Climate change and technological advancement have become major concerns in the contemporary business environment, as they influence organizational performance directly or indirectly ([40]; [24]). In recent years, the adoption of computerized accounting information systems (CAIS) has emerged as a key factor in enhancing operational efficiency and performance in manufacturing firms ([9]; [56]). According to the [88], inefficient accounting and financial reporting systems contribute to 15% - 25% operational losses in manufacturing globally. In Ghana, many manufacturing firms still rely on manual or semi-automated accounting systems, which affect cost control, decision-making, and overall firm performance ([60]; [40]). The continuous rise of digital technologies and the need for accurate financial information have urged firms to adopt CAIS to improve accounting accuracy, real-time reporting, and strategic decision-making ([90]).

Beyond improving financial reporting, CAIS provides firms with tools for enhanced monitoring, data integration, and predictive analytics, enabling managers to identify inefficiencies, control costs, and respond proactively to market changes ([44]; [33]). The integration of CAIS in manufacturing processes also facilitates compliance with regulatory requirements, reduces the risk of fraud, and improves transparency in operations, which in turn strengthens stakeholder confidence ([6]; [3]). Furthermore, CAIS allows firms to generate comprehensive performance metrics that support strategic planning, resource allocation, and operational optimization ([69]). However, while these potential benefits are widely acknowledged, the extent to which CAIS translates into tangible performance improvements in Ghanaian manufacturing firms remains underexplored ([63]; [89]; [30]). The effectiveness of CAIS usually depends on contextual factors such as organizational culture, employee IT competence, and the level of technological infrastructure available to support its implementation ([52]).

Consequently, understanding the mechanisms through which CAIS influences firm performance, as well as the conditions under which these systems are most effective, has become a critical area of research ([41]; [57]). This is particularly important in developing countries like Ghana, where resource constraints, technological gaps, and varying managerial capabilities may limit the potential impact of CAIS on operational and financial outcomes ([68]; [67]). Investigating these dynamics can provide valuable insights for managers, policymakers, and stakeholders seeking to leverage digital accounting technologies to enhance performance, ensure cost efficiency, and maintain competitiveness in the manufacturing sector.

Recent studies have suggested that CAIS adoption can improve cost efficiency, as automated systems reduce errors, minimize resource wastage, and streamline accounting processes ([9]; [44]). By automating routine accounting tasks, CAIS reduces the reliance on manual record-keeping, which not only lowers operational costs but also enhances the accuracy and reliability of financial information ([64]). Improved cost efficiency, in turn, can positively impact organizational performance by lowering operational costs, enhancing resource allocation, and providing managers with timely insights for strategic decision-making ([71]; [10]). However, despite the recognized benefits of CAIS, few studies have examined the mediating role of cost efficiency in the relationship between CAIS and firm performance, leaving a significant gap in understanding the mechanisms through which technology adoption influences performance outcomes, particularly in the context of emerging economies like Ghana ([70]; [32]).

Moreover, technological infrastructure, including IT hardware, software, network systems, and cybersecurity measures, has been highlighted as a critical factor that can strengthen or weaken the effectiveness of CAIS ([82]; [55]). Firms with robust technological infrastructure are better positioned to leverage CAIS for real-time reporting, advanced analytics, and integrated decision-making, whereas firms with weak or outdated infrastructure may struggle to realize the full benefits of computerized accounting systems ([18]; [51]). This suggests that the impact of CAIS on both cost efficiency and overall firm performance is contingent upon the presence of adequate technological support, which acts as a facilitator or constraint in the adoption process.

Extant literature further suggests that organizational and environmental factors, such as managerial support, employee IT competency, training, and technological readiness, can influence the effectiveness of CAIS in improving cost efficiency and firm performance ([21]; [11]). Firms that invest in employee training, system customization, and integration of CAIS with other enterprise systems are more likely to achieve higher performance outcomes. Despite the recognition of these factors, there is limited research that simultaneously examines the mediating role of cost efficiency and the moderating role of technological infrastructure in explaining the link between CAIS and firm performance in manufacturing firms, particularly in Ghana. This represents a critical knowledge gap, as understanding these mechanisms can provide practical insights for managers to optimize resource allocation, design effective accounting systems, and develop strategies that enhance operational efficiency and competitiveness.

