Single-source PCB under flood risk, with a Scope 3 obligation attached
Calder Electronics assembles control boards in Guadalajara. One custom PCB is 22% of a $41M cost of goods sold ($9.0M annual spend) and comes from a single supplier whose plant sits on a flood plain outside Kaohsiung; there is no long-term agreement. In seven years the plant has been down twice, once for nine weeks. Calder's line stops when boards run out after six weeks of cover, costing about $310,000 a week in lost contribution. Options on the table: (1) qualify a second source in Penang - 14 months and $350,000 of qualification cost, then a 7% unit-price premium on the 35% of volume transferred, with the new supplier able to surge to 60% within six weeks of a declared disruption; (2) hold eight weeks of boards as a strategic buffer - about $1.38M of working capital at a 10% carrying rate, plus roughly 20% annual write-down risk on older board revisions; (3) extend the Kaohsiung contract to five years for a 5% price cut, with a supplier-funded business-continuity programme. The largest customer now requires Scope 3 Category 1 reporting and a 25% supply-chain carbon cut by 2030.
1. Compare the three options on expected cost and on risk exposure, and recommend one.
Read the worked answer
Disruption cost first. The observed history is two events in seven years, about 0.29 a year. A three-week event costs 3 x $310k = $930k of lost contribution plus roughly $250k of expediting and premium freight, so about $1.18M. Expected annual loss is therefore roughly $290k/yr (0.29 x $1.18M); at a conservative 15% annual probability it is still around $177k. The mean understates the problem - a nine-week event is a $2.8M hit against a $41M COGS business, and it is the balance sheet, not the mean, that absorbs it.
Option 1 - second source: 7% x $3.15M = $220k/yr recurring, plus $350k of qualification amortised over five years = $70k/yr. Total about $290k/yr. The surge right to 60% plus a 35% standing allocation covers about 95% of normal demand, which is what makes the option real rather than cosmetic.
Option 2 - buffer: 10% x $1.38M = $138k carrying, plus 20% write-down = $276k. Total about $414k/yr, and it covers only eight weeks - less than the outage already experienced - with no residual asset.
Option 3 - extend and take the cut: saves about $450k/yr, but keeps 100% concentration, and the 5% is most likely funded out of the supplier's margin and its ability to invest in flood mitigation. It also raises switching costs for five years.
Recommendation: option 1, with a reduced three-to-four-week buffer bridging the 14-month qualification gap. On expected value it is near a wash with the buffer, but it buys coverage beyond eight weeks, a price benchmark, and capacity that survives a repeat event.
2. How should the dual-sourcing arrangement be structured and contracted?
Read the worked answer
Splitting 35/65 by unit volume across every part number is the lazy version and it satisfies nobody. Structure it deliberately:
- Split by part family, not by percentage of every part. Qualify the second source on the highest-volume, highest-risk boards where the $350k is amortisable; single-source the low runners and buffer those in stock. Not every SKU deserves two sources.
- Buy capability, not a certificate. Require first-article approval, a three-lot pilot, yield data, and periodic re-validation. A second source that cannot build the board to spec is a name on a slide.
- Contract terms that matter: a capacity reservation (35% standing) with a surge obligation (60% within six weeks of a declared force majeure at the other site), a documented price index, a right to audit the business-continuity plan, quarterly capacity attestation, and dual tooling ownership so the tooling is not stranded at one site.
- Keep the incumbent engaged. A 60/40 split with a rotating swing volume preserves the relationship and gives both suppliers a reason to perform. Giving 5% to a new supplier usually means giving it nothing.
- Governance. A quarterly dual-source review covering allocation, price premium, and readiness drills, plus one annual test where a small volume is actually built at the second site so the capability does not wither between disruptions.
Finally, be honest about what the split does not solve: a total Kaohsiung outage still leaves 40% of normal supply unmet until the surge comes through, so the bridge inventory and the surge clause have to be designed together.
3. How does the customer's Scope 3 Category 1 requirement interact with this sourcing decision?
Read the worked answer
Category 1 (purchased goods and services) is usually the bulk of an electronics assembler's footprint, and PCBs are energy-intensive to fabricate - so this is not a reporting formality, it is a sourcing criterion.
Three interactions with the decision just made:
- Dual sourcing can raise emissions. Two qualified sites, two tooling set-ups, two logistics lanes. Quantify the delta before claiming a carbon win; often it is small relative to the fabricator's electricity source.
- Disruption emissions dominate. A single air-freight expedite of boards during a Kaohsiung outage can exceed a year of routine inbound trucking. Reliability investment is also a decarbonisation strategy, which supports option 1 rather than undermines it.
- Data quality is the real deliverable. A spend-based estimate is not a number the customer will accept. Build a data-quality ladder: spend-based, then industry average, then supplier-specific (facility electricity and fuel), then product-specific. Require facility-level energy reporting and a stated allocation method in the Penang tender so the qualification decision is made once, not twice.
Practical plan: target supplier engagement where it matters (the top 20 suppliers will be roughly 80% of Category 1), write a renewable-electricity requirement and an annual reporting obligation into both PCB contracts, and set the 25% target against a measured baseline with the methodology disclosed. Then re-run the sourcing decision with carbon included as a scored criterion, not as an afterthought bolted onto the award letter.