1 Introduction

Since the overthrow of Ghana's first president in 1966, coup d’états became a major repertoire for rationalization against corruption until 1992 when there was a restoration to democracy. This is mainly because the devastating effect of corruption on development in general is perceived to be significantly high. That notwithstanding, real GDP growth per capita has been fluctuating in Ghana. In 2008, Ghana recorded its first double-digit growth rate of 14.5% after independence [1]. Nevertheless, the aftermath of the commercialization of oil discovery saw Ghana’s GDP shrink further by 7 per cent despite rebasing the economy [2]. Ghana’s vegetation cover continues to reduce at a frightening rate as a result of over-exploitation entrenched in rent-seeking. Thus, in less than 50 years, it is estimated the primary rainforest of Ghana has reduced by 90%, whereas in the last 20 years (1990–2010), the country lost more than 2.5 million hectares or 33% of its cover to illegal mining [3]. At the same time, the inflow of foreign direct investments (FDI) to Ghana has also been fluctuating. For instance, over the last 9 years, the value of net inflows has increased from $144,970,000.00 in 2005 to $3,293,430,000.00 in 2012. Heightened interest has emerged about which sectors of the Ghanaian economy have been the recipients of these mammoth investments in foreign capital inflows and how has that influenced the spate of corruption through rent-seeking tendency. Ghana is currently ranked 70th position out of 180 countries and territories with a score of 43 on the corruption perception index globally. The spate of corruption has stagnated in the last four years with no sign of improvement in the anticorruption strategies [4] as shown in Fig. 1. Therefore, this study is guided by the following research questions: (1) What has been the effect of corruption on different development outcomes in Ghana? (2) Do the positive and negative aspects of corruption and corruption controls elicit an equal impact on Ghana’s development trajectory? Is there evidence to suggest the operation of MNEs through FDI has environmental consequences? This study is driven by these concerns and more.

Fig. 1
figure 1

Source: Author’s construct

Score changes for corruption in Ghana from 2013 to 2023.

Full size image

The corruption-economic development nexus is well established at the micro and macro levels [5,6,7]; yet on sustainable development, few empirical studies present country-level evidence. Most studies on the corruption–sustainable development nexus have been explored at the macro level [8,9,10]. It is instructive to point out that, most of these studies that purport to have explored the nexus between corruption and sustainable development failed to incorporate a one-stop measure of sustainability but rather resorted to different measures such as GDP growth rates, food insecurity, poverty outcomes, income inequality, public debt among others without recourse to the environment at the macro level. GDP growth rate is not a sustainable measure of development as it has always been a gauge for economic development in most research studies; therefore, these studies may have suffered from a methodological gap. Moreover, no study present country-level evidence on a resource-rich economy such as Ghana. This has important implications in extending the literature on the corruption-sustainable development nexus and conscientize policymakers on the need to gauge development trajectory on sustainable measures that incorporate the fundamental dimensions of sustainability rather than the traditional measure of growth and development.

This paper therefore examines the effects of corruption on different development outcomes with emphasis on the symmetric and asymmetric effects of corruption on well-being. It is instructive to note that corruption as a concept is broad and multidimensional which implies there is no one-size-fits-all approach to perceiving it. The form it assumes is indicative of its complexity and this could serve as a potential source of asymmetry owing to the imperfect information regarding it. The strength of this study is in three folds: (1) the study adopts a measure of sustainability that is broad and encompassing of the basic principles of sustainability as presented by [11] through the use of the genuine wealth per capita (GW) in comparison with the traditional measure of development measured by GDP growth per capita; (2) the study also presents a detailed account of the anticorruption policy initiatives by successive governments in clamping down on the menace of corruption in Ghana, and (3) the study applies an estimation technique that captures structural breaks and at the same time present a linear and the nonlinear bounds testing procedure in the analysis of level relationships. This has important implications to resource-rich economies in sub-Sahara Africa, particularly Ghana which depend mostly on its natural resource endowment for development and growth. This is against the backdrop that [12] have recently documented a nonlinear relationship between natural resources and sustainability in sub-Saharan Africa. Understanding further the channels of this phenomenon might give policymakers a clear understanding when it comes to decision-making.

The study is organized as follows: section one is the introduction, which spells out the rationale and the policy problem statement. The second section is the literature survey; which looks at the theories and concepts including the empirical linkages that have been established. A stylized fact about the anticorruption strategies adopted by successive governments is also explored in this section. The method and design are presented in section three together with a brief description of the nature of the data. The study results and discussion are concurrently presented in section four. The study concludes in section five with policy implications.

