Irrational Exuberance Redux?

There are no free lunches. Every time there is exuberance in any bidding for public contracts, it is prudent to step back and reflect. 

The aggressive bids based on over-optimistic traffic growth and low fuel price forecasts associated with road and power sector PPPs of the last decade, the get-at-any-cost bids in telecom 3G spectrum auctions, and the race to the bottom with solar tariffs are some recent examples from India. In each case, the triumphalism following the contracting and resource mobilisation proved short-lived and the longer term costs were considerable. The legacy of the first was over-leveraged infrastructure companies and banks with non-performing assets, that of the second telecom companies with no option but to skimp on capacity investments, and the third one is currently playing itself out. 

Last week the National Highways Authority of India (NHAI) signed a 30 year concession agreement with Macquarie Infrastructure and Real Assets (MIRA), world's largest infrastructure fund manager, to monetise a bundle of nine fully operational highways involving 648 km in Andhra Pradesh and Gujarat. This is the first concession under the Toll-Operate-Transfer (TOT) model which was approved by the Cabinet in 2016 to monetise 75 national highway stretches of 4500 km. MIRA will make an upfront payment of Rs 96.815 billion in return for maintenance of the highways and the 30 year toll collection rights. 

MIRA's bid was striking in its exuberance. Against the NHAI concession value of Rs 62.58 bn (assuming an IRR of 12-13%), MIRA bid Rs 96.815 bn, a premium of 55%. It was also 30% higher than the next highest bid.

It is of course possible that this first bundle had all great highway assets with very high traffic growth potential and MIRA was only betting on its realisation. But that does raise the question of how the government, which technically has more information than the concessionaire, came out on the other side of the information asymmetry and left money on the table with its low concession value. Or was it done strategically to tantalise market interest in light of the big pipeline of such offerings?

How do we assess such deals? Is it a case of MIRA sensing money of the table that nobody else saw? Or one where MIRA bid sensed several attractive medium-term exit options?

The table below captures the spectrum of choices that governments and concessionaires have with such contracts.

Clearly, the overall welfare maximising option (option IV) is to have the concessionaire making a commercially viable enough bid and government being able to realise a reasonable price for the asset. Given the information symmetries and uncertainties of a long-term contract, this is difficult to achieve in the real world. In the circumstances, options II and III are practical realities, and the objective should be to get as close as possible to option IV. 

Based on the experience of such projects from across the world, let us get a few practical considerations out of the way. One, 30 years is a long period and will encompass multiple business cycles and their respective downturns. Traffic flows and interest costs will accordingly vary widely. The upsides have to more than compensate for the downsides. Two, if the downsides persist and in case of doubts on medium-term commercial viability, the concessionaire will invariably seek renegotiations, and most likely get it. Three, MIRA may be the current owner, but can and will exit (presumably there is a lock-in period) in due course and there will be multiple ownership transfers over the concession period. The MIRA Fund from which resources were drawn to pay for the monetised highways would typically have a ten year life, after which it pays out and closes.

In the circumstances, there are two concerns. The first concerns the discount factor and the second over likelihood of asset stripping.

Let us examine the first. The two biggest drivers of commercial viability are traffic projections and interest rates used to discount the cash flow. One can assume that MIRA has made its offer based on its best assumptions of the traffic projections (based also on data made available by NHAI), and to that extent is completely accountable for the associated risks. 

But the interest rate used in discounting is determined by NHAI and can be just as critical. As Ajay Jain of Indusbridge Capital Advisors has written in Bloomberg Quint, the nominal interest rates in India since the millennium have varied between 14.5% and 4.25%. Getting this right can be the difference between winner's curse (for the concessionaire) and leaving money on the table (for the government).

If the discount rates are too low, then this would imply a higher estimate of the Net Present Value (NPV) of the cumulative life-cycle cash flows. This, while giving governments high revenues would also force bids which tread on commercially questionable terrains. But, as experience from across the world and across sectors shows, it is only a matter of time before the concessionaire throw up their hands and seek renegotiations. And again experience shows, the net outcome is always bad for the tax payers. On the other hand if the discount rate used is on the higher side, the NPV estimate would be lower. This would leave money on the table and generate a windfall for the concessionaire.

In this context, the Value for Money (VfM) analysis of UK's Private Finance Initiative (PFI) projects by the country's National Audit Office is instructive,

Making changes to the discount rate applied to future costs can also affect which financing route is assessed as VfM. The VfM assessment compares private finance costs with a government discount rate of 3.5%, which is 6.09% with inflation, known as the Social Time Preference Rate (STPR), which is higher than government’s actual borrowing costs. The higher the rate applied, the lower the present value of future payments. For example a payment of £100 in 12 years will have a present value of just £49 when discounted by the STPR. Discounting using a lower discount rate, which compares private finance with the actual cost of government borrowing, results in fewer private finance deals being assessed as VfM. Using a fixed discount rate, set in 2003, means that the VfM assessment does not reflect the additional cost of private finance above the prevailing cost of government borrowing. In the current low-interest-rate environment it is possible to privately finance projects below the 6.09% rate. When this is the case private finance will be assessed as costing less than public finance even though the actual long-term cash costs of debt servicing and repayment will be higher than government debt costs.

As the graphic below shows, the government's actual cost of borrowing has consistently and significantly been lower than the discount factor used in making VfM assessments of PFI projects.

