This study examines the impact of digital banking adoption and the COVID-19 crisis on the profitability of Swedish banks, testing whether digital banking mitigated the pandemic’s adverse effects.
Using a panel dataset of Swedish banks, the study employs ordinary least squares, fixed effects (FE) and feasible generalized least squares (FGLS) models to analyze the effects of digital banking and the COVID-19 crisis on banks’ profitability. It further evaluates digital banking’s moderating role during the COVID-19 crisis.
Digital banking exhibits a modest positive association with profitability, though its economic magnitude is marginal. The COVID-19 crisis has had no significant direct impact, and digital banking does not mitigate pandemic-related profitability shocks. Moreover, past profitability and capital adequacy emerge as robust predictors of profitability.
These insights inform strategies for balancing technological innovation with financial stability in turbulent economies. Banks operating in advanced digital economies, like Sweden, may prioritize factors such as capital adequacy and persistent profitability during crisis times, while digitalization should be treated as part of a broader strategy rather than a single profit generator.
This study contributes to the literature by examining the interplay between digital banking and the COVID-19 crisis in the unique context of Sweden, offering new insights into the role of digitalization in crisis resilience and challenging prior findings on the protective effect of digital banking during economic downturns. By triangulating results across multiple estimators, the study advances methodological rigor in profitability studies, highlighting the resilience of traditional bank fundamentals over digitalization during systemic shocks.
1. Introduction
The COVID-19 crisis emerged as the most substantial economic shock to the global economy and its sectors since the Second World War (Naseer et al., 2023a, b). During the COVID-19 crisis, digital banking experienced massive growth throughout all life domains. Social distancing combined with restricted movement forced the closure of physical bank branches, which in turn increased digital banking service requirements (Dadoukis et al., 2021; Doran et al., 2022; Kwan et al., 2024; Silva et al., 2023). Survey results revealed significant growth in first-time online accounts alongside mobile deposits and payments during the pandemic, while web and mobile banking usage increased substantially (Troy, 2020). It is often discussed that rapid digital banking adoption during the pandemic served as a measure to reduce the negative financial impact on banks’ performance. However, empirical studies on the issue are scarce.
Previous research has mainly analyzed the separate influence of the COVID-19 crisis and digitalization on banking performance. The various metrics, such as return on assets (ROA), return on equities (ROE) and cost-to-income ratio, have been employed to measure the impact of the COVID-19 crisis on banks’ profitability because of its diverse aspects. Research on this topic frequently demonstrates that the COVID-19 crisis adversely affected banking financial performance (Elnahass et al., 2021; Hladika, 2021; Kozak, 2021). Regarding digital banking, previous research indicates that digital banking positively impacts bank profitability through various studies (e.g. Campanella et al., 2017; Del Gaudio et al., 2021; Nguyen-Thi-Huong et al., 2023; Rega, 2017; Tunay et al., 2015). Research about the combined effect of digitalization and the COVID-19 crisis on bank performance has recently appeared mainly in developing country contexts. The studies indicate that digitalization positively impacted banks’ performance throughout the COVID-19 pandemic (e.g. Dadoukis et al., 2021; Kwan et al., 2024; Silva et al., 2023).
Against this background, this study aims to explore both the separate and combined effects of the COVID-19 crisis and digital banking on Swedish banking profitability through empirical research. The unique combination of no lockdown policy and advanced digital banking infrastructure makes Sweden an ideal case to study digital banking and the effects of the COVID-19 crisis on banking profitability. During the pandemic, Sweden maintained a distinctive approach by choosing not to impose lockdown measures. Sweden is a digital nation with advanced skills and early adoption of digital banking services. At the same time, Internet usage reaches 100% among its 16–64 age population, and digital payment apps attract 90% of its population (Svenskana och Internet, 2024). The Swedish banking sector has strong supervisory systems and resistance to outside disturbances (Andersson and Jonung, 2024; Lindblom et al., 2011). During the COVID crisis, the Swedish Central Bank (Riksbank) conducted extensive asset purchases alongside interest rate reductions to support the financial industry (Andersen et al., 2022). International comparisons show that Sweden maintained a minimal fiscal burden of less than 3% of GDP through its less stringent measures. However, other countries such as the UK, Italy and France experienced much larger deficits of 27%, 17 and 16%, respectively, of their GDP (Andersson and Jonung, 2024).
