Is the Book Judged by Its Cover? Unveiling the Impact of Corruption on Foreign Direct Investment in the PALOP Economies

Abstract

This paper analyzes the impact of corruption on foreign direct investment (FDI) in the Portuguese-speaking African countries (PALOP) economies between 2006 and 2018. The focus lies on Angola, Cape Verde, Guinea-Bissau, and Mozambique since, according to Transparency International, they exhibit intermediate to low levels on the Corruption Perceptions Index. Despite sharing historical and cultural ties, as former Portuguese colonies, no research has focused on the impact of corruption on FDI in the PALOP economies, to the best of our knowledge. To accomplish this, we use an Instrumental Variables Fractional Probit Regression applied to data from the World Bank Enterprise Surveys, which gather information for 2180 firms. The results show that, on average, corruption does not significantly affect FDI in PALOP economies. Trade, credit, and firm size emerge as key FDI determinants, while investment levels and tax rates are not relevant. Corruption has negligible effects on FDI in manufacturing but boosts FDI in services. Interestingly, while corruption has no significant effect on FDI for small and medium firms, a positive, significant impact is revealed for large firms. Finally, corruption’s overall FDI impact is the same across PALOP countries, except in Angola, where it negatively influences FDI compared to Mozambique.

1. Introduction

With the unprecedented growth of globalization, major economies and multinational corporations have a clear interest in investing economically and financially in various countries, as shown by the flow of financial capital, the value of the stock of capital that the investing firms accumulate, and the flows of income from the investments (Lipsey, 2000). According to the Organization for Economic Cooperation and Development (OECD), foreign direct investment (FDI) is “a category of cross-border investment made by a resident in one economy with the objective of establishing a lasting interest in an enterprise that is resident in an economy other than that of the direct investor” (OECD, 2025, p. 17). It has been shown to be a pivotal element in international economic integration since it creates “stable and long-lasting economic links between economies” and it is key in the transfer of technology between countries, promotes international trade through access to foreign markets and can be an important vehicle for economic development, namely in Africa (OECD, 2025).

Nevertheless, when pondering whether to invest in a new market, investors can either be attracted or deterred by certain factors, one of them being a country’s level of corruption (Barassi & Zhou, 2012). According to Transparency International (2024), corruption refers to the “abuse of entrusted power for private gain”, as well as a deliberate strategy driven by self-interest, which stems from situations of asymmetrical power in which an economic agent holds authority over others (Svensson, 2006; Kolstad & Søreide, 2009).

In recent years, corruption in African countries has been rising (Yahyaoui, 2023) and, as stated by Transparency International (2024), most of them show stagnation and poor performance, with an unchanged regional average score of 33 (out of 100) on the Corruption Perceptions Index, a widely used composite indicator that captures perceived levels of corruption based on expert assessments and business surveys, where lower scores indicate higher perceived corruption. Despite their notorious corruption challenges, the evolution of Africa’s formerly underdeveloped markets makes it tougher for foreign investors to ignore them. In fact, the significant impact of FDI on enhancing development is becoming widely acknowledged, particularly in the context of the changing economic landscape of Africa: not only does it attract large amounts of capital, but it also introduces advanced technology, managerial expertise and provides access to global economies, thereby playing a crucial role in driving economic growth and integration into the international market (Dupasquier & Osakwe, 2006).

The primary aim of this analysis is to unveil whether corruption poses a challenge to FDI in the Portuguese-speaking African Countries (PALOP) economies, namely Angola, Cape Verde, Guinea-Bissau, and Mozambique. The focus lies specifically on these countries because, according to Transparency International, they exhibit intermediate to low levels on the Corruption Perceptions Index’s scale of 0 (highly corrupt) to 100 (highly clean) (Transparency International, 2023). Based on the Corruption Perceptions Index scores, Angola, Guinea-Bissau, and Mozambique typically record values between 20 and 30, whereas Cape Verde attains substantially higher scores around 60 (Transparency International, 2023). Additionally, these countries share historical and cultural ties as former Portuguese colonies; beyond language and cultural proximity, they also exhibit similar administrative practices, regulatory environments, and legal traditions that shape firms’ interactions with public authorities. In these settings, corruption tends to materialize through firm-level, discretionary encounters (such as licensing, taxation, customs, and regulatory compliance) rather than solely through macro-level institutional indicators. Prior research highlights that these types of direct firm-state interactions are particularly relevant for understanding how corruption affects business decisions and foreign ownership at the microeconomic level (Fisman & Svensson, 2007; Kaufmann & Wei, 2000; L. Martins et al., 2020). This makes the PALOP economies a suitable setting to analyze how firms’ direct exposure to corruption influences foreign ownership decisions, while limiting confounding variation arising from significantly different regulatory, cultural, or administrative systems. Yet, despite these factors, no research has specifically focused on the impact of corruption on foreign direct investment in these economies, to the best of our knowledge.

In addition, this context is especially interesting because it allows corruption to be examined as a firm-level constraint rather than as a purely country-level characteristic. The relative similarity of the business environment across PALOP countries, as well as the shared cultural, historic and social backgrounds these countries have as former Portuguese colonies, creates scope for identifying heterogeneous effects of corruption across firms, sectors, and sizes. These dimensions are often emphasized in the literature on the impact of corruption on FDI but rarely examined within a comparable setting (Cuervo-Cazurra, 2008), as many studies aggregate countries that differ greatly in their political, historical, social, and business backgrounds (e.g., L. Martins et al., 2020; Fisman & Svensson, 2007; Cerdeira & Fortuna, 2025).

