Accounting, Finance and Computational Intelligence

The misery index is a key indicator for assessing a society's economic condition and its impact on public welfare. An increase in this index signals an intensification of economic problems that affect the well-being of individuals. Consequently, the misery index functions as an important tool for policymakers to adopt appropriate decisions aimed at improving economic conditions. Particularly during periods of crisis and economic recession, this index serves as a warning signal, highlighting the urgent need for corrective policy measures. One of the variables influencing the misery index is the exchange rate. Given the significance of the misery index and recent exchange rate surges in recent years, the present study evaluates the causal relationship between the exchange rate and its volatility with the misery index over the period from Spring 2001 to Summer 2024. Considering the numerous economic fluctuations in the country and the likelihood of structural breaks, the study employs the Fourier approximation to account for structural breaks in the stationarity tests of the variables, cointegration, and causality tests. In addition, the GARCH method is used to estimate exchange rate uncertainty for the purpose of examining the causal relationship between exchange rate uncertainty and the misery index. The results indicate a unidirectional causal relationship from the exchange rate to the misery index and a bidirectional causal relationship between exchange rate uncertainty and the misery index. Therefore, exchange rate instability can have damaging effects on the economic welfare of the population (misery index). Since external shocks primarily influence the national economy—and subsequently inflation and unemployment—through the exchange rate and its fluctuations, it is essential to prioritize the monitoring and forecasting of such shocks and the use of tools that can enhance the resilience of the domestic economy against them.

The capital market is one of the most critical sectors of the economy, and its status is closely linked to the overall economic structure of a country. Due to the transparency and speed of transactions, the stock exchange is recognized as a significant investment option. In other words, an active stock market facilitates corporate financing and channels idle and often unproductive small-scale capital into productive ventures. In most countries, stock indices are considered one of the most reliable primary indicators for evaluating financial markets. The Tehran Stock Exchange's overall index reflects the general trend of Iran’s capital market and indicates the broader status of the investment landscape. Therefore, understanding and analyzing the overall index and its fluctuations can assist capital market participants in making more informed decisions. Additionally, the stock index is widely used by financial researchers to compare the performance of the capital market with other markets and to measure the impact of various factors on market returns. This study is applied in terms of its objective and descriptive-analytical in terms of its nature. It falls within the category of ex-post facto research (using data from March 20, 2012, to March 19, 2024). The estimation method employed in the study is the Structural Vector Autoregression (SVAR) approach. The present article examines behavioral modeling of stock index volatility with a focus on market risk and trading volume fluctuations. Initially, the research model is specified, followed by stationarity testing using the Phillips-Perron method. Next, cointegration tests are conducted using the Johansen-Juselius approach. Subsequently, the model is estimated, and stability tests, impulse response functions (IRF), and variance decomposition analyses are carried out. The results of model estimation indicate that the coefficients of most key variables influencing the Tehran Stock Exchange’s overall index volatility align with the theoretical foundations of the subject. The main variables that are essential and interpretable within the results of the SVAR model include impulses originating from market risk premium, global gold ounce price fluctuations, exchange rate fluctuations, Bitcoin price volatility, OPEC oil price volatility, and fluctuations in stock trading volume. Based on the estimation results, Bitcoin price and market risk premium have no significant impact on stock index volatility. However, the other explanatory variables in the model significantly affect stock index fluctuations at a 95% confidence level.

It can be confidently stated that sudden crashes in financial markets over the past hundred years have been among the most significant events in human societies—events that, in addition to inflicting substantial losses on a vast group of investors, have led to major decisions. The occurrence of recent global financial crises and the subsequent abrupt collapse in stock prices of companies in financial markets, which caused significant losses for numerous investors, has attracted the attention of many financial researchers and scholars toward the topic of financial market crashes and their prediction. The sharp decline in financial market prices causes substantial losses to investors’ wealth and undermines their trust in capital markets. This study investigates the estimation of the financial market crash rate model in Iran, with an emphasis on the dynamic behaviors of the free float stock index and dividend yield, using time series econometrics. The study period spans from 1996 to 2023 in Iran. The model estimation was conducted using the Autoregressive Distributed Lag (ARDL) technique. This study is applied in its objective and descriptive-analytical in nature, and it falls under the category of ex post facto research. According to the results of the model estimation, the coefficient of the free float stock index is −0.173730, with a corresponding p-value of 0.0046, indicating a statistically significant impact at the 95% confidence level on the financial market crash rate in Iran during the study period. The coefficient of the dividend yield index in the model is −0.213467, with a corresponding p-value of 0.0061, demonstrating that the dividend yield index also has a statistically significant impact on the financial market crash rate in Iran within a 5% error margin. The coefficients of the variables for the Top 50 Companies Index, the Financial Index, and the Industry Index (with one lag) are −0.949789, −0.755780, and −0.514797, respectively. The associated p-values are 0.0018, 0.190, and 0.0212, respectively, indicating that these variables also exert a negative and significant impact on the financial market crash rate in Iran at a 95% confidence level.

Banks play a vital role in creating liquidity and transforming risk in the economic operations of every country, both at the micro and macro levels. Commercial banks use relatively liquid liabilities to finance relatively illiquid assets, thereby releasing their individual liquidity to maintain the normal functioning of the financial system and promote economic development. Accordingly, the present study aims to identify the factors influencing the effectiveness of liquidity risk management in Iraqi banks. Relying on domain knowledge analysis and the qualitative content analysis model, the factors influencing the effectiveness of liquidity risk management in Iraqi banks were identified. Then, through a persuasive Delphi survey, 17 experts in the field of risk management were selected using a non-random method, and the most effective indicators for measuring the variables were evaluated and refined using the fuzzy network analysis model. Based on the final analyses, the effectiveness of risk management encompasses ten influential factors: organizational literacy, alignment and adaptability, competitive pressure, complexity, government support, managerial capabilities, market uncertainty, relative advantage, technical competencies, and stakeholder engagement. The measurement components for each of these factors were identified and refined. This study identified the factors affecting the effectiveness of liquidity risk management in Iraqi banks and subsequently proposed an effective model for measuring technical components.

