Haute Lumière

Commerce · V.09 · MMXXVI · daylight

La Bourse  /  Volume V  /  Nº V.09  /  Workbook — the executive

WORKBOOK — THE CORPORATE EXECUTIVE

Chapter V.09 · Health as a Return on Capital

For the person with a P&L, a benefits line, a vendor renewal on the calendar and a board that has heard the 3.27 : 1 number. This workbook is ninety days of work and it ends with a paper you can table.

---

THE PREMISE, STATED COMMERCIALLY

You are almost certainly buying at least one thing whose stated return cannot be supported, and you are almost certainly under-buying at least three things whose return can. This workbook is a reallocation, not a cut. In the worked case below the spend goes up and the defensible return goes up considerably more.

Three commercial facts frame it.

One. The randomised evidence on comprehensive wellness programmes is close to null on cost. Two large trials — 12,459 employees randomised in one, 160 worksites and 32,974 employees in the other — found no significant effect on medical spending, utilisation, absence or performance. The Illinois intervals exclude 78.0 per cent of the published savings estimates that preceded them.

Two. The published 3.27 : 1 asserts that a 150.00 USD programme removes 490.50 USD of medical cost per employee — 6.97 per cent of the 7,034 USD an employer pays for single coverage after the 1,401 USD worker contribution comes off an 8,435 USD premium. Say the claim in that form to a board once and the conversation changes permanently.

Three. You keep 14.44 per cent of the long-run health value you create, at a median tenure of 3.90 years. Your hurdle is therefore 6.92 x society's. That is not a reason to stop. It is the reason your portfolio should be weighted to short-payback health spending and to instruments that move the ownership.

---

PART ONE — DISCOVERY

Days 1–30

Exercise 1.1 — The three baselines (one week, with your controller)

Pull and date three files. Nothing in this workbook works without them.

Then compute two things immediately: your median employee tenure, and your voluntary turnover rate by site. They are the inputs to everything below.

Exercise 1.2 — The appreciative sweep (three conversations)

Ask three managers, separately and in these words: tell me about a time somebody stayed, or came back, or got better, because of something we did. Write what they say verbatim. You are looking for the programmes that are already working and have never been costed — which, on the evidence in this field, is where your real return already lives.

Exercise 1.3 — Open the vendor file (half a day)

Take every health or wellbeing vendor you currently pay and write, for each: the annual cost, the stated return, and the comparison group that return was computed against. Where the answer to the third column is "participants versus non-participants", write those four words in full. You are not cancelling anything yet. You are building the page that makes the reallocation obvious.

---

PART TWO — THE ARITHMETIC

Days 31–45

Exercise 2.1 — Decompose your own vendor's number (one week)

Ask the administrator for participant and non-participant mean claims in the year before the programme began. This is a routine request and it is the whole exercise.

Worked, on the chapter's case: after, 5,100 USD against 6,600 USD — a gap of 1,500 USD, which across 2,800 participants on a 750,000 USD programme is 5.60 : 1. Before, 4,900 USD against 6,100 USD — a gap of 1,200 USD. The causal estimate is 300 USD, the saving 840,000 USD, the return 1.12 : 1, and the naive figure is 5.00 x too large.

Write your own version of those two blocks on one page. That page is the most valuable document you will produce this quarter.

Exercise 2.2 — The regression-to-the-mean check (2 hours)

If any programme enrols on prior spend, compute what its saving would be with zero effect. On a mean of 6,000 USD, a standard deviation of 12,000 USD and a year-to-year correlation of 0.35, the top decile averages 27,060 USD and is expected to average 13,371 USD next year unaided: an apparent 13,689 USD a head, 50.6 per cent, from arithmetic.

Put that figure beside your vendor's reported saving before you read theirs.

Exercise 2.3 — Power every endpoint before you buy it (2 hours)

  n per arm = 2 x 7.8400 x SD^2 / effect^2          (means)
  n per arm = 7.8400 x 2 x p(1-p) / effect^2        (proportions)

Spending: SD 12,000 USD, effect 300 USD, needs 25,088 per arm and 50,176 in total — 10.04 x a firm of 5,000. Turnover: twenty per cent base, four-point effect, needs 1,568 per arm and 3,136 in total, which the same firm clears 1.59 x over.

Rule for your procurement policy, in one line: no health programme may be bought on an endpoint this business could not detect.

Exercise 2.4 — Your capture ratio, computed and adopted (1 hour)

λ = ln 2 / median tenure; capture = r / (r + λ). At 3.90 years and 0.03 the capture is 0.1444, or 14.44 per cent, and the hurdle multiple 6.92 x. At a tenure of 5.00 years it is 0.1779, a hurdle multiple of 5.62 x, and a corporate hurdle of 9.0 per cent becomes a social hurdle of 50.59 per cent.

Add one line to your capital paper template: duration of benefit, share retained by this entity, and who holds the rest. It costs a template change and it permanently stops the firm paying for other people's assets by accident.

Exercise 2.5 — Presenteeism, banded or struck (90 minutes)

If any business case in your building contains a presenteeism figure, require the multiplier to be printed beside it with a band. On a 60,000 USD salary and 5.0 per cent reported impairment, a ten per cent improvement is worth 150.00 USD at a multiplier of 0.5, 300.00 USD at 1.0 and 450.00 USD at 1.5 — returns of 1.00 : 1, 2.00 : 1 and 3.00 : 1 against a 150.00 USD programme. A 3.00 x spread on an undefended parameter is not a forecast. Band it or strike it.