2 Related literature and hypothesis development

2.1 The corruption – development perspective

Development theories are manifold but in contemporary times, the neoclassical and the endogenous growth theories have dominated the discussion on income convergence. Robert Solow and Trevor Swan are credited for their contribution to the neoclassical theory; which emerged from simplistic Harrod-Domar models in the 1950s [13]. According to this theory, three factors (i.e., Labour, capital, and technology) are necessary for a growing economy. However, the neoclassical school of thought counter-argued that the view that temporary equilibrium is different from long-term equilibrium does not require the three factors espoused by the Harrod-Domar models. Endogenous growth theory on the other hand argues that technological advances do not fall from heaven (exogenous) as in the neoclassical theory. Rather, these factors should be internally (endogenously) developed and determined. Convergence of incomes has never been achieved and inequality continues to linger on allowing development advocates to call for an alternative to this measure. Proponents of such departure included [14], who called for a comprehensive measure of well-being that incorporates elements in the entire social system. Myrdal classified the social system into economic and non-economic factors. The corruption–development perspective is thus divided into two different outcomes: (1) economic development and, (2) sustainable development.

The perspectives on economic outcome and corruption are further divided into two strands (grease vs. sand). While the former documents a positive relationship between corruption and economic development [15], the latter have otherwise empirically proven that the menace indeed has a negative effect by reducing growth in per capita on a country-wide level [16]. GDP per capita is an indicator of a country’s standard of living, which most of the time is a measure of economic development or growth. The effect of per capita income on economic growth has been extensively discussed in related studies within the growth nexus. However, corruption can mitigate against the achievement of a uniform distribution of wealth [17]. But in terms of general outlook, per capita income could paint a rather positive picture concerning the performance of an economy disregarding inequality. When considered this way, it could be hypothesized that:

H1: Corruption could have a positive significant effect on growth by assuming incomes are converging.

The other perspective which is on sustainability, started as a concept from a simple idea in 1987 by the Brundtland Commission on development that is intergenerational in orientation [18]. The idea has received numerous contributions from practitioners, scholars and policymakers after the launch of the Sustainable Development Goals (SDGs) in 2015. Operationalising the core principles of sustainability has attracted an avalanche of research outputs on the sustainability strand of development [8, 9, 11, 19]. The early contributors to the concept of sustainability identified three different perspectives; which include economic perspective, ecological perspective and social perspective. These perspectives operate in unison without one exerting any negative effect on the other. The principles of sustainable development hold that there should be a coordinated approach to tackle social inequalities and the damage to the environment without compromising on a sound economic pillar [11, 20]. Others have cautioned that sustainable development should not be viewed as a one-size-fits concept. There are some intrinsic weaknesses in the approach that have to do with the persistent environmental, social and economic problems. Rather, there has been a call to see the concept as a framework to mitigate these implementation deficits [21]. As a result, five dimensions of sustainability have been identified including place or environment that informs one's identity, fact and behaviour as the first three dimensions. The fourth dimension is permanence or changes and improvement and the person's corner is the fifth dimension [21].

For instance, the work of [22], who used genuine wealth growth per capita as a proxy for sustainable development, observed that corruption has a negative effect on sustainable development. The study however emphasized that such a negative effect is well appreciated when institutional variants such as democracy and governance were used as control variables. For instance [17] documents a similar relation when he correlated wider corruption indices with genuine wealth per capita. In examining the relationship between corruption, the resource curse, and genuine savings [23], argued that institutional quality weakens development. Wealth per capita measures the change in total assets taking into account population changes [17, 24]. This measure is an extension of genuine savings. Four critical adjustments are made to the traditional genuine savings proposed by [25]. For operational purposes, an economy is said to be on a sustainable development path when the inter-temporal social welfare (the measure of the present discounted value of social welfare attained at each future date) does not decrease over time along the chosen path [17]. Other scholars such as [12] document the nonlinear relationship between natural resources and sustainability in sub-Saharan Africa. Empirically, it is negatively correlated with corruption. By implication, corruption reduces inter-temporal welfare. This paper thus envisages:

H2: Corruption could dampen sustainability through the shrinking of wealth per capita.

