Two Cases from Hong Kong's IP Financing Sandbox: A Patent and a Brand

Hong Kong launched its IP Financing Sandbox in December 2025, run by the HKMA together with the Commerce and Economic Development Bureau and the Intellectual Property Department. The problem it set out to solve is concrete: many small and medium enterprises hold technology and brands but no physical asset to pledge, so banks find it hard to lend. To borrow, someone first has to be able to say what that intellectual property is worth.

My involvement runs on two levels. On the professional side, I head a valuation and transaction advisory department regulated by the Royal Institution of Chartered Surveyors (RICS), and I am an MRICS. Since early 2025 I have taken part, as a representative of RICS, in the Bureau's and the Department's discussions on the IP valuation guideline. The sandbox's Operating Principles require the valuation to be carried out by an independent IP valuation service provider, and the provider must be a member of a recognised local or overseas professional body. RICS is one of them.

On the practical side, the first batch of seven pilot cases completed credit approval in 2026, with loans from HK$1 million to HK$39 million, spanning electronics, construction, toy manufacturing and medical devices. Two of the seven independent valuations were carried out by me and my team. By coincidence, those two were the smallest and the largest of the seven, the two ends of the range. This column is about what those two cases involved.

I cannot name the clients or the banks. The amounts and details below are given in broad direction.

Two cases: one patent, one brand

The first client is a Hong Kong listed company, valuing a Chinese invention patent. The technology is a lithium carbon capacitor, used in the battery modules of electric bicycles.

One detail I had not expected: this patent does not make its money by selling a product. It makes it by renting. It does not sell a battery and collect once. It rents the batteries out and collects the rent period by period, for every module deployed. Revenue is therefore the cumulative number of batteries rented multiplied by the rent per period, a curve that stacks upward, not a sales curve.

The second client is a long established Hong Kong toy manufacturer that spent years as an original equipment manufacturer for international brands, and had always earned from manufacturing. This was the first time it priced its own brand: two trademarks and a design application, for a new designer toy brand.

The two differ greatly in scale. The first valuation ran into the hundreds of millions. The second into the millions. The discount rates came out around thirty per cent for one, and about twenty five per cent for the other. The first took about three weeks from the client's first email to a signed report. The second took about two months from the first email to the report reaching the bank, and the extra time was mostly spent waiting on communication and internal process between the bank and the client.

A listed company, pricing its most valuable asset. An SME, pricing its intellectual property for the first time. Put together, the two draw the two ends of this policy.

Why the income approach, and why excess earnings

Internationally there are three routes to valuing intellectual property. The income approach is the most used: discount the returns from the asset over its economic life to present value. The market approach looks at transactions in similar assets, but every piece of intellectual property is unique, rarely sold on its own, and the transaction data is not public, so it is used least. The cost approach asks what it would cost to recreate an asset of equal utility, and suits assets that do not directly generate returns and have no comparables.

Both of my cases used the income approach, and the same method within it: the multi-period excess earnings method (MPEEM).

The logic sounds simple. The profit of a business is not earned entirely by its intellectual property. Plant, inventory, workforce and other intangible assets all contribute. Each contributing asset must be given the return it deserves and deducted in turn. What remains is the excess earnings, and once discounted, that is the value of the intellectual property.

One point deserves a pause. Even the workforce counts as a contributing asset, and must be given a return. In my case that workforce numbered more than a hundred, and was valued by replacement cost, that is, what it would take to recruit and train the same team, after tax. So valuation is not simply discounting profit. It is separating out, layer by layer, who did what.

The discount rate: not one number, but built item by item

The discount rate is the cell most easily misunderstood in this kind of valuation. It is not a number dropped in. It is assembled item by item.

The risk free rate: Hong Kong government bonds for Hong Kong, mainland government bonds for the mainland. The market risk premium: the recommended figures market institutions publish each year. The country risk premium: added for the political, economic and currency risk of where the business operates. The industry beta: unlevered from public industry data and relevered. The size premium and the company specific premium: one for the market, one for the risks this company alone carries. Together these build the cost of equity, which is weighted against the cost of debt by the capital structure to give the weighted average cost of capital, WACC.

Then one more layer, particular to this kind of valuation. Patents and trademarks carry more risk than physical assets, so an additional IP risk premium is added on top of WACC. In one of my cases it added roughly a tenth. That layer cannot be waved away with a sentence about intellectual property being riskier. The reasons must be written on the worksheet, one by one. The premium is not decoration. Whether it was justified is a question a bank can ask the moment it opens the file.

The hardest part is not the arithmetic

If I had to put the experience of these two cases into one line: the hard part is not the arithmetic. It is the questions that come before the arithmetic.

The first layer is legal, and legal comes before the numbers. Is this intellectual property real, and is it clean? That must be settled first. Has the patent application been granted? Is the chain of title complete? Have the assignments from employees and contractors been done? Are there joint inventors? Has it been pledged or licensed to anyone else? And one point many overlook: owning a patent does not mean you are free to manufacture. You may be standing on someone else's patent. Ownership and the right to use are two different things.

