Q1
What do we call the two variants — and is single-line / multi-line right?
✓ You askedRobinson on 27 August whether the two variants should simply be single-line and multi-line; Werner called those “the buzzwords”, and Mondello wanted the split framed as mono-line versus cross-sell rather than as regulatory labels. ✓ We built the plain gloss v5.0 now carries wherever the pair appears: clv_base is “the customer as they are today” and clv_with_growth is “plus expected future purchases”. The single-line / multi-line pair was worked through and declined in §3.2, with the reason stated.
Now pick the pair the published paper and toolkit carry.
Why single-line is the wrong word for the base variant. The base value is the customer as they stand — every product they already hold, renewed. On the reference book 250 households hold three or more products and 1,545 hold two; each of them has a base value that is already multi-line. Calling it single-line would tell a reader that a three-product household's base CLV counts one policy. What the with-growth variant adds is not more lines but future lines: the products the customer does not hold yet.
A
Keep clv_base / clv_with_growth, with the gloss — the customer as they are today versus plus expected future purchases, stated at first use and in the glossary. Ratified verbally on 21 August; this adds the words a first reader needs.recommended
B
Rename to single-line / multi-line — the Call #7 suggestion. Short and familiar, but wrong for the 250 three-plus-product and 1,545 two-product households whose base value already spans several lines.
C
current-holdings / with-growth — says exactly what the base variant contains, at the cost of a longer column name and a third rename in one summer.
Q2
Fixed expense: Kozlowski #112 versus Robinson #111
✓ You askedKozlowski in Word comment #112: “someone needs to pay for fixed expenses. Leaving it out would overstate values.” Robinson had said in #111 that fixed expense should stay out of a customer-level CLV. ✓ We built both views rather than choosing: the marginal view (variable expense only, the 16 August binding decision) and a named fully-allocated view through Economics.fixed_expense_per_term, with one worked example on the ten-year renewer of paper §3.4 (hand-worked in supplement Appendix F.4) and a book total under each.
Now pick the default view, and whether the second one is reported.
What the two views do. The §3.4 ten-year renewer is worth $782.14 net of acquisition on the marginal view and $278.88 fully allocated at an illustrative $75.00 per term — a difference of $503.26, which is $75.00 × the ten-year annuity factor 6.7101. On the in-force book (3,458 customers) the totals are $13,842,714.22 against $12,973,898.83, a gap of $868,815.39; 93.72% of the marginal value is retained. The charge is a constant per expected renewal, so it moves levels, not the loss-ratio relativities.
A
Marginal default, fully-allocated view beside it — marginal for rating decisions (fixed cost cancels out of every relativity), fully-allocated for valuing a book (a buyer inherits the overhead). Both reported, each exhibit labelled with its view.recommended
B
Fully-allocated default — Kozlowski's position taken whole. Every customer-level figure in the paper moves down by a constant per expected renewal, and the 16 August decision is reversed.
C
Marginal only — Robinson's position taken whole; the fully-allocated view stays in the framework doc and the planner but leaves the paper.
Q3
Which use case leads: rating decisions or valuing a book?
✓ You askedKozlowski on 27 August: “I'm more of an M&A guy … could this stuff be used to figure out the value of a book of business?” Werner added that if it serves both uses, say so regardless of the primary focus. ✓ We built the boxed What this paper is, and is not in §1.2, which names both uses, and a new §9.6 Valuing a book of business that gives the book-value reading of the existing results — the scenario planner's book-average mode already computes it.
Now pick how the paper frames its primary use.
A
Rating decisions primary, book valuation named as the second use with its own §9.6 — the RFP asked for a ratemaking method, and the filing-side exhibits serve it; the valuation reading uses the same number summed rather than compared.recommended
B
Equal billing — two use cases of one method, presented in parallel from the abstract onward. Truer to how the group talked on the 27th; further from the RFP's wording.
C
Rating only — keep §9.6 as an aside and the abstract on ratemaking alone. Cleanest scope; loses the reader Kozlowski described.
Q4
Basis names: ratify revenue-neutral / level-effect
✓ You askedRobinson on 21 August whether the filing/planning split in §3.5 was necessary at all, and Mondello said he could not follow it. ✓ We built the v3.3 rewrite: the section keeps the paper's main tenure caution, gains its motivation (McNulty 2013; CAS Price Optimization Working Party 2014) and its baseline sentence, and both bases are renamed to say what the arithmetic does — revenue-neutral and level-effect. Carried from the 27 August poll (Q1), where time ran out before a vote.
For the minutes.
At v3.3 the two bases read: revenue-neutral, 1-yr cohort +2.85% and 10+ yrs −7.43%, book total unchanged; level-effect, +17.28% and +4.84%, book +$2,615,287. Parameter and values unchanged — naming only. Four passages still read “provisional” until this is minuted.
A
Ratify revenue-neutral / level-effect and lift the provisional flags.recommended
B
Other names — write in the pair you would use; the rename is one sweep and the parameter does not change.
Q5
Reserve — does a 59-page paper plus a 51-page technical supplement serve you better than one 87-page document?
✓ You askedWerner on 27 August: “capture my attention so I hand it to them and say Do this”; Mondello: “first I want to understand it very simplistically, then the complicated coding part.” Call #7 asked for simpler, not shorter. ✓ We built v4.0 as one 87-page document that answered both inside a single binding, then v5.0, which executes the same decision in two: a paper of 59 pages — plain language, no formulas or software vocabulary in the body, 17 figures of which eight are new schematic diagrams, every section closing with a one-line Further detail pointer — and a technical supplement of 51 pages whose Part I carries the paper's own section numbers with the formulas, derivations, full tables and every provenance line, and whose Part II holds the notation, data dictionary, sensitivity grid, estimation details, reproduction guide, all 13 hand-worked exhibits and the 78-term glossary. Built from the same pinned numbers — not one changed; the drift gate reports 0 unmatched literals across the paper (822) and the supplement (1,094). An amendment adopted on your review, not drift.
Now pick — reserve; answer here if time allows, otherwise by email.
A
Yes, keep two documents — the paper is the one Geoff hands to a team; the supplement is the one the team opens. Same numbers, same section numbers, so nothing is lost between them.recommended
B
No, one document — return to a single binding at v4.0's length, with the reader's guide marking the two paths.
C
Two documents, but move X back into the paper — write in X (a formula, a table, a model-family subsection, a hand-worked exhibit); the section numbers match, so any piece moves in one step.