The study analyzes data from 19 Swedish banks spanning from 2007 to 2022 using ordinary least squares (OLS), feasible generalized least squares (FGLS) and fixed effects (FE) models to handle heterogeneity and endogeneity concerns. The analysis shows that bank profitability demonstrates a statistically meaningful but economically limited positive relationship with digital banking through Internet banking utilization rates. The COVID-19 crisis fails to generate any statistically relevant effects on profitability. This study diverges from previous research by demonstrating that the interaction between digitalization and the COVID-19 crisis remains statistically insignificant, meaning digitalization failed to reduce or intensify the pandemic’s influence on bank profitability. The study’s results match Sweden’s special situation during the pandemic due to its non-lockdown approach and existing advanced digital banking system, which likely minimized COVID-19 disruptions in bank operations. The study supports most previous studies because it shows that past profitability and capital adequacy consistently produce strong and meaningful effects on present bank profitability.
The results from this study provide valuable theoretical and practical insights about digital banking alongside external crises and their impact on bank profitability. Digital banking produces a small yet positive impact on profitability, supporting previous research because digitalization optimizes operations while extending reach to customers, thus maintaining bank performance regularly. The insignificant adverse effect of the COVID crisis on profitability, together with the non-existent protective impact from digital banking, contradicts previous research that shows crises negatively affect financial institutions and digitalization serves as protection during crises. Theoretical models of digital transformation in banking need to include contextual elements such as strong digital infrastructure and supportive regulations because these factors reduce the role of digital banking as a crisis management tool. Practically, the results indicate bank digitalization should be treated as part of a multifaceted strategy instead of being seen as an independent profit generator. Capital adequacy and previous performance are more vital determinants of bank profitability than digital transformation during crises such as the COVID-19 pandemic. Swedish banks operating in advanced digital systems must maintain operational stability, but less developed markets should accelerate digital growth to reduce upcoming operational challenges.
In the following sections of this paper, a review of previous literature is provided in Section 2. Section 3 explains the research methodology, including the sample, the variables and the econometric model. The analysis results are presented in Section 4, and Section 5 discusses the results, their theoretical importance and their limitations. Section 6 explicitly discusses this study’s practical implications for managers, policy makers and regulators.
2. Literature review
Previous research relevant to this study can be grouped into three areas: the effects of digital banking on banks’ profitability, the impact of the COVID-19 crisis on banks’ profitability, and the effects of digital banking and the COVID-19 shock on banks’ profitability.
Regarding the effects of digital banking on banks’ profitability, many previous studies that measured the direct impacts of digital banking on banks’ profitability using measures such as ROA or ROE have reported a positive effect (e.g. Campanella et al., 2017; Del Gaudio et al., 2021; Nguyen-Thi-Huong et al., 2023; Rega, 2017; Tunay et al., 2015). For example, the studies of Campanella et al. (2017), Del Gaudio et al. (2021) and Tunay et al. (2015) on 28 European countries or the study by Rega (2017) on 38 European banks have all found a positive effect of Internet banking adoption on banks’ profitability. Similar results have also been observed in countries with less developed Internet infrastructure. For example, Nguyen-Thi-Huong et al. (2023), employing a dataset of 32 commercial banks in Vietnam for 2010–2021, found that a 1% increase in digitalization led to a 0.2–0.6% increase in banks’ profitability. The study by Potapova et al. (2022) on 16 Russian banks revealed that banks that have more transactions through digital channels have a higher ROA. The research conducted by Kumar (2022) on Indian banks from 2009 to 2019 also found a positive relationship between the number of online transactions and the ROA. However, at a broader systems level, research cautions that advances in digitalization should be balanced against monetary and financial-stability considerations (Belke and Beretta, 2020). Digitalization is a complement, not a substitute, to sound fundamentals.
