The Leverage Machine: How Thames Water Turned a Monopoly into a Debt Vehicle
Download PDF
Executive Summary
Thames Water is the largest water and wastewater company in the United Kingdom and the most indebted. For more than two years its future has hung between a creditor-led restructuring and the government’s special administration regime, a form of temporary public ownership. The arguments have focused on who should bear the losses. This paper asks a prior question: how did the debt arise, and what does that imply for the choices now?
- The debt has grown much faster than the assets. Net debt rose from 3.4 billion pounds in 2006 to 10.8 billion in 2017 and 19.1 billion in March 2026. Gearing, the ratio of net debt to the regulatory capital value, rose from about 57 per cent to 83 per cent and then 87 per cent, far above the 60 per cent assumed by the regulator.
- The regulatory model rewarded leverage. From 2005 to 2020, the return allowed by Ofwat on the regulated asset base exceeded the real yield on ten-year index-linked gilts by between 3.6 and 5.5 percentage points. When the allowed return far exceeds the cost of debt, every pound of equity replaced by debt raises returns to shareholders.
- The debt is, to a large extent, the dividends. About 2.7 billion pounds of dividends were paid between 2007 and 2017. If they had been retained, and the debt they replaced had not been incurred, net debt in 2026 would be about 5.4 billion pounds lower at a 5 per cent interest rate, and gearing about 62 per cent, close to the regulatory norm. Compounded at the same rate, the dividends paid since 1990 exceed the entire current debt.
- Customers have paid more. The price of water supply and related dwelling services relative to all consumer prices rose by about 150 per cent between 1989 and 2026, including a real increase of 27 per cent since 2024 as the 2025 to 2030 price control raised bills to fund investment.
- The stakes of the restructuring are large for customers. The creditors’ March 2026 proposal included a write-off of 9.4 billion pounds, 3.35 billion of new equity, a new 6.55 billion debt facility and 20 billion of investment to 2030. At a 6 per cent interest rate, each billion pounds written off saves about 60 million pounds a year, or 3.75 pounds per person served.
The lesson extends beyond Thames. A regulatory model that sets returns on the whole asset base while allowing companies to choose their capital structure invites financial engineering whenever allowed returns exceed the cost of debt. Whatever happens to Thames Water, the regulator’s framework needs to remove that incentive.
1. A Failure Foretold
When the water industry of England and Wales was privatised in 1989, the companies were sold without debt, with large investment programmes and a regulator, Ofwat, charged with setting prices every five years. Thames Water was owned by its own shareholders, then by the German utility RWE from 2001, and from 2006 by a consortium led by the Australian bank Macquarie, which sold its last stake in 2017. Since then the company has been owned by a group of pension and sovereign funds, which have taken no dividends but have seen the value of their equity wiped out.
By 2024 Thames could no longer finance itself. In February 2025 the High Court approved an emergency loan of up to 3 billion pounds from senior creditors. In March 2026 those creditors made what they called a best and final offer, and in June 2026 the government objected to the deal, while Parliament’s environment committee recommended considering special administration. In late October 2026, with restructuring talks still continuing, S&P Global Ratings downgraded the company further.
Table 1. Thames Water in numbers
| Indicator | Value |
|---|---|
| People served | about 16 million |
| Net debt, 31 March 2026 | 19.1 billion pounds |
| Gearing (net debt to regulatory capital value) | 86 to 88 per cent |
| Ofwat notional gearing | about 60 per cent |
| Implied regulatory capital value | about 22 billion pounds |
| Dividends paid 2007 to 2017 | about 2.7 to 2.8 billion pounds |
| Dividends paid 1990 to 2022 | about 7 billion pounds |
| Shareholder dividends since 2017 | none |
Sources: House of Commons Environment, Food and Rural Affairs Committee (2026); company and press reports; author’s calculations.
2. The Leverage Machine

Figure 1 shows the central fact. Between 2006 and 2017 the regulated asset base roughly doubled, from about 6 to 13 billion pounds, as the company invested; net debt more than tripled, from 3.4 to 10.8 billion. Since 2017 debt has continued to grow faster than assets, reaching 19.1 billion against an implied asset value of about 22 billion. The company has, in effect, borrowed against almost all of the value of the assets its customers have paid for.
2.1 Why leverage paid
Under the regulatory model, Ofwat sets an allowed return on the regulatory capital value, the weighted average cost of capital, assuming a notional capital structure. Companies are free to choose their actual structure. If the actual cost of debt is below the allowed return, a company that replaces equity with debt keeps the difference. The incentive is strongest when interest rates are low.

