A constant among variations
Bruce Hansen, Professor of Economics, University of Wisconsin, Madison
Some people believe that economics is not a true science, because it is impossible to do controlled experiments and the results of any experiment cannot be reproduced, because there are too many variables.
This does not mean that economists cannot make a worthwhile contribution to knowledge and help to improve the world for everyone.
It also does not mean that they do not have any data; in fact, the world is awash with economic data. If economists can make sense of it, they can advise policy-makers on ways to intervene and protect us from the slings and arrows of outrageous fortune or, if not, then maybe make us materially better off.
The problem is that it is extremely difficult to sort the wheat from the economic chaff. It is well known that you can correlate things where there is no causal relationship.
For example, there is a positive correlation between purchases of ice cream and crime levels (because crime levels go up in hot weather). In a world so full of variables, economists need tools to tell them whether what they are observing and drawing conclusions from is real or not.
This is where Professor Bruce Hansen comes in. He is a mathematician and Professor of Economics at the University of Wisconsin in Madison. He has created statistical analysis tools that are used by the top economic theorists, and is one of the most cited economists in the world. While he may not be a household name outside the world of economics, his work is used by people who advise the major world banks and government institutions, whose decisions have a drastic effect on our lives.
His head of department, Professor Ken West, says that while Hansen is extremely modest, he is in fact one of the leading researchers in his field. He has made a major contribution to the development of many of the most important economic theories. West adds that he is an extremely good teacher, very popular with students and the department is extremely happy to have him.
He says: ‘Bruce is widely known for his work on non-linear time-series. He has developed and applied techniques for a class of models known as threshold autoregressions, in which the response equation is a potentially discontinuous function of the response variables.
This class of models is widely popular among applied economists, but much of this research was quite undisciplined. Bruce developed rigorous statistical methods to test for the presence of nonlinear (threshold) effects, for estimation of these models, and for accurate reporting of estimation uncertainty. His methods are now quite popular in this area.’
Hansen was born in Los Angeles, California. His father is a retired engineer and his mother a home-maker. He was particularly good at mathematics at school, but it seemed a little dry to him. He says: ‘I didn’t have to study hard at school; I sailed through school, because I found it easy. I was excelling at mathematics, but I was bored by it because it was taught as this dry stuff that only teachers understand. So when it came to college I decided I would not study mathematics; I wanted to change the world. I went to Occidental College in Los Angeles and got involved in radical politics. Half my time was spent studying philosophy and half studying economics. I ended up with a degree in economics, because I started talking to the faculty and they told me I had rocks in my head. There were very few openings in philosophy and I was not going to be one of the five people a year who got positions in that field.
‘I don’t regret studying philosophy… it teaches you clear deductive thinking. I was very politically active; at that time we were still worried about nuclear weapons. The year I went to college was when Reagan was elected. I was involved with the leftist leaning social movements on the campus, but nothing very serious.’He graduated summa cum laude even though he was still cruising. He worked for a year in a private sector bank and hated it so much he decided he never wanted to work in the private sector again. He applied to graduate school and took some evening courses in mathematics to brush up.
‘The people at Yale admitted me on a lark; I had good grade scores, but they admitted me at the bottom end. I had to really knuckle down.’
He started developing an interest in econometrics, where economics meets statistics. His thesis adviser, theorist Peter Philips, was a very dynamic figure who had a strong influence on Hansen. He worked with Philips on co-integration, which had first been proposed by Clive Granger, but was being developed at Yale. This looks at measurements that appear to track each other, but where there is no obvious causal link between them. They worked out methods of estimating models for various measurements that seem to have this cointegration.
He says: ‘These methods allow you to distinguish purely spurious trends. Some things track each other, like the decline of piracy and the rise in global temperature, but the two are clearly not connected. If you apply these statistical techniques, you can disentangle things that are genuinely connected from those that just appear to follow each other.
‘My research has wandered all over the place since then. My first job was at the University of Rochester, where I worked with Adrian Pagan. He was on the more applied end of econometrics and he pushed me to always think about why you do what you do – not to think about the mathematics, but to think about why we should care.
He kept hitting me over the head with this for three years. Between these two, Philips and Pagan, I created my personal philosophy as to how I did my research. Firstly, always employ a high degree of mathematical rigour and theoretical excellence, but at the same time always focus on what is important – why would economists care? I make sure that my research is always relevant to what economists need as well as having a high mathematical content.’
