Podcast

Important Information:
Not All Career Ladders Are Created Equal
Important Information:
Katarína Borovičková and Claudia Macaluso describe their research on differences in wage growth between workers as they progress in their careers. Borovičková and Macaluso are economists at the Federal Reserve Bank of Richmond.
Related Links
Transcript
Tim Sablik: My guests today are Katka Borovičková and Claudia Macaluso, both economists in the Research department at the Richmond Fed. Katka and Claudia, welcome to the show.
Katarina (Katka) Borovičková: Thank you very much for having us.
Claudia Macaluso: It's a pleasure to be here.
Sablik: Today we're going to be discussing your recent research looking at how career changes affect worker earnings. You recently published an Economic Brief article that discusses a concept that I think should be familiar to a lot of people, which is the job ladder.
The conventional wisdom has long been that workers progress in their careers by moving from lower paying jobs to higher paying ones, either at the same company or by switching employers; in other words, by climbing the corporate ladder step by step.
Labor economists have studied this idea for decades and we'd need a much longer show to get into all of that history. Could you briefly share some of the evidence that economists have accumulated over the years that supports the job ladder theory of wage growth?
Macaluso: It's very hard to explain differences that we see in the cross-section of wages in the way we observe them. We do observe some things about workers, say their age, their gender, their education. These things go some way to explain how much they earn, but not all the way. In fact, we would expect these kind of factors to explain much more.
The job ladder framework really came about to try to explain these differences. Maybe they climbed more steps. They climbed them faster or slower. So, it was essentially born for explaining unexplained wage growth.
Borovičková: The idea of the job ladder has been supported pretty well with the empirical data. The first paper goes back to a seminal paper of Topel and Ward from 1992. We use longitudinal data where you can observe workers over time for several years, then you can study what these workers have been doing. Workers have changed jobs quite frequently, especially in their early careers. And why they changed the job was that they actually got wage growth.
Macaluso: When we look back at the literature, there's very robust evidence that this mechanism is in place for young workers — this idea that you change jobs to get a better wage, a better pay. There's some work by Derek Neal, for example, in the mid '90s showing that fairly clearly with data from the United States.
Sablik: One thing you mentioned in your Economic Brief article is that there's an open question about the job ladder theory as to whether it accurately describes the experience of all workers. Claudia, what did you find in your paper when you examine the lifetime wage growth for different types of workers?
Macaluso: In a nutshell, we found that the job ladder is not the same for all workers.
When I look at people at the peak of their career, 50-55, some of them have done very well — they sit at the top of the earnings distribution. Some not so much — they sit at the bottom. So, I see rich workers, median workers, [and] poor workers in terms of their earnings. And I'm going to look back and try to understand which kind of jobs, which kind of employment transitions happened to get you where you are at the age of 55.
When we do this exercise and we plot wage growth going backward, we see that wages grow much faster for rich workers than they do for poor workers. Now this may sound obvious. This could have been completely flat. It could have been that you are at the top at 55 and were at the top when you started at 20. But this is not what we see. The dispersion at young ages is much smaller.
The steps that people climb or not are really important, and that's exactly what we wanted to study: this mobility, these steps up or down or steps to nowhere.
Sablik: Katka, maybe you can get into a little bit more about what you found when you looked at career mobility, or that movement along the job ladder for these different workers.
Borovičková: What we find is that the way these different workers move along the ladder is very different.
The first difference is how quickly people move through the ladders. How often do people make the step to a different rung? We find that the people at the bottom of the earnings distribution change jobs relatively frequently, even more so than the people at the top of the distribution, which might sound surprising in the view of the idea that every step to a different rung is a step upward.
We find that workers at the bottom of the distribution move through the steps frequently, but the steps they are making are very different. The ladder of the guys in the top of the distribution is upward sloping, is leaning against the wall. So, every time they make a step, it's a step upward. The ladder of the workers at the bottom of the distribution is lying on the ground. They do make steps through the ladder, but this ladder is not bringing them to the better jobs. This is the key difference in how the career is progressing for workers at the bottom of the distribution versus the workers at the top of the distribution.
Sablik: So, if changing jobs doesn't necessarily lead to higher wages, what are some other factors that could explain what you're seeing in the data?
Macaluso: That's exactly the question that we want to ask. It is a puzzle.
As Katka was saying, workers at the bottom of the earnings distribution move across jobs, but they're going nowhere. Their ladder is flat on the ground. So, we want to understand why is the ladder that way? To do that, we set up what we call a structure model, which is sort of a little lab in which we can model the economy and move the little pieces, just like you would in a lab experiment.
