SECTOR

Econometrics: What has changed?

Just de Groot, Denise van Dijk
06 January 2020
Interview

An interview with Zanders co-founder Rob Naber

In today's world, econometrics can be found in every facet of our lives. However, this wasn't always the case, as the field has changed tremendously over the past half century. We interviewed Rob Naber of Zanders advisory to find out more. Rob Naber studied econometrics at the University of Amsterdam and graduated in 1988. He started his career in sales at ABN AMRO banking, later becoming one of the first quantitative risk managers at the dealingroom of that bank, before founding Zanders alongside Chris Zanders, and has been a prominent figure in the treasury and risk industry for the last 25 years. 

What was your experience studying econometrics?

What was the study like?

My study year finally consisted of roughly 12 guys and 1 lady. The study I did was called “Econometrics, Operations Research and Actuarial Sciences”. 

I started in 1981 and at that time the old study scheme (kandidaats/doctoraal) was still in place. In 1982 it changed into what we called the “Tweefasen structuur”, the predecessor of the BAMA structure. My study program comprised of 3 years “kandidaats” (or what we now call a bachelor degree) and roughly 3 years what we called “doctoraal” (what we now call master). So the full nominal study was roughly a 6 year program. At that time at the UvA, econometrics was an “inter-faculty” of the mathematical and economical sciences. This implied that our program was nearly the full program of mathematics and economics together, whereby mathematics were partially replaced by purely econometrical courses. Often we were not able to go to the lectures of the economics classes because they coincided with the mathematical program, so we just had to take the exams. Furthermore, the way people at that time worked at the economics faculty completely differed from the (small) mathematics faculty, this was sometimes difficult to deal with. Next to that we also had interfaculty econometrics courses, further complicating things. So we had the mathematical faculty as our main faculty and on top of that the economics faculty and the interfaculty of econometrics. 

Back then, we had to program all our own stuff and run it in a batch on a mainframe.

The study was more theoretical than the courses nowadays. When we finally got to the more practical side, we had the problem of computing; please bare in mind that PC’s were not available at that time. Nowadays, most students can just get their hands on a computer and do it with some statistics software. Back then, we had to program all our own stuff and run it in a batch on a mainframe. For instance, we had to program our own OLS or matrix computations. So everything was a lot more theoretical and a lot more work for any practical application, which could be one of the reasons why I went into sales right out of college. I was done with the “theoretical” stuff. 

So, it was a considerable workload for a study, which was only augmented by the completely different ways of thinking of mathematicians and economists. Combine that with the organisational mess of the three faculties and you have yourself quite a difficult study. I even passed a few exams I didn’t need to, because of that opaqueness.

Did you specialise?

My mathematical specialty was numerical mathematics. Why? It was a course from the doctoraal, or masters, which I already did in my kandidaats, or bachelor, because I simply passed the wrong exam. The confusion of adhering to three faculties in action.

There were four specializations at that time: Mathematical Economy, Econometrics (basically Time Series), Operations Research and Actuarial Sciences. You had to do one main test in all of them, but you had to do 2 in your chosen specialization, which was mandatory. My specialization was Operations Research, which is relatively rare nowadays, then it was called “besliskunde’’, or decision making theory. 

What did you miss in your study?

To me, it is very crucial that I studied econometrics. If I could I choose again, I would study it again. Only not to specialize in Operations Research. Instead, I would do Time Series. 

At that time I was thinking I wanted to work for a car factory. The assembly line in a car factory is a very intricate process and everything needs to be assembled right on time, otherwise you just get a massive backlog from the part manufacturers. Add to that that every car is different, and you have yourself a very difficult process. How to optimize and run it was my interest at that time.

He rather had his oldest son become a doctor.

Were you really one of the first students to finish the econometrics study?

No, but my brother is; I have a much older brother. He started his econometrics study in 1965. Which was, I believe, two years after the study started in Amsterdam. He was one from the first batch of econometricians in Holland. Nowadays, this seems like a positive thing to most of us. However, our father was not really enthusiastic since the study was not known back then in 1965. He rather had his oldest son become a doctor.

