Measuring Technostress in Corporate Culture: Insights from the 10-K Annual Reports

Abstract

This study introduces an innovative approach for quantifying the technostress phenomenon, drawing on textual narratives from the firm’s annual report. Based on a dataset covering the Standard and Poor’s 500 (S&P 500) index firms, we analyze 2532 10-K annual reports and highlight the key contributors of technostress across six different dimensions of technostress using a combined score. A major advantage of the new six-dimensional scoring framework is that it offers a set of objective metric proxies to capture technostress without bias, utilizing a refined list of 42 key clues derived through factor analysis. Also, it adopts natural language processing, revealing hidden patterns and anomalies that indicate technostress. We further validate this framework by applying fixed-effect regression models to examine the impact of technostress on productivity. The main results imply that the four technostress dimensions presented in techno-risks, insecurity, uncertainty, and invasion negatively impact firms’ productivity. This framework offers practical implications for firms, allowing them to generate a rich profile concerning the degree of technostress associated with existing practices, highlighting the crucial need for advanced interventions, facilitating comparisons with other firms from the same or different industries, as well as cross-country comparisons.

1. Introduction

Information and Communication Technologies (ICTs) have become an indispensable component of our daily lives, revolutionizing the way we work, learn, and interact with each other. While these technological innovations bring speed, convenience, efficiency, and communication, they have invaded each aspect of our lives, removing the boundaries between personal and professional lives, bringing challenges known as the technostress phenomenon.

Over the past few years, digital transformation has become a global strategic initiative for many firms, encompassing functionality for information management, performance measurement, and decision making (Obaid, 2024). Digital transformation is not only the introduction of novel technologies such as artificial intelligence, big data, machine learning, cloud computing, robotic automation process, and blockchain (Silva et al., 2024), but also, it is a journey that includes management change, process reengineering, and technology implementation (Liew et al., 2022). Hence, many professions are anticipated to lose relevance or experience significant transformation in the digitalization process (Obaid, 2024). Furthermore, these technologies trigger massive changes in the working environment, which lead to numerous outcomes on employees, including stress (Gerekan et al., 2024).

Researchers began to explore the multidimensional technostress phenomena, delving deeper into its definition, dimensions, and implications. This term was first proposed by (Brod, 1984), who defined technostress as a modern disease resulting from the difficulty of adapting to new computer technologies in a healthy way; this disease indicates resistance to accepting new technologies or excessive attachment to computer technology. Technostress can be divided into techno-eustress, which is the bright side of technology use or positive stress, and techno-distress, which is the dark side of adopting advanced technology or negative stress, and this classification depends on the way people perceive technostress as a stressor (Sanjeeva Kumar, 2024).

Tarafdar et al. (2007) introduced one of the widely adopted frameworks for technostress, which is primarily grounded in the transactional theory of stress and the Job Demand–Resources model, to capture the distress arising from employees’ inability to cope with new technology in the workplace. This framework identifies five core technostress creators: techno-overload, techno-invasion, techno-complexity, techno-insecurity, and techno-uncertainty, which contribute to the negative psychological and behavioral impacts of technology use. Despite its widespread adoption in the literature, this framework fails to consider the systemic risks associated with technology adoption. Cremer et al. (2022) confirmed that the vast development in digitized technologies intensifies the propensity for cybercrimes. This limitation motivates authors to extend the conceptual framework of technostress to address this limitation through suggesting a new perspective to address the techno-risks, such as privacy violations and breaches, digital dependency, and cybersecurity threats that contribute to cybercrimes.

Numerous studies, on the other hand, adopt the quantitative approach, mainly based on surveys (e.g., Dutta & Mishra, 2024; Consiglio et al., 2023; La Torre et al., 2020; Boyer-Davis, 2019; Turel & Gaudioso, 2018; Tarafdar et al., 2007) to measure technostress, however, the dependence on surveys as the data collection method has weaknesses, as sample estimates are subject to sampling error (Stern et al., 2014), a shortage of responses from all individuals in the sample (Ponto, 2020). Also, offline surveys are costly (Jones et al., 2013), and online surveys lack privacy and confidentiality (Nayak et al., 2019).

Another group of studies implemented the qualitative approach, utilizing interviews as the data collection method, such as Ioannou (2023), Castro Rodriguez and Choudrie (2021). This approach has some disadvantages, since the process is time-consuming, with reliability and validity challenges (Glassman et al., 2020), as well as the results cannot be generalized (Alsaawi, 2016), and analyzing the linguistic data of the interview is challenging (Alshenqeeti, 2014). In addition, the interviewer himself can be a weakness if he/she lacks interview skills or shows subjectivity and bias (Hofisi et al., 2014).

Finally, a third group of studies undertook a mixed-method approach that fundamentally depended on both surveys and interviews for measurement (e.g., Schmidt et al., 2021; Califf et al., 2020; Tarafdar et al., 2020; Sellberg & Susi, 2014; Ayyagari et al., 2011). Conversely, the combination of findings from both approaches is challenging for researchers (Z. Li & Liu, 2021; Hong et al., 2020).

Given the limitations associated with these approaches to measure technostress phenomena, a crucial need for more research efforts devoted to developing a new method for measuring the technostress-related themes is realized. This study applies content analysis by utilizing a Python-based Streamlit application that eliminates subjectivity and bias. It is a novel avenue that has the potential to measure the degree of technostress in annual reports. It allows the quantitative assessment of the whole text, through assigning scores to coding units and compiling the counts to compute the degree of intensity across each category (Beattie et al., 2004). Hence, this study takes a step on that path and develops a new six-dimensional scoring construct for the holistic content analysis of the technostress phenomenon in the US firms’ 10-K annual reports and proposes a digital-assisted methodology for operationalizing this framework. A 10-K report is a comprehensive annual report required by the SEC for publicly traded companies to disclose their financial performance, business operations, risks, and audited financial statements. Figure 1 presents the key constructs of the technostress measurement framework. Moreover, this new methodology enables firms to generate a rich profile regarding the degree of technostress associated with their current practices.

To validate the technostress measurement approach, first, the list of key clues acts as a crucial input into the natural language processing (NLP)/Python to analyze 2532 10-K annual reports, generating word frequency counts, which validate the credibility of the key clues measurement approach. This is followed by the second validation, which investigates both the aggregated and individual-level impact of technostress dimensions on firms’ overall productivity.

