1996–2007 to investigate the level of corruption against the type of political regime in the country of origin of a foreign investor. They found that ‘foreign investors from less corrupt and more democratic countries tend to invest in less corrupt and more democratic countries while foreign investors from more corrupt and non-democratic countries tend to invest in more corrupt and less democratic countries’.

Conversely, Egger and Winner (2005) used a large panel of 73 developed and less-developed countries over the period 1995–1999 and found a positive impact of corruption on growth both in the short term and long term, thus supporting the ‘helping hand’ view. Quazi et al. (2014) also found that corruption facilitates FDI inflows in Africa.

Zheng and Xiao (2020) used a principal-agent model to examine ‘the conditions under which corruption prompts investment’. They investigate three policies which can be used to control corruption: strengthening monitoring, increasing compensation and enhancing accountability. The theoretical model suggests that strengthening monitoring could help in mitigating corruption but at the cost of reduced investment. This theoretical result is further confirmed in the empirical investigation of this paper where the authors found a negative correlation between infrastructure investment and anti-corruption efforts.

Bayar and Alakbarov (2016) also found similar results suggesting that control of corruption and the rule of law had no significant impact in attracting FDI for 23 emerging market economies (2002–2014). Belloumi and Alshehry (2021) used a panel of GCC countries over the period 2003–2016 and found corruption to be neutral regarding FDI inflows for GCC countries. More specifically they did not find a significant relationship between corruption and FDI inflows. However, this study finds corruption to have a positive impact on domestic investment. The authors argue that ‘bribery acts can help in overcoming the administration bureaucracy and inefficient regulations’.

Petrou and Thanos (2014) suggest that perhaps the relationship is non-linear, which could accommodate both sides of the theoretical arguments. Interestingly, in using firm-level data for 131 banks in 40 host countries, the empirical findings of Petrou and Thanos (2014) support a U-shaped relationship, implying that at low to moderate levels of corruption, a grabbing hand view is supported, while at a high level of corruption the helping hand view is supported. Hopkin and Rodriguez-Pose (2007) found that financial development followed by an increase in corruption tends to reduce FDI inflows.

3.2 What Do the Data Tell Us?

To further understand the dynamics between the ‘grabbing hand’ and ‘helping hand’ views in the context of growth-corruption and investment-corruption scenarios, I will present some data and use empirical analysis. For both analyses, data are sourced from the corruption perception index (CPI) published by Transparency International (2022) over the period 2000 to 2022.18 The index provides a ranking and a country specific score. A country’s score is the perceived level of public sector corruption on a scale of 0 to 100, where 0 means highly corrupt and 100 means very clean.19 Following Petrou and Thanos (2014), the CPI scores used in this chapter are reversed to be more intuitive. For ease of interpretation, these scores are converted into ascending order by taking a difference from 100. In this way, a clean country gets a 0 score and a highly corrupt country gets a 100 score. This way an increase in score is an indication of increasing corruption and vice versa. This helps us to directly compare the CPI score with increase in growth and FDI.

To provide a comparison between low and high corrupt countries I split the entire group into the top 20 least corrupt countries and the bottom 20 most corrupt countries.20 This section investigates how corruption impacts growth and FDI between the two groups. Figure 2.1 shows that the top 20 countries on average experienced a somewhat downward trend in corruption (which means an increase in CPI score) over the sample period (Fig. 2.1a).21 However, improvement in corruption (a decline in CPI score) is significantly visible in the bottom 20 countries (Fig. 2.1b).22 This trend becomes clearer when the CPI for the top 5 and bottom 5 countries is plotted (see Fig. 2.2a and b). For developing countries, this trend could be explained by measures adopted by these countries to combat corruption.

Fig. 2.1
Two multi-line graphs of C P I and G D P growth versus years from 2000 to 2020. The curves are represented c p i stop 20, G D P P g r top 20, and F D I top 20. C p I stop 20 denotes a high (2020, 22) and (2000, 78) respectively. The values are approximate.