This paper contributes to the literature in various ways, offering insights to academics, investors, and policymakers, and thereby enhancing the understanding of FDI dynamics in developing economies. Firstly, we focus the analysis on a group of countries largely ignored by the literature, to the best of our knowledge. In addition, while the literature on the impact of corruption on FDI generally uses country-level data, we resort to firm-level data, allowing for firm, as well as country, heterogeneity in the analysis. We employ data from the World Bank Enterprise Surveys (WBESs), which gathers microdata for firms between 2006 and 2018, with a total of 2180 observations, allowing for a more disaggregated analysis than with macro-level, country data. Furthermore, our methodology considers the fact that FDI is measured as a fractional variable, i.e., a variable that varies between zero and 1 (or 0% and 100%). While the literature generally uses OLS or other methods for continuous, unbounded variables as the estimation method, we depart from that and explicitly consider a framework appropriate for the FDI variable: the Instrumental Variables (IVs) Fractional Probit Regression, which also accommodates the existence of endogeneity between corruption and FDI. We also consider in our methodology that differences between manufacturing and services activities, as well as between small, medium, and large firms, may lead to different impacts of corruption on FDI. The results obtained with this procedure have relevant policy implications, as countries wishing to attract FDI ought to introduce measures that should consider the effect of corruption on foreign direct investment.

Henceforth, the paper is structured as follows: Section 2 reviews the relevant literature on corruption and FDI, highlighting the gaps the present study aims to fill. Section 3 includes a detailed explanation of the methodology used. Section 4 covers the results of the analysis and offers insights into the implications of our findings. Lastly, Section 5 concludes.

2. Literature Review

FDI determinants have been extensively studied in the economics literature due to the advantages it may offer to African countries. The impact that foreign direct investment has on economies is intricately influenced by a combination of factors such as economic policies, market size, political and macroeconomic stability, good infrastructure, an educated labor force, openness to trade and governance effectiveness (Asiedu, 2006; Dupasquier & Osakwe, 2006; Adinda, 2018; Sabir et al., 2019; R. Martins et al., 2021; Cerdeira & Fortuna, 2025).

At a macroeconomic level, Mijiyawa (2015) emphasizes the importance of return on investment and the size of host countries’ domestic markets: firstly, FDI flows are more likely to be routed to countries that can offer greater returns on investment because FDI activities should be economically and financially advantageous; secondly, nations that have a significantly large domestic market tend to capture more FDI, particularly if domestic market needs are satisfied. Boğa (2019) has highlighted other variables such as natural resources, GDP growth and domestic credit, as determinants of FDI inflows in sub-Saharan Africa. Asiedu’s (2006) research also underlines macroeconomic stability, evaluated by the standard inflation rate, infrastructure development, human capital, as well as openness to FDI, generally measured in terms of share of trade in GDP, as variables that policymakers can directly influence to attract FDI. In addition, Zghidi et al. (2016) consider that nations that encourage increasing freedom in economic activities tend to foster a significant presence of multinational corporations, thus intensifying FDI inflows.

Dupasquier and Osakwe (2006) have further concluded that foreign investors’ reluctance to invest is due to the high degree of uncertainty in the region, exposing firms to risks, while also introducing the concepts of lack of transparency, as well as political and economic uncertainty, arguing that their impact on FDI inflows into African economies depends on the type and sources of FDI. Thus, economic uncertainties emerge as a major obstacle when fused with political instability and the burden of external debt (Lemi & Asefa, 2003). In fact, Malikane and Chitambara (2017) refer to the importance of greater democratic institutions and political systems for absorbing FDI, with Knack and Keefer (1995) further corroborating that well-established institutions can help lessen the negative effects of corruption on FDI by promoting transparency, accountability, and legal certainty for foreign investors. Asiedu (2006) equally concludes that disorganized and weak institutions, characterized by corruption and inadequate contract enforcement, as well as political instability, provoked by coups, assassinations and revolutions, inhibit FDI.

At a microeconomic level, R. Martins et al. (2021) and Cerdeira and Fortuna (2025) concluded that trade, investment, credit, firm size and human capital can be defined as fundamental FDI determinants: corporations that prioritize higher levels of exports, enhance investment, and ensure credit lines tend to have increased levels of FDI. This further corroborates that larger enterprises have the capacity to attract more FDI when compared to smaller ones, and in terms of workforce, higher levels of human capital positively influence foreign investment. Nevertheless, the impact of these variables on FDI can differ, depending on each African country’s economic and political landscape, which explains the predominant influence of one variable relative to the others (R. Martins et al., 2021). Wei (2000) further argues that increasing tax rates on multinational corporations is associated with decreasing levels of FDI.

The relationship between corruption and foreign direct investment is a complex one, with effects seen in both developed and developing nations. In the context of Africa, studies indicate that corruption can be understood through three main channels (Amegavi et al., 2022): within bureaucratic systems, considering corruption thrives in environments that lack supervision and regulation; external controls, e.g., assurance of free press and an impartial judiciary which can help reduce corruption by exposing governmental wrongdoing and posing a risk of prosecution for corrupt officials (Zakharov, 2019; Debski et al., 2018; Lederman et al., 2005); lastly, culture and colonial history also indirectly impact corruption, since greater ethnic diversity may lead to disorganized corruption (Shleifer & Vishny, 1993). Thus, corruption stands out as an element known for its ability to either discourage or sometimes inadvertently attract FDI based on the nature and predictability of corrupt practices (Barassi & Zhou, 2012).

