Annual 10 Million KRW Installment Savings vs. ETF Auto-Trading: 3-Year Simulation Results
I compared personal installment savings with monthly ETF investing using data from the past three months. To cut to the conclusion, in a volatile market like this year's, the savings account was better in terms of interest income stability, but on a 3-year annualized basis, the ETF recovered faster after making up its drawdown.
The comparison criteria are as follows:
- Installment savings: 4.2% fixed annual rate, 500,000 KRW per month, tax preference applied
- ETF: S&P 500 tracking product, 500,000 KRW per month, dividend reinvestment assumed
- Period: rolling simulation using 2021–2024 data
What was interesting was that the ETF posted negative returns in exactly two periods, yet recovered its principal within six months both times. In contrast, the installment savings had severely poor liquidity due to early termination penalty conditions.
My conclusion is: if your drawdown tolerance is 20% or more, ETFs are better; if it's below 10%, savings/deposit accounts are better. Personally, I'll put together an asset allocation table by age group in the next post.
6 answers
Agreed, the conclusion is clean.
Well, if it's a three-year simulation, isn't the sample size too small?
I've actually done it for 3 years... I once broke a savings account due to the early termination penalty. With ETFs, even when they're down, I can feel them recover at some point if I just hold on, so I ended up sticking with it for the long haul.
Using your own drawdown tolerance as a criterion is good, but the problem is that you don't know your drawdown tolerance until you actually experience losses. In simulations, people who can withstand a 20% drop often struggle at just 8% in real life. The age-based allocation table is also fine, but it seems more realistic to set the risky asset allocation based on the amount of loss you can afford.
Does the dividend reinvestment assumption account for fees?
Isn't 4.2% on a savings account an amazing rate? Is there even a product like that on the market right now? Isn't that old data? Also, if it's a rolling simulation, the periods overlap, so they're not statistically independent. You should be careful when interpreting the results.