The Value of Complementary Coworkers

In today’s world, most workers are highly specialized, but this specialization can come at a cost – especially for those on the wrong team. New research by Growth Lab Research Director Frank Neffke assesses the importance of the skills of coworkers. Finding coworkers who complement and not substitute one’s skills can significantly impact earning potential. The impact is equal to having a college degree. Coworker complementarity also drives careers and supports urban and large plant wage premiums.

Explore more of the research on the value of complementary coworkers. 


Katya Gonzalez-Willette: Hello and welcome to the Growth Lab at Harvard University's weekly podcast. In today's world, most workers are highly specialized, but this specialization can come at a cost, especially for those on the wrong team. New research by Growth Lab Research director Frank Neffke assesses the importance of the skills of coworkers. Finding coworkers who complement and not substitute one's skills can significantly impact earning potential. The impact is equal to having a college degree. Coworker complimentarity also drives careers and supports urban and large plant wage premiums. 

Chuck McKenney: Thank you, Frank, for joining us. Your latest research on skills recently published in the journal Science Advances reveals the importance of teams and coworkers when it comes to productivity, earning potential and stays of employment. Can you tell us why it is important to study skills and teams? 

Frank Neffke: So the way we have been thinking about skills or what economists typically call human capital is that they are something that belongs to an individual. So we have skills and we sell them to firms. When we work for a firm, basically we offer them our skills. And if we have a high level of skills, or maybe if you have a high level of education, those skills are more valuable to the company than if you have lower levels of education. So in that sense, there are two things that are implicit in such a worldview. First is a very binary view of what skills are they are either high or low. And the other thing is that they are an attribute of an individual. And I think both are not that helpful if you really want to understand how skills work in a modern economy. So let me first get to the binary part. In reality, workers are not high or low skilled. Workers have skills that cannot be ranked along one particular metric. They come with different types of skills. So architects and accountants have both very high level skills, but they have very different skills as well. Just like carpenters and bakers may have similar level of skills, but they cannot do each other's job very easily. So that's the first part. The second thing has to do with the fact that if you just think from the individual, you're missing an important part about skills. So why do carpenters and bakers have different skills? And why do architects and accountants have different skills? Well, the reason is that they specialize in something. So they took a long education and long training, long work experience to become good at a particular set of tasks. But now that we become specialized, we do not just gain the skill, we also lack a whole lot of other skills. And if we want to make something complex, then typically we don't have the skills anymore to do that. And we need the help of others who have skills that are not ones that we have, but that would be needed to make the product that we're trying to make in a team based setting. So nowadays, to achieve almost anything because we're so specialized, people will have to work in teams in which others know the things that we don't know how to do. But we still need to be done to produce a good or service. So, for instance, if you're a Mason and you want to build a house, you will also need electricians and architects. You might need fire inspectors. You need a whole lot of other people to help you out with that. So in this world, a specialist, what is actually important is not just the skills that you have yourselves, but the skill set you can mobilize through others who have different skills. So by working together, you can actually mobilize a whole lot of different skills and add them to your own to be able to produce the complex products and goods that modern societies produce on a regular basis. 

Chuck: So what are some of the major takeaways of your research on skills? 

Frank: Well, I think that the single most important takeaway in the paper is that although your own skills are important, they matter for your career and for the wage that you will earn. What is equally important is how well they match the skills of your coworkers. What the paper argues is that coworkers can actually do two different things. They can either provide the skills that would be useful to combine with yours, but that you do not have yourself. So those would be workers who compliment your skills. But coworkers can also provide skills that are very similar to the ones you have. And in that case, these coworkers would be your substitutes in your firm. Now, complimentary coworkers are typically good for your wages and good for your career prospects, whereas workers who can substitute you are not good for your career. They typically tend to be associated with lower wages and also with you leaving the establishment where you work sooner. So what the research stresses is that your coworkers should be different but relevant. So the skills that your coworkers have there should be different from yours, but they should still be relevant to the kind of things that you are trying to achieve. Now, this is actually this is related to something that others have pointed out about themes and skills, but it sheds a slightly different light on these issues. So on the one hand, people often think that the best place to go if you're a specialist is to go to another place with many of such specialists. So if you are a graphic designer, the best place to go is the place where they have a large number of graphic designers. Now, the problem with that is that in that place, there are a lot of people from whom you could learn undoubtedly, but there are also a lot of people who can do your job. So you are easily substituted by your coworkers. So that might be less than ideal. Instead, if you were to work in a firm that hires not just graphic designers, but also software developers, database managers and so on, in that place you would benefit from the fact that these other workers have skills that are useful when combined with your skills, but they cannot easily substitute for your skills. So that is actually much better environment. So another thing that people have often said about teams is that the diversity of people in a team makes the team more productive and more innovative. And I think that's true. It's just slightly more nuanced, according to this research, than just saying we need large set of diverse people with diverse backgrounds. Actually, what you need is complementary people. So complementary people are diverse. But they are also relevant to one another. 