The complex nature of corruption serving as both an obstacle and an enabler of FDI highlights the intricacies involved in comprehending its effects on investments. Experts commonly consider that corruption poses a barrier to FDI by increasing expenses, raising risks, and amplifying uncertainties when investing in countries. The mainstream consensus is that corruption erodes the foundation upon which trust and predictability in international investments stand. Habib and Zurawicki (2002) offer compelling empirical support for this argument, illustrating how differing levels of corruption in home and host countries effectively act as a burden on foreign investors, which decreases the overall appeal of a country as an investment destination. This perspective aligns with Castro and Nunes (2013), with the results suggesting that in countries where corruption is lower, FDI inflows are greater, because of the increase in a favorable climate for investors. Also, Zhao et al. (2003) reveal that the presence of high corruption and low transparency significantly hinder the inflow of FDI to host countries.

Following this stream of thought, Kaufmann and Wei (2000) found that companies that pay more bribe money waste more time negotiating with bureaucrats, thus suffering a bigger intrusion in equilibrium. Simultaneously, Wei (2000) found that corruption—more broadly interpreted as “poor public governance”—has a significant and negative effect on FDI, and acts as a deterrent to investors, “by adding to the cost and unpredictability of doing business” (Wei, 2000). It is further argued that increasing corruption levels in multinational firms directly correlates to a reduction in inward FDI (Wei, 2000). Likewise, Hakkala et al. (2008) underline the negative effect of corruption on the probability of a firm investing. The arguments put forth suggest that corruption not only diminishes the attractiveness of a market but also directly impacts the bottom line of potential investment returns, leading to a reticence among foreign investors to commit capital, which is explained by the uncertainty regarding the costs of operation in the country (Kaufmann, 1997).

Adding to the discussion, by emphasizing the need to distinguish between the forms of corruption when evaluating its impact on FDI, Cuervo-Cazurra (2008) argues that it is not the level but rather the type of corruption that affects FDI. He revealed that although each type will have a negative influence on FDI, pervasive corruption, with its predictable nature, is a more significant deterrent than its arbitrary counterpart—corruption that is uncertain (Rodriguez et al., 2005). Furthermore, Freckleton et al. (2012) advanced that the economic environment is a key factor in determining how corruption impacts overseas investments, with the specificity that lower levels of corruption enhance the impact of FDI on economic growth.

While it is evident that an ample body of literature underlines the negative impacts of corruption on FDI, a substantial fraction argues otherwise. The counter-narrative that positions corruption as a facilitative element for foreign direct investment gains further traction when delving deeper into the complexities of global investment landscapes. Considering the ceaseless debate about whether corruption “greases” or “sands the wheels” of economic growth (Bardhan, 1997), the “greasing the wheels” perspective argues that in environments where regulatory and procedural clarity is lacking, and where bureaucratic inefficiencies create significant barriers to market entry and operations, corruption can paradoxically become a tool for efficiency. Supporting the latter hypothesis, research by Egger and Winner (2005) and Barassi and Zhou (2012) confirmed earlier studies by Leff (1964) and Huntington (1968), concluding that in the presence of regulations and other administrative controls, corruption can act as a “helping hand” to foster FDI. Leff (1964) argued that “corruption may introduce an element of competition into what is otherwise a comfortably monopolistic industry, and payment of the highest bribes becomes one of the principal criteria for allocation” (see Kaufmann, 1997, pp. 114–115). Adding to this notion, Bardhan (1997) further indicates that even though corruption can increase uncertainty amongst investors, “in the presence of rigid and inefficient bureaucracies, the informal payments to bypass red tape can improve overall efficiency”.

All things considered, Africa’s heterogeneous economic landscapes offer a perspective from which to examine how corruption impacts FDI, with the results undoubtedly varying among different sets of countries (Jalil et al., 2016). Through Westerlund (2007) and ARDL panel cointegration tests, Jalil et al. (2016) determined that corruption seems to have a positive long-run effect on FDI in Africa, because bribes alleviate the impact of institutional inefficiency, supporting previous research conducted by Barassi and Zhou (2012). Conversely, studies also show that corruption negative effects on FDI can pose a serious impediment to economic growth in the African economies, acting as the “grabbing hand”, since investors are unwilling to invest in countries where corruption levels are high (Yahyaoui, 2023), and where corruption is perceived as a tax on investors’ income, deterring them from investing in that territory (Zallé & Ouédraogo, 2021). However, as shown by Bouchoucha and Benammou (2018), with results supporting the work of Egger and Winner (2005), the control of corruption positively affects FDI inflows in Africa.

Although the literature appears to be divided into two groups, one supporting the “helping hand” and the other sustaining the “grabbing hand” perspectives, there seems to be a third group that meets halfway. Results suggest that the first can become the latter when weak governance and incapable institutions become competent and efficient. In this case, the link between corruption and FDI in Africa would emulate the example of other countries where corruption has negative effects (Quazi et al., 2014). In his study with a panel of 30 Sub-Saharan African (SSA) countries, including Angola, Mozambique and Guinea-Bissau, Gossel (2018) concluded that corruption serves as a “helping hand” for FDI to overcome feeble institutional capabilities, but as democratic capital accumulates, it becomes an obstacle, i.e., “sands the wheels” for foreign investment. Democracy has been further proven to increase net inflows of FDI in SSA, while corruption averts it, not only in the host country but also in the neighboring ones (Zallé & Ouédraogo, 2021).

