Navigating the Factor Zoo with Matthias Hanauer

[00:00:00] Welcome to excess returns where we focus on what works over the long term in the markets.

[00:00:05] Join us as we talk about the strategies and tactics that can help you become a better

[00:00:09] long-term investor. Justin Carboneau and Jack Forehand are principals at Validia Capital

[00:00:13] Management. The opinions expressed in this podcast do not necessarily reflect the opinions of

[00:00:17] Validia Capital. No information on this podcast should be construed as investment advice.

[00:00:21] Securities discussed in the podcast may be holdings of clients of Validia Capital.

[00:00:24] Hey guys, this is Justin. In this episode at Robico. Jack and I and our audience, I think really value individuals and investors like you that have one foot. You can piggyback on this, but it's like you have one foot in the empirical finance academic grounding world with the research you're doing, but you're also a practitioner. You

[00:01:44] guys run billions and billions of dollars in initial paper, maybe coming up with a new data set and then leveraging on this data set multiple times. So most of it is quite spontaneous. So some

[00:03:00] of it is like we want to, yeah, we see something in the market,, it was asked questions and said, it's not for this paper, but interesting idea. So usually it comes from one idea to the other. And then if you do it for multiple years, you have a lot of papers in parallel. And with some papers, it goes faster, but that is the multi-year process. So in the end, you see the end product,

[00:04:20] but typically it's sometimes a shorter process behind it.

[00:04:23] Sometimes it can be several years behind it.

[00:04:26] Yeah, that's great.

[00:04:26] I mean, you can long-short factors that there's like long lag has higher returns than the short lag, then we require usually some type of robustness. You can think about robustness over

[00:05:41] time, maybe not in every year, but title. It's one of my, you know, a lot of times we get these dry titles in these research papers. But I would say that went along with AQR's size matters if you control your junk or two of my favorite paper titles. So congratulations on that. Thanks. Yeah, as we dig into the paper, I want to talk a little about the process of compressing the FactorZoo

[00:07:03] because you had to quote the paper that you compress the FactorZoo by explaining the available alpha factor that has like a 1% return every month with zero variance. So every month 1% it will be have like a great return, zero variance and this factor would not be picked up by a PCA. And actually some PCA papers call these type of factors weak factors. Well for us it would be a great factor. Having 1%

[00:08:22] return every month without any risk. What is a bit different in our approach is that we really control for all the other factors that have been published before, whereas maybe some of these papers did a bit of cherry picking, controlled for some factors,

[00:09:41] but not all. So we took all the factors we want to measure it and doing little changes to that, testing then the robustness across regions,

[00:11:00] maybe if you have limited time series data

[00:11:03] and maybe having then an independent verification

[00:11:05] of the results by the factor value. Can you talk about in academic research,

[00:12:20] when they say value-weighted factors,

[00:12:22] what are they talking about?

[00:12:23] Yeah. Typically, when. So most academic research then would be done with valuated factors, is that correct? Yes. So it's typically in our paper this factor sous-zip or honey this with the GRS statistic, but it's also the same factor that has the highest significance of the alpha for the CAPM. So then we added this factor to the CAPM, and one iteration further, we checked

[00:15:01] which factor of the equal weighted factors itself had a high alpha. So the equal weighted alpha was about 50% higher than for the cap value weighted factors. And on the other side, the alphas were a bit more uncorrelated because we see that the sharp ratio, the tangency portfolio sharp ratio of the equal

[00:16:21] weighted factors was about two times or three times higher

[00:16:25] than the sharp ratio of the tangency portfolio intrinsic value to market value factors, or an enhanced book to price factor, because this intrinsic value is based on a residual income model that starts with book value plus some information about future expected profitability. So these were enhanced versions of existing variables,

[00:17:42] also like momentum was then picked,

[00:17:44] because we started, we have,

[00:17:46] once you have value in the model,

[00:17:47] typically the model number of factors, and that it needed much more factor to explain really all the alpha and the factor.

[00:19:02] So I want to move to your next paper, which is actually an excellent paper. It's called Resurrecting the Value Premium,

[00:19:04] which obviously all of us hope is going to happen very soon.

[00:19:07] But maybe it has interesting going back a bit how pharma French define value. So they take the book to price ratio. We all know that. But when measuring the book to price ratio, they define it only once a year at the end of June. They don't take the market capitalization. So the price they

[00:20:22] have to pay for the stock at the end of June. And so, but then when you now also use mentor, for instance, in your portfolio, or if you blend momentum value scores to come up with your portfolio, then actually it turns out that this more timely value factor is better because it's more negatively correlated

[00:21:41] with value and it's then also not redundant anymore

[00:21:45] in such a spending type of regression. that we take just a four-factor model and use this one. So in your paper, you had basically three steps you use to try to improve the value factor. And one you've already alluded to a little bit, which is this idea of additional metrics. Is there anything else we have to cover there before we get to the other ones, or have you pretty much covered that? I think it's interesting to look at which factors we use there.

[00:23:00] So I think we also use book-to-price.

