Why (I Think) It’s Cheaper to Use Your Card

When I went to Korea two summers ago, many advised me against using my Visa debit card to make purchases. Instead, they said, I should withdraw cash from the ATMs inside every convenience store, where I'd avoid my bank's foreign transaction fee of three percent. Their advice was wrong, and I have an Excel spreadsheet to prove it.

The important detail here is that when you access a foreign account from a Korean ATM, you pay a flat fee of ₩3000 (a little less than $3). Now, the ATM's withdrawal limit is ₩100,000 per transaction, so if you withdraw the maximum each time, the percentage cost of the transaction fee bottoms out at three percent—exactly what you'd pay most American banks in foreign-transaction fees. (This can vary greatly depending on your bank, of course. This guy wrote in 2013 that most American banks charge a flat fee rather than a percentage, but my bank is on his list and has apparently changed its model since. Citibank, on the other hand, advertises that they charge no fee whatsoever.) I assume this is the reasoning my friends had in mind when they told me to use the ATM: the fees are transparent and static, and as long as you withdraw the ₩100,000 limit every time, the surcharge is no worse than the bank's. It made sense to me, and I stuck to that advice for the whole summer.

However, when I got home, I checked my bank statement to find a nasty surprise: I was charged my bank's foreign-transaction fee on top of the ATM's built-in fee for each of my withdrawals. Effectively, I was paying ₩103,000 × 103% = ₩106,900 for every ₩100,000 withdrawal.

And that's not even the whole story, because I was paying in dollars, not won. As it turned out, the ATMs I used in Korea consistently undervalued the US dollar by about one percent in addition to applying their ₩3000 fee. I discovered this because my bank listed the foreign-transaction penalty separately on my statement—meaning I could see how much the ATM "billed" my bank before the bank applied its own conversion fee. So, I pulled my bank statement into a spreadsheet, looked up the true exchange rate for each of my transaction dates, and did a little bit of algebra to find the factor by which the Korean ATMs tweaked the value of the US dollar. There was some ambiguity in the math because I didn't know the exact times of the transactions; thus, the ATM's stretch coefficient (the proportion by which it overvalued the Korean won) varied between 1.006 and 1.015. But I mitigated the effects of this hourly exchange-rate variation by averaging over all the withdrawals I made. That average was quite close to 1.01, or a one-percent undervaluation of the US dollar. For the stats people: a one-sample T test rejected the null hypothesis (stretch coefficient = 1) with p = 0.0011. For the non-stats people: that means I'm pretty sure something's going on!

Once I accounted for all these various fees, I realized that by following past tourists' advice, I paid more than a seven-percent surcharge to access my money in Korea. It's better to just use your Visa.

If this were a snappy travel blog, I'd end the post here without any qualifications. But because I have an academic heart, I feel compelled to call attention to a critical limitation of my research (did you spot it already?): I don't know what exchange rates my bank would follow if I paid directly at the register. If it's the rate they charge for traveler's checks—a serious price gouge—the ATM strategy could still prove to be more cost effective. But at least I'm armed with something the anti-credit-card windbags lack: a kickass spreadsheet for comparing currency-conversion techniques.