Written evidence submitted by T Martin
Blaiklock, Consultant, Infrastructure & Energy Project Finance |
The Treasury Committee has requested comments with
respect to issues relating to the use of PFI. This paper represents
my response to that Inquiry.
I have been a consultant, banker and practitioner
in infrastructure and energy project finance,PFI, PPP and
the like,for the last 35 years or more, with both UK and
wide overseas experience. I also regularly give seminars on these
topics internationally, eg World Bank/IFC, EBRD, etc.
1. What are the strengths and weaknesses of
different public procurement methods?
PFI V. CONVENTIONAL GOVERNMENT PROCUREMENT
[OF CAPITAL ASSETS FOR PUBLIC SERVICE DELIVERY]
|1. Embraces private capital (equity) resources for public investment
2. Can be structured as "off balance sheet" for Government (see below)
3. Facilitates technological and operational innovation by PFI concessionaire
4. Ensures cost control of investment (public sector projects historically have cost over-runs).
5. Locks-in asset maintenance regime over project life-cycle
|1. Complex structure & documents for PFI|
2. PFIs take twice as long to arrange
3. High up-front costs for PFI: twice as much as conventional procurement. Hence, PFI only suitable for investment projects exceeding, eg £40 million.
(NB some up-front costs can, on occasion, be capitalised into the funds raised)
4. Reduces competition: high costs and complexity means only major companies can afford to bid for such concessions
5. Not suitable for PFI concessions when the underlying output specification may change over time, eg IT
6. Demands significant senior staff attention
7. Difficult to resolve in the event of default or failure
2. If PFI debt had been on-balance sheet rather than off-balance
sheet, would PFI projects have been used so much? How should PFI
deals be accounted for?
underlying accounting rules as to what is "on" and "off"
balance sheet are unclear, but the problem is not UK-specific.
International initiative is required.
UK Government has employed a flawed methodology for determining
whether to support PFI-type deals or not. This has led to PFI
deals being undertaken, when conventional analysis would not have
2.1 There are two main types of PFI (or, now
generically termed, "PPP"s):
(a) the delivery of a public service by a private
(PFI) concessionaire, when the Government or its agents purchases
the service, often in the form of an "availability"
payment, eg an NHS hospital or a school; and
(b) the delivery of a public service by a private
concessionaire, when the user or consumer pays for the service
delivered, eg a cash toll-road.
In both cases, lenders will typically contribute
70-90% of the funding for such deals. Further, lenders, under
conventional project financing techniques, will take a lien over
the project assets, for as long, at least, as their loan is outstanding.
Hence, if the concessionaire fails to perform, the
lenders technically can take over ownership of the assets. Alternatively,
they will demand repayment of their loans, if the Government wishes
to take the assets back into public ownership. For international
PFI/PPP-type deals, lenders will insist on international arbitration
provisions to bolster their security.
2.2 Overall, therefore, PFI deals can create:
liability for Government under structure (a) above. In other words,
PFI is being used like a credit card for Government. It removes
the financial obligations for the purchase of an asset from Capital
to Current Account (
we all know how easy it is to overspend on one's credit card!!);
liability under both (a) and (b) above, if the underlying project
is an essential service, which Government cannot fail to deliver
to the public, funded either as a PFI or conventionally.
2.3 The question arises as to whether such contingent
liabilities should be recorded in National Accounts. Unfortunately,
the accounting rules are a mess! It is also ironic that the architects
of PFI/PPP have been the international accountants, who have made
fortunes from creating such obfuscation in the first place, failing
to adjust the accounting rules to meet modern corporate and financial
structures, and then advising proponents of such structures how
to avoid tax!!
An anecdote: in
1995-96 I applied for a post in the PFI Executive and subsequently
was interviewed by HM Treasury officials in Whitehall. At the
end of the interview I was asked if I had any questions for them.
I responded: "who is to count up the contingent liabilities
of all the PFI deals (being promoted at that time)"? Embarrassed
smiles all round. Silence! End of interview. No job offer!
2.4 In February 2004, Eurostat published a Decision,
which stated that:
If a PPP Concessionaire carries both the construction/completion
and demand or "availability" risk, then the transaction
can be considered "off" balance sheet for the host government.
If not, then the transaction will be deemed "on" balance
sheet for the host government.
2.5 In the above, it is clear that the determinant
for this on/off decision is the interpretation of what "represents
"risk" in any situation. Unfortunately, risk,uncertainty,is
a subjective, not objective, value!!
2.6 Secondly, the determination depends on a
judgement,not an objective measure,as to how much
risk is transferred from the public to the private concessionaire.
