Raffaella Sadun is a Professor of Business Administration at Harvard Business School and has co-founded several large-scale projects to measure management practices and managerial behavior in organizations. She spoke with SSRN about the influence of management on productivity in organizations and how evolving technologies, such as AI, fit into that equation.
Q: You’ve done a lot of different work studying how management and organizational factors contribute to the productivity of corporations. What drew you to this to begin with, and how has that work continued to evolve with the changing times?
A: I started this line of work in 2003, when I was a PhD student. I was very interested in firm country-level growth from a macroeconomics perspective and understanding why it is that in some countries, the productivity grows. In the long run, this influences almost everything. When I was growing up as a researcher, you could, for the first time, go from this macro-level data on GDP and total factor productivity to very granular information on firms. It was the early stages, not the very beginning. There had already been some work using the U.S. Census, but… for the first time, you could really appreciate how individual firms contributed to macroeconomic growth. I thought that this was just the coolest thing ever that you could do.
Then it became even cooler, because I did my PhD studies right at the point in which you could do even more than looking at total factor productivity of a firm: you could open the black box of a firm and understand how the firm was managed. All this work on management and the organization – it’s called the World Management Survey – it’s really an attempt of going from just outcome measure to measures like productivity and total factor productivity, profits and so forth, to managerial inputs. This was getting a sense of how resources were allocated inside the firm but also how the firm recognizes if there are problems, how they solve problems, whether they have targets, and more importantly, how they manage people. Do they have a way of understanding who is a good performer and who is not, and how do we help people grow or learn new things inside the firm?
The connection then became even more interesting, because you went from macro to micro at the firm level, getting all this internal, soft stuff but measured quantitatively in a rigorous way, across many thousands of organizations. This was just a blast, and continues to be a blast for me, because you can understand much more about the world using this approach.
Q: What are the challenges to collecting this kind of data, especially at such a large scale, looking at so many different countries?
A: It’s a managerial challenge. I work on this with a large team of people, and I think that we did drink our own medicine. The primary challenge really starts with the design of the survey and understanding if you’re measuring something that is really an input, first of all, and also whether it’s something that you can measure with some precision. From a quantitative perspective, you don’t want to get stuff that is too trivial, so that everybody does it, or too hard, so that nobody can do it. You want to make sure that you’re measuring something where there is a good chance of variation, but also something that can be measured with enough precision and quality that these comparisons across and within countries are meaningful.
The way in which we approached this was to hire people – in the first waves of this research, they were MBA students who had some exposure to management education – and we would work with them throughout the summer to help them understand what we wanted to measure. We had our own management systems, where every three analysts would have a manager, and they would talk with the managers to debrief after every interview. Every interview was listened to by two people, so that it could be calibrated. We had monthly check-ins as the data was being collected. We were really trying to create standards for data collection that would allow us to then scale this measurement across many thousands of organizations.
Q: In collecting data, is there anything that has been particularly surprising to you, either about the process itself or the results the data has yielded?
A: This is a line of research that is infinite. One of the interesting things that continues to surprise me today is the fact that even when this sort of managerial know-how should be trivially adopted and known by everybody, to this date, we see this variation in adoption of basic management practices. This can be surprising, because we know that these things are correlated, in some studies done by my co-authors, also causally related to performance at the firm level.
Eventually, this builds to macroeconomic differences. This is so interesting because it tells you something about the frictions in knowledge diffusion. The same way we know that there are diffusions of new techniques, new technologies, new ways of doing things, there are also the frictions in the diffusion of this managerial technology and understanding what shapes it: whether it’s an issue of information, whether it’s incentives, or whether it’s culture, both inside and outside the firm. I think this is the best thing to study, because it’s a connection between economics, management, sociology, and history.
Q: You’ve written and contributed to many papers on SSRN related to these subjects of management and productivity. You also have dived into how technology contributes to these factors, and most recently, the effects of AI. One of the recent, highly downloaded papers that you are an author on is “The Cybernetic Teammate: A Field Experiment on Generative AI Reshaping Teamwork and Expertise,” which explores how AI fits into the collaborative work environments of organizations. How do the results of this study fit within the idea of AI as empowering workers, rather than replacing them?
A: This is a very salient problem, this issue of whether AI is a complement or a substitute. This is a line of work that I think is really needed, because at this point, there is a lot of anxiety about the impact of AI. I think that in many technology circles, there is almost a foregone assumption that, of course this is going to be bad for people. There are lots of discussions of entry level jobs that are going to disappear. I think that getting a very good sense of what these technologies can actually do is critical for us not to be panicking about it, and also for us to be able to shape how this technology is adopted.
There are studies that show this technology is very good at balancing the knowledge of less experienced workers with more experienced workers. For example, there is a beautiful paper recently published on the Quarterly Journal of Economics that has shown that the rollout of a copilot – something that allows call center agents to answer questions and gives them suggestions on what to tell angry customers – has a strong impact on the productivity of the agents, especially the agents with less experience. This paper has a similar flavor, because it tells you that the copilot helps experts: not call center agents, but marketing or R&D experts. In a way, it’s similar, because it helps these experts bridge across silos of different expertise when you have to work on an innovation, typically it’s a combination of R&D and marketing work. That’s why they have teams, because they need this different knowledge and expertise to come together. The paper shows that an R&D person with a copilot, or a marketing person with a copilot will be able to deliver innovation ideas that are perceived to be of the same quality as two human experts. This is helping you bridge across knowledge silos. You’re going across these different domains of expertise, which could be really powerful for the experts themselves.