2.2 Foreign direct investment, financial innovation and development

Foreign direct investment (FDI) is considered a necessary form of capital inflow to developing economies primarily because such a channel may not be liable to crises and sudden stops as is of aid and grants. Additionally, there are no visible strings attached to such kind of investment. The launch of the SDGs in 2015 and the changes in the climatic conditions lately present a significant challenge to developing economies, particularly those in sub-Saharan Africa due to low socioeconomic baseline. The influx of FDI within an economy may be well appreciated when there are a significant number of multinational enterprises (MNEs) operating outside the financial system which happen to be the traditional recipient of FDI. Thus, the activities of FDI (MNEs) elicit two different behaviours in the ecosystem that feed into two different pollution hypotheses: (1) the pollution halo hypothesis, and (2) the pollution haven hypothesis. The pollution halo perspective is the view that MNEs may share superior technologies such as green technology, and energy-saving technology among a host of other technologies with the host country to enhance the welfare of the host country [26]. The pollution haven hypothesis on the other hand documents that MNEs relocate to other jurisdictions to saddle the host country with pollution, while reducing the spate of pollution for the country of origin [27,28,29]. This action may be detrimental to the well-being of the host country, while the country of origin may benefit.

The theory of creative destruction holds unique relevance to development from the broader perspective as it expounds on how capitalism enhances growth and development through innovation and entrepreneurship. That notwithstanding, recent research works have begun exploring the environmental consequences of financial innovation through technology adoption on sustainability. Energy-intensive digital currencies like cryptocurrencies have been the main point of emphasis in most studies. These studies argue that energy consumption and carbon pollution are associated with cryptocurrency mining and use [30, 31]. Others such as [32] have warned that Bitcoin mining emissions alone could push global warming above the critical 2 degrees Celsius threshold. Other scholars such as [12] argue that the interaction between financial development and natural resources worsens the ecosystem. The effect of these variants of financial innovation extends beyond localized contamination as it translates into a surge in greenhouse gas (GHG) emissions. Other studies have acknowledged the positive effect of financial innovation on environmental sustainability and development and other studies have pointed to no effect on growth [33,34,35,36,37]. In China, a study on the nexus between financial development and carbon emission using NARDL revealed that the asymmetric form of financial development elicits equal proportionate effects on environmental quality [38]. In South Africa, [39] argued that financial development stimulates carbon and ecological footprints in the short run; but in the long run, it dampens sustainability.

Realizing this grim reality, this study extends our understanding of how MNEs (FDI) and financial innovation have impacted development outcomes differently in Ghana. These variables, it must be said are used as a control to gauge the behaviour of firms and technology in the nexus between corruption and development outcomes broadly.

2.3 Stylized facts of anti-corruption policy initiatives in Ghana

Ghana’s anticorruption policies unofficially took effect during a brief spell by the AFRC in 1979. The June 4th uprising by the Armed Forces led to the ousting of Gen. Fred William Kwasi Akuffo’s Supreme Military Council. The “House Cleaning Exercise” initiated by the Armed Forces Revolutionary Council (AFRC) Government led by Flt. Lt. Jerry John Rawlings was chiefly justified in the name of fighting corruption. In addition to these efforts, major blue-prints (manifestoes) of successive political parties have captured an aspect of these anti-corruption strategies, including promises to fight corruption, with each party pledging on rally platforms to fight corruption better than their opponents. Other anti-corruption strategies have focused on the moral aspect of society where the help of religious and community leaders has been the target to guide citizens to uphold the values of integrity and to manifest high moral ethics in their personal lives. Ghana ratified other international pacts on corruption such as the UNCAC and the AU Convention in 2005, and the ECOWAS Protocol in 2003 [40]. Regardless of these developments, corruption remains a reality.

Establishing constitutional and statutory bodies with vested powers has been pencilled as some of the anti-graft measures adopted by the country. Bodies such as the Commission on Human Rights and Administrative Justice (CHRAJ), the Economic and Organized Crime Office (EOCO), the Finance Intelligence Centre (FIC), and the Fair Wages Salary Commission are the foremost anti-corruption institutions in Ghana. These institutions are there to support the efforts of traditional law enforcement agencies such as the Ghana Police Service (GPS) and the Bureau of National Investigations (BNI). As part of efforts to ensure that the mandates of these institutions are strictly adhered to and followed through, the Government of Ghana in 2011 launched its first major anti-corruption blue-print dubbed the “National Anti-Corruption Action Plan (NACAP)” to appraise and evaluate the activities of these vested and constitutional bodies. The plan outlines its key strategy for appraising and evaluating the pro-activeness of these institutions by offering a framework to effectively mobilize broad public support and resources for anti-corruption activities in a focused and sustained fashion [40]. Since its establishment, the agency has mildly made some progress in the fight against corruption. It has emerged that more than 202 people are currently serving jail sentences, 336 are awaiting trial, assets have been frozen in 46 cases and the EOCO has interrogated more than 30 people. Public sentiment regarding the fight against corruption is seen to be sluggish hence the need for an independent investigator who cannot be tagged with party affiliation to bite hard on clamping down on the menace of corruption.