As a result, the basis for this Decision is flawed, notwithstanding
that many non-EU, as well as EU, countries now use this definition
for PFI!!! [It is a convenient definition for governments!].
2.7 As to whether a transaction should be deemed
"on" or "off" balance sheet is in essence
a financial decision, so should be based on financial, not physical
or commercial, criteria. It comes as no surprise, therefore, that
many governments (including the UK!) by using the current Eurostat
definition have hidden many transactions "off" balance
sheet, whereas they surely should be "on"!!
The classic case has been Network Rail, which receives
80% public subsidy for its operations, yet whose £20 billionodd
of debt is considered "off-balance sheet" by HM Treasury
on the grounds that its debt is a contingent liability for Government,
which only becomes a direct liability, if Network Rail defaults.
2.8 Many commentators have recognised the problems,
but no agreed answer has emerged, not even to at least publicly
record the data. Just as banks have been asked to clean up their
balance sheets in recent times, so one might claim governments
should do so too.
2.9 However, if this is to include the contingent
liabilities arising from PFI/PPP, so it also should cover government
participation in para-statal organisations and their debt, eg
Manchester Airport; Electricité de France, Schipol Airport,
etc., which represent contingent liabilities for national/regional
2.10 Whatever is decided, it certainly would
be prudent for all Governments and their agencies, which they
control financially or corporately, to publish such contingent
liability data as a start.
2.11 Finally, it should be noted that Greece,
Spain, Portugal and Ireland have all in recent years been very
active in PFI/PPP. Germany have not!.
2.12 The Inquiry asks whether UK Government has
been influenced by balance sheet issues in promoting PFI/PPP.
Possibly. Only Whitehall can confirm this.
However, what has influenced the decision in the
past has been the flawed methodology used by HM Treasury to assess
whether to pursue conventional or PFI-type procurement for a particular
public service investment.
2.13 When undertaking investment analysis of
a range of project opportunities, the conventional method for
determining the value of one opportunity against another is to
calculate the Present Value of each option. As PFI opportunities
can often span many years, eg 20-30 years, the discount rate chosen
to bring future values back into today's terms is key.
2.14 Conventional investment theory dictates
that the discount rate to be used is the opportunity cost of capital,
eg the interest rate for 30 year government bonds, by way of example.
It is, of course, important to choose a discount rate which reflects
the overall period of the analyses.
2.15 A second issue is how to treat inflation.
Inflation exists in every jurisdiction come what may! If one is
undertaking investment analyses of options which include finance,
eg PPPs, then the cost of finance (eg interest rate) has to be
included. As market interest rates reflect actual, or nominal,
rates, the analysis should be undertaken in nominal terms too.
This reflects reality, as we all see it.
2.16 The UK Government, when it undertakes analysis
of investment options for public service asset projects, uses
data in the calculations quoted in "real" terms, ie
ignoring inflation and the cost of money. In this respect, the
UK is out of step with most, if not all, other countries.
2.17 Via the "Green Book" HM Treasury
determines the discount rate to be used to express future values
in today's terms. From 1991 to 2003 the discount rate was 6% "real",
which, with inflation through that period being, say, 2-3%, gives
an approximate equivalent "nominal" rate of 8-9%.
From 1991-2003 the cost of 30 year UK Government
bonds ("gilts") varied, but on average was 5.5-6%. This,
of course, is a "nominal" value.
Hence, one can see that, by using a 6% "real"
discount rate for such analyses, the UK Government was imposing
a higher rate of discount than conventional investment analysis
would suggest, ie 8-9% versus 5.5-6%; a differential of 3-3.5%.
2.18 The consequence of using a too high discount
rate was twofold:-
firstly, this choice favoured PFI options
against conventionally funded alternatives. In cash-flow terms
under a PFI the payments to be made by the purchaser (ie Government),
which include the cost of funding the underlying asset, are later
in the Concession period than for conventional funding, just like
for a credit card.
If one calculates the value
of a 3-3.5% differential in Present Value terms over 30 years,
this shows an advantage to the PFI option amounting to 35-37%
of the Present Value purely by using such artificially high discount
rate. The result was that a number of deals, eg West Middlesex
Hospital PFI, were undertaken as PFIs, when in cost terms it would
have been cheaper to fund them conventionally; and
secondly, if the underlying cash-flows
from the analysis are used for budgeting purposes at a later date,
they will underestimate the payments to be made to a PFI Concessionaire
as they exclude the impact of inflation. This may well be the
main reason why many NHS Health Trusts, who undertook PFI projects
some years ago, now find they are very expensive, ie the Trusts
Fortunately, inflation has been
relatively low in recent times, so this impact has been limited.