As can be seen, a considerable number of empirical studies explore the relationship between corruption and FDI. Nonetheless, there seems to be a gap in the literature, a blank space that has yet to be filled, since, to the best of our knowledge, no study has been conducted vis-à-vis the impact of corruption on foreign direct investment in the PALOP economies. Despite focusing their research on the link between PALOP’s corruption levels and the entry modes of Portuguese multinational companies, Grande and Teixeira (2012) equally noted that lower levels of corruption are more prone to attract FDI, therefore concluding that perceived corruption does hinder foreign investment in the PALOP countries.

Thus, the literature review suggests that the impact of corruption on FDI remains theoretically and empirically ambiguous. Several studies conclude that corruption discourages FDI by increasing transaction costs, uncertainty, and operational risks for foreign investors (Wei, 2000; Asiedu, 2006). In this context, corruption acts as an implicit tax on investment and reduces the attractiveness of host countries. Accordingly, our first hypothesis reflects this dominant view in the literature:

H1. 

Corruption has a negative impact on foreign direct investment.

Nevertheless, the literature also suggests that the impact of corruption on FDI may depend on contextual and firm-specific factors. In environments characterized by bureaucratic inefficiencies and regulatory complexity, corruption may allow firms to circumvent administrative barriers, potentially facilitating investment (Leff, 1964; Egger & Winner, 2005; Barassi & Zhou, 2012). Also, the effect of corruption is unlikely to be the same across sectors. For instance, services typically rely more on licenses, permits, and direct interaction with public authorities than manufacturing activities. As a result, the effect of corruption on FDI may differ systematically across these sectors. We then consider the following hypothesis:

H2. 

The impact of corruption on foreign direct investment is different between manufacturing and services sectors.

As discussed above, firm size may also be a relevant mediator of the impact of corruption on FDI. Larger firms typically possess greater financial resources, bargaining power, and experience in dealing with complex regulatory environments, which may allow them to better absorb or manage corruption-related costs (Hakkala et al., 2008; L. Martins et al., 2020). In contrast, small- and medium-sized firms may be more vulnerable to corruption, facing higher relative costs and greater uncertainty. Thus, we consider that:

H3. 

The impact of corruption on foreign direct investment is different for small and medium firms and for large firms.

These hypotheses reflect the heterogeneous mechanisms emphasized in the literature on the impact of corruption on FDI and motivate the empirical strategy adopted in this study, which explicitly allows the impact of corruption to be different across sectors and firm sizes.

3. Data and Methodology

This study uses firm-level data from the World Bank Enterprise Surveys, which gathers microdata for firms. The sample of firms in this research focuses on the PALOP economies, namely Angola, Cape Verde, Guinea-Bissau, and Mozambique. The data analyzed span the period between 2006 and 2018. This time frame is because the database solely contains data spanning these years, more specifically, 2006, 2007, 2009, 2010, and 2018. While data are available for several years, it is not possible to track the same firm over time; i.e., the database consists of repeated cross-sections rather than panel data. Each firm is included in one of the following 11 sectors, comprising manufacturing and services activities: food; furniture; garments; hospitality and tourism; manufacturing; metals, machinery, computers and electronics; other manufacturing; retail; services; other services; and other sectors.

The database comprises a total of 2180 observations, as depicted in

Table 1. Observations by country, by sector and by year.

	2006	2007	2009	2010	2018	Total
Angola	425	0	0	360	0	785
Manufacturing	215	0	0	136	0	351
Services	210	0	0	224	0	434
Cape Verde	0	0	156	0	0	156
Manufacturing	0	0	68	0	0	68
Services	0	0	88	0	0	88
Guinea Bissau	159	0	0	0	0	159
Manufacturing	50	0	0	0	0	50
Services	109	0	0	0	0	109
Mozambique	0	479	0	0	601	1080
Manufacturing	0	341	0	0	287	628
Services	0	138	0	0	314	452
Total	584	479	156	360	601	2180
Manufacturing	265	341	68	136	287	1097
Services	319	138	88	224	314	1083

. As verified, Angola and Mozambique account for most of the data, with the latter holding nearly fifty percent of the total observations and the most recent ones.

In order to comprehend the impact of corruption on foreign direct investment, the dependent variable (FDI) is defined as the proportion of private foreign ownership of a firm, a variable that varies between zero and 1 (or 0% and 100%). The independent and control variables considered in the analysis are the following:

♦

Corruption, a dummy variable equal to 1 if the firm considers that corruption is a major obstacle to its activity and zero otherwise. According to Escresa and Picci (2017), differences between the Public Administration Corruption Index (PACI) and perception-based measures are not systematically driven. Thus, a perception-based corruption variable like the one we employ in this study may be a good proxy for actual levels of corruption and has been employed extensively in the literature (see, e.g., L. Martins et al., 2020). Regarding this variable, studies tend to diverge between those who argue that corruption hinders FDI (see Wei, 2000; Asiedu, 2006; Sabir et al., 2019; Yahyaoui, 2023), and those who defend that corruption can positively impact foreign investment (Bardhan, 1997; Gossel, 2018).

♦

Trade, measured by the percentage of a firm’s sales that are exported. A quite extensive body of literature has identified trade to be a pivotal FDI determinant (Asiedu, 2006; Boğa, 2019; R. Martins et al., 2021).