Drawing on the Resource-Based View (RBV; [15]), which emphasizes that firms achieve competitive advantage through unique, valuable, and inimitable resources, this study posits that CAIS represents a strategic organizational resource that can enhance cost efficiency and firm performance ([15]). In addition, technological infrastructure acts as a boundary condition that can either amplify or attenuate the impact of CAIS on organizational outcomes, influencing both the direct and indirect pathways through which CAIS affects performance ([62]). Based on this theoretical lens, we propose a moderated-mediation framework in which cost efficiency mediates the relationship between CAIS and firm performance, while technological infrastructure moderates both the direct effect of CAIS on performance and the indirect effect via cost efficiency. By examining these mechanisms, the study contributes to the literature on accounting information systems, organizational performance, and technology adoption, while providing practical guidance for managers seeking to leverage digital tools and infrastructure to achieve superior performance in Ghanaian manufacturing firms.

In this study, the subsequent section presents the literature review and hypotheses development, followed by the research methods. The results and discussion are then provided, and finally, the study concludes with the implications of the findings for theory, practice, and policy in the context of Ghanaian manufacturing firms.

The study examined the relationships among computerized accounting information-system (CAIS) capability, cost efficiency, technological infrastructure, and performance in Ghanaian manufacturing firms. The sampling frame comprised manufacturing firms operating in Ghana that had adopted computerized accounting information systems and had personnel with adequate knowledge of their accounting systems, technological infrastructure, cost-management practices, and organizational performance. A purposive sampling technique was employed to select eligible manufacturing firms from the Greater Accra, Ashanti, Central, Eastern, and Western Regions of Ghana. The participating firms operated in the food and beverages, automation, ceramics, wood manufacturing, and fabrics subsectors.

One knowledgeable key informant was selected from each participating manufacturing firm. Respondents were drawn from accounting, finance, operations, and managerial roles and were required to have worked with their respective firms for at least one year to ensure sufficient familiarity with the organization’s accounting systems and technological environment ([53]). Accordingly, each usable questionnaire represented a distinct manufacturing firm, and the empirical unit of analysis was the firm, represented by a single knowledgeable key informant. This design avoided the problem of multiple respondents being nested within the same organization and therefore did not require within-firm aggregation statistics such as rwg, ICC(1), or ICC(2). Firm performance was assessed through the perceptions of the knowledgeable organizational informant and should therefore be interpreted as key-informant-assessed firm performance rather than as performance derived directly from audited financial statements or other archival measures.

A questionnaire-based survey technique was employed for data collection. Initially, 1000 manufacturing firms were approached, with one questionnaire administered to an eligible key informant in each selected firm. Data collection was conducted between December 2025 and February 2026. Of the 1,000 questionnaires distributed, 652 were returned. Following screening for incomplete and invalid responses and the matching of responses across the survey waves, 638 questionnaires were retained for the final analysis, resulting in an analytical sample of 638 manufacturing firms and 638 key informants. Participation was voluntary, while anonymity and confidentiality were assured throughout the data-collection process ([42]).

To reduce the potential influence of common method variance, the questionnaire was administered in two waves separated by 30 days ([86]; [28]). In Wave 1, respondents provided information on computerized accounting information-system capability and technological infrastructure. Thirty days later, Wave 2 collected information on cost efficiency and performance of the manufacturing firms from the same organizational informants. Responses from the two waves were matched at the respondent and firm levels before the final dataset was constructed. The temporal separation was introduced primarily as a procedural remedy to reduce respondents’ ability to cognitively connect the predictor measures with the mediator and outcome measures.

Importantly, the 30-day time lag should not be interpreted as establishing a causal sequence between cost efficiency and firm performance because both constructs were measured in Wave 2. Similarly, the study did not measure the timing of CAIS adoption or post-implementation adjustment. Consequently, the observed negative relationship between CAIS capability and cost efficiency is interpreted as an empirical association rather than evidence that CAIS adoption initially disrupts firms’ cost structures. Questionnaire items were also randomized, and unrelated filler questions, such as “I enjoy outdoor activities,” were incorporated to reduce respondents’ ability to infer the proposed relationships among the study constructs ([33]). Harman’s single-factor test further indicated that the first factor accounted for 31.12% of total variance, below the 50% reference point, suggesting that common method variance was unlikely to dominate the observed relationships.

With respect to demographic characteristics, 65.83% (N = 420) of the respondents were male, while 34.17% (N = 218) were female. In terms of educational qualification, 46.08% (N = 294) held a bachelor’s degree in accounting, finance, engineering, or related disciplines, 35.11% (N = 224) possessed a master’s degree, while 18.81% (N = 120) held professional or diploma-level qualifications. Regarding age distribution, 61.29% (N = 391) of respondents were between 30 and 39 years, 26.18% (N = 167) were above 40 years, and 12.53% (N = 80) were between 25 and 29 years. Based on work experience, 58.15% (N = 371) had between 1 and 3 years of experience, 24.61% (N = 157) had between 4 and 7 years, while 17.24% (N = 110) had more than seven years of experience with their current organizations.