In this model, we equip workers with different characteristics. They have some version of baseline ability, which we call human capital. They also have different abilities to learn and to accumulate even more of that human capital. You may think of cognitive skills or manual skills, the kind of skills that would help in production of objects or in the carrying out of one's job.
There's also another component of human capital, which we think of as non-cognitive skills. Imagine that this is a proxy for reliability or dependability. We think that these really approximate well the separation rate: how likely is a person — conditional on their other abilities — to keep a job, to keep being employed for a long time? It turns out that even though it is important that workers are able to take advantage of learning opportunities in their job, none of this learning can happen if workers don't manage to stay in the job very long.
Borovičková: So, maybe to provide a little bit more details about how we come up with this conclusion that Claudia was mentioning, one experiment we are doing is to think about what if all workers are the same, either in terms of their ability to learn or in terms of their ability to keep the job.
We found that just shutting down the differences in the learning ability closes the gap between the top and the bottom workers by about half. So, 50 percent of the differences that we have measured in the data could be attributed, according to our model, to the differences in the learning. We conduct a similar experiment, but now shutting down the differences in the ability of keeping jobs. So, if we say all workers have the same ability of staying in the same job for a specific amount of time, then we find even more than 50 percent of the gap would have been closed.
Sablik: Have any other researchers found similar differences in workers' career paths?
Borovičková: There is a recent paper by Ozkan, Song and Karahan that also studies the heterogeneity in the job ladders. What they conclude is that people at the bottom part of the distribution are there because they have very high risk of separating to unemployment. So, they are not able to stay in a job for a long time and learn, and they frequently move through employment and unemployment. At the top part of the earnings distribution, they say the differences are mostly attributed to the differences in the learning abilities.
Macaluso: What we really want to stress is, great, so now we know what the source is. It's a combination of learning ability and the possibility of staying in a job for a sustained amount of time. Now what? Now what do we do? In our paper, we try to think about that very explicitly by adding one extra ingredient. We allow workplaces to be different in the kind of opportunities that they offer to accumulate human capital, to learn new things.
I think this resonates with the experience of most people. In some jobs, it's all very routine. There aren't many new things that sort of pop up on your desk. You use the skills you have, you do your job, you go home. That's it. But in other jobs, you are constantly bombarded with new things, new technologies: robots, AI, automation, different software. And those are both challenges, of course, but there are also opportunities to learn.
We want to take these ideas seriously and see if learning on the job, coupled with those other differences that we were talking about, can make a difference. We find some role for it, but most importantly we want to think about it for policy interventions.
Sablik: That's a great segue. You mentioned there's things that employers can do. What are the potential things that policymakers can do, based on your findings?
Macaluso: We were partly motivated by a set of policies that are already in place. Sometimes they are referred to as "good match" policies.
The idea is that some people are high ability but unlucky. It happens to the best of us — you get a bad shock and things get on a bad path for you. Is there a chance to redress that path? Could we provide a match with a good job to workers who so far have been unlucky in terms of their wages?
This is, in fact, a policy that exists. Some of these take the form of apprenticeships, which is exactly matching young workers with what is seen as a good job opportunity for them. Some of them take the form of retraining after a layoff episode. Some of them are grants to match underrepresented degrees with firms that provide high earnings growth.
In our model, we investigate these types of policies. In particular, there's one exercise that I find very interesting. We provide to a worker at the age of 25, one of those that are unlucky so they end up poor at that point in their life, we give them the chance to meet a firm that has high wage growth. We show that this can improve the earnings. This "good employer" effect is much larger if there is some stability to the match, which we think speaks directly to the kind of policies that the Department of Labor, for example, has in place right now.
Sablik: Katka, what are your next steps with this research?
Borovičková: We think that the model in this paper is very useful for answering some other questions.
One thing that they would like to address is a longstanding question in economics which asks, "How much of the differences that we observe are due to initial conditions versus luck?"
The initial conditions [are] everything that has happened to you before you start, before you enter the labor market. This could be the type of college you have attended, the type of family background you were raised in.
Luck is what happens to you just by chance. For example, there are two graduates from the same college with the same degree. When one of them goes on the labor market, there is a job opening at Google. But when the other goes out on the labor market, no such opening exists. The first one is lucky to get the job at Google and his career can start off very well. The other one was just unlucky, not because he would not be qualified for the job — he was as qualified as the first candidate. That vacancy just was not open at the time. Sometimes you just are at the right place at the right time but sometimes you are not.
Sablik: Super interesting. Katka and Claudia, thanks so much for joining me today to discuss this work.
Borovičková: Thank you very much for having us.
Macaluso: Yeah, thanks so much for having us.