Dr Duarte da Silva on Anglo (pt 2): The squandered house

In Part Two of his Anglo American exposé, Germiston-based strategist Dr Duarte da Silva asks what a chief executive faithful to Ernest Oppenheimer’s rules would have built. His answer: cancel the US$10.5 billion of buybacks, never sell the gold, and buy out the minorities in AngloGold, Amplats, Kumba and Gold Fields at the 2008 and 2013 lows. By his arithmetic, that company would be worth about US$156 billion today, against the real US$52 billion. It would still own the Witwatersrand gold mines and the Bushveld platinum, and its head office would be in Johannesburg rather than bound for Vancouver. For South Africa, it is a costly road not taken..By Dr Duarte F da Silva*.What a competent chief executive, steeped in the Oppenheimer method, would have donePart One is a post-mortem. It says what four chief executives did with Ernest Oppenheimer's house and what each decision has since cost, and it stops where a post-mortem stops, at the body. But a ledger of what was lost is only half an answer, because the question a shareholder wants answered is what a competent executive would have done instead, with the same cash, in the same years, facing the same prices, and what that house would be worth now. That is a harder question, and it is the one the four boards never put to themselves. Part Two puts it. The answer, worked line by line at the closes of 22 October 2026, is a company worth about US$156 billion, third in the world, against the US$52 billion that exists. The gap is the opportunity. It was not lost to the cycle, or to the country, or to bad luck. It was squandered, in the plain sense of the word, by people who had the founder's rule on the wall behind them and spent twenty years doing the opposite of it.The method is narrow on purpose, so that it can be attacked. I give the chief executive no foresight and no genius. Every decision in the Part One ledger stands except two: the buybacks are cancelled and the gold shares are not sold. The dividend is held at half from 2006 to 2013. Minas-Rio is bought and built as it was, because a house that holds the cycle also makes mistakes at the top of it. De Beers is bought from the family as it was, because buying the diamonds when the family sells them is what the method says to do; what the house did with them afterwards is another matter, and Section XI takes it up. The money those two abstentions free is spent, at the market's own prices, on the minorities of the companies the house already controlled and on the one gold company it had let slip. A thirty per cent premium is charged on every purchase. No synergy is counted, no dividend received in the interval is counted, and two timings are priced: the method applied in the years the boards actually acted, at the top, and the method applied as a house applies it, at the bottoms of 2008–09 and 2013.THE RULEI. Four rules, and what they would have meant in the chairErnest Oppenheimer never wrote a strategy. He wrote one sentence, in 1954, about profits earned 'in such a way as to make a real and permanent contribution' to Southern Africa, and he left one rule about diamonds, that the only way to make a stone valuable is to make it scarce. The rest he built rather than wrote: gold, diamonds, platinum, coal, copper and industry under one roof, with a finance arm to move money between them, so that whichever metal was out of favour could be carried by whichever was not, and bought from those who could not carry it. Harry ran the house that way for twenty-five years, through Sharpeville, Soweto and the state of emergency, and bought when others sold. A chief executive who had absorbed that record would have carried four rules into the office. Never sell the asset that is out of favour, because that is the day the house exists for. Never buy your own paper, because a mining house that can find no ground to buy has stopped being a mining house. Consolidate what you already control before you buy what you do not understand. And where the house owns enough of a metal to matter, set the price, because a custodian who lets the buyer name the price of what he owns has given up the one thing De Beers was built to do.The first three rules can be priced, and the next nine sections price them. The fourth cannot, and Section XI says why it has to be stated anyway. What follows is written as if a competent chief executive, steeped in that method, had held the office from the strategic review of April 2006 to the present, with the same facts in front of him that Trahar, Carroll, Cutifani and Wanblad had, and had applied the rules and nothing else.Never sell what is out of favour. Never buy your own paper. Consolidate what you control before you buy what you do not understand. Set the price of what you own enough of. Four rules, applied to the same cash in the same years.THE CASH THEY HADII. Where the money went, 2004 to 2013Between 2004 and 2013 the boards of Trahar and Carroll had about US$31 billion to deploy over and above what the mines needed, and the table says where it went. The gold shares brought in US$6.4 billion. The buybacks took US$10.5 billion, all of it between February 2006 and the suspension of February 2009, at an average of US$49.50 a share that the shares would not see again for a decade. Dividends over the ten years came to about US$10.5 billion, with a special dividend in 2006 for good measure. Minas-Rio cost US$4.65 billion to buy and De Beers US$5.2 billion. Net debt, which had been US$3.3 billion at the end of 2006, was US$11 billion two years later, and that debt is the hinge of the whole story: it is