The textual analysis of 10-K annual reports reveals the key contributing factors for each of the six dimensions of technostress. Since the disclosure in 10-K reports provides a promising pathway for gaining insights into the effects of operational intentions on firms’ performance, this disclosure also influences employment quality (Tan et al., 2023; Cooper et al., 2022). Furthermore, the Securities and Exchange Commission (SEC) regulations require management to disclose material risks associated with cybersecurity incidents (U.S. Securities and Exchange Commission [SEC], 2023), which clearly includes technostress as an important operational stressor. Therefore, 10-K annual reports serve as a reliable proxy for measuring firm-level technostress, as they capture the aggregated firm experience that influences employment quality and stress.

Further examination of the six dimensions of technostress confirms that they reduce firm productivity in the short-term, and this impact grows significantly over the years, acknowledging the cumulative and intensive influence of technostress over time. Specifically, regarding the impact of each technostress dimension individually, the results conclude that techno-risks, insecurity, uncertainty, and invasion unfavorably influence the firms’ overall productivity.

Therefore, this study contributes to the academic literature in four different ways. First, the new scoring six-dimensional framework offers a set of objective metric proxies to measure technostress, which allows a powerful test of various research questions relating to the technostress phenomenon. Second, the development of this new scoring framework acts as a practical tool for measuring technostress-related themes across different firms. Third, it allows firms to benchmark their current practices, and the associated technostress effects enable them to compare themselves with other firms in the same or different industries, also permitting cross-country comparisons. Fourth and last, the utilization of an advanced Python-based Streamlit application that adopts natural language processing, to conduct content analysis, reveals hidden patterns and anomalies that might suggest technostress.

The rest of the study is organized as follows: Section Two reviews the existing literature. Section 3 describes the research design, outlines the basic principles of content analysis, the data sources, and the samples used in the study. Section 4 validates and tests the application of the digital-assisted framework to measure the degree of technostress across US firms. Section 5 includes robustness checks. Finally, Section 6 wraps up with insights into practical applications, study limitations, and ideas for future research.

2. Literature Review

The new advancements in technologies trigger great changes in the working environment, facilitating people’s lives; however, they have brought about negative consequences, such as technostress (Gerekan et al., 2024). Arnetz and Wiholm (1997) defined technostress as the state of mental and physical stimulation experienced by certain employees who rely heavily on computers in their work tasks. In addition, it has unfavorable impacts on thoughts, behaviors, attitudes, and the body as a whole (Shu et al., 2011). Furthermore, Ragu-Nathan et al. (2008) explained the various traits that contribute to the technostress phenomenon in the work environment, which are the heavy reliance of employees on ICT applications that require continuous update, also the increase in the level of sophistication of ICT_s that creates a gap between the level of knowledge needed to undertake tasks and the level of knowledge among employees, and finally changes in work environment and organization culture. However, Salazar-Concha et al. (2021) confirmed that people react differently to internal and external alterations; hence, there are also positive stressors, in addition to the negative ones, that trigger joy, satisfaction, and facilitate decision-making.

Five categories of stressors contributing to the negative experience of technostress were first introduced by Tarafdar et al. (2007). Their examination is crucial, as they create a widespread perturbation or distress, leading to negative consequences for both individuals and firms (Turel & Gaudioso, 2018). These five technostress creators are techno-overload, where users feel overburdened to work quicker for long hours. While techno-complexity refers to the sophisticated technology that threatens users’ skills. Whereas techno-invasion is constantly connected, it disturbs the work-life balance. Moreover, techno-insecurity refers to employees’ fear of being replaced by skilled workers. Finally, techno-uncertainty refers to the continuous changes and updates in technologies that cause anxiety among employees.

Hence, Salazar-Concha et al. (2021) implemented a literature review based on bibliometric analysis over the years 1982–2017, and realized that the works by Tarafdar et al. (2007, 2015) each receive more than 100 citations, reflecting a significant impact among academic researchers. In a similar vein, the study by Borle et al. (2021) conducted an electronic literature review and identified 21 articles that met the eligibility criteria and measured technostress using the five key technostressors. Therefore, these five stressors are widely accepted and embraced in the academic literature as the appropriate way to measure technostress (Dutta & Mishra, 2024; Gerekan et al., 2024; Harris et al., 2022; Camacho & Barrios, 2022; La Torre et al., 2020; Wu et al., 2020; Boyer-Davis, 2019; Khedhaouria & Cucchi, 2019; Gaudioso et al., 2017; Jena, 2015; Srivastava et al., 2015; Day et al., 2012; Ayyagari et al., 2011).

The smart technologies and digitization have escalated the likelihood and severity of cybersecurity crimes (Cremer et al., 2022). Y. Li and Liu (2021) discussed the most common methods used by cybersecurity criminals are denial of service, man in the middle, malware, and phishing. A recent report issued by Deloitte (2024) suggests that the average cost of data breaches by ransomware is US$4.91 million. The rapid growth of technologies and digitization has escalated the likelihood and severity of cybersecurity crimes (Cremer et al., 2022). Consequently, cybersecurity has become an essential part of management and accounting practices, helping firms safeguard networks and sensitive data from unauthorized access, theft, and damage (Al-Shattarat et al., 2025). In 2021, the cybersecurity attacks on Colonial Pipeline and SolarWinds significantly restricted the US government’s access to critical information from key industries, threatening both national and the country’s economic stability (Snider et al., 2021).

Hence, Z. Li and Liu (2021) discussed the most common methods used by cybersecurity criminals are denial of service, man in the middle, malware, and phishing. Also, a recent report issued by Deloitte (2024) suggests that the average cost of data breaches by ransomware is US$4.91 million. As technology continues to progress, firms are under pressure to balance the dual challenge of improving operational efficiency along with mitigating the various cybersecurity risks through information security technologies and systems (Al-Shattarat et al., 2025). Employees interact with these technologies daily. Adapting to these complex and unfamiliar security systems imposes an extra workload and uncertainty due to the security protocols. Ultimately, this fosters technostress intensified by role stress and diminished employees’ organizational commitment (D’Arcy et al., 2014; Hwang & Cha, 2018).

Despite the importance of the above-explained five stressors, they fail to capture an important stressor: techno-risks, which refer to the anxiety that users experience due to actual and potential threats associated with technology use, specifically cybersecurity data breaches and unauthorized hacks. Therefore, this study modifies the five dimensions of stressors proposed by Tarafdar et al. (2007) by incorporating a novel techno-risks dimension to better capture cybersecurity threats within firm contexts.