In 2017, the office of the special prosecutor was established from an act of parliament (Act 663, 2018) by the Akuffo Addo-led government [41]. Consequently, the Office of the Special Prosecutor (OSP) serves as the autonomous body to handle sensitive cases that the Attorney General would otherwise be ineffective at handling. Since its inception, its key achievements have been among other things the following: (1) the disruption of counterfeit foreign currency manufacturing network; with a discovery of over USD$40 million counterfeit dollars; (2) the suspension of TOR-Torentco Deal (USD$22 million lease agreement between TOR and Tema Energy and Processing limited); (3) closed pathways for corrupt practices and tax avoidance; (4) youth in an anti-corruption campaign that uses interactive based pedagogy based on the learner-led model; and (5) direct recovery that included the suspension of auction goods at the port, and recovery of over GHS 1 million from the Labianca Company Limited scandal [42].

3 Research design and method

3.1 Research approach, design, and data

This study employs a quantitative approach with the experimental research design type to inquiry. According to [43], experimental design is the process of undertaking a study in an objective and controlled manner; so that precision is attained and specific conclusions drawn from the hypothesized statements. Thus, its existential purpose is to establish the effect that an independent variable has on a dependent variable.

The data for this study is primarily sourced from credible agencies such as the International Monetary Fund (IMF), The World Bank (WB), ALFRED, and Transparency International (TI) from 1980 to 2023 for the World Development Indicators (WDI), 1995 -2023 for the TI data on corruption. The work file structure of the dataset is annual or yearly thereby making it a time series with a data point of 1498 and 44 years of data observation. It is worth noting that the study has two dependent variables: (1) Economic development, and (2) sustainable development. Economic development is measured by growth in per capita income; while the measure of sustainability is genuine wealth per capita. The study also controls for firm behaviour using foreign direct investment and financial innovation using active mobile money users. Other measures that are policy-oriented such as control of corruption are intermittently introduced into the equation progressively.

Table 5 in the appendix presents the descriptive summary statistics of the variables at each level. The skewness and kurtosis values were also found to be within the acceptable threshold. On average, the mean score of public sector corruption was significantly high at 3.69 on a scale of 0–10 (\(\approx 36.9\%)\) with a corresponding rate of dispersion of 0.30. Control of corruption, on the other hand, recorded a -0.15 (SD = 0.125). Concerning the development outcomes, it turns out that Per capita GDP Growth exhibits the highest score of 4.52% (SD = 3.58) whereas the lowest score is observed on Genuine Wealth per capita -1.74% (SD = 5.36). The variable FDI on the other hand scored 2.98% on average with a standard deviation of 2.83%. The measure of financial innovation averaged 515.46 per 1000 active mobile money users with a standard deviation of 342.98 per active user.

3.2 Model specification and estimation technique

The characteristics of the data warrant the use of an estimation technique that can deal with small sample observation. Therefore, this study uses the ARDL bound test for the following reasons: (1) able to deal with a small sample [44]; (2) does not need the variables of interest to be in the same order but rather a combination of integration at levels I (0) and I (1), and (3) can eliminate issues of serial correlation and endogeneity. Thus in this study, following [45], the ARDL approach to cointegration involves estimating the conditional error correction version of the ARDL model for both real per capita GDP and corruption on one side, and Wealth per capita and corruption on another side. In both cases, the study gauges for firm’s behaviour using capital inflows and financial innovation using the number of active mobile money users as a proxy:

