نوع مقاله : مقاله پژوهشی

نویسندگان

• امین احمدپور ¹
• سیده محبوبه جعفری ²
• فاطمه صراف ¹

¹ دانشگاه آزاد اسلامی، واحد تهران جنوب

² دانشگاه آزاد اسلامی واحد تهران جنوب،تهران ،ایران

چکیده

در این تحقیق به بررسی تأثیر تحریم‌ها، بر فرار مالیاتی معاملات اشخاص وابسته بر اساس ترکیب رویکرد گراف کاوی-فراابتکاری فازی پرداخته شد. در این پژوهش تعداد1،780 شرکت دارای معاملات وابسته شامل 523 شرکت واقع در مناطق آزاد تجاری و 1،257 شرکت واقع در خارج از مناطق آزاد که دارای عضو هیأت مدیره مشترک و فعالیت اقتصادی تولیدی یا بازرگانی بوده انتخاب شده‌اند. دراین پژوهش داده‌های مالی ومالیاتی سال‌های 1395 لغایت 1399 ازاظهارنامه‌های مالیاتی و سامانه‌های سازمان امور مالیاتی کشور مورد استفاده قرار گرفته است. این پژوهش از نظر هدف، کاربردی می‌باشد. جهت برآورد مدل از نرم افزار پایتون و پکیج NetworkX و متلب 2021 بهره گرفته شده است. جهت پیش‌بینی فرار مالیاتی معاملات اشخاص وابسته از الگوریتم‌های فراابتکاری فازی غیرخطی از نوع لجستیک نوع 3 شفر نوع 4 بهره گرفته شد. بر اساس نتایج رویکرد مدل جگوار در مدل‌سازی گراف مابین شاخص‌های معاملات اشخاص وابسته با فرار مالیاتی از دقت بالاتری برخوردار بود؛ بر اساس نتایج معاملات اشخاص وابسته قبل از ورود مدل به تحریم موجب افزایش 389/0 درصد در فرار مالیاتی و بعد از ورود تحریم موجب افزایش 414/0 درصدی در فرار مالیاتی گردید. به عبارتی ورود تحریم به مدل موجب گردیده است؛ معاملات بیش‌تری در حوزه معاملات وابسته مشکوک در حوزه فرار مالیاتی شناسایی نموده است.

کلیدواژه‌ها

• گراف‌کاوی
• مدل جگوار
• گروه ‌های مشکوک
• فرار مالیاتی ‌
• معاملات اشخاص وابسته

موضوعات

• اجتناب از مالیات
• حسابداری مالی
• گزارش های حسابداری

عنوان مقاله [English]

The Impact of Sanctions on Tax Evasion in Related-Party Transactions: A Hybrid Graph Mining-Fuzzy Metaheuristic Approach

نویسندگان [English]

• Amin Ahmadpour ¹
• Seyedeh Mahboobeh Jafari ²
• Fatemeh Sarraf ¹

¹ Islamic Azad University, South Tehran Branch, Tehran, Iran

² Islamic Azad University , South Tehran Branch,

چکیده [English]

This study investigates the impact of economic sanctions on tax evasion facilitated through Related-Party Transactions (RPTs) in Iran. Utilizing a novel hybrid framework that integrates graph mining, Principal Component Analysis (PCA), and advanced fuzzy metaheuristic optimization, we analyze financial data from 1,780 companies (2016-2020). Graph mining is employed to map and detect suspicious transaction networks, particularly those involving Free Trade Zones (FTZs). A sanctions intensity index is constructed using PCA from 10 macroeconomic variables. The core predictive modeling leverages a Jaguar-optimized Type-3 Sheffer-like Type-4 fuzzy logic system to handle data uncertainty and non-linear relationships. Results indicate that sanctions exacerbate RPT-based tax evasion, increasing its magnitude from 0.389% to 0.414%. The proposed Jaguar model demonstrated superior performance with 98.8% accuracy (MSFE: 0.012), significantly outperforming traditional detection methods. Post-sanctions network topology analysis revealed a marked increase in suspicious clusters and nodes, with prevalent evasion patterns including multi-layer transfer pricing and abnormal profitability in FTZ subsidiaries. This research offers a robust, scalable tool for tax authorities to prioritize audits and enhances the understanding of how macroeconomic shocks influence illicit financial behaviors within corporate networks.

Introduction

Economic sanctions are coercive measures imposed by states to restrict international activities of target nations, offering a lower-risk alternative to military conflict (Cordesman et al., 2011). Iran exemplifies this, facing escalating sanctions that incentivize tax evasion through Related-Party Transactions (RPTs). Under sanctions, firms exploit legal gaps and accrual accounting to manipulate profits (Abeysekera, 2003; Arabi et al., 2018), transforming Iran’s financial market into a complex network (Soleimani et al., 2014). Traditional analytical methods fail against such complexity, while metaheuristic models excel. Graph mining uniquely uncovers hidden dimensions in sanctioned markets by analyzing network structures and variable relationships (Hu et al., 2022), especially where information asymmetry impedes tax authorities (Iacovacci & Lacasa, 2019; Yang & Xu, 2024).RPTs occur in nested networks with non-linear relationships (e.g., shared boards, cross-ownership) (Ruan et al., 2019). Sanctions amplify complexity through layered tactics like free trade zones (FTZs) and multi-layer transfer pricing (e.g., sequential sales at non-arm’s length prices) (Chan et al., 2016; Tian et al., 2016). Non-disclosure of ~68% key RPT information (e.g., pricing logic) exacerbates tax avoidance (Barokah, 2013), enabling profit shifting to foreign affiliates and eroding tax bases (Yang & Xu, 2024).Although RPTs can be economically justified (Gordon et al., 2004a), they risk abuse for private gain (Djankov et al., 2008; Barokah, 2013). In Iran, firms use subsidiaries in FTZs (e.g., Kish, Chabahar) and transfer pricing under Article 132-T of Iran’s Direct Taxation Law to shift profits: e.g., selling goods below market to affiliates, which then export at global prices, registering profits offshore. Weak oversight and fragmented databases hinder monitoring, but Iran’s Taxpayers’ Integrated System (TIS) provides foundational data for analysis.This study proposes a novel framework combining graph mining (to detect high-risk FTZ firms) and Type-3 Sheffer-like Type-4 fuzzy logic (to model tax data uncertainty) optimized by the Jaguar metaheuristic algorithm. It identifies suspicious groups exhibiting structural (e.g., nested ownership) and behavioral (e.g., abnormal pricing) tax evasion patterns, aligning with Iran’s Comprehensive Tax Plan for risk-based audits.
Research Questions:

