Investors should always strive to understand observed performance

Investors should always strive to understand observed performance

02-12-2020 | Intervista
Mathijs van Dijk is Professor of Financial Markets at the Rotterdam School of Management, Erasmus University. We spoke with him about recent factor performance and some of the challenges raised by the pressing need to take sustainability into account.
  • Yann Morell Y Alcover
    Morell Y Alcover
    Investment Writer

Speed read:

  • Recent performance does not necessarily undermine factor investing
  • Climate change has become a major risk for investors
  • Sustainability integration implies important challenges

Over the past few years, factor investors have been through a rough patch, with several popular time-tested factors posting disappointing results. How do you view these developments?

“As an academic, I don't necessarily follow short-term factor performance very closely. That said, I think that a couple of years of disappointing performance do not necessarily undermine the whole idea of factor investing.”

In financial markets, there is no ‘law of nature’ that you can trust to persist. As a result, we need to base our decisions to a large extent on past data

“In financial markets, there is no ‘law of nature’ that you can trust to persist. As a result, we need to base our decisions to a large extent on past data. So, if I were an institutional investor, I would still make long-term backtests, such as the ones that factor investors apply, an important basis for my work.”

Yes. But from that perspective, do the last three years of data mean anything?

“That’s difficult to say. The premise behind backtesting is that we don’t know expected returns and take historical realized returns as an estimate. But there is a lot of noise in the realized returns we observe. Take Netflix or Tesla. These stocks achieved stellar returns over the past decade, but I'm pretty sure those returns were not expected ex ante.”

“This was actually the key idea of a study1 I co-authored: the difference between expected and realized returns. Realized returns are the sum of expected and unexpected returns. From the perspective of factor investing, we want to measure expected returns.”

“But we have unexpected returns all the time, because some companies do unexpectedly well, while others have a lot of bad luck. And the picture gets blurred. Unexpected returns simply drive a wedge between expected and realized returns. In our study, we suggested an approach to clean the historical data a little bit to get closer to expected returns.”

“Standard backtesting considers everything we observe as expected returns. If certain stocks have higher average returns over a long period, then we say: OK, they have higher expected returns. And the implication is that that going forward, the expected returns on those firms will also be higher. But that's a tricky exercise.”

“The examples of Tesla and Netflix illustrate this very well. I don't think that anybody expects Tesla to show the same performance in the next 10 years.”

Tieniti aggiornato sull'universo quantitativo
Tieniti aggiornato sull'universo quantitativo

Maybe Elon Musk…

“Yes, maybe Elon Musk. But I think most investors do not. And therefore, backtests need to look at the largest possible number of stocks over very long periods of time so that unexpected events may cancel each other out a little bit. And even if the last three years were pretty extreme, I wouldn't expect the overall conclusion on factor performance over the past decades to change that much.”

“Also, if you pursue a factor investing strategy, there is a good chance that some of the premiums you are aiming to harvest are, at least partially, a compensation for risk. And if that’s true, you need to consider that for a few years, that risk may materialize.”

“That said, the study I just referred to is only a starting point. I would not recommend applying this method and basing an investment decision on that. I see it more as an interesting tool to help investors get a better understanding of what happened over the past few years, rather than a well-fleshed out method. At the end of the day, investors must always strive to understand observed performance and make the call whether it was perhaps unexpected or whether it really represents a shift in expected performance going forward.”

How do you view innovations such as big data or artificial intelligence for the future of quant investing?

“My view is that these techniques are definitely worth exploring, but that it’s not so easy to get things right. Conceptually speaking, the idea of aggregating data from a variety of sources and seeing if and how they influence returns and, moreover, doing this in an unstructured and non-linear way, sounds appealing.”

“But the huge downside is that you never really know what’s going on – and this was actually my own experience in a paper I co-authored on machine learning on an altogether different topic.2 With old-school linear regression analysis, one at least has the advantage of knowing what you're doing: you know what you put in and what you get out.”

The Covid-19 shock has made even more obvious the need to take sustainability into account. One key aspect is climate change. How do you view this risk? What would be the main challenges for investors in addressing it?

