Dynamic capabilities are aligned with the organization’s dynamic vision, based on the evolution of resources and skills ( Wang & Ahmed, 2007 ). Thus, a company that is fully capable of integrating, building and reconfiguring its internal and external competencies to cope with a rapidly changing environment is likely to have a better competitive edge. In this sense, dynamic capabilities in the context of digitized tax procedures, refer to the ability to develop skills in response to changes in manual methods, to the ability to acquire and develop resources to respond to the digitization process. Teece et al. (1997) proposes a 4-step process that has been widely adopted in the literature. The first stage consists in observing and assessing the environment with a view to identifying market opportunities or technological innovations to be seized. Once an opportunity has been identified, the second step is to implement the appropriate means to seize it, through a process of resource acquisition. This is followed by the renewal of resources and, finally, the reconfiguration of internal resources. Dynamic capabilities, which apply to managerial issues and disruptive market innovations, are therefore a source of performance for companies that manage them strategically. Dynamic capabilities and performance appear as two intrinsically linked and even inseparable notions, insofar as the former serves to explain the sources of the latter. We therefore put forward the following hypotheses:

H1: Dynamic capabilities have a positive and significant influence on business process performance.

The relationship between dynamic capabilities and information systems is an integral part of a company’s core digital strategy. The use of digitized tax procedures therefore implies that digitalization is effectively integrated into these processes, i.e., that it enables tax-related information to be collected, summarized and forwarded to the tax authorities for processing. The dynamic capability of companies to mobilize and deploy the information, knowledge and skills they need to keep pace with changes in the tax environment, and to make effective use of the tax authorities’ electronic services, is therefore of paramount importance. We put forward the following hypothesis.

H2: Dynamic capabilities have a positive and significant influence on the use of digitized tax procedures.

Delone and McLean (2016) presents the “use” as the degree of integration of information systems into the business processes of the user and the work routine of each individual. This construct measures the use of tax e-services, developed thanks to ICT and implemented by tax authorities. In this sense, the permanent use of digitalized tax procedures implies a change in the company’s processes, working methods and organization, in other words, a restructuring of internal processes that can lead to the achievement of the objectives set by the company in terms of time, cost and quality optimization. Therefore, we propose the following hypothesis:

H3: The use of digitized tax procedures has a positive and significant impact on the performance of business processes.

The technology boom in sub-Saharan Africa has lagged behind other continents, possibly due to the region’s economic, technological and infrastructural backwardness. Nevertheless, African nations are determined to catch up and have made strides in improving Internet access, cell phone usage, and digital transformation. This provides economic agents with opportunities to create new markets and extend existing ones. Singh and Sahu trace the first attempts at e-governance projects back to the 1970s. Developed countries quickly grasped the importance of these changes and embraced them, benefiting today from the positive spin-offs of digitized procedures. Sub-Saharan African nations were slow to embrace the digitization of public services, with tax digitization projects not gaining traction until the 2010s. Nonetheless, sub-Saharan African countries have been implementing measures to digitalize their tax administration processes over recent years, and Cameroon is no exception to this trend.

The digitization of tax procedures in Cameroon was executed through the launch of the “Fiscalis” platform in 2014 by the Directorate General of Taxation (DGT) under the supervision of the Ministry of Finance. This platform offers several tax services, such as remote declaration, remote payment of taxes due, and retrieval of supporting documents issued by the DGT, including tax notices and tax clearance certificates ( DGI, 2023 ). The need for change in the tax administration of Cameroon stems from the necessity to revitalize public services in this area, given that the tax environment is a crucial aspect of the country’s business climate. Cameroon is currently ranked 181st out of 190 countries on the “payment of taxes” indicator, according to the Doing Business ranking of 2020. This ranking justifies the need to implement policies aimed at restructuring Cameroon’s tax environment by means of dematerializing tax procedures. After several years of operation of online tax activity platforms in Cameroon, it is timely to evaluate their impact on the performance of SMEs.

1) Questionnaire design

We selected Cameroonian SMEs that benefit from digitized tax services as our target population, as our study focuses on the impact of the use of digitized tax procedures on the business process performance of Cameroonian companies. Within Cameroon’s entrepreneurial ecosystem, these companies are the most significant taxpayers. Moreover, e-governance in Cameroon is undergoing significant transformations. Several e-governance projects, whether focused on modernizing tax and customs services or facilitating the payment of government department services through electronic payments, are in development.

To conduct this quantitative study, we developed a questionnaire. It was designed based on a literature review that centered on theories and concepts related to the research. The questionnaire has three parts, each measuring a variable of the model we are assessing (see Appendix). The first part discusses the dynamic capabilities of the company. It consists of 21 measurement items distributed amongst four sub-constructs. The second part assesses the use of digitized tax procedures using seven items. The third section addresses business process performance, which is evaluated through 18 items categorized into four sub-constructs. 7-point Likert scale that ranges from “completely disagree” to “completely agree” as a measurement of each item ( Boone & Boone, 2012 ).

2) Pre-test phase

It is important to carry out a pre-test to verify the comprehensibility of the first draft of the questionnaire ( Taherdoost, 2019 ). We aim to determine whether the items that measure the constructs are easily understood by the respondents. To achieve this, we gathered a sample of ten individuals comprising five master’s students, two civil administrators, and three contractors. At the end of their evaluations, they confirmed that the questionnaire was comprehensible and unambiguous.