The data were initially screened for missing values, outliers, normality, and multicollinearity, as violations of these assumptions could bias the results. Following the criteria suggested by [25], no missing values were detected in the dataset. This may be attributed to the fact that the questionnaires were personally administered, and respondents were requested to review their responses before submission. Outliers were examined using the stem-and-leaf technique, and no extreme cases were identified. Normality was assessed through probability-probability (P-P) plots, histograms, and the values of skewness and kurtosis, which were found to be within acceptable thresholds (±1 for skewness and ±3 for kurtosis), as recommended by [22]. Multicollinearity was examined using correlation coefficients and variance inflation factors (VIFs), and none of the correlations exceeded the critical value of 0.85, indicating that multicollinearity was not a concern ([80]). The descriptive statistics has been presented in Table 1.

Table 1.Correlation, mean and standard deviation.

The descriptive statistics and correlation matrix are presented in Table 1. The mean scores were CAIS capability (M = 2.874, SD = 1.199), cost efficiency (M = 4.236, SD = 0.913), technological infrastructure (M = 4.477, SD = 0.967), and perceived firm performance (M = 2.815, SD = 1.143). Thus, respondents reported comparatively lower CAIS capability and perceived performance than cost efficiency and technological infrastructure. CAIS was negatively correlated with cost efficiency (r = −0.591) and technological infrastructure (r = −0.666), and positively correlated with perceived performance (r = 0.450). Cost efficiency was negatively correlated with perceived performance (r = −0.493) and positively correlated with technological infrastructure (r = 0.392), while technological infrastructure was positively correlated with perceived performance (r = 0.492). The negative bivariate CostEf-performance correlation differs from the positive conditional structural path reported below; this pattern should be interpreted as a change in the conditional association after other predictors and interactions are included, rather than as evidence of a temporal process.

To test the proposed hypotheses, the study employed the PROCESS option within SmartPLS version 4.1.4, as it is appropriate for estimating complex models involving mediation and moderation in prediction-oriented research ([34]). The analysis followed a two-stage approach, beginning with the assessment of the measurement model to ensure reliability and validity, and subsequently proceeding to the structural model assessment for hypothesis testing.

The results revealed that CAIS capability does not have a significant direct association with perceived firm performance (H1 not supported). This finding challenges the assumption that possessing or using an accounting information system is sufficient, by itself, to produce superior performance. From an RBV perspective, CAIS may be a potentially valuable technological resource, but its performance value depends on how it is combined with complementary organizational capabilities ([15]). This interpretation is consistent with studies arguing that IT investments do not guarantee improved performance without alignment with organizational processes and capabilities ([13]; [59]; [29]) and with [20] productivity-paradox argument. The finding differs from evidence of a positive CAIS-performance relationship reported by [39], [1], and [37], suggesting that contextual and complementary-resource conditions may shape the realized value of CAIS.

Cost efficiency had a significant positive association with perceived firm performance (H2 supported). This is consistent with the RBV argument that superior performance can arise from organizational capabilities that enable more effective deployment of resources ([15]). In the present model, cost efficiency represents such a capability because it reflects waste reduction, cost control, and efficient resource use. The result is consistent with [27] and [58] and also accords with [65] cost-leadership logic, under which cost advantages can support stronger margins and competitive positioning. The result should nevertheless be interpreted as a conditional association in the PLS-SEM model rather than as evidence that cost-efficiency improvements necessarily precede performance changes.

The mediation analysis identified a statistically significant negative indirect association between CAIS capability and perceived firm performance through cost efficiency. Thus, the study provides evidence that cost efficiency transmits part of the association between CAIS and performance, but the direction is contrary to the theorized beneficial pathway. H3 is therefore only partially supported: mediation is present statistically, but the hypothesized positive mechanism is not. From an RBV perspective, the result cautions against assuming that a technological resource will automatically be associated with a stronger complementary capability. It is compatible with [50] argument that resource possession and capability deployment are distinct. Although [66] and [14] highlight operational-efficiency mechanisms linking systems and performance, the negative sign in the present study should be described only as an adverse observed association. The data do not establish that CAIS causes an initial cost disruption or that efficiency subsequently recovers, because CAIS adoption timing and post-implementation adjustment were not measured.

