Yep, Telly has it right.
Another way to look at it, is just how accurate 4 minutes in a month really is (glass half full).
4 minutes divided by the minutes in a month (60min * 24 hrs * 30 days) = 93 millionths.
Or 93 ppm accuracy.
Or 99.9907% accurate.
Or .0093% inaccuracy.
Amazing just how good a dirt-cheap device is. Try to buy a torque wrench, or a tire pressure gauge, or a meat thermometer, or a bathroom scale that is 99.9907% accurate! That would weigh a 175 pound person to within about 1/4 of an ounce!
They could design in a better part, but it would still be off a minute every month or two. So, if you care, you would still be adjusting it. To get much better, you really need to get to some pretty fancy and somewhat costly circuits; temperature and voltage compensation, maybe a little computer chip to track all the non-linearities. It just isn't worth it for most products, esp when most people change the time twice a year anyhow. And, as stated above, there are options if you really want/need the accuracy.
I'd rather see them spend (and charge) a few more bucks to make sure the clock does not lose it's settings on every little power glitch. That is infuriating to me, yet, you can rarely find out from the product literature if it has that feature or not.
-ERD50