Cynthia Carroll

The Professor's Questions

1. What should Cynthia Carroll do about the safety issues at Rustenburg?
2. What should Cynthia Carroll do to achieve her goal of "Zero Harm" for all of Anglo American employees, taking into account all relevant stakeholders?

My Answers

Q1. What should Carroll do about safety at Rustenburg?

Shut Rustenburg — all nine shafts — now, as her own decision, and use the stoppage to retrain every one of the 30,000 workers before anyone goes back underground. Not a speech, not a task force, not a new procedure. A stop.

Why a stop and not something softer:

• The baseline itself is the problem. 29 deaths by June 27, 16 of them at Rustenburg. Anglo averaged 46 deaths a year from 2002 to 2006. South Africa as a whole was losing a miner every 44 hours. Anglo is not worse than its peers — and that is the point. If "no worse than the industry" is acceptable, 46 deaths a year is acceptable.
• Rustenburg is getting worse, not better. Exhibit 8: Rustenburg's LTIFR went from 1.84 (2006) to 2.28 (2007). Group LTIFR went from 0.94 to 1.16. Potgietersrust, an open pit, sits at 0.26 with zero deaths. Depth explains the gap; it does not excuse it.
• Words have already been tried. The 2003 industry summit promised a 20% cut in fatalities every year; by 2007 it was 20–25% behind target. The board's own language was that procedures were in place "but the attitude was not where it needed to be." Front-line supervisors were "the most resistant to change." One more message from a four-month-old CEO is one more sentence in a language that only speaks downward.
• Only a decision that costs money is believed. Rustenburg makes about $8M of revenue a day. Fixed costs are large, so the loss "flows right to the bottom line." A week ≈ $56M; a month ≈ $240M. Rustenburg is 35% of Platinum's steady-state operating profit; Platinum earned $2,398M in 2006, so Rustenburg is roughly $840M a year, or about $2.3M of profit a day. A month of that is about 2.4% of the group's 2006 operating profit ($9,832M). It hurts. It is survivable. And it turns "unacceptable" from an adjective into a number that every tribal chief understands.
• If she doesn't stop it, the government can. Section 54 of the 1996 Mine Health and Safety Act lets an inspector halt a mine or part of a mine. A CEO who stops her own mine owns the story; a CEO who gets stopped looks like every other South African miner.
• The problem is low-tech, which means it is people. Moody-Stuart: "We had made a lot of progress using technology... Where we were still having problems was with low-tech incidents." Mud rush, conveyor belt, blasting barricade. You can't engineer your way out of these in a week — you have to change what 30,000 people and their supervisors do every shift. That takes everyone's attention at the same time. A shutdown buys exactly that.

How to do it, so the stop is a test and not just silence:

1. Decide it with Havenstein, announce it herself. Anglo Platinum is majority-owned and separately listed; Havenstein is the CEO. Going around him at month four is a governance cost. So: tell him first, insist on the outcome, let him execute. But the public message is hers — "I will not support operations that are killing people."
2. Pay the workers through the stoppage. 70% illiteracy, many migrants, many paid by shift. An unpaid shutdown is a strike you started. The wage answer has to exist before the announcement, not after.
3. Keep the mine alive while it is idle. The case is clear: an idle deep mine can get more dangerous because rock conditions change while no one is watching. So the stop is short, planned, with skeleton crews on ventilation, pumping and ground-support inspection. Restart is shaft by shaft, each one signed off against written criteria.
4. Retrain every person — supervisors first. Supervisors are the resistance and the bottleneck. Retraining is not a new rulebook read aloud in Fanagalo. It is: every worker has the right and the duty to stop unsafe work; every supervisor is accountable for using that information rather than punishing it. Train in the workers' own languages (Setswana for two-thirds of Rustenburg), visually, not in 2,000 words of commands.
5. Bring the union and the government to the table during the stop. NUM, Solidarity, the Department of Minerals and Energy. Not after the restart — during. The shutdown is the one moment when all three have a reason to sit down.
6. Publish the numbers. Fatalities and LTIFR, mine by mine, every month. Define in advance what "the shutdown worked" looks like: no fatalities in the first N months after restart, LTIFR back under 2006 levels, supervisor turnover, and whether the dialogue became two-way.

The honest costs: lost production in a record platinum market; shareholders in Johannesburg who want dividends and institutions in London who want to know why a first-year CEO is burning cash on a problem "everyone has"; Havenstein's authority; and the risk that supervisors simply wait it out. She should name all four in her own announcement. A leader who names the costs before her critics do is harder to argue with.

Q2. How does she get to "Zero Harm" across Anglo American, with all stakeholders in view?

Rustenburg is the shock. Zero Harm needs a system: one standard for the whole group, personal accountability at every level, a workforce that is allowed to say no, and three parties — company, unions, government — owning the same number. The stop buys attention for perhaps six months. What she builds in those six months decides whether the deaths come back.

