Additional menu

Karsten Philip Aichholz

Spreadsheets In, Awesomeness Out.

Debitum Replied: 13 Pages on How Forests Work. Zero on Where the Money Went.

posted on

On March 26, Debitum Investments posted a 13-page statement (as well as in their own group) in response to my original Debitum Investments article.

I took a closer look to see if it changes anything I wrote about them.
(tl;dr: Not really. In fact, it might have made it worse. But it’s a fun read, so keep going.)

Let’s start with what they got right and I got wrong. There are two points:

  1. LFDF can’t pledge assets to someone else. The article stated: “Because investors hold no mortgage over it, nothing prevents LFDF from pledging the land to someone else.” Debitum cites Sections 11.16.2 and 11.16.3 of the LFDF prospectus, which contain negative pledge covenants. I’ve verified this against the prospectus. Section 11.16.3 states that LFDF will not “create or permit any Security over the Issuer’s assets” until full redemption. My statement that “nothing prevents” was wrong.
  2. The Intelligent Innovations pledge goes the other way. The article described Intelligent Innovations as holding a EUR 120 million pledge on LFDF shares. Debitum says the pledge is granted by Intelligent Innovations, pledging its LFDF shares to DN Trustee for investor benefit. The pledge registry confirms this: Intelligent Innovations is the pledgor, DN Trustee is the pledgee. I got the direction wrong.

I thank Debitum for taking the time to point out those inaccuracies.

The mistakes in the article have been fixed.

This consisted of deleting two sentences.

With that out of the way, let’s look at the rest.

The Timber Defense

Debitum’s response opens with its strongest card. Five pages, bolded text, an independent valuation, and the CEO of LFDF explaining how forests work.

The argument: land registry prices show only the land component. Timber (biological assets) is a separate, invisible value that explains the gap between what the family companies paid and what LFDF paid.

They’re right that timber can be a significant portion of a forest asset’s value.

I know that, because I wrote it myself in the same article they’re responding to:

“under Latvian Generally Accepted Accounting Principles (GAAP), inventory at cost can include timber purchase agreements, forestry management fees, and service commissions, none of which appear in the land registry. LFDF is a forestry fund. Timber rights may represent a significant share of the asset value.”

Debitum repeated that part in their response.

But I had another one, right after:

“That’s the optimistic reading. The less optimistic reading notes that the same network already collecting a 50% markup on land sales would also be in a position to collect commissions for ‘brokering’ deals between two companies that sit in the same office, or management fees for services rendered by entities sharing the same owner.”

That part they didn’t repeat. I wonder why.

The Latio Valuation

As part of their statement, Debitum commissioned a valuation from Latio, a reputable Latvian firm, for one of the seven properties in the 1,143% markup case study: “Laši,” cadastre 68860020099. Latio’s assessment: EUR 30,000 total market value (EUR 8,700 land, EUR 21,300 timber). EUR 6,976 per hectare.

Debitum valued one property out of seven that were all part of the same purchase contract.

To explain why that’s a problem, I need to back up a bit.

For the original article, I pulled the property register files. It’s about EUR 2 each and I pulled 680 of them. Then I manually copied over the price that LFDF paid, the company they bought it from, as well as the amount that company paid when they bought it. In addition, for each of those, I calculated how many days the company held it before it was sold to LFDF.

I didn’t do that because I enjoyed financially supporting Latvia’s property register or because I liked copy-pasting 3 days straight to get the necessary information. I did it because any partial excerpt didn’t make any sense.

When you look at how LFDF assigns prices to individual properties within a contract, the numbers are all over the place. The average markup across all JUNO-to-LFDF transactions is around 50%. But within a single contract, individual property markups can range from -76% to +2,567%. Some properties appear to be sold below what JUNO paid for them. Others at multiples. It’s not clear how LFDF distributes the purchase price across the properties in a contract. What is clear is that the contracts as a whole usually come out at a markup – 50% if you calculate across all properties surveyed.

This wild variation is overwhelmingly a network phenomenon. Non-network sellers show some spread within contracts, but the largest is 387 percentage points. For network sellers, 40 out of 82 multi-property contracts mix gains and losses across spreads exceeding 500 percentage points. The widest: 4,624 percentage points in a single BONO contract.