♦

Investment, a dummy variable equal to one if the firm purchased any new or used fixed assets (e.g., machinery, vehicles, equipment, land, or buildings) and zero otherwise. Several studies have concluded that investment may boost FDI, being a relevant FDI determinant in many cases (Asiedu, 2006; Du et al., 2012; Cerdeira & Fortuna, 2025).

♦

Credit, a dummy variable equal to 1 if the firm has access to a line of credit and zero otherwise. The literature has shown that easier access to finance is associated with higher FDI inflows (Boğa, 2019).

♦

Manager’s experience, defined as the years of the top manager’s experience working in the firm’s sector and considered as a proxy for human capital, which several studies identify as a key FDI determinant (Asiedu, 2006; Cerdeira & Fortuna, 2025).

♦

Inadequate workforce, a dummy variable equal to one if the firm identifies an inadequately educated workforce as a major constraint to the firm’s activities and zero otherwise. Studies have shown that an educated labor force leads to increased levels of FDI (Asiedu, 2006).

♦

Tax rates, a dummy variable equal to 1 if the firm perceives tax rates to be a major constraint and zero otherwise. This variable shall be taken into consideration, given that research has concluded that higher tax rates on firms deter FDI (Wei, 2000).

♦

Political instability, in lieu of weak institutions, a dummy variable equal to 1 if firms identify political instability as a major constraint and zero otherwise. Research has concluded that poor institutionalization and an unstable political scenery tend to dissuade FDI (Wei & Shleifer, 2000; Asiedu, 2006).

♦

Size, determined by the number of workers of the firm. Research has shown that larger companies tend to attract more foreign investment, because internationalization is more easily achievable for them, in contrast to smaller companies (R. Martins et al., 2021).

♦

Age, determined by the number of years since the firm began its operations. Given the existence of learning effects, age is a proxy for the firm’s experience and may be a relevant determinant of FDI (Fan & Wang, 2021; Cerdeira & Fortuna, 2025).

Table 2. Descriptive statistics.

Variable	Obs.	Mean	Std. dev.	Min	Max
FDI	2169	0.144	0.319	0	1
Corruption	2156	0.120	0.325	0	1
Trade	2174	3.994	15.529	0	100
Investment	2174	0.397	0.489	0	1
Credit	2162	0.112	0.316	0	1
Manager’s experience	2142	13.815	10.022	1	60
Inadequate workforce	2169	0.195	0.396	0	1
Tax rates	2158	0.280	0.449	0	1
Political instability	2097	0.231	0.422	0	1
Age	2130	12.885	11.313	1	110
Size	2164	37.725	87.894	1	2000

exhibits the descriptive statistics of the variables in the analysis. The dataset exhibits some variability as the standard deviation for FDI is higher than the corresponding mean. Overall, 20.96% of the firms are at least partially owned by foreign agents, while 79.04% are domestic firms (i.e., with the proportion of private foreign ownership equal to zero). On average, 14.4% of the firms’ capital is owned by foreign agents, with some firms being fully owned by domestic agents and others wholly owned by foreign agents.

As for the independent variables, 12% of the firms report that corruption is a major obstacle to the firm’s activity. On average, around 4% of the firms’ sales are exported, although the variability is very high, as indicated by the corresponding standard deviation. Around 39.7% of the firms purchased fixed assets and 11.2% of the firms have access to a line of credit. On average, the top manager has nearly fourteen years of experience in the firm’s sector. However, the amplitude of values is high, considering that, in some cases, the manager only has one year of experience, while in others about sixty years. As for the workforce, 19.5% of the firms identify an inadequately educated workforce as a major constraint to the firm’s activities. Interestingly, less than half of the firms consider political instability as a pressing issue, nor do they perceive tax rates to be a significant constraint. In fact, 23.1% of the firms identify political instability as a major constraint, while 28% of the firms consider tax rates to be a relevant obstacle to the firm’s activity. Finally, the average age of the firm is around 13 years, and the average number of workers per firm is around 38. However, the data exhibits high variability in these variables, with age varying from 1 to 110 years and the number of workers ranging from 1 to 2000.

We aim to analyze the impact of corruption on FDI. Our dependent variable, FDI, is measured as the percentage of a firm owned by foreign investors, ranging between 0% and 100% (or between 0 and 1). Because of this, we cannot use Ordinary Least Squares (OLS) for estimation. OLS assumes a continuous dependent variable that is not constrained within a specific range, which is not the case here. Additionally, using OLS could lead to predicted values falling outside the [0, 1] interval, making it unsuitable for our analysis. Given these constraints and the nature of FDI, we adopt the fractional probit regression model, which provides an appropriate framework for estimating the conditional mean of FDI (see, e.g., Wooldridge, 2010), $E\left({C_i,F}_i,X_i\right)$:

$$E\left({C_i,F}_i,X_i\right)=\Phi (\alpha +\beta C_i+\gamma F_i+\delta X_i),$$

(1)

where ${\mathit{F}\mathit{D}\mathit{I}}_{\mathit{i}}$ is the percentage of the firm i owned by foreigners; ${\mathit{C}}_{\mathit{i}}$ represents the corruption variable, as defined above; ${\mathit{F}}_{\mathit{i}}$ represents the vector of independent variables at the firm level, including trade, investment, credit, manager’s experience, inadequate workforce, tax rates, political instability, age and size; ${\mathit{X}}_{\mathit{i}}$ consists of the sector and countries dummies; and $\mathit{\Phi }\left(.\right)$ denotes the standard normal cumulative distribution function.

