The Hydropower Financial Model

First published 23 Aug 2026 · Last verified 29 Aug 2026

Lesson 43.1 — The Anatomy of a Hydropower Financial Model

A hydropower financial model is, at its core, a very long spreadsheet that tries to answer one question: will the water flowing down a Himalayan river generate enough cash, over enough years, to repay the people who financed the dam, turbine, and powerhouse — and still leave something for the equity owners? Everything else in this chapter is detail layered onto that single question.

Before going further, it helps to name the building blocks, because hydropower modelling in Nepal has its own vocabulary that differs in important ways from modelling a factory, a hotel, or a toll road.

A hydropower project has two distinct life phases, and the financial model must treat them completely differently. The first is the construction period — typically three to six years for a run-of-river project in Nepal, longer for a storage scheme — during which the company spends money on civil works, tunnels, penstocks, turbines, and transmission lines, but earns no revenue at all. The second is the operating period, which begins on the Commercial Operation Date (COD) — the date the project starts legally and physically delivering electricity under its Power Purchase Agreement — and typically runs 25 to 30 years, matching the tenor of the PPA and often the remaining life of the generation license.

During construction, all costs are accumulated on the balance sheet as Capital Work in Progress (CWIP) — an accounting bucket that holds every rupee spent on an asset that isn't finished yet and therefore isn't yet generating revenue or being depreciated. Think of CWIP as a construction ledger: cement, steel, turbine payments, contractor bills, consultant fees, and — critically — the interest paid on loans drawn to fund all of that, all sit in this bucket until the plant is commissioned. Once COD is reached, CWIP is "capitalised" — converted into a fixed asset on the balance sheet — and depreciation begins.

That interest cost deserves its own name because it behaves unusually: it is called Interest During Construction (IDC), and unlike interest paid on a working operating company, it is not expensed through the profit and loss statement while the project is being built. Instead, it is added to the cost of the asset itself. A farmer who borrows money to plant an orchard doesn't pay themselves back out of income that doesn't exist yet — the interest on that loan effectively becomes part of the cost of establishing the orchard, recovered later once the trees bear fruit and are sold. Hydropower IDC works the same way: the interest accrued on construction-period debt is capitalised into the project cost, increasing total CAPEX, and only starts hitting the income statement as depreciation and interest expense once the plant is operating and earning revenue.

Once operations begin, the model shifts its attention to four moving parts that repeat every year for the life of the PPA: revenue (driven by how much water flows and what NEA pays for the electricity it buys), operating costs (a relatively small, fairly predictable line for a run-of-river plant with almost no fuel cost), the fiscal regime (royalty payments to the Government of Nepal and income tax to the Inland Revenue Department, both of which change shape at fixed milestones in the project's life), and debt service (principal and interest paid to the lenders who financed construction). The output that lenders, equity investors, and regulators all watch most closely is the Debt Service Coverage Ratio (DSCR) — a single number, recalculated every year, that says how many times over the project's cash generation could cover that year's loan repayment.

A useful way to hold the whole model in your head is to picture a household budget stretched across three decades. In the early years, the household (the project company) is paying off a large home loan (construction debt) while its income (PPA revenue) is still building up. Government dues (royalty and, later, tax) start small and grow as the household matures and its temporary exemptions (the tax holiday) expire. The household's ability to comfortably make its loan payment every year, even in a bad year, is what a bank actually underwrites — not the average year, but the worst plausible year. That underwriting instinct — plan for the dry year, not the average year — turns out to be the organising principle behind nearly every convention described in this chapter, from how tariffs are structured to how hydrology is forecast to how DSCR covenants are set.

Lesson 43.2 — The PPA Tariff Structure: Dry Season, Wet Season, and Escalation

Nepal's rivers are governed by the monsoon. From roughly mid-June to mid-November, glacier melt and monsoon rain swell river flows several times over; from December through May, flows fall to a fraction of their wet-season level. Because most Nepali hydropower to date has been run-of-river (RoR) — meaning the plant has little or no reservoir and generates power more or less in proportion to whatever water happens to be flowing past the intake at that moment, rather than storing water to release on demand — a run-of-river plant's output is not steady through the year. It might run near full capacity for six months and at a fraction of capacity for the other six.