what forced the sales at the bottom in 2009, 2015 and 2016, and it is the first thing the method removes..Sources and uses, actual and under the method. Market values from companiesmarketcap.com year-end series (AngloGold US$13.0bn end-2006, US$11.9bn end-2007, US$9.9bn end-2008; Gold Fields US$2.4bn end-2013; Sibanye about US$0.9bn), Anglo Platinum's 2008 report (237.1m shares; Anglo at 76.5% in 2007 and 79.6% in 2008) and JSE prices (about R1,150 at end-2007, about R470 at end-2008, at R6.85 and R9.30 to the dollar), Kumba at about R50bn at end-2008 with Anglo at 63 per cent. Dividends from Anglo American's declarations 2004–13. A 30 per cent premium on every purchase. Author's arithmetic.Two things in that table settle the rest. The first is that the method applied with the boards' own timing, buying the gold and platinum minorities in 2006 and 2007 when the boards were placing shares instead, costs about US$27 billion and carries net debt of about US$27 billion into 2010. It is the fair reading of 'what if they had consolidated instead of buying back', it would have survived, because Anglo paid US$11 billion of debt down to US$1.4 billion in two years on the cash flow of 2010–11, and it ends in the same company. But it is not what a house does. A house does not buy at the top at all. It cancels the buyback in 2006, pays half the dividend, holds its gold, and waits. In the last quarter of 2008, with AngloGold at a third of its 2006 price, Amplats at a third of its 2007 price and Kumba down by half, it buys all three sets of minorities for about US$13 billion with the premium, which is within a few hundred million dollars of what the boards had spent on their own shares eighteen months earlier at the top. The whole programme, with Gold Fields in 2013, costs about US$17 billion. The buybacks not made and the dividend held at half pay US$14.5 billion of it; the rest is debt, and it is carrying a mine..The same money, spent two ways. The boards deployed about US$31 billion of discretionary cash between 2004 and 2013, US$10.5 billion of it on their own shares at the top. The method deploys about US$33 billion of the same money and ends up owning all of AngloGold, all of Amplats, all of Kumba, all of Gold Fields and Sibanye's mines.WHAT WOULD NOT HAVE BEEN SOLDIII. The parcels that stay, and what each is worth todayThe first rule asks nothing of a chief executive except that he sit on his hands, which makes it the easiest to apply and the most damning to have broken. Every parcel of gold, platinum and coal the four boards sold went because it was out of favour, and out of favour is the one condition under which the method forbids a sale. The table lists them with what each is worth at the 22 October 2026 close. Nothing in the right-hand column required the house to be clever. It required the house to do nothing..What abstention alone would have kept: the value on 22 October 2026 of each parcel the boards sold or gave away, at the closes of that day. The sum is the US$37 billion Part One calls 'what Anglo let go'. Sources as in Part One.Thirty-seven billion dollars is the price of a chief executive who could not sit still, and the figure flatters him. It leaves out the US$1.2 billion of dividends the AngloGold shares have paid their new owners, the Gold Fields dividends, the Sibanye shares that fell into Gold Fields holders' laps in 2013, Thungela's R22 billion, and the platinum dividends of 2021 and 2022, when Amplats paid out more in two years than in the previous ten. It leaves out the US$10.5 billion of buyback money that would still be in the house, and the interest on it. And it leaves out the thing the method was for, which is the next section: the money not spent on the company's own shares, spent instead on the ground the house already half owned.WHAT WOULD HAVE BEEN BOUGHTIV. Consolidating what the house already controlledIn 2006 Anglo American controlled, without owning outright, the largest gold company in the world, the largest platinum producer in the world and the largest iron-ore producer in Africa. It held 42 per cent of AngloGold Ashanti after the first placing, about three-quarters of Anglo Platinum and 63 per cent of Kumba Iron Ore. Each minority was a claim on earnings the house had found, sunk and managed, held by shareholders who had paid nothing for the finding, and each could be bought in the ordinary way, with a premium and a scheme of arrangement. The second and third rules point at those minorities before anything else: buy your own ground before you buy a stranger's, and buy it when it is cheap. The boards did the reverse. They bought the one thing that gives a house no more ground than it had, its own paper, at the top, and then bought a stranger's iron-ore project in Brazil from a promoter.AngloGold first. At the end of 2006 the whole company was worth US$13.0 billion and at the end of 2007 US$11.9 billion, so the 58 per cent the house did not own would have cost about US$9.4 billion with a thirty per cent premium in the years the boards were placing shares instead. In the last quarter of 2008, after the crash, the whole company was worth US$9.9 billion and the 58 per cent about US$7.5 billion with the premium. That is the quarter in which Anglo's board was preparing to sell its last 39.9 million shares to John Paulson at US$32. A house following the method stands on the other side of that trade: it is the buyer of the 58 per cent, from funds that have to sell, at the price Paulson paid. The scheme closes in the first half of 2009 with gold at nine hundred dollars. The house then owns the whole of a company that mined 4.6 million ounces