3. Research Design

3.1. Technostress Measurement

The technostress variable in this study is conceptualized as a multidimensional construct, captured through a composite index consisting of six distinct dimensions: techno-overload, techno-invasion, techno-complexity, techno-insecurity, techno-uncertainty, and techno-risks. These dimensions collectively capture various technological stressors that firms suffer. To empirically measure these constructs, a Python-based (version 3.11) content analysis was conducted on the textual data of 10-K annual reports. This approach involved searching for a predefined set of keywords associated with each technostress dimension. The frequency of these keywords was then quantified and aggregated by dimension, with higher counts indicating elevated levels of technostress with the firm’s disclosure.

Content analysis is one of the most popular quantitative analysis methods, which has extensive historical applications in sociology, psychology, communication, journalism, and business (Duevel, 2019). Quantitative content analysis involves methods for reducing and transforming text into a unit-by-variable matrix, and analyzing that matrix to test hypotheses using a predefined set of codes on the set of data (Serafini & Reid, 2023). In other words, it identifies the frequently occurring terminology from a text, calculates relationships through linkages between the terminologies, and maps them into research streams according to their relatedness or co-occurrence (Klarin, 2024).

In contrast to other data-generating techniques, content analysis offers various advantages. First, unlike interview techniques, content analysis produces an unobtrusive measure in which the sender and the receiver of the message are not aware of what is being analyzed (Weber, 2011). Second, content analysis is a structural system for categorization of data, enabling statistical analysis, and assuring the validity and reliability of the results (Serafini & Reid, 2023). Third and last, content analysis is capable of processing a large amount of data (Krippendorff, 2019).

For the purpose of conducting the content analysis, we follow the guidelines outlined by (Serafini & Reid, 2023). First, we identified the research question: how the six-dimensional technostress measurement framework can be used to quantify the degree of technostress embedded in corporate culture, as expressed in the content of 10-K annual reports. This research presents a new method for measuring technostress using NLP/Python content analysis. Although the core of content analysis is matching keyword dictionaries, this tool integrates multilevel context analysis to differentiate positive, neutral, and negative words using semantic similarity modelling (Term Frequency-Inverse Document Frequency (TF-IDF) cosine similarity), co-occurrence-based context analysis, and adaptive parameter tuning for context sensitivity. This results in capturing the semantics along the document and prevents overgeneralization, which ensures that keywords are not treated identically.

Second, construct the data corpus. Once the main research question is determined, the researchers have to determine which data sets would be analyzed. Hence, the researchers chose a sample of the US firms’ annual reports.

Third, establish the object of the study, to analyze the annual reports as a suitable data corpus. We have to define what is meant by technostress and its dimensions to include in the analysis. In this step, we followed the same procedure introduced by Hussainey et al. (2003). We identified the preliminary list of key terms associated with technostress. To generate this list, we reviewed the existing literature, surveys (e.g., Berger et al., 2024; Dutta & Mishra, 2024; Nastjuk et al., 2024; Tarafdar et al., 2007), and annual reports. We then wrote notes on the key terms associated with technostress and its dimensions. After that, we measure how often each of the key terms associated with technostress ends with a list of key clues reflecting the categories of technostress.

Fourth, developing the initial categories, in this step, we decided to extend the work offered by Tarafdar et al. (2007) to include a new category presented in techno-risks as a novel stressor added to the initial five categories. For each category, a set of key clues was identified from the prior technostress literature and refined through an iterative content analysis and factor-analytic screening process to ensure both conceptual relevance and empirical distinctiveness.

Table 1. Technostress dimensions: list of initial key clues.

Technostress Dimension	Key Clues	Technostress Dimension	Key Clues
Techno-overload	Overload	Techno-insecurity	Job security
	Workload		AI
	Work pressure		Machine learning
	Information overload		Talent
	Big data		Workforce reduction
	Large volume		Automation
	Productivity demands		Robot
	Volatile		New recruits
	Competitive	Techno-uncertainty	Uncertainty
	Speed		Development
	Scalability		Rapid
	New trends		New laws
	New technology		Regulatory changes
	Technology-driven tasks		System upgrade
	Multitasking		Change
Techno-invasion	Remote		Training
	24/7		Unauthorized
	Global supply chain	Techno-risk	Cyber security
	Digital transformation		IT Control
	Work-life balance
	Connectivity		Hack
	Intrusion		Data security
	Virtual		Breach
	After work hours		Vulnerability
Techno-complexity	Complexity		System failure
	Digitization		Leakage
	IT		Technology risks
	ICT		Confidentiality
	Cryptocurrency		Privacy risks
	Blockchain		Security risks
	ERP		IT fatigue
	Cloud		Technostress
	Technical skills		Attacks
	System updates
	User difficulty	Total key clues	68

presents the six categories and their initial key clues concluded from prior work in technostress.

As part of the keyword refinement process, an exploratory factor analysis using principal component analysis was conducted on the initial set of technostress-related keywords. This step aims to recognize a reduced set of variables that captures the common characteristics of the original variables (Forina et al., 2005). Hence, we follow Forina et al. (2005) and Chin et al. (1997) and adopt a Varimax rotation as a common procedure in factor analysis, which allows factors to be correlated. During the phase of initial item development, the factor loadings of an item that is 0.60 or above are considered acceptable (Hurtt, 2010; Chin et al., 1997). Based on the results of the rotated matrix presented in

Table 2. The rotated matrix components’ response to technostress.

	Component
	1	2	3	4	5	6
Connectivity	0.891
Regulatory changes	0.861
Global supply chain	0.857
Data security	0.845
Competitive	0.812
Training	0.806
Uncertainty	0.799
Remote	0.791
Complexity	0.786
Unauthorized	0.733
Leakage	0.73
System failure	0.725
Big data	0.705
Attacks	0.681
Workforce reduction	0.607
Digitization	0.607
System updates
technostress
New technology
Rapid
Automation		0.916
Scalability		0.89
Cloud		0.885
Machine learning		0.807
24-Jul		0.799
Breach		0.774
Confidentiality		0.739
Virtual		0.723
Volatile		0.694
Technical skills	0.634	0.635
Changes		0.615
Speed
Vulnerability
IT
Overload
Work-life balance
User difficulty
Productivity demand
Job security
Information overload
Technology-driven tasks
Cryptocurrency			0.798
Cyber security			0.733
Development			0.71
New trends	0.612		0.697
IT fatigue
Intrusion
Work pressure
Multitasking
Hack
After work hours
Technology-driven tasks
Technology risk				0.858
ERP				0.763
Workload				0.664
Talent	0.641			0.652
New technology			0.613	0.624
AI				0.606
Security risks					0.89
Large volume					0.844
IT Control					0.832
New laws
Robot
ICT
System upgrade
Privacy risks						0.928
Digital transformation						0.774
Blockchain
Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization.
Rotation converged in 6 iterations.