The impact of the COVID-19 crisis on banks’ profitability has been measured using several different metrics, such as ROA, ROE and cost-to-income ratio, to capture its multifaceted nature. The results of these previous studies have also generally indicated that the COVID-19 crisis has had an adverse effect on banks’ profitability (e.g. Elnahass et al., 2021; Gazi et al., 2022; Hladika, 2021; Kozak, 2021). For instance, Elnahass et al. (2021) found a negative growth in the ROA and ROE in the early years of the pandemic, along with higher levels of insolvencies and the cost-to-income ratio. Other studies, such as Kozak (2021), have found that supply-side constraints, such as reduced loan demand and restricted borrowing, negatively affected net interest income and fee income. Furthermore, crisis-related credit and liquidity risks were also observed to have increased; for instance, banking profits in Croatia decreased by 53.1% in 2020 (Hladika, 2021). Similarly, Tran et al. (2022) established that the pandemic brought about more accounting risks and more volatility in returns, with a one percent increase in COVID-19 cases reducing the z-scores of banks by 2.51%. Colak and Öztekby (2021) also reported that smaller banks with a low capital base were more vulnerable to funding risk. The researchers also found that the variation in regulatory frameworks affects resilience; for instance, more stringent pre-pandemic capital and liquidity requirements made some regions perform better (Danisman et al., 2021), whereas banks in the European region faced severe challenges of sovereign risk and high levels of non-performing loans (Gazi et al., 2022).
Although most studies have analyzed the effects of the COVID-19 crisis or digitization on various banks’ profitability metrics, studies investigating the combined effects of the two on banks’ profitability and their intricate relationship are very scarce. The few studies in this area have reported a positive impact of digitization on banks’ profitability during the COVID-19 crisis (e.g. Dadoukis et al., 2021; Kwan et al., 2024; Silva et al., 2023; Stefanovic et al., 2021). For instance, Stefanovic et al. (2021) observed that banks that were more focused on digitalization and sustainability in their strategies were more likely to be profitable during the COVID-19 crisis. Another study by Dadoukis et al. (2021) on the US banks found that banks with higher pre-pandemic IT investments were less affected by the crisis. Kwan et al. (2024), in a study of banks in the US during the pandemic, found that banks with better digital banking services attracted more customers and significantly higher increases in deposit flows during the COVID-19 crisis. In a study by Doran et al. (2022) on 10 Central and Eastern European Union countries (CEEC) during the 2010–2021 period, they observed that an increase in the use of Internet banking and the security of bank servers produced positive effects on the performance measures of banks such as ROA, ROE and non-performin loan (NPL). Complementing these findings, additional studies emphasize that digital maturity moderated COVID-related profit pressures (e.g. Ghani et al., 2022; Dinu and Bunea, 2022). Consistent with this, the COVID period cross-country evidence shows that policy support and pre-existing adoption materially shaped the surge in digital payments (Mansour, 2022), highlighting how demand- and supply-side frictions determine the realized “digital buffer.”
At the same time, the magnitude of any digital buffer appears context dependent. Where lockdowns were stringent and branch access severely constrained, digital channels plausibly substituted for physical networks and yielded larger short-run benefits. In contrast, in economies with very high pre-COVID digital penetration and limited operational disruption, incremental insulation from digitalization may be modest, with profitability primarily driven by fundamentals such as capital strength, asset quality and profit persistence. Moreover, previous research links digital economy performance to institutional and regulatory quality (Aldieri et al., 2025), suggesting that strong institutional baselines, like Sweden’s, condition how far digitalization can mitigate crisis period shocks.
Taken together, prior work indicates that digitalization is generally associated with higher profitability and can mitigate the disruptions during the COVID-19 period. Still, the effect varies with lockdown severity, baseline digital adoption and institutional quality. This creates a clear gap for Sweden, a digitally mature, non-lockdown economy with robust supervision, where the counterfactual disruption averted by digital channels may be limited. Our study addresses this gap by estimating the direct association of digital adoption with profitability and the moderating effect of digitalization on the COVID-19 period profitability within a bank-panel framework that accounts for persistence and FE.
3. Research methodology
3.1 Sample
The research data was obtained from Swedish commercial banking institutions. The annual reports of these institutions served as the source for obtaining bank-specific data. The financial statements, including income statements and balance sheets, provided detailed information to us. The analysis excluded banks with less than ten years of data and any missing cases to handle missing data and outliers. The research includes 19 commercial banks from 2007 to 2022, resulting in 304 observations. The analysis incorporates micro- and macro-level factors through industry-specific data from credible institutions such as Statistics Sweden (SCB) and the Swedish Central Bank (Riksbanken). The extensive dataset provides sufficient evidence to demonstrate bank profitability patterns.