Nepal Electricity Authority (NEA), which under the current market structure is the sole legal buyer of wholesale electricity from private hydropower developers under a Power Purchase Agreement (PPA) — a long-term contract fixing the price and terms under which the developer sells all its output to NEA — has responded to this seasonal imbalance by paying two different prices for the same electricity depending on when it's delivered. This is the single most important feature of Nepali hydropower revenue modelling, and it is worth understanding through an everyday analogy before touching the numbers.

Think of a vegetable farmer near Kathmandu who grows tomatoes. During the monsoon, every farmer's tomatoes ripen at once, the market floods, and prices fall. During the dry winter months, few farmers can grow tomatoes at all, so the ones who can — perhaps because they have irrigation or a greenhouse — sell into a scarce market and command a much higher price for the identical vegetable. NEA's tariff structure recognises exactly this same seasonal scarcity in electricity: during the dry season (roughly mid-November to mid-May), when most run-of-river plants are producing well below their rated capacity, the electricity that is available is scarce and valuable to the grid, so NEA pays a dry-season tariff that is markedly higher per unit. During the wet season (roughly mid-June to mid-November), electricity is comparatively abundant because every run-of-river plant in the country is running near full output at once, so NEA pays a lower wet-season tariff for the same kilowatt-hour.

Illustratively, a typical recent PPA for a small-to-medium run-of-river project has set the dry-season tariff at roughly NPR 8.40 per kWh and the wet-season tariff at roughly NPR 4.80 per kWh — meaning the dry-season rate is close to 75% higher than the wet-season rate for physically identical electricity. This differential is deliberate policy: it is NEA and the Government of Nepal trying to incentivize developers toward storage and peaking projects (which can shift wet-season water to dry-season generation) and to compensate run-of-river developers fairly for the fact that their most valuable production window is also their leanest one.

On top of the base tariff, most current PPAs include an escalation clause — a pre-agreed annual increase in the tariff rate, applied for a limited number of years from COD, after which the tariff is held flat (in nominal terms) for the remainder of the PPA term. A common structure seen in recent NEA PPAs escalates both the dry- and wet-season tariffs by roughly 3% per year for the first eight years of commercial operation, after which the tariff freezes at whatever level it reached and stays there — in nominal rupee terms — for the remaining 22 or so years of the agreement. This escalation exists partly to compensate developers for inflation during the early operating years and partly as a negotiated feature of the standard PPA template; it is not indexed to actual inflation and does not continue indefinitely, which is a detail every model must get exactly right, because forgetting to flatten the escalation after year eight can overstate 20+ years of revenue by a very large margin.

Because tariffs are fixed in the PPA rather than freely negotiated each year, and because NEA is a government-owned, government-backed offtaker, the PPA also functions as the project's primary credit support — lenders are effectively underwriting NEA's payment obligation as much as they are underwriting the river's flow. This is why PPA terms such as the Required COD (a contractual deadline for reaching commercial operation, missing which can trigger liquidated damages or, in newer PPA templates, a reduction in the number of escalation years the developer receives) and the presence or absence of a Deemed COD clause (a provision that lets the developer start earning revenue on schedule even if delay is caused by NEA's own grid connection not being ready) matter enormously to bankability. Sponsors reviewing a PPA before financial close should treat these clauses with the same seriousness as the tariff numbers themselves, because a technically attractive tariff can be worth very little if a construction delay outside the developer's control simultaneously costs escalation years and produces near-zero revenue for months on end.

Lesson 43.3 — Hydrology Risk: P50, P90, and the Honest Revenue Forecast

Every hydropower revenue projection ultimately rests on one uncertain input: how much water will actually flow down this specific river, in this specific location, in each future year. Unlike a toll road (where traffic count is uncertain but knowable within a fairly narrow band once the road is open) or a factory (where output is a management choice), a run-of-river plant's energy output is dictated by nature, and nature does not deliver the same flow every year. Some years bring an unusually wet monsoon and abundant dry-season flow; other years bring drought.

Hydrologists address this by studying decades of historical flow records — where available, gauge data going back many years, sometimes supplemented by rainfall-runoff modelling where direct river gauge history is thin — and producing not a single "expected" energy number but a probability distribution of possible annual energy outputs. From that distribution, engineers extract specific reference points known by their exceedance probability: the probability that actual annual energy generation will equal or exceed a given figure.