, after eliminating 26 items that do not load significantly, 42 items were retained for further analysis. Moreover, the Kaiser–Meyer–Olkin measure of sample adequacy yields a value of 0.916; this high value indicates that the data are very suitable for factor analysis. Furthermore, the p-value = 0.000, indicating a significant correlation among the tested variables, which suggests that the data have sufficient relationships among the keywords to justify factor analysis.

Table 2 presents the results of exploratory factor analysis, reducing 43 technostress keywords to six empirically derived dimensions that explain 91.6% of the total variance. Principal component analysis with Varimax rotation identified (1) System Stress, (2) Technology Adoption Stress, (3) Cybersecurity Risk, and (4) Workload Pressure. Keywords loading < 0.7 on all factors were excluded, enhancing construct validity and achieving dimensional reduction. Although factor analysis is used as an appropriate tool to refine the list of key clues, the resulting factor structure did not perfectly match the theoretically established technostress dimensions and prior empirical studies. Since the main objective is to measure technostress based on the well-established theoretical framework, the final set of key clues was therefore retained and organized according to the original six predefined dimensions. This approach ensures consistency of measurement with the extant technostress framework and preserves the conceptual validity of the constructed categories (Bondanini et al., 2020).

Table 3. Final list of technostress key clues.

Technostress Dimension	Key Clues	Technostress Dimension	Key Clues
Techno-overload	Workload	Techno-insecurity	AI
	Big data		Workforce reduction
	Large volume		Automation
	Volatile		Machine learning
			Talent
			Regulatory changes
			Changes
			Training
	Competitive	Techno-uncertainty	Uncertainty
	Scalability		Rapid
	New technology		Regulatory changes
Techno-invasion	Remote		Training
	24/7		Development
	Global supply chain	Techno-risk	Unauthorized
	Digital transformation		Cyber security
	Connectivity		Attacks
	Virtual		Data security
Techno-complexity	Complexity		Breach
	Digitization		System failure
	IT control		Leakage
	Cryptocurrency		Confidentiality
	ERP		Security risks
	Cloud		Technology risks
	Technical skills

reports the final list of key clues of technostress aligning with theoretical lens.

Fifth, developing the analytical template, one of the most advanced analytical techniques that enables the analysis of a data corpus, is the Python-based Streamlit application that employs natural language processing for the analysis of the textual narrative in the annual reports. Hence, this technique enables the processing of numerous annual reports, identifying the synonyms, calculating the word frequency and counts, generating breakdowns for each file, and providing professional reports including document statistics and total count across the uploaded files.

Sixth, pilot testing of the key terms to achieve the validity of the proposed measure of technostress. The researchers randomly selected five annual reports and revised each annual report individually and compared the results.

Table 4. A sample of technostress themes from the annual reports.

Technostress Dimension	Proposed Key Clues	Analytical Notes
Techno-overload	Workload	The increase was primarily due to higher labor-related costs driven by overtime and contractor usage to support the increased workload.
	Volatile	The Company believes, in general, gross margins will be subject to volatility and downward pressure… in competitive markets. The electric utility industry is subject to various factors that may result in volatile fuel costs.
	Competitive	The markets for the Company’s products and services are highly competitive… aggressive pricing. The electric utility industry is highly competitive and subject to evolving regulatory requirements.
	Scalability	Our manufacturing operations must maintain the capacity and scalability to meet fluctuating global demand for our products.
Techno-invasion	Remote	Some of our employees continue to work in remote or hybrid arrangements
	Global supply chain	We are subject to risks associated with the global supply chain, including delays and increased costs. Abbott’s global supply chain is subject to risks including disruptions from natural disasters, geopolitical events, and supplier issues.
	Work-life balance	We provide various options to assist with career growth and development…We also provide volunteer opportunities and volunteer grants, as well as $10,000 of charitable giving matching annually, through the ONEOK Foundation.
Techno-complexity	Complexity	The complexity of global regulatory requirements presents ongoing compliance challenges. The complexity of our energy delivery systems requires significant ongoing maintenance and investment.
	IT control	We rely on information technology (IT) systems to operate our business and maintain customer data. …the Company’s business and reputation are impacted by information technology system failures and network disruptions.
	ERP	Oracle Fusion Cloud Enterprise Resource Planning (ERP), which is designed to be a complete and integrated ERP solution to help organizations improve decision-making and workforce productivity and to optimize back-office operations by utilizing a single data and security model with a common user interface.
	Cloud	The Company’s cloud services store and keep customers’ content up-to-date and available across multiple Apple devices and Windows personal computers. We believe that our Oracle Cloud Services offerings are opportunities for us to continue to expand our cloud and license business.
Techno-insecurity	Automation	Automation in manufacturing processes improves efficiency but may require significant capital investment.
	AI	Abbott is exploring the use of artificial intelligence (AI) in diagnostic technologies.
	Machine learning	Machine learning algorithms are being incorporated into some of Abbott’s diagnostic platforms.
	Talent	Failure to attract and retain a qualified workforce could have an adverse effect on our business.
	Workforce reduction	Our recruitment strategy is focused on hiring a workforce to meet our business objectives, including critical skilled trade roles. A shortage of skilled labor may make it difficult for us to maintain labor productivity and competitive costs.
	Regulatory changes	Changes in regulatory requirements could impact the timing and cost of bringing products to market.
		Changes in tax laws and regulations may adversely affect our financial condition…
	Training	Abbott provides ongoing training to employees to ensure compliance with quality and regulatory standards.
	Development	Research and development expenses were $2.8 billion in 2023, reflecting continued investment in innovation.
Techno-risk	Unauthorized	The Company and its business partners and customers could be subject to unauthorized access…
	Cyber security	We have implemented cybersecurity measures to protect our IT systems from unauthorized access. The Company and its business partners and customers could be subject to unauthorized access, cybersecurity threats, data breaches and other security incidents…
	Attack	Cybersecurity attacks could result in operational disruptions or unauthorized disclosure of sensitive information. Attacks are expected to continue accelerating in both frequency and sophistication. …hackers and other malicious actors…
	Breach	…cybersecurity threats, data breaches and other security incidents… A data breach could have a material adverse effect on our reputation and financial condition.
	System failure	…the Company’s business and reputation are impacted by information technology system failures and network disruptions.
	Leakage	If any of our systems are damaged, fails to function properly or otherwise becomes unavailable, we may incur substantial costs to repair or replace them and may experience loss or corruption of critical data and interruptions or delays in our ability to perform critical functions, which could affect adversely our business and results of operations.
	Confidentiality	The misappropriation, corruption or loss of personally identifiable information and other confidential data from us or one of our vendors could lead to significant breach of notification expenses…
	Security risks	Abbott faces security risks related to both its physical facilities and IT systems. Security risks include both physical and cybersecurity threats to our assets and systems.

shows the NLP syntactical patterns of technostress to discover the way technostress is exhibited in corporate disclosures. Then, we adopted the final analytical template to analyze a sample of 2532 10-K annual reports.