$$\Delta Y_{t} = \beta_{0} + \mathop \sum \limits_{i = 1}^{p} \beta_{i} {\Delta }Y_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \gamma_{i} {\Delta }COR_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \delta_{i} {\Delta }FDI_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \tau_{i} {\Delta }FIN\_INNO_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \sigma_{i} {\Delta }COR\_CONT_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \rho_{i} {\Delta }D_{t - 1} + \lambda ECT_{t - 1} + \varepsilon_{t} { }$$
(1)
$$\Delta GW_{t} = \beta_{0} + \mathop \sum \limits_{i = 1}^{p} \beta_{i} {\Delta }GW_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \gamma_{i} {\Delta }COR_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \delta_{i} {\Delta }FDI_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \tau_{i} {\Delta }FIN\_INNO_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \sigma_{i} {\Delta }COR\_CONT_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \rho_{i} {\Delta }D_{t - 1} + \lambda ECT_{t - 1} + \varepsilon_{t}$$
(2)

where \(\Delta\) is the difference operator, \(y\) is per capita real GDP, \(GW\) is the measure of sustainability (Genuine wealth per capita), \(COR\) is an index of corruption, COR_CONT is a measure of the anti-corruption strategies and \(FDI\) is firm behaviour measured by the inflows of capital. FINN_INNO is a measure of financial innovation. D is a dummy variable for the breakpoints, where \({D}_{t}=1,\) if \(t\ge \text{the breakpoints}\) and \({D}_{t}=0\), \(t=\)otherwise indicating no break. The long-run dynamics are captured by the coefficients \({\beta }_{i},{\gamma }_{i}, {\tau }_{i}, {\sigma }_{i}, {\rho }_{i}\text{ and} {\delta }_{i}\) for \(i=\text{1,2}, \dots 5.\) ECT is the error correction term that captures the long-run relationship between the variables and their coefficients, \(\lambda ,\) which measures the speed of adjustment to long-run equilibrium at any time of uncertainty.

Following the work of [46], the study moves a step further to estimate the nonlinear effect of the variable of interest in a Nonlinear Autoregressive Distributed lags (NARDL) by decomposing the independent variable of interest corruption (COR) into two sets of negative and positive signals denoted COR− and COR+ respectively. Thus, the decomposition series can be expressed as follows:

$$NEG\left( {COR} \right)_{t} = \mathop \sum \limits_{s = 1}^{t} COR_{s}^{ - } = \mathop \sum \limits_{s = 1}^{T} Max\left( {\Delta COR_{s} , 0} \right)$$
(3)
$$POS\left( {COR} \right)_{t} = \mathop \sum \limits_{s = 1}^{t} COR_{s}^{ + } = \mathop \sum \limits_{s = 1}^{T} Max\left( {\Delta COR_{s} , 0} \right)$$
(4)

Equations (3) and (4) are included in Eqs. (1) and (2) respectively to form the NARDL model expressed as follows:

$$\Delta Y_{t} = a_{0} + \mathop \sum \limits_{i = 1}^{p} a_{i} {\Delta }Y_{t - 1} + \mathop \sum \limits_{i = 0}^{p} a_{i}^{ - } {\Delta }COR\left( {NEG} \right)_{t - 1} + \mathop \sum \limits_{i = 0}^{p} a_{i}^{ + } {\Delta }COR\left( {POS} \right)_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \delta_{i} {\Delta }FDI_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \tau_{i} {\Delta }FIN\_INNO_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \sigma_{i} {\Delta }COR\_CONT_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \rho_{i} {\Delta }D_{t - 1} \, + \lambda_{1} Y_{t - 1} + \lambda_{2}^{ - } NEG\left( {COR} \right)_{t - 1} + \lambda_{2}^{ + } POS\left( {COR} \right)_{t - 1} + \lambda_{3} COR\_CONT_{t - 1} + \lambda_{4} FDI_{t - 1} + \lambda_{5} FIN\_INNO_{t - 1} + \omega_{t }$$
(5)
$$\Delta GW_{t} = a_{0} + \mathop \sum \limits_{i = 1}^{p} a_{i} {\Delta }GW_{t - 1} + \mathop \sum \limits_{i = 0}^{p} a_{i}^{ - } {\Delta }COR\left( {NEG} \right)_{t - 1} + \mathop \sum \limits_{i = 0}^{p} a_{i}^{ + } {\Delta }COR\left( {POS} \right)_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \delta_{i} {\Delta }FDI_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \tau_{i} {\Delta }FIN\_INNO_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \sigma_{i} {\Delta }COR\_CONT_{t - 1} + \mathop \sum \limits_{i = 0}^{p} \rho_{i} {\Delta }D_{t - 1} + \lambda_{1} GW_{t - 1} + \lambda_{2}^{ - } NEG\left( {COR} \right)_{t - 1} \, + \lambda_{2}^{ + } POS\left( {COR} \right)_{t - 1} + \lambda_{3} COR\_CONT_{t - 1} + \lambda_{4} FDI_{t - 1} + \lambda_{5} FIN\_INNO_{t - 1} + \omega_{t}$$
(6)