Trend of CPI, GDP growth and FDI/GDP. a Countries with top 20 CPI (least corrupt) b Countries with bottom 20 CPI (most corrupt)

Fig. 2.2
2 multi-line graphs of the top 5 and bottom 5 corrupted trend of C P I versus years from 2000 to 2020. a. It depicts c p i s Finland, New Zealand, Denmark, Sweden, and Singapore. b. It depicts c p i s Cameron, Nigeria, Kenya, Zimbabwe, and Uzbekistan. Kenya denotes a low at (2020, 60). The values are approximate.

Trend of CPI over time (2000–2022). a Top 5 (least corrupt) b Bottom 5 (most corrupt)

Finally, I use scatter plots to further explore these relationships. Figure 2.3 shows a direct relationship between corruption and growth. A downward trend is evident for the top 20 countries (Fig. 2.3a) which implies that GDP growth improves as CPI moves downward (towards less corruption). One would expect a similar trend for the bottom 20 countries. However, the trend line is upward sloping, suggesting that higher corruption leads to higher economic growth (see Fig. 2.3b). This perhaps supports the ‘helping hand’ view. A similar picture emerges in Fig. 2.4 through a trend line between CPI and FDI (FDI/GDP). Figure 2.4a shows a downward trend indicating a positive relationship between less corruption and high growth. However, Fig. 2.4b has an upward slope, suggesting that corruption helps to attract more FDI which, again, supports the ‘helping hand’ view.23

Fig. 2.3
2 scatterplots of c p i s top 20 and c p i i s bottom 20 versus G D P P g r top 20 and G D P P g r bottom 20. a. Two curves depict c p i s top 20 denote a high at (negative 3, 20.5), and fitted values. b. The curves depict c p i s bottom 20 and fitted values high at (6, 75.5). The values are approximate.

Scatter plot CPI versus GDP growth a Countries with top 20 CPI (least corrupt) b Countries with bottom 20 CPI (most corrupt)

Fig. 2.4
2 scatterplots of c p i s top 20 and c p i i s bottom 20 versus F D I top 20 and F D I bottom 20. a. 2 curves depict c p i s top 20 denote a high at (2, 20.5), and fitted values low (14, 14.5). b. The curves depict c p i s bottom 20 and fitted values high at (5.5, 76.5). The values are approximate.

Scatter plot CPI versus FDI a Countries with top 20 CPI (least corrupt) b Countries with bottom 20 CPI (most corrupt)

Finally, the analysis attempts to verify the observation made by Mauro et al. (2019).24 For this plot the data on average government spending on education for the top 20 and bottom 20 countries over three periods (2000, 2010 and 2021) is presented in Fig. 2.5. Surprisingly, for the top 20 (least corrupt) countries, public spending on education shows an increasing trend from the years 2000 to 2010. However, the same declined significantly in the year 2021. A very similar trend is observed in the bottom 20 (most corrupt) countries where public spending on education increased between the years 2000 and 2010 but then declined. The most striking observation is that during the year 2021, the top 20 countries’ allocation on education (as a percentage of GDP) was even less than the bottom 20 countries. As discussed in Sect. 2.2, this significant decline was initially thought to be due to the fact that during COVID-19 more resources were diverted towards governmental emergency support programmes for small businesses as well as towards relatively poor segments of society who suffered the most during the pandemic. For this reason, the sample data cease beyond 2019 to isolate the effects of COVID-19 (see Fig. 2.A1). Interestingly the trend was still the same, though the decline is not as significant as observed in 2021. Also note the expenditure on education in 2000 and 2019 is almost the same for the top 20 countries, although it has increased for the bottom 20 countries for the same period. This evidence supports the view of diverted spending on priority areas during the COVID-19 pandemic. Figure 2.A1 also confirms a decline in spending on education in 2021 (as compared to 2019). It also confirms that the spending on education in 2019 was higher for the top 20 relative to bottom 20 countries.

Fig. 2.5
A bar graphs depict the government expenditure. The two sets represent the top 20 and bottom 20 countries. 2010 denotes a high 6.5 and 4.4 respectively. 2021 is low at 2.5 and 2000 is low at 2.8 respectively. The values are approximate.