The final seventh step involves the findings and implications. Content analysis using NLP/Python is applied to a sample of 2532 annual reports, highlighting the various terms and bigrams of technostress, with their frequency counts in the annual reports. As well as the number of 10-K reports that include these terms and bigrams. The following

Table 5. Summary of top technostress terms and bigrams used in the construction of the technostress measure.

Terms and Bigrams	Frequency	The Percentage of Total Technostress Clues Count	No. of Reports
Workload	230	0.013%	223
Big data	283	0.25%	206
Large volume	140	0.02%	201
Volatile	4540	0.79%	1718
Competitive	39,107	7.26%	2540
Speed	3362	0.66%	1255
Scalability	445	0.19%	316
New technology	1007	0.066%	801
Remote	3321	0.613%	1438
24/7	271	0.038%	343
Global supply chain	1269	0.269%	631
Digital transformation	536	0.089%	326
Connectivity	2657	0.432%	737
Intrusion	772	0.132%	579
Virtual	2547	0.312%	842
Complexity	5622	0.967%	67
Digitization	191	0.032%	208
IT control	2495	23.17%	77
Cryptocurrency	197	0.0164%	107
Blockchain	189	0.0277%	164
ERP	1651	0.183%	330
Cloud	21,561	3.369%	1541
Technical skills	154	0.0205%	183
Automation	3935	0.822%	967
AI	8752	1.341%	673
Machine learning	1266	0.208%	568
Talent	9464	1.768%	1795
Workforce reduction	403	0.068%	315
Uncertainty	15,691	2.905%	2443
New laws	1833	0.335%	1130
Regulatory changes	1904	0.384%	1071
Training	12,186	2.119%	2133
Unauthorized	10,817	1.876%	2429
Cyber security	1381	0.273%	600
Attacks	12,275	2.306%	2345
Data security	3684	0.60%	1362
Breach	11,461	1.635%	3425
System failure	245	0.043%	273
Leakage	253	0.049%	1686
Technology risks	289	0.058%	254
Confidentiality	3215	0.540%	1434
Security risks	1089	0.023%	837

Table 5 shows the list of key terms and bigrams that are used to identify technostress in textual narratives of 10-k annual reports. Column 2 captures the total number of key terms and bigrams, column 3 includes the percentage of total technostress clues count to standardize the frequencies of word count, while column 4 indicates the number of reports that include the key terms and bigrams. Both statistics are based on a sample of 2532 annual reports.

shows a summary of NLP/Python content analysis as follows:

3.2. Sample and Data Collection

Our data set covers a sample of 422 US industrial firms randomly selected from the S&P 500 over the years from 2017 to 2024. However, we exclude 72 financial firms (SIC code 6000–6999) and 10 firms with incomplete data over the years of the sample. Moreover, we eliminate years 2020 and 2021, from the analysis to isolate the impact of the COVID-19 pandemic on our sample, since the COVID-19 pandemic presents an extraordinary outlier that deviates from the normal condition. The study of Bahamondes-Rosado et al. (2023) and Molino et al. (2020) acknowledged that during COVID-19 pandemic, workers experience heightened technostress due to various factors such as lack of control over working hours, work–family conflict, and their unfamiliarity with remote working arrangements. We draw data from the intersection of Edgar and Compustat databases, resulting in a final sample of 2532 observations.

4. Validation of the Proposed Measure

Validating Textual Analysis Through Operational Impact

In order to strengthen the validity of the textual analysis results, this study examines the potential association between the extent of technostress disclosure in the annual reports and the operational variable, which is firm productivity. Prior research suggested that technostress affects organizational outcomes as productivity, employee well-being, and organizational stress (Berger et al., 2024; Gerekan et al., 2024; Boyer-Davis, 2019). Linking the frequency of key technostress cues to operational variables, such as a firm’s productivity, helps assess whether financial reporting language and disclosures reflect actual operational pressures or communication practices. Model (1) explains the relationship between technostress frequencies and productivity as follows:

Productivity = α0 + β1 Tech_Stress + β2 Firm_Size + β3 IW500 + β4 LEV + β5 CAP + β6 Report_Complex+ β7 Ass_Turover + ε

Model (1)

where productivity is the dependent variable measured by revenue growth and return on equity for robustness check. While Tech_Stress refers to the percentage of technostress key clues coverage in the annual reports divided by the length of the annual report, measured using the textual analysis of the annual reports. Firm_Size represents the firm size measured by the natural logarithms of total assets (Eltamboly, 2025), whereas, IW500 refers to the firms’ IT capability measured using a 0, 1 variable, taking 1 if the firm is listed in the IW 500 index and 0 otherwise (Guo et al., 2021; Hoffman et al., 2018). Also, CAP is the market cap of the firm. Moreover, Report_Complex refers to the complexity of the annual reports measured by the length of these reports, as the total number of words (Nazari et al., 2017). Finally, Ass_Turover means the percentage of total revenues divided by total assets (Ahmad et al., 2023) and Big 4 measured as a dummy variable takes the value of 1 if the audit firm is one of the Big 4, 0 otherwise.

Table A1. Descriptive statistics.

Variable	Obs.	Mean	Std. Dev.	Min	Max	Jarque–Bera	Probability
ROA	2534	23.22938	1.318457	3.175039	27.19735	46,161.69	0
Tech_Stress	2534	220.2253	110.4422	10	968	2376.09	0
Firm_Size	2534	10.39781	0.599114	8.228264	12.60237	89.808	0
IT_Capability	2534	0.418706	0.493444	0	1	423.5788	0
LEV	2534	0.832715	19.62108	−550.685	432.2	29,139,053	0
CAP	2534	24.165	1.18134	4.323735	28.89587	118,901.1	0
Report_Complex	2534	10.80343	0.858798	1.791759	13.27872	422,298.8	0
Ass_Turover	2534	0.986977	0.120156	0	2	453,183.3	0
Big 4	2534	0.636771	0.601111	0	8.126312	37,842.47	0

Continuous variables are winsorized at the 1st and 99th percentiles.

and

Table A2. Correlation matrix.