To avoid over-parameterization and to ensure parsimony, the number of dummies is set to a maximum of 9. The maximum lag length is set to 4. The study adopted the general-to-specific modelling approach presented by [39], with the optimal lag length determined by the Akaike info Criterion (AIC). Another investigation is done to explore the causal relationship among these variables. To end, several other diagnostic checks are conducted on normality, serial correlation, RESET for model misspecification and the CUSUM test of stability for the predicted model.

4 Empirical results and discussion

Before applying the ARDL and NARDL regressions, the study performed some preliminary tests that included unit root and cointegration tests. The Zivot and Andrew test of unit root is used in this study as opposed to other conventional unit roots due to its ability to address structural break issues in the series. The null hypothesis of the test implies the presence of a unit root in the series whereas the alternative hypothesis suggests otherwise.

The results as indicated in Table 1 specify a mixed integration at the level and first difference. This means GDP per capita income growth rate is integrated at both levels and first difference, but Genuine wealth per capita is integrated at the first difference. The other policy variables such as FDI, corruption, and control of corruption are all integrated at first difference. Thus, the study finds evidence of the complex nature of the dynamic properties of the variables, with a mixture of I (0) and I (1) series but with none of the variables integrated at the second difference I (2). Given the uncertainty concerning the time series properties of the variables, the decision to use the ARDL approach is tenable.

Table 1 Zivot-Andrews unit root results
Full size table

The break points in the series signify major economic and political reforms in the country (see Fig. 1). For instance, the break year for GDP per capita at the level and first difference captures the period of famine, the economic recovery programme (ERP), and the structural adjustment programme (SAP) in the early 80 s initiated by the World Bank and the International Monetary Funds. The break year for FDI in 2005 and 2008 marked the period of the enactment of a legislative instrument (L.I. 1817) by the Ghana Investment Promotion Centre to reintroduce and promote tourism. The break year for corruption in 2013 and 2016 captures the period of political reform dominated by elections and electoral petitions in the law court. The break year for corruption control in 2005 and 2006 marked the periods Ghana ratified other international pacts on corruption such as the Coalition – Association for the Implementation of the UN Convention against Corruption (UNCAC) and the AU Convention in 2005, and the ECOWAS Protocol in 2003 [40]. The break year for financial innovation in 2012 is the period service operators began with momo-based micro-loans to subscribers. These reforms did open the economy up for capital inflows and the easy movement of funds within and outside the economy.

Table 2 presents the ARDL bound test results. The F-statistics helps in deciding the presence or otherwise of cointegration when the bound test is performed. The rule of thumb states that the F-statistics must be greater than the critical values (10%, 5%, 2.5%, and 1%) to reject the null hypothesis of no integration. The study finds evidence of a long-run equilibrium relationship as the values of the F-statistics (21.657 and 7.915) are greater than the critical values at 10%, 5%, 2.5% and 1% respectively for the respective development outcomes. Therefore, it is safe to theorize that the variables are cointegrated for the different development outcomes. The study moves to estimate the long-run and short-run equilibrium relationship of the cointegrated variables and the degree of adjustments in Table 3.

Table 2 ARDL Bounds test for co-integration
Full size table
Table 3 ARDL Long-run and short-run outputs for development outcomes
Full size table

