A forward-looking reserve turns two resurfacing years that would break the covenant into twelve years that clear it, at a price equity rarely sees itemised.
A major maintenance reserve account is sized as a rolling share of forecast lifecycle costs: typically 100 per cent of what falls in the next twelve months, two thirds of the following twelve and one third of the twelve after that, topped up every year. On an illustrative road with $60.0m of resurfacing in years 4, 9 and 12, this turns a minimum DSCR of 0.40x into 1.20x. It is not free: equity gives up 6.09 of present value at 10 per cent for the smoothing.
Worked in full in The Infrastructure Investment Analyst by Julian R. Sterling, with every figure reproduced in a free workbook.See the book on Amazon →
The road is fictional and figures are in $m. CFADS before any maintenance and debt service are held flat so that only the reserve moves the coverage ratio.
| Input | Value |
|---|---|
| CFADS before major maintenance, each year | 40.0 |
| Debt service, each year | 25.0 |
| Resurfacing, year 4 | 24.0 |
| Partial works, year 9 | 6.0 |
| Resurfacing, year 12 | 30.0 |
| Look-forward period | 3 years |
| Equity discount rate | 10% |
Without maintenance the DSCR is 40.0 / 25.0 = 1.60x every year. Interest on the reserve balance is ignored to keep the arithmetic visible; a real model credits it and reduces the transfers.
Target balance at year end = Σ cost of each future event × weight
Weight = 100% if the event falls next year, 66.7% if in two years, 33.3% if in three, 0 beyond
End of year 1: 24.0 × 33.3% = 8.0 End of year 2: 24.0 × 66.7% = 16.0 End of year 3: 24.0 × 100% = 24.0
Excel: =SUMPRODUCT(Costs,(Event_year-t>=1)*(Event_year-t<=N)*(N-(Event_year-t)+1)/N)
Transfer = target this year − target last year + withdrawal this year
DSCR = (CFADS − transfer) / debt service, with the maintenance paid from the reserve
Year 1: transfer 8.0, DSCR = (40.0 − 8.0) / 25.0 = 1.28x
Year 4: 24.0 withdrawn, target falls to 0, transfer 0, DSCR 1.60x instead of (40.0 − 24.0) / 25.0 = 0.64x
| Year | Spend | Target balance | Transfer | DSCR with MMRA | DSCR without |
|---|---|---|---|---|---|
| 1 | 0.0 | 8.0 | 8.0 | 1.28 | 1.60 |
| 2 | 0.0 | 16.0 | 8.0 | 1.28 | 1.60 |
| 3 | 0.0 | 24.0 | 8.0 | 1.28 | 1.60 |
| 4 | 24.0 | 0.0 | 0.0 | 1.60 | 0.64 |
| 5 | 0.0 | 0.0 | 0.0 | 1.60 | 1.60 |
| 6 | 0.0 | 2.0 | 2.0 | 1.52 | 1.60 |
| 7 | 0.0 | 4.0 | 2.0 | 1.52 | 1.60 |
| 8 | 0.0 | 6.0 | 2.0 | 1.52 | 1.60 |
| 9 | 6.0 | 10.0 | 10.0 | 1.20 | 1.36 |
| 10 | 0.0 | 20.0 | 10.0 | 1.20 | 1.60 |
| 11 | 0.0 | 30.0 | 10.0 | 1.20 | 1.60 |
| 12 | 30.0 | 0.0 | 0.0 | 1.60 | 0.40 |
| Total | 60.0 | 60.0 | min 1.20 | min 0.40 |
Transfers add up to exactly the 60.0 spent: the reserve changes when the cash leaves, never how much. Without it, the year 12 resurfacing alone would take coverage to 0.40x, a default under almost any covenant. With it, the tightest years are 9 to 11, at 1.20x, when the largest event, the 30.0 resurfacing in year 12, is built up at 10.0 a year.
| Look-forward, years | Largest transfer | Minimum DSCR | PV cost to equity at 10% |
|---|---|---|---|
| 1 | 30.0 | 0.40 | 2.85 |
| 2 | 15.0 | 1.00 | 4.42 |
| 3 | 10.0 | 1.20 | 6.09 |
| 5 | 14.4 | 1.02 | 8.42 |
A one-year rule just moves the spike back a year: the $30.0m is reserved in year 11 and coverage falls to 0.40x then. A five-year rule should smooth more, but here the reserve starts empty at financial close, so year 1 has to catch up 14.4 for an event only three years away, and coverage drops to 1.02x. Longer look-forwards work only if the opening balance is funded from the construction budget. The right period is the one that clears the lender's minimum DSCR at the lowest present-value cost, which is a calculation, not a convention.
Lenders usually want the lifecycle forecast behind the target certified by a technical adviser and refreshed periodically. The reserve is only as good as the cost profile: a resurfacing that slips forward a year with the reserve two thirds funded lands on the distribution line.
Treating the reserve as costless because the money stays in the project. It does not stay in the equity's pocket. Every transfer is cash that would otherwise have been distributed years before the spend. On the illustrative road the three-year rule costs equity 6.09 in present value at 10 per cent, 21.4 per cent of the 28.50 present value of the maintenance itself. A bid model that deducts lifecycle costs in the year they fall, while the financing terms require a reserve, overstates the equity value by that amount. The same timing error appears in CFADS if the reserve transfers are left out of the line lenders actually test; the definition is worked in how to calculate CFADS in project finance, and the coverage measured over the whole tenor in how to calculate the LLCR.
The book's toll road model, with its resurfacing year and the debt sculpted around it, is in the free workbooks for this case.
A blocked account into which the project pays a share of upcoming heavy maintenance, such as resurfacing or turbine overhauls, so that the cost is spread over the years before it falls. In the illustrative road a $24.0m resurfacing in year 4 is funded at 8.0 a year over years 1 to 3 instead of hitting year 4 alone.
The cash is the project's own, so the cost is timing: it leaves the distribution line earlier than the spend. On the illustrative road, a three-year look-forward costs equity 6.09 in present value at 10 per cent, 21.4 per cent of the 28.50 present value of the maintenance itself. A two-year look-forward costs 4.42.
It rarely smooths anything. Funding 100 per cent of the cost in the year before it falls simply moves the spike back a year: the illustrative road's $30.0m resurfacing is reserved in year 11, and the DSCR that year falls to 0.40x, exactly the figure the reserve was meant to avoid.
The book's toll road case, the Calder Toll Road, carries a resurfacing in year four and has its debt sculpted in chapter 4 of The Infrastructure Investment Analyst. The book takes the same case from first principles to the decision, chapter by chapter, and every figure it prints is a live formula in the free companion workbooks.
Get the book on Amazon →Free companion files
Also on Amazon UK · Amazon Germany · Amazon France · Amazon Canada
Reading guide: infrastructure, data centres and energy → · All 1340 articles →
If this book helped, or didn’t, a few lines on Amazon are worth more than they look: they are what the next reader goes on. Write a review. The workbook stays free either way.