that year, including every South African mine that AngloGold would otherwise sell to Harmony over the next eleven years, and it never sells them, because a house does not sell Mponeng for two hundred million dollars at the bottom of anything.Amplats second, and this is the purchase the four boards would find hardest to explain, because the house had been at three-quarters for a generation and the boards themselves took it from 76.5 to 79.6 per cent in 2008 without finishing the job. At the end of 2007 Amplats was worth about R270 billion, US$40 billion, and the quarter the house did not own would have cost about US$12 billion with the premium, more than the whole buyback, which is why the 'same years' version strains the balance sheet. At the end of 2008 Amplats was worth about R110 billion, US$12 billion, and the fifth the house still did not own would have cost about US$3.2 billion. Amplats had paid its shareholders R12.3 billion of dividends in 2007 and R13.8 billion in 2008, and the minorities took a fifth to a quarter of that, three billion rand a year, for owning a slice of mines the house ran. Buying them in the last quarter of 2008 for US$3.2 billion recovers that leakage within a decade on the 2007–08 run-rate alone, and it means that in 2025 there is no Valterra to demerge and no 19.9 per cent to place at R845 eleven weeks before the price went vertical. The house owns the whole of the largest platinum producer on earth through the 2026 half-year in which it earned R33 billion.Kumba third, and briefly, because Kumba is the one asset the boards got right by keeping it, and the method only finishes what they started. Anglo held 63 per cent at the end of 2008, when the whole company was worth about R50 billion, US$5.3 billion, and the 37 per cent minority would have cost about US$2.5 billion with the premium. Kumba has paid a dividend every year since; the minorities' share of them over seventeen years is several times that price. The house that buys all three sets of minorities in the last quarter of 2008 spends about US$13 billion, which is the buyback money, and comes out of the crisis owning outright the three best businesses it had, instead of 99 million of its own shares parked in three charitable trusts.Gold Fields last, and later, because the method buys when the market is giving things away and Gold Fields was not cheap until 2013. The house has kept its fifth from 2004, so it is the largest shareholder throughout. When Nick Holland unbundles Kloof, Driefontein and Beatrix into Sibanye in February 2013 at a first-day value of about US$900 million, and the gold crash of that spring takes Gold Fields itself from US$8.95 billion to US$2.4 billion by the year's end, the house is the obvious buyer of both. The 80 per cent of Gold Fields it does not own costs about US$2.5 billion with the premium and the whole of Sibanye about US$1.1 billion: US$3.6 billion for South Deep, Tarkwa, St Ives, Agnew, Damang, Cerro Corona, Kloof, Driefontein and Beatrix, paid from a dividend held at half since 2010. On 22 October 2026 Gold Fields alone was worth US$31.3 billion. The house now owns, under one roof, AngloGold, Gold Fields and the old GFSA mines: every large gold mine in the Witwatersrand basin, the Ghanaian and Australian mines of both companies, and Cripple Creek and the Boddington stake before AngloGold sold them. In 2013 that is about 7.5 million ounces a year, level with Barrick for the largest gold producer on earth, and the house has paid about US$11 billion for the part of it that it did not already own..The four consolidations under the method, at the prices the market offered at the bottoms of 2008–09 and 2013, and what the same holdings are worth on 22 October 2026. Sources: companiesmarketcap.com year-end values; Anglo Platinum annual report 2008; JSE price history; Gold Fields and Sibanye announcements, 2013. Author's arithmetic.Seventeen billion dollars, most of it the buyback money, bought at the bottom what is worth sixty-one billion today. The boards spent the same money on their own shares, and the shares are worth less than they paid.THE DEBTV. How the balance sheet carries itThe treasurer's objection is the balance sheet, so here is the path, year by year, with nothing hidden. The boards took net debt from US$3.3 billion at the end of 2006 to US$11 billion at the end of 2008 by spending US$10.5 billion on buybacks and US$4.65 billion on Minas-Rio against US$6.4 billion of gold proceeds. Under the method there are no buybacks and no gold proceeds, the dividend is halved, and the house pays its share of AngloGold's 2008 rights issue, about US$0.7 billion. It enters the last quarter of 2008 with net debt of about US$4 billion and the same Minas-Rio obligation. The three consolidations of that quarter cost about US$13 billion, and with the Paulson parcel never sold the house ends 2009 owing about US$18 billion against the actual US$11.3 billion. That is a real difference and I do not wave it away: about two and a half times the 2009 EBITDA of a company that had just suspended its dividend, and a rating agency would have had something to say about it.Three things carry it. The first is what the boards themselves did next: Anglo paid net debt down from US$11.3 billion at the end of 2009 to US$1.4 billion at the end of 2011 out of operating cash flow, because the recovery of 2010 and 2011 was the strongest in a generation. The same cash flow, with the whole of AngloGold's, Amplats' and Kumba's free cash inside the house rather than two-fifths, four-fifths and two-thirds of it, retires most of the extra seven billion by the end of 2011 and the