Big business advocates of more immigration love to say that 52 percent of the Silicon Valley startups were associated with immigrants, as mentioned in an earlier blog "Start-up Founders, Patents, Lies, Damn Lies, Statistics and Migration Policy".
As I pointed out then, the 52 percent is not the percentage of Silicon Valley startup founders; it is the percentage of such firms that had one or more immigrants among their founders, and each firm can have one to multiple founders. Such studies never calculate the percentage of founders who are immigrants, a percentage sure to be substantially less than 52 percent.
Well, a reader asked about similar claims of immigrant influence in other groups of corporations, claims that always are used to support further increases in the levels of immigration, and the number of vulnerable foreign workers to be admitted to work for big corporations.
The most quoted of such studies was done a couple of years ago by the big-business oriented Partnership for a New American Economy (PNAE, founded by New York's billionaire mayor, Michael Bloomberg). Its headline finding was:
More than 40 percent of the 2010 Fortune 500 companies were founded by immigrants or their children.
Now, 40 percent is not quite so impressive as 52 percent but it would seem to indicate that first- and second- generation immigrants are more likely to start big corporations than the rest of us.
A careful examination of that study – something I should have done at the time – shows nothing of the kind.
Yes, many immigrants were involved in the early stages of many of these corporation, but what the study also shows is the manipulative skills of its (unnamed) authors.
There are, upon careful examination, four separate and distinct research maneuvers - questionable ones – on display here:
- statistical concepts;
- statistical comparisons;
- definitions of "founders" and, in at least one instance,
- the definition of "immigrants and their children".
Statistical Concepts. The basic problem with this study, as with the Silicon Valley study, is the definition of "companies with immigrant founders." It counts the Fortune 500 companies with one or more immigrant founders or founders who were children of immigrants; it does not count, as a responsible study would, the total number of founders and then compare that number to the immigrants, and/or children of immigrants among them.
The whole study, in short, was being played with a loaded deck.
Statistical Comparisons. The study then compares the 40 percent of the firms that were founded by one or more immigrants and their children to this statement: "Even though immigrants have made up only 10.5 percent of the American population since 1850 ..."
So, we see immediately that the authors are comparing the apples of firms founded by immigrants and their children to the oranges of the percentage of immigrants in the population on average since before the Civil War. (Never mind that immigrants now constitute about 13 percent of the U.S. population and that percentage will soar if the Senate's omnibus bill is signed into law.)
Bearing in mind that the basic statistical concept is somewhere between flawed and fraudulent, what happens when you compare the percentage of such firms to the percentage of the total population that is either foreign born, or has one or two foreign-born parents? This is not done, as far as I could see, in the PNAE study.
The current percentage of the U.S. population that is either foreign-born, or born to one or two foreign parents is about 24.5 percent according to a February 2013 report of the Pew Charitable Trust. That figure was considerably higher at the turn of the last century (34.5 percent), and was something like 25 percent, on average, for the 20th century. The 40 percent of the firms founded by one or more migrants - a dubious number at best - should be compared to 25 percent of the population, not to 10.5 percent of it.
But it gets worse.
Definitions of Founders. Clearly, the broader the definition of "founder" the larger the number of them, and thus the chances that one of them will be either an immigrant or the child of one. While the authors' exact methodology on this point is not described, the listing of companies and founders in the appendix suggest some rather casual definitions were used from time to time in this study.
There are two kinds of loose definitions. First, how do you define the founder of a unitary firm (one that is not a conglomerate), and secondly how do you define a founder of the second type of firm? Without examining all 204 firms on these variables, I dipped into just two of them, and found immediate problems, perhaps symptomatic ones.
So who is a founder of a unitary firm? Let's look at Intel.
As my colleague, John Miano, pointed out to me, Intel's founders were two native-born Americans, but immigrant Andrew Grove (listed as a founder in the PNAE document) was an early employee, not a founder. It is not clear how often this stretch is used in the study.
And who counts as a founder of a conglomerate, such as Pentagon contractor United Technologies? This gets complicated. UT was once a part of an even bigger mega-corporation that included both Boeing and United Airlines, till that was broken up by the New Deal, something I just learned yesterday.
If you turn to UT's corporate history, you will see a series of pictures of the firm's pioneers, starting with E.G. Otis of elevator fame (1853), William Courier, the inventor of air conditioning (1902), Igor Sikorski, the helicopter guru (1923), and Francis A. Pratt and Amos Whitney (1925), who made significant advances in airplane engines. One of these five luminaries, Sikorski, was born overseas so UT becomes, in the authors' eyes, an immigrant-founded company. Even though he shows up seventy years after another founder started a major segment of the firm.
Finally, there is the question of how you define an immigrant or a child of one. Although it did not impact the statistics, the study (in a footnote) said that Oracle co-founder Larry Elison "was adopted by his grandfather, an immigrant from the Crimea in Ukraine." Elison was born in New York City to parents who presumably were American-born (or else the study would have mentioned their birth place) who then abandoned him. Two other biographies of him say that he was adopted by his great-uncle and his great-aunt.
Since Elison was born in 1944 the adopting great-uncle must have emigrated from either Czarist Russia or the USSR, many decades before Ukraine became a separate nation. (Elison's nation of origin is listed in the appendix as Russia). I cite this example to show the degree of flexibility in the definitions used in this study, as well as to note two apparent errors in a single, eight-word phrase.
So Elison is listed as a founder of Oracle who was a child of immigrants. This does not tilt the statistics in this case, because another Oracle founder was born to parents who migrated from Iran.
Policy Considerations. The PNAE study uses a combination of facts about some immigrant corporate founders and a pallet loaded with questionable research techniques in an attempt to show that migrants are more likely to found big corporations than natives, but, on close examination, it fails to do so.
Further, such studies are often trotted out in support of a "need" for more immigration, generally, with little recognition that the corporate founders cited were all admitted under current or then current immigration laws.
Why change our laws when we are getting such good results with the ones we have?