Well, tfa says it pretty clearly: When asked to estimate the real value of TVs priced at 4998.10, 5000 and 5012.56 the middle one was estimated nearly $500 lower than the others.
Same goes for houses and other situations where you're likely to be conceived as having rounded up.
This says it clearly for TV prices in the interviewed customer segment and house prices in the interviewed customer segment. It doesn't mean that other situations don't do better with other prices.
Working in IT I have some friends who would pay $503.71 more for the same TV (compared to the price tag on the TV itself) if you present them a complex mathematical formula for why that TV should be valued higher, but who wouldn't pay $503.70 more because that's not what the formula says.
What I want to say by that example: Different people have different thinking and test your customers, don't assume you are right because a book says so.
Interesting. Wasn't something like this in the undercover used car salesman article a while back as well? You should always come up with weird numbers so the customer doesn't think you made it up.
Same goes for houses and other situations where you're likely to be conceived as having rounded up.