rest during 2012, before the De Beers cheque. The second is the dividend, suspended in 2009 under the method as it was in fact, and held at half when it resumes, so the house keeps about US$2 billion more than the boards did over 2010–13. The third is that the assets bought pay. AngloGold paid roughly US$200 million a year in dividends in 2010–12; Kumba has paid every year, and its 2011–12 payouts alone were R27 billion; Amplats, thin in 2009–10, returned to large payouts later and paid its shareholders more than R80 billion in 2021 and 2022, a fifth of which, under the method, comes home rather than leaking to minorities. By the time Gold Fields comes up at the end of 2013 the house is at about US$14 billion of net debt against the actual US$10.7 billion, with three wholly owned subsidiaries the actual history never had, and Gold Fields' US$3.6 billion is paid from the dividend saving and a year of their cash. The 2015–16 downturn is then faced with a little more debt than the boards faced and a great deal more earnings, because the platinum and gold are the house's own; Rustenburg is not sold at the bottom, because the house does not need Sibanye's R1.5 billion. Woodsmith is still bought, because the method confers no foresight, and it is still written down.Seven billion more debt at the end of 2009, every dollar of it carrying a mine instead of a treasury share, and gone by 2012 on the cash flow the boards actually had.THE HOUSE AT EACH DATEVI. What the company looks like in 2009, 2013, 2016, 2021 and 2025Stand at the end of 2009 and look at it. The house owns all of AngloGold Ashanti, with Mponeng, TauTona, Savuka, Moab Khotsong, Great Noligwa, Kopanang, Tau Lekoa and the Vaal River surface business still inside it, together with Geita, Obuasi, Iduapriem, Siguiri, Sadiola, Morila, Sunrise Dam, Cerro Vanguardia, Cripple Creek and the Boddington stake; all of Anglo Platinum; all of Kumba; the Queensland and South African coal; Minas-Rio half built; 45 per cent of De Beers; and a fifth of Gold Fields. It has handed Mondi to its shareholders, which was right, and sold Tongaat and Hulamin, which was well timed. It owes about US$18 billion and it has never bought one of its own shares. Mark Cutifani, then running AngloGold, is the chief executive of a subsidiary rather than a candidate from outside.In 2012 the house buys the Oppenheimers' 40 per cent of De Beers, as it did, for US$5.2 billion, and carries it. In 2013 it buys the rest of Gold Fields and all of Sibanye, and its gold division is now AngloGold, Gold Fields, Kloof, Driefontein and Beatrix: about 7.5 million ounces a year, every large mine in the Witwatersrand basin, and the largest gold producer in the world alongside Barrick. In 2016, cut to junk with the rest of the sector, it sells the niobium, the phosphates and the Lafarge stake as it did, but not Rustenburg, because the platinum earnings are now wholly its own and R1.5 billion is not a price for mines. In 2018 AngloGold, as a subsidiary, does not sell Moab Khotsong to Harmony, and in 2020 it does not sell Mponeng; the subsidiary's chief executive, whoever he is, does not get to call the deepest mine on earth non-core, because the parent's rule is that it is the core. Thungela is demerged in 2021 on climate grounds, which the method does not forbid, but it is demerged at a valuation, not with a cheque attached, and the house keeps a third. In 2024 BHP arrives. The board that owns all of AngloGold, Gold Fields and Amplats does not receive a £38.6 billion offer; it receives one for a company worth something nearer double that, or it receives none, because BHP cannot afford it. There is no Valterra demerger in 2025 and no placing at R845. The platinum earnings of the 2026 half-year, R33 billion, are the house's..The house at each date under the method, against the actual at the start. Holdings from Anglo American annual reports 2006–2025 and the transaction announcements in Part One. Net debt from the results; the method's figures are the author's.WHAT IT IS WORTH TODAYVII. The valuation, line by line, on 22 October 2026The valuation adds to Anglo American as it is the value of everything the method keeps and everything it buys, each at the close of 22 October 2026, with nothing counted twice. Anglo as it is, US$52.0 billion, already contains Kumba, Minas-Rio, Quellaveco, the copper and De Beers at the price now being discussed, and it is the base. AngloGold, wholly owned, is the whole company at that day's value, US$43.1 billion; the 42 per cent Anglo once held is inside that figure and is not added again. The South African mines AngloGold never sold are valued at what they are worth inside Harmony today, three-fifths of Harmony's production profit applied to its US$10.2 billion, US$6.1 billion; AngloGold's US$43.1 billion does not contain them, so they are added. Gold Fields, wholly owned, is US$31.3 billion, and Sibanye's three gold mines are the Introduction Paper's sum-of-parts for that division, US$1.3 billion. Valterra, wholly owned and never demerged, is US$18.5 billion, and the 51 per cent demerged and the 19.9 per cent placed are inside it. Rustenburg, never sold, is a quarter of Sibanye, US$1.6 billion. Thungela is counted at its whole value of US$1.2 billion, which overstates the third kept and understates the dividends by more. The Kumba minorities are the 30 per cent Anglo still does not own, US$1.3 billion. Mondi is excluded, because it was delivered to shareholders and the method would have done the same. De Beers is neutral: held rather than sold, at about the value the sale implies..The house on 22 October 2026, built line by line: US$52 billion as it is, US$104 billion of what the method keeps and buys, about US$156 billion in all, with BHP, China Shenhua and Rio Tinto marked.The sum is about US$156 billion, and the kind of number it is matters as much as its size. It is nominal, at one day's prices, and it counts no dividend received between purchase and valuation on any line: not the US$1.2 billion the AngloGold shares have paid, not Thungela's R22 billion, not the fifth of R80 billion that Amplats paid in 2021 and 2022, not Kumba's seventeen years, not Gold Fields' US$2.5 billion. Count those and the house is comfortably past US$170 billion. It charges no synergy, although a company that owned AngloGold, Gold Fields and Sibanye's mines in one basin would have closed three head offices and four plants. It assumes the house ran each business no better than its actual owners did, which for the South African gold mines is a conservative assumption, since Harmony ran them better than AngloGold had and the house could have hired Harmony's people. And it assumes the gold price of 22 October 2026, which is where the bear will stand, and where the next section meets him..Third in the world, where it is fourteenth. The league table at the 22 October 2026 close with the method's Anglo placed on it at US$156 billion, and at US$130 billion if gold were US$3,000.On the league table of 22 October 2026 the house is third: behind BHP at US$197 billion and China Shenhua at US$160 billion, a whisker ahead of Southern Copper at US$155 billion and well clear of Rio Tinto at US$137 billion. Its gold division alone, US$82 billion, is second only to Newmont by value and level with Newmont in ounces, and it is the largest platinum producer on earth, and it still owns the diamonds. It is run from Johannesburg, because a house that owns the Witwatersrand, the Bushveld and the Northern Cape does not move its head office to the country of a copper company it has merged with; the Teck merger, if it happened at all, would have happened the other way round. The actual Anglo American is fourteenth, behind a royalty company with forty employees, with its gold in Denver, its platinum in other hands, its diamonds for sale at a twelfth of what it paid, and its head office on its way to Vancouver.About US$156 billion, third in the world, a gold division the size of Newmont's, run from Johannesburg. The actual company is US$52 billion, fourteenth, and leaving.SENSITIVITIESVIII. What moves the number, and by how muchThe premium does not move it. At twenty per cent the four consolidations cost about US$15.5 billion and at forty per cent about US$18 billion; the value today is the same either way, because the value is what the assets are worth now, not what was paid for them. The timing moves the cost and not the value: the 'same years' version pays US$27 billion for what the house version pays US$17 billion for, carries about ten billion more debt into 2010, and ends in the same place. The dividend moves the debt: a dividend paid in full rather than at half needs about US$4 billion more borrowing over 2006–13, which the cash flow of 2010–11 also retires, a year or so later. None of those changes the rank.The gold price moves it, and that is the right place for the bear to stand. About half of the US$156 billion is gold: AngloGold, Gold Fields, the ex-AngloGold mines inside Harmony and Sibanye's gold division, US$82 billion between them at a gold price of about US$4,040. If gold were US$3,000, which it last was in the spring of 2025, those four lines would be worth about thirty per cent less, with the leverage a gold equity carries to its metal, about US$57 billion, and the house about US$130 billion: fifth in the world, between Rio Tinto and Newmont, still two and a half times the actual Anglo. If gold were US$2,000, which it last was in 2023, the gold lines would be worth perhaps US$35 billion and the house about US$108 billion, seventh, ahead of Freeport, still double the actual. There is no gold price in the last decade at which the method's Anglo is worth less than the real one, because the real one sold its gold at prices from four hundred to nine hundred dollars and gold has not been back there since. Platinum moves it less, because Valterra is US$18.5 billion of the total, and the copper and iron ore are the same in both companies..Sensitivities. The gold lines are AngloGold Ashanti, Gold Fields, the ex-AngloGold mines inside Harmony and Sibanye's gold division, scaled with the gold price and the leverage gold equities carry to it; the rest of the house is Anglo as it is plus Valterra, Rustenburg, Thungela and the Kumba minorities, held constant. Ranks from the 22 October 2026 league table. Author's arithmetic.THE SHAREHOLDER'S ANSWERIX. What the shareholders got, both waysThe fairest objection to everything above is the shareholder's, and it deserves its own section rather than a line in the next one. Part Two has priced two companies by their market value on one day, and market value is not what a shareholder lives on. The real Anglo paid out. It paid US$30.2 billion in dividends and US$12.8 billion in buybacks since the London listing, and it handed its owners the shares of Mondi, Thungela and most of Valterra, which they could keep or sell. A holder who sat through the whole cycle did not end up with US$52 billion of market value and nothing else; he ended up with the market value and twenty-seven years of cheques. The method's Anglo, by contrast, held a dividend at half for eight years, bought nothing back, and kept Thungela and Valterra inside the house. So the honest comparison is total return, value plus