3.2 Econometric model
The assessment of bank profitability regarding internal and external factors employs a linear model, which serves as the baseline model according to the literature (e.g. Athanasoglou et al., 2008; Dietrich and Wanzenried, 2011; García-Herrero et al., 2009). The baseline model appears in Equation (1), where
represents the profitability of bank i at time t measured by return on assets, ROA,
c indicates the constant term,
denotes the effect of digital banking with coefficient β1,
is a dummy variable for the COVID period with coefficient β2,
represents a vector of bank-specific variables with coefficient ,
captures macroeconomic and industry-specific variables with coefficient , and
indicates the error term capturing for other unobserved factors.
To examine whether digitalization altered profitability specifically during the COVID period, we augment the baseline with an interaction term. The coefficient represents the moderating effect of digital banking on the COVID-19 crisis.
We also control for unobserved, time-invariant factors, such as the specific bank or institutional characteristics, by estimating a FE model. This is shown in Equation (3) below, where is the bank-specific FE:
In addition to OLS and FE, we estimate a FGLS model to address heteroscedasticity and autocorrelation issues that can produce biased results in cross-sectional panel data (Wooldridge, 2002). The maintained Equation and error structure are:
where:
, with ,
E = 0,
Var
Cov
Equations (4) specify that, beyond absorbing time-invariant bank heterogeneity via , the idiosyncratic error follows a bank-specific AR(1) process with bank-specific variance and no cross-bank correlation. FGLS addresses this by transforming and weighting the system using estimated and (e.g. a Prais–Winsten–type transformation), delivering more precise inference. We therefore present FGLS estimates alongside OLS and FE to triangulate results under complementary assumptions about the disturbance process.
3.3 Variables
Table 1 illustrates all the variables that the study employed. ROA is the measure of bank profitability used in this study, as recommended by many previous studies (e.g. Almaqtari et al., 2019; Athanasoglou et al., 2008), since it shows how well a bank uses its assets to generate profits. The digital banking variable (DIG) is operationalized through the proportion of the Swedish population that uses online banking services. The effect of the COVID-19 crisis is measured by a dummy variable (COV), which is equal to 1 for 2020–2022 and 0 for other periods. The following bank-level, macro-level and industry-level control variables were included.
List of variables
| Variable | Definition | Proxy/Measurement |
|---|---|---|
| ROA | Profitability | Return on assets (ROA): net income/total assets |
| L.ROA | Lagged profitability | Capture the influence of past profitability |
| DIG | Digitalized banking | Share of population using internet banking |
| COV | COVID crisis | Dummy variable: 1 for period 2020–2022, 0 otherwise |
| CAP | Capital ratio of the bank | Equity/Total Assets |
| SIZ | Size of the bank | Natural logarithm of total assets |
| AGE | Age of the bank | Natural logarithm of years since establishment |
| GDP | Real GDP | Inflation-adjusted GDP (economic output) |
| MSP | Broad money supply (M3) | Growth of cash, deposits and other liquid assets |
| BSD | Banking sector development | Banks total assets/GDP |
As for the bank-level factors, we used capital adequacy (CAP), bank size (SIZ) and age (AGE). The previous research findings suggest that a higher capital ratio is associated with a higher profitability (Borio et al., 2017; Edin et al., 2025; Goddard et al., 2004). The effect of bank size on profitability is not apparent; some studies indicated that size has a positive impact on profitability (Borio et al., 2017; Demirgüç-Kunt and Huizinga, 1999), while others suggested that profitability increases with size at the beginning and then decreases (Athanasoglou et al., 2008; Cardone-Riportella et al., 2013). Likewise, the effect of a bank’s age on its performance is not apparent; some works found that age has a positive impact on the performance of a bank (Berteji and Hammami, 2016; Derbali, 2015), while others suggested that new banks are more efficient than old banks (Beck et al., 2005; Dietrich and Wanzenried, 2011). In addition, previous research has established that bank profits will likely carry over to the next period (Edin et al., 2025; Dietrich and Wanzenried, 2011; Trujillo-Ponce, 2013). In this regard, we also include lagged profitability and L.ROA in our model.