I had to pull every single property LFDF currently owned to make sure I got the right number.

Let’s get back to that valuation.

There’s a reason I quoted seven properties in the table in my article. Those were all part of the same purchase contract. Which means there is probably significant variation between the prices assigned within that contract. Following the pattern established above, some will be well below overall value, some way above. If individual property markups go from -76% to +2,567%, picking any single property out of that collection is potentially misleading.

Debitum picked a single property.

I like cherries as much as the next guy. But I try not to pick them when doing due diligence.

Interestingly enough though, even for that one handpicked example, the valuation and Debitum’s comments raise more questions than they answer:

  • JUNO paid EUR 1,422 to the previous owner
  • 150 days later, LFDF paid JUNO EUR 17,000. In the absolute best case scenario, that payment was only for the land. Let’s give Debitum the benefit of the doubt and say that was the case.
  • Latio says the property is worth EUR 30,000. However, only EUR 8,700 was for the land.

What does that tell us?

  1. JUNO gets fantastic deals (bummer that this family company isn’t liable to investors).
  2. Even in the valuation that Debitum presented, the land is worth half of what LFDF paid for it.

Even in the most charitable reading, the valuation is not making the point that Debitum thinks it does: 

It shows a 95% markup.

Cadastral Values

Speaking of land values, Debitum’s statement had an interesting claim: “In the table of the seven mentioned properties, the acquisition price shown represents only the cadastral value of the land plot.” It continues: “the cadastral value … can be significantly lower (sometimes up to 10 times lower) than the actual market value.”

Mistaking cadastral value for purchase price?

Boy would that be embarrassing if I had made that mistake.

Except I didn’t.

The Latvian land registry (Zemesgrāmata) records the contractual amount stated in the purchase agreement submitted for registration. Every entry references the specific contract: “2024.gada 12.janvāra pirkuma līgums Nr. PL-12-01/2024/2. Amount: 2,360.00 EUR.” That’s the price from the purchase agreement. Not a cadastral assessment. Not a tax valuation.

I know this because I have the registry entries for all 652 properties out of 680 total that showed a price. Every price in the article was extracted from the “Pamats” (basis) field of the land registry record, which cites the purchase agreement and the amount. Cadastral values exist separately in Latvia (set by the State Land Service for tax purposes) but they are not what appears in these records.

To settle this, I obtained the official cadastral values for all seven properties from Latvia’s State Land Service open data (Valsts zemes dienests, published March 18, 2026). Here they are, next to the contractual prices from the land registry:

CadastreNameCadastral ValueJUNO PaidLFDF Paid
44720030034VitasEUR 537EUR 537EUR 16,500
44700030270KangariEUR 479EUR 350EUR 7,000
88700020041Jaunčakšu mežsEUR 893EUR 893EUR 14,000
62720040007StrēļiEUR 1,116EUR 1,116EUR 14,500
68860020099LašiEUR 722EUR 1,422EUR 17,000
68600010432Vecie RunčiEUR 802EUR 4,388EUR 52,000
66880100012VecupītesEUR 380EUR 1,750EUR 9,000

Sources:

Three things jump out.

First, the cadastral values and JUNO’s purchase prices are different columns. They’re close for some properties, far apart for others. The land registry records the transaction price, not the cadastral value. Debitum’s central claim (“the acquisition price shown represents only the cadastral value”) is factually wrong.

Second, for three of the seven properties (Vitas, Jaunčakšu mežs, Strēļi), JUNO paid exactly the cadastral value to the cent. Then JUNO sold them to LFDF at 13-31x that price.

Third, compare the columns. JUNO’s purchase prices average EUR 1,494. The cadastral values average EUR 704. LFDF’s prices average EUR 18,571. JUNO buys at or near the cadastral floor. LFDF pays 12x more.

If Debitum’s argument is that the registry prices are just cadastral values, the data shows the opposite: JUNO is buying at cadastral prices and LFDF is paying far above them. The markup is real, measured between two contractual prices recorded in the same registry, and for three of these seven properties, JUNO’s side of the transaction is literally the tax-assessment minimum.

The Timber Markup

One possible explanation: JUNO paid the cadastral-value price for the land and separately purchased timber rights in a transaction not recorded in the registry. If so, JUNO could show that documentation. They haven’t. In five pages about timber value, they don’t produce a single timber purchase contract.