[00:23:03] We use this more devil variant of book-to-price using the most recent market capitalization shows and therefore our measure of choice is book to price. But yeah this what is the best measure over a certain sample period is also quite sample specific. For instance if you would run the same tests as final French 96 with all the data after the publication to now, extra I think earning surprise was better than book to price. So therefore we prefer And we also thought about including one measure from looking at the payout policy of a company and not just using dividend yield, but also share buybacks and share issuance to come up with something what we call net payout yield, but it's also very similar to shareholder yield. So having these different perspective to measure value.

[00:25:43] Why do you think price to book is still so widely used?

[00:25:46] Like when I go look behind the scenes at like value VTFs, is the best of these four value factors that we discussed. So the second thing you looked at in the paper was this idea of risk management. And this kind of gets at an issue that a lot of value investors debate, which is this idea of if there's a lot of cheap companies in one particular industry or sector, do I overweight that sector or do I stay with the sector weightings and then try to find value within there? So can you talk a little bit about what you found there?

[00:27:01] So what we say in the paper, we academic studies in the US are using, and this Chris Combus that universe, so then it's about 1,000, 1,500 stocks in the 1960s that grows until 6,000 stocks in the end of the 1990s.

[00:28:20] And nowadays we have around 3,000 stocks.

[00:28:23] But within this big universe, there are let's focus on these stocks. These are roughly stocks like in the MSCI USA, the investible universe for many institutional investors. And these are around 600 stocks, take this universe, but then don't apply evaluating this universe, but more equilating,

[00:29:40] so that we're just not depending on a few in autumn 2020. And I think since then, we saw then some recovery.

[00:31:00] But I think you want to talk more about what was happening between 2018 and 2020? value of value, the value spread factor. So some people said that value might be overcrowded. It's known. Everybody knows it. Therefore, it stopped working. But we were wondering if this is really the case. And therefore, we looked at also what we call the value of value, the value spread. So how much the valuation of value

[00:32:20] compared to the history has changed.

[00:32:22] Per definition, value stocks are always

[00:32:25] cheaper than their expensive counterparts. correct for the changes in the evaluation spread. So the returns that we saw there were mainly driven by multiple expansion for crows stocks, correcting for these changes in valuation. We saw that value was actually not underperforming in this period at all, having maybe a small positive performance, yes,

[00:33:40] and then a slightly negative performance in the other unit process.

[00:33:44] But most of the underperformance just came from valuation changes.

[00:35:03] And now a few, this was unsustainable. Because if it would have been the fundamentals that it would scatter plot where you looked at the valuation spread versus the value return. So can you just talk to what sort of the data sort of shows us there? So what we do there is this is not just the US, but this is an all country universe, so developed and emerging markets. So, and what we use there, we have these annual value returns on the X-axis or

[00:36:22] sorry, on the Y-axis, these are annual 12 month value returns.

[00:37:21] I think this is quite interesting. Actually, this corrective for this valuation spread,

[00:37:26] the value return is higher than the realized return

[00:37:28] because over the full sample period,

[00:37:30] we see a small widening of the valuation spread.

[00:37:35] I'm just curious, what are your thoughts on factor timing?

[00:37:37] Because when we get into spreads and stuff like that,

[00:37:39] people always have these ideas that,

[00:37:40] well, when the spread's wide or whatever,

[00:37:42] I can take advantage of that,

[00:37:43] I can go into the factor, I can time this.

[00:37:46] What are your thoughts on that

[00:37:47] and if that really works in the real world? there's now more attractive than it was before. After 2022, I thought the relationship between growth and value and interest rates, I thought I was right about my theory, which is interest rates go up, growth stocks should go down, value should outperform. And that happened for basically one year.

[00:39:01] But after this year, I think some investors

[00:39:03] are sort of asking or questioning that.

[00:39:06] But one of the most factors in Japan, maybe the best factor working in Japan. Or if you think about the early 2000s when we had this value rally after the dotcom bubble

[00:40:20] and initially it was not that interest the US, even better in international markets. And then yeah, why is this the case? Because if you look at many value portfolios, then actually they underperform the broad market index, be it the MSCR world or the S&P 500.

[00:41:40] And actually what I think is the case this year, then you can get this more pure effect. If you're more like having a concentrated value portfolio, then it also depends a lot how these McNissifhand seven stocks are doing. Yeah, for sure. Yeah, you can, as value investors, say, okay, the time is finally here, and then be disappointed

[00:43:03] in the next month.

[00:43:05] But I don't know, the on past realized returns. So this sounds a bit like the disclaimer, past performance is not a guarantee for future performance, but I would like to elaborate a bit more. So can think about of a bond, yeah, when bond yields are going up as in the last two years, then it's bad for realized returns.

[00:44:20] So the bond prices go down.

[00:44:22] And, but actually you had bad past

[00:44:26] or negative past returns, for the other factors that you already have in the portfolio. If people want to follow you, your research, follow you on Twitter or X or whatever we're calling it, where can they go to learn more? So I'm on Twitter and I'm on LinkedIn. So both platforms, you can just find me by my name.

[00:45:40] So Hanawa Matias is on Twitter and on LinkedIn.

[00:45:44] It's Matias Hanawa.

[00:45:46] Great. Thank you very much.

[00:45:47] We really appreciate it.

[00:45:48] Happy holidays.