Chuck: Were you able to determine the impact on wages for someone who's on a team with complimentary coworkers versus someone who's working with a bunch of workers who are substitutes? 

Frank: Yes, definitely so a lot of research has focused on the relation between wages and the complimentarity to coworkers. Now it turns out that if you have a lot of complimentary coworkers, you typically earn a higher wage and the increase in wage is comparable to the increase in wage if you go from a high school to a college degree, when it comes to substitute substitutes, do more or less the opposite. If you have a lot of substitutes around you, your wage typically is substantially lower. And as a consequence, you're also more likely to leave the establishment where you're working. 

Chuck: Can you elaborate on the data set and the networks that you constructed in the research? 

Frank: So the real challenge in the research was to figure out which workers are complements and which workers are substitutes for each other. There is no variable in a data set that tells you about this particular feature of your coworkers. You just know what a coworker has in terms of skills, but not what the relation of the skills are to your skills. So that is something we have to figure out. So the hardest thing in the research was finding out which skills are complementary and which skills are substitutes for one another. And to do so, we actually relied on a very large data set. In Sweden, we had administrative data on the entire population of Sweden and some 9 million individuals for about 10 years of their working life. And we knew not just the level of education ahead, but the exact type of educational degree they acquired, so we know that somebody has a post-secondary degree in accounting or college degree in architecture, post-secondary degree in food preparation and so on. The other thing that we knew about workers was the kind of job they held. So the occupation in which they were employed and the work establishments where they were employed. So we knew exactly what people did and with whom they worked. And that last part was particularly important because the idea behind measuring complementarity of coworkers, which educations would be useful if they are used together? Well, we just ask firms because firms are in the business of making useful teams. So if we ask a lot of firms, which people would you put together in one team, we will get an answer that will tell us about how complimentary coworkers are. So we didn't really ask firms. We just looked at what they did in practice. And we found that there are very strong patterns in the types of educational backgrounds that are hired in the same establishments. And by counting how often two educational backgrounds, so two educational degrees would be present in the workforce of one and the same establishment. We could get a sense of which educations are often combined by firms. Now, if you compare that against some random benchmark to see whether these core occurrences of educations are significant in a statistical sense, you get an idea of, what we call in the paper, which educations are synergistic. Now, the problem is that firms will not just hire complementary coworkers. So teams will very often also have duplicated workers. So if there's a lot of work for workers with a specific skills, you will need to hire more of those workers. So many of the people in your team will actually not be complements to you, but they will be your substitutes. They will know how to do the same tasks. So we also needed to figure out which educations allow you to carry out the same tasks. Now we have actually information about that because we know which occupations, people held and occupations are sort of bundles of tasks. So by asking which educations allow you to get into the same types of occupations we could back out, which educations prepare you to do the same tasks? So which educations are substitutes to one another? Now complimentarity is basically the access synergy. Given these substitutability of your coworkers, so what we ask this, how synergistic are you to your coworkers? How often do workers with these particular educations work together? Controlling for how similar these educations are? And then you see that there are certain skills that are very different, but still very often used together. And those are the skills that we're calling complements, given that you have nine million people in hundreds of thousands of establishments. You can actually get a pretty accurate measure of which educations are preparing workers to be substitutes for each other and which educations are preparing workers to become complements. 

Chuck: And you've constructed a tool which identifies right coworkers and wrong coworkers. 

Frank: Yes, so we visualize these educations in a network. These networks, they show all of the educational specializations you can get in Sweden - there are almost 500 different degrees. We connect two degrees in this network. If they are either very synergistic to each other or if they're very substitutable, one another. So there is a synergy network and there is a substitutability network. Now you can look up each education and find which of my most synergistic educations starting from my own education. 

Chuck: How does this research relate to, say, an urban or a large firm wage premium? 