everything paid out, on both sides, and the chart draws it..What the shareholders got, both ways. Everything the shareholder body has received since the London listing, as it happened and under the method: market value on 22 October 2026 plus every dividend, buyback and demerged share, nominal, against the US$27 billion the company was worth at its first London year-end.As it happened, the shareholders have about US$112 billion to show for twenty-seven years: US$52 billion of market value, US$30 billion of dividends, US$13 billion of buyback cash, and about US$16 billion of demerged shares and the dividends those shares have paid since. Under the method they would have about US$187 billion: US$156 billion of market value, US$26 billion of dividends, which is every dividend the real company paid less the US$4 billion held back in 2006–13, and Mondi, which the method demerges as the boards did. The method pays out less in cash, by about US$17 billion, and leaves the shareholder US$75 billion better off, because the US$17 billion it kept bought what is now worth US$61 billion and the dividends the real company paid were, in the years that mattered, paid out of the proceeds of the gold.Two things in that arithmetic lean against the method rather than for it. The first is the dividend. A house that owned the whole of AngloGold, Amplats, Kumba and Gold Fields from 2013 would have had their entire cash flow to distribute, and the Amplats payouts of 2021 and 2022 alone, more than R80 billion to all shareholders, would have been the house's to pass on; the chart gives the method's Anglo the real company's dividends and not a cent more. The second is the buyback, which is counted at face value as money returned, when a buyback at US$49.50 returns money only to the shareholder who sells and leaves the one who stays holding a larger share of a smaller company. The holder through the cycle, the one this section is about, saw none of the US$12.8 billion; it went to the people who left. Take the buyback out of his column and the real company gave him about US$99 billion, and the method about US$187 billion, which is where the fourteenth-against-third gap in market value reappears, almost undiminished, in the only number a shareholder keeps..Total return to the shareholder body, both ways, nominal. Sources: Anglo American results 1999–2026; closes of 22 October 2026; Thungela dividend history; Section VII for the method's market value. The method's dividends are the real company's less the US$4 billion held back; the dividends the method's wholly owned subsidiaries would have passed up are not counted. Author's arithmetic.Value plus every cheque: US$112 billion as it happened, US$187 billion under the method. The real company paid out more cash and left its owners US$75 billion poorer, because the cash it paid out was the gold.THE OBJECTIONSX. What the four boards would say, and the answerThey would say, first, that minorities cannot be bought at the bottom, because the bottom is when nobody has money. The house had the money, because it had not spent it on its own shares; that is the whole purpose of the first two rules. Anglo found US$10.5 billion for buybacks in 2006–08. The method finds the same US$10.5 billion eighteen months later and spends it on something else. The only thing the house needed in the last quarter of 2008 was a board that had not already spent the cash, and that is a decision, not a market condition.They would say, second, that a scheme of arrangement for AngloGold's minorities in early 2009, in the worst credit market in seventy years, would have failed or been blocked. Perhaps the first attempt. But the house was the largest shareholder, the price was a third of what the same shares had fetched two years before, and the sellers were funds that needed cash; and had it failed in 2009 it would have succeeded in 2013, when AngloGold was worth US$4.7 billion and the 58 per cent would have cost US$3.5 billion. The method is patient. The boards were not, which is why they were on the wrong side of the trade in both years.They would say, third, that a wholly owned AngloGold and Gold Fields would have been run worse than the independent companies that went to New York and Perth and quadrupled. Possibly. But the quadrupling was the gold price, which the house would have owned all of, and the South African mines the two companies sold or gave away are worth ten times inside Harmony and Sibanye what either company received for them, which is the clearest evidence in this ledger that the independent companies ran the South African ground worse than a house would have. The house would also have had, in 2013, the one answer to AngloGold's failed split that nobody at AngloGold could give: it would not have needed one.They would say, fourth, that political risk justified leaving South African gold and platinum. Part One has met this with the dates: the sales began before the nationalisation debate and continued for thirteen years after the ANC dropped it, and the buyers who took the other side of the political bet were South African companies that stayed and are now worth more than Anglo. The method's answer is shorter. Harry Oppenheimer stayed through Sharpeville and the state of emergency and bought. A house that leaves at the first talk of nationalisation is following the opposite of his rule.They would say, fifth, that this is hindsight, because nobody in 2008 knew gold would be four thousand dollars. Conceded, and the sensitivities are there for that reason: at two thousand dollars the method's Anglo is still double the real one, because the real one sold at four hundred