I did finish my study as one of the first of the faculty doing an analysis at a company rather than doing a theoretical thesis. My thesis was done at Akzo in Germany. Nowadays it is more common to write your thesis at a company. 

"Did you make a misspelling in your study?"

What parallels or differences are there between econometricians of
today as opposed to econometricians of your time?

General view of econometrics

Back then, econometrics was not really known. I was for instance interviewed by the HRM of Amro bank in 1988 and they asked: “did you make a misspelling in your study?”, it turned out that only economics was known to them. When I started Zanders in 1994 along with Chris Zanders, I told him we needed to focus on the mathematical side of the business, because that is the future. The importance of econometrics dramatically increased with the introduction of faster and more powerful computers. Mostly because this made measuring, recording and storing large amounts of data far easier. This was hardly possible before 1990, because we simply lacked the computing power.

The study

What I think is missing in the study, is balance sheet reading and financial analysis in general. After this study nobody seems to be able to do this anymore. Business administration, or economic concepts in general, are left behind. For example, the EBITDA is something every economy student learns and knows about, but econometrics students mostly have never even heard about it. I would suggest doing a little bit more on the economical views in your studies. After all, the study is called econometrics and not mathematics.

Natural vs Guided

Nowadays, a lot of models used by banks to quantify risks are profoundly the same, which creates a systematic risk on its own. Example: consider one is observing how many eggs a blackbird lays. I can take a sample from the blackbird population by going through a certain area with blackbird nests and checking the amount of eggs laid. With this data, I can make a statistic with 95% surety to predict the minimum number of blackbird eggs. However, if one interacts with the blackbirds behaviour by for example feeding them for more eggs and removing eggs when there are more than 6, then the original model will become invalid. This is simply because it is not a natural process anymore.

Quite an answer, eh?

This analogy can easily be extended to the financial market, because everybody acts on the same, prescribed (!) models. This results in everyone viewing the same information and making the same analyses. Hence everyone acts to changes in a similar way and does the same deals. In short, it has become a guided process and not a natural process anymore (the latter being the basis of our statistical theory).

In the 90’s and early 00’s, econometrics could still be used to observe natural processes, such as the stock or currency markets. Nowadays it can hardly be used as such, because markets have become “steered”. This explains the many 6σ events recently seen in financial markets.

Econometrics is a wonderful tool when applied to natural processes, but when everyone uses it, it creates guided processes. This will for certain lead to problems and a disbelief in our skill set! “Quite an answer, eh?”

Regardless, econometrics is a nice study.

Furthermore, most people don’t realise they are overusing models. They simply do the math and get an answer, which they deem sufficient. In particular, this also extends to artificial intelligence, since it automates fitting data, which makes the same mistake. Everybody will end up doing the same at the end without questioning. So, you can expect AI to be removed from most fields in a decade, because they automate a faulty process. “Regardless, econometrics is a nice study.”

What would be the most impactful change caused by econometrics in
your sector?

Every banking and insurance regulation involves econometrics. It goes very, very far. For example, if I supply goods, and you don’t pay me in thirty days, you have become a dubious debtor. You often see this in the annual reports of companies, since they have a certain turnover and a part of that turnover will become a dubious debtor (because the person delivered to is broke or bankrupt or whatever). Currently, you have to model every client and determine the probability of defaulting. i.e. not fulfilling your payment requirements. With this, you can calculate that the percentage of dubious debtors has lessened, based on the fact that it could be due to pay in a month or so.

Even for dubious debtors, even there, you see econometrics being applied. Same goes for accountancy, they have to be able to do econometric analysis on nearly all parts of the balance sheet. At universities of applied sciences (HBO’s), you nowadays too have to learn SPSS to practice mathematical methods. The impact is huge. Being a pro at it would be perfect and understanding the underlying processes is very valuable, since understanding a model can be far more valuable than applying it.