Government expenditure on education (% of GDP)

Next, the analysis focuses on public expenditure on health. This comparison is presented in Fig. 2.6. The figure demonstrates that the picture is significantly different between the two groups. There is evidence of reasonably high and accelerated spending on health in the top 20 (least corrupt) countries. The figure also shows that public spending on health increased from over five percent of GDP to above six percent between the years 2000 and 2021. However, the evidence suggests that the bottom 20 (most corrupt) countries do not spend much on health and the growth is also extremely slow. Public spending on health in the most corrupt countries rose from around 1.5% of GDP in the year 2000 to just below two percent in the year 2021.

Fig. 2.6
A bar graph titled Government Expenditure on Health % of G D P. The two sets represent the top 20 and bottom 20 countries. 2021 denotes a high of 6.1 and 2000 low at 1. 2021 is high at 1.5 and 2000 is low at 0.5. The values are approximate.

Government expenditure on health (% of GDP)

3.3 Empirical Analysis

To gain more insight into the corruption–growth–FDI relationship and verify the ‘grabbing hand’ versus the ‘helping hand’ view, I will perform some econometric tests. To do so I use the following two specifications; one for economic growth and another for FDI. The underlying model with fixed effects takes the form:

$${{\text{g}}}_{it}={\upbeta }_{0}+{\upbeta }_{1}{{\text{CPI}}}_{it}+{\upbeta }_{2}{{\text{FDI}}}_{it}+{\mathrm{\gamma Controls}}_{it}+{\upeta }_{{\text{i}}}+{\uptheta }_{t}+{\upvarepsilon }_{it}$$
(2.1)
$${{\text{FDI}}}_{it}={\upbeta }_{0}+{\upbeta }_{1}{{\text{CPI}}}_{it}+{\upbeta }_{2}{{\text{g}}}_{it}\left(-1\right)+{\mathrm{\gamma Controls}}_{it}+{\upeta }_{{\text{i}}}+{\uptheta }_{t}+{\upvarepsilon }_{it}$$
(2.2)

Equation (2.1) is to see the impact of corruption (CPI) on economic growth (g). In Eq. (2.1), git is economic growth measured by the growth of the real GDP per capita in country i at time t. CPI is the value taken from the corruption perception index. As stated above, the converted CPI scores range from 0 to 100, where 0 means clean or no corruption and 100 means the highest level of corruption.25 FDI is foreign direct investment and is measured as a ratio of FDI to GDP. I also use a number of control variables. These include Gov_exp (government expenditure to GDP ratio), GFCF (a proxy for domestic investment which is measured as a ratio of gross fixed capital formation to GDP) and INF (to measure inflation). Finally ηi is a country-specific fixed effect, θt is a time effect, and εit is a multivariate normally distributed random disturbance. Equation (2.2) is to see the impact of corruption on foreign direct investment. Here, one period lag of growth is used in the specification. A fixed effects model, rather than a random effects model, is estimated as the ηi’s are likely to represent omitted country-specific characteristics which are correlated with other explanatory variables.

Data on GDP growth, FDI and all control variables are taken from the World Bank World Economic Indicators. Data on the corruption perception index is taken from the Transparency International website.

For empirical estimation I use two variants of the model. The first is as stated in Eqs. (2.1) and (2.2). I also add an interaction term in Eq. (2.1) to further explore the joint effect of CPI and FDI on growth. Hence the equations are modified as follows:

$${{\text{g}}}_{it}={\upbeta }_{0}+{\upbeta }_{1}{\text{CPI}}{}_{it}+{\upbeta }_{2}{{\text{FDI}}}_{it}+{\upbeta }_{3}{\text{CPI}}.{{\text{FDI}}}_{it}+{\mathrm{\gamma Controls}}_{it}+{\upeta }_{{\text{i}}}+{\uptheta }_{t}+{\upvarepsilon }_{it}$$
(2.3)

All three equations are estimated using three samples: sample 1 combines data on the top 20 and bottom 20 countries, sample 2 uses the top 20 countries and sample 3 only uses the bottom 20 countries.

Following the theoretical literature and our discussion in Sect. 2.3, I expect a positive relationship to exist between CPI and growth which would indicate that an improvement (decrease) in corruption would yield higher growth. I also expect higher FDI to enhance growth. Based on the empirical evidence, Gov exp is expected to negatively impact growth. INF is also expected to have a negative impact on growth. As for the FDI equation, I expect that improvement (decrease) in corruption would help to attract more foreign direct investment, hence a positive value for the relevant parameter. However, the result would depend on whether the ‘helping hand’ view is stronger than the ‘grabbing hand’. Further, the interaction term is expected to take a positive or negative value depending on the above view as well.