Start with a fact about the organisation: Anglo is "a collection of mining businesses that operated independently." The business heads "don't" meet. "If they were asked to do something around safety... they would tell you to get lost." Zero Harm is impossible in a federation. So the first move is structural, not motivational.

1. The board — get the mandate in writing, then spend it. - The board told her safety "had improved but that I might want to attend to it." Mild. She should go back and make them say the hard version: Zero Harm is a group goal, the CEO owns it, and the board will accept the production cost of getting there. Peter Woicke's safety and sustainability committee becomes the place where this is reviewed first, every meeting, before the numbers. - Tie her own pay and every business-unit CEO's pay to safety outcomes — fatalities and LTIFR — not as a small modifier but as a gate.

2. The business heads — end the federation on safety. - One Anglo American safety standard, written at group level, mandatory in all 40 countries. Fatal-risk standards for the handful of things that actually kill people (fall of ground, transport, machinery, blasting, mud rush). Non-negotiable. Local rules can be stricter, never looser. - Make the "tribal chiefs" meet. A quarterly safety review where every business CEO presents his own numbers to his peers. Public comparison does what memos cannot. - Replace people who won't comply. The case says senior management was "on board"; the test is whether a chief who misses the standard and shrugs keeps his job. One visible exit does more than ten speeches.

3. Middle management and front-line supervisors — the real battlefield. - Ndlovu: "It was our front line supervisors that were the most resistant to change." Woicke: procedures were in place, attitude was not. So: supervisor KPIs change — a share of their evaluation is safety, including leading indicators (near-misses reported, stop-work calls honoured), not just whether anyone died this month. - Train supervisors to lead a two-way conversation, and give workers something to say back. The "visible felt leadership" she brought from Alcan — executives underground, asking questions — only works if the people above the supervisor model it first.

4. The workers — turn the language around. - Fanagalo has 2,000 words, "most of them commands," and operates "in the imperative. The dialogue is one way." You cannot build Zero Harm in a language that cannot carry a question. Shift training and safety communication into workers' own languages; retire Fanagalo as the safety language over time. - Give every worker the right to refuse unsafe work, protected from retaliation, and make supervisors report how often it was used. If the number is zero, the right doesn't exist. - Use what already works. The HIV/AIDS programme — free anti-retrovirals, 110,000 people tested voluntarily — proves that this workforce will trust the company when the company acts first and visibly. Safety can borrow that trust; it should also borrow the method: voluntary, peer-led, in the community.

5. Unions and government — from adversaries to co-owners of the number. - NUM and Solidarity: invite them into writing the standard, not just reviewing it. Solidarity's McMaster has already named the problem ("the human element was rarely if ever addressed"); he is a potential ally. - The South African government: the relationship is strained; the board will judge her on fixing it; Section 54 is their weapon. Turn it around — propose a tripartite safety agreement: company, unions, Department of Minerals and Energy, one target, one scoreboard, published together. The 2003 summit failed because it was companies promising to each other. Three parties watching the same number is harder to let slide. - The PIC — the government's own fund is a shareholder. A safety agenda the government co-owns is also a shareholder relations move.

6. Shareholders and analysts — make the business case explicit. - Three kinds of shareholders: old South African holders who want dividends; London institutions holding Anglo because it's in the FTSE 100; emerging-market investors. Citigroup says Anglo could cut $560M–$1.1B of costs. None of these groups will applaud lost production. - Her argument, said the same way every time: "business performance follows safety performance." Then prove it: every fatality is a stoppage, an investigation, a Section 54 risk, supervisor turnover in a hot labour market, and a licence-to-operate question in a country that already distrusts the industry. Report safety alongside the financials, every quarter. Shareholders accept a cost they can see and a plan they can track.

7. The industry and the communities — lead outward. - Carroll's own framing: "this was not just about Anglo American; it was about the responsibility of the mining industry as a whole to the communities in which its companies operated." Share the standards with competitors. Being first non-South African, first woman, four months in — she has no history to defend. That is what lets her say what the industry hasn't said about itself.

8. Herself — be the standard. - 20-day tours, 14 countries, "she gets energy from other people." Keep going underground. Make the first question on every site visit a safety question, in front of the general manager, every time. Culture is what the CEO asks about first. - Keep "Zero Harm" as the stated goal even though an executive told her "it will never happen in our lifetime." The moment she accepts a non-zero number, she has named how many deaths are fine.

Sequence and timeline. Months 0–2: Rustenburg stop, retraining, tripartite talks begin. Months 2–6: group standard written and signed by every business CEO; pay changed; right-to-refuse policy issued; monthly public reporting starts. Months 6–18: tripartite agreement signed; first full-year numbers; first consequence for a non-compliant leader. The test: fatalities under 46 (the 2002–2006 average) in year one, and falling every year after, with LTIFR moving the same way.

Where it can fail. Supervisors wait it out; production pressure in a record platinum market pulls everyone back; the government uses the opening to tighten rather than partner; Havenstein and other unit heads treat group standards as London interference. Her mitigation for all four is the same: make the number public and make the consequences personal.