At the macro and industry levels, we control for real GDP (GDP), the growth of money supply (MSP) and the development of the banking sector (BSD). The results of previous research have indicated that there is a positive relationship between economic growth and profitability due to higher credit expansion, investment and quality of borrowers (Athanasoglou et al., 2014; Bikker and Hu, 2002; Demirgüç-Kunt and Huizinga, 2001). Results regarding the effect of MSP on bank profitability are not precise, with both positive impacts (e.g. Haron, 1996; Kutsienyo, 2011) and negative impacts (e.g. Badaruddin and Ariff, 2009). Finally, the sophistication of the banking sector, captured through a measure of banking sector development, is assumed to support better bank performance by increasing market opportunities, improving resource utilization and encouraging innovation (Tan and Floros, 2012; Tan, 2016).
4. Empirical results
The descriptive statistics of all the variables are presented in Table 2. The mean of ROA is 0.018 with a low standard deviation, but the minimum and maximum values vary considerably. This means that all the banks have positive returns, but some have had better returns than others at some point in time. The adoption of digital banking services (DIG) has an average of 78.812, representing the good digital infrastructure in Sweden. Figure 1 further compares the tendencies of digital banking adoption to the mean ROA of the banks. The use of digital banking services increases steadily in Sweden, which is known for its digital development during the sample period. ROA first declines sharply with the financial crisis of 2008, while digital adoption keeps increasing. Since 2013, profitability has shown some improvement, but it does not always follow the rising trend of digital adoption. This gap suggests that while the use of digital banking may have advantages, other economic factors and the characteristics of the bank also affect the profitability measures.
Descriptive statistics
| Variable | Obs | Mean | Std. Dev | Min | Max |
|---|---|---|---|---|---|
| ROA | 304 | 0.018 | 0.039 | −0.009 | 0.525 |
| DIG | 304 | 78.812 | 7.944 | 57.000 | 87.000 |
| COV | 304 | 0.187 | 0.391 | 0.000 | 1.000 |
| CAP | 304 | 0.120 | 0.089 | 0.005 | 0.834 |
| SIZ | 304 | 23.974 | 2.352 | 19.375 | 28.686 |
| AGE | 304 | 78.868 | 68.408 | 4.000 | 195.000 |
| GDP | 304 | 2.856 | 4.325 | −5.800 | 11.200 |
| MSP | 304 | 7.325 | 4.218 | 1.800 | 16.300 |
| BSD | 304 | 2.202 | 0.255 | 1.757 | 2.852 |
Note(s): ROA: Return on Assets (ROA), DIG: Digital Banking, COV: COVID Criss Dummy, CAP: Capital Adequacy, SIZ: Bank Size, AGE: Bank Age, GDP: Real GDP, MSP: Money Supply, BSD: Banking Sector Development
The horizontal axis of the line graph shows years, from 2007 to 2022, with tick marks at 2005, 2010, 2015, 2020, and 2025. The vertical axis ranges from negative 4 to 4 in increments of 2. The graph has two lines: a black line labeled in the legend as “R O A (mean)” and a red line labeled as “D I G.” The black line starts at (2007, 3.2), drops sharply to (2009, negative 0.28), rises slightly at (2013, negative 0.13), increases to form a sharp peak at (2014, 1.03), drops to (2020, negative 0.91), and ends at (2022, negative 0.74). The red line starts at (2007, negative 2.68), increases in a concave down manner to (2011, negative 0.048), and continues with a small hump to (2015, 0.15). It increases to form a peak at (2017, 1.09), and drops to end at (2022, 0.67).Adoption of digital banking in relation to banks’ profitability. Source(s): Author’s own work
The horizontal axis of the line graph shows years, from 2007 to 2022, with tick marks at 2005, 2010, 2015, 2020, and 2025. The vertical axis ranges from negative 4 to 4 in increments of 2. The graph has two lines: a black line labeled in the legend as “R O A (mean)” and a red line labeled as “D I G.” The black line starts at (2007, 3.2), drops sharply to (2009, negative 0.28), rises slightly at (2013, negative 0.13), increases to form a sharp peak at (2014, 1.03), drops to (2020, negative 0.91), and ends at (2022, negative 0.74). The red line starts at (2007, negative 2.68), increases in a concave down manner to (2011, negative 0.048), and continues with a small hump to (2015, 0.15). It increases to form a peak at (2017, 1.09), and drops to end at (2022, 0.67).Adoption of digital banking in relation to banks’ profitability. Source(s): Author’s own work
The correlation matrix appears in Table 3, while Table 4 presents the variance inflation factor (VIF) test results for multicollinearity. The analysis reveals moderate relationships between most variables without any strong cases of severe multicollinearity. The multivariate analysis reveals more potent DIG (digital banking) and COV (COVID crisis dummy) effects on profitability because these variables show low correlations with other variables. The positive correlation between ROA and L.ROA indicates that banks that performed well in the past tend to maintain strong current profitability levels. The positive correlation between ROA (profitability) and CAP reaches 0.645, which indicates that banks with better capital ratios tend to achieve higher profitability.
