Cross-border banking and bank stability: evidence from Sub-Saharan Africa

Abstract

Based on bank-level data from 29 Sub-Saharan African countries between 2005 and 2019, we apply panel fixed effects (FE) and two-step system GMM estimators to investigate whether increased cross-border banking affects domestic banking sector stability. We find significant evidence that the stability of banks in host countries declines with an increased presence of foreign banks—and the impact is more pronounced on banks that are small and less efficient. The stability impact of foreign banks is also found to depend on the quality of governance institutional factors in the host country. The findings shed some important insights on the downside of financial liberalisation policy in developing countries and the need for increased cross-border collaboration between home and host supervisory authorities in the SSA region—especially in jurisdictions where the foreign bank affiliates are systemically important. The domestic supervisory authorities thus need to effectively manage the inherent trade-off between reaping the benefits from international financial integration while effectively safeguarding domestic banking systems against cross-border contagion and fragility.

Introduction

Cross-border banking, i.e. the presence of foreign banks as either subsidiaries, affiliates or branches in a host country, is a prominent feature in most Sub-Saharan African (SSA) countries.¹ Statistics from the World Bank’s Global Financial Development Database [3] show that the relative share of foreign banks (in terms of assets) in SSA substantially grew from about 30.6% in 2000 to 66.3% in 2005 and stood at around 64.2% in 2019.² In particular, after the global financial crisis (GFC), foreign banks with headquarters in Africa—the commonly called Pan African Banks (PABs), have become more dominant in the region than international banks from outside Africa following de-risking by latter banking groups. Figure 1 shows the trend in foreign bank penetration in SSA since 2005.

Fig. 1

Source: Global Financial Development Database (2019) and Bankfocus

Trend in average foreign bank penetration in SSA (2005–2019).

The traditional view has been that foreign banks bring benefits to host countries, such as increasing competition and access to financial services and stabilising credit quantity during domestic turmoil [4,5,6]. Foreign banks also promote efficiency in host countries’ banking systems by enhancing the quality of financial regulations, imparting specialised knowledge and expertise and introducing modern technology [7, 8]. Foreign banks can contribute towards financial stability in the host market because they are better capitalised, liquid and well-diversified [1]. Banking sector solvency and liquidity improve because foreign banks are better capitalised than their domestic peers and depositors’ trust in the stability of foreign institutions makes local bank run less likely [9]. The foreign banks can also mitigate the risk of sudden stops and capital flow reversals as parent banks will provide the needed international liquidity in crisis periods to safeguard their investments in the respective host countries [10]. Given the low levels of financial intermediation in most banking sectors of developing countries, many small economies thus stand to gain substantially from cross-border banking in terms of financial deepening and increased outreach if they take advantages of the financial innovation and efficiency gains that foreign banks can bring.

However, the increased expansion of these banks across countries also implies that risks can easily be transmitted across borders—the affiliates of the foreign banks could thus serve as a channel of contagion among countries [11, 12]. The expansion of foreign banks also shifts responsibilities of bank supervision from the home countries of foreign banks to host supervisory agencies. The domestic supervisory authorities thus face two policy objectives with inherent trade-offs: reaping the benefits from international financial integration while effectively safeguarding domestic banking systems against cross-border contagion and fragility. To date, there is still ongoing debate in both academic and policy circles as to whether the benefits (in terms of competition and efficiency) that are associated with globalisation through cross-border banking outweigh the cost of greater financial instability. The expansion of foreign banks across countries thus raises relevant policy questions about the role played by these banks, specifically, whether allowing their participation introduces more instability in the host countries.

This paper investigates how increased foreign bank penetration affects domestic banking sector stability in SSA. It specifically examines whether the impact is heterogeneous across the different types of banks or is conditional on the quality of governance institutional factors in the host country. The focus on SSA is relevant from two perspectives: First, most firms in SSA are bank dependent for capital financing, as capital markets are still undeveloped in most countries. Ensuring a stable and sound banking sector is thus an important policy objective for the authorities in the region. Previous episodes of bank failures and crises in Africa (for example, Nigeria (1999; 2009–2011), Zambia (1995–1999), Zimbabwe (1995–1999), Kenya (1993–1995), Cote d’Ivoire (1988–1991) and Ghana (2017–2020) resulted in huge economic costs and significant losses, setting back financial development.³ Secondly, the region has one of the highest penetrations of foreign banks in the world and most subsidiaries of the foreign banks are also systemically important (in terms of share of assets, deposits and payment/settlement transactions) in the host countries where they operate [9]. Financial stability risks in this case are elevated in two ways: first, the financial sectors of home countries of some PABs may be fragile and vulnerable—for instance, due to changes in commodity prices arising from limited economic diversification [8].⁴ This implies that the parent banks may not be in a position to provide the required financial support to their subsidiaries in times of need. Second, the legal framework on consolidated supervision in most SSA countries is generally inadequate and in some cases non-existent [13]—yet the increased cross-border banking connectivity has in some cases, resulted in creation of complex, interrelated banking structures that require adequate supervision on a consolidated basis. Despite the limited effects of the GFC on SSA banking systems, the crisis laid bare the challenges and constraints that supervisory authorities can face to resolve cross-border shocks that simultaneously disrupt banking systems in multiple countries.

