Haute Lumière
Commerce · III.06 · MMXXVI · daylight
ASSESSMENT · Chapter III.06 — Mycelial Finance
Three instruments: a ten-point quiz, eight reflection questions, five essay prompts. The quiz checks comprehension rather than recall. The reflections are private and first-person. The essays are arguable from more than one side.
THE QUIZ — ten points
Four on recall.
1. Define a common mycorrhizal network precisely, and say why two trees colonised by the same fungal species may not be connected.
A CMN exists when the hyphae of one fungal individual connect the roots of two or more plants. Shared species is not shared genet: many ectomycorrhizal fungi form small, transient individuals, so two neighbouring trees can carry the same species and no shared network. One mark for the genet distinction — it is the distinction that collapses most of the popular claim.
2. State the three claims Karst, Jones and Hoeksema (2023) evaluated, and the verdict on each.
(i) CMNs are widespread in forests — insufficiently supported; structure has been genotype-mapped in five studies across two forest types, two tree species of about 73,300, three fungal species, with two studies showing actual link continuity. (ii) Resources move through CMNs and improve seedling performance — insufficiently supported; of 26 field studies, every result can be explained without invoking a network, because barriers that block hyphae also block soil solution. (iii) Mature trees preferentially send resources and defence signals to their offspring — no peer-reviewed published evidence.
3. What did Klein, Siegwolf and Körner measure in 2016, and what word did they use for it?
Roughly 40 percent of fine-root carbon in neighbouring mature trees derived from a labelled donor, about 280 kg C per hectare per year, bidirectional, over five years in a real forest. Their word was trade. Credit any answer noting that the route — hyphae, overlapping root spheres or soil solution — was not isolated.
4. State the assessment-cost floor and what sets it.
The cost of deciding whether to lend does not fall with the size of the loan. At an assumed $3,000 per relationship assessment and a 9.0 percent margin on principal (6.0 percent spread × 1.5 years), the breakeven ticket is $33,333; at $25 by machine it is $278. One mark for the mechanism, one for either figure.
Four on application.
5. A colleague proposes a "solidarity fund" in which members lend to each other because they care about each other, citing the wood wide web. Using only the surviving biology, give the strongest correction — and then the strongest version of their proposal.
The surviving biology does not support solidarity: Kiers (2011) shows reciprocal, delivery-proportional reward, and Whiteside (2019) shows fungi moving phosphorus to where the exchange rate is better. That is discriminating trade, not sharing. The strongest version of their proposal keeps the topology — small transfers, many counterparties, decision at the edge — and replaces solidarity with an observable local signal and an intermediary on risk. Full marks require the second half: the correction is only useful if it makes their idea better.
6. A platform proposes to originate 4,000 loans, charge a 2 percent origination fee, hold no loans on its own balance sheet, and pass all credit risk to retail lenders. Diagnose it.
Volume is the only variable the platform controls, and its revenue rises with volume regardless of outcome — so origination standards have nothing holding them. The correction is a retained vertical slice, pari passu, written into the first document. The stronger answer notes that RateSetter, LendingClub and Zopa all resolved this the same way between 2020 and 2021, by becoming principals rather than by abandoning distributed origination.
7. You are asked to add a two-page application form to a network whose assessment currently costs about $40 of a node's time. What is the real cost of the form?
If the form takes an hour of staff time to process, assessment moves from roughly $40 toward the officer figure, and the breakeven ticket rises with it — potentially from a few hundred dollars toward tens of thousands, which excludes the borrowers the network exists to reach. Credit any answer that names the failure mode: the cost is invisible on the P&L and shows up as a quietly rising minimum loan size.
8. Why is a 96 percent repayment rate not, by itself, evidence that a local assessment model outperforms a bank?
Because it says nothing about the pool, the term, the loss given default, the correlation of losses, or the unpriced labour. A group meeting weekly costs about 260 person-hours a year against a $1,500 group book — 17.3 percent of principal at a $1.00 shadow wage, against a stated rate of 20 percent. A high repayment rate bought with unbilled time is not a cost advantage; it is a cost transfer.
Two that require the arithmetic to be done.
9. A network can assess a borrower for $1,200. Its net spread is 5.0 percent a year and its average loan runs 2.0 years. Below what ticket size can it not operate? Show your working, and say what two moves change the answer.
Margin on principal = 5.0% × 2.0 = 0.10. Breakeven ticket = $1,200 / 0.10 = $12,000. Two moves: cut the assessment cost (at $40 the floor falls to $400) or lengthen duration (at 4 years the margin doubles to 20 percent and the floor halves to $6,000). Credit widening the spread as a third, with the observation that it is the one move that changes who is willing to borrow.