But when inflation is high, under-budgeting arising from the use
of a high discount rate could be very significant.
2.19 In 2003, Government changed the discount
rate from 6% to 3.5% "real". Given inflation was low
at 2-3%, this discount rate thereafter reflected nominal rates
prevailing at the time.
To counterbalance this abrupt change,
HM Treasury introduced the concept of "Optimism Bias"
to reflect, as they thought, the inherent under-estimation of
costs that Government departments had demonstrated over past decades.
A consultant report on a number of major projects undertaken previously
showed various grades of cost over-runs, and so the empirical
concept of Optimism Bias was introduced as a multiple to be applied
to departmental cost estimates (the Public Sector Comparator)
to cover this possibility. The minimum multiple was 1.24, ie a
24% increment for simple projects, whereas for complex projects
the multiple could be 1.5-1.6 reflecting 50-60% cost over-runs.
2.20 It is well known universally that Governments
often over-spend on projects. However, no other government has
formalised the over-runs into a "cross the board" regulation
as has the UK through the application of Optimism Bias. Other
governments prefer to evaluate these possibilities through conventional
sensitivity testing on "base case" cost estimates.
Arguably, the project data used
to develop the UK Optimism Bias multiples was an inconsistent
sample. Furthermore, if a multiple of 40% were to apply to any
project estimate, could not a strong project manager be employed
to control costs for 10% of the cost, plus paid another 10% as
a bonus for success, and the sponsor would still be better off
[Another flaw in the methodology
was identified in that HM Treasury applied Optimism Bias before
risk analysis, whereas some agencies, eg Network Rail, applied
Optimism Bias after risk analysis. Mathematically the answers
are different. When this anomaly was raised via the Regulator
with HM Treasury and Network Rail, they both claimed they were
right and that the answers were the same anyway!!]
2.21 The overall net effect of the introduction
of Optimism Bias was to largely compensate for the errors which
arose through the use of an erroneous discount rate for the years
Today, there is greater realism
in the UK Government's approach to this topic, but there is, on
occasion, a reluctance to use these quantitative tools. Nevertheless,
the UK Government remains out of step with other governments on
this methodology, which, while not perfect, is probably the best
that can be derived. 70-80 countries worldwide now are considering
PPPs, but evaluate the opportunities in "nominal" terms.
2.22 Obviously, other factors such as efficiency,
innovation, etc., have to be taken into account before arriving
at a final figure for "Value for Money".
[NB. when the UK's new Green Book was published in
2003 there was an error in a quoted discount rate. The error remained
for a further five years!]
3. How far can risk really be transferred
from the public to the private sector?
3.1 The private sector will be happy to assume
the risks they can manage and control, eg design, materials, construction,
operations. There will, however, always be some risks which Government
has to shoulder, eg land purchase for a PFI road project. Should
the private sector be asked to take risks outside their control,
the cost to Government is likely to be high.
4. Are there any particular kind of risk which
are particularly appropriate for transfer through PFI deals, or
particular projects suited for PFI?
4.1 See above comment in Sec 3.
5. What state guarantees are explicit in PFI
5.1 See Sec 2 above. Technically, state (financial)
guarantees play no part in PFI transactions. Investors and lenders
look to the cash-flows for repayment of their capital. However,
on occasion, when the creditworthiness or financial sustainability
of the Government counterpart in a PFI transaction is in doubt,
the PFI concessionaire may require explicit "guarantees"
from Central Government underpinning the commercial undertakings
of such PFI counterpart.
6. In what circumstances are PFI deals suitable
for delivery of services?
6.1 Certain types of project are more suitable
for PFI procurement than others. Much depends on the inherent
project risks. Typically, "accommodation"-type projects,
where the underlying demand and service output over 30 years does
not change, are suitable for PFI treatment, eg hospital accommodation;
schools; university accommodation; prisons; Government offices.
Interestingly, in the years 1995-2003, more than 50% of UK PFI's
were in this sector.
6.2 Other sectors can be characterised as follows:
||not suitable as PFIs; too many risks|
|Light rail||||possibly, as Availability payment PFIs.
|Motorways||||yes, as Availability payment PFIs
|Estuarial bridges & tunnels||
||suitable for PFI, either as toll projects or against Availability payment
|Urban roads||||not suitable for PFI
|Waste treatment plants||
||suitable as PFIs|
|Water treatment plants||
||suitable as PFIs|
|High-tech buildings [eg NPL]||
||not suitable for PFI|
|Airports & ports||
||potentially PFI, but high infrastructure costs
[NB. need access infrastructure]
|Hospitals including clinical services||
||not suitable for PFI.|