We find significant evidence that the stability of banks in host countries declines with an increased presence of foreign banks—and the impact is more pronounced on banks that are small, less efficient and less capitalised. The current study thus contributes towards generating new evidence on the stability implications of cross-border banks in developing countries and sheds important policy insights on the downside of financial liberalisation policy in developing countries. It also highlights the need for increased cross-border collaboration between home and host supervisory authorities in the SSA region—especially in jurisdictions where the foreign bank affiliates are systemically important.

The rest of the paper is organised as follows: Sect. 2 provides an overview of foreign banks and bank stability in SSA over the past two decades. Section 3 summarises the relevant literature, while Sect.  4 describes the variables and data that were used in the study. Section 5 provides the study methodology, and in Sec. 6 we discuss the empirical results. Section 7 provides the study conclusion and policy implications.

Overview of foreign banks and bank stability in SSA

There have mainly been two categories of foreign banks operating in SSA: those originating from jurisdictions outside Africa and those emanating from within Africa. The former banks had established the presence mostly in countries that were either former colonies of Britain, France or Portugal—although, over time, they have also expanded beyond their colonial roots [9]. The second group consists of African cross-border banks, commonly referred to as Pan African Banks (PABs). The holding companies of these banks are headquartered in an African country with various subsidiaries spread across different countries in Africa. Most of the PABs originate from South Africa, Morocco, Kenya and West Africa states (e.g. Nigeria and Togo)—see Table 1. Consequently, the share of foreign banks in these countries is considerably lower—ranging between 20 and 30 per cent [3]. However, given the ownership linkages, these countries are the main sources of the possible transmission of cross-border banking risks to different parts of the continent and beyond.

Table 1 Major Pan-African Banks in Africa.

According to the World Bank Development Database, the countries that experienced significant entry of foreign owned banks between 2005 and 2019 include Malawi (from 43 to 62.5%); Mali (from 38 to 75%); Namibia (from 43 to 60%); Rwanda (from 38 to 62%); and Senegal (from 64 to 85%). In other countries (e.g. Zambia, Burkina Faso, Benin, Ivory Coast), the percentage of foreign banks remained high throughout the period at over 80%, while the banking systems of Madagascar, Lesotho and Eswatini were completely (100%) dominated by foreign banks. In terms of systemic importance, the IMF Africa Department (2015) estimates that foreign banks in SSA control more than 30% of total assets and customer depositors in most jurisdictions. The supervisory oversight model in most countries is such that the foreign subsidiaries are licensed as separate banking entities—making them subject to regulatory rules independent from their parent holding companies. Although this arrangement helps to ring-fence domestic risk, it also has implications for cross-border banking supervision, especially in times of crisis management or resolution. If collaboration is absent or weak between home and host regulatory agencies, the resolution of cross-border banks could become problematic. The situation could be aggravated if the distressed bank is considered systemically important in the host market but not within the banking group itself.

Overall, banks in SSA are characterised by higher capitalisation and excessive liquidity [3]. The average capital ratio for SSA banks between 2010 and 2019 was about 18% and significantly surpassed that prescribed by the Basel Committee on Banking Supervision (BCBS) of 8.0%. The average liquidity ratio (as measured by liquid assets to short-term labilities) for SSA banks was 76.7%, compared to below 35.0% for banks from other developing regions. Table 2 shows comparative financial soundness indicators in SSA and other global regions.

Table 2 Selected financial soundness indicators—SSA and other global regions (2010–2019).

The above stability indicators, however, mask a serious financial stability problem that has characterised the banking sectors of SSA countries since the GFC: banks in SSA have historically, on average, recorded high levels of non-performing loans (NPLs) compared to those from other developing regions.⁵ Increased cross-border banking implies that credit shocks can easily be transmitted across countries. Given the nascent levels of consolidated supervision in most developing countries and the large and visible costs of financial instability from the previous episodes, the avoidance of banking crises or failures is thus a high priority policy area for the authorities in the region.