Variables	1	2	3	4	5	6	7	8	9
ROA	1
	-----
Tech_Stress	−0.02974	1
	−1.49721	-----
Firm_Size	0.734703	0.075961	1
	54.49597	3.833351	-----
IT_Capability	0.014091	−0.02799	−0.08941	1
	0.709103	−1.40909	−4.5172	-----
LEV	0.027402	−0.01942	−0.00517	0.044929	1
	1.37934	−0.97757	−0.25998	2.263078	-----
CAP	0.695888	0.097923	0.619525	0.001848	0.021952	1
	48.75907	4.951177	39.71309	0.092996	1.104854	-----
Report_Complex	−0.00966	0.121182	0.022349	−0.02494	−0.02355	−0.03216	1
	−0.48619	6.143048	1.124835	−1.25557	−1.18507	−1.61924	-----
Ass_Turover	0.121798	−0.07324	−0.04091	−0.00561	−0.02629	0.040619	−0.0132	1
	6.174724	−3.69512	−2.06032	−0.28203	−1.32345	2.045592	−0.66441	-----
Big 4	0.188944	−0.03703	0.137717	0.078686	0.001498	0.137169	−0.01545	0.005521	1
	9.681854	−1.86436	6.996436	3.971719	0.075357	6.968082	−0.77767	0.277798	---

in Appendix A depict descriptive statistics and a correlation matrix among the tested variables. Continuous variables are winsorized at the 1st and 99th percentiles. The findings indicate a significant correlation coefficient among the examined variables. Also, we investigate the data for potential multicollinearity and find that the greatest VIF among the examined variables is 3.046, which is below 10, the recommended value (Abdallah & Eltamboly, 2022).

Table 6. Results of validation analysis.

Variables	Model (1) Revenue Growth	Model (2) ROE	Model (3) Lagged 2
Tech_Stress	−0.001481 *** (−9.40645)	−0.00218 ** (−2.33123)	−0.00631 ** (−2.2344)
Firm_Size	1.147389 *** (34.5865)	−0.0733 (−0.08449)	−0.71245 *** (−4.6392)
IT_Capability	0.119569 ** (3.829824)	−0.18198 (−0.89015)	−0.06344 (−0.84153)
LEV	0.001384 (0.416473)	0.285272 ** (2.575353)	5282.602 *** (7.774141)
CAP	0.416473 *** (25.08076)	0.046975 (0.666771)	1.190006 *** (2.780128)
Report_Complex	0.00232 (0.128464)	−0.03388 (−0.66581)	−0.02407 *** (−3.47823)
Ass_Turover	0.273204 *** (10.68288)	0.068838 (0.363501)	0.073523 *** (3.676603)
Big 4	0.714439 *** (5.434125)	0.187274 (0.836811)	0.251825 *** (3.948869)
C	0.61219 (1.574044)	1.770008 ** (2.966364)	17.7374 *** (11.91317)
Year Fixed Effects	Yes	Yes	Yes
Firm Fixed Effects	Yes	Yes	Yes
Adj. R-squared	0.671759	0.461977	0.448345
F-statistic	635.4293	5.828718	4.044966
Prob(F-statistic)	00000	0.0000	0.0000
Durbin–Watson	2.263651	1.513625	1.57598
VIF.	3.046542022	1.858657	1.170746
Obs.	2532	2532	2532

Notes: Values are significant at *** p < 0.01, ** p < 0.05; t-values are in two parentheses. Table 6 presents the regression results for Tech_Stress. Columns (1) and (2) use Productivity as the dependent variable. Column (1) presents the results of the impact of the overall Technostress on Revenue Growth. Column (2) reports robustness-check results using ROE as an alternative measure of productivity, and column (3) replaces the dependent variable with lagged productivity for two-year lags.

shows the main results of multivariate analysis using a Fixed-Effect Regression model. Four separate regression models were conducted to examine the impact of technostress on the firms’ productivity. The first model is conducted to investigate the impact of technostress measured by a Python-based Streamlit application on the firm’s productivity, while the second one suggests a proxy to measure productivity to confirm and strengthen the previous findings of the first model. Model 3 considers the lag of productivity for two years to address endogeneity and control the unobserved factors.

The findings of the first model in Table 6, imply that technostress significantly decreases firm productivity at a coefficient of −0.001481 and t-value of −9.40645 (p-value < 1%). This result agrees with prior studies, which suggest that technostress is a huge issue with negative consequences on productivity and job outcomes (Berger et al., 2024; Dutta & Mishra, 2024; Saganuwan et al., 2013; Tarafdar et al., 2007). Also, the results of Model (2) confirm the findings of Model (1), at a coefficient of −0.00218 and a t-value of −2.33123 (p-value < 5%). Moreover, Model (3) strongly confirms the previous results at coefficients of −0.00631, also t-values of −2.2344 (p-values < 5%), reinforcing the long-term impact of technostress. Therefore, the results of Models (1) and (2) suggest a significant impact of technostress on productivity in the short term, while this impact becomes more significant in Model (3), suggesting the cumulative intensive technostress effect over time. Finally, Table 5 also presents the results of other control variables across the four models that contribute to the multivariate analysis.

The results confirm the negative effect of technostress on productivity in both the short and long term, with a deteriorating effect over time. This highlights the idea that technostress acts as an ongoing drag on firms’ efficiency. Urging the need for policy interventions to mitigate the technology-related stressors for sustained economic competitiveness.

By delving deeply into the most contributing dimensions of technostress to firms’ outcomes, we further investigate the impact of each dimension separately on firms’ productivity, as reported in

Table 7. Technostress dimensions contributing to firms’ productivity.

