Our recently published article in the eJournal of Tax Research describes how Chi-square Automatic Interaction Detection (CHAID) analysis, a type of decision tree modelling, was used to study the tax compliance costs for over 10,000 individual taxpayers in South Africa. Tax compliance costs are an economic burden on society and can negatively affect taxpayer behaviour. The study aimed to provide a deeper, more detailed understanding of what determines these costs compared to other traditional methods like regression analysis.
Research Approach: Why CHAID?
Traditionally, studies on tax compliance costs primarily relied on regression and descriptive statistics. However, the researchers chose to employ the CHAID technique, alongside multiple linear regression, because CHAID offers several benefits over more common techniques, particularly for this type of study. These benefits include:
- Non-parametric and Non-linear: It works well with different types of data (including categorical variables) and doesn’t require the data to follow a standard distribution (normality).
- Segmented Insights: It divides the data into smaller subgroups based on statistically significant characteristics.
- Visual Output: It creates an easily understandable tree diagram that shows the determinants and their interactions with the dependent variable (tax compliance costs).
- Handles Complex Data: It can determine both linear and non-linear relationships and is not troubled by missing data.
Data Collection
The data were acquired from an online questionnaire distributed by the South African Revenue Service (SARS) to a stratified random sample of individual taxpayers in 2019. After data cleaning, the analysis was based on 10,260 fully completed questionnaires. The sample was found to be a good representation of the general population of individual taxpayers in South Africa. The questionnaire covered various factors (possible determinants) of tax compliance costs including: employment status; who completed the tax return; demographics (age, gender, location, education, income level); and scale items (taxpayer perceptions on tax legislation complexity, SARS service quality, appeals, audits/penalties, and SARS communication).
Key Findings from CHAID Analysis
The CHAID analysis revealed that tax compliance costs vary significantly across taxpayer segments and are influenced by factors such as employment status, income level, type of assistance, perceptions of service quality, and complexity of legislation.
It was found that employment status was the strongest predictor: self-employed taxpayers incurred the highest costs, while retired individuals without active income had the lowest. Other influential factors included:
- Type of assistance (paid help significantly increased costs)
- Income tax bracket (higher brackets correlated with higher costs)
- Perceptions of SARS service quality and communication (those with very positive perceptions had costs approximately half the cost of those with a less positive perception)
- Education level and gender (in certain subgroups)
- Complexity of tax legislation (especially for high-income full-time employed/unemployed taxpayers).
In respect of post-filing costs, these were most affected by SARS service quality and negative experiences with audits and appeals.
These insights go beyond what traditional regression methods can offer, enabling a more targeted approach to reducing compliance burdens.
Conclusion and Implications
The CHAID analysis provided nuanced insights into taxpayer compliance costs, revealing that these costs are not uniform but vary significantly across demographic and behavioural segments. Unlike traditional regression, CHAID allowed identification of specific taxpayer profiles most burdened by compliance costs.
The results of the CHAID analysis provided a granular level of insight not possible with other techniques. By identifying specific groups associated with high compliance costs (e.g., self-employed individuals with negative perceptions of SARS communication, or high-income individuals who find legislation complex), the revenue authority can target support initiatives to reduce the compliance burden for those specific taxpayers, which in turn can enhance overall compliance.
The following implications for Revenue Authorities were also highlighted:
- Targeted Support: Revenue Authorities should focus on self-employed taxpayers and high-income earners who perceive legislation as complex or service quality as poor.
- Service Quality Improvements: Revenue Authorities should enhance communication and consultation processes to reduce costs and improve taxpayer experience.
- Simplification Initiatives: Streamline legislation and filing processes for groups identified as most affected.
- Education and Assistance Programs: Offer tailored guidance for taxpayers with higher education levels and those relying on paid help, as these groups incur higher costs.
- Post-Filing Interventions: Improve audit and appeal procedures to minimize additional compliance burdens.
By applying these insights, revenue authorities can reduce compliance costs strategically, fostering greater voluntary compliance and trust in the tax system.
Recent Comments