Literature review

Theoretical literature on the implications of foreign banks in the host country is often analysed from three perspectives: (i) the impact on banking sector stability and/or shock transmission to domestic markets, (ii) the impact on access to finance/credit intermediation and (iii) the impact on banking sector efficiency and competition. Our objective in this paper is mainly aligned with the first strand of the literature.

Foreign banks can contribute towards financial stability in the host market because they are better capitalised, liquid and well-diversified [1]. This makes them more stable credit providers in case of shocks or disruptions in the host country. These banks also tend to have advanced risk management skills and often bring higher regulatory standards because they come from advanced economies with more stringent regulatory regimes [8]. There are, however, contrasting views in the literature that suggest other mechanisms or forces through which the beneficial effects of foreign banks in the host market could potentially be off-set. First, domestic banks may be adversely affected by the shift of customers after the entry of foreign banks [14]. On the one hand, foreign banks may “cherry-pick” good, less risky customers—leaving the domestic banks with mainly high risky clients [15]. If the domestic banks’ advantage over ‘soft information” in the market is not adequate to shield them from customer defaults, their asset quality would then decline, resulting in increased financial instability in the host country. On the other hand, because foreign banks have superior service and global reputation, depositors may transfer their savings out of domestic banks and into the foreign ones—the so-called flight-to-quality effect [16]. As a result, domestic banks may have to incur higher costs in order to either attract more deposits or replace deposits with other sources of funding. The rising cost of raising deposits could in turn, thus, compel domestic banks to increase their lending rates—resulting in adverse selection [14]. Increased foreign bank penetration may therefore make domestic banks more vulnerable as a result of pressure from both sides.

Second, higher presence of foreign banks may increase competition in the host market and reduce banks’ franchise value due to lower profitability [14]. The traditional view that higher franchise value would limit banks’ incentive to take excessive risk would no longer be effective under such circumstances as banks are induced to assume more risk and search for “higher yield” to restore their profits. Increased competition from foreign bank entry could thus result in worsening of the quality of the asset portfolio in the domestic market and hence exacerbate instability [17]. Foreign bank presence can also be detrimental for financial inclusion if a cream-skimming effect occurs and reduces the global access to credit [18]. Such phenomenon happens if foreign banks cherry-pick more profitable and transparent clients, pushing domestic banks to reduce their activities.

Empirical evidence is also mixed. While several pre-crisis studies tend to find evidence of stabilising effect of foreign banks (e.g. [19, 20] and [21]), most post-crisis studies find that increased presence of foreign banks is negatively associated with financial stability (see for example [22,23,24]. Studies that adopt a more granular approach by focusing on transmission of cross-border liquidity shocks from international banking groups also find that foreign banks have a destabilising effect on domestic bank stability [2, 25]. Other studies examine the stability implications of cross-border banking in terms of increase or decrease in credit risk. For instance, Beck and Brown [26] found that foreign banks in Central and Eastern Europe exhibited “cherry-picking” in their lending behaviour. Claessens and Van Horen [27] document that foreign banks only seem to have a negative impact on credit in low-income countries, in countries where they have a limited market share, where enforcing contracts is costly and where credit information is limited. De Haas and Van Lelyveld [28] found that multinational banks increase the risk of importing instability from abroad. The authors observed that during the global financial crisis, the multinational bank subsidiaries slowed down credit growth about twice as fast as domestic banks—especially for subsidiaries of banking groups that relied more on wholesale-market funding. Leon and Zin [18] found that in general, the presence of Pan-African banks increased access to credit to firms in Africa—regardless of the firm size.

A few other studies establish that the stability impact of foreign banks appears to be context specific. For instance, Dwumfour [23] observed that cross-border banks increase domestic banking stability only in periods of crises. In normal times, a higher level of cross-border banking is associated with low capital ratios and higher non-performing loans—both of which are indicators of higher banking instability. In their study [24] find that overall, increased presence of PABs in the West African Economic and Monetary Union (WAEMU) was linked to increased banking sector volatility—however, different results were observed when the analysis was disaggregated: banks with French origin appeared to be more stable when competition was high. Kusi et al. [29] find that foreign bank presence promotes banking stability in Africa, but the positive effect is only enhanced in countries with strong corporate governance institutions. In his study of determinants of bank stability in Africa, Ozili [30] find that countries with greater foreign bank presence experience fewer non-performing loans ratio and higher banking solvency. The findings however are dependent on the banking stability proxy employed and the period of analysis, i.e. pre-crisis, during-crisis or post-crisis. The World Bank in its 2013 Financial Sector Assessment Program (FSAP) for countries in the East African Community (EAC) documents that the performance of foreignbanks in these countries differed based on their origin. Subsidiaries of banks with headquarters in EAC had reported strong profitability performance compared to subsidiaries of foreign banks from other regions. Cihák et al. [31] find that increased cross-border banking tends to improve system stability in a country where the banking sector has relatively few linkages to other banking sectors, but only to a certain extent. After that, increases in cross-border banking linkages start to have negative effects. Kimunio et al. [32] found that foreign-owned banks had no important direct effect on financial stability in Kenya—instead, financial stability was observed to be more depended on the credit policies of the banks and their balance sheet structures, regardless of the ownership type.