Variable	Model (1)	Model (2)	Model (3)	Model (4)	Model (5)	Model (6)
Techno-overload	−0.01099 (−1.33633)
Techno-invasion		−0.06671 ** (−2.18679)
Techno-complexity			−0.00646 (−0.91534)
Techno-insecurity				−0.01073 *** (−3.51852)
Techno-uncertainty					−0.031467 ** (−2.403959)
Techno-risk						0.00014 *** (−4.2684)
Firm_Size	−0.02004 (−0.09287)	−0.17506 (−0.97601)	−0.28871 (−1.22905)	−0.29299 (−1.23156)	0.336683 (0.145037)	0.000538 (1.053121)
IT_Capability	0.312841 ** (2.565339)	0.286006 ** (2.551512)	0.285656 ** (2.551361)	0.285659 ** (2.551427)	−1.068729 (−0.680302)	−5.42 × 10−7 (−0.07787)
LEV	0.150148 (0.868963)	0.166711 ** (2.377836)	0.058223 (0.847637)	0.058697 (0.85781)	0.229796 (1.346195)	0.003236 ** (2.020165)
CAP	−0.07326 (−0.66315)	−0.02756 (−0.54249)	−0.04081 (−0.88697)	−0.04391 (−0.96725)	−0.984641 (−1.093158)	1.40 × 10−7 (0.885502)
Report_Complex	−0.09183 (−0.25562)	0.03648 (0.180057)	0.049026 (0.246336)	0.051402 (0.259828)	−0.000139 *** (−3.1948)	0.002927 (1.397936)
Ass_Turover	0.256859 (1.324924)	−0.38243 (−0.9766)	−0.01209 (−0.07741)	−0.0201 (−0.12282)	−2.633196 ** (−2.026122)	−0.00782 ** (−2.19672)
C	−2.67929 (−0.86566)	−3.24619 ** (−2.7013)	−0.78589 (−0.75004)	−0.77921 (−0.74389)	35.51408 (1.63093)	−0.05706 (−1.41213)
Year Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
Firm Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
Adj.R2	0.503717	0.460097	0.459785	0.499794	0.291431	0.814958
F-statistic	6.692012	6.256657	5.743039	5.285412	3.055343	25.60266
Prob(F-statistic)	0.000	0.000	0.000	0.000	0.009045	0.000
Durbin–Watson stat	1.543719	1.523781	1.523395	1.523348	1.936821	1.141354
VIF.	2.014979	1.852185	1.851115	1.999176	1.411295	5.404179
Obs.	2453	2453	2453	2453	2453	2453

Notes: Significance levels: *** p < 0.01, ** p < 0.05; t statistics in parentheses. Table 6 presents the regression results for technostress dimensions individually, with Revenue Growth as a measure of Productivity. Column (1) presents the results of the Techno-overload dimension using all firms. Column (2) shows the results of the techno-invasion dimension, column (3) depicts the results techno-complexity dimension. The results of the techno-insecurity dimension are presented in column (4). Columns (5) and (6) explain the findings of techno-uncertainty and techno-risk dimensions, respectively.

. The findings of model (1) acknowledge that techno-overload, which captures the stress from excessive technology-induced workload demands, has a negative but insignificant impact on firms’ productivity, with a coefficient of 0.01099 and a t-value of −1.33633. While techno-invasion reduces productivity by overwhelming employees’ tasks and information, its effect has a significantly negative impact with a coefficient of −0.06671 and a t-value of −2.18679. However, the techno-complexity dimension, which stems from the complicated nature of technology, has a negative but insignificant impact on productivity, with a β-value of −0.00646 and a t-value of −0.91534.

Moreover, techno-insecurity, which refers to anxiety from perceived job threats due to technology, has a negative and significant impact on productivity at a β-value of −0.001703 and a t-value of −3,051,852. This implies that insecurity decreases productivity due to higher employee turnover intentions, reduced motivation, or defensive behaviors like resistance to technology adoption.

In the same vein, stress from unpredictable technological changes (tech uncertainty) yields a negative and significant impact on firms’ productivity, with a coefficient of −0.031467 and a t-value of −2.403959. This suggests that uncertainty disrupts planning and focus, decreasing productivity via increased anxiety or adaptation costs. Similarly, the new dimension of technostress, techno-risk, which encompasses threats such as cybersecurity risks or data breaches, has the most significant and negative association with productivity, with a β-value of −0.00014 and a t-value of −4.2684. These results indicate that perceived techno-risk is the strongest among technostress dimensions, significantly hindering productivity by diverting resources to reduce caution that slows operations, or due to the incidence of cybersecurity attacks that unfavorably impact the firms’ operating systems.

To sum up, these findings reported that technostress dimensions differently affect firms’ outcomes. Four of six technostress dimensions, overload, insecurity, uncertainty, and risk, negative impact on productivity, calling for advanced interventions such as programs for employee support, risk assessment, and technology design. These results provide novel evidence that technostress, as a silent factor in the modern workplace, urges firms to balance technological advancement with employee well-being for sustained outcomes. These results coincide with the empirical evidence from previous research (e.g., Zhang et al., 2022; Borle et al., 2021; Ingusci et al., 2021) that technostress dimensions exert varying effects on employee outcomes, with techno-overload and techno-insecurity consistently associated with heightened work exhaustion and behavioral stress. While techno-overload and -uncertainty frequently show insignificant relationships.

5. Robustness Test

To further check the validity of our main findings, we conduct a comprehensive set of robustness tests that address concerns regarding endogeneity, heteroscedasticity, serial correlation, and model specification. These tests demonstrate that our core result, which indicates a positive association between firm productivity and technostress, remains statistically significant and economically meaningful across alternative empirical tests. First, we address the potential reverse causality by employing lagged productivity measures and firm fixed effects. Also, we apply clustered standard errors at the firm level to account for heteroscedasticity and within-firm serial correlation, following Petersen (2009).

Model (2) explains the reverse-causality controlled model as follows:

Productivity_it = α₀ + β₁ Tech_Stress_it+1 + β₂ Controls_it + Firm FE+ Year FE + ε_it

Model (2)

Tech_Stress_it = α₀ + β₁ Productivity_{_}lag1+ β₂ Controls_it + Firm FE+ Year FE + ε_it

Model (3)

where Tech_Stress_it+1 is the technostress forward, and Productivity_lag1 represents 1 year lagged productivity measures (Rev growth_t−1 and ROE_t−1).

Table 8. Results of potential reverse causality with clustered standard errors at the firm’s level.