They do, however, explain why they can’t share the contracts. The response states that “detailed valuation data is typically not publicly disclosed due to competitive and commercial sensitivity.”

I’m not asking for the valuation methodology. I’m asking what they paid. The purchase contracts would show who sold, who bought, how much land, how much timber, what price. That’s not exactly the secret Coca-Cola formula.

When the buyer’s money comes from retail investors, transparency about what was paid is not optional.

But here is the bigger problem with the timber defense. If timber is a separate, significant value component not captured in the land registry, then there were timber transactions on both sides: when JUNO bought from the original owner, and when LFDF bought from JUNO. The article measures only the markup visible in the land registry. If Debitum is right that timber represents the majority of a forest property’s value, the question is: did JUNO mark up the timber too?

If they did, the 50% average markup the article documents is the floor, not the ceiling. The total extraction from investors would be larger, not smaller. Debitum’s timber defense doesn’t shrink the problem. It raises the possibility that the problem is bigger than what the public records show.

If they didn’t, if JUNO passed timber through at cost, that would be straightforward to demonstrate with the purchase contracts.

Where’s the Breakdown?

The article asked where LFDF’s EUR 24.6 million in inventory went. Debitum’s response gave five pages about how forests contain timber.

That was never the question. The question was: how much went to land, how much to timber, how much to management fees, and how much to the five family companies that sold 81% of it?

Two days later: still no breakdown.

“34 Cents Per Euro Goes to Insiders”

Their response to the headline finding is a single paragraph restating the timber argument. Perhaps they don’t want to say more, because if the markup wasn’t just on the land but also on the timber, the real number would be higher. The timber defense doesn’t reduce the 34 cents. It raises the possibility that it’s worse.

Governance: Latvia Is Small

Debitum’s governance defense comes in two parts. First, a list of internal controls: AML checks, conflict-of-interest registers, Credit Committee approval, UBO verification. Second, a broader argument about Latvia being a small market where “it is common that companies have worked with each other historically.”

The article doesn’t claim that professional overlap is prohibited. It claims that the overlapping relationships were not disclosed in filings that legally require such disclosure. The question was never “do these people know each other?” The question is: why do their annual reports say they don’t?

False Filings: The Wrong Standard

This is where the response falls apart.

Debitum writes: “disclosure is required only where there is actual ownership, control, or significant influence. Beneficial ownership is typically defined at 25% or more ownership or control.”

That’s the AML beneficial ownership threshold from Directive 2015/849. It determines who qualifies as a “beneficial owner” for anti-money laundering purposes. It has nothing to do with related-party disclosure in annual reports.

The article’s fourteen false filing claims are based on IAS 24, the international accounting standard for related-party disclosure. Latvian accounting law defines “related parties” by direct reference to IAS 24 (and, separately, through the Latvian Annual Financial Statements Law, Article 1, Section 1(18), which defines related parties through shared management and family). Neither standard uses a 25% threshold. IAS 24 explicitly includes: key management personnel, close family members of key management personnel, and entities controlled by any of those people.

Under IAS 24, the relationships documented in the article aren’t borderline cases:

  • Galvanovskis controls AS JUNO. His wife (close family member under IAS 24) controls BONO. Entities controlled by close family members of a person who controls another entity are related parties. Textbook.
  • Andžejevskis married into the Galvanovskis family, owns Foresto, and chairs BONO’s board. The close-family relationship combined with control of Foresto and a board chair role at BONO creates exactly the kind of influence IAS 24 is designed to capture.
  • Melderis simultaneously serves on the boards of AS JUNO (a major seller to LFDF), LFDF itself (the buyer), and Intelligent Innovations (LFDF’s shareholder). IAS 24 does note that shared directorships alone don’t automatically create a related-party relationship. But Melderis isn’t just a name that appears on two letterheads. He has a management role in the entity buying land, the entity selling it, and the entity that owns the buyer’s shares. If that combination doesn’t warrant disclosure, the standard has no teeth.
  • Lezdiņš is Head of Sales at JUNO and owns Baltic Terra Capital. If his role at JUNO constitutes key management personnel, his ownership of BTC makes it a related party.