Typical power curve

If you look at the insurance industry, it is exactly the same. A good example would be a project we did together with an insurance company. A bank looks for certainty of repayment of its loan and we needed to fund a windturbine. A windturbine does not produce if there is no wind at all or if the wind is too strong. We searched for an insurance for these circumstances to get a better (more sure) bank loan. You need wind statistics, information on the exact production formula of the windmill, and so on, to calculate the expected loss of not producing electricity. The insurance company also performed this calculation and came to a comparable answer as we did. Then, to come to the insurance premium, they multiplied that very precisely calculated number by simply two (!). 

So, you calculate very, very precise, and then the premium is simply calculated “times two”; but we got the insurance for the bank loan!

To return to the question, econometrics is very important in our field of expertise, where we do treasury and quantitative risk management. This is the field of expertise where companies are confronted with financial markets and movements that one has to anticipate on. Half of our Zanders crew of 200 people, or maybe more, studied econometrics. 

Therefore, our unique selling point is that we are knowledgeable in both the heuristic and numeric side of the economics. This allows for a view deeper than the application of our mathematical models, which is important in this econometrics filled world.

Last words to readers

If you started studying again today, would you still study econometrics. If so, in what
would you specialise?

I would still study econometrics, because I like math and economics. It makes you, in general, smarter than the average student (but not necessarily a better person..!). If you ask me what I would do in hindsight, I would not do Operations Research again, but I would choose to do Time Series because I use that a lot more in my work. Also, I refused an offer to go to INSEAD post-doc, which is something I regret with hindsight.

What would you have gained from attending an institution such as INSEAD?

I learned entrepreneurship by doing, but if I had taken these courses, it would have shortened my learning by four or five years. Furthermore, the people attending those courses are usually very driven and bound to be successful. However, only doing INSEAD or IMD isn’t worthwhile, having a strong theoretical background is very important as a basis.

Would you recommend econometrics students to broaden their horizons?

If you studied econometrics, you are good at making models and doing the math. You are not better at sales, leading people, spotting opportunities, etc. In aggregate, you are just better at “math”. That is a mistake I see a lot of econometricians making, thinking they are better at everything than others, that is not the case. People that studied Econometrics are the best at analysing data, making models, thinking behind the data and understanding data series.

I think, if you want to be a successful business man or woman, studying econometrics and doing a business administration, INSEAD, IMD afterwards is the best way. Because you also learn doing business, managing people and so on. If you can do that, if that is your disposition, then it would be ‘the shit’. That means, studying, working three years, and doing another training to enhance your skills. At least, that’s my opinion.

What other fields would you recommend to newly-minted econometricians?

I would say people have forgotten what the basis of econometrics is, it’s statistics. Analyse the problem at a more fundamental level, think behind the data. Don’t just do the math.

What I think what you will need to see in your study, is to check whether the data derive from a natural or indirectly guided process. 

A nice field for more research would be to check whether a process is natural or not.

8c8S7Bmwixz2j90gcm1ffMrwVcmsShY-p_1N15vbPvmgWOW_X9KYjpGFWGvEUs9V36ctugVoAMO5CxMQfMX0VRhJNhWstcqqsPn4_70by7kziU956JFwaBP8ZjFHuUrNYMGr-28bCGt53WEK3Q

For example, take a graph of the dollar against the euro, with its two six sigma events. Many model this process by cutting out the six sigma events, denoted in red, plotting the rest of the data on a normal distribution, and using e.g. a jump diffusion/poisson processes to explain the six sigma events. I think however that the basis just isn’t a natural process anymore…

You’re fitting more and more details, while the process is not suitable for statistics. A nice field for more research would be to check whether a process is natural or not.

Studying is not only about passing exams. There is much more to it.

Do you have any last advice?

Do other things next to your study. Do a lot of fun things, so you can grow as a person. Try things such as courses from business administration, to broaden your horizon. Studying is also about making friends, and getting to meet new people. In the end you make a lot of friends that you can easily call for doing business, which itself can be a great boon. It is not a big deal when it takes a bit longer to finish your study. If can you go abroad for half a year or so, just do so. Studying is not only about passing exams. There is much more to it. 

close
Forgot password