Dawie Roodt and Garron Whitesman warn that a steep new levy could drive betting offshore. Their warning deserves scrutiny, but so does the argument that gambling harm is overstated. South Africa’s proposed online gambling tax poses a difficult question: can government reduce the damage from betting without pushing vulnerable players towards operators it struggles to police? That question remains open. Finance Minister Enoch Godongwana said on 21 October that consultations were continuing, with the final design to be announced in the February 2027 Budget . The announcement concerned the policy’s next stage, rather than the commencement of a new levy. In their discussion, economist Dawie Roodt and gambling lawyer Garron Whitesman argue that a punitive tax could weaken licensed businesses while leaving demand for gambling intact. Roodt approaches the issue through economic freedom and the consequences of changing incentives. Whitesman focuses on the commercial response: reduced spending, less attractive offers and customers moving to offshore competitors. Their perspectives matter. Roodt is Efficient Group’s chief economist . Whitesman’s practice represents gambling operators and suppliers , an industry connection relevant to assessing his argument. Neither expertise nor commercial proximity settles the policy question. The evidence has to do that work. First, get the tax base right Treasury’s November 2025 paper proposes an additional national tax of 20% on gross gambling revenue from online betting, including interactive gambling . Gross gambling revenue means stakes less winnings paid out. Operating expenses have not yet been deducted. The proposal would sit alongside provincial gambling taxes. In a simple illustration, R1,000 in stakes and R900 in payouts leave R100 in gross gambling revenue. A 20% levy on that amount is R20. The remaining R80 must still cover other taxes and operating costs. The assumed payout is illustrative, not an industry average. That distinction also explains why multiplying existing provincial tax receipts by 20% would not estimate the proposed levy. The relevant starting point is taxable gambling revenue, rather than tax already collected. The offshore warning has substance, and limits Whitesman’s strongest argument concerns substitution. If licensed operators respond to higher costs by worsening their offers, some customers may seek better terms elsewhere. His warning is principally about players migrating to existing offshore businesses, rather than established international companies deciding to operate illegally. But a plausible mechanism is not a measured South African outcome. How many players would switch, how much they would spend and how effectively enforcement could constrain that movement remain essential questions for any forecast of the tax’s effects. The Netherlands offers a useful warning. In its April 2026 monitoring release, the Dutch gambling regulator estimated that 91% of players used only legal providers, while the legal market accounted for 53% of gambling losses . Those percentages measure different things. A market can retain most players while losing a substantial share of spending to illegal operators. The regulator also reported that average monthly losses per player had fallen following protective measures, although they rose somewhat during the second half of 2025. It identified several possible explanations for changes in the legal market’s share, including player-protection rules. The findings support concern about illegal supply; they do not isolate taxation as the cause or establish that regulation achieved nothing. Britain illustrates another distinction. Its remote gaming duty rose from 21% to 40% from April 2026 . A separate 25% remote betting rate is scheduled for April 2027, with exceptions including UK horseracing. Treating every gambling product as subject to one identical increase obscures the policy choices. The British government says the higher gaming rate is intended partly to discourage companies from steering customers towards more harmful products. That is a stated objective, not proof of success. Equally, the existence of a tax increase is not proof that the resulting market will collapse. Treatment figures cannot measure the whole problem The discussion is less convincing when it uses the number of people receiving treatment to question the scale of gambling harm. Treatment records describe people who reached a service. They do not count everyone experiencing damage, everyone unable or unwilling to seek help, or affected family members. The World Health Organization describes harm below the threshold of a clinical disorder , including money diverted from essential household spending. It also identifies stigma as a barrier to seeking help. A person need not appear in a counselling register for gambling losses to affect food, rent or relationships. This does not establish a particular South African prevalence rate. It establishes why