How much of the racial wage gap is due to differing job recommendations from friends?

This paper estimates the extent to which inequality in jobs found through friends can account for the aggregate wage gap between black workers and others in the US. Data from the NLSY79 are used to estimate a job search model in which individual productivity is distinguished from social capital by comparing the wages and frequency of jobs found directly with those of jobs found through friends. Jobs found through friends tend to pay more, but this premium is lower for black workers; the difference can account for more than a tenth of the racial wage gap.

Here is the paper, by Nicholas Tenev, a job candidate from the University of Wisconsin.


This paper is a variant of the meme "Friends don't let friends watch Friends". Think about it and see if you agree. The corollary is that moving from a ghetto would increase your wages and enrich your life experiences. And they award PhD's over this kind of research?

Research confirming one's intuition can still be useful. Sometimes your intuition is wrong, and intuition is not quantified.

The median household incomes for Asians, Whites and Blacks are:




This correlates with rates of marriage and stable two-parent families. The most glaringly obvious reason for economic disparity seems to be the one that we never talk about.

Life is one damn correlation after another.

Such is life in Tyler's blog.

Chip, The study is of persons of the same skill level, black and white; ie, it controlled for skill levels.

But if it accounts for the number of working adults in a Median household, then the number is lower for Asians than Whites.

So that is what America has become: a country where nepotism and cronyism rule. Either you bow before the cool kida and play ball their way or you don't play ball at all!

This is putting the horse before the cart. I'm not going to recommend someone who's incompetent. It's entirely possible that blacks get fewer personal job recommendations because they're less qualified and their friends know it. The paper makes no attempt to address this hypothesis.

"This is putting the horse before the cart": that's precisely what you are meant to do unless you have a desire to teach horses to push.

+1 Yeah, but don't get too close to the rear end of the horse either.

Doug, The paper is not about getting fewer job recommendations, but rather getting job recommendations that explain part of the gap. The paper controls for skill levels between black and white persons.

Bill, there's a lot more to job competence than quantified skill level.

Yes, there is experience. You gain experience by working, and you get hired without experience through connections. If the group of people you have connections with have businesses, entrepreneurial, have jobs, etc, then you have opportunities that aren't available to someone who doesn't, all else being equal.

Doug, You are really trying hard. The paper controlled for skill level, that is, black and white persons had the same skill level.

Now you slither away by arguing "Oh, it's not skill level, it's only something only I can measure, like something "more than competence" in a skill level; stick with the study, and not what you imagine and can only claim without proof.

derek, I don't understand your point. They are equal skill levels. What you are saying is what the paper is saying: connections matter. The premiums are lower for blacks.

Bill, I don't know who Doug is but if you believe that job performance is only about quantified information, consider that it has been quantified that IQ matters more for job performance than the amount of experience a person has. If what you see as "skill level" is the only thing that matters, the IQ aspect wouldn't be true. And that's just one example of the many factors that can affect job performance besides what the paper qualifies as "skill level", along with other things like personality traits, cultural factors, health, and home life. Incidentally, there is a large average IQ gap between white and black Americans which can impact job performance in a way that educational background sometimes doesn't quantify. And the social problems in black communities (evidenced by stats and experience living around sizable black populations) suggest that they may be less likely to get along with co-workers or their boss. If I have a friend who is qualified on paper but just a bad employee then I'm not going to recommend them.

Anon, You don't have any basis for claiming that an equally skill black and an equally skilled white person have different IQs, and that there differences are race based.

But, I do have evidence of your racial beliefs.

Even economists know that measured skill levels are not the only predictors of competence and success on the job. They are just the easiest to control for. There are also areas like non-cognitive ability which are almost never properly controlled for. Family influences both before and after acquring different skill levels are relevant as are home peer effects (how your friends affect your work output). Think about something like resilience -- what happens when you get an unexpected bad shock (like a poor evaluation) or an unexpected large promotion. Your peers (i.e. friends) have potential influence on this as well. It is well known that in many third world countries the lone person who is more successful than the other members of their extended family is beholden to the rest of the family, second cousins and all. Thus, the friends proxy could be picking up the extended influence of one's peers. Yet the authors assume it is just the limited connections of their poorer friends that is driving their result. They don't even think deeply about how endogenous the peer effects are. Are black people willing to mingle freely with all races and develop peers on economic rather than racial lines more likely to do well? That is relevant for business success. Just as in the old days it was legal to ask if a man was married to see if he was a stable, normal member of the community.

" The paper controlled for skill level, that is, black and white persons had the same skill level."

It certainly claimed to. But it would be more accurate to say that it used the same deeply flawed methods that everyone else does.

Pretty small effect size.

Not if its you. Also, this is wage gap, not employment gap.