Variables	Panel A: Tech_Stress Forward		Panel B: Productivity Lag 1
	(1)	(2)	(1)	(2)
Tech_Stresst+1	3.529473 (0.653175)	0.000707 (0.789018)
Rev growtht−1			0.030848 (0.63616)
ROEt−1				0.023517 (1.154867)
Firm_Size	−4.670012 (−1.97137)	−2.15399 ** (−2.31159)	−0.14592 ** (−3.49248)	−4.750012 * (−1.91859)
IT_Capability	−0.64409 (−1.12278)	0.345928 (0.554371)	−2.98123 * (−1.8049)	−0.62207 (−0.90666)
LEV	0.217223 (0.597096)	0.000778 (0.275381)	0.430796 (0.582473)	0.226458 (0.637409)
CAP	0.000233 (0.201016)	0.157467 (1.099826)	−0.00051 (−0.2196)	0.000222 (0.327962)
Report_Complex	0.191818 (1.064525)	0.120472 (1.331389)	0.308073 (0.707427)	0.202068 (0.807663)
Ass_Turover	−0.02038 (−0.40393)	−0.0476 (−0.20424)	0.145258 (0.878958)	0.024437 (0.352478)
Big 4	−1.41949 (−5.8412)	3.302582 * (1.670937)	−0.11725 (−0.41508)	−1.41731 ** (−2.77275)
C	0.10738 (0.461543)	17.51315 * (1.800241)	2.986881 *** (9.417875)	0.065275 (0.187537)
Cross-section fixed (dummy variables)	Yes	Yes	Yes	Yes
Period fixed (dummy variables)	Yes	Yes	Yes	Yes
Cross-section weights (PCSE) standard errors and covariance (d.f. corrected)
Adj. R-squared	0.259222	0.383047	0.333971	0.289562
F-statistic	1.555969	2.219458	2.229635	1.461256
Prob(F-statistic)	0.000001	0.00000	0.000000	0.000026
DW	1.484854	2.325093	2.105547	1.613674
VIF.	1.349932	1.620869	1.501436	1.407582
Obs.	1477	1453	1477	1453

T-statistics in parentheses are based on firm-clustered standard errors. *** p < 0.01, ** p < 0.05, * p < 0.1. DW denotes the Durbin–Watson statistics for serial correlation.

reports the results for the reverse causality tests (i.e., whether low-performing firms disclose more technostress). Panel A is like our baseline model except that the technostress indicator (Tech_Stress) is one-year forward instead of lagged. Panel B considers technostress disclosure as the dependent variable and regresses it on the lagged ROE and revenue growth dummy to test if poor performance predicts higher stress disclosure. Within each panel, financial control variables (firm size, LEV, Ass_Turover, CAP), corporate governance variables (Big 4), and year fixed effects are introduced into different specifications.

The key findings indicate that the forward technostress is positive but insignificant in model 1 and 2, which reveal that there is no reverse causality. Neither lagged revenue growth nor ROE predicts future technostress disclosure. Firm-clustered standard errors address serial correlation and heteroscedasticity according to Petersen (2009). Panel A shows lagged ROE and lagged revenue growth positively predict future technostress disclosure. Panel B finds no significant effect of past poor productivity on current stress disclosure (p > 0.10), confirming our main result is not driven by low-performing firms disclosing more stress.

6. Conclusions

The new advancements in digitized technologies such as big data, cloud computing, blockchains, and artificial intelligence have become a strategic imperative across firms worldwide. One of the main challenges that face the implementation of these technologies is technostress. Various scholars investigate this technostress phenomenon, mainly measured by adopting the five dimensions proposed by Tarafdar et al. (2007) framework using questionnaires or interviews; however, these methods are characterized by subjectivity and respondents’ bias. Hence, this study presents a multidimensional framework using content analysis, which consists of the five dimensions introduced by Tarafdar et al. (2007) and a proposed new dimension of technostress, namely techno-risk, to capture the anxiety raised by different threats of technology adoption in business, such as data breaches and hacks.

Furthermore, this study introduces a new method to measure technostress, utilizing a refined list of 42 key terms and bigrams derived through factor analysis, and analyzes texts from 10-K annual reports using a Python-based Streamlit framework. The major orientation of this approach focuses on assessing the degree of technostress faced by firms, especially distress from their narrative disclosure. This approach represents a practical tool that includes objective metrics to capture technostress across firms and adopts natural language processing to implement content analysis, uncover hidden patterns and irregularities that signify technostress.

The main results of the textual analysis of the sample 10-K reports specify the primary contributors to technostress across each of the six dimensions of technostress. The findings present the key contributors to technostress with their frequency counts from the annual reports of US firms listed in the S&P Index.

We further validate the resulting list of key clues by investigating the aggregated and individual impact of technostress dimensions on the firm’s productivity. The main results imply that technostress is a vital factor that negatively affects firms’ outcomes, where technostress decreases the firm’s productivity in the short term, and this impact significantly increases, confirming the cumulative severe effect of technostress over the years. Particularly, regarding individual technostress dimensions, our findings indicate that techno-risk is the strongest factor in affecting firms’ outcomes, followed by insecurity, uncertainty, and invasion, respectively.

This novel multidimensional technostress framework offers various implications for firms and their employees. As it helps firms to quantify the technostress associated with their existing technological practices, they can tailor specific training programs to enhance employee satisfaction and sense of security. Also, firms will be able to pinpoint the main technostress dimensions that trigger stress; hence, they will develop appropriate technostress inhibitors to better manage stress. Moreover, this framework enables firms to assess the levels of technostress against industry benchmarks, allowing them to highlight potential areas for improvement. Therefore, firms will be able to garner the benefits of technological tools and better manage the technostress in their workplace to maintain a competitive edge and guarantee their resilience.

This study has a few limitations. First, the list of clues is written in the English language, which constrains its applicability to countries using different languages and different report formats. Hence, we recommend future research to translate the list of clues into various languages to enhance its global application. Second, the list of clues was developed and tested using 10-K reports, which limits its implementation in regions that adopt a different report format. So, we recommend future studies to apply this list in different report formats to enhance the possibility of its implementation worldwide. Third, this study analyzes firms’ annual reports written in formal and structural language, measuring high-level aggregated insights of technostress in corporate culture, without direct employee surveys. This may not fully capture management emotions toward technology, such as stress or anxiety that may be evident through verbal communication. Additionally, financial reports may hide certain challenges and emphasize positive outcomes to avoid a stock price crash, potentially limiting the accuracy of stress indicators derived from textual analysis. Future research could combine textual analysis with surveys or interviews to reveal management attitudes toward technology. Fourth and last, this framework tests the impact of technostress on a sample of US industrial firms, which may limit its applicability in different contexts. Thus, we recommend future research to investigate the impact of technostress within the developing countries’ context.