to nine hundred. But the method asks for no view on the gold price. It asks only that a house which has just been handed gold, platinum and iron ore at a third of the price of eighteen months before, with the cash in hand, buys rather than sells. The four boards sold. They did not need foresight to do otherwise. They needed the founder's rule, which was on the wall.THE PRICEXI. The fourth rule: passiveness is not a strategyEverything above can be priced because it is about ownership, and ownership has a market value on a given day. The fourth rule is about price, and it cannot be priced, which is why it is stated here on its own, and emphatically, rather than folded into the sum. Ernest Oppenheimer's De Beers was not valuable for owning diamond mines; Botswana and Russia own diamond mines. It was valuable because from 1934 it sold most of the world's rough through one channel at prices it set, and held stock off the market when demand fell, so that the stone stayed scarce and the price held through the Depression, the war and the Soviet entry. That is the function the family performed for eighty years and the listed company gave up in two steps: in 2000, when De Beers abandoned the single channel for 'supplier of choice' and let the price find its own level, and after 2012, when the professionals who had just paid the family US$5.2 billion carried the diamonds at full value for eleven years while a laboratory stone took the engagement-ring market and nobody in the house lifted a finger for the price of the natural one. Keeping De Beers was consistent with the method. Letting the market set the price of a thing the house owned most of was the reverse of it, and it is the part of the De Beers story the ledger cannot score, because the counterfactual is a price that was never defended.I put no number on what a defended diamond price would have been worth, and nobody should. What can be said is that the company bought the custodian's chair for US$5.2 billion and then declined to sit in it, and that a chief executive steeped in the method would have understood that the chair was the asset and the mines were the furniture. The same applies, with more force and with the arithmetic already done, to platinum. Under the method the house owns the whole of Anglo Platinum from 2008, the largest producer of a metal of which South Africa holds 83 per cent of the world's reserves, in a country whose law, as The Richest Beggar on Earth set out in the first book, offers eight lawful instruments by which the owner of that ground can stop taking the price it is shown: a Johannesburg price venue, an export consortium for the minor metals under the block exemption, a sovereign reserve that holds metal off the market around cost, a variable export duty on concentrate, security-for-price collars, a regional compact with Zimbabwe, demand creation, and a golden share in refining. None of them asks the house to do anything Ernest did not do with diamonds in 1934. A house that owned all of Amplats and sat on the Bushveld with Impala and Northam as the only other producers that mattered was the natural convener of every one of them, and the platinum price that sat at US$1,000 for a decade while the house sold its mines into it was, in large part, the price of nobody convening anything.Passiveness is not a strategy, and I say it without hedging. The four boards treated the price of gold, platinum and diamonds as weather, to be forecast or endured, and the ledger shows what forecasting it cost them. The founder treated the price of the one metal he controlled as a thing the custodian sets, and his company was the most valuable mining house in the world for seventy years on the strength of it. The method's Anglo, with the whole of Amplats, the whole of AngloGold and Gold Fields and the diamonds, would have been the custodian of three of the four precious metals, and a competent chief executive would have used the chair. The value of that sits on top of the US$156 billion, not inside it, and it is the only line in Part Two whose size I cannot state and whose sign I have no doubt about.The company bought the custodian's chair for US$5.2 billion and declined to sit in it. Passiveness is not a strategy, and the owner of 83 per cent of a metal's reserves does not have to take the price it is shown.THE VERDICTXII. The opportunity, and who squandered itA competent chief executive steeped in the Oppenheimer method would have done four things the four boards did not, and nothing they did except the two mistakes I leave in to be fair. He would not have bought one share of the company's own paper, in any year, at any price, because a mining house that buys its own shares has announced that it can find no ground to buy, and the ground was all around it. He would not have sold a share of AngloGold or Gold Fields, or a tonne of platinum or coal, in a year in which it was out of favour, because that is the year the house exists to hold it. With the money those two abstentions freed, and a little debt, he would have bought at the bottoms of 2008–09 and 2013 the parts of AngloGold, Anglo Platinum, Kumba and Gold Fields that the house did not already own, because the first duty of a house is to own what it controls. Minas-Rio he would have bought and regretted, as they did. De Beers he would have bought from the family, as they did, and then done the one thing they did not: sat in the custodian's chair and defended the price, of the diamonds and of the platinum, with the instruments the law and the house's own history put in his hands.That company, on 22 October 2026, is worth about US$156 billion before a