The second layer is remaining economic life, the shorter of legal life and economic life. A standard patent runs up to twenty years, a short term patent eight, a registered design twenty five, and a registered trademark can be renewed indefinitely. The protection can be long. The economic life is often far shorter. A toy design is in fashion for a few years. That answer is not in the statute.

The third layer is attribution of revenue: which part was earned by the intellectual property, and which by the plant and the people. The fourth is the forecast itself. A revenue curve is rarely flat, and every segment of that curve needs a reason.

The second case illustrates the first layer best. The client thought one patent application would be enough, that the several designs were much the same. My team and I took a different view. The company's licence documents gave it the right to produce, not the right to apply. That became the first reason in the risk premium. Put another way, what we were valuing was not really a patent. It was a contract. A licence document is not intellectual property. It is a contractual right, and it can only be valued separately under specific conditions.

What the bank looks at

Many enterprises assume the number on the valuation report is the answer. The bank does not see it that way.

What a bank cares about usually comes down to these: whether title is clean, whether renewals have been paid on time, whether there is a clear line between the intellectual property and the revenue, whether there is licence income, whether there is a dispute or a challenge to the right, and whether the report itself stands up to being read. In my two cases, what the banks asked about most was that line to revenue. Licence income is the hardest of all, because someone willing to pay a royalty is market evidence.

One more point must be put plainly: the valuation amount is not the loan amount. The bank applies its own loan to value, and looks at the borrower's cash flow, other collateral and the legal status of the intellectual property itself. Whether to lend, how much, for how long and at what rate is the bank's decision.

There is an institutional reason here that every enterprise hoping to borrow against intellectual property should know. Every loan a bank makes requires capital to be held against the risk, and eligible collateral reduces the charge. Basel III's CRE22 recognises only a few classes of collateral: cash, gold, rated debt securities, and equities in major indices. Intellectual property is not on that list. A loan secured on intellectual property therefore counts, for capital purposes, as if it were unsecured, and ties up more capital. Capital has a cost, and that cost returns in three places: pricing, loan to value, and tenor. Borrowing that is expensive, small and short is not an attitude. It is arithmetic. Banks are not unwilling. The framework does not encourage it, which is precisely why a sandbox is needed.

That said, banks do read, and they read well. Not long after the second report was delivered, the bank's colleagues raised two questions about a page of assumptions. The first: revenue is forecast to jump in a certain year, what are those new designs? The second: revenue jumps sharply but promotion costs do not rise with it, which does not fit the usual marketing spend pattern for new intellectual property. My team and I replied in writing the same day. A bank that asks is a good sign. It means it read the report properly. A report that can withstand that reading is a professional one.

What an enterprise should prepare

Gathering the work from both cases, I would offer three sets of preparation to any enterprise thinking of borrowing against its intellectual property.

The legal set. The chain of title must be clear: who invented, who applied, whether there was an assignment. The assignment documents from employees and contractors are the piece SMEs most often lack.

The numbers set. Revenue must be pointable: which product line, which contract, from which year. Where a group has subsidiaries, costs may sit in a subsidiary and the parent sees only a processing fee, and that layer must be explained. Renewal fees must be paid on time. A patent whose renewal has lapsed does not exist in law.

The process set. The appendix to my guideline carries an information request list you can prepare against. The most important line is this: do not wait until the day you need the bank's money to price your intellectual property for the first time. A valuation has a shelf life. The government also runs a Pilot Patent Valuation Support Scheme, offering one to one matching funding of up to HK$80,000 per enterprise.

The piece Hong Kong is missing

Policy, standard, quality, market: all four now exist in some form. Policy is led by government, with the Policy Address, the sandbox and the matching funding. Standard is the international valuation standards plus the local professional bodies. Quality is patent quality analysis. Market is the banks, with the sandbox, the patent box and the related tax provisions pushing it along.

One piece in the middle is missing: a local professional register and professional body. The International Valuation Standards Council sets standards. It is not a professional body, it has no individual members and no disciplinary power. Hong Kong still has no professional body that specifically governs intangible asset or business valuation. To borrow, someone has to be able to say what it is worth. Someone also has to be able to say who is qualified to say it.

What the two cases mean

The first case matters because it proves the route can be walked. For a patent to move from technology to an independent valuation a bank will read, someone has to tell the legal, technical and financial stories as one story. The second case matters because it lowers the threshold. A valuation in the millions is not a large number for a company with more than a hundred million in revenue, but it was the first time this company had priced its own intellectual property. For an SME, that is the starting point.

IP financing will not succeed on a policy alone. It needs people who can write the standard, people who can do the valuation, people who can read the report, and people willing to price a collateral nobody has seen before. These two cases are only the beginning.