Franz Kafka’s The Trial narrates a parable that I have never quite been able to put down since law school, and I found myself thinking of it again while reading paragraph 6 of this judgment. A man from the country comes to the door of the Law and asks to be let in. A doorkeeper stands there, not refusing him outright, only telling him: not yet. The man waits. He waits for years, bargaining, growing old on a stool by the door, watching the door stand open the whole time, admission always theoretically available and never given. At the end, dying, he asks why no one else has ever come to this particular door. The doorkeeper tells him the door was made for him alone, and now that he is finished, it will be shut.
I do not want to overemphasise the comparison, because the man hidden in the middle of Kafka’s chapter nine never had an underlying claim worth anything. Stichting Rabobank Foundation had a very good one: an admitted debt of USD 230,868.51, correspondence in which both the borrower and the guarantor acknowledged owing it, and a High Court that struck the whole thing out before trial on the strength of a doorkeeper’s objection that Parliament, on a fair reading of the text, never authorised (para 16). That is the case in one sentence. Everything else is analysing how the Court of Appeal got there, whether I think it got there correctly, and what it left for the rest of us to sort out.
The facts, briefly
Stichting Rabobank Foundation v Mwangi & another [2026] KECA 1550 (KLR) was handed down by the Court of Appeal at Nairobi [“the Court”] (Korir, Ndolo & Hassan, JJ.A.) on 31st July 2026, reversing the High Court judgment (Mongare, J.) of 25th July 2024. More specifically, in October 2016, Stichting Rabobank Foundation, a Netherlands-incorporated entity, advanced roughly USD 180,116 to AVA Chem Limited under a Financial Support Agreement, with the company’s director, Christopher Irungu Mwangi, standing as guarantor. The parties restructured repayment in 2018. Both the company and the guarantor later acknowledged the debt in correspondence dated 28th April 2022. When repayment stalled, Rabobank sued in the Commercial Division of the High Court in September 2022 for USD 230,868.51 (paras 9–13).
AVA Chem did not contest owing the money. Its defence, raised by way of preliminary objection dated 24th May 2024, was founded on two propositions: that Rabobank was not a juristic person capable of instituting the suit, and that the Financial Support Agreement amounted to carrying on business in Kenya in contravention of Part XXXVII of the Companies Act, 2015, in particular section 974, without the registration the statute requires (para 15). Mongare J. agreed in July 2024, holding that non-registration was itself dispositive of locus standi, and struck out the suit (para 16). The Court of Appeal reversed on 31st July 2026.
The findings
I want to be precise about what the Court held, because the press coverage I have seen collapses three distinct holdings into one headline about “foreign lenders winning.”
First, the Court separated concepts the High Court had examined together without apparent difficulty: legal personality, capacity to sue, locus standi in the conventional sense, and statutory registration, which is a compliance question and nothing more (paras 40–43). A company incorporated in the Netherlands does not become a legal nullity the moment it crosses into Kenyan territory. That much I regard as uncontroversial, and overdue as a matter of stated Kenyan doctrine, since I have seen this same conflation surface in registration disputes under other statutes without a Court of Appeal bench saying so this plainly.
Second, the Court read section 974 for what it says rather than for what a worried borrower’s counsel might wish it said. Subsection (1) prohibits an unregistered foreign company from carrying on business in Kenya. Subsection (3) attaches a penalty to contravention. Nowhere does the section say that an unregistered company ceases to exist, that its contracts are void, or that it may not sue. Since Parliament specified a consequence for breach, the Court declined to graft on a second, harsher one that the text does not contain (paras 44–48). This is ordinary statutory construction, expressio unius in substance if not in name, and I find it hard to fault on the words of the section as drafted.
