What return should you actually assume in Nepal? The number your SIP and retirement math depends on

A reader sent me his retirement spreadsheet in Bhadra. Rs 25,000 a month into a mutual fund SIP from age 30, 12% return, 4% inflation, retire at 55 with Rs 4.7 crore. The arithmetic was correct. The two assumptions were doing all the work, and neither came from anywhere in particular. The 12% was the number the fund's app had pre-filled. The 4% was "roughly what inflation is now".

Change 12 to 9 and 4 to 6 and the same spreadsheet gives Rs 2.8 crore that buys what about Rs 65 lakh buys today. Same saver, same discipline, roughly a third of the outcome. Nobody in Nepal publishes a "long-run expected return" the way US planners lean on the S&P 500's ten percent, so people borrow the US figure, or the fund marketing figure, or whatever FD rate they last saw. This post does the work once: what the Nepali record actually shows for each asset, and what number to type into the box.

The assumption matters more than the saving rate

Before the evidence, the reason it matters. Take Rs 10,000 a month for 25 years and vary only the return.

Standard future-value-of-an-annuity arithmetic, monthly compounding; the formula is in the SIP post. The gap between 8% and 12% is bigger than the gap between saving Rs 10,000 and saving Rs 19,000 a month at 8%. Whoever picks the return has picked the answer.

The rule of 72 is the quick check: money doubles in 72 ÷ rate years. At 12% that is six years and four doublings in 25; at 6% it is twelve years and two doublings. Which is why the mis-set return in that spreadsheet mattered more than any other line on it.

Start with inflation, because every return is measured against it

A return only means something after prices. NRB's annual average consumer price inflation, fiscal year by fiscal year:

The simple average of those ten years is about 5.6% (my arithmetic; FY 2020/21 is omitted because the published figures conflict). The last fiscal year's 3.08% was the lowest in the series and it is already gone: the year-on-year print was 5.14% in mid-July 2026 and 5.96% in mid-August, against 1.68% a year earlier.

Longer memory says the decade was mild. NRB's own inflation report tabulates period averages of 9.88% for 1973/74 to 1989/90, 7.64% for 1990/91 to 2005/06 and 7.41% for 2006/07 to 2022/23. The rupee's peg to the Indian rupee imports India's inflation, which was 4.82% in August 2026, and a plan that runs thirty years should not assume Nepal stays below its own long-run average. Use 6%. Use 5% if you want the optimistic case and say so in the spreadsheet.

Deposits: the rate you see is not the rate you get

The fixed-deposit rate feels like the one number everybody knows, and it is the one that has moved most. NRB publishes the weighted-average rate commercial banks actually pay on fixed deposits every month. The mid-July readings:

Sources: NRB Banking and Financial Statistics for Poush 2076, Poush 2079 and Asar 2083.

Average the ten mid-July FD readings and you get about 7.9% (my arithmetic). Knock off the 6% TDS and it is roughly 7.4%, against 5.6% inflation: a real return near 1.8% across the decade, earned entirely in the two liquidity crunches of 2018 to 2019 and 2022 to 2023. Today the best individual FD among the twenty commercial banks pays between 3.91% and 5.25%, the weighted average is 4.95%, and inflation is 5.96%. The real return is negative and where to park money has become a live question.

Government paper tells the same story from the other side. The 91-day Treasury bill yielded 6.35% in mid-July 2023 and 2.32% in mid-July 2026. The Citizen Savings Bond paid 11.50% in early 2023 and 6.5% on the 2087 issue sold in late 2025. Bank debentures went from 10.75% (NMB, April 2023) to 7% (Nepal SBI, July 2025) to 6.25% on Prime Bank's mid-2026 issue.

Planning number: 6% nominal for deposit-type money over a full cycle, which is a little below the decade average because the last decade included two spikes that may not repeat. Real return: call it zero. Deposits are for certainty, and how much of your money should sit there is a question about your horizon, not about the rate.

Equities: 9.4% before dividends, and the entry point decides everything

Nobody publishes a total-return index for NEPSE, so the record has to be assembled. Investopaper's history of the index puts it at 176.3 in FY 2053/54 (1996/97) and 1,102.64 on 5 March 2019, "a gain of 525% over 22 years which is 8.69% compounded annually", with the explicit caveat that this "excludes the dividend return received by investors". Carry the series forward to the 2,647.22 close on 18 September 2026 and thirty years of compounding comes to roughly 9.4% a year, still before dividends (my arithmetic: 2,647 ÷ 176.3 over 30 years).

That is the long run. The medium run depends entirely on when you started.

Dated levels from Investopaper, the all-time-high report, and a 2024 Nepali journal table of month-end closes; CAGRs are my arithmetic.

Two things sit on top of the price line. Dividends: Nepali companies pay heavily in cash and bonus shares, and both raise your holding beyond what the index shows. Bank shares currently yield around 4% at market prices, the whole market closer to 1 to 2%; across the thirty years a 2 to 4% annual dividend layer on the 9.4% is a reasonable reading, which puts the historical total return in the low teens. And valuation: the market traded at a P/E of 38 against its historical average of 31 in late 2025. Buying above the average multiple has historically meant below-average forward returns, which the 2016 and 2021 rows above demonstrate.

Mutual funds are the same asset with a manager in between. ShareSansar's study of matured closed-end schemes found five-year schemes paid an average 25.25% a year on their Rs 10 par (Siddhartha Investment Growth Scheme-1 returned 242% in total), while seven-year schemes averaged 12.61%, and at Asar-end 2083 14 of 43 closed-end schemes traded below par. The oldest open-end fund has compounded at about 8.26% over seven years, which is the honest number to hold against the 12.74% the marketing quotes.