Frank: So this was actually a surprising finding in the paper. It is a well-known fact that large cities pay higher wages than small cities. Boston pays a much higher wage. And then Worcester, which is near by and large, firms also pay higher wages than small firms. The explanations that people have been giving for that tend to be very different. For instance, in a large city, you would have a higher learning rate than in a small city because some more people from whom you can learn. People have said that large firms are just very productive and they share part of this productivity with their workers. There are a number of other explanations that have been given for this. But what this paper actually suggests is that complimentarity to coworkers could explain both of these premiums. So what happens if you move to a large city is that it's easier to find the right coworkers because a large city has many different workers with many different skills. So it's easier to find the team in which you would be the complementary coworker. Similarly, in a large firm, the division of labor can be much deeper. Meaning that people in a large firm can specialize much more on the tasks that they're really good at, but they can only do that if there are coworkers who do the other tasks. So if there are enough complementary coworkers and what we find in the paper is actually that if you control for the complimentarity to your coworkers, both the large plan premium and the urban wage premium become much less pronounced. So to a large extent, complimentarity explains why these premiums exist. But not only that, it turns out that both premiums are very dependent on having complementary coworkers. So the people who work in a large city but do not work with complementary coworkers, they get hardly any return to working in a large city so full of people have many complementary coworkers. The urban wage premium is about 9 percent, meaning that for every doubling of the city size, wages go up by 9 percent. If you compare that to people who work in the same large cities but do not find complimentary coworkers for then the urban wage premiums only 1 percent. Similarly, if you look at people who work in large establishments but then do not find complimentary coworkers, they tend to also not get a very high return to working in that large plant. 

Chuck: Does everyone benefit equally from complementary teams? 

Frank: No and that was another surprising finding. The returns for complimentarity are actually highly uneven and this was surprising because in principle, if I am complementary to you, you are complementary to me. That's the way we constructed these measures. So there was a built in symmetry in this relation of complementarity. However, it turns out that people with a high level of education have a much higher return to complimentarity than people with lower levels of education. So this suggests that the people who really benefit from working complementary teams are people with the highest levels of skills. So why would that be? Well, we can only speculate. But one plausible reason for this is that what complimentarity does is that it increases the productivity of the team as a whole. So it basically increases the amount of output a team can produce, but it doesn't tell us anything about how this extra output will be divided. We know that complimentarity increases the size of the pie, but we do not know who will get which share of this pie. 

Chuck: So how does this research fit into what you're doing at the growth lab and what might be next? 

Frank: Well, that's an excellent question. So the reason why we looked at this complementarity of teams is that we are very interested in understanding how growth works. So the way we have been thinking about why some places grow faster than others is that some places just have more capabilities than others. So let me explain that a bit. The way we think about economic production is that if you want to produce something, let's say a car, you needs not just capital labor, but you need a whole lot of different types of capability. So you needs dedicated machines. You might need research institutes, but you also need a whole lot of different types of workers. So you need car engineers, you need car designers, you need people who do the marketing of the car. You need the accountants taking care of the bookkeeping in the home and so on. So, so far, we have mostly looked at what that means for economic growth. And they are. The idea is that for countries, but also for cities or firms, it's easiest to move into new products if these products require many of the same capabilities to the ones that you have already. So if you want to diversify as a country, it's easier to use the capabilities that are embedded in your current workforce than to train workers to get new capabilities. And that's why we see that most countries, but also cities and regions and firms, they diversify in a very predictable way. They diversify into new activities at a very late to the old ones. If they want to diversify into something that is less related, they typically have to get capabilities from elsewhere. So that explains why we were very interested in looking at migration and how firms invest in foreign markets, because very often some firms invest in foreign market, they do not just bring capital, they also bring a team of specialists to this new location that allows this new location to diversify into things that without the help of these foreign firms that couldn't have done. So the question that we had an answer is, is it enough to have just a lot of different capabilities? Does the coordination of these capabilities into teams, does that happen automatically? So the missing piece of the puzzle this research provides is that workers themselves become very dependent on each other in very complex economies, because these complex economies, they organize they coordinate the capabilities that are distributed across a lot of different workers into teams that then can produce high value products. Now, this got us actually very interested in the question of coordination. And how economies historically and also nowadays have managed to coordinate workers and teams of workers into networks that produce output. So currently we're using this line of research to explore how the US actually managed to become the frontier of technological progress and the economic powerhouse out of this. Nowadays, where the organization mechanisms could be funded, but they could also be cities. We are also very interested in thinking about the future of work along those lines because what the future of work people have looked at, the future of work in terms of which skills are going to be replaced by machines. But this is not the only thing that happens when a technology changes how we do our work. It actually also changes how we can coordinate different skills in different teams. So we're also trying to understand how jobs have changed throughout history and how different skills are being combined in different ways over the past 60, 70 years or so. 

Chuck: Thanks, Frank. Fascinating research. 

Frank: Thanks Chuck. We should do this again. 

Katya: If you want to learn more about the Growth Lab's latest research and events, please visit See you next week.