Third, and this is the part that will be most impactful in future litigation, the Court held that “carrying on business in Kenya” is a fact-sensitive inquiry into the nature, continuity, and territorial connection of the entity’s activities, not a label triggered automatically by cross-border lending; no single factor, including a solitary transaction, is necessarily decisive (paras 49–53). On section 974(2) specifically, the Court accepted that “includes (but is not limited to)” makes the statutory examples non-exhaustive, but held that a loan and a debenture are not invariably synonymous, and declined to treat the mere advancement of cross-border credit as automatically “offering debentures in Kenya” (paras 54–56). That question is not something a court can resolve on a preliminary objection where the underlying facts, on the location, frequency and character of the appellant’s activities, are contested rather than admitted (paras 51–53, 57–58). On that basis, the suit was reinstated for trial before a different judge, with the correctness of the preliminary objection alone now before the Court and the merits of the debt claim expressly left open (paras 67–68, 75).
The English comparison, where it helps, and where it does not
Kenyan commercial counsel quote the English position too casually, usually to say “well, England doesn’t require this,” without checking what England actually does. For this commentary, the comparison applies both ways.
England has had a registration regime for overseas companies with a UK establishment since the Companies Act 1985, now under Part 34 of the Companies Act 2006 and the Overseas Companies Regulations 2009 (SI 2009/1801). On the bare registration and filing obligations under Parts 2 and 3, non-compliance is treated exactly the way the Court of Appeal treated section 974(3): a summary offence carrying a fine, nothing more, and no suggestion that an unregistered overseas company loses its capacity to sue.
But English law does not stop there. Part 7 of the same Regulations, governing trading disclosures, contains a real court-access consequence, and a calibrated one. Regulation 66 provides that proceedings brought to enforce a contractual right, where the company failed to comply with the trading-disclosure requirements at the time the contract was made, must be dismissed if the defendant shows either that it had its own claim under the contract that it was unable to pursue because of the claimant’s non-compliance, or that it suffered financial loss connected with the contract because of that non-compliance, unless the court is satisfied that it is just and equitable to let the proceedings continue.
That is a genuinely measured middle position between “non-registration is fatal” (the High Court’s view here, and the Milimani Small Claims Court’s view of unlicensed lenders a fortnight later) and “non-registration is legally irrelevant to standing” (where our Court of Appeal ultimately landed). It is not automatic because it requires the defendant to show actual prejudice, and even then the court retains a discretion to let the claim proceed.
I do not say the Court of Appeal was wrong to decline to invent such a route out of section 974, which contains no equivalent of regulation 66’s prejudice-and-discretion structure and gives a court no textual permission to create one. But I do think the judgment would have been stronger, and more useful to the bench that has to apply it next, had it said so explicitly: that Kenyan law currently has no regulation 66, that this is a legislative gap rather than a judicial choice, and that Parliament could close it if it wished, without reopening the capacity question the Court has now closed. As it stands, a future defendant facing a genuinely well-founded prejudice argument, evidence that it lost the chance to bring its own claim, or suffered quantifiable loss, because it could not locate or serve the foreign claimant, has nothing in Kenyan statute to hinge that argument on. Rabobank deals with the blunt instrument without offering the fine one in its place.
On the substantive “carrying on business” question, the comparison is kinder to the Court of Appeal. English courts grappling with essentially the same phrase, for jurisdictional and enforcement purposes rather than under section 974’s type provisions, have long applied something close to the two-limb test from Adams v Cape Industries plc [1990] Ch 433: a fixed place of business maintained for more than a minimal period, from which the company’s own business is transacted. Our Court of Appeal’s list of relevant factors- the place where contracts are negotiated and concluded, the existence of an office or agents, the duration and continuity of activity, the degree of commercial presence- mirrors that English jurisprudence closely, if more loosely stated (para 50). The purposes are not identical, however. Adams was decided on jurisdiction over a foreign defendant rather than a domestic registration threshold, so the cross-reading is an analogy rather than an authority, but it gives Kenyan counsel something concrete to argue from rather than the open-ended “nature and extent” language of section 974(2) standing alone.