The objective of the present study is to predict audit risk using the Long Short-Term Memory (LSTM) algorithm in companies listed on the Tehran Stock Exchange and to compare its results with those of other deep learning algorithms. To achieve this objective, a total of 1,650 firm-year observations (150 companies over 11 years) were collected from the annual financial reports of companies listed on the Tehran Stock Exchange during the period from 2013 to 2023. In this study, four deep learning algorithms—including Long Short-Term Memory (LSTM), Support Vector Machine (SVM), Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN)—were utilized. Additionally, to select the final research variables for model construction, the two-sample mean comparison test method was applied. The results of the deep learning algorithms show that the overall accuracy of the LSTM, SVM, CNN, and RNN algorithms was 99.1%, 89.6%, 85.8%, and 96.4%, respectively, indicating that the LSTM algorithm has the best performance and the CNN algorithm the poorest performance in predicting audit risk. In other words, the results demonstrate the superior efficiency of the Long Short-Term Memory (LSTM) algorithm compared to other deep learning algorithms. Therefore, in companies listed on the Tehran Stock Exchange, the LSTM algorithm provides the most efficient model for audit risk prediction. The findings of this study can offer useful insights into enhancing the prediction of audit risk and minimizing errors in evaluating financial statement information, improving the assessment of audit evidence based on data, and facilitating auditors’ ability to issue more reality-based opinions.

The objective of this study is to identify and explain the most probable future options related to the auditing profession over the next decade. In terms of nature, this study follows a mixed-method approach (qualitative–quantitative); in terms of purpose, it is applied; and in terms of methodology, it is descriptive–analytical. In the qualitative section, scenario analysis was used to determine the key driving forces influencing futures studies of information technology development in the auditing profession. In the quantitative section, the Delphi method was applied in two stages. The statistical population of the study includes all academic and professional experts in the field of information technology development in auditing—this includes senior managers at the Audit Organization, faculty members, and professors of accounting and auditing departments who have published articles or supervised theses on topics such as factors influencing the auditing profession, the future of accounting and auditing, and information technology development. To select the sample, a purposive non-random sampling method was applied, utilizing the theoretical saturation technique. After conducting 13 interviews, data redundancy was observed, and to ensure saturation, interviews were extended to 15 participants. To ensure data validity in the qualitative section, peer review criteria were used. For reliability assessment, an intercoder reliability index was calculated. In the quantitative section, content validity was confirmed through a survey of eight academic experts, and Cronbach's alpha coefficient was used to measure the reliability of the data. The two-stage Delphi analysis results revealed the acceptance of 16 driving forces in the domain of information technology development in the auditing profession. Interpretation of the findings indicates that with technological advancement in the future, clients will have more confidence in automated auditing than manual auditing. Due to the broader fraud detection capabilities of artificial intelligence, the relationship between auditors and clients will become more strained. Additionally, because of increased automation (resulting in easier and less costly processes), audit clients will find the current pricing of auditing services less reasonable. Therefore, despite the challenges posed by emerging technologies, they are expected to serve a supportive role for auditors.

The present study aims to explain the post-purchase behavior model with the objective of enhancing profitability, using the Importance–Performance Analysis (IPA) approach. This research is applied in nature, employs an exploratory-explanatory approach, and is conducted using qualitative methods. The statistical population consists of experts and professionals active in the home appliance industry, and a purposive snowball sampling technique was employed to select 10 participants for the study. To analyze the qualitative data obtained from interviews, content analysis and grounded theory methodology based on the Strauss and Corbin approach were applied, including open coding, axial coding, and selective coding. Based on this process, 25 components were identified and categorized under 9 main dimensions affecting the post-purchase behavior pattern of consumers of Iranian products aimed at improving profitability. The findings indicated that pre-purchase expectations and the perceived image of Iranian brands have a significant impact on post-purchase behavior, which in turn plays a key role in fostering loyalty to domestic brands. Furthermore, brand loyalty—mediated by contextual and intervening conditions such as social media and customer support—leads to both customer and corporate profitability. The Importance–Performance matrix analysis also revealed that variables such as social media and customer support services are in a favorable position in terms of both importance and performance. However, the variable "perceived image of Iranian brands," despite its high importance, shows relatively weak performance and requires strategic revision. Accordingly, reconfiguring the marketing system of domestic brands—by focusing on enhancing customer support services, redefining brand image, effectively managing social media, and adjusting consumer expectations—can pave the way for long-term and sustainable profitability in Iran’s competitive home appliance industry.

This study addresses the challenges of sentiment classification arising from the vastness of textual data by introducing a three-phase adversarial fine-tuning framework. In this framework, a base model is trained using a transformer encoder, multi-head attention, and a BiLSTM layer on preprocessed data, followed by the creation of a systematic generator that applies fourteen controlled perturbations at varying intensities to the adversarial datasets. In the base article, stage-wise training was conducted sequentially on preprocessed data and on different levels of adversarial datasets, demonstrating significant improvements particularly on a noisy dataset (level 3). Overall, this approach effectively enhances the robustness and reliability of sentiment classifiers in big data contexts with corrupted text. In the proposed method, by incorporating an optimized RCCN network and multi-agent SVM-based clustering, factors were re-clustered, resulting in the proposed algorithm achieving an accuracy rate of 97.86%.