Q: In a previous interview, you mentioned that you think one of the biggest misconceptions about AI is that it’s just about the technology. Could you talk a little bit more about this idea?
A: This is regarding AI, and you could use exactly the same line of arguments that I am going to use for previous technological waves: we could talk about PCs, software, or even electricity, if you wanted to, depending on how far back you go. Technologists will talk about just the technology. Organizational economists, or people that are more grounded in firms, will talk about the ways in which the technology is adopted inside an organization. You can have an incredible technology at your fingertips, but if it’s not well-integrated with the organization, people will not use it or they will not use it well.
What you want to understand is how to integrate the new innovations with your production systems, your skills, the ways in which people do their job every day. With AI, there are particular skills that are very human and organizational that will become critical for how this technology is used. The first one is the ability to know your limits. That’s a very human skill, because potentially, these are technologies that give us access to expertise at a fraction of the cost. It’s like the tacit expertise that was embedded in humans before – now it’s embedded in a machine. You need to have judgment to understand the quality of that expertise, and you also need to have the humanity to ask questions and really make good use of this gift that you now have.
The second one is the ability to adapt. This is such an early stage for this technology that the use cases have not yet been completely discovered. So, you will see the hype cycle, where every day there is a new application, something that sounds incredible, but at the end of the day, how this technology will impact firms will depend on how each individual firm will be able to use that technology to achieve their goals or their strategy. Everything is firm-specific. This is the time where you need people who are willing to adapt and who are willing to explore and understand how to experiment with this technology and find new sources of value creation. These are not things that can be automated. You need people to be at the front line.
Q: What do you think holds back some organizations from experimenting and adapting?
A: I think that there are humane and rational factors. There is a lot of uncertainty still on exactly how the technology is going to be useful to individual cases. It’s something that you would say, “Well, maybe it’s better to wait and see, let other people figure it out, and then I jump in.” This is, I think, very dangerous right now, because I expect that a lot of these applications are going to be firm-specific. If you don’t do it, nobody else will do it for you.
Another point, which is real for this wave of technology, and was probably not there in previous waves, is fear of being replaced. This is precisely because AI is able to provide answers, and in some tasks, the answers are much better and more reliable than the experts. I’m not sure that everybody likes that, because that makes you feel irrelevant. The key there is to understand the value of judgment. It’s like having a very smart person in the room that needs to be managed, and you still have a role as a human.
Q: A lot of the work you do is very timely and relevant, so instead of asking about all of it, I’d love to know: are there any of your papers or ongoing works – on SSRN or otherwise – that you’d like to highlight?
A: I am really passionate about a new line of work on training and reskilling, which is related to the adoption of AI. I have a new working paper called “Training Within Firms” that I love because it goes inside three large organizations, and it asks the question: why is it that often workers are offered opportunities for learning and training, and we see that there is low, or very heterogeneous, take up across different parts of the firm? The bottom line from this paper is the importance of the middle manager, which is the person whom the workers report to, who’s not really at the top of the company, but is closest to the frontline. The paper shows that there is a real difference in the extent to which middle managers can be coaches and mentors for frontline employees, and these differences matter for both training take up as well as the performance of firms when there is stress in the system, when we use some shocks to simulate these moments of heightened production.
I think that’s important, because it tells you that firms still need humans. Learning as an adult is tough, and you need a person to motivate others. You need the person that is able to communicate the commitment of the firm to employees. It’s not enough to say it from the top: you really need people on the ground that are able to be credible in investing in people’s skills and careers.
Q: How do you view SSRN’s place in scholarship right now?
A: I think the fact that you have this very large repository where the latest cutting-edge papers are published is super important right now. I expect that, with the advance of technology that allows us to make good use of all this knowledge, summarize it for us, and help us understand how useful it is, it will be increasingly important to have this big repository of knowledge.
More About Raffaella Sadun
Raffaella Sadun is Charles E. Wilson Professor of Business Administration at Harvard Business School, a Co-Chair of Harvard Business School’s Project on Managing the Future of Work, and co-PI of the Digital Reskilling Lab. Her research focuses on managerial and organizational drivers of productivity and growth in corporations and the public sector. She co-founded several large-scale projects to measure management practices and managerial behavior in organizations, such as the World Management Survey, the Executive Time Use Study, and the first large scale management survey in hospitals, MOPS-H, conducted in partnership with the US Census Bureau. Sadun currently co-leads the Digital Reskilling Lab at HBS and also serves as director of the of the National Bureau of Economic Research Working Group in Organizational Economics, and is faculty co-chair of the Harvard Project on the Workforce. Sadun has authored articles published in well-known journals and was recognized for the best article published in the Harvard Business Review in 2018 and 2023. She received the honor of Grande Ufficiale dell’Ordine “Al Merito della Repubblica Italiana,” the highest-ranking order awarded by the President of the Italian Republic for “merit acquired by the nation” in 2021. In 2022 she was awarded the Prize “Fondazione de Sanctis per le Scienze Economiche.”
You can see more work by Raffaella Sadun on her SSRN Author page here