There are a lot of anecdotal data points about the sweatshop end of the H-1B spectrum, but quite a few of us benefit from being at the flip side: real, legitimate employers want to ensure we get an H-1B so they take the reasonable salary requirement seriously and offer at or above the market average. H-1Bs are subject to a quota which is usually exhausted within the first few days of new availability each year, so you get only one shot at this. If you have a specific person you actually want to bring in, you don't take any chances.
If it were only the sweatshop kind of people, that'd lower the average. If it were only the specific offer kind of people, that'd raise the average.
I know a lot of people have an emotional answer about what they really, really believe is true on how that averages out, but I'd love to see some real data.
If I had to guess, I'd imagine that the overall H-1B average offer isn't as far below the market average as concentrating on the sweatshops would imply, and that the quota makes the total number of H-1Bs entering the pool each year have only a negligible effect. It's at most 65,000 people a year across all industries; there are believed to be at least 6.7m tech workers in the United States. Even if every single H-1B were for a tech job, would an influx of less than 1% each year really have much effect?
H-1Bs are time limited, so even if you assume that everybody stays for the totality of their visa, obtaining the maximum available extension, and that everybody who came in was a tech worker, that's still less than 6%.