Finally, I investigate if the corruption–growth and corruption–FDI relationships are linear or non-linear. To do so I add the squared term of CPI in the model. Accordingly, the two equations take the following form.

$${{\text{g}}}_{it}={\upbeta }_{0}+{\upbeta }_{1}{{\text{cpi}}}_{it}+{\upbeta }_{2}{{\text{cpi}}}_{it}^{2}+{\upbeta }_{3}{{\text{FDI}}}_{it}+{\mathrm{\gamma Controls}}_{it}+{\upeta }_{{\text{i}}}+{\uptheta }_{t}+{\upvarepsilon }_{it}$$
(2.4)
$${{\text{FDI}}}_{it}={\upbeta }_{0}+{\upbeta }_{1}{{\text{cpi}}}_{it}+{\upbeta }_{2}{{\text{cpi}}}_{it}^{2}+{\upbeta }_{3}{{\text{g}}}_{it}\left(-1\right)+{\mathrm{\gamma Controls}}_{it}+{\upeta }_{{\text{i}}}+{\uptheta }_{t}+{\upvarepsilon }_{it}$$
(2.5)

4 Discussion of the Results

Here I discuss the results of the estimated model.

4.1 The Corruption–Growth–FDI Relationship

Table 2.2 reports the results of the combined sample (the combined top 20 and bottom 20). The empirical evidence strongly supports the view that more corruption enhances economic growth (model 1(a)). This result is consistent with the ‘helping hand’ view which claims that corruption facilitates economic activities and could lead to higher growth. The results also suggest a strong positive impact of FDI inflows on economic growth. Both Gov_exp and INF exert a negative (weak significance) impact on growth while the impact of gross investment (GCF) is strong. The ‘helping hand’ view is further confirmed when an interactive term is added in the regression model (model 2(a)). Now we find that CPI still has a positive impact on growth, although it loses its significance. Further, the interaction term emerges as a strong positive value indicating that an increase in corruption enhances FDI thus leading to higher growth. Other variables show a similar pattern as per model 1(a), except that GFCF is not significant anymore.

Table 2.2 Fixed effect estimation: combined sample

Perhaps the above discussion should be taken with caution. The results represent a sample which combines the two extreme and opposing ends of corruption—the top 20 and the bottom 20 countries. To further evaluate these results, let’s move to the next model which uses the sub-sample of the top 20 (least corrupt) countries. The results in Table 2.3 (model 1(c)) support a ‘grabbing hand’ view where the parameter for CPI is negative (though not significant) while FDI has a positive and significant impact on growth. The inclusion of the interactive term does not change the sign or significance of the relationship between corruption and growth (model 2(c)).

Table 2.3 Fixed effect estimation: Top 20 CPI (least corrupt countries)

Finally, I use the sub-sample of the bottom 20 (most corrupt) countries. The results (see Table 2.4) are very similar to the whole sample. CPI again has a strong positive impact on growth, indicating that more corruption leads to high growth (model 1(e)). FDI is positive but no longer significant, while other variables have a similar relationship as discussed above. Inclusion of the interactive term gives the most surprising results (model 2(e)). Now both CPI and FDI have a strong positive relationship with growth while the interactive term has a negative and weakly significant impact. This, perhaps, suggests that corruption has a negative impact on FDI though this has subsided with the direct positive impact on growth.

Table 2.4 Fixed Effect Estimation: Bottom 20 CPI (Most Corrupt Countries)

4.2 Digging Deeper: Is the Corruption–Growth–FDI Relationship Non-linear?

Given these surprising results, I will further explore whether the corruption–growth relationship is in fact non-linear. This can be verified by estimating Eqs. (2.4) and (2.5). The results of the whole sample can be seen in 1(b) of Table 2.2. The results clearly confirm a non-linear relationship. The linear coefficient of CPI maintains a negative (but not significant) sign, whereas the non-linear (squared term of cpi) is positive and strongly significant. This means that as corruption changes from low to moderate levels, growth falls to its minimum; but when corruption changes from moderate to high levels, growth accelerates.