Where I disagree with the judgment, at least in part
Two things trouble me, and I say this as someone who thinks the outcome, on these facts, was correct.
The first is the complete silence on illegality doctrine, and here I want to be more careful than the judgment itself was tempted to be, rather than only asserting the answer I would prefer. AVA Chem’s argument, unclothed of its logical apparel, was not really a section 974 argument at all; it was an ex turpi causa argument dressed in registration language, namely that a party operating outside a Kenyan regulatory framework should not get the assistance of a Kenyan court in enforcing what it earned by operating outside that framework. The Court of Appeal never engages that argument on its own terms. Instead, it disposes of the “non-existent person cannot sue” submission in three short paragraphs (paras 59–61) without asking whether a genuinely well-pleaded illegality defence, premised not on section 974 but on the general law relating to contracts made in the course of unlawful business, might survive its reasoning even where a bare non-registration argument does not.
I do not think the answer to that question is as settled in Kenyan law as either side would like. English law itself abandoned a rigid rule in favour of Patel v Mirza [2016] UKSC 42, which asks courts to weigh the underlying purpose of the prohibition transgressed, whether denying the claim would enhance that purpose, any other public policy consequences of denial, and whether denial would be a proportionate response, rather than outrightly barring any claim that touches an illegal act (para 101). In my detailed review, I have not found a Kenyan appellate decision squarely adopting that trio of considerations in place of the older, stricter formulations of ex turpi causa non oritur actio still pleaded in our courts, which makes the doctrinal landscape AVA Chem’s argument would have to navigate considerably less settled, in either direction, than “it probably would survive” suggests. What I can say with more confidence is that it is a pending and separate question from the one Rabobank decided, and the Court’s total silence on it leaves it entirely open for the next case. It is worth noting that the Court itself offers the requisite approach for testing that gap: earlier in the judgment it warns that the label attached to an objection does not determine its character, and that an objection dressed as a statutory bar remains unsuitable for preliminary determination if it in truth depends on disputed facts (para 34). That is not a novel proposition; it is the ordinary rule from Mukisa Biscuit Manufacturing Co. Ltd v West End Distributors Ltd [1969] EA 696, that a preliminary objection raises a pure point of law argued on the ground that the other side’s facts are correct, and cannot properly be used to resolve what is a contested question of fact. Applied consistently, that rule operates against the Milimani rulings’ “illegality” label just as much as it outweighs AVA Chem’s “locus standi” formulation, since whether a given lender’s activities in fact amount to unlicensed banking business is no more a pure point of law than whether Rabobank’s lending amounted to carrying on business in Kenya. The Court failed to sufficiently deploy that approach to appraise the illegality question, because AVA Chem never squarely raised it.
The second is smaller but, for a court that says it wanted to lay down principles “given the importance of the issue and the possibility of recurrence” (para 65), avoidable. The eight numbered principles at paragraph 66 are genuinely useful, and I expect to see them pleaded from both sides within the year. But principle (vi), that non-registration does not “without express statutory language or necessary implication” deprive a company of standing, would have been considerably more useful to trial courts if the Court had spent even a paragraph on what would count as such an implication, rather than leaving the phrase as operative towards all the function it will inevitably be asked to perform in the next unregistered-claimant case that comes up.
The Milimani rulings, and why the resemblance is closer than it first looks, but not exact
Here I can speak from the primary record rather than press paraphrase, since both rulings are now indexed. On 17th July 2026, a fortnight before Rabobank, Resident Magistrate Gladys W. Kiamah, sitting at the Milimani Small Claims Court, struck out two debt-recovery suits brought by digital lending platforms against individual borrowers: Tri-State Capital Limited v Kang’ethe [2026] SCC 161 (KLR), where the claimant sought Kshs 500,000 on a facility of Kshs 213,500 secured against the respondent’s motor vehicle registration KBZ 817X (para 1), and Mombo iCapital Limited v Kariuki [2026] SCC 168 (KLR), where the claimant sought Kshs 162,297 on a Kshs 65,000 loan carrying Kshs 13,650 interest and a 3.5 percent weekly default penalty (para 1). Both rulings contain eight paragraphs and, paragraph for paragraph, the reasoning is nearly identical.