“Climate change is one of the major sources of systematic risk for investors. We have a combination of both physical and transition risk, that could potentially have a serious and pervasive impact on companies and financial markets.”

“For investors, there are two main challenges. The first one is that, in this case, backtesting is unlikely to be effective, because the events we're talking about are completely new. I don't want to downplay today’s storms and heatwaves, but I think there's much more coming our way. So, historical data will be of little use and the best you can do is to work on possible scenarios.”

“The second challenge is the lack of reliable data, even on key aspects such as carbon footprints, for example. It is now clear that carbon data available from various data providers can be very messy. It’s self-reported, is usually only available for large firms and extrapolated to smaller ones, and there's very little consistency across the different data providers.”

“And then, even if you know the carbon footprint, it's still a very poor proxy for carbon risk going forward, because companies are bound to differ a lot in the way they tackle this issue. For instance, some may already be working to reduce their emissions, while others may be completely unaware. Also, reducing carbon footprint is easier for some industries than others.”

Our discussion on ESG should be backed up by evidence and reason, not wishful thinking’

You are also looking at other aspects of sustainability, such as ESG scores. Could you explain your work?

I am currently working on a research project to test whether stocks with higher ESG scores really did perform better during the Covid-19 crisis, as some media and industry reports suggested. The background is that I am skeptical of the simplistic story that investing in stocks with higher ESG scores systematically leads to higher returns.”

“That may have been the case over the past decade, but my interpretation is that there has been a large demand effect at play. Valuations of stocks with high ESG scores have likely been pushed up, which would suggest that, according to the standard finance theory, expected returns for these stocks might in fact be lower going forward.”

“Of course, the current trend may continue in the near term and investors may be still able to benefit from it. But, ultimately, I suspect achieving higher returns is not the most compelling argument for ESG investing. Perhaps reducing risk would be a better a reason. This is something I am also looking into.”

“I am very much in favor of ESG investing because the financial sector has a critical role to play in the transition towards a more sustainable world, and we badly need it. But I also think that we should have a discussion backed up by reason and evidence, rather than by simple wishful thinking.”

1 Hou, K. and Van Dijk, M. A., 2019, “Resurrecting the Size Effect: Firm Size, Profitability Shocks, and Expected Stock Returns”, The Review of Financial Studies.
2 “Machine learning shows that ecology eclipses culture in predicting societal complexity” (working paper, with Thomas Pollet)

Gli argomenti collegati a questo articolo sono:


Confermo di essere un cliente professionale

Le informazioni e le opinioni contenute in questa sezione del Sito cui sta accedendo sono destinate esclusivamente a Clienti Professionali come definiti dal Regolamento Consob n. 16190 del 29 ottobre 2007 (articolo 26 e Allegato 3) e dalla Direttiva CE n. 2004/39 (Allegato II), e sono concepite ad uso esclusivo di tali categorie di soggetti. Ne è vietata la divulgazione, anche solo parziale.

Al fine di accedere a tale sezione riservata, si prega di confermare di essere un  Cliente Professionale, declinando Robeco qualsivoglia responsabilità in caso di accesso effettuato da una persona che non sia un cliente professionale.

In ogni caso, le informazioni e le opinioni ivi contenute non costituiscono un'offerta o una sollecitazione all'investimento e non costituiscono una raccomandazione o consiglio, anche di carattere fiscale, o un'offerta, finalizzate all'investimento, e non devono in alcun caso essere interpretate come tali.

Prima di  ogni investimento, per una descrizione dettagliata delle caratteristiche, dei rischi e degli oneri connessi, si raccomanda di esaminare il Prospetto, i KIIDs delle classi autorizzate per la commercializzazione in Italia, la relazione annuale o semestrale e lo Statuto, disponibili sul presente Sito o presso i collocatori.
L’investimento in prodotti finanziari è soggetto a fluttuazioni, con conseguente variazione al rialzo o al ribasso dei prezzi, ed è possibile che non si riesca a recuperare l'importo originariamente investito.