In summary, if the government has used a lower discount factor, it has inflated the cash flows estimates, and set the stage for renegotiations. In contrast, a higher discount factor deflates the cash flows, and leaves money on the table. Only time will tell which of these two scenarios best fit the first round of TOT roads monetisation.

This brings us to the second, and more important concern, around asset stripping. This can arise through multiple pathways. One way, as the example of UK's water utility privatisation shows, can be by leveraging up to finance opex and maintenance, making disproportionately large payouts to the investors from profits accrued, gradually diluting equity, and finally selling off to the next buyer and exiting with handsome returns. Another option is to cut down on expenses by skimping on operational expenses, socialising externalities wherever possible, and stretching the boundaries on the interpretation of the pre-defined maintenance schedules. 

Both approaches are straight out of the playbook of Macquarie itself during its ownership of Thames Water, UK's largest water utility with more than 9 million consumers. In the 2006-17 period of its ownership of Thames Water, "Macquarie paid itself and fellow investors £1.6bn in dividends, while Thames Water was loaded with £10.6bn of debt, ran up a £260m pension deficit and paid no UK corporation tax". Macquarie exited by progressively selling off its stake to an emergent group of private equity, pension funds, and sovereign wealth funds. 

Consider this scorecard from the monetisation of UK's water utilities,

The owners of the nine companies — many of which are overseas investors, including sovereign wealth funds — paid out £18.1bn in dividends in the 10 years to 2016, even though post-tax profits amounted to £18.8bn during the decade, according to the Greenwich University researchers’ analysis of their financial reports... Consequently, almost all capital expenditure has been financed by adding to the companies’ debt piles. Collectively these now stand at a towering £42bn... Three companies — Anglian Water Group, Severn Trent Water and Yorkshire Water Services — have paid out more in dividends than their total pre-tax profits over the past decade... Greenwich researchers said the cost of maintaining and improving water and sewer infrastructure has been paid for almost entirely by an increase in debt, which has risen from almost zero at the time of privatisation to nearly £40bn in 2016. The interest payments on the debt are higher than what would be paid by the public sector, which can borrow more cheaply. Together, the £1.8bn in dividends and the extra £500m of debt interest payments each year pushed up bills for each of England’s 23m households by about £100 a year.

Now that asset monetisation is here to stay, what are the lessons? Some suggestions

1. For a start, it is important to attract the right kinds of investors in these assets. The financial logic of investing in such long-gestation assets is the relative stability and not the size of returns. Pension funds, insurers, sovereign wealth funds and so on, who value reasonable returns which are stable and long-term, are the most appropriate direct or indirect (through the likes of  infrastructure funds) investors in such assets. Private equity and other similar investors, who elevate returns over stability and tenor, are likely to have a different set of incentives with potential to engender distortions.

It is possible to encourage self-selection among investors by prioritising stability and tenor. Clarity in the principles of renegotiations, a very high likelihood event in long-term concessions and one fraught with significant uncertainties, can be an important signal to lower risk and promote stability. I had blogged earlier with a few suggestions on renegotiations. On the upside, it may also be useful to cap certain types of windfalls, like in the UK, by forcing the concessionaire share profits with the government. In simple terms, investments in operational infrastructure facilities should become a boring financial asset!

2. On a related note, the vast majority of capital for such monetised assets would and should come from domestic institutional investors and not foreign infrastructure funds. I have already blogged here, here, and here about the limited depth of dry powder among global infrastructure funds to invest in infrastructure in countries like India. More importantly, these assets generate revenues in rupee and it is only appropriate that they be financed by domestic capital, so as to avoid currency liability mismatches and the resultant higher cost of capital.

It is therefore an opportunity to deepen and broaden the financial markets, especially the pension and insurance businesses as well as the bond market. Complementary reforms to incentivise these market participants to absorb the supply of these instruments may be necessary. The promotion of Infrastructure Investment Trusts (InvITs), announced in the Union Budget 2018-19, is a step in the right direction. However, the government may need to walk the talk with this so as to ensure that this does not become one more in the long-list of less than effective reforms initiated over the last two decades to broaden and deepen the capital markets.

3. More specifically, given the experience of infrastructure privatisation and asset monetisation from across the developed world, the concession agreements should seek to build safeguards that explicitly guard against asset stripping. It may be useful to have clearly defined financial ratios and floors/thresholds to limit excessive debt accumulation and equity dilution, as well as maintain oversight on pension and other statutory payouts. 

4. It is never a good thing to base a 30 year contracting decision on a single discount factor. Given the impossibility of reliably predicting long-term rates, this is pretty much a lottery for both the government and the concessionaire. The longest credible enough measure of long-term interest rates are the 10 year Treasury bonds.

One option would be to do periodic discounting resets, say once in 7 years, based on the expected long-term interest rates (not the prevailing bank rates). The payouts could then be made as instalment bullet payments, which however may come in the way of meeting the government's fiscal requirements and may also not align life-cycle incentives. The other option would be to have the initial payment which is based on discounting with the rate for the ten-year T-Bonds and then adjust the initial payment every time the reset happens. The adjustment could be in the form of a one-off term payment (by either the government or the concessionaire) based on how the discount factor varies.

Update 1 (19.12.2018)
Expectedly, after the first flush of irrational exuberance and as reality dawns, the demand for the second round of bidding for the Toll-Operate-Transfer (TOT) roads appears to be muted. This round proposes to concession out 586 km of national highways across Rajasthan, Gujarat, W Bengal, and Bihar, at a base bid price of Rs 5362 Cr for a period of 30 years.