Before touching the merits in either case, the learned magistrate recorded that the claimant “is not licensed as a digital credit lender by CBK,” and held that section 3 of the Banking Act (Cap 488), which requires all persons or entities carrying on banking or financial business in Kenya to be licensed by the Central Bank, means that conducting lending business without such licensing “amounts to an illegality and economic risk” (para 2). The stated rationale is protective: a party who undertakes “regulated financial activities without the requisite authorisation cannot invoke the aid of the Court” to enforce rights arising from unlawful activity (para 3). For that proposition both rulings rely on Rei-Iman v Succeed Capital Limited (Civil Appeal E1374 of 2024; [2025] KEHC 19316 (KLR)), where the High Court held that allowing an unlicensed person to conduct a banking-type business would be against public policy, with the result that a suit filed by such a lender “was a nonstarter since the Respondent lacked the legal capacity to sue” (para 4). Applying that, the magistrate found in each case that the claimant had not demonstrated the legal capacity and regulatory authority necessary to engage in the lending activities disclosed in the statement of claim, and concluded, in materially identical terms across both rulings, that the claimant “lacks the locus standi to institute or file a suit” (paras 5–7). Both suits were struck out with no order as to costs; the underlying debts never determined on their merits (paras 6, 8).
I should highlight that Rei-Iman is itself a single High Court decision (Chigiti, J.), delivered on an appeal from a lower court, and its own use of “illegality” was tested against the traditional public-policy formulation of ex turpi causa rather than against anything resembling a purpose-and-proportionality analysis. It has not, so far as I have been able to establish, been considered by the Court of Appeal. A magistrate applying it as settled ground does not make it so; it makes it a doctrine built on foundations no higher court has yet examined.
Set the two rulings side by side, and the resemblance is real: an entity that has not complied with a Kenyan licensing requirement is told it cannot use the courts to enforce what it lent. That is exactly the inference the Court of Appeal rejected in Rabobank, paragraph after paragraph: a registration or licensing failure leads to penalty, the judgment says, not to a court-access disability Parliament never enacted.
There is a principled way to analyse the two rulings apart, and it deserves to be taken more seriously than being mentioned in passing. Section 974 is a company-registration statute with its own express, limited penalty, aimed at regulatory oversight of foreign entities; the Banking Act’s section 3 is a market-conduct licensing statute aimed squarely at protecting individual borrowers from unregulated, often predatory lending. If one takes the purpose of the prohibition seriously, as English law now does through Patel v Mirza, the two statutes may not point the same way. Denying Rabobank’s claim would have done nothing to advance section 974’s purpose, which is about the state’s regulatory framework for foreign entities, not the fairness of the underlying loan. Denying an unlicensed digital lender’s claim, by contrast, arguably advances the Banking Act’s protective purpose directly: it removes the financial incentive to lend without the authorisation that exists to protect borrowers such as the respondents in these two cases from unregulated, often punitive credit terms. On that reading, carrying on an unlicensed lending business is a materially different wrong from failing to register a foreign company, and Rabobank does not necessarily speak to it.