In general, although there appears to be extensive work on how foreign banks affect credit quantity and quality in developed countries and other emerging regions, research on the financial stability implications of foreign banks in SSA is relatively scanty. The few available studies also do not explore the heterogeneity in the stability impact of foreign banks. The empirical approach adopted in the current study allows for examining whether the different types of banks in the domestic market are affected differently when there is increased penetration of foreign banks. Such a heterogeneous analysis has important policy implications for supervisory authorities—in terms of both cross-border supervisory cooperation and informing the appropriate approach to regulating banks in the host country. Our study also explores whether the quality of governance institutional factors in the host country conditions the stability impact of foreign banks. In this way, our work contributes towards a more comprehensive understanding of stability implications of foreign banks in the SSA region.

Variables and data

Bank stability

We follow extant banking literature (see for example [33,34,35]) and measure banking sector stability by the Z-score. The Z-score is an indicator of insolvency or bankruptcy and shows the extent to which the value of assets must fall before a bank becomes insolvent. The score is derived by applying the following formula:

$$Z_{it} = \frac{{{\text{ROA}}_{it} + {\text{EA}}_{it} }}{{\sigma {\text{ROA}}_{it} }}$$

(1.1)

where ${\text{ROA}}_{it}$ is the return on assets, ${\text{EA}}_{it}$ is the equity-to-assets ratio, and $\sigma {\text{ROA}}_{it}$ is the volatility in assets (measured by the standard deviation of ROA). The subscripts i and t show bank and year, respectively. Higher indices of the Z-score imply more stability and vice versa. In order to account for the time volatility in the denominator of the Z-score, we calculate the standard deviation of the ROA based on a three-year rolling window [33]. Unlike other measures (such as the distance to default and equity volatility), which can be calculated only for listed banks, the Z-score can be used to measure the probability of default even for unlisted banks. Most banks in our sample are not listed, the Z-score is thus a more appropriate stability measure in this regard. However, for robustness checks, we follow literature (see for example, [33, 35]) and also employ two alternative stability indicators: the return on assets (ROA) ratio and the ratio of loan loss reserves to gross loans (LLR). Lower or negative values of ROA show that a bank can fall into the threat of bankruptcy. Therefore, when the ROA decreases the chances of insolvency (instability) increases and vice versa. On the other hand, higher values of LLR indicate poor credit quality, hence more risk, i.e. less stability.

Cross border banking

In line with extant literature, we employ three measures cross-border banking: (i) the percentage of assets held by foreign banks in the host country, (ii) the percentage of foreign owned banks to total number of banks in the host country and (iii) a foreign ownership dummy that takes a value of one for each foreign-owned bank in the sample at each point in time and zero otherwise in line with the approach adopted by [36]. The first two measures, though related, allow for capturing two slightly different dimensions of foreign banks. For instance, the argument in literature that foreign banks enhance competition in host markets [37] could better be explained if these banks control a bigger share of the banking market in the domestic country. In that case, the foreign banks would be in direct competition with domestic banks. The latter measure on the other hand, can be more insightful in explaining the “cherry-picking” behaviour of foreign banks [38] or shifting of customers from domestic to foreign banks even when the latter do not have a larger market share [39]. The third indicator allows for observing the impact of cross-border banking at bank level (as opposed to the first two measures that are both measured at the country level). This variable thus identifies the direct effect of foreign ownership on the stability of individual banks in the host market [36].