When estimating our model, we need to consider not only the nature of the dependent variable, but also the potential endogeneity of corruption. The relationship between corruption and FDI is such that corruption can influence FDI, but FDI can also shape levels of corruption (Cerdeira & Lourenço, 2022; Lourenço & Cerdeira, 2024). This means that corruption is an endogenous covariate, and we must account for this in our estimation procedure. To address the issue of endogeneity, we use the fractional probit regression method, as previously discussed, and incorporate instrumental variables (IVs). The fractional probit framework is well-suited for handling endogenous regressors (see, e.g., Wooldridge, 2010). This approach requires selecting an appropriate instrument for corruption, i.e., a variable that is correlated with corruption but uncorrelated with the error term. Following the approach of Fisman and Svensson (2007), we use industry-location averages of corruption as our instrument. The logic behind this choice is that corruption is influenced by both firm-specific factors and broader industry-level conditions. Since industry-level factors are external to individual firms, they serve as a valid instrument.

One question that may arise with this identification strategy is whether the corruption instrument may affect FDI directly, rather than primarily through firm-level corruption. We argue that the exclusion restriction is plausible conditional on the controls and fixed effects included in the model. All specifications include year, sector, and country fixed effects, which account for systematic differences in investment attractiveness across industries, time, and countries, such as sector-specific institutional features, macroeconomic conditions, or regional investment climates. Consequently, identification relies on variation in corruption within sectors and within countries, rather than on cross-sectoral or cross-country differences. In addition, the models include a rich set of firm-level controls commonly identified as key determinants of FDI, such as access to finance, investment or export orientation. These controls further mitigate concerns that the instrument may proxy for broader investment conditions rather than corruption itself. This method has been widely applied in the literature (e.g., Fisman & Svensson, 2007; Ha et al., 2021; L. Martins et al., 2020; Cerdeira & Fortuna, 2025).

In addition to the regression for analyzing the impact of corruption (our baseline model), we estimate the model considering different impacts of corruption for manufacturing firms and services firms, as well as for small and medium firms (SME) and large firms. We also interact the corruption variable with dummies for countries to evaluate the potentially different impacts of corruption in each country, that is, we augment the regressions by including four additional independent variables defined as the interaction between the corruption variable and each country dummy. All regressions include sector, year and country fixed effects: each firm is included in one of 11 sectors so each model includes 10 dummy variables for sectors (with the remaining sector being the baseline sector); as there is data for 5 years, 4 dummy variables for year are included in the models (with the remaining year being the baseline year); as we consider 4 countries in the analysis, 3 dummy variables for country are included in the models (with the remaining country being the baseline country). All specifications include robust standard errors to account for heteroskedasticity.

4. Results and Discussion

The estimation results are presented in

Table 3. Estimation Results.

	Baseline Model	Manufacturing vs. Services	SME vs. Large Firms
Corruption	0.911 (0.777)	−2.252 (1.585)	0.692 (0.710)
Corruption × Services		3.909 ** (1.569)
Corruption × Large Firms			3.946 ** (1.725)
Trade	0.006 *** (0.002)	0.005 *** (0.002)	0.005 ** (0.002)
Investment	0.044 (0.067)	0.061 (0.062)	−0.003 (0.065)
Credit	0.299 *** (0.101)	0.151 (0.131)	0.335 *** (0.095)
Manager’s experience	−0.005 (0.004)	−0.007 * (0.004)	−0.003 (0.004)
Inadequate workforce	0.021 (0.100)	0.036 (0.094)	0.027 (0.093)
Tax rates	0.026 (0.080)	0.058 (0.076)	0.035 (0.076)
Political instability	−0.157 * (0.090)	−0.157 * (0.089)	−0.165 * (0.087)
Age	−0.009 ** (0.004)	−0.006 (0.004)	−0.011 *** (0.004)
Size	0.002 *** (0.001)	0.002 *** (0.000)	0.000 (0.001)
Constant	−1.237 *** (0.120)	−0.949 *** (0.254)	−1.083 *** (0.155)
Wald ${\mathit{\chi }}^{\mathbf{2}}$ test	115.40 ***	226.41 ***	130.79 ***
Observations	1969	1969	1969

Note: *** (**) [*] statistically significant at 1% (5%) [10%]. Robust standard errors in parentheses.

.1 All regressions include sector, year and country fixed effects. Three sets of results are displayed: the baseline model; a model in which we add the interaction between the corruption variable and the variable Services, a variable that takes the value 1 if the firm is included in a service sector and zero if the firm is included in a manufacturing sector; and a model in which we add the interaction between the corruption variable and the variable Large Firms, a variable equal to 1 if the number of workers of the firm is higher than 100 and zero otherwise. All models include 1969 observations and are globally significant at the usual significance levels, as shown by the Wald ${\mathit{\chi }}^{\mathbf{2}}$ test.2

As for the baseline model, the results suggest that the impact of corruption on FDI in the PALOP economies is negligible, as the corresponding coefficient for corruption is not significant. Thus, H1 is not confirmed. This result differs from several studies that conclude there is a negative impact of corruption on foreign direct investment (see Habib & Zurawicki, 2002; Zhao et al., 2003; Asiedu, 2006; Cuervo-Cazurra, 2008; Castro & Nunes, 2013; Sabir et al., 2019; Yahyaoui, 2023), as it fuels the uncertainty of operating costs in the country (Kaufmann, 1997) and the unpredictability of business activity (Wei, 2000). While these conflicting results may be a consequence of including different countries and years in the analysis, as the impact of corruption on FDI depends on these factors (e.g., see Lourenço & Cerdeira, 2024), our result also contrasts with the findings of Grande and Teixeira (2012) concerning corruption’s negative impact on FDI in the PALOP economies. This paper uses a different dataset for the PALOP economies and a logistic model with the dependent variable equal to 1 if the firm chooses FDI to enter the market and zero if other entry modes are chosen. In addition, the corruption measure used in Grande and Teixeira (2012) relies on the Corruption Perceptions Index, an indicator at the country level. We reason that the use of a macro indicator for corruption, instead of a measure that (potentially) varies across firms in the same country (as we consider in our analysis), may help to explain the contradictory results, in particular when the data and methodology (and the nature of the dependent variable) are different in these two studies.

