At UCLA Anderson, the “Forecast” does not refer to the weather. Rather, it refers to the venerable UCLA Anderson Forecast, the school’s research group that has produced lower-case “f” forecasts for the U.S. and California economies for nearly seven decades. The Forecast is widely read and, more important, widely respected for its independence and accuracy; dozens of private- and public-sector entities rely on its quarterly releases for their own planning and analysis. The center has recently undergone some personnel changes, with longtime national forecaster David Shulman’s retirement in June 2020 — not to mention the additions of economist Leila Bengali last year and senior economist Leo Feler last summer, the pair joining director Jerry Nickelsburg, economist William Yu and Professor Emeritus Ed Leamer on the research side.
Q: The UCLA Anderson Forecast has undergone a few changes in the past year or so, with longtime national forecaster and even longer-time friend of the Forecast David Shulman stepping away. Jerry, could you please introduce the newest members of the team, who join you and William Yu?
Jerry Nickelsburg: Over the past year we’ve added two outstanding economists to give more depth to the Forecast: Leila Bengali and Leo Feler. David Shulman retired, so we needed additional firepower for the U.S. forecast. But we made the additions also because we are increasingly reaching out across Anderson to other centers and academic units and across the university to bring in content that relates the macro-forecast to the changing face of business. With Leo and Leila on board, it gives us the depth by having at least two members of the Forecast working side by side on each and every forecast. In the past, David (or director emeritus Ed Leamer) would write the national forecast and I would write the California forecast. Now, I’ll team with both Leo and Leila on the U.S. and California. The additions also allow us to drill down regionally within California and to some regions outside of California.
Leila Bengali: My forecast focus is the California economy, and I’m also recalculating the model that we use to create the California forecast. I’m starting to take over more of the work that we’re doing in the Bay Area, with the goal of eventually expanding our footprint there. We have a smaller model of the Bay Area economy and we also work with a number of local government agencies there to do forecast reports or projections and analysis.
JN: Leila, why don’t you explain your approach to economics and behavioral economics and how that can influence the Forecast in a positive way?
LB: My research in graduate school centered on behavioral economics, which is a field that combines insights from psychology and economics. I’ve never heard of a subtopic called “macro-behavioral” yet, but I think there are a number of interesting avenues here. To have an understanding of when people are likely to make decisions that maybe aren’t what you’d predict just from a classical economic perspective might help inform the art of forecasting, where you have your models, but you also have to think about what’s driving individuals to make decisions.
Q: Leo, since you’re the newest member of the team, could you recap where you were last and how it feels to join the Forecast?
Leo Feler: I did my Ph.D. at Brown and specialized in urban and labor economics. From there, I joined Johns Hopkins as an assistant professor, teaching courses in applied econometrics, banking and finance. I then transitioned to Boston Consulting Group and did a series of projects on consumer behavior.
We were working with a retailer that has 9,000 pharmacies, and we would actually be able to run randomized control trials. The wealth of data was phenomenal. We could see whether patients were diabetic, based on the medications they would take — and then whether or not they would purchase Gummy Bears at the checkout counter, and then what kind of interventions could we give these individuals to ensure that they were well informed about the interactions with the behaviors and the medications they were taking.
I met the Forecast team a few years back and was fascinated by the work that they do, the policy implications of the Forecast, and the conversations that the Forecast facilitates. That we get to engage in and help shape and then help communicate those ideas to a broader audience — I was really enamored by the idea of being able to partake in that.
Q: The Forecast is one of the highest profile units on campus. Are you prepared to have your work digested by such a large audience?
LF: I’m very excited about that. Actually, it’s one of the reasons I wanted to be part of the Forecast. It gives you this platform, in part because of what you’re writing and then in part because it’s UCLA Anderson.
LB: The opportunity and privilege to be able to speak with authority to the broader public is something that I value and take seriously. I want to make sure that what I am saying and conveying is the truth to the best of my ability to understand it.
Q: How does the economic forecast come together?
JN: The nature of forecasting is, you begin with the most recent data, as complete as you can find. You compare it to past circumstances that are similar, and a forecast basically says if the future is like the past, then this is what we expect.
The next step in that process is to look at how today is different from the past. Now we’re moving from that statistical [information], taking the past and extrapolating the data to the future. But we’re asking, how is today different from the past? This is where the art of forecasting comes in: inferring how a forecast should differ from what the statistical model just spits out.
LF: We start with a baseline forecast and we adjust the inputs and assumptions to reflect our beliefs about how the economy will evolve. This is where the art comes in. The science encompasses these equations that exist and the relationships between different inputs and outputs.
We look at both historical trends and brand new data that’s coming out, and we try to think about how we need to adjust based on our perception of what the future will be like. We change some of the inputs of the model and rerun it to see what numbers it will give us and whether or not those numbers are a sensible expectation, given the news and information that we have available.
Q: Does it make you uneasy when your forecast stands apart from the “consensus”?
JN: When we look at what others are saying, we want to understand whether they are using simply the output from the model, or do they have reasoning we ought to be thoughtful about? Internally is where we do have to come to a consensus on all of our forecasts. We listen to each other and we try and think about to what extent we agree. And to what extent is this simply our personal view, or is it where the data lead us? We have some spirited discussions and they bring us to a forecast that we, individually and collectively, say, yes, this seems like the likeliest outcome, even though individually we might think it’s a little too optimistic or a little too pessimistic.
Ed Leamer used to say years ago that he would rather have the story correct and the forecast end up being wrong than have the forecast numbers correct but not comporting with economic theory. So the important thing here as we look at the U.S. and the California forecasts is to be able to trace the “why” of the forecast by using good, solid economics.
Q: Leo and Leila, could you have picked a crazier time to start forecasting here at Anderson with the pandemic, the unique economic conditions and the election?
LF: The nice thing is that now everyone is really paying attention to this. I get to come in at a moment when there will be both heightened scrutiny and heightened interest, and, I think, a lot of understanding even if, as Jerry mentioned, our ideas are correct but our numbers aren’t exactly there. It’s an exciting time to be coming in.
LB: We aim to be helpful. If you’re redesigning a model at a time when there are extremes, as is the case now, it’s probably a good time to test your new model. If your model produces good predictions when the economy experiences large fluctuations, then it’s probably going to work well during more normal times.