We also consider other factors that may affect banking sector stability, namely size, bank concentration (share of assets held by the three largest banks), funding structure (ratio of non-deposit funding to total funding), leverage (equity-to-assets ratio), GDP growth and the regulatory environment. These variables have been selected among the most popular indicators employed in other empirical studies on the foreign bank-stability relationship. Large banks benefit more from market power and diversification and often possess advanced risk management systems [7, 40]. As a result, large banks are likely to post and enjoy more stable earnings and hence be more financially stable than smaller banks. Larger banks could however be more risky when they behave in a manner that they are “too-big-to-fail” or “too-important-to-fail” [41]. A higher level of bank concentration may induce instability if it results in collusion among banks and higher loan rates [17], but can also improve stability if the banks are more efficient and profitable in a concentrated market [42]. Higher reliance on non-deposit funding may increase bank stability if the sources are less volatile; however, it could increase instability if deposits are more stable than other short-term borrowings. Lower leverage (high equity-to-assets ratio) provides a better cushion to banks for unforeseen losses [43]—resulting in less risk. Higher GDP growth rate enhances the profitability of banks due to general improvement in income levels in a country; hence, a positive relationship is expected with bank stability [41]. However, bank managers also tend to relax credit underwriting practices when the economy is expanding [44]—as such, a booming economy could be associated with more instability in the banking sector. A sound regulatory environment promotes banking stability by creating predictable and effective remedies to commercial disputes and bank operations [45]. On the other hand, cumbersome, protracted or unclear banking rules and legal frameworks negatively affect banking operations and economic growth. Table 12 contains a summary of all the variables that were used in the study and their sources.

Data

The data on individual banks and cross-border banking indicators were sourced from Bankfocus, Global Financial Development Database and Claessens and Van Horen database [27]. The country-level variables were obtained from IMF International Financial Statistics and World Bank Governance Indicators database. The sample comprised 270 banks from 29 countries between 2005 and 2019—see Table 13 for list of the countries. Table 3 shows a descriptive summary of the variables. It shows that there are wide variations in cross-border banking penetration in SSA. The distribution of the Z-scores also highlights some significant variations in banking sector stability levels including technically insolvent banks as evidenced by some negative Z-scores.

Table 3 Summary of descriptive statistics.

Table 4 shows the correlation among the variables. The Z-score and the cross-border banking measures are negatively correlated with each other and statistically significant at the 5% level. There is little correlation among the explanatory variables—multicollinearity should thus not be a problem in the estimations.

Table 4 Correlation matrix.

Methodology

The baseline empirical model is estimated at the bank level and specified as follows:

$${\text{STAB}}_{ijt} = \alpha_0 + \alpha_1 {\text{FBP}}_{jt - 1} + \mathop \sum \limits_{k = 2}^N \theta_k {\text{BANK}}_{kit - 1} + \mathop \sum \limits_{k = 1}^2 \gamma_k {\text{COUNTRY}}_{kjt} + v_i + \mu_{it}$$

(1.2)

where i, j and t represent bank, country and time, respectively. ${\text{STAB}}_{ijt}$ represents stability (Z-score), ${\text{FBP}}_{jt}$ is a measure of foreign owned banks (i.e. either ratio of foreign banks to total number of banks in the host market, or the share of assets held by foreign banks in the host country or a foreign bank ownership dummy that takes the value of one when a bank is foreign-owned and zero otherwise). As the impact of foreign bank penetration on domestic banking risk more likely emerges after a time lag, one-year lagged values of foreign penetration are used in the estimation. ${\text{BANK}}_{kit}$ are bank-specific characteristics; ${\text{COUNTRY}}_{jt}$ represents country-specific factors; $v_i$ is the bank-specific fixed effect; and $\mu_{it} ,$ is the error term. Equation 1.2 is estimated with 1-year lags of bank-specific variables to mitigate endogeneity [33].

We employ a bank-specific panel fixed-effects estimator with Driscoll–Kraay standard errors that are robust to heteroscedasticity, within-panel serial correlation as well as cross sectional dependence in the panel error residuals [46]. In the current study, cross-sectional dependency could, for instance, emerge because of, among other factors, trade between nations, common financial integration, commodities and monetary shocks, and other unobserved cross-border spillover effects. The panel fixed-effects estimator is chosen not only because it is commonly adopted in extant literature but also because of some of its merits. First, as we are using bank-level panel data, fixed effects model allows unobservable bank-level individual effects, which may be heterogeneous across banks and constant over time. Second, fixed effects model allows the bank-level time-invariant effects to be correlated with explanatory variables, which is supported by the result of Hausman test.

It can however be argued that in estimating Eq. (1.2), identification issues could arise due to the dynamic nature of bank stability and the potential endogeneity of the foreign ownership variables [36]. Banking systems might be characterised by information opacity, regulatory reforms, market power and relationship lending—all of which may cause persistence in the cost structure and profitability of banks—and hence, on stability. Endogeneity of the foreign ownership variables can arise from both reverse causality and omitted variable bias. For instance, foreign banks may be more inclined to enter markets where domestic banks are more fragile, since they incur lower costs for mergers and acquisitions [14]. This reverse causality leads to biased results. To account for these dynamics and endogeneity problems, an alternative econometric methodology is employed to serve as a robustness check to the baseline results—the two-step system Generalised Method of Moments (GMM) in which Eq. (1.2) is estimated in its dynamic form by including the lag(s) of the dependent variable as an additional explanatory variable.