Table 2. Allowed return and the real risk-free rate
| Price control | Allowed return (real) | Basis | Real 10-year gilt yield, average | Spread (pp) |
|---|---|---|---|---|
| PR99, 2000 to 2005 | 4.75% | RPI | 2.19% | 2.6 |
| PR04, 2005 to 2010 | 5.1% | RPI | 1.49% | 3.6 |
| PR09, 2010 to 2015 | 4.5% | RPI | -0.20% | 4.7 |
| PR14, 2015 to 2020 | 3.74% | RPI | -1.71% | 5.5 |
| PR19, 2020 to 2025 | about 2.96% | CPIH | -1.17% | 3.1 |
| PR24, 2025 to 2030 | about 4.0% | CPIH | 1.53% | 1.5 |
Note: allowed returns are approximate vanilla WACC figures; spreads use RPI-equivalent returns. The PR24 average risk-free rate covers 2025 and 2026 only.
Table 2 shows that the spread between the allowed return and the real risk-free rate widened from about 2.6 points at the turn of the century to 4.7 points in 2010 to 2015 and 5.5 points in 2015 to 2020, as real gilt yields turned negative while allowed returns fell only slowly. In those years a water company could borrow at a real cost of 1 per cent or less and earn 4 to 5 per cent on the assets financed. The Macquarie era of 2006 to 2017 coincided almost exactly with the widest spreads. As one former regulator put it, the generous settlement of 2004 was left uncorrected for a decade.
The spread closed sharply after 2022, when real yields rose by more than 4 points in two years. That reversal is what broke the model at Thames: debt refinanced at high rates could no longer be serviced from returns set when rates were low, and with gearing near 90 per cent there was no equity buffer to absorb the shock.
3. The Debt That Dividends Built
How much of today’s debt can be traced to the money taken out of the company? A simple counterfactual answers the question. Assume that each pound of dividends was in effect financed by an extra pound of debt, which then accumulated interest. Removing the dividends, and the interest on them, gives the debt the company would have had if it had retained its earnings.

Table 3. Counterfactual net debt in 2026 (GBP bn)
| Scenario | Interest at 3% | Interest at 5% | Interest at 6% |
|---|---|---|---|
| Actual net debt | 19.1 | 19.1 | 19.1 |
| Without 2007 to 2017 dividends | 15.0 | 13.7 | 12.9 |
| Without any dividends since 1990 | 5.5 | -2.8 | -8.9 |
| Memo: gearing without 2007 to 2017 dividends | 68% | 62% | 59% |
Note: gearing on an implied regulatory capital value of 22 billion pounds.
The results are stark. Without the dividends of the Macquarie era, Thames Water’s net debt in 2026 would be between 12.9 and 15 billion pounds, and its gearing between 59 and 68 per cent, close to the 60 per cent that the regulator considers prudent. At a 5 per cent interest rate, the value of all the dividends paid since 1990 exceeds the company’s entire current debt. The calculation is stylised: it ignores tax, assumes that every pound of dividends displaced a pound of debt and that investment would have been the same. But it shows that Thames Water’s insolvency is not mainly the result of the investment it made; it is the result of how that investment was financed while money was paid out.
4. What Customers Paid

Customers have paid for both the investment and the returns. The CPI component for water supply and related dwelling services, which also includes sewerage and refuse collection, roughly doubled relative to the overall price level between privatisation and 2010. It then stayed broadly flat in real terms for a decade, as Ofwat tightened allowed returns. In 2025 and 2026, the price control for 2025 to 2030 raised bills sharply to fund a large investment programme: the relative price index rose by 27 per cent between 2024 and 2026, to about two and a half times its 1989 level.
The combination is the core of the political problem. Customers are now paying higher real bills to finance investment that past bills were meant to support, while the company’s debt, built partly to pay dividends, absorbs a growing share of revenue in interest.
5. The Choice Now

Two broad routes remain. In the creditor-led restructuring proposed in March 2026, creditors would write off about 9.4 billion pounds of debt, inject 3.35 billion of new equity and provide a new debt facility of 6.55 billion, with 20 billion pounds of investment pledged to 2030; in exchange they sought regulatory concessions, including relief from fines. In special administration, a court-appointed administrator would run the company while its debts are restructured, with temporary public funding and the prospect of a larger write-off and a sale to new owners.

