GroundedPriyatham Chimmani ↗
On this page
Decision tool - arithmetic only, nothing uploaded

Price a safety control twice, and compare the two paybacks

Enter what the incidents cost you in a year, what the control costs, and how much risk you assume it removes. The page returns payback, ROI, a sensitivity tornado and a 2,000-draw simulation for two accountings side by side: booked cost lines only, and booked plus the human-capital lines nobody invoices.

Everything runs in your browser. Nothing is uploaded and nothing is stored beyond your theme choice.

Open the calculator Read the story
What it shows, on the loaded example
3.41Years payback
Booked cost lines only 1.59Years payback
Human-capital lines priced

The same $150,000 loading-station control against a 3-year hurdle. It fails on the left count and clears on the right. Nothing about the equipment changed.

Worked example from Brandon (2025), loaded in the calculator below

Model4 lines
DiscountingNone
Risk reductionYour input
Empirical claimsNone
01

Walk through one incident

Three steps, one set of numbers. Pick a scenario, watch what the two accountings book, then see whether the control pays for itself under each.

Step 1

One incident

Start from the published example or an illustrative site, or edit the numbers yourself in the calculator below.

Enter your own →
Step 2

What gets booked, versus what it actually cost

Same incident, two ledgers. The traditional bar counts only the line items a ledger tracks; the human-capital bar adds the cost that does not get an invoice.

What gets booked
-
What it actually cost
-

-

Step 3

The investment

Same capital cost, same risk reduction, same horizon. Only the cost side changes.

Booked costs only

-

-
Human-capital costs included

-

-

-

02

The calculator

Enter annual incident cost in a typical year without the control. Left is input, right is arithmetic.

Start from a preset
About the presets
Only the first is published data: Brandon (2025), $55,000 / $118,000 annual cost, 147% / 315% five-year ROI. The other two are made-up inputs. Edit any field and the page recomputes.

Investment

Directly booked costs

both accountings
Subtotal$0

Human-capital costs

second accounting only
Subtotal$0

Side by side

-

Traditional accounting

-yr payback

-
Annual incident cost
-
Annual saving
-
Residual cost
-
5-yr ROI (cumulative)
-
5-yr net ROI
-

Human-capital-inclusive

-yr payback

-
Annual incident cost
-
Annual saving
-
Residual cost
-
5-yr ROI (cumulative)
-
5-yr net ROI
-

-

Source: your inputs above, evaluated with the four-line model in src/capitals.py. Not discounted.

Payback vs hurdle
Traditional
-
Human-capital-inclusive
-

Orange line is your 3-year hurdle. Shorter bar is better; a hatched bar never pays back.

Why the two verdicts differ
-
Risk reduction is an input, not an output The page does arithmetic on the number you type. Section 02 shows how much rests on it.
03

How much the answer moves

Three views of the same arithmetic. Each shows how far the verdict travels when one number changes.

shorter payback (better) longer payback (worse) Dashed line: current answer.
How to read the bars
Each bar spans the payback when that one input moves to each end of its range, everything else held. The dashed line is the current answer. One-at-a-time analysis understates joint risk.
04

Method, and what this does not do

Four lines of arithmetic. No discounting, degradation, maintenance or residual value: absent, not hidden.

The arithmetic and its limits
  • The arithmetic. annual saving = annual incident cost × risk reduction; payback = capex ÷ annual saving; ROI = (annual saving × horizon) ÷ capex. That is the whole model.
  • ROI is cumulative benefit over investment, not net return. 147% means benefits reached 1.47× capex. Net ROI is always exactly 100 points lower.
  • Nothing is discounted. A year-5 dollar counts like a year-1 dollar, which flatters long horizons.
  • Benefits are flat and permanent. No degradation, maintenance or residual value.
  • Human-capital lines are estimates, modelled, not invoiced. The second accounting inherits that uncertainty.
  • Counting more cost can never weaken the case. Every added line is non-negative, so the broader accounting always pays back at least as fast. That is structure, not evidence. Treat it as a reason for scepticism.