As more investors have turned their attention to indexing, researchers are turning their attention to the best ways to do it. Traditionally, index funds are constructed as market-capitalization-weighted indices to capture the aggregate activity of the market.
But is emulating the market a desirable objective? Some are starting to question whether it is. In fact, using ever more sophisticated tools and methods, researchers are postulating a number of alternatives, ranging from the tilted index to smart alpha to fundamental indexing.
Back in the early 1990s, Eugene Fama and Kenneth French led the way with their work identifying common risk factors in the returns on stocks and bonds. In the 2000s, Robert Arnott, Jason Hsu, and Philip Moore added their scholarship on robust factors. Countless others have contributed further insights.
Recently, Lidia Bolla, CFA, managing partner at Algofin in Switzerland, authored “Fundamental Indexing in the Global Bond Markets: The Risk Exposure Explains It All” for CFA Institute Financial Analysts Journal®. In her paper, Bolla takes the basic model of factor exposure and applies it to the concept of fundamental indexing in the bond market.
The composite index Bolla created produces an annual average outperformance of 101 basis points (bps) per year versus the market-cap-weighted benchmark from 1990 to 2014. In the world of bonds, that’s a big difference. But the question remains: Are these new approaches to investing, in fact, better than traditional market-cap indexing, or are they just different? In short, is this outperformance really alpha?
To get to the bottom of this question (and many others), I recently interviewed Bolla about her take on the pros and cons of fundamental indexing in the bond market.
Ron Rimkus, CFA: In just a few sentences, please describe fundamental indexing for our readers.
Lidia Bolla, CFA: Fundamental indexing is a type of passive investing. Compared to traditional indexing, it is not weighted according to the market cap of the stocks or bonds. Rather, its security weighting scheme is driven by fundamental factors. We wanted to test the benefits of moving away from market capitalization, but still indexing.
The four factors you used in your research were GDP, size of population as proxy for labor force, the square root of land as proxy for resource richness, and energy usage as a proxy for technological progress. Why did you choose these as fundamental factors? Were you constrained in your approach based on the data available? Are there other factors that were overlooked or unavailable to you?
These factors were chosen for continuity’s sake with the large and growing body of work on the topic. These were essentially set in place by Arnott’s research, but many others have followed suit. I wanted to make sure that our work would not be singled out due to the usage of alternate factors. Many other authors use the same factors, so keeping [the] discussion consistent with [the] body of research was important.
None of these variables take into account valuation, spreads, or any definition of opportunity. Is that a problem?
Not explicitly using valuation factors is a strength — so it’s neutral on valuation. The work is based on Fama and French’s factors and Arnott’s factors that seem to best describe the drivers of returns. These factors work because they are unique and ultimately deliver better performance.
Of course, your target for comparison is market-cap indexing. Market-cap indexing invests based on inclusion in an index and weights securities by market-cap representation in the index. In essence, the higher the price, the greater the weight. So it naturally has a bias toward momentum. Therefore, there is a departure from fundamental analysis of issuers. Some suggest that the more popular such indexing becomes, the greater the departure of valuations from fundamentals. Would you expect the same phenomenon with fundamental indexing of bonds? Why or why not?
Moving away from valuation is what matters. There is no proxy for valuation here. If based on valuation, the index naturally moves toward issuers with large amount of debt/deficits.
Your approach gives equal weighting to each of the four factors in the model. Why? On what basis?
Consistency with the work of previous authors.
You re-weight the portfolio once per year on 1 January. Did you test re-weighting on other dates?
Yes, we tried re-weighting schemes for the first day of each of the 12 months in the calendar and it made no difference. In the 12 different months we reached [the] same conclusions each time.
In essence, your fundamental weighting scheme gives greater weights to large, more productive economies. Presuming these larger, more productive economies performed well in the past, why should we expect larger, more productive economies to perform well in the future?
Not a bias toward large economies . . . GDP is only one of the factors. Market-cap weights have bias toward large economies too. One of the factors is also land area. The land area index performed quite well . . . Mexico and Canada, for instance, get a large weighting based on physical size, but not necessarily from a large economy . . . therefore [it is] not necessarily driven by large economies.
Have you considered a fundamentals-friendly policy index weighting? For instance, countries like Mexico and Venezuela might have substantial energy resources, but due to nationalization and politicization, they are grossly underutilized. Therefore, perhaps bad policies receive low weights and good policies receive high weights? Perhaps weight the markets according to the degree of freedom in the market?
That’s an interesting idea. Perhaps we will look into it.
Is the study really just skewed by the presence and weighting of the United States and perhaps a handful of other large economies that have engineered steadily declining rates between 1990 and 2014, leading to stronger bond markets there than in other countries? Perhaps this is suggesting alpha where none exists? Perhaps try the study again without the United States and Europe?
In five recent papers, only a handful of countries drive the results. In fact, in each, the weighting on Japan has had a big influence on outperformance. So no, the weighting of the US has not had a huge impact.
Figure 5 in your paper illustrates that the main performance drivers over time were an underweight in Japanese bonds (contributing roughly 35 bps) and overweights in Australian, Canadian, Mexican, and Polish bonds (whose contributions ranged from 6 to 13 bps each). As you highlight in the paper, Japan’s sovereign debt was downgraded during this period while the other four countries were upgraded. It just so happens that a country in which the composite had a significant underweight experienced a credit downgrade, while several countries in which the composite had significant over-weights experienced credit upgrades. By implication then, unexpected changes in credit ratings happened to be congruent with your over and under weightings. By the same token, couldn’t this work in the opposite fashion going forward?
Yes, this is a real risk . . . not explainable in other studies.
Have you considered running the numbers on countries that didn’t experience significant unexpected credit rating changes? What would you expect the results to look like then?
No, it’s simply not one of the many steps we undertook, but that is a good idea.
Did Japan show a strong contribution because the market-value-weighted index showed relatively weak results and the land area/energy resources are small relative to its economy? In other words, did the composite index construction reduce the negative impact of Japan?
Your data suggest that convexity differences accounted for the greatest discrepancies in weighting schemes. Is this true?
Yes, skewed to countries with higher credit risk.
Was the fundamental composite comprised of higher (or lower) bond ratings?
Fundamental portfolios had higher credit risk than market portfolios. The market cap index had an average credit rating of 1.91 and the fundamental composite index was 1.99 (1 =AAA and 2 = AA+). This is a statistically significant difference and contributor to performance . . . On Table 4 the unexplained variance or alpha is 101 bps and in Table 7 it falls to 90 bps, so there are 11 bps that fall away due to default/credit risk.
Are you satisfied that your results are different from spurious correlation?
Yes, once we take into account other risk factors, alpha vanishes. This paper suggests that performance is driven by risk exposures. So, fundamental indexing is not necessarily better than market-cap indexing, but it does give investors an easy way to gain exposure to these risk factors, if that’s what they seek.
So, there you have it. Yet another study finding that alpha disappears when risk factors are taken into account. Maybe active investors struggle to outperform because they are simply playing the wrong game. Thank you for your time, Lidia.
If you liked this post, don’t forget to subscribe to the Enterprising Investor.
All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer.
Image credit: ©Getty Images/liangpv