Results and discussion

The results from the baseline regression model (1.2) are reported in Table 5. Three sets of regressions are estimated: Model 1 uses ratio foreign banks to total banks in a given country (i.e. the number of foreign owned banks divided by the total number of all banks in the host country); model 2 uses share of assets held by foreign banks in the host country; and model 3 uses a foreign ownership dummy at bank-level. As explained in the preceding section, the impact of foreign bank penetration is likely to be felt with a lag—hence our use of lagged cross-border banking indicators in all the models.⁶ The dependent variable in models 1–3 is the Z-score. For robustness purposes, we also present, in models 4–6, the results when using an alternative measure of bank stability—the return on assets (ROA). Models 1–3 and 4–6 thus differ from each other by using different cross-border banking measures.

Table 5 Baseline results of the impact of cross-border banking on banking sector stability in SSA (2005–2019)

The results show that the coefficient of cross-border banking is negative and statistically significant in all the models. Because higher Z-scores indicate more stability, the results show that increased cross-border banking reduces banking sector stability in the host country. The findings are in line with the hypothesis that the risk-increasing effect attributable to cross-border banking in the domestic market, outweighs its potential beneficial impact. This could partly reflect the shifting of customers or more intensive competition or disadvantages on innovative financial services. The significance of the foreign bank presence variable (% number of foreign banks in host country) and the foreign ownership dummy suggests that the mere presence of foreign banks—even without controlling a larger market share, is a strong factor to induce risk taking behaviour in the host market. Some studies [14, 37] report similar findings.

As for control variables, it is first observed that bigger banks are less stable (negative co-efficient of size in all the models)—supporting the too-big-to-fail hypothesis [41]. The impact of capital is positive and significant in all the models—suggesting that higher capitalisation reduces insolvency risk. As discussed in Sect. 2, the relatively higher equity-to-assets ratios in most SSA countries help to cushion bank losses and hence improve overall bank stability [35]. A diversified funding base increases bank stability as evidenced by positive and significant coefficient of the fund structure variable. The coefficient of bank concentration is negative and significant in all the models—interpreted as evidence that increased concentration in the banking market is associated with less banking sector stability. This finding is consistent with the structure-conduct-performance hypothesis. Some studies report similar findings [34, 44]. Finally, banks are less stable when the economy is booming as evidenced by a negative and significant coefficient of GDP growth in all the models. The outcome is consistent with the view that bad debts in the banking system increase when the economy is doing well because bank managers tend to relax credit underwriting practices during such times [44].

Robustness checks

In addition to using an alternative bank stability measure (i.e. the ROA), we also perform two further robustness checks: first, we re-estimate Eq. 1.2 in its dynamic version using 2-step system GMM estimator to check whether our findings are robust to alternative estimation techniques; and second, we use another risk measure—the ratio of loan loss reserves to gross loans (LLR). Unlike the Z-score, higher LLR ratios imply higher risk, i.e. less stability. Exogenous and endogenous variables are classified based on theoretical considerations and other bank risk-related empirical literature (see for example [47]. Current and future bank strategic and business decisions are usually guided by or based on past bank performance. This implies that current and future bank-specific variables are correlated—the current bank performance can thus impact future performance. For this reason, the bank-specific variables are treated as weakly endogenous and instrumented with their first lags. On the other hand, and as stated in Sect. 5 above, there is the possibility of reverse causality between bank stability and foreign bank variables. Supervisory authorities also often enact new rules or policies when they observe fragilities in the banking system—risk levels thus inform or drive regulatory initiatives. For these reasons, the cross-border banking and regulatory variables are treated as endogenous and instrumented with their second and higher lags. Year dummies and the macro-economic variable are treated as exogenous. The key assumption under system GMM is that the lags of instrumented variables are the only available instruments [48]. However, to avoid instrument proliferation, the lag depth of the instruments is restricted and the GMM estimates are reported with “collapsed” instruments. The Hansen statistic and the second-order autocorrelation (AR) test is then applied to verify the validity of the instruments. The results are shown in Table 6 (with the Z-score as the dependent variable in models 1–3 and the LLR ratio as dependent variable in models 4–6). We still find significant evidence that cross-border banking negatively affects banking sector stability—similar to the baseline results in Table 3.⁷