The most commonly cited figures are P50, P90, and P99 (both P90 and P99 matter in Nepali practice, so both are explained here):

• P50 is the energy output that has a 50% probability of being equalled or exceeded in any given year — in plain terms, the "median" year, exceeded half the time and not reached half the time. It is the best single estimate of the long-run average and is the number typically used for equity investors' base-case return calculations.
• P90 is the energy output that has a 90% probability of being equalled or exceeded — a deliberately conservative figure, because it is only not achieved in roughly one year out of ten. Lenders overwhelmingly prefer P90 (sometimes an even more conservative P99) as the basis for sizing debt and testing DSCR, precisely because a bank cares far more about surviving a bad year without defaulting than about capturing the upside of a good one.

Return to the household analogy from Lesson 43.1: a family planning a home loan should size their monthly repayment against their income in a lean month, not their income in a bonus month, because the loan has to be paid every month regardless of how good that particular month turns out to be. A bank underwriting a hydropower loan does the same thing with river flow — it wants to know that the project can still make its debt payment even in a year that's drier than nine years out of ten, which is exactly what the P90 estimate represents.

The practical modelling implication is that a properly built Nepali hydropower model should carry at least two separate energy-generation scenarios side by side: a P50 case used to project the "expected" return to equity investors (their IRR, payback period, and dividend capacity), and a P90 (or more conservative) case used to size the debt facility and to test whether DSCR stays above the lender's minimum covenant even in a dry year. A model that only shows a single, optimistic energy figure — often quietly closer to P50 or even better — and applies it to both the equity case and the lender's covenant test is a red flag that either the sponsor has not done rigorous hydrology work or is presenting an intentionally flattering picture. Institutional investors reviewing a hydropower deal memorandum should always ask which exceedance probability underlies every energy figure quoted, because "annual generation of X GWh" is a meaningless claim without that qualifier attached.

It is also worth noting that reservoir and peaking projects — schemes with enough storage capacity to hold back water and release it on demand, rather than simply passing through whatever flow arrives — behave differently. Their Plant Load Factor (PLF), sometimes called capacity factor, which measures actual energy generated over a period as a percentage of the theoretical maximum if the plant ran at full rated capacity every hour of that period, is a design choice as much as a hydrological outcome: a peaking plant is deliberately built to run at a lower average PLF (often 30-45%) while concentrating its generation into the highest-value hours of the day and the highest-value season, whereas a pure run-of-river plant's PLF (often 45-60% depending on the river's flow variability) is simply whatever the river delivers, hour by hour, with limited ability to shift timing. Both figures matter, but they answer different questions: PLF tells you how intensively the installed capacity is being used; the P50/P90 energy figures tell you how much total annual energy that translates into, and with what confidence.

Lesson 43.4 — Construction-Period Mechanics: CWIP, IDC, and the Drawdown Schedule

Returning to the construction phase introduced in Lesson 43.1, it is worth walking through exactly how CWIP and IDC flow through a model, because getting the mechanics right materially changes the final project cost — and therefore the tariff and equity return needed to make the project viable.

A typical Nepali run-of-river project is financed with a debt-to-equity ratio in the region of 70:30 or 80:20 for the construction period, reflecting both the risk appetite of Nepali commercial banks (which have historically been the dominant lenders to domestic hydropower, often through syndicated loan consortia given the size of individual projects relative to any single bank's lending limits) and increasingly, for larger schemes, international development finance institutions and export credit agencies. Equity is typically drawn down first or pro-rata with debt, and debt is drawn down against certified construction progress, milestone by milestone, over the construction period.

Every rupee of debt drawn during construction starts accruing interest immediately, but the project isn't generating any revenue to pay that interest out of. Two things can happen to that accruing interest, and the choice matters enormously to the model: either the developer pays it in cash out of a separate reserve (uncommon for early-stage projects with no revenue), or — the standard treatment — the interest is capitalised, meaning it is added to the outstanding loan balance (and to CWIP) rather than paid, so it compounds and gets repaid later out of operating cash flow, alongside the principal.

Walk through a simplified illustration. Suppose a 25 MW run-of-river project has a base construction cost (civil works, electromechanical equipment, transmission line, land, and development costs) of NPR 5.4 billion, financed 70:30, so debt is NPR 3.78 billion drawn progressively across a four-year construction period. If the average outstanding balance during construction carries an interest rate of roughly 10.5% per year (fairly typical for a Nepali hydropower term loan denominated in NPR), and the debt is drawn evenly, the capitalised interest accrued over those four years might add somewhere in the region of NPR 500-650 million to the final project cost — pushing total CAPEX from the "base" NPR 5.4 billion figure to roughly NPR 6.0 billion once IDC is included. That NPR 600 million is not paper money — it becomes real debt principal that has to be repaid out of operating revenue for the next 15 years, and it is why disciplined construction scheduling (avoiding delays that stretch out the interest-accrual period) is one of the single highest-value levers a developer has over total project economics.