cent of the dividends it would have collected: third in the world, a gold division the size of Newmont's, the largest platinum producer on earth, the diamonds still in the house and the head office still in Johannesburg. The real company is worth US$52 billion, fourteenth, with its gold in Denver, its platinum placed thirteen months before the price went vertical, its diamonds for sale at a twelfth of what it paid, and its head office bound for Vancouver, and it paid its four chief executives about £150 million for the difference. The difference is a hundred billion dollars, and it is not luck and not the gold price, because the method wins at every gold price of the last decade. It is four rules, written down or lived by the man who built the house, that the people who inherited it did not read. That is what a squandered opportunity looks like when it is priced: not a mine that failed, or a bet that went wrong, but a house that was handed the best ground on three continents and the rule for holding it, and chose its own paper instead..NO BUYBACKS. NO GOLD SALES. THE DIVIDEND AT HALF. THE MINORITIES OF ANGLOGOLD, AMPLATS, KUMBA AND GOLD FIELDS BOUGHT AT THE BOTTOM FOR ABOUT US$17 BILLION, MOST OF IT THE BUYBACK MONEY. MINAS-RIO AND DE BEERS LEFT AS THEY WERE.THAT HOUSE IS WORTH ABOUT US$156 BILLION ON 22 OCTOBER 2026, THIRD IN THE WORLD, WITH A GOLD DIVISION THE SIZE OF NEWMONT'S, BEFORE A CENT OF THE DIVIDENDS IT WOULD HAVE COLLECTED. THE REAL COMPANY IS WORTH US$52 BILLION AND IS FOURTEENTH. COUNTING EVERY DIVIDEND, BUYBACK AND DEMERGED SHARE, THE SHAREHOLDERS GOT ABOUT US$112 BILLION FROM THE REAL COMPANY AND WOULD HAVE HAD ABOUT US$187 BILLION FROM THE OTHER. AT A GOLD PRICE OF US$2,000 THE HOUSE WOULD STILL BE DOUBLE THE REAL ONE. A HUNDRED BILLION DOLLARS WAS SQUANDERED BY PEOPLE WHO HAD THE FOUNDER'S RULES ON THE WALL, AND THE FOURTH RULE, THAT THE CUSTODIAN SETS THE PRICE, IS WORTH MORE THAN THE SUM AND CANNOT BE COUNTED..*Dr Duarte F da Silva • Capital Markets Strategist • Managing Director, Northbound Processing • Germiston, October 2026.SOURCESAnglo American plc. Final results 2005 (22 February 2006: total 2005 dividend 123 US cents including a special dividend of 33 cents, US$480 million; US$1 billion buyback; net debt US$4,993 million), results 2006 to 2013 (dividends per share 108c, 124c, 44c, nil, 65c, 74c, 85c, 85c; net debt US$3.3bn at end-2006, US$11.0bn at end-2008, US$11.3bn at end-2009, US$1.4bn at end-2011, US$8.5bn at end-2012, US$10.7bn at end-2013; EBITDA US$11.8bn in 2008 and US$7.0bn in 2009); Annual Report 2008 (63 per cent of Kumba Iron Ore; AngloGold Ashanti stake reduced to 11.8 per cent after year end); announcement of the sale of 39.9 million AngloGold Ashanti shares to Paulson & Co at US$32, 17 March 2009. London.Anglo Platinum Limited. Annual Report 2008 and corporate overview, June 2009 (Anglo American at 76.5 per cent in 2007 and 79.6 per cent in 2008; 229.6 million and 237.1 million shares in issue; dividends paid R12,276 million in 2007 and R13,816 million in 2008). Valterra Platinum. Results for the six months ended 30 June 2026; dividend declarations 2021–22. JSE price history for Anglo Platinum at the ends of 2007 and 2008 (about R1,150 and about R470).companiesmarketcap.com. Year-end market values of AngloGold Ashanti (2004–2015 and October 2026), Gold Fields (2003–2015 and October 2026), Anglo American, BHP and Rio Tinto (1999–2026), and the ranking of mining companies by market value at the 22 October 2026 close.Kumba Iron Ore. Annual Reports 2008–2012 (shares in issue; Anglo American's holding; dividends paid) and dividend history 2007–2025. Johannesburg.AngloGold Ashanti. Rights offer of July 2008 (US$1.7 billion); annual reports 2009–2013 (production 4.6 million ounces in 2009; dividends paid 2010–12). Gold Fields Limited. Unbundling of Sibanye Gold, 29 November 2012, and Sibanye's listing of 11 February 2013; results for 2013 and 2025. Sibanye-Stillwater. Results for the six months ended 30 June 2026. Harmony Gold Mining Company. Results for the year ended 30 June 2026 (production profit by operation; Mponeng free cash flow R9,444 million). Johannesburg.The Gold Papers, Book One, Paper 31 Part II: The Richest Beggar on Earth (the owner of 83 per cent of PGM reserves takes the price; the eight lawful instruments) and Paper 31 Part I: The Platinum Corollary. Northbound Processing, 2026.The Money Metal, Introduction Paper: The Four Who Stayed (draft 4, 23 October 2026): the leavers' ledger and the sum-of-parts for Sibanye's gold division. Paper 7: The House That Sold Its Gold. Paper 8: The Second House. Northbound Processing, Germiston, October 2026.Oppenheimer, N. 'See what is right, and be brave.' Farewell address, Vergelegen, 30 November 2011, quoting Sir Ernest Oppenheimer's chairman's statement of 1954. Epstein, E. J. 'Have You Ever Tried to Sell a Diamond?' The Atlantic, February 1982 (Ernest Oppenheimer on scarcity).Exchange data for 22 October 2026: Anglo American 3,625p (US$52.0bn); AngloGold Ashanti US$85.19 (US$43.1bn); Gold Fields US$35.14 (US$31.3bn); Valterra Platinum R1,165.16 (US$18.5bn); Harmony R269.50 (US$10.2bn); Sibanye-Stillwater R38.43 (US$6.7bn); Thungela R144.17; Kumba R218.27 (US$4.3bn); Newmont US$120bn; gold about US$4,040; GBP/USD 1.3385; USD/ZAR 16.49. Dr Duarte F da Silva, Northbound Processing, Germiston, 24 October 2026.

Dr Duarte da Silva on Anglo (pt 2): The squandered house

Related Stories

Duarte Da Silva: The destructIon of Anglo American

Dr Duarte da Silva: The richest beggar on earth

Dr Duarte da Silva: The Platinum corollary

Dr Duarte da Silva: Sibanye's Burnstone decision