10. A book of 4,000 loans has a 6 percent default rate and a 55 percent loss given default. Compute the expected loss, and the one-in-a-hundred-year loss at ρ = 0 and at ρ = 0.04. What does the comparison decide?
Expected loss = 6% × 55% = 3.3%. One loan's loss sd = 0.55 × √(0.06 × 0.94) = 13.06%. Portfolio sd = 13.06% × √(1/4,000 + ρ(1 − 1/4,000)). At ρ = 0: sd = 0.207%, and the 1-in-100 loss is 3.3% + 2.326 × 0.207% = 3.78%. At ρ = 0.04: sd = 2.620%, and the 1-in-100 loss is 9.39%. The tail nearly triples on one assumption that nobody measured, which is why ρ belongs on the front page of the credit paper with its evidence beside it. Credit any answer noting the normal approximation is optimistic, because a real loss distribution is right-skewed.
REFLECTION — eight questions, for one person and a pen
These are not for a room. Write them by hand if you can; the slowness is the point.
- Which idea are you currently carrying because you heard it three times rather than because you read the source? Name one, specifically, and decide whether you will go and read it.
- Where in your own life do you give something you describe as generosity, and receive something in return that you have never named? Name it. Is the arrangement worse for being named, or only less comfortable?
- Who has extended you credit — money, time, benefit of the doubt — on a local signal rather than an assessment? What signal were they reading?
- Think of the last time you refused somebody who wanted something from you. What did the refusal cost, and where did that cost sit?
- Where are you running redundancy against a risk that is actually correlated — several versions of the same protection against the same shock?
- What is the assumption in your own plans that everything else moves with, and have you ever written down the evidence for it?
- Recall an arrangement of yours that survived a bad year because it could stretch rather than break. What made it able to stretch?
- What would you be willing to put on risk beside somebody else, to be believed by them? Answer concretely, with an amount.
ESSAY PROMPTS — five
Each is arguable from more than one side. Each requires at least one source the chapter cites and at least one it does not.
1. The correction and the story. Karst, Jones and Hoeksema (2023) found no peer-reviewed evidence for the claim that mature trees preferentially feed their offspring — a claim now carried by best-selling books, documentaries and forest policy argument. Argue either that popular scientific narrative is a net benefit even when it runs ahead of the evidence, because it directs attention and funding toward real systems, or that it imposes a cost that falls on the next generation of researchers and on the policy built in the interval. Use Karst et al. and Simard (2021) directly, and one source on science communication or replication that the chapter does not cite.
2. Is the market frame the right one for a symbiosis? Kiers et al. (2011) describe the mycorrhizal relationship as a biological market with reciprocal reward. Argue whether importing market vocabulary into symbiosis is an advance in precision or a smuggled ideology that shapes what gets measured. Engage Kiers directly, and at least one published critique of biological market theory that the chapter does not cite.
3. Did peer-to-peer lending fail? Between 2020 and 2021 RateSetter, LendingClub and Zopa all ceased retail peer-to-peer intermediation. The chapter argues this was consolidation into the mycelial structure rather than away from it, because distributed origination survived and only the fee-earning pipe was abandoned. Make the opposite case: that the retail lender's disappearance is the substance of the failure and the rest is a rebrand. Use the FCA's PS19/14 and one account of a specific platform's economics that the chapter does not cite.
4. The unpriced labour of local assessment. The chapter computes joint-liability group meetings at roughly 260 person-hours a year against a $1,500 group book — of the same order as the interest charged. Argue either that this labour is genuinely productive and its invisibility is merely an accounting failure, or that it is a real cost borne disproportionately by poor women and that a cheaper automated assessment is the more humane instrument. Use Yunus (1999) and Banerjee, Duflo, Glennerster and Kinnan (2015), and one source on the time burden of microfinance participation that the chapter does not cite.
5. Who should hold the correlation risk? The chapter shows that at ρ = 0.05 a 10,000-loan book is barely safer than a 1,000-loan book, and that the provision fund is recapitalised out of lenders' returns when the correlated loss arrives — as RateSetter's was in May 2020. Argue where that risk should sit: with the retail lender who chose the exposure, with the intermediary who originated it, or with a public backstop. Use Vasicek (2002) or Ostrom (1990) as your analytical frame, and one source on deposit insurance, securitisation retention rules or credit guarantee schemes that the chapter does not cite.