Technological infrastructure positively moderated the association between CAIS capability and perceived firm performance (H4 supported). In substantive terms, the estimated CAIS-performance relationship becomes more positive as technological infrastructure increases. This finding is consistent with RBV resource-complementarity logic: infrastructure can act as a complementary asset that increases the usefulness of CAIS ([15]). It also accords with Contingency Theory, which predicts that the effectiveness of organizational systems depends on fit with contextual conditions ([26]). Prior work by [85] and [48] similarly suggests that stronger IT infrastructure helps organizations derive greater value from information systems.

The moderated-mediation result requires a more qualified interpretation than originally reported. The index of moderated mediation was statistically significant and positive, but the conditional indirect effect remained negative at low, mean, and high levels of technological infrastructure and became progressively less negative as infrastructure increased. Thus, technological infrastructure attenuates an adverse CAIS → cost efficiency → performance association; it does not transform the pathway into the hypothesized stronger positive mediation. H5 is therefore not supported in its hypothesized direction, even though the existence of conditional indirect effects is supported statistically. From an RBV perspective, the result still illustrates the importance of complementary assets because stronger infrastructure reduces the magnitude of the adverse indirect association. From a Contingency Theory perspective, it shows that the strength of the indirect relationship depends on technological context rather than being uniform across firms.

At the low level of technological infrastructure (M − 1 SD = 3.510), the standardized conditional indirect effect was the most negative (β = −0.023, t = 3.396, p = 0.001). This indicates that the adverse indirect association between CAIS capability and perceived performance through cost efficiency is strongest when technological infrastructure is relatively weak. Consistent with Contingency Theory, the result suggests that contextual technological conditions shape how CAIS capability is associated with internal efficiency and performance ([26]). However, the time-lagged survey evidence does not identify the specific operational causes of this pattern; explanations such as poor integration or inadequate IT skills should therefore be treated as possible mechanisms for future testing rather than as observed causes.

At the mean level of technological infrastructure (M = 4.477), the conditional indirect effect remained negative but was smaller in magnitude (β = −0.018, t = 3.250, p = 0.001) (see Table 5). The pattern indicates attenuation of the adverse indirect association as infrastructure improves. This is compatible with RBV arguments that the value of a technological resource depends on complementary resources and capabilities ([15]), but it should not be interpreted as evidence of a measured transition from early inefficiency to later efficiency because the study did not observe CAIS adoption stages over time.

At the high level of technological infrastructure (M + 1 SD = 5.444), the negative conditional indirect effect was weakest (β = −0.013, t = 2.700, p = 0.007). Thus, stronger technological infrastructure reduces, but does not eliminate or reverse, the adverse indirect association. This pattern is consistent with the general argument that robust technological environments can improve the fit between information systems and organizational processes ([36]). The finding therefore supports a contingency interpretation of the model while also showing why the original directional claim for H5 must be rejected: higher infrastructure attenuates a negative mediated relationship rather than strengthening a positive one.

This study contributes to the information-systems and accounting-technology literature by refining the Resource-Based View (RBV) and Contingency Theory in the context of CAIS capability. First, the insignificant direct CAIS-performance path indicates that a technological resource is not necessarily associated with superior perceived performance in isolation. This is consistent with the RBV proposition that resources create value through deployment and complementarity rather than mere possession ([15]; [50]). In this sense, CAIS is better conceptualized as an enabling resource whose performance association depends on accompanying organizational capabilities and technological conditions.

Second, the study adds nuance to capability-based RBV reasoning by showing that the observed indirect association through cost efficiency is significant but negative. Rather than validating a uniformly beneficial resource → capability → performance chain, the findings demonstrate that the sign of an indirect relationship may differ from the sign anticipated theoretically. The contribution is therefore not evidence that CAIS initially disrupts cost structures and later produces efficiency gains; that temporal process was not measured. Instead, the study shows that the relationship between a technological resource and a cost-efficiency capability can be adverse in the observed data and should be interpreted conditionally rather than assumed to be value enhancing.

Third, the study extends Contingency Theory by showing that technological infrastructure conditions both the direct CAIS-performance association and the indirect CAIS-cost efficiency-performance association. The conditional indirect effect becomes less negative as infrastructure increases, indicating that technological context changes the magnitude of the mediated relationship. This provides a more precise contingency contribution than a general claim that infrastructure simply ‘strengthens’ the mediation: the empirical pattern is attenuation of an adverse pathway.

Finally, integrating RBV and Contingency Theory shows that CAIS-related performance associations depend jointly on resource complementarity and contextual fit. Stronger technological infrastructure is associated with a more favorable direct CAIS-performance slope and with a smaller negative indirect effect through cost efficiency. The combined evidence therefore supports a configuration-based interpretation in which the consequences of CAIS capability vary according to the complementary technological environment, while avoiding a causal claim that the infrastructure necessarily produces later efficiency gains.