These findings are consistent with Petrou and Thanos (2014) who also found a U-shaped relationship between corruption and investment. These findings support a ‘grabbing hand’ view26 at the low to moderate levels of corruption and a ‘helping hand’ view27 at high levels of corruption. The empirical results did not provide any such evidence when a sample of the top 20 (least corrupt) countries was used (see Table 2.3, model 1(d)). In fact, the results are inconclusive as parameters of both the linear and non-linear terms are statistically insignificant.28 Testing the bottom 20 (most corrupt) countries provides the most interesting results (see Table 2.4, model 1(f)). This results in the linear coefficient of CPI being positive and significant while the squared term is negative and significant. This means that at low to moderate levels of corruption, growth increases and reaches a maximum level; but at high levels of corruption, growth falls. This is evidence of an inverted U-shaped relationship between corruption and growth where a ‘helping hand’ view is supported at the low to moderate levels of corruption while a ‘grabbing hand’ view is supported at the high levels of corruption. This result is opposite to the findings of Petrou and Thanos (2014). This is perhaps due to the nature of the weak regulatory structure, bureaucratic red tape and lack of a formal infrastructure in developing countries. Bribery and corruption help to bypass such constraints to attract FDI leading to high growth. However, over time, countries develop better infrastructure, regulatory controls take place and governance structure is improved, leading to a decline in corrupt practices but higher growth.

4.3 The Corruption–FDI Relationship

This section discusses the effect of corruption on foreign direct investment. The results using a combined sample are presented in Table 2.2, model 3(a) and appear to be less strong as no variable other than GFCF is statistically significant. Nevertheless, there is a negative relationship between CPI and FDI which suggests that an improvement in corruption (less corruption) helps to attract more FDI in the mixed sample, a view supported by the proponents of the ‘grabbing hand’. Similar to the growth equation, the results also support a U-shaped relationship between corruption and FDI (see Table 2.2, model 3(b)). This result is further strengthened when a sample of top 20 countries is used. Now the CPI shows a negative and statistically significant relationship with FDI (Table 2.3, model 3(c)). There is no evidence of a non-linear relationship in this case. Interestingly the results for the bottom 20 countries are consistent with the previous findings of supporting a ‘helping hand’ view. Now the CPI has a positive and significant relationship with FDI suggesting that corruption helps to facilitate FDI (Table 2.4, model 3(e)). The results in Table 2.4, model 3(f) do not support a non-linear relationship between corruption and FDI for the most corrupt countries.29

Finally, the analysis focuses on a sample of Asian countries.30 For analytical purposes, a randomly selected sample of 33 Asian countries was used to continue the above analysis.31 These results are reported in Table 2.5.32 The results of the linear model support a ‘helping hand’ view for the corruption–FDI relationship. However, the results did not find a non-linear relationship between CPI and FDI. Although the parameter for CPI has a positive sign while the CPI squared term has a negative sign, in both cases these parameters are statistically insignificant.

Table 2.5 Fixed Effect Estimation: Sample of Asian Countries

5 Conclusions

Corruption has been perceived as a major problem impacting the growth and investment in a country, especially developing countries. The empirical literature is divided on the impact of corruption. Some scholars argue that corrupt practices serve as a helping hand by removing the obstacles towards investment while others oppose it. There are also a few papers arguing about the nature of the corruption–growth and corruption–FDI relationships, namely whether they are linear or non-linear. This chapter has revisited these issues in the context of the least and the most corrupt countries. First, I provided an in-depth literature review on this issue. I then discussed in detail the cost of corruption supplemented with some actual numbers from different countries. Next, I performed some econometric tests to verify the relationship between corruption, growth and FDI. For an empirical investigation, I used panel data for the top 20 least corrupt and the bottom 20 most corrupt countries over the period 1995–2022 and applied the fixed-effects estimation technique. I also expanded the analysis by taking a sample of 33 Asian countries.