Do economic sanctions increase RPT-based tax evasion?
How can advanced data analytics identify and model these hidden patterns?
Theoretical Framework

2.1. Related-Party Transactions (RPTs)
Per Iranian Accounting Standard 12 (Audit Organization, 2020), RPTs involve entities with control/influence over financial decisions. Key groups include:

Parent/subsidiary entities under shared control.
Key management personnel and relatives.
Entities with significant economic/management ties.

Two theoretical perspectives exist:
- Agency Theory:RPTs enable opportunism by insiders (Jensen & Meckling, 1976), e.g., underpriced asset sales (Cheung et al., 2006).
- Efficiency View: RPTs reduce transaction costs (Gordon et al., 2004a) but require disclosure to mitigate information asymmetry (Kohlbeck & Mayhew, 2010).
Empirical evidence confirms RPTs facilitate tax avoidance via transfer pricing (Harris et al., 1993; Jian & Wong, 2010), especially in low-tax jurisdictions (Barker et al., 2016).
2.2. Sanctions’ Economic Impact
Sanctions restrict input access, raise production costs (Parsa et al., 2013), contract import-reliant sectors (Caetano et al., 2023), and reduce total factor productivity (Nosratabadi, 2023). They incentivize shifting activities to the informal economy, causing technical inefficiency (Markus, 2024).

Methodology

3.1. Data & Variables
- Dependent Variable: Tax evasion, measured by the tax gap (difference between declared and final tax) per OECD standards (Slemrod & Weber, 2012).
- Independent Variable: RPT volume (Iranian Accounting Standard 12).
- Moderator: Sanctions index (PCA-derived from 10 macroeconomic variables, Table 1).
Data: 16,756 RPTs from 1,780 Iranian firms (2016–2020), including:

523 firms in FTZs (zero tax rate under Article 132-T).
1,257 non-FTZ firms with shared boards.

Financial data (net sales, COGS, operating profit) sourced confidentially from Iran’s National Tax Administration (INTA).
3.2. Integrated Framework

Graph Mining:


Construct transaction networks (nodes = firms; edges = RPTs weighted by price deviation).
Identify high-risk clusters(e.g., firms in FTZs with below-market pricing).


PCA for Sanctions Index:

- Combine 10 macroeconomic variables (e.g., oil exports, currency volatility) into a unified index.
- 2 principal components explain 85% variance (Table 1, Chart 3).

Fuzzy Metaheuristic Optimization:

- Apply Type-3 Sheffer-like Type-4 fuzzy logic to model data uncertainty (e.g., transfer pricing discrepancies).
- Optimize via Jaguar algorithm (multi-objective: minimize prediction error [MSFE], maximize detection accuracy).
- Output: Dynamic risk index (transaction volume, price deviation, geographic concentration).

Results & Discussion

- The analysis confirmed that sanctions significantly intensified RPT-based tax evasion, elevating its level from 0.389% (pre-sanctions) to 0.414% (post-sanctions). This 0.025% increase, though seemingly small, represents a substantial rise in hidden economic activity within the constrained environment.
- The Jaguar model achieved 98.8% accuracy (error rate: 0.012), outperforming traditional methods (40% vs. 74.6% detection rate).
- Graph analysis revealed post-sanctions topological shifts: increased suspicious nodes/clusters (Chart 4).
- Key evasion patterns:
- Multi-layer transfer pricing (e.g., mother → FTZ subsidiary → export).
- Abnormal profitability in FTZ subsidiaries.
- Geographic concentration in low-tax areas.

Conclusion & Policy Implications

5.1. Key Findings
Sanctions intensify RPT-based tax evasion by incentivizing complex, hidden transaction networks. The integrated graph-fuzzy-jaguar framework proves superior to linear models in detecting evasion under data uncertainty.
5.2. Innovations
- First application of Type-3 fuzzy logic in taxation.
- Dynamic risk index for audit prioritization.
- Operational compatibility with INTA’s existing systems (e.g., TIS).
5.3. Recommendations
- To INTA:Integrating the model into a blockchain-based real-time monitoring platform and Develop an AI dashboard with risk-tiered visualization (green/yellow/red).
- Domestic Policy: Mandating disclosure of transfer pricing logic and topological RPT networks and establishing a National Networked Data Analysis Center.
- International Cooperation:Leveraging double-taxation agreements for cross-border data exchange.
- Future Research: Extending the model to multinational contexts and designing "tax resilience indices" for sanction-affected economies.

کلیدواژه‌ها [English]

• Graph Mining
• Jaguar Algorithm
• Tax Evasion
• Economic Sanctions
• Related-Party Transactions