The results also show that trade, credit, and size are statistically significant and positively impact FDI in the PALOP economies. Thus, the larger the firm, i.e., the number of employees, the more likely it is to attract FDI, which aligns with previous research (R. Martins et al., 2021). At the same time, access to credit and trade boost FDI, suggesting that these variables are pivotal determinants of FDI (Asiedu, 2006; Boğa, 2019; R. Martins et al., 2021). Results concerning trade might be explained by Angola’s dependence on the oil sector (World Bank, 2022) and Mozambique’s exports of aluminum, coal, and natural gas, among others (República de Moçambique, 2016).

On the other hand, our findings indicate a negative impact of age on FDI, suggesting that young, recent firms have been able to attract high proportions of FDI in these countries. In addition, political instability also may reduce FDI, as the corresponding coefficient is negative (although it is statistically significant only when considering a 10% significance level). As for investment and tax rates, the estimates in Table 3 are not statistically significant, i.e., these variables do not have a relevant impact on FDI in the PALOP economies. This result also holds for variables concerning the firm’s human capital, i.e., manager’s experience and inadequate workforce.

Table 3 also shows the results for the models that include an interaction term between the variables Corruption and Services and Corruption and Large Firms. Our estimates reveal a differentiated impact of corruption on FDI in manufacturing and services sectors: while the impact in manufacturing sectors is, on average, negligible, there is a positive impact of corruption on FDI in services. This is in line with the “greasing the wheels” perspective, which suggests that in countries where regulations are unclear and bureaucratic inefficiencies make it difficult to enter and operate in the market, corruption can serve as a tool for efficiency (Egger & Winner, 2005; Barassi & Zhou, 2012; Leff, 1964; Huntington, 1968). In addition, the effect of corruption is different for small and medium enterprises and for large firms. Corruption does not significantly impact FDI in small and medium firms, but the effect is positive and statistically significant for large firms, a finding in line with the previous literature (L. Martins et al., 2020; Lourenço & Cerdeira, 2024).3 These results confirm H2 and H3.

The models estimated in this study are non-linear regressions. As such, the estimated coefficients presented in Table 3 are not the marginal effect of the independent variable in the conditional mean of the dependent variable. We display these marginal effects in

Table 4. Marginal Effects.

	Baseline Model	Manufacturing vs. Services	SME vs. Large Firms
Corruption	0.196 (0.182)	−0.581 (0.506)	0.163 (0.182)
Corruption × Services		0.101 * (0.060)
Corruption × Large Firms			0.093 ** (0.046)
Trade	0.001 *** (0.000)	0.001 *** (0.000)	0.001 ** (0.000)
Investment	0.009 (0.014)	0.016 (0.017)	−0.001 (0.015)
Credit	0.064 *** (0.023)	0.039 (0.029)	0.079 *** (0.025)
Manager’s experience	−0.001 (0.001)	−0.002 * (0.001)	−0.001 (0.001)
Inadequate workforce	0.004 (0.021)	0.009 (0.024)	0.006 (0.022)
Tax rates	0.006 (0.017)	0.015 (0.020)	0.008 (0.018)
Political instability	−0.034 * (0.020)	−0.040 * (0.022)	−0.039 * (0.021)
Age	−0.002 ** (0.001)	−0.001 (0.001)	−0.002 ** (0.001)
Size	0.001 *** (0.000)	0.001 *** (0.000)	0.001 (0.002)

Note: *** (**) [*] statistically significant at 1% (5%) [10%].

. The results for the baseline model show that, on average, an increase of 1 percentage point in Trade leads to an increase of 0.1 percentage points in FDI, ceteris paribus. On average, access to a line of credit improves FDI by 6.4 percentage points, while political instability may reduce, on average, FDI by 3.4 percentage points (although it is statistically significant only when considering a 10% significance level), all other things being equal. As for the models with the interaction terms, corruption in the services sector may boost FDI by 10.1 percentage points, on average, compared to the manufacturing sector (although it is statistically significant only when considering a 10% significance level). Regarding the impact of corruption across firms of different dimensions, the results indicate that corruption fosters FDI in large firms by 9.3 percentage points compared to small and medium firms.

These results should be interpreted with caution and do not imply that corruption is efficiency-enhancing or desirable from a policy perspective. In fact, they are consistent with second-best mechanisms emphasized in the literature: in environments characterized by regulatory complexity and discretionary enforcement, some firms—namely large firms and those operating in service activities with frequent interactions with public authorities—may be better able to navigate bureaucratic frictions through informal channels. While this does not reflect a welfare-improving role of corruption, it highlights differences in firms’ capacity to absorb, anticipate, or negotiate corruption-related costs (Shleifer & Vishny, 1993; Wei, 2000). In addition, while corruption appears to be associated with higher foreign ownership in specific sectors or firm sizes, that does not imply the absence of substantial distributional and efficiency costs as corruption may advantage firms with greater bargaining power or political connections but, at the same time, raise entry barriers for smaller and less connected firms. As a result, firm-level “greasing” effects can coexist with negative aggregate and long-run consequences for institutional quality and investment efficiency (Svensson, 2006).