Table 6 Results of the impact of cross-border banking on banking sector stability in SSA (2005–2019) using system Generalised Method of Moments (GMM)

Heterogeneity analysis

After finding a consistent negative relationship between bank stability and foreign bank penetration, we explore the heterogeneity of this impact. We specifically examine whether the impact is: (i) conditioned on the quality of governance in the host country; and (ii) what types of domestic banks are affected more by foreign bank penetration. Some studies, see for example [37, 49, 50] adopt a similar approach. We employ three governance institutional factors, namely the voice of accountability, control of corruption and political stability. These variables were selected based on their relevance to banking sector stability and are among the most popular governance indicators employed in other empirical studies. Voice of accountability reflects the extent of freedom enjoyed or expressed by the citizens and investors (including the freedom to participate in economic and business activities). The index on control of corruption depicts how those in power or authority use their power or mandate for private benefit at the expense of the general public. Political stability depicts the possibility of political unrest and/or violent acts driven by politics. Each factor is measured as an index on a scale of − 2.5 to 2.5, with lower values showing poor quality of governance and vice versa. To capture their conditioning impact, each factor is interacted with either indicator of cross-border banking. As the governance indicators are highly correlated with each other, they are employed one at a time. A composite index to capture the overall conditioning impact of the quality of regulation and the governance on banking stability in the host country is then constructed through principal component analysis (PCA). The idea here is to examine the overall (average) conditioning impact of the institutional governance factors. The regression results showing the conditioning impact of governance institutional factors are shown in Table 7 (when using % number of foreign banks), Table 8 (when using foreign ownership dummy) and Table 9 (when using share of foreign banks' assets). A consistently negative impact of foreign banks penetration on domestic banking stability is observed in all the regressions—similar to those reported in Table 5. The coefficients of the interaction terms in Table 8 are all found to be negative and statistically significant. In Table 7, the interaction terms are significant in model 1 (political stability x foreign banks) and model 3 (accountability x foreign banks), while in Table 9, the interaction term is significant in model 2 (corruption x foreign assets). The coefficient of the interaction term in the last column (i.e. the PCA constructed variable) is also significant in Tables 7 and 9. Overall, the results can be interpreted to mean that higher quality of governance enhance the effect of cross-border banking. In other words, the negative impact of increased foreign bank penetration on banking sector stability in the host country is more pronounced when there is more political stability, less corruption and more economic freedom.

Table 7 Results of the impact of cross-border banking (measured using % number of foreign banks) on banking sector stability in SSA (2005–2019) when conditioned on institutional factors

Table 8 Results of the impact of cross-border banking (measured using a foreign ownership dummy) on banking sector stability in SSA (2005–2019) when conditioned on institutional factors

Table 9 Results of the impact of cross-border banking (measured using share of foreign assets) on banking sector stability in SSA (2005–2019) when conditioned on institutional factors

Next, we explore whether different types of banks in the host country are impacted differently by the presence of foreign banks. We follow the approach by [37] and focus on three aspects: size, efficiency and capitalisation. The banks are first categorised into large and small banks based on whether their size is beyond the median of the sample distribution. Banks whose assets are above the median are classified as large, while those below the median are classified as small. Second, the banks are categorised into 2 sub-samples based on their efficiency level using the same criteria as before. Efficiency is measured as the ratio of total operating costs to total assets [14]. Ratios above the median indicate less efficiency, while those below the median indicate higher efficiency. Finally, the banks are classified based on their capitalisation levels (equity-to-assets ratio) using the same criteria as before. In each of the above three categories, a binary dummy variable is constructed, which is equals to 1 (0) if a bank is classified as a large (small), efficient (less efficient) and capitalised (less capitalised). Including the interaction of this dummy with the indicators of foreign bank penetration, the estimations of Eq. (1.2) are then repeated. The results are reported in Table 10—Part 1 (for larger versus small banks), Part 2 (for efficient versus less-efficient banks) and Part 3 for (capitalised versus less capitalised banks).

Table 10 What types of domestic banks are more affected by cross-border banking penetration?