The empirical findings of this chapter support a ‘grabbing hand’ view for the least corrupt countries while a ‘helping hand’ view is supported for the most corrupt countries. I also found the relationship is non-linear, suggesting that for the most corrupt countries, corrupt practices do facilitate investment at the initial stage and may help to attract FDI. However, over time, when a better regulatory structure is developed, corrupt practices hurt growth and FDI. These are interesting results and consistent with similar findings by Egger and Winner (2005) and Quazi et al. (2014) who also found that ‘corruption is a stimulus for FDI’ thus supporting the ‘helping hand’ view. The results are, however, contrary to Petrou and Thanos (2014) who also support a U-shaped relationship but found that at low to moderate levels of corruption a ‘grabbing hand’ view is supported while at high level of corruption the ‘helping hand’ view is supported. The tests of a small sample of Asian countries are, in general, consistent with the above results. These are interesting findings and could lead to further research in this area.

However, these results should be taken with caution. First, as stated in note 19, this study used data for a combined sample over the period 2000–2022. A more appropriate way would be to use data starting in 2012. However, this would leave a shorter timespan to perform a meaningful analysis. Future research could focus on the CPI data starting in 2012. Second, the study used a sample of the 20 least corrupt and 20 most corrupt countries, and not all countries. Hence, these results may not be generalised. Third, since countries differ significantly in terms of their economic and political environment and regulatory structure, a country-specific study would shed better light on the impact of corruption. However, due to limited data availability on the corruption perception index, such estimations may suffer from a small sample bias at present and this may be better done when more data is available. Nevertheless, the findings of this chapter do add new evidence to the existing literature. Finally, I will make some recommendations for policymakers and respective governments:

  • There is a need to further improve the definition of corruption. The current measure of corruption (CPI) is more suitable for developed countries but not for the developing world, thus increasing the risk of under- or over-estimation of corruption scores.

  • Institutional development, more transparent policies and better communication among government departments and ministries are needed to combat corruption, especially in sectors with a high potential for corrupt practices such as procurement, privatisation and foreign direct investment.

  • Technology should be used in government transactions to keep track of them and reduce the incidence of corruption.

Notes

  1. 1.

    See Bardhan 1997 and Buchanan 1997 for more details.

  2. 2.

    For example, UNCAC Article 15 defines bribery as ‘[t]he promise, offering or giving, to a public official, directly or indirectly, of an undue advantage, for the official himself or herself or another person or entity, in order that the official act or refrain from acting in the exercise of his or her official duties’. Similarly, Article 17 states ‘that the embezzlement, misappropriation or other diversion by a public official for his or her benefit or for the benefit of another person or entity, of any property, public or private funds or securities or any other thing of value entrusted to the public official by virtue of his or her position’. Article 19 defines ‘abuse of function’, ‘when committed intentionally, [as] the abuse of functions or position, that is, the performance of or failure to perform an act, in violation of laws, by a public official in the discharge of his or her functions, for the purpose of obtaining an undue advantage for himself or herself or for another person or entity’ (UNODC 2004).

  3. 3.

    UNDP 2018.

  4. 4.

    Baker 2009. Also see Johnsøn and Nils (2015) for more details on the impact on foreign aid due to corruption.

  5. 5.

    UNODC 2022; Csonka and Salazar 2021; Gaspar et al. 2020.

  6. 6.

    Locatelli et al. (2017) used an institutional theory to study the impact of corruption in public sector projects and megaprojects. The study observed that ‘corruption worsens both cost and time performance, and the benefits delivered’.

  7. 7.

    Some examples of these pending claims indicating corrupt practices in awarding contracts and the execution of projects include oil and gas related investment (Caistor and Villarán, 2006; Werlin, 1994), land development projects (Czech Republic, 1997) and infrastructure development (Costa Rica, 1997; Pakistan 1997).

  8. 8.

    A study in Brazil found evidence that where federal transfers to local governments for education spending are partially lost to corruption, dropout rates are higher and test scores worse.

  9. 9.

    In the case of Thailand, reforms started even earlier, around the 1950s.

  10. 10.

    Lambsdorff (2003) found that law and order in a country is crucial in attracting capital.

  11. 11.

    See Lambsdorff (2006) for a detailed review of the empirical literature.

  12. 12.

    Ang (2020) compares the economic performance of China (among the most corrupt countries) with that of the United States (among the least corrupt countries) from 1995 to 2016 and shows how both superpowers managed to achieve a remarkable GDP of around USD11 trillion during the period. This obviously negates the common belief of a negative impact of corruption on growth. He believes that this is due to the mechanism of the construction of the commonly used corruption perception index (CPI) and provides further explanations to this debate.