Once COD is reached, CWIP — including all that capitalised interest — is transferred onto the balance sheet as Property, Plant and Equipment, and depreciation begins. Nepali tax law generally allows hydropower generation assets to be depreciated for tax purposes using a written-down value (declining balance) method under a specified depreciation pool and rate set by the Income Tax Act, distinct from whatever depreciation policy the company uses for its own financial reporting (which may use straight-line depreciation over the asset's useful economic life, often estimated at 30-35 years or the PPA term, to better match the pattern of revenue generation for investors). This creates a familiar situation in project finance: tax depreciation (used to compute taxable income and hence actual cash tax paid) and book depreciation (used to compute reported accounting profit) diverge, and a careful model tracks both separately, because it is the tax depreciation schedule — combined with the tax holiday discussed in Lesson 43.6 — that determines actual cash tax outflows, which is what matters for DSCR.

Lesson 43.5 — The Debt Service Coverage Ratio: The Lender's Lens

If there is one number that a Nepali hydropower lender watches more closely than any other, it is the Debt Service Coverage Ratio, or DSCR. In its simplest form:

DSCR = Cash Flow Available for Debt Service (CFADS) ÷ Debt Service (Principal + Interest due that period)

Cash Flow Available for Debt Service (CFADS) is, roughly, operating revenue minus operating costs minus royalty payments minus cash taxes paid — in other words, the actual cash the project generates in a given period before anything is set aside to repay lenders. Debt Service is the total principal repayment plus interest due to lenders in that same period. A DSCR of 1.30x means the project generated 1.30 rupees of available cash for every 1 rupee it owed lenders that period — a comfortable cushion. A DSCR of exactly 1.00x means the project generated exactly enough cash to make its loan payment with nothing left over — an uncomfortably tight position that would alarm any lender, because it implies zero margin for a bad month, an unplanned repair, or a slightly drier-than-expected season.

Nepali commercial bank hydropower loan agreements typically embed a minimum DSCR covenant — a contractual promise, tested at each debt service date (often semi-annually) using the P90 energy case, that DSCR will not fall below a stated floor, commonly somewhere in the 1.20x to 1.30x range, sometimes distinguished between a "minimum" DSCR tested at each period and a slightly higher "average" DSCR tested across the life of the loan. Breaching the minimum covenant does not necessarily mean default outright, but it typically triggers a cash sweep or dividend lock-up — a restriction preventing the company from distributing dividends to equity shareholders until DSCR recovers above the threshold — and repeated or severe breaches can escalate to an event of default under the loan agreement, giving lenders the right to accelerate the loan or take other remedial action.

Lenders often go a step further and require DSCR-sculpted repayment — meaning the principal repayment schedule itself is deliberately shaped (rather than a simple straight-line or equal-instalment schedule) so that debt service is lower in years when cash flow is naturally tighter (early operating years, before escalation has fully worked through, or years when the tax holiday is expiring and cash tax first bites) and higher in years when cash flow is stronger. This sculpting is precisely why understanding the seasonal tariff structure, the escalation schedule, and the royalty and tax step-changes described elsewhere in this chapter matters so much: a debt schedule that ignores the timing of these cash flow shifts risks being unbankable even if the project's average, multi-year economics look perfectly healthy.

The table below illustrates a simplified DSCR projection for a hypothetical 25 MW run-of-river project — "Himal Khola Hydropower" — over its first eight years of commercial operation, the period during which its PPA tariff escalation is still active. The example assumes energy generation held at the P90 case (35 GWh dry-season, 83 GWh wet-season, 118 GWh total per year), a starting dry-season tariff of NPR 8.40/kWh and wet-season tariff of NPR 4.80/kWh escalating 3% annually, O&M costs of NPR 60 million in Year 1 escalating 6% annually, royalty at a representative NPR 150/kW capacity charge plus 1.85% of gross energy revenue, no cash income tax (the project is assumed to be within its income tax holiday period, discussed in Lesson 43.6), and a level annual debt service of NPR 480 million.