The results show that the coefficient of the stand-alone penetration indicator (i.e. the baseline impact on small, less efficient and less capitalised banks) is negative and statistically significant in all the regressions, while its interaction with a dummy is significant only in Part 3—a finding that is interpreted in the subsequent section below. There are two explanations to the former results: first, the risk of small and less-efficient banks in the host market increases with the entry of cross-border banks; and second, the increase of cross-border banks’ market share and their mere number exert similar impact on small and less-efficient banks. Smaller banks usually possess less market power—they are thus mostly affected by competition arising from the entry of, and loss of market share to foreign banks. The smaller banks therefore lose their profit share at a relatively larger rate than bigger banks. The less-efficient banks are also the most vulnerable and disadvantaged when foreign banks enter the domestic market as they find it difficult to manage and contain operational costs [37]. Foreign banks have an overall advantage in terms of operating more efficiently than domestic banks—because group synergies in part allow the foreign banks to access relatively cheaper advanced banking technology and resources.

The results in Part 3 can also be interpreted in two ways: First, the consistent negative coefficient of the penetration variable in all the models shows that both, capitalised and well capitalised banks are vulnerable to the adverse risk effect arising from increased presence of cross-border banks in the host market. However, the positive coefficient of the capitalised dummy shows that foreign banks’ market share and their mere number seemingly exert differential impact on capitalised and less capitalised banks. Capitalised banks’ risk is more significantly sensitive to share of assets, while the variation of less-capitalised banks’ risk is more closely associated with mere presence (ownership) of foreign banks.⁸ A possible explanation for this outcome is that the market dominance of well-capitalised banks would only be seriously threatened when foreign banks effectively capture a sizeable portion of the market (higher share of assets or number of foreign banks). Because foreign banks boost competition in the host market [39], profitability of well capitalised banks would decline at a higher pace relative to less-capitalised banks. However, the sheer increase in the presence of foreign competitors—even without having to change the market structure by considerably increasing their asset share, might still cause a significant “cherry-picking” and/or “flight-to-quality” effect on the clients of less-capitalised banks—hence the significantly higher adverse impact still observed in model 3.

Finally, we explore the heterogeneity of the impact of cross-border banking across the banking sectors in three different regions in SSA: Southern African Development Community (SADC), West Africa and East African Community (EAC). The results are presented in Table 11. The negative stability implication of increased foreign bank penetration is more pronounced in SADC and West African region—where, as highlighted in Sect. 2, the share of foreign banks is predominant in most of the countries. On the contrary, the EAC is largely dominated by domestic banks—the impact of foreign banks in this region is thus less buttressed.

Table 11 Impact of increased cross-border banking across the different SSA regions

Conclusion

This paper investigated how increased penetration of cross-border banks affects banking sector stability in SSA. The paper finds significant evidence that increased cross-border banking decreases banking sector stability in the host country—and the impact is more pronounced on banks that are small and less-efficient. The stability impact of foreign banks is also more pronounced when the political environment is stable and investors/citizens enjoy more economic freedom.

Three significant policy implications flow from the findings. First, the overall finding that cross-border banks decrease banking sector stability in the host country suggests that host regulatory authorities need to establish proper channels of collaboration with the home regulatory authorities of foreign subsidiaries. In other words, there is need for enhanced consolidated supervision of banking groups in SSA as an effective remedy to mitigating cross-border contagion and systematic failures. In particular, supervision of banks needs to be strengthened at the national level and cooperation enhanced among home and host countries of cross-border banks. Home–host supervisory collaboration is more paramount in cases where foreign bank affiliates pose systemic risk in the host market—even when they may not be of systemic importance within the banking group itself. In order to effectively carry out consolidated supervision, authorities should improve the availability and regular exchange of relevant information—for example, through Memoranda of Understanding and/or Colleges of Supervisors. In addition to being essential for consolidated supervision, such information and regular cooperation would also allow regulators to move towards a risk-based approach to supervising cross-border banks.

Second, the findings shed some important insights on the downside of financial liberalisation policy in developing countries. Policy makers need to be aware and mindful that allowing entry of foreign banks as part of financial liberalisation is associated with some undesired effects in the host market. The domestic supervisory authorities thus need to effectively manage the inherent trade-off between reaping the benefits from international financial integration while effectively safeguarding domestic banking systems against cross-border contagion and fragility. On the other hand, if the authorities’ objective is to ensure increased impact of foreign banks, then enhancing the quality of governance institutions in the host country is critical. Specifically, ensuring political stability and creating an economic environment that allows investors and the citizens to exercise more freedom are important institutional governance arrangements that enhance the impact of foreign banks.

Table 12 List of variables, definitions and sources

Table 13 Average foreign bank penetration by country (2005–2019)

Finally, the finding that banks of different characteristics in the host country are affected differently by the presence of foreign banks suggests that purposeful regulatory strategies are needed to offset the negative effects of cross-border banking. For instance, measures or policies that improve bank efficiency (for instance through cost minimisation) are critical in mitigating the negative effect of foreign banks in the local market.