Debitum’s response cites a 25% AML threshold that has nothing to do with any of this. IAS 24 asks whether the people involved have control, significant influence, or close family ties. A spouse is enough. A wife who owns the company that sells EUR 3.7 million in land to the entity her husband’s company also sells to is enough. Debitum’s response cites a standard that doesn’t apply to the question the article asks.

Debitum then offers a partial concession: “LFDF is related to AS Juno (through a shared board member) and Intelligent Innovations (through shareholding).” They acknowledge those two. For the rest, they state: “Neither LFDF nor Terra are related parties to Debitum or other parties listed in the article.”

The article lists fourteen specific false filings. The response addresses one (LFDF/JUNO, which it concedes) and half-addresses another (BTC, which it denies). It says nothing about the remaining twelve. Nothing about BONO’s eight years of blank related-party fields while carrying EUR 2.8 million in related-company receivables on its own balance sheet. Nothing about JUNO’s EUR 5.3 million in undisclosed related-party debt. Nothing about Juno Finance’s annual report that denies having related companies on the same page where its P&L reports related-party interest income.

Thirteen pages. Fourteen false filings. Twelve of them are not even mentioned once in their statement.

Collateral: Right, But

On collateral, I already noted the correction above. But Debitum’s own prospectus tells the rest of the story.

Section 4.4.3: “The Collateral does not include the Land Portfolio meaning that in the Notes are not secured with the Land Portfolio and in case of enforcement of the Collateral or in case of insolvency of the Issuer in respect to the Land Portfolio the Noteholders would be regarded as unsecured creditors.”

Section 4.4.4 goes further: the commercial pledge explicitly excludes real estate.

So the full picture is: LFDF can’t voluntarily pledge the land to someone else (negative covenant, corrected above). But if LFDF becomes insolvent, investors are unsecured creditors with respect to the land. The protection is contractual, not proprietary. A negative covenant protects against voluntary acts by the company. It doesn’t protect against insolvency, and it doesn’t give investors a secured claim on the assets.

The primary asset of a forestry fund sits outside the secured claim. Debitum’s own prospectus says so. The negative pledge covenant doesn’t change that.

And investors wouldn’t be the only unsecured creditors. At year-end 2024, LFDF’s balance sheet showed EUR 5,163,401 owed to related companies, 55% of its entire balance sheet at the time. Since then, LFDF has grown from EUR 9.3 million to EUR 43.9 million in total assets. The 2025 audited report hasn’t been filed yet, so we don’t know what the current related-party debt figure is. But in an insolvency, investors would be sharing the unsecured pool with the same family companies that sold LFDF the land at a 50% markup.

PEP: Not the Point

Debitum responds that “PEP status on its own does not indicate wrongdoing” and that prospectus disclosures don’t typically include AML classifications.

The article doesn’t claim PEP status is wrongdoing. It notes that the LFDF prospectus has a section titled “Beneficial Owner Interest Risk” that describes risks arising from the beneficial owner’s interests. That section does not mention that the beneficial owner is the sitting Parliamentary Secretary at the Ministry of Finance. A prospectus section specifically about beneficial owner risk that omits the most material fact about the beneficial owner is an omission.

Sandbox: Trust Us

Debitum claims the Sandbox relationship is “transparent and publicly disclosed.” They say SPVs have direct claims against borrowers, Sandbox has pledged receivables, and ZIdea loans are not securitized into the ABS and therefore don’t expose investor funds.

That last point is actually worse than it sounds. If the ZIdea loans were securitized, investors would at least have a claim on them through the SPV. By keeping them off the ABS, Sandbox is lending EUR 290,000 at 1.66% from its own balance sheet to an entity owned by the platform’s founder. That’s value being removed from Sandbox at below-market rates. Sandbox is the entity investors actually have exposure to. Anything that weakens Sandbox’s financial position weakens the investors’ position. The “not securitized” defense doesn’t protect investors. It just means they have no claim on the loan that’s draining the company they depend on.

They also don’t address the buyback clause enforceability. If a borrower defaults, Sandbox is supposed to buy back the loan. But Salmiņš owns both Sandbox and the platform. The person deciding whether to enforce the buyback is the same person who would have to pay for it.

Asset Valuation: General Forestry Facts

Debitum cites certified valuators (“where applicable”), 7% annual forest growth, and average Latvian forest prices of EUR 4,000-5,000 per hectare.