The findings suggest that managers should not assume that stronger CAIS capability alone will be associated with higher performance. The direct CAIS-performance path was insignificant, while the indirect path through cost efficiency was negative. Managers should therefore monitor both system use and cost-efficiency indicators when evaluating CAIS investments and should investigate whether implementation, integration, training, process redesign, or other organizational conditions are associated with adverse cost outcomes in their own firms. These mechanisms were not directly measured in the present study, so they are practical diagnostic possibilities rather than demonstrated causes. The positive cost-efficiency-performance path also indicates that cost-management capability remains important when firms seek to realize value from accounting technologies.

Technological infrastructure is a particularly important complementary condition. The positive CAIS × technological-infrastructure interaction indicates that the CAIS-performance slope becomes more favorable as infrastructure rises, while the conditional indirect analysis shows that stronger infrastructure attenuates the negative mediated association through cost efficiency. Managers should therefore evaluate hardware, software, connectivity, security, and integration capacity alongside CAIS capability. Importantly, the results do not show that high infrastructure converts the mediated pathway into a positive effect; rather, they show a smaller adverse indirect association. Policy support for technological readiness may therefore help firms reduce constraints around digital accounting systems, but causal policy claims should await longitudinal or quasi-experimental evidence.

Several limitations qualify the findings. First, the study used a two-wave, 30-day time-lagged employee survey rather than a single-wave cross-sectional design. This time separation helps reduce same-occasion common method covariance, but it does not establish the timing of CAIS adoption, the duration of system use, or a causal post-implementation adjustment process. Second, all focal constructs were self-reported, and the performance construct represents employee perceptions. Because the current manuscript does not report the number of firms, respondents per firm, firm identifiers, within-firm agreement statistics, or cluster-adjusted estimation, the results should not be interpreted as aggregated firm-level effects. Third, the manuscript must report wave-specific sample sizes, matching procedures, and attrition to make the final N = 638 reproducible. Fourth, the study focuses on Ghanaian manufacturing settings and may not generalize to sectors or countries with different institutional and technological conditions. Finally, cost efficiency is the only mediator and technological infrastructure is modelled as an overall construct, so other mechanisms and specific infrastructure dimensions remain outside the model.

Future research should use longer longitudinal designs that record CAIS adoption dates, implementation stages, and repeated measures of cost efficiency and performance so that claims about adjustment, disruption, recovery, or temporal ordering can be tested directly. Studies seeking firm-level inference should sample multiple respondents per firm, report the number of firms and cluster sizes, establish within-firm agreement and ICCs before aggregation, and use multilevel or cluster-robust estimation where appropriate. Multi-source designs that combine employee perceptions with objective financial and operational performance measures would also strengthen construct validity. Future work could further examine alternative mediators such as decision-making quality, innovation capability, organizational learning, or agility, and disaggregate technological infrastructure into components such as system integration, connectivity, scalability, and cybersecurity to identify which complementary resources most strongly condition CAIS outcomes.

The authors acknowledge the valuable support and cooperation of the manufacturing firms and respondents who participated in the study. The authors are also grateful to colleagues and academic mentors whose comments and suggestions contributed to the development of the manuscript. Any errors or omissions that remain are the responsibility of the authors.

Conceptualization, Fredrick Asare-Appiah and Cornelius Adorm-Takyi; methodology, Fredrick Asare-Appiah, Cornelius Adorm-Takyi and Emmanuel Yaw Arhin; software, Fredrick Asare-Appiah; validation, Cornelius Adorm-Takyi, Emmanuel Yaw Arhin and George Tackie; formal analysis, Fredrick Asare-Appiah; investigation, Fredrick Asare-Appiah and Michaella Selasi Genego; resources, Fredrick Asare-Appiah, Edward Marfo-Yiadom and Michaella Selasi Genego; data curation, Fredrick Asare-Appiah and Michaella Selasi Genego; writing—original draft preparation, Fredrick Asare-Appiah; writing—review and editing, Cornelius Adorm-Takyi, Emmanuel Yaw Arhin, George Tackie, Edward Marfo-Yiadom and Michaella Selasi Genego; visualization, Fredrick Asare-Appiah; supervision, Cornelius Adorm-Takyi, George Tackie and Edward Marfo-Yiadom; project administration, Fredrick Asare-Appiah and Cornelius Adorm-Takyi. All authors have read and agreed to the published version of the manuscript.