Matrix of correlations
| Variables | (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| (1) ROA | 1.000 | ||||||||||
| (2) DIG | −0.095 | 1.000 | |||||||||
| (3) COV | −0.096 | 0.353 | 1.000 | ||||||||
| (4) L.ROA | 0.757 | −0.178 | −0.089 | 1.000 | |||||||
| (5) CAP | 0.645 | −0.035 | −0.009 | 0.569 | 1.000 | ||||||
| (6) SIZ | −0.191 | 0.148 | 0.117 | −0.192 | −0.567 | 1.000 | |||||
| (7) AGE | −0.317 | 0.119 | 0.085 | −0.310 | −0.013 | −0.211 | 1.000 | ||||
| (8) GDP | −0.013 | 0.321 | 0.002 | −0.072 | 0.025 | 0.046 | 0.038 | 1.000 | |||
| (9) MSP | −0.046 | 0.360 | 0.652 | −0.020 | 0.032 | 0.098 | 0.079 | 0.055 | 1.000 | ||
| (10) BSD | −0.014 | 0.510 | −0.036 | −0.032 | 0.015 | 0.063 | 0.056 | 0.003 | 0.174 | 1.000 |
Note(s): ROA: Return on Assets (ROA), DIG: Digital Banking, COV: COVID Criss Dummy, CAP: Capital Adequacy, SIZ: Bank Size, AGE: Bank Age, GDP: Real GDP, MSP: Money Supply, BSD: Banking Sector Development
Multicollinearity test
| Variable | VIF | 1/VIF |
|---|---|---|
| CAP | 2.29 | 0.436 |
| COV | 2.10 | 0.477 |
| DIG | 2.07 | 0.484 |
| MSP | 1.90 | 0.528 |
| L.ROA | 1.81 | 0.552 |
| SIZ | 1.71 | 0.584 |
| BSD | 1.62 | 0.617 |
| AGE | 1.23 | 0.811 |
| GDP | 1.22 | 0.822 |
Note(s): ROA: Return on Assets (ROA), DIG: Digital Banking, COV: COVID Criss Dummy, CAP: Capital Adequacy, SIZ: Bank Size, AGE: Bank Age, GDP: Real GDP, MSP: Money Supply, BSD: Banking Sector Development
Table 5 presents the results obtained from the baseline models, which include OLS, FE and FGLS. Digital banking (DIG) shows no statistical significance in OLS. Still, it demonstrates a 5% level of significance in FE and a 10% level in FGLS, indicating positive effects of digital banking service adoption on Swedish bank profitability. The economic importance of digitalization appears restricted, which means digital banking’s impact on profitability remains small or depends on specific circumstances. However, the COVID-19 crisis (COV) shows no significant effects across all three models, reflecting the Swedish banking sector’s resilience during the crisis.