To examine the varying impact of corruption on FDI across different PALOP countries, we build upon the baseline model in Table 3 and estimate the country-specific effects of corruption on FDI using different reference countries. To obtain this, we interact the corruption variable with a country-specific dummy variable, systematically rotating the baseline country to capture all possible effects. This approach enables us to analyze and test how corruption impacts FDI in each country relative to the chosen baseline. The results are presented in

Table 5. Differences in countries’ impact of corruption on FDI.

	Angola	Cape Verde	Guinea Bissau	Mozambique
Angola	---	−2.507 (12.646)	−0.198 (2.006)	2.063 *** (0.598)
Cape Verde	2.507 (12.646)	---	2.434 (12.146)	5.631 (12.717)
Guinea Bissau	0.198 (2.006)	−2.434 (12.146)	---	2.261 (1.991)
Mozambique	−2.063 *** (0.598)	−5.631 (12.757)	−2.261 (1.991)	---

Note: *** (**) [*] statistically significant at 1% (5%) [10%].

. Each country appears in columns, while the respective baseline country is displayed in rows.

The estimates suggest that the impact of corruption on FDI is, on average, the same across PALOP countries, with the exception of the differentiated impact in Angola and Mozambique. In fact, corruption seems to have a more positive impact on FDI in Mozambique than in Angola, i.e., the impact of corruption on FDI in Angola is negative compared to Mozambique’s case.

Considering the years 2006 and 2010 for Angola (see Table 1), the Corruption Perceptions Index exhibits a score of 2.2 on a scale of 0 (highly corrupt) to 10 (highly clean) in 2006 (Transparency International, 2006), which later decreased to 1.9 in 2010 (Transparency International, 2010). This decrease in ranking and worsening corruption may help explain why investors have been dissuaded from investing in the country compared to alternative PALOP countries, as corruption usually raises the uncertainty and cost of operating in the country (Kaufmann, 1997). Additionally, it seems corruption in Angola typically involves foreign multinationals and top-level domestic and foreign officials, as in the case of the major corrupt arms deal referred to as “Angola Gate” (Chêne, 2011), which may hinder other investors from associating themselves with the country.

5. Conclusions

This study seeks to analyze the impact of corruption on foreign direct investment in the PALOP countries. We utilize data from the World Bank Enterprise Surveys (WBESs) from firms spanning the years 2006 to 2018, comprising a total of 2180 observations. We adopt the IV Fractional Probit Regression framework, which effectively addresses the nature of the FDI variable and the potential endogeneity between corruption and FDI.

The empirical analysis indicates that, on average, corruption does not have a significant impact on FDI in PALOP economies. The results also show that trade, credit, and firm size are relevant determinants of FDI in these countries. On the other hand, investment and tax rates have a negligible influence on FDI. Interestingly, the impact of corruption varies across sectors: while it hardly affects FDI in the manufacturing sector, it boosts FDI in the services sector. The effect of corruption on FDI also differs depending on the size of the firm: for small- and medium-sized firms, corruption does not seem to significantly impact FDI; however, for large firms, the effect is positive and statistically significant. When looking across PALOP countries, the overall impact of corruption on FDI seems to be similar, except for Angola and Mozambique, as corruption appears to have a negative impact on FDI in Angola in comparison with the corresponding impact in Mozambique.

These results highlight the need to distinguish between short-run firm-level responses to weak institutions and their broader implications for economic governance. While corruption may function as a way to “get things done” for some firms in highly regulated or discretionary contexts, it simultaneously signals institutional deficiencies that ultimately deter investment, reduce competition, and weaken trust in public institutions (Habib & Zurawicki, 2002; Wei & Shleifer, 2000). Thus, these results highlight the importance of institutional reforms that limit discretionary contacts between firms and public authorities, particularly in service activities and in dealings with large firms. Measures that increase transparency, simplify regulatory procedures, and standardize enforcement are more likely to foster sustainable foreign investment than reliance on informal arrangements, whose benefits are unevenly distributed and institutionally fragile (Shleifer & Vishny, 1993; Svensson, 2006).

The results of this study have relevant policy implications, raising awareness around the challenges these countries face in the present day and providing insights to enable them to introduce attractive measures that will result in an easier and steadier inflow of foreign direct investment. Notwithstanding, this research holds a few limitations. Firstly, PALOP countries consist of more than the four countries included in our analysis. In fact, Equatorial Guinea and São Tomé and Príncipe were not included in the analysis because of data limitations: although it became an observer member in 2014, Equatorial Guinea only became a fully-fledged PALOP member during the FORPALOP Virtual Summit in 2021, and the WBES database does not have information for this country from 2021 on; in the case of São Tomé and Príncipe, the database did not present any data for the country at all. Also, more recent firm-level data are mainly available for Mozambique, which represents a larger share of observations in the later years; this reflects data availability rather than sampling choices and suggests caution when interpreting the findings as fully representative of the most recent period across all PALOP countries. In addition, this paper uses a database with only cross-sectional information for firms, which does not allow for longitudinal analysis. Future research that uses panel data would be useful to explore the impact of corruption on FDI, controlling for country and firm heterogeneity over time.