I am not persuaded that distinction fully survives the Court of Appeal’s own reasoning, but I hold that view more tentatively than I would have a paragraph ago. Both Milimani rulings transform a regulatory shortfall by the claimant into a threshold standing bar rather than a merits defence available to the defendant, which is the same invention the Court of Appeal criticised as an impermissible judicial graft onto legislative silence. Read paragraphs 2 and 3 of either Milimani rulings again with that criticism in mind: the magistrate moves directly from “not licensed” to “illegality” to “cannot invoke the aid of the Court,” without pausing on whether the Banking Act itself attaches that consequence to non-licensure, in much the same sequence the Court of Appeal faulted the High Court for following under section 974. Nothing in the Rabobank judgment mentions the Banking Act, CBK licensing, or illegality doctrine, and the Court pointedly declined to find, one way or the other, whether Rabobank itself was carrying on business in Kenya. A borrower facing a CBK-licensing objection now has a sound, post-Rabobank argument that locus standi and regulatory compliance are conceptually distinct questions under the Banking Act too. Whether that argument succeeds will likely turn on how firmly the word “illegality” in Rei-Iman and in the Milimani rulings is read: as a substantive contract defect going to the purpose of the Banking Act, which plausibly survives Rabobank’s reasoning on ordinary illegality principle, or as a standing bar dressed in illegality language, which almost certainly does not.
My final take
I think Rabobank is correctly decided on its own facts, and I think it is a necessary correction: the High Court’s approach had been hardening into a real obstacle for offshore credit into Kenya, and a foreign entity’s undisputed debt should not evaporate because of a filing it forgot to make at the Registrar of Companies. On that, I am fully with the Court of Appeal, and I say so having read the commentary from Spencer West, Bowmans, and DLA Piper Africa/IKM Advocates, the last one co-authored by my Parklands classmate and roomie Leonard Thuo, all of whom collectively highlight the ruling, correctly in my view, as restoring predictability for offshore lenders and guarantors funding Kenyan borrowers without a local branch.
The correction also arrives against a documented account of first-instance decisions going the other way, the same Rabobank ruling at first instance among them, which Cliffe Dekker Hofmeyr and Afriwise had each indicted, when the High Court ruling first came down, as reading “carrying on business” too loosely against the statutory examples in section 974(2). I read the Court of Appeal’s judgment as vindicating that criticism rather than breaking new ground of its own.
But I do not think the judgment is the complete statement of principle it presents itself as being. It resolves the registration question carefully and, in my view, correctly. It leaves the illegality question, which is the one informing the reasoning in the Milimani rulings and every unlicensed-lender dispute now filed through the courts, entirely untouched. It tests neither Rei-Iman’s foundations nor the purpose of the Banking Act against the reasoning it applied to section 974, and offers no middle ground of the kind English common law has curated into its own trading-disclosure regime.
Kafka’s doorkeeper never refused the man from the country outright. He let him wait, bargaining, growing old on a stool, in front of a door that was, on the doorkeeper’s own account, made for him and him alone, until the man died without ever finding out whether the answer would have been yes. Rabobank at least ends that particular wait: a foreign creditor with a genuine claim against a Kenyan borrower now knows the registration objection alone will not be allowed to keep it on the stool indefinitely. That much the Court did decide, solidly and finally.
But the door Rabobank closes was only ever the registration door. The illegality door, and the CBK-licensing door the Milimani litigants are sitting in front of, were never opened, never examined, and never even acknowledged as separate doors in this judgment. Advocates on both sides of the next unregistered or unlicensed-claimant dispute should read Rabobank for exactly what it decided, no more, and should test rather than assume whether its reasoning travels to a licensing statute the Court never mentioned. Whether it reaches that far is the genuinely open question this judgment leaves behind. But I surmise it won’t stay that way for long. Somebody is filing that appeal already, and until that appeal is heard, the unlicensed lender and the borrower resisting it are exactly where Rabobank itself was for two years: waiting outside a door nobody has yet told them is open or shut.
This briefing is for general information only and does not constitute legal advice. Paragraph citations in parentheses refer to the numbered paragraphs of the Rabobank judgment discussed in the sections, as it appears on Kenya Law. The Milimani rulings in section “The Milimani line” are each cited by their own internal paragraph numbers (1–8); other UK judgments, discussed elsewhere in the commentary, are cited by their own separate paragraph numbering.
Gody Mwango is an advocate at Mwango Law Advocates, specialising in constitutional litigation, judicial review, and commercial law. He is the founder and managing editor of Mwango Law Review.