Two features of this table are worth pausing on. First, DSCR rises steadily over the eight-year window purely because the PPA's escalation clause lifts revenue by 3% a year while debt service is held flat and O&M rises more slowly than revenue — this is exactly the kind of pattern lenders like to see, since it means the tightest DSCR year (and the year most exposed to a hydrology shortfall) is Year 1, right at 1.28x, just above a typical 1.25x covenant floor. Second, notice how thin that Year 1 margin really is: if actual Year 1 generation came in even 5% below the P90 estimate used here, CFADS would fall by roughly NPR 31 million, pulling DSCR down toward 1.21x — below a 1.25x covenant in many loan agreements. This is precisely why lenders insist on the P90 (not P50) energy case for exactly this calculation, as discussed in Lesson 43.3, and why the earliest operating years of any Nepali hydropower project are the ones examined most anxiously by both lenders and equity investors alike.

Lesson 43.6 — Royalties, Taxes, Depreciation, and the Levelized View

The Government of Nepal collects two distinct charges from every operating hydropower project, both called "royalty" but structured very differently, and both step up sharply once a project passes its fifteenth year of operation.

The capacity royalty is a fixed annual charge per kilowatt of installed capacity — it is payable regardless of how much energy the plant actually generates that year, much like a fixed land-use or resource-access fee. The energy royalty is a percentage of the plant's gross energy revenue (or, in some formulations, of the value of energy generated) — a variable charge that rises and falls with how much electricity the plant actually produces and sells.

Under provisions that have applied in recent years (subject, as with all Nepali tax and royalty rates, to revision through the annual Finance Act and Electricity Regulation Commission decisions, so any sponsor or investor should always confirm current rates before modelling a live transaction), a broad picture looks like the table below. Rates differ depending on whether the project sells domestically or is licensed to export power, and whether it is a run-of-river or storage-type scheme, reflecting government policy to charge storage/peaking and export-oriented projects a higher royalty given their generally larger revenue potential.

Alongside royalty, hydropower companies benefit from one of the more generous tax concession regimes available to any sector in Nepal, reflecting the government's long-standing policy priority of encouraging private investment into electricity generation given Nepal's enormous underexploited hydropower potential relative to installed capacity. Under provisions applicable to projects reaching commercial operation within specified windows set by successive Finance Acts, hydropower generation companies have typically been entitled to a substantial income tax holiday — commonly structured as a full (100%) rebate on corporate income tax for an initial period (often ten years) from the start of commercial operation, followed by a partial rebate (commonly 50%) for a further period (often five more years), after which the company reverts to paying tax at the prevailing standard corporate rate on income (currently 20% for most companies, prior to any further sector-specific adjustment) for the remainder of the PPA term. On top of the income tax holiday, hydropower companies commonly benefit from reduced customs duty on imported plant, machinery, and equipment (since Nepal has essentially no domestic turbine or major electromechanical equipment manufacturing base) and exemption from value-added tax (VAT) on qualifying imported equipment not produced domestically, both of which reduce effective CAPEX rather than operating cash flow.

Because the tax holiday, the royalty step-up, and (for projects with PPA escalation) the tariff freeze all occur at different points in the project's life, a single year's numbers — even a well-chosen "representative" year — tell you very little about whether the project is a good investment across its full 25-30 year PPA term. This is why practitioners rely on levelized analysis: rather than looking at cash flow year by year, a levelized metric compresses the entire multi-decade stream of costs or revenues into a single equivalent constant figure, using a discount rate to make cash flows from different years comparable. The most common such measure in energy finance generally is the Levelized Cost of Energy (LCOE) — the constant per-kWh price that, if received every year for the life of the plant, would exactly cover all its capital and operating costs (discounted back to construction start) — which is useful for comparing the underlying economics of different technologies or different sites on an apples-to-apples basis, independent of any specific PPA's actual (non-constant, seasonally split, escalating-then-flat) tariff schedule.

For an equity investor specifically, the levelized concept extends naturally into the project's equity Internal Rate of Return (IRR) — the single discount rate at which the present value of all future equity cash flows (dividends received, plus any terminal or exit value) exactly equals the initial equity investment. Because Nepali hydropower has such a distinctive cash flow shape — years of zero revenue during construction, tight but improving DSCR and modest dividends during the tax-holiday years while debt is being serviced, a step-down in retained cash around the tax-holiday expiry, and a further step-down in net revenue after the year-15 royalty increase, but by then typically with the original debt substantially or fully repaid — investors evaluating a specific project's equity IRR should always ask to see the full year-by-year projection across the entire PPA term, not merely a "steady-state" year plucked from the middle of the schedule, since no single year in a Nepali hydropower project's life is actually representative of the whole.