3) Minimum sample size determination

Afterwards, we determine the minimum sample size through the GPower 3.1.9.2 software, which has the particularity of calculating the statistical power for various statistical tests ( Schoemann et al., 2017 ). By inserting parameters such as effect size, probability error, the power of the effect, and the number of predictors or latent explanatory variables, we obtained the sample size necessary to test the model effectively. For this, with a total of five (05) predictors (lower constructs), a probability of error α = 0.05, an effect power (1 − β) = 0.80, and an effect size of f² = 0.15, the minimum total sample size for this study is ninety-two (92) individuals.

4) Pilot test

To ensure the reliability and construct validity of the study, we conducted a pilot test after obtaining a sample of 45 responses. The benefit of pilot testing is that it helps us rectify the questionnaire by removing measurement items that are statistically unreliable or rephrasing them, if possible, to make them more understandable ( Hair Jr. et al., 2014 ). After analyzing the sample data, no changes were made as the results indicated that both the measurement items and the model are statistically reliable and valid. So, we continued our data collection.

5) Data collection phase

We designed the questionnaire on Word and printed several copies. We conducted an on-site visit directly to the SMEs office. We made sure to clarify the purpose of the study, explain the anonymity of the questionnaire and assure the respondents that the data would be used solely for academic research purposes. We administered the questionnaire to individuals responsible for tax compliance within the company after their agreement to participate in the study. Each SME completed one questionnaire. 200 questionnaires were distributed among SMEs and we received a total of 143 responses that represented 143 different SMEs. Thus, a response rate of 71.5%. Data was collected between November 2021 and April 2022.

6) Common method bias

To detect the possible effect of common method bias (CMB), we performed the full collinearity test Kock (2015) . This test suggests that the presence of a variance inflation factor (VIF) greater than 3.3 indicates pathological collinearity and also contamination of the model by CMB. After we conducted the full collinearity test, the results showed that all the VIF values were lower than 3.3, thus confirming the absence of CMB and the reliability of the data collection.

The data for this study were analyzed using partial least squares structural equation modeling (PLS-SEM) using SmartPLS 4 software ( Ringle et al., 2022 ). This method has the advantage of evaluating complex models with small sample sizes ( Hair et al., 2019 ). It is particularly well known in research in marketing, management, entrepreneurship, information systems, etc. We chose a higher-order construct model ( Sarstedt et al., 2019 ) to reduce the number of relationships in the model and to evaluate the most important concepts of this research. Thus, dynamic capabilities and business process performance are higher-order constructs, each explained by four lower-order constructs. We used the disjoint two-stage approach ( Sarstedt et al., 2019 ). In the first stage, we evaluated the reflective measurement model of the lower-order constructs, and in the second stage, we used the latent variable scores of the lower-order constructs to create and estimate the higher-order constructs. We then evaluated the measurement model and the structural model of the research model.

The results show that the majority of SMEs surveyed are private limited companies (54%). 48% have between 6 and 20 employees. Most of the managers responding to the survey were men (70%). 46% are aged between 31 and 40, and 34% have an executive function ( Table 2 ).

The purpose of this assessment is to check the model’s constructs for their reliability and validity. To measure internal consistency, we look at the outer loading, rho_A, and composite reliability (CR), which must be 0.7 or greater. To evaluate convergent validity, we check the average variance extracted (AVE), which should be 0.5 or greater ( Hair et al., 2019 ). In the first stage, we checked these criteria for the lower-order components. Items with outer loading values below 0.7 and/or VIF values above 3 were removed to improve the reliability and avoid multicol-linearity issues. Since the lower-order component model fulfilled all criteria, we utilized their latent variable scores to create the model displayed in Figure 1 during stage two. The assessment of the reflective measurement model of the higher-order constructs adhered to the same standards. Table 3 shows that the model is reliable and valid since all criteria met the required values.

Source: Authors.

Discriminant validity was assessed by the Fornell-Lacker criterion ( Fornell & Larcker, 1981 ), which suggests that the square root of the AVEs, along with the diagonal, must be greater than all the values below and left. Table 4 shows that discriminant validity is not a problem in the research model.

With the satisfactory measurement model in place, the subsequent step is to assess the structural model. In this regard, we first examine collinearity to ensure that the results are not biased. We evaluated the Variance Inflation Factor (VIF) values, which indicate probable collinearity issues when they exceed 5 and possible collinearity issues when they range from 3 - 5. Ideally, the VIF values should be closer to 3 or lower ( Hair et al., 2019 ). The results indicate that there is no issue with collinearity in the research model, as the internal VIF values for measurement items in the lower-order constructs are below 3.

Next, we examined the model’s explanatory power using R². R² values of 0.75, 0.50, and 0.25 are considered substantial, moderate, and weak, respectively ( Hair et al., 2019 ). Results ( Table 5 ) indicate that 78.6% (R² = 0.786) of the business process performance construct is explained by dynamic capabilities and digitized tax procedures use. Additionally, the use of digitized tax procedures construct was explained by dynamic capabilities at 26.2% (R² = 0.262).

Finally, the bootstrapping method was used to determine the significance and statistical relevance of the path coefficients. The results ( Table 6 ) confirm all hypotheses, demonstrating that dynamic capabilities have a positive and significant influence on business process performance (H1; β = 0.675, p < 0.001) and the use of digitized tax procedures (H2; β = 0.511, p < 0.001). The use of digitized tax procedures has a positive and significant influence on business process performance (H3; β = 0.299, p < 0.001).

^****p < 0.001; ^***p < 0.01; ^**p < 0.05; ^*p < 0.1; n.s. not significant.

The results also show that the use of digitized tax procedures completely mediates the relationship between dynamic capabilities and business process performance. In other words, the dynamic capabilities of SMEs have an indirect influence on the performance of business processes when these SMEs make use of electronic tax services ( Table 7 ).

The estimation model is shown in Figure 2 below.