It is. Nearly 90% of the wage gap is not explained by job recommendations from friends. What is the primary explanatory factor? Surely not this.

Ah, discrimination? Take a resume and put the name "DeShawn" in it as a substitute for your own name and see what it gets you.

Yes. But that is not a reason why this research and its small effect size is interesting.

Crisare, Please tell me what you think is interesting, and support it with facts.

I wonder if they ever consider changing their name?

Do they have a higher rate of name changes?


Let's play a game.

Tyrone (a black name) is not selected for an interview because of his name being associated with black people. Tyrone changes his name to Tyler, gets offered an interview, and shows up in person for the interview.

What result. If he wasn't going to be interviewed because he had a black name, will showing up for the interview as Tyler overcome this person's prejudice.

It depends on what the HR people are controlling for. If they refuse to hire someone called DeShawn because he is Black, turning up to the interview as Tyler is unlikely to help.

If, however, they are looking for something else, let's suggest someone who is likely to be disruptive, then having a name associated with Black Nationalism is likely to be a red flag. Having a middle class White name is not. So turning up to the interview as Tyler ought to have little effect. It may be as simple as a class marker - a DeShawn is likely to be poor and poorly educated. A Tyler is more likely to know which fork to use with fish. Again there should be no effect at interview.

We do not know which effect matters because everyone just assumes it is racism. There is no basis for this. Jessie Jackson boasted that he used to spit in the White people's food when he worked as a young man. I would suggest that when employers choose to hire someone called Consuela who is just off the truck from Guatemala rather than someone called Latisha they are not being racist. They are being entirely rational.

Here's Bill, whining and nagging like a bitch as he asks for "facts."

There are millions of people for whom putting down DeShawn (which is a stupid name by the way) would improve odds of hiring/being admitted.

So Much and Laderff,

So Much: You drew from that person's name that he was a Black Nationalist and that he is likely to be poor. Your detective skills from a name is amazing. What am I to presume from the name: So Much for Subtlety? Let me guess.

Laderff, Oh, the meme of the discriminated against white guy he loses his job to Deshawn because he has a black name.

You and So Much ought to get together.

Maybe you could form an employment agency.

Maybe you could, idiot: there are apparently millions of underpaid women and "minorities" (i.e., not Asians) waiting for you to profit by. Let me know when your firm gets off the ground.

I suppose it really depends on whether or not independent recruiters are counted as "friend" recommendations. Or for that matter, considered at all.

How would you like not to be rewarded for your productivity or skills by 10%?

You're exaggerating the effect size. The claim is that recommendations explain 10% of the *gap*, not 10% of anybody's entire income. So it would account for 10% of the difference between $16/hr and $24/hr for men (or about $0.80/hr) and 10% of the difference between $15/hr and $19/hr for women (or about $0.40/hr).

You are deliberately misreading what I said. I was responding to prior comments, including Crisare which explicitly is in reference to the gap. I also pointed out that it does not cover periods of unemployment. or job training or skills development. Over a lifetime, or even in the short term, this is a big difference, and it is only an AVERAGE.

I don't know how big the gap is so I'm not sure how significant 10 of the gap is.

Nevertheless my threshold for an interesting finding is much higher than explaining only ten percent of a given inter group difference.

First, like someone who mistakes "money" for "compensation". Secondly, like someone who will be looking for a better job. Third, like someone who thinks that averages apply to individual data points.

There is tons of research on this subject. all of which confirms this result. Look up the work by Granovetter, and if you want to read more I would suggest Jackson's Social and Economic Networks and the SAGE Handbook of Social Network Analysis.
Chapter 22 entitled Labor Markets and Referrals in the Handbook of Social Economics. Jackson et al eds. summarizes the research.

Frankly, I am surprised by some of the comments above.
Don't let intuition get the best of you. Stick with facts and read some research on the subject.

It's not what you know but who you know.

I knew that.

In Hollywood it's not who you know but who has known you.

What is a "friend" in these cases?

In my firm, the warehouse manager hires people he knows from Home Depot. These are all "friends" but he hires the ones he knows show up on time, work hard, etc. He can even know that they are "better" than their resume might suggest.

That said, he has also ixnayed some resumes of "friends" who he knows have problems.

The funniest one was a guy who went crazy at Home Depot and ate a bite out of everyone's sandwiches in the break room fridge. Yeah, that's a deal breaker.

When my stepson rose to managerial level at his business, he quickly hired his best friend and his sister's boyfriend. And when both turned out to be bozos, he fired them so quickly their heads spun - and without apology or remorse. That's when I knew the kid was going to succeed in life. And he has.

But does he still have any friends?

The knowledge you share really changes me in life, I sincerely thank you for the things you have done, sure your blog will help more people.

Comments for this post are closed