  13. 13.

    Petrou and Thanos 2014.

  14. 14.

    Also see Shleifer and Vishny 1993 for more details on this argument.

  15. 15.

    Javorcik and Wei 2009; Uhlenbruck et al. 2006; Voyer and Beamish 2004; Wei 2000; Wei 1997; Mauro 1995.

  16. 16.

    Egger and Winner 2005; Wheeler and Mody 1992; Lui 1985.

  17. 17.

    Barassi and Zhou 2012; Helmy 2013; Al-Sadig 2009; Habib and Zurawicki, 2002; Nguyen and van Dijk 2012.

  18. 18.

    Transparency International started publishing CPI data in 1995. However, not many countries were listed in the report until 2000. Hence, I use 2000 as the starting point in our analyses.

  19. 19.

    From 1995 to 2011, Transparency International used an old methodology for measuring CPI using a scale from 0 to 10, where 0 represents very high levels of perceived corruption. I have converted the scale from the old data to the new by simply multiplying old values by 10. This may not be the ideal method, but perhaps the only way to do so. However, Transparency International suggests that due to a change in their methodology in 2012, results from before that year cannot be compared. Only CPI results from 2012 onwards can be compared (Transparency International, 2020). For empirical analysis, I have conducted some robustness checks, which is a point also raised in Chap. 3 (Jetin et al.). See note 29 for further details on empirical analysis.

  20. 20.

    The list is provided in the Appendix in Tables 2.A1 and 2.A2.

  21. 21.

    This can also be verified by Table 2.A1 where the top 15 of the top 20 least corrupt countries in 2000 experienced an increase in their CPI (converted) scores in 2022, indicating a gradual increase in corruption. In the case of Canada and Iceland, according to CPI scores, corruption seems to have increased significantly.

  22. 22.

    This is consistent with the numbers reported in Table 1b. The numbers suggest that CPI improved (a decline in corruption) between 2000 and 2022 in all the bottom 20 (most corrupt) countries. The only exceptions are Zimbabwe and Venezuela, where corruption has increased significantly.

  23. 23.

    Ang (2020) found a similar trend using a cross-section of selected countries in 2016 (see Fig. 1.1, p. 3).

  24. 24.

    As stated in Sect. 2.2, Mouo et al. (2019) observed that public sector spending is one of the main sectors for attracting corruption.

  25. 25.

    This helps us to interpret the CPI parameter with a positive (increase in corruption) or a negative sign (decline in corruption).

  26. 26.

    Javorcik and Wei 2009; Uhlenbruck et al. 2006; Voyer and Beamish 2004; Wei 2000; Wei 1997; Mauro 1995.

  27. 27.

    Egger and Winner 2005; Wheeler and Mody 1992; Lui 1985.

  28. 28.

    Barassi and Zhou 2012; Helmy 2013; Al-Sadig 2009; Habib and Zurawicki 2002; Nguyen and van Dijk 2012.

  29. 29.

    As stated in note 19, Transparency International suggests that due to a change of methodology, ‘CPI scores before 2012 are not comparable over time’. However, this leaves us with a shorter timespan to do a meaningful analysis. To avoid this problem, I have used data from 2000. To check the robustness of the results, I then replicated the same analysis using data from 2012. The results, in general, are similar to the combined dataset. We, however, observe that due to lack of sufficient data, some variables are not statistically significant, though still having the same signs. The only notable difference is model 3 for the bottom 20 countries where the non-linear model is now statistically significant and suggests a U-shaped relationship. This will be further investigated in a separate paper.

  30. 30.

    The UN classification of ‘Asian region’ is used. According to the UN regional classification and definition, regions included under ‘Asia’ are: Southern Asia, West Asia, South-Eastern Asia, Eastern Asia, Central Asia, Other Asia and Other Non-Commonwealth (United Nations n.d.).

  31. 31.

    Initially, the sample consisted of 48 counties. A few countries were removed for various reasons, leaving a sample of 33 countries over the period 2000–2022. See the Appendix for a list of sample countries.

  32. 32.

    I also have added population as another control variable.