A Retirement Calculator Says You'll Be Fine. A Monte Carlo Simulation Says About 1 in 4 Runs Ran Out of Money
7 min read
Loading...
7 min read
You are 60. You have Rp2 billion saved. You plan to live on Rp80 million a year, and raise that amount 3% every year for inflation.
You type this into a retirement calculator. It assumes your money earns 7% a year. It says you will still have about Rp4.1 billion at age 90. Green tick. You are fine.
Now run the same plan through a Monte Carlo simulation. About 23 out of every 100 runs ended with the money gone before 30 years were up. Same savings, same spending, same average return. Different answer.
This post explains why, in plain words, and how to read a "probability of success" number.
A Monte Carlo simulation is a way to test a plan by trying it thousands of times with random luck.
Think of rolling a die. You cannot predict one roll. But if you roll it 10,000 times, you learn a lot about the pattern.
For retirement, each "roll" is a made-up 30-year future. In each future, the market has good years and bad years in a different order. The computer checks if your money lasts. Then it does it again. And again. 20,000 times.
At the end you count. If your money lasted in 15,400 of 20,000 runs, your plan has a "success rate" of 77%.
A basic calculator uses one number: the average return. It says "7% every year, forever."
Real markets never do that. One year is +25%. The next is -15%. The average may still be 7%, but the bumps matter.
Why? Because you are taking money out while the bumps happen. A bad year early in retirement hurts more than a bad year late. When your balance is down and you still withdraw, you sell more of your pot at low prices. It cannot fully recover. This is called sequence of returns risk, and we explain it in our sequence of returns post.
The calculator cannot see this. It plays only one smooth path. A Monte Carlo run plays thousands of bumpy ones.
Everything below is NetWort's own arithmetic. It is not a forecast and not a real client.
The 7% and 12% are round numbers we picked to look like a mixed stock and bond portfolio. They are not a measured result for any fund or for the IDX. Change them and the answer changes.
| Method | Result after 30 years |
|---|---|
| Calculator, 7% every year | Rp4.13 billion left |
| Monte Carlo, middle run (median) | Rp2.40 billion left |
| Monte Carlo, runs that ran out of money | 23.4%, about 1 in 4 |
| Monte Carlo, when failed runs ran out | Median year 25 |
The median is the middle result: half of the runs ended above it and half below.
Two things stand out.
First, the calculator was too happy. Its answer, Rp4.13 billion, is far above the Monte Carlo middle.
Second, the failure is not at the end. In the failed runs, the money was gone around year 25, at about age 85. That is exactly when you are least able to go back to work.
To make this less abstract, here are three single runs from the same setup, using a different random seed.
All three used the same average return. The only difference was the order of good and bad years. That is the whole story of Monte Carlo in one table.
A result like "77% success" is easy to misread. Here is how to think about it.
It is not a promise. It means that in 77 out of 100 pretend futures, the plan worked. The real future is one single run.
It depends on your inputs. We changed only one thing at a time, using the same method:
| What we changed | Runs that ran out of money |
|---|---|
| Ups and downs of 8% a year | 9.1% |
| Ups and downs of 12% (base case) | 23.4% |
| Ups and downs of 16% a year | 35.6% |
| Withdraw Rp60 million (3% start) | 5.7% |
| Withdraw Rp70 million | 12.6% |
| Withdraw Rp80 million (base case) | 23.4% |
| Withdraw Rp90 million | 36.2% |
A bumpier portfolio and a higher withdrawal both raise the failure rate a lot. A 12% swing is not a rule. It is a guess you should test with your own number.
It ignores what you would really do. Real retirees spend less after a bad year. A simulation that never changes your spending is stricter than real life. That is a reason to treat the failure rate as a warning light, not a verdict.
Many people aim for a high rate, but there is no official line. Different planners use different cut-offs. We did not find one agreed number that fits everyone, so choose a level you can live with.
Be honest about the limits.
If you want the simpler rule of thumb first, read our four percent rule post. It is the single-line cousin of what we just did.
The biggest input above was the swing of 12% a year. Volatility is the technical word for that swing: how much a price moves up and down.
You can see your own. In NetWort, the Portfolio Health card on your Dashboard shows your portfolio's volatility, next to your return, Sharpe ratio and maximum drawdown. Open it and compare your volatility with the 8%, 12% and 16% rows in the table above. That tells you which row of the table looks most like you. NetWort does not run a Monte Carlo retirement test itself yet, so use the table here as the guide.
To see the inflation and interest-rate backdrop that makes the 3% assumption realistic or not, check the macro page.