For customers, the size of the write-off matters most. At an interest rate of 6 per cent, each billion pounds of debt written off saves about 60 million pounds a year in interest, equivalent to about 3.75 pounds per person served. The 9.4 billion write-off in the creditors’ proposal would save about 560 million pounds a year, or about 35 pounds per person. A deeper write-off in special administration would save more, but would involve transition costs and temporary public funding, and would raise the cost of capital for other water companies whose creditors would reassess the risk of lending to the sector.
Table 4. The options compared
| Feature | Creditor restructuring | Special administration |
|---|---|---|
| Debt write-off | About 9.4 billion pounds proposed | Potentially larger, decided in administration |
| New money | 3.35 billion equity, 6.55 billion debt facility | Temporary public funding |
| Regulatory concessions | Sought, including relief from fines | None required |
| Control | Creditors become owners | Administrator, then sale |
| Risk for the sector | Lower immediate contagion | Higher perceived risk for lenders to water companies |
| Precedent | Owners and lenders recover more | Losses imposed on creditors |
Note: author’s summary of publicly reported proposals and of the statutory regime.
6. Beyond Thames: Fixing the Model
The Thames case reflects a design flaw in water regulation, not only the behaviour of one set of owners. Three reforms follow from the analysis.
- Link allowed returns to actual capital structure. When companies gear above the notional level, the benefit of cheaper debt should be shared with customers, or the allowed return reduced, so that leverage does not generate excess returns.
- Cap gearing and dividends. A binding ceiling on gearing, with dividends blocked above it, would prevent a repetition. Ofwat’s licence changes since 2022 move in this direction; the Thames history shows they came too late.
- Index allowed returns to market rates. The widest spreads of the 2010s arose because allowed returns adjusted slowly to the collapse in real yields. Mechanical indexation of the cost of debt and equity to market rates would remove the arbitrage in both directions.
7. Conclusions
Thames Water was not ruined by investment alone. It was turned into a debt vehicle by a regulatory model that, for fifteen years, allowed returns far above the cost of debt and left companies free to replace equity with borrowing. The dividends of the Macquarie era alone explain the difference between Thames’s current gearing and the regulatory norm. Customers have paid twice: through bills that rose far faster than prices in general, and now through the cost of resolving a debt they did not choose. Whatever form the restructuring takes, its central question is how much of that debt is written off, and therefore how much of the cost of past financial engineering falls on lenders rather than on customers.
Methodological Appendix
A1. Spread between allowed returns and the risk-free rate
Allowed returns are the vanilla real weighted average cost of capital set at each price review, as reported in regulatory and consultancy publications. PR19 and PR24 figures, set on a CPIH basis, are converted to an approximate RPI basis by subtracting 1 percentage point. The risk-free rate is the annual average of the Bank of England’s 10-year real zero-coupon gilt yield (RPI-linked) over each control period.
A2. Counterfactual debt
Dividends: 2.7 billion pounds over 2007 to 2017 (about 245 million a year) and about 4.1 billion over 1990 to 2006, the difference between the reported total of about 7 billion pounds for 1990 to 2022 and the Macquarie-era payments. Each annual dividend is compounded to 2026 at 3, 5 or 6 per cent and subtracted from net debt of 19.1 billion pounds. The regulatory capital value of 22 billion pounds is implied by 19.1 billion of net debt and 87 per cent gearing.
A3. Limitations
The analysis relies on publicly reported aggregates rather than the company’s regulatory accounts, and the dividend series is approximate. Allowed returns are approximate and not strictly comparable across controls. The counterfactual assumes that dividends were debt-financed one for one and that investment and tax would have been unchanged. The CPI component used for bills includes sewerage and refuse collection and covers all water companies, not only Thames.
References
House of Commons Environment, Food and Rural Affairs Committee (2026), Report on Thames Water.
Ofwat, Final determinations, PR99 to PR24.
Oxera (2018), Ofwat’s PR19 methodology: what has changed?
Office for National Statistics (2026), Consumer price inflation, September 2026.
Bank of England (2026), Yield curve data: real zero-coupon gilt yields.
Driss EFFINA, PhD
Independent Researcher, Global Strategy Files
The opening pages of our four country reports: United States, United Kingdom, Japan and China. A free PDF, sent to your inbox.
You will also receive occasional analysis from Global Strategy Files. Unsubscribe in one click, at any time.