In an efficient, transparent labour market where a person's productivity was obvious to buyers, I would expect you would be right.

Personally, I believe the market is so opaque; the water so turbid that the cost of living difference is almost noise compared to the value I can get by finding good people who other companies don't notice.

I think the thing is that nearly everyone trains to send false signals in this market. It's extremely difficult to measure a persons ability at an interview. (hell, it's no cakewalk to measure a persons ability at all, but I have a hard time imagining a worse way to evaluate non-sales candidates than an interview.)

I'm spending a lot more effort finding the person than most companies do, and that's a lot easier for me to do in America. Almost everyone I've hired that worked out well was a friend or a friend of a friend. I've tried many foreigners; the only one that was really good was someone who I knew personally (she worked with me in the US at my dayjob, then worked for me remotely after she went back home.)

I don't think this is due to a dearth of good people overseas; I believe this is just because the hard part is figuring out who is good and who is not, and really without working with someone for a while, that's an extremely difficult thing to do.

As for experience, yeah, these people were more valuable after working for me than before; but at least some of these people had decent paying jobs before, then lost them in the downturn, lost hope, and ended up doing menial jobs.

One was the classic example; this guy happened to be my roommate at one point. He was obviously brilliant, but he practically radiated self-doubt. He had this slouch that took a few inches of his height and make him look a little like he thought you were going to hit him.

But he was obviously brilliant, and had worked as a C/C++ programmer for a while, he even had some open-source code out there. But he lost his job during the .com crash and had been subsisting since on savings, menial jobs and an occasional elance-type gig.

oh man, and his lack of confidence absolutely killed him there. He had this one client who'd call him up for hours every night and kept adding features and changing requirements. He said he made $2/hr on the gig when he said he was quitting. At that point I think he was into me some for rent, so I offered to hire him at $40/hr. I called up his customer and explained that I'd get the work done for $60/hr, but he couldn't talk to my friend, and that he'd be paying for phone, time, too (I was, well, quite a bit younger at the time. 22? 23? and thought $60/hr was a fine wage for yelling at/getting yelled at by some asshole. I was just figuring out the whole 'if you are arrogant and aggressive, people give you what you want' thing.) The job got done, and everyone got paid, and my ego got stroked. (I mean, yeah, the wages were not awesome, but eh, when you are that age, it's certainly a living wage)

Really, once you got down to coding, my friend's apparent self-doubt evaporated like it was just an illusion. "I don't write segmentation faults" he insisted. But he did really badly in interviews. But my point was that he was really good before he worked for me; but that fact was obscured by the year or two of downtime after the .com crash. My friend eventually got noticed by a real recruiter, and got a full-time job shortly thereafter. he currently works for some compiler company or something making rather a lot more money than I do. (more than I was making when I left my full-time job... my current business pays me, ah, mostly in equity.)

obviously, one anecdote does not equal statistical significance; I'm just saying, I've seen people with experience fall behind in the 'interview arms race' and end up underemployed, to the detriment of their potential employers. Because it is so hard to sort the really good people from the mediocre or the useless, there is huge value that can be had correcting the market's mistakes.

After reading this, I'm pretty sure we generally agree but I didn't really say much in my first comment. Anything approaching full time work is really going to work differently than Rent-A-Coder style projects. I would expect the costs for my mythical Romanians to approach the costs of an American like your roommate, though their actual hourly rate might be cheaper.

I will try to be more explicit next time I make a throwaway one-line comment.