The relationship between corruption and GDP per capita growth meets priori expectations, as it is positive and significantly related. This result supports the grease in the wheels' hypothesis advanced by [47]. The relation strengthens with the introduction of measures to control corruption in the long run in Table 3 Panel 1-A; but in the short run (Panel 1-B), there is evidence to suggest corruption control may reduce the convergence of income. This is possible as there is cause to believe the approach to fighting corruption in the Ghanaian context by the Attorney General and the Office of the Special Prosecutor is fraught with interference. Critics have accused the special prosecutor of window dressing; while neglecting the main actors in the menace. Foreign capital inflow through the activities on MNEs (FDI) on the other hand had been insignificant in both the short and long runs. However, in the long run, it is worth emphasizing on the negative relationship of FDI on income convergence albeit it is insignificant. Generally, FDI impacts positively on the host countries' economy; although the growth is not robust as there are job losses through the displacement of local nationals by foreign firms including cultural effects in the short and long runs. The impact of financial innovation through mobile money users on growth is inconclusive in the long-run. In the short run, it is seen to dampen income convergence. The strict regulation of the financial technology (FIN-TECH) sector may have contributed to its usage as it became a conduit for the repatriation of funds invested in the economy. This result, however, does not support the assertion by [37]; who found no evidence of financial innovation impacting economic growth in Ghana. Throughout the results, the dummy variable is insignificant but is accounted for in the model. The coefficients of the first lagged error correction model (ECM); which is negative capture the extent to which GDP per capita growth adjusts to the deviation from the short-run equilibrium and is around 27.6%. The diagnoses of the linear ARDL model, as inferred from the adjusted R-square, show that the model explains 61.2% of the variance and the model fitness deduced from the F-statistics (p = 0.000) suggests the prediction value of the model. The model stability is creditably good as the p-values of the autocorrelation test, heteroskedasticity test (the ARCH EFFECT), and Jarque–Bera normality are insignificant. This implies the model does not suffer from serial correlation or heteroscedasticity problems and the distribution has normally distributed residuals. Graphically, the CUSUM and CUSUM of squares (see Fig. 2) are between the critical bounds at a 5% level of significance. This is a confirmation of the accuracy of the short and long-run coefficients.

The study replicates the same process to capture the effects of corruption and corruption controls on genuine wealth per capita (GW) when development is conceptualized based on sustainability. In the long-run, as shown in Table 3 Panel 1-C, the first and fourth lags of corruption significantly impact sustainable development negatively. This result dovetails with the finding of [22] on the depressing outlook of corruption on GW in a dysfunctional institutional setting. Corruption controls do not significantly impact GW. However, FDI and financial innovation impacts negatively on GW and this provides evidence in support of the pollution haven hypothesis. It also captures the detrimental effect of financial innovation on emissions. These results support the finding by [11, 22, 25] who found the activities of MNEs (FDI) and financial innovation to be depressing to environmental sustainability.

The short-run results in Panel 1-D are slightly different as the focus is on the first to the third lagged effects of financial innovation on GW. The relationship is simultaneously negative but insignificant and positive and significant respectively. This result is similar to the study by [48]; whose result was observed to be inconclusive on the impact of financial innovation on sustainable development. The speed of adjustments in the short-run equilibrium is significantly well over 75% with the expected negative sign. A summary of the diagnoses report is quite similar to the results under economic outcome as the model is stable because the p-values of the autocorrelation test, Jarque–Bera normality are insignificant. The model fitness inferred from the F-statistics is significant (p = 000) and the cross-variable explanation was well over 71.6%. The CUSUM and CUSUM of squares (see Fig. 1B) are between the critical bounds at a 5% significant level. This is a confirmation of the accuracy of the short and long-run coefficients.

Table 4 summarizes the NARDL estimation for the different development outcomes. The second lag of GDP growth per capita significantly impacts economic development with a negative sign to suggest income convergence in Panel 2-A. This result supports the neoclassical growth theory on income convergence [13]. FDI does not impact per capita growth but has the expected negative sign. There is evidence to suggest that corruption and corruption controls in the asymmetric form have significant effects on GDP per capita. The fourth lag of corruption and corruption control is positively and significantly related to per capita GDP in the general NARDL model. The long-run results in Panel 2-B are in tandem with the general output as the relationship is positive. In the short-run, however, there is divergence as corruption and corruption controls dampen GDP per capita and are both significant from the first to the third lags. These results confirm the bound test results that suggest a relationship between the variables of interest.

Table 4 NARDL estimation output, Long-run and short-run results for Development outcomes
Full size table

In Panel 2-C, FDI negatively affects GW in the general output of the NARDL model. The first lag of FDI including the third and fourth lags are significantly related to the reduction in the intertemporal welfare of Ghanaians. This result further supports the pollution haven hypothesis and empirically syncs with the assertion of [28]. The asymmetric analysis establishes a strong effect of corruption and corruption controls on economic development in the long run but with a weak effect in the short run. Yet, on sustainability, corruption weakens genuine wealth per capita in the short run, but the long-run asymmetric effect is inconclusive (see Panel 2-D). Ghana has good blueprints on corruption on paper, but in reality; the commitment is questionable. Therefore, it does not come as a surprise that at one point, the strategies to combat corruption seem to be at work; at another time it is otherwise. This has the propensity to affect the intertemporal well-being of the masses.

