Every rate and cash flow so far has been nominal, stated in rupees or dollars whose purchasing power erodes. A real rate or cash flow is measured in constant purchasing power. The bridge is the Fisher equation (Irving Fisher, 1930):
(1 + r_nominal) = (1 + r_real) × (1 + π) where π = inflation over the period.
How do you calculate the real interest rate?
Solve for the real rate and you get r_real = (1 + r_nom)/(1 + π) − 1. The intuition takes two steps. Your money grows at r_nom in rupee count, but each rupee shrinks at π in what it buys, so your purchasing-power growth is the ratio of the two.
The famous approximation r_real ≈ r_nom − π comes from expanding the product and dropping the cross-term r_real·π. Whether that's safe depends entirely on the inflation level:
| Setting | r_nom | π | Exact real | Approx. | Error |
|---|---|---|---|---|---|
| US-like | 4.3% | 2.5% | 1.756% | 1.8% | ~0.04 pp, harmless |
| India-like | 7.0% | 5.0% | 1.905% | 2.0% | ~0.10 pp, visible |
| High-inflation EM | 60% | 50% | 6.667% | 10% | 3.3 pp, catastrophic |
So use the approximation for talking and exact Fisher for computing. At India-level inflation, always compute exactly. In anything high-inflation, make no exceptions.
Advertisement
India vs US real interest rates: a worked example
The India–US contrast, worked. Use typical recent-decade levels. For India, take a 10-year G-sec at ~7.0% and CPI at ~5%. For the US, take a 10-year Treasury at ~4.3% and CPI at ~2.5%. These are illustrative, so pull today's numbers. G-sec yields come from FBIL/RBI's DBIE or the NSE, and India CPI from MOSPI. Treasury yields come from treasury.gov or FRED series DGS10, and US CPI from BLS/FRED. Here's the arithmetic:
- India real rate = 1.070/1.050 − 1 = 1.90%
- US real rate = 1.043/1.025 − 1 = 1.76%
This result should rewire your intuition. India's rates look enormously higher, 7% against 4.3%, but almost the entire gap is inflation compensation. The real reward for waiting is within 15 basis points of the US level. Nominal-rate differences across countries mostly reflect inflation differences. Nobody is handing out a free lunch. It's also the seed of why the rupee structurally depreciates against the dollar by roughly the inflation differential, the international Fisher relation, which the macro phase develops properly. For a market-based shortcut to expected π, compare a nominal bond with an inflation-indexed one of the same maturity: TIPS in the US, RBI's inflation-indexed bonds in India. The yield gap is the breakeven inflation the market itself prices.
Why must nominal cash flows use nominal rates?
The consistency law, the one rule that is never optional:
Discount nominal cash flows at nominal rates. Discount real cash flows at real rates. The two, done right, give identical answers. Mixing them is not a small error: it compounds with the horizon.
Now prove it by construction. The numbers are synthetic, and it's the most instructive table here. A project produces cash flows worth ₹100 crore/year in today's purchasing power for 10 years. Real CFs are flat at 100, and nominal CFs grow with π = 5%. With a nominal discount rate of r_nom = 12%, r_real = 1.12/1.05 − 1 = 6.667%.
| Method | Cash flows | Rate | PV |
|---|---|---|---|
| Correct (nominal) | 105, 110.25, … 162.9 (growing 5%) | 12% | ₹713.3 cr |
| Correct (real) | 100 flat | 6.667% | ₹713.3 cr, identical to the paisa |
| Wrong mix A | 100 flat (real) | 12% (nominal) | ₹565.0 cr, understates 21% |
| Wrong mix B | growing 5% (nominal) | 6.667% (real) | ₹918.0 cr, overstates 29% |
Mix A, flat "today's rupees" cash flows discounted at a nominal WACC, is the most common DCF error on the subcontinent. An analyst forecasts "₹100 crore a year, conservative" in their head, in today's money. Then they discount at a 12% nominal rate that already contains 5% inflation compensation. Inflation gets charged in the rate but never credited in the flows.
At US inflation the same sloppiness costs a fifth as much. So Indian analysts have to be stricter about Fisher than their US counterparts. The same level of care isn't enough. By year 10 at 5% inflation, the nominal and real versions of the same cash flow differ by 63% (1.05¹⁰ = 1.629). At 2.5% they differ by 28%. The error grows with horizon, and valuations run over 10+ year horizons.
Nominal or real in a DCF, and does currency matter?
Here are the practical defaults you'll carry into building a DCF. Work nominal, because statements, market rates, and comparables are all nominal, and let your forecasts contain inflation. Revenue growth is "volume + price," where price growth reflects π. Go real only when the whole analysis is real end to end. That happens in some infrastructure and regulatory work, and in high-inflation economies where nominal forecasting is meaningless.
There's a cross-currency corollary. A ₹ cash flow carries ~5% inflation and a $ cash flow ~2.5%, so ₹ streams take ₹ discount rates and $ streams take $ rates. Discount an Indian company's rupee FCF at a US-derived dollar WACC and you embed a 2.5-point inflation mismatch that compounds forever. You'll formalize currency-matching when you build the cost of capital, and terminal-value g caps differ by currency for the same reason.
Is headline GDP growth a safe terminal growth rate?
Last, connect it backward. The ~7% "India growth story" number in headlines is real GDP, 7.7% in FY2025-26. Nominal GDP grew 8.9% that year and has averaged roughly 10 to 11% over the longer run. Those are MoSPI's provisional estimates of 5 June 2026, on the new 2022-23 base year, checked October 2026, so verify both. Neither figure is a terminal-growth cap. Today's nominal growth describes today's India, and it belongs in the explicit forecast window, where a company can ride it for a decade. A terminal growth rate describes a mature economy long after that window closes. Its cap is the risk-free rate in your own stack, as the perpetuity section set out in full. A rupee terminal g of 5 to 6% looks conservative beside a 9 to 11% headline. But on the cleaned rupee stack it's at or above the ceiling and needs defending. Real-versus-nominal confusion works both ways, and so does horizon confusion.