The baseline models
| OLS | FE | FGLS | |
|---|---|---|---|
| DIG | 0.000 | 0.000** | 0.000* |
| COV | −0.003 | −0.000 | −0.000 |
| L.ROA | 0.331*** | 0.158*** | 0.473*** |
| CAP | 0.145*** | 0.261*** | 0.047*** |
| SIZ | 0.001*** | −0.006** | 0.000** |
| AGE | −0.003*** | −0.011 | −0.001** |
| GDP | −0.000 | −0.000 | 0.000 |
| MSP | −0.000 | −0.000 | −0.000 |
| BSD | −0.003 | −0.003 | −0.001 |
| Cons | −0.040** | 0.120** | −0.017** |
| R-square | 0.66 | 0.64 | |
| F value | 63.32 | 50.85 | |
| F (sig) | 0.000 | 0.000 | |
| Hausman test χ2 | 42.28 | ||
| Hausman test (sig) | 0.000 | ||
| Wald χ2 (9) | 315.27 | ||
| Wald χ2 (sig) | 0.000 | ||
| n | 285 | 285 | 285 |
Note(s): ROA: Return on Assets (ROA), DIG: Digital Banking, COV: Covid Criss Dummy, CAP: Capital Adequacy, SIZ: Bank Size, AGE: Bank Age, GDP: Real GDP, MSP: Money Supply, BSD: Banking Sector Development
Where *** means a level of significance of %1; ** means a level of significance of %5; and * means a level of significance of %10
The variable lagged profitability (L.ROA) demonstrates high statistical significance at the 1% level across every model, indicating that profitability tends to persist from one period to the next. The bank-specific variables capital adequacy (CAP) and size (SIZ) demonstrate significance in all models, where CAP shows a positive relationship with profitability. Still, SIZ produces conflicting results, positive in OLS and FGLS but negative in FE. The study period indicates that macroeconomic variables (GDP, MSP and BSD) have no significant impact on profitability. The Hausman test (χ2 = 42.28, p = 0.000) confirms that the fixed effect model performs better than the random effect model with R-squared values at 0.64, which indicates a strong model fit. The FGLS model provides additional robustness against panel-specific disturbances.
The results in Table 6, which examines the moderating effect of digital banking (DIG) on the COVID crisis (COV), provide further insights into their combined impact on Swedish banks’ profitability (ROA). The interaction term DIG × COV is negative across all models (−0.001 in OLS, −0.003 in FE and FGLS). Still, it lacks statistical significance, which means that digital banking did not substantially offset any potential adverse effects of the crisis. The variables like L.ROA and CAP continue to show strong positive effects (significant at 1%). The model fit, R-square of 0.66 in OLS, 0.64 in FE and diagnostic tests (e.g. Wald χ2 of 317.44 in FGLS) remain robust.
The moderating effect of COVID on digital banking
| OLS | FE | FGLS | |
|---|---|---|---|
| DIG | 0.000 | 0.001** | 0.000** |
| COV | 0.061 | 0.273 | 0.261 |
| DIG × COV | −0.001 | −0.003 | −0.003 |
| L.ROA | 0.331*** | 0.157*** | 0.474*** |
| CAP | 0.145*** | 0.261*** | 0.047*** |
| SIZ | 0.001*** | −0.006** | 0.000** |
| AGE | −0.003*** | −0.011 | −0.001** |
| GDP | −0.000 | −0.000 | −0.000 |
| MSP | −0.000 | −0.000 | −0.000 |
| BSD | −0.003 | −0.004 | −0.002 |
| Cons | −0.040** | 0.122** | −0.018** |
| R-square | 0.66 | 0.64 | |
| F-value | 56.78 | 45.70 | |
| F (sig) | 0.000 | 0.000 | |
| Wald χ2 (9) | 317.44 | ||
| Wald χ2 (sig) | 0.000 | ||
| n | 285 | 285 | 285 |
Note(s): ROA: Return on Assets (ROA), DIG: Digital Banking, COV: Covid Criss Dummy, CAP: Capital Adequacy, SIZ: Bank Size, AGE: Bank Age, GDP: Real GDP, MSP: Money Supply, BSD: Banking Sector Development
Where *** means a level of significance of %1; ** means a level of significance of %5; and * means a level of significance of %10
Overall, the results reveal how digitalization provided benefits, yet these advantages were insufficient to overcome the general market challenges during the pandemic.
5. Conclusion
This study examined whether digitalization cushioned Swedish banks’ profitability during the COVID-19 period using a panel of 19 banks over 2007–2022 and three complementary estimators (OLS, FE, FGLS). The results are consistent across models: digital banking adoption is associated with a small, positive effect on ROA; the COVID dummy has no direct statistically significant effect; and the DIG × COV interaction is negative but insignificant, indicating that higher digital adoption did not materially mitigate pandemic period profitability pressures in Sweden’s high-digital, non-lockdown context. Profitability displays persistence (lagged ROA is strongly positive), capital adequacy is a robust positive predictor, bank size shows model-dependent signs and macro/industry controls are muted. Together, the evidence points to fundamentals (capital strength and earnings persistence), rather than incremental digitalization, as the primary drivers of profitability during systemic stress in digitally mature settings.