The post-estimation diagnosis of the non-linear ARDL model residuals for the long and short runs asymmetric relation present no evidence of serial correlation and heteroskedasticity as both the Breusch-Godfrey serial correlation LM test and ARCH effects were insignificant. The model stability is creditably good (see Fig. 3).

5 Conclusion and policy implications

This study demonstrates that the effect of corruption on development might be dissimilar depending on how development is perceived. This study uses the linear and nonlinear ARDL bounds test estimation technique to cointegration with annual time series data (1980–2023) from WDI, ALFRED/IMF and Transparency International. Two different hypotheses are formulated from the literature and tested to the affirmative. The symmetric analysis reveals that corruption stimulates economic development in the long run, but reduces economic progress in the short run. However, the symmetric effect of corruption on sustainability is consistently negative in the short and long runs. This outcome supports the findings of [16, 46]. Moreover, financial innovation and FDI are the possible channels by which the intertemporal welfare of Ghanaians could be reduced significantly. The behaviour of FDI (MNEs) and financial innovation negatively impacts GW. This supports the pollution haven hypothesis and the detrimental effect of financial innovation on the greenhouse effect. These results support the findings by [11, 22, 25, 36, 37] who found the activities of MNEs (FDI) and financial innovation to be depressing to environmental sustainability. The asymmetric analysis establishes a strong effect of corruption and corruption controls on economic development in the long run but with a weak effect in the short run. Yet, on sustainability, corruption weakens genuine wealth per capita in the short run, but the long-run asymmetric effect is inconclusive. This supports studies such as [17, 38]. Ghana has good blueprints on corruption on paper, but in reality, the commitment is questionable. Therefore, it does not come as a surprise that at one point, the strategies to combat corruption seem to be at work; at another time it is otherwise. This has the propensity to affect the intertemporal well-being of the masses.

. The results espoused by this study have important policy implications for practitioners and policymakers in sub-Saharan Africa and Ghana in particular. Without a shred of doubt, corruption is inimical to sustainability, but for its lag effect, it may seem to grease the wheels of development. Therefore, policymakers and development practitioners must pay attention to the lag effect of corruption on any discourse that focuses on development broadly. Moreover, decision-makers must intensify their seriousness and pay attention to the SDGs with recourse to the expedition of trial on environmental-related corruption cases. Moreover, the work of anticorruption agencies must also evolve to focus on the natural environment. The office of the special prosecutor must go beyond investigating political abuses and witch-hunting to dealing with all manner of environmental-related abuses. In the event of the heightened incidence of public sector corruption, there is an urgent need to institute measures to sanitize the public space through an effective anticorruption court to fast-track the trial of environmental-related corruption cases without delay to serve as a deterrent. Strict adherence to environmental policies by MNEs must also be encouraged to stem the tide of the greenhouse effect (GHG).

While the results presented in this study offer new insights into the corruption–-development perspectives, it is equally important to highlight its shortcomings. To begin with, the findings in this study are limited to the Ghanaian context for the period considered. Secondly, it limited itself to the perceived corruption in the public space and not the actual act of corruption. Future studies can rely on data from the International Country Risk Guide (ICRG) to replicate the procedure used in this study on Ghana or any emerging economy in Africa. The ICRG publishes actual counts of corruption monthly for over 170 countries and territories globally. Thirdly, this study did not estimate the multiplier effects of corruption and corruption controls on the different perspectives of development. Future studies can explore the asymmetric effect of corruption on sustainability with the multiplier effect to indicate the adjustment behaviour of the indicators of sustainability to a unit change in positive and negative components of corruption and corruption controls. Other diagnostic checks for the acceptance or rejection of the effect of the long and short run using the coefficients of the Wald test could be done to complement the multiplier graph.

Future studies can also explore a broader set of controls that are institutionally embedded to capture the stringency of government policy in addressing policy issues related to sustainability at the micro level in any of the countries in sub-Saharan Africa. A different measure of sustainability such as the Sustainable Development Index (SDI), the Biocapacity Reserve/deficit (BCR) and the Environmental Sustainability Index (ESI) published by the United Nations and the Global Footprint Network could also be used to explore this relationship. There is also the need to investigate the specific items in the constituent of financial development and how they affect sustainability in Africa. Country-level evidence should be encouraged to bring to the fore the downside risk of embracing technology without recourse to environmental consequences in the financial sector.