The South African general anti-avoidance rule and historical case law

In the absence of case law where the current GAAR has been subject to interpretation and application by the courts (in the context provided by the judgment in the ABSA SCA case), hindsight cannot be used to determine whether the current GAAR represents an improvement over the previous GAAR. To determine whether the weaknesses of the previous GAAR were addressed by the 2006 amendments, further research of the efficacy of the current GAAR was necessary.

The method employed for court case selection, termed purposeful maximal sampling, chooses cases aligning with the research objectives. Another rationale for selecting a case is if it represents a critical case of the study. The selection process incorporated both purposeful maximal sampling as well as cases that were considered to be critical cases.

The population of case law, serving as the foundation for selecting relevant cases, was sourced from the LexisNexis database of South African Tax Cases Reports (LNTC Reports) as of 14 February 2024. The population of case law was derived by combining the results of two different searches on the LNTC Reports and eliminating duplicate cases from the search results as follows:

1. The search term “103(1)” was applied to the LNTC Reports, after filtering for: “Cases”, “South African Tax Case Reports” and “Tax cases” to identify cases where the provisions of the previous GAAR were considered in the courts. This search yielded a total of 71 cases.
2. The search term “tax avoidance” was applied to the LNTC Reports, after filtering for: “Tax cases”; “Avoidance of tax” and “Tax avoidance”. This search yielded a total of 51 cases.

After combining the results and removing the duplicates, the population of cases for the purposes of this study totalled 94 cases.

In refining the selection of cases pertaining to tax avoidance through the application of critical case law principles, the following qualitative criteria were employed to select cases aligning with the study's objectives:

1. Only those cases included in the LNTC Reports that had come before the South African courts were selected. Consequently, cases from other jurisdictions such as the Rhodesian High Court, the Zimbabwean Special Court and the Zimbabwean High Court were excluded from the selection process to maintain the study's focus and scope.

2. Cases not considering or applying the previous GAAR were excluded. Therefore, cases relating to anti-avoidance rules in the form of assessed losses (such as those outlined in Section 103(2) of the Income Tax Act) were excluded from the selection. These cases were eliminated in order to maintain the scope of the study and limit the study to that of the South African GAAR. In addition, cases where the current GAAR was considered were removed from the population to limit the study to that of the previous GAAR.

3. Cases that did not involve the avoidance of normal tax (for example, Secondary Tax on Companies and Value-Added Tax) were excluded. This exclusion is necessary due to the fact that the Commissioner is only able to apply Section 80B of the South African GAAR to normal tax matters. Consequently, any cases concerning the avoidance of taxes other than normal tax are excluded for the purposes of this study.

4. In applying the principles of purposeful maximal sampling, the scope and focus of the study was refined by excluding all cases presented in the Tax Court, as they do not create judicial precedent. Consequently, these rulings are only legally binding on the parties involved in the specific cases (van Coller, 2023). Therefore, cases brought before the Tax Court have been excluded for the purposes of this study.

Applying the above qualitative elimination criteria assisted in the unbiased selection of cases from the population. the file reflects the total population of cases and the analysis of those that met the elimination criteria described above.

NOTE: This data is not yet linked to an academic article but forms part of a large masters project where students will have this figshare data linked to the mini-dissertations (after the examination process is completed) that will be uploaded to figshare.