Our results both accord with and qualify prior evidence. The small but positive association between digitalization and profitability is consistent with studies documenting efficiency and reach gains from online and mobile channels (e.g. Campanella et al., 2017; Del Gaudio et al., 2021; Rega, 2017; Nguyen-Thi-Huong et al., 2023), although the economic magnitude in Sweden is modest. By contrast, we do not find the broadly reported COVID period profitability deterioration (e.g. Elnahass et al., 2021; Kozak, 2021; Hladika, 2021), a difference plausibly linked to Sweden’s non-lockdown stance and robust supervisory and monetary backstops during 2020–2022. Most notably, our insignificant DIG × COV term diverges from evidence that digital maturity cushioned performance in settings with stricter mobility restrictions or lower pre-pandemic digital penetration (Dadoukis et al., 2021; Doran et al., 2022; Silva et al., 2023; Stefanovic et al., 2021; Kwan et al., 2024). We interpret this discrepancy as context-dependence: in a digitally saturated, operationally undisrupted environment, the marginal protective effect of digital channels is limited, and fundamentals, capital strength and earnings persistence dominate short-run profitability dynamics.
The results generate important theoretical implications about how digital banking interacts with external crises and bank profitability. The theoretical models of banking digital transformation must consider country-specific elements such as mature digital systems and favourable regulatory environments. These factors reduce digital banking’s crisis-mitigating power. Research on digital strategy efficacy for bank performance during economic shocks needs to investigate how institutional and market-specific factors interact with digital transformation initiatives.
While the study reveals significant findings about digital banking and crisis management, it also has specific research constraints that guide future studies. The single proxy for digital banking (DIG), which tracks Internet banking adoption, fails to represent the full scope of digitalization since it does not measure fintech investments or advanced technologies like AI and blockchain. Future research should adopt broader measurement approaches to describe digital transformation more accurately. The findings about the COVID-19 crisis (COV) effects and the DIG × COV interaction may not generalize because they result from Sweden’s exceptional no-lockdown strategy and digital system development, thus necessitating future studies across different economic settings. Moreover, the study examines only the COVID-19 crisis between 2020 and 2022, therefore omitting possible delayed profitability impacts, and future studies should analyze extended periods or examine how additional crises affect digitalization-performance relationships.
6. Practical implications
The study also provides critical implications for bank managers and policy makers who face challenges in digital transformation and crisis resilience. Digital banking adoption maintains a steady positive relationship with profitability. Yet, its lack of significant moderating power in the COVID-19 crisis indicates that digital transformation may not provide sufficient protection from systemic disturbances. Banks should dedicate ongoing funds to digital infrastructure development since these investments create individual benefits yet need complementary capital reserves and operational flexibility to absorb unexpected disruptions. The continuous impact of previous profitability levels demonstrates the need for businesses to adopt strategic planning and focus on efficiency improvement. The COVID-19 crisis has had a minimal direct effect on results, which indicates that economic disturbances primarily affect banks through secondary economic mechanisms, thus requiring organizations to adopt flexible risk management systems. The strategic approach to financial management, combining digital tools with strengthened financial stability alongside adaptive governance, proves essential for maintaining profitability across everyday and crises.
For regulatory bodies and policymakers, the results caution against assuming that digital readiness insulates bank profitability during system-wide stress in high-adoption environments. Prudential authorities should prioritize well-calibrated capital and liquidity buffers and ensure that supervisory stress tests emphasize margins, credit losses and operational continuity, not just channel substitution. Policy efforts that build “digital public goods” (secure digital identity, interoperable payments, open-banking rails and cyber defence) are valuable complements. Still, they do not replace the need for strong fundamentals. In less digitally advanced regions, targeted policies that accelerate safe digital uptake (infrastructure, inclusion and consumer protection) may deliver larger stabilization benefits in a crisis. Across contexts, statistical agencies and supervisors can improve risk monitoring by collecting bank-level, harmonized measures of digital intensity (e.g. active-user ratios, IT/cyber investment) so that macroprudential policy can distinguish where digitalization materially shifts risk and where traditional buffers remain the primary line of defence.
