Governing by concession: the backlog and the rule of directive

South African immigration law is supposed to be found in the Immigration Act and its Regulations. In practice, a large share of what decides whether a foreign national may stay, travel or work is now found elsewhere: in a sequence of ministerial directives, concessions and circulars, most of them issued to manage the consequences of delay. This commentary looks at that system critically. The Department of Home Affairs has made real progress on its backlog. But a state that governs its immigration system through emergency concessions, renewed every few months for four years, has a rule-of-law problem as well as an administrative one.

What the numbers show

The scale of the problem is not in dispute. In August 2024 the Minister, Dr Leon Schreiber, described a backlog of more than 300,000 pending cases.¹ A dedicated clearance team processed applications at speed, and by early 2025 the Department reported that more than 90 per cent had been cleared, with reports of nearly 94 per cent by the end of March 2025.² By September 2025 a ministerial directive stated that the backlog in the visa and permitting system had been eradicated, while acknowledging that waiver and appeal backlogs remained.³

Those remaining backlogs matter. In August 2026 the Department told Parliament's Select Committee on Security and Justice that 21,847 appeals were pending at the end of July 2026, of which 13,148 were already overdue. Its stated capacity is between 13,000 and 15,000 applications a month, and its target is to clear all overdue applications by 31 March 2027.⁴ The pattern is familiar to practitioners. Clearing first-instance decisions quickly produces more refusals, more refusals produce more appeals, and the appeal system becomes the new bottleneck.

Clearing a backlog at speed does not end it. It often moves it to the appeals desk.

Four years of concessions

Running alongside the backlog is a chain of concessions protecting people whose applications were pending. Since at least 2022 the Department has repeatedly extended the validity of existing visas, or protected applicants from being declared undesirable if they left the country, for people with pending long-term visa, waiver or appeal applications lodged through VFS Global.⁵ From September 2022 the extensions came roughly every six months: to March 2023, December 2023, June 2024, December 2024, March 2025, September 2025 and March 2026.

The most recent, Immigration Directive 7 of 2026, broke the rhythm. It extends protection for 15 months, to 30 June 2027, for waiver applications, long-term visa applications and long-term visa appeals pending at 30 March 2026.⁶ Permanent residence applicants are excluded, as they have been since 2025, as are short-term visitor extensions. Other directives deal with specific problems: Directive 10 of 2026 waived the medical report for applications made in South Africa, Directives 6 and 11 of 2026 gave relief to people stranded by airspace closures in the Middle East, and Directive 13 of 2026 waived the sixty-day filing rule.⁷ Circular 8 of 2026 then published target turnaround times, such as four weeks for critical skills and general work visas and eight months for some permanent residence categories.⁸

Governing by directive

Each of these instruments is defensible on its own. Together they raise three concerns.

The first is legal basis. Several directives, including Directive 13 of 2026, rely on section 31(2)(c) of the Immigration Act, which allows the Minister to waive prescribed requirements for good cause. The long-running concessions refer to section 30(1)(h), the undesirability provision, read with regulation 27(3), essentially undertaking not to apply it.⁹ Neither provision was obviously designed to suspend, for years at a time, the ordinary rules on visa validity for whole classes of people. When a waiver power becomes the main instrument of policy, the line between administering the law and rewriting it starts to blur.

The second is certainty. Coverage has shifted repeatedly and sometimes quickly. In September 2025, for example, permanent residence appeals were included in Directive 22 of 2025 and then removed by an amended version a week later.¹⁰ Long-term visa applicants were dropped from one directive and restored in the next. Relief depends on cut-off dates and on holding a VFS receipt. For an employer planning a secondment, or a family deciding whether a parent can travel to a funeral, that instability is itself a harm. Legal certainty, the idea that people can know in advance the rules that apply to them, is a core element of the rule of law.

The third is accountability. A regulation must be published and, under the Act, made after consultation. A directive can be issued, amended and withdrawn with little of that. The result is that some of the most consequential rules in the system are the least scrutinised.

The courts and the duty to decide

For individuals, the main remedy against delay remains litigation. Section 6(2)(g) of the Promotion of Administrative Justice Act allows review of a failure to take a decision, and section 6(3) allows an applicant to approach court where no decision has been taken within a reasonable time. The High Court has used those powers. In Ziaul v Minister of Home Affairs [2024] ZAGPPHC 1316, a case about unanswered asylum seeker visa extensions, the court ordered the Department to respond within ten days and decide within fourteen.¹¹ In Nlandu v Minister of Home Affairs [2025] ZAGPPHC 1351, where an asylum appeal decision had gone uncommunicated for about eleven years, the court substituted its own decision.¹² In 2026 the Department was held in contempt for failing to comply with a court order on citizenship.¹³

These cases are important, but they also show the limits of individual litigation. Each one helps one applicant who can afford to go to court, and the order is often needed only because the ordinary system failed. A case in the Western Cape in October 2026 was dismissed because the decision had been made in Pretoria and the court lacked jurisdiction, a reminder that even the remedy has its own procedural traps.¹⁴

What lasting reform would require

The revised draft White Paper of December 2025, approved by Cabinet in April 2026, proposes significant structural change, including a single independent review and appeals authority, specialised immigration courts and administrative fines instead of undesirability bans.¹⁵ Those proposals address the right problems. Until they become law, three steps would reduce the dependence on directives:

• Turn recurring concessions into regulation. If protection for pending applicants is needed every six months for four years, it is not an emergency. It is a rule and should be written as one, with clear criteria.
• Publish binding service standards. Circular 8 of 2026 is a good start. Turnaround times should be enforceable and reported on, so that delay has consequences inside the system rather than only in court.
• Fix the appeals stage. Better first-instance decisions with clear reasons would reduce the number of appeals. Faster decisions that create more appeals do not.

What applicants and employers should do now

• Check which directive applies, and its cut-off date. Coverage depends on when the application was lodged and on its category.
• Keep the VFS receipt. It is usually the document that proves the protection applies.
• Permanent residence applicants should keep a valid temporary visa. They are excluded from the main concession.
• Consider legal action where delay is unreasonable. The courts have shown they will order decisions within fixed periods. Take advice on the right court before filing.

Conclusion

The Department deserves credit for reducing a backlog that once ran into hundreds of thousands of files. But the method of managing what remains has become a problem in its own right. A system that depends on concessions renewed every few months, issued under waiver powers and changed at short notice, protects applicants only at the cost of legal certainty. The next stage of reform should not be measured only by how many files are closed. It should be measured by how much of the system can once again be found in the law itself.