Planning number: 9 to 10% nominal for a diversified Nepali equity holding including dividends, 3 to 4% real at 6% inflation. Use the bottom of that range if you are buying at today's multiple, and give it fifteen years. Do not use 12% for money you will need in under ten.

Gold: the decade's winner is the hardest to plan on

Gold's record is spectacular and mostly not repeatable. FENEGOSIDA's reference price was Rs 55,800 a tola in 2073 BS; the record was Rs 339,300 on 29 January 2026; the price on 24 September 2026 was about Rs 299,100, per FENEGOSIDA's daily rate. From 2073 to now is roughly 5.4x in ten years, about 18% a year (my arithmetic), the figure that has made gold the decade's winner in every look-back.

Unpick it. Gold is priced in dollars, and the rupee went from Rs 106.73 to the dollar in July 2016 to Rs 152.89 in September 2026, about 3.5% a year of depreciation (my arithmetic). That part of gold's Nepali return is really a currency return, and it is the part you can reasonably expect to continue while Nepal's inflation runs above the dollar's. The other 14 points a year came from a global gold surge, dollar gold rising 20% in the single year to mid-July 2026. Surges end. The January-to-September 2026 drop from Rs 339,300 to Rs 299,100 is 12% in eight months, on an asset people buy for safety.

Planning number: inflation plus rupee depreciation, so around 6% nominal, zero real, with wide error bars in both directions. Hold gold for what it does to the portfolio's volatility and its dollar exposure, and size it in tolas rather than as a return engine.

Land: there is no number you can defend

Kathmandu land is the asset every family believes in and the only one with no measured series. What exists: NRB's proposed real-estate index reported 26.45% average real-estate inflation in FY 2076/77, falling to 12 to 16% the following year; an NRB study from 2011 quoting the IMF that Valley prices had "quintupled in some areas"; and NRB's latest quarterly data showing transaction value up 87% nationwide in FY 2025/26, with Kathmandu Metropolitan City up 217%. Transactions, not prices. The widely repeated figure of a 20-odd percent fall in Valley prices from the 2023 peak comes from property blogs with no stated method.

Land probably compounded in the low double digits over the past two decades in well-located Valley plots, and I cannot show you a table that proves it. That is the point. A planning assumption you cannot source is a guess, and the illiquidity, the 7.5 to 10% capital gains tax and the months-to-sell reality covered in the look-back post make land the wrong place for a number that needs to be reliable.

Planning number: for the house you live in, nothing; it is consumption, not return. For investment land, inflation plus two or three points, and only for money you will not need for a decade.

Retirement funds: the guaranteed slice is sliding

The mandatory and semi-mandatory schemes are where most salaried Nepalis' long-term money actually sits, and their rates have followed deposits down.

The EPF and SSF figures for the current year are not cleanly published; the two EPF numbers above conflict, and SSF has declared nothing recent that could be found. The direction is unambiguous. These schemes invest in the same deposits and bonds whose rates have halved, and their tax-deductible status, not their return, is the reason to fill them. CIT vs PF vs SSF covers the choice.

Planning number: 5% nominal, which is below inflation on today's print. The tax deduction on the way in is where this money earns its keep.

The house numbers

Put together, for a Nepal-resident saver who cannot legally buy foreign stocks:

These are my planning judgements drawn from the records above, not forecasts. A 60/40 split of equities at 10% and deposits at 6% compounds at about 8.4% nominal, which is about 2.3% real at 6% inflation and 3.2% at 5% (my arithmetic). Call the defensible band 3 to 4% real for a diversified Nepal-only portfolio.

That has a consequence for this blog's own earlier posts. The Coast FIRE post uses 5% real as its central case; treat that as the equity-heavy, good-entry-price scenario. The FI roadmap shows a 7% real scenario and already warns it is optimistic; on the record above it is a US figure, roughly the S&P 500's 6.8% real since 1957, and Nepal has not matched it over any full cycle. The 3 to 3.5% withdrawal rate the retirement posts recommend is the correct partner to a 3 to 4% real return.

The same spreadsheet, honestly filled in

Back to the reader's plan: Rs 25,000 a month for 25 years.

Future value of a monthly annuity, deflated by the inflation factor over 25 years; my arithmetic throughout.

The bottom two rows are not a reason to stop. They are the reason to save Rs 35,000 instead of Rs 25,000, or to plan on 60 instead of 55, or to accept a smaller retirement than the app promised, decided now rather than discovered at 54. A plan built on 3% real that gets 5% has a pleasant problem. A plan built on 8% real that gets 3% has a house sale.

What you actually need to know

1. Use 6% inflation and think in real terms. The decade averaged 5.6%, the long run 7 to 10%, and the latest print is 5.96%. Every return in the plan is only worth what it clears after that.
2. Deposits are zero real, equities 3 to 4% real, everything else unproven. NRB's FD rate averaged 7.9% but is 4.95% now; NEPSE compounded 9.4% before dividends over thirty years but starts today at a P/E of 38; gold's 18% was half currency and half a surge that has already given back 12%.
3. A 60/40 Nepal-only portfolio earns about 2.5 to 3.5% real. Build the retirement number on that, pair it with a 3 to 3.5% withdrawal rate, and treat 5% real as the good scenario and 7% as a foreign one.

Want your own spreadsheet checked against these numbers? Email parjanya57@gmail.com with the monthly amount, the horizon and the return you typed in, and I'll send back the same table for your figures.

This post is part of the Nepal Money Basics guide — the investing section.