---

PART THREE — DESIGN

Days 46–60: the reallocation and the instrument

Exercise 3.1 — Randomise the next rollout (one decision, one meeting)

Whatever is being deployed in waves in the next two quarters, assign the wave order at random and document the allocation before the first wave starts. Marginal cost: zero. What you buy for that zero is a randomised control group, which is the difference between the evidence in this chapter's Discovery and the evidence in its Arithmetic.

Exercise 3.2 — The reallocation table (one week)

Build it on evidence, not on enthusiasm.

Exercise 3.3 — The turnover case, which is where your money is (half a day)

  5,000 employees, voluntary turnover 20.0 %        =  1,000 separations a year
  replacement cost 21.4 % of a 60,000 USD salary    =     12,840 USD each
  annual cost of voluntary turnover                 = 12,840,000 USD
  intervention at 120.00 USD a head                 =    600,000 USD
  four points of turnover avoided                   =        200 separations
  saving                                            =  2,568,000 USD
  return                                            =       4.28 : 1
  net in year one                                   =  1,968,000 USD

Note what makes this case strong and say it in the paper: the endpoint is read from payroll, the effect size is inside the range a randomised trial has produced, and the whole return lands inside the 3.90-year lease. It is the one health investment whose payback is shorter than your capture problem.

Exercise 3.4 — Draft the pooled instrument (one week, with treasury)

The capture arithmetic says the wrong party holds the asset. The structure that moves it: a multi-employer portable health-capital pool, funded by a shared-savings note, paid against a randomised control.

• 14 employers, 41,000 covered lives, 210.00 USD a life a year: 8,610,000 USD invested annually.
• Capture rises from 14.44 per cent at 3.90 years of firm tenure to 34.18 per cent at 12.00 years of trade tenure — 19.74 points, a multiple of 2.37 x — and the hurdle falls from 6.92 x to 2.93 x.
• At a verified reduction of 320.00 USD a life, gross saving 13,120,000 USD: 1.52 : 1, a rate of 52.4 per cent, 43.4 points of headroom over a 9.0 per cent cost of capital.
• Waterfall: 75.0 per cent of verified saving — 9,840,000 USD — to the note until repaid; 3,280,000 USD retained by the pool from year one.
• Security: the verified saving stream alone. If it does not reduce claims against its randomised control, there is nothing to repay.

Exercise 3.5 — The audit and accounting conversation (one meeting, early)

You may not capitalise a workforce; IAS 38 and its equivalents forbid recognising an internally generated intangible of that kind. So the employer contribution is operating expense and the asset sits in the trust as a funded position against a future claim stream. Have this conversation before the paper, not after. It is a welfare-trust conversation, which your auditors have annually.

---

PART FOUR — DESTINY AND DELIGHT

Days 61–90

Exercise 4.1 — Into the standing pack (one conversation) Three lines, monthly: voluntary turnover by site, workers' compensation cost per 100 employees, and claims per member per month against the randomised control where one exists. Anything reviewed monthly persists.

Exercise 4.2 — Approach the stop-loss carrier (two meetings) Your stop-loss carrier already holds the tail of your claim distribution and is the one counterparty whose horizon is longer than 3.90 years. Ask for a shared-savings rider before you ask a bank for a note.

Exercise 4.3 — The one page, the one person (day 90) Baseline, intervention, randomised control, result, and what it implies at ten times the size. One page, to whoever controls the next allocation.

Exercise 4.4 — Delight, for a firm (ongoing) Stop requiring the small human things to carry a ratio. The shift somebody needed to move, the extra week for the man whose wife is ill. Fund them plainly because they are right and they cost little. A thing you do plainly is easier to keep than a thing you have to keep proving, and the proving was always the fragile part.

---

THE FAILURE MODES, NAMED

The baseline built afterwards. Then the 1,200 USD that was always there is counted as the 1,500 USD that was earned, and the number is 5.00 x too large.

The unpowered endpoint. At 5,000 employees and a spending endpoint you are 10.04 x short. Any result you report is noise, including a flattering one.

The three numbers added together. Claims, absence and presenteeism summed into one total, which is then defended as though the softest were as solid as the hardest.

The good programme killed by its own business case. A wellbeing programme oversold on cost, cut when the cost effect does not arrive. This is the most expensive failure in the field and it is caused entirely by the paper, never by the programme.

The pool with no control group. Every clause of the instrument above is ordinary except one: payment measured against randomly assigned sites. Lose that clause and you have rebuilt the literature you were escaping.

---

BOARD PAPER TEMPLATE

1. What we spend now and what it returns, with the comparison group for each stated return named in full.
2. The decomposition, in the chapter's two blocks: the after-gap, the before-gap, the naive return, the difference-in-differences return.
3. The powered endpoint list. What we can detect at our size, and what we are therefore declining to buy a claim about.
4. The reallocation, with the randomised evidence beside each line.
5. The turnover case, at 4.28 : 1 and 1,968,000 USD net in year one.
6. The capture position: we retain 14.44 per cent; here is who holds the rest and what we propose to do about it.
7. The instrument, at 1.52 : 1 with 43.4 points over a 9.0 per cent cost of capital, paid against a randomised control.
8. The one line we are asking for: randomise the order of the next rollout.

---

APPRECIATIVE QUESTIONS FOR YOUR LEADERSHIP TEAM

1. Where has this company already improved somebody's health or retention, and which of us can name the number?
2. Which of our current health spends would we keep even if it returned nothing, and what does that tell us about how to fund it?
3. If every health paper carried its endpoint, its power and its control group, what would we stop arguing about in this room?
4. What is being rolled out in waves next quarter, and what would it cost us to randomise the order?
5. Who outside this company holds the rest of the health value we create, and which of them would pay us for it?
6. If we were the best employer in our sector at this in three years, what would a visitor notice in the first hour?