“Where applicable” is doing heavy work. How often? For which transactions? All 652? The 7% growth figure from LFDF CEO Edgars Birks refers to biological growth of the asset (mean annual increment), not financial arbitrage. Markups of 96.9% annualized over 203 days are not generated by trees growing.

Refinancing: Maybe

Debitum points out that Note refinancing is standard practice in bond markets. They’re right. “Refinancing” is also standard practice in other things, but let’s not go there.

Long story short: something happening elsewhere says nothing about what it means at Debitum.

Ukraine: Not the Issue

Debitum provides context on the DN Funding Alpha situation: martial law prevents servicing, Motor Finance is not bankrupt, the position is restructured rather than defaulted. The war is real and the martial law context is fair.

But the article’s finding wasn’t about Ukraine. It was that the platform moved the obligations into a subsidiary to preserve its “zero defaults” track record, and that DN Alpha’s 2024 annual report filed with blank related-party fields despite being 100% owned by DN Operator. The response addresses the war. It doesn’t address the accounting.

“Zero Compliance Breaches”

Debitum defines “zero compliance breaches” as “no confirmed violations of regulatory or AML requirements identified by the Bank of Latvia or through our internal compliance monitoring.”

By that standard, every company has zero compliance breaches until someone catches them. That’s not a compliance record. That’s the absence of enforcement.

They also don’t address the Trust Score methodology that gives network companies A and B+ while independent issuers get B and C.

Bono House: Repaid, But How?

Debitum states: “No investor funds are currently exposed to Bono House.”

Bono House had no revenue and negative equity of EUR 488K. It repaid EUR 990K to investors. Debitum confirms it was repaid. They don’t say how.

The balance sheet shows EUR 1.5 million in unnamed borrowings. If that lender is itself funded by investors on the same platform, the exposure didn’t disappear. It just has one more entity between the investor and the problem. No direct exposure, Debitum would say.

What Thirteen Pages Don’t Mention

Twelve sections. Not one of them addresses:

  • The 203-day average holding period and the 96.9% annualized return for network sellers vs 5.6% for non-network sellers
  • Five undisclosed intermediaries in the LFDF prospectus (289 properties, EUR 5.9 million)
  • Henrijs Jansons’ simultaneous board seats across DN Operator, Sandbox, and all six SPVs, all dismissed on a single date
  • The shared accountant preparing both sides of every transaction and producing reports that carry the header of the wrong company
  • EUR 905K in network lending to Sandbox (AS JUNO EUR 280K + Ozolu meži EUR 625K)
  • Sandbox’s EUR 60M blanket pledge registered 37 days after the balance sheet date, undisclosed in the annual report
  • The Trust Score methodology that gives network companies A ratings and independent issuers B/C

These findings are sourced to public registries. The land registry records what it records. The annual reports say what they say. The pledge registry is the pledge registry.

The Scorecard

Debitum published thirteen pages. They identified two errors in a 4,500-word article. Both have been corrected. Both consisted of single sentences.

Everything else in the response either restates what the article already acknowledged (timber is a real component of forest value), cites the wrong legal standard (AML thresholds for an IAS 24 question), or simply doesn’t address the finding at all (twelve of fourteen false filings, the shared accountant, the 203-day flip).

The article raises fourteen false filings. The response redefines “related party” using anti-money laundering thresholds instead of the accounting standard that Latvian law actually requires, then concludes there is nothing to disclose.

The article documents a 50% markup by family intermediaries across 484 matched-pair transactions. The response explains that forests contain timber, which the article already acknowledged on line 158, then presents that explanation as if it addresses the markup.

The response does drop new LFDF figures: EUR 43.9M total assets, EUR 10.85M net turnover, EUR 1.58M net profit for 2025. These figures are interesting. They don’t address the markup or the self-dealing.

Debitum closes with an invitation: “We welcome investors and opinion leaders to visit our financing partners and gain a deeper, first-hand understanding of how these businesses operate.”

I would welcome something simpler. A breakdown of where the EUR 24.6 million went. The purchase contracts for both sides of any of the 484 properties where the intermediary bought and LFDF bought the same plot. And an answer to the questions their thirteen pages didn’t mention.