His research has been widely used by applied economists, ranging from those working inside banks to the Federal Reserve and World Bank. Although it appears esoteric to the outside world, it is all about giving those applied economists a better tool to view the data and decide what is real.
He stayed at Rochester for about five years. During this time he met his wife, who was a PhD student. She eventually needed a job and took one in Boston, so Hansen had to find a job in Boston. He contacted Boston College and they offered him a full professorship.Things did not work out there for his wife and so they decided to move together to Wisconsin, where they both work in the Economics Department.
He says: ‘Boston was a wonderful environment, because there are so many colleges there and there is a lot of interaction. I used to go to seminars at Harvard and MIT regularly. People came to these from all over the area. Boston is a very rich academic environment; it’s like having Oxford and Cambridge in the same town.’
Hansen found Wisconsin a better research base than Boston College, because it attracts very good graduate students. It did not faze him at all moving from the cosmopolitan atmosphere of Boston to the mid-west.
He says: ‘From a lifestyle point of view we were sceptical. But now that we are here we cannot imagine going back. Madison is a city of 200,000 people and the county has about 400,000 people; it sounds very small, but it’s upscale. It’s intellectual and a fairly well-off area. It’s very liberal and so the activities of the people here are conducive to a good lifestyle. In contrast, Rochester, New York has twice as many people, but about half as much stuff to do, because it’s more of a working class area. Madison is a very good place to live for someone who has a reasonable income and likes cultural pursuits.
It’s also a very inexpensive place to live. Affordability in money and time is great; you can get almost anywhere in town in 15 minutes. One of the beauties of this town is that while the university has a huge presence, it is also the state capital and there are a lot of other things going on in the town. So it’s actually a very nice place, and Chicago is two and a half hours’ drive.’
Hansen took over the position of Arthur Goldberger, who was a legendary econometrician, and the department told him they were looking to rebuild after some setbacks. As a state university, it has some challenges at budget time, but Hansen has found it a good fit.
His research has changed over the years. He spent a lot of time looking at threshold models. A common illustration of these is to look at whether companies behave differently if they are financially constrained against companies that are not constrained.
You then define the level of constraint by whether or not certain parameters are above or below certain thresholds (for example, debt to equity ratio). He has also done work on modelling discontinuities; for example, the sudden rise in unemployment during a downturn in the economy. He says: ‘My work has been trying to find ways to estimate and test models of such discontinuities. I got motivated to do this because I saw people estimating these models and I said “what nonsense”. They have no evidence that there are such discontinuities, so let’s be rigorous about it, rather than use these fancy models until we have evidence that there are such discontinuities.
Gradually it became a kind of advocacy – here are the methods and here are the ways that you use them – because people wanted methods to estimate these kinds of models. That has had quite a big impact.’
Hansen develops his own code to test models on simulated data sets and freely publishes it to the academic community, something which has made him very popular with colleagues. It is open source and he develops his programmes in an interactive way with the users who give him feedback on its application to particular datasets.
Although it is based on economics, it is actually very abstract statistical theory that happens to be useful in economics. His work has brought a great deal of mathematical rigour to the work of leading applied economists.
For example, there is the theory of ‘The Great Moderation’, which postulates a structural change in the business cycle in the early 1980s, something which has caused a lot of excitement among economists. The statistical work that lies behind this theory has relied heavily on the tools that Hansen created to give it a firm statistical foundation.
He says: ‘My recent work has been on averaging and combination of models. Forecasters use models and they make different forecasts; various sources report a combination of different forecasts, which turns out to be a good thing to do, but the question arises as to what weight should be assigned to particular forecasts. There has been a lot of interest in using Bayesian methods in combination. I thought a lot of the work had been non-rigorous so I set about trying to do better. The question is how to pick the weights, but I have found out something about how to pick the weights according to particular criteria. It turns out to be more complicated than just basing weights on what has happened in the past.’
Hansen realises that he could have become extremely rich if he had ever decided to become one of the ‘rocket scientists’ of the Wall Street hedge funds. But he is actually quite happy doing what he is doing right now. Wisconsin has worked well for him and while he does not say he will stay there forever, so long as there is work to do and good quality research students to help, he does not feel in any hurry to get back to the bright lights of the big city.