Re:The role of high-skilled foreign labor in startup perform...
by orange_joe on Monday August 23, 2021 @01:21PM
(#28279114)
This paper seems to suffer from selection bias, as the examine the a large universe of VC backed companies. Not all VC companies can afford H1-B employees and this implies that only better funded companies are more likely to succeed. There’s a vast difference in amount of funding a VC backed startup can raise, so it doesn’t really make sense to compare a company that raised 100K through YC and a pre-IPO unicorn. The paper didn’t seem to control for this when I looked through the first few pages.
[ Parent
| on HN ]
Re:The role of high-skilled foreign labor in startup perform...
by shalmanese on Monday August 23, 2021 @03:57PM
(#28281129)
The paper uses the H1B lottery process as the randomization so there's no selection bias. All the companies studied applied for H1B employees, just some got approval and others got rejected at random by the lottery.
> In our sample of VC-backed startups, we observe that winning more H-1B visa lotteries enhances startup performance. Specifically, a one-standard-deviation increase in the likelihood of winning an H-1B visa lottery increases the number of VC financing rounds in the three years after the lottery by 7.1%, the number of patents filed by 6.2%, the number of claims per patent by 9.7%, and the likelihood of an IPO by 8.8%.
Re:The role of high-skilled foreign labor in startup perform...
by orange_joe on Monday August 23, 2021 @10:57PM
(#28284658)
good point, I didn't see that, so its still better than I expected. That being said, The Lottery isn't truly random. Hiring a Master's Degree holder(higher salary than the bachelors student), applying multiple years (Company longevity usually means better financials) and filling out the forms with correctly with a skilled immigration team (again, more financial requirements) all improve the odds. So you're definitely right that the selection bias isn't as bad as I originally thought, but it still seems to be present.
Re:The role of high-skilled foreign labor in startup perform...
by marcinzm on Monday August 23, 2021 @07:54PM
(#28283557)
I'm guessing they accounted for applicants with advanced degrees who have a separate cap or did that not exist during the time of the study?
Re:The role of high-skilled foreign labor in startup perform...
by rightbyte on Monday August 23, 2021 @02:50PM
(#28280285)
Ye the authors are taking correlation for causality big time. It is just a bad paper, judging from the abstract.
Re:The role of high-skilled foreign labor in startup perform...
by bjornsing on Monday August 23, 2021 @05:33PM
(#28282272)
What? This is one of few papers I’ve ever seen (on this kind) that can make a causal claim. The visa lottery essentially makes it a Randomized Controlled Trial (RCT), the gold standard of science.
Re:The role of high-skilled foreign labor in startup perform...
by wins32767 on Monday August 23, 2021 @06:28PM
(#28282819)
It's randomly controlled within startups that apply to sponsor visas. That implies some level of initial success since visa sponsorship is a huge pain in the ass.
Re:The role of high-skilled foreign labor in startup perform...
by johncolanduoni on Monday August 23, 2021 @07:01PM
(#28283145)
But it doesn’t include startups that didn’t apply to sponsor visas in the study, so isn’t that sufficient for randomization purposes?
Re:The role of high-skilled foreign labor in startup perform...
by koolba on Tuesday August 24, 2021 @12:10AM
(#28285010)
It doesn’t show that it’s a net plus vs not applying. It shows that winning the lottery that you choose to enter is better than not winning.
Re:The role of high-skilled foreign labor in startup perform...
by rightbyte on Tuesday August 24, 2021 @06:52AM
(#28287101)
Are companies randomly given lottery tickets normalized by employee count or do e.g. bigger successful companies get more tickets?
Re:The role of high-skilled foreign labor in startup perform...
by thunkshift1 on Monday August 23, 2021 @06:29PM
(#28282827)
Did you read the paper or just the abstract?