Faced with Thames Water’s financial difficulties, and the potentially very negative consequences for investors with shares in the holding company controlling this business, many stakeholders seemed surprised by the size of the loss in value recognised at the end of 2022, while transactions carried out in recent years were based on high and even rising valuations.
However, careful consideration of trends in both Thames Water’s cost of capital and volatility, calculated using an appropriate risk model, would have made it possible to anticipate the operator’s financial difficulties.
Indeed, Thames Water’s volatility and value-at-risk have always been high in relation to various infraMetrics benchmarks, including its closest peers, namely other regulated utilities companies. However, the risks have doubled over the past 10 years, as shown in Table 1. Meanwhile, logically, this risk was reflected in the value of the cost of capital and therefore in the discount rate of future cash flows. As a result, its true fair market value has fallen sharply over the past 10 years.
Thus, over the period 2013-2017, under Macquarie’s ownership, the price-to-sales ratio of Thames Water dropped by 58 per cent and EV/EBITDA went down by over 25 per cent, while all the other benchmarks improved in value. In other words, the loss that was booked in December 2022 did not happen overnight.
Clearly, the failure to recognise this loss in value in investors’ financial statements at the right time raises the question of how to value this type of asset, and in particular the methodologies used to calculate the fair market value which, in the case of infrastructure, are based on discount rates that take no account of the reality of the risk factors to which the investments are exposed, nor of the market dynamics of the premiums associated with these risks.
Table 1: Risk & Valuation Metrics for Thames Water Utilities, UK Water, Global Utilities and the infra300® market index
| Date | Segment | Volatility* | Value-at-Risk** | Median Price/Sales | Median EV/EBITDA |
| 2013 | Thames Water | 18.9% | -33.1% | 3.73 | 10.30 |
| Global Regulated Utilities | 13.4% | -11.3% | 2.22 | 8.92 | |
| infra300 | 8.3% | -1.2% | 2.19 | 9.69 | |
| 2017 | Thames Water | 38.6% | -68.1% | 1.77 | 8.13 |
| Global Regulated Utilities | 13.8% | -11.1% | 2.44 | 12.33 | |
| infra300 | 8.8% | -3.0% | 2.62 | 11.83 | |
| 2020 | Thames Water | 38.0% | -63.0% | 1.94 | 8.45 |
| Global Regulated Utilities | 13.0% | -8.5% | 2.42 | 11.69 | |
| infra300 | 8.6% | -1.7% | 2.79 | 12.00 | |
| Q1 2023 | Thames Water | 37.9% | -64.5% | 1.54 | 7.69 |
| Global Regulated Utilities | 12.9% | -12.7% | 2.06 | 12.81 | |
| infra300 | 10.4% | -7.0% | 2.93 | 11.66 |
Source: infraMetrics.
*Standard deviation of monthly returns over a historical 10-year period.
**Gaussian value-at-risk at 97.5% confidence interval.
An anomaly or a poorly measured systematic risk?
Today, the Thames Water event is often presented as a special case. Indeed, a financial structure favouring debt, a dividend policy inconsistent with the economic performance of the business and furthermore constrained by the securitisation of the dividends, and the regulator’s decision to no longer take account of the project’s real cost of capital when setting tariffs, all played a particular role in the risk of this investment.
However, it would be regrettable to limit the risk of investing in unlisted infrastructure to purely idiosyncratic considerations. Thames Water is not an isolated case. Table 2 provides some significant examples of the materialisation of extreme financial risks in the infrastructure investment class. Lessons for the whole unlisted infrastructure asset class should be drawn from the Thames Water event.
Table 2: Impairments, Defaults and Related Volatility of Infrastructure Projects
| Company | Sector | Country | Event Detail | Value Before the Event (USD) | Price After the Event (USD) | Drop in Price | Volatility Before the Event | ||
| Company | Segment | infra300 | |||||||
| Bluewaters Power Station | Coal-fired power | Australia | Default in 2012 |
87,929,231 | 63,280,539 | -28% | 21.2% | 10.9% | 10.3% |
| Line 9 Metro Arganda | Urban mass transit | Spain | Impairment in 2020 |
51,568,739 | 33,186,324 | -36% | 21.8% | 13.1% | 8.9% |
| A-70 Circunvalacion de Alicante | Motorway | Spain | Default in 2012 |
25,195,207 | – | -100% | 21.6% | 9.9% | 6.6% |
| Nottingham Express Transit | Urban light rail | UK | Default in 2018 |
474,227,049 | 340,903,055 | -28% | 27.4% | 11.2% | 8.3% |
| Robin Hood Airport Doncaster Sheffield | Airport | UK | Impairment in 2013 |
182,580,630 | 53,902,406 | -70% | 21.5% | 13.8% | 10.3% |
Source: infraMetrics.
The segments used to qualify the business risk are those determined in the TICCS framework and representative of the sector and operational risk of the infrastructure investment in question.
Volatility is calculated over periods of 2 years prior to the occurrence of the event affecting the infrastructure investment under consideration.
In particular, managers and investors need to stop thinking that the volatility of this asset class can be measured with a global appraisal-based index calculated from very backward-looking and de facto smoothed valuation data. An annualised volatility of 3-4 per cent and a Sharpe Ratio of more than 3 tell you nothing about the risks in infrastructure investment. Furthermore, as can be seen in Table 1, not only is the volatility of unlisted infrastructure, as represented by the infra300 index, reasonably higher in general, but it can also vary significantly depending on the business segment (in this case, regulated utilities for Thames Water).
This link between observed volatility and default risk is not unique to Thames Water. Table 2 shows that, before they suffered a major impairment or even default, the investment projects under consideration experienced high volatility calculated on the basis of fair market value. This volatility, as with all other asset classes, was predictive of the risk of loss in value. As such, it is much higher than not only the volatility of global equity infrastructure represented by the infra300 index, but also that of the TICCS segment representing the sector and business risk to which each of the projects in question was exposed.
This reality of the risks involved in investing in unlisted infrastructure must be taken into account both when allocating to this asset class and when assessing a specific project.
Measuring risks for what they are represents the first and indispensable step towards good risk management. Moreover, this transparency lends credibility to an asset class with attractive financial characteristics, whether it be comparing its risk-adjusted performance with that of other private or public asset classes, or considering its diversification and, of course, liability management capabilities.
At a time when all stakeholders, including regulators, agree on the importance of increasing the proportion of private assets in pension fund allocations, transparency on the risks of these investments is essential.
Noël Amenc is Associate Professor at EDHEC Business School (Singapore).
