Is it possible the size & price factor premiums could grow?

[Tyr0ne]

I finally came across Effective Diversification in a Three-Factor World. Highly recommended for anyone who hasn't read it yet (its short too).

Let's assume that same [size & price premium] relationship will continue in the future. (Note that since size and price are risk factors, we do not know that this relationship will continue).

It is clear to me that the premiums for small and value stocks could remain constant, shrink to nothing, or reverse entirely, but my question is, is there any likelihood at all that these premiums could actually grow? This seems completely counter intuitive, but its hard to say the market is "too efficient" when such a distinct premium appears to have existed over such a long period.

My guess is that with how much more widely known these premiums have become in recent [years? decades? (or have they always been "widely known?")], there is only one direction they can go. I guess a relevant question may be, is there evidence to suggest that the premiums are gradually shrinking, using something like 20-year rolling returns?

This may be a stupid question, please humor me.

[larryswedroe]

Well the value premium in ex ante sense has certainly increased in recent years due to its underperformance,
(:-))
The fact that a premium exists and is discovered doesn't mean it has to shrink. If it is a risk premium it's discovery should not cause any particular change. If it's behavioral (anomaly) it might, but even that is not certain because humans seem to keep repeating bad behaviors and there are limits to arbitrage and use of margin/leverage.

So yes they both could grow, but if you think they are risk stories then they could shrink but should not disappear any more than beta premium would be expected to disappear

Larry

[EDN]

As Larry alluded to, the size of the forward looking expected premiums grows after periods of poor returns, such as a long stretch of "underperformance" like the late 1990s or a huge market sell-off like 2008. In order for them to be materially higher during a set period of time, one of a few conditions would need to occur:

1. the start of the period coincides with a wide spread between each side (small/large or value/growth) of the factor
2. a significant compression in the spread between each side
3. a permanent elimination of the return premium as prices converge to a one factor world where all risk is the same (a function of the market) which would require a large one-time capital gain to value and small stocks relative to the market so they all trade at the same cash-flow adjusted multiples.

For value, #1 probably applies today. If we see a very strong period of factor returns, it will be because #1 and #2 occur. I doubt #3 will ever happen.

Now you didn't ask this, but it probably bears mentioning that there is a way to potentially get more out of the premiums. By adding a portfolio that holds only companies with high profitability (which is the 4th "factor"), you have an investment with an expected return just as high as if you are holding a portfolio of the lowest priced value stocks and higher than a portfolio of the smallest stocks -- but profitability as a factor tends to do well when value does not, so there is a diversification benefit there when you include value and profitability in a balanced portfolio. OR, you can just avoid buying the value stocks that are the least profitable. A broader mix of value stocks (say cheapest 35% of companies) with high profitability has much higher expected returns than a more focused mix of value stocks (say cheapest 20% of companies) that include unprofitable stocks.

Nothing is set in stone. The basic principles are in place and have been solid for decades, but thinking continues to evolve as we seek to better understand market prices.

Eric

[Random Walker]

I think the size and value premiums will always be there but their sizes will vary over time. Someone got a Nobel Prize I believe for showing that the expected return for a company is the same as its cost of capital. As long as a bank lending money will perceive a small company as more risky than a big company and a distressed company as more risky than a growing one, smaller and distressed companies will be paying higher interest rates to borrow money. And that means the prices of their stocks should reflect that risk, resulting in higher expected returns. The size of the premia will vary, but should always be there. And just as it is impossible to time the market in general, it is very difficult or impossible to time the changes in the small and value premiums.
By the way, it is the lack of correlation between these risk factors and their variability that makes diversifying across them so valuable from my point of view. Volatility of an individual asset class in isolation ain't so good. Volatility of an asset class as a component of a diversified portfolio can be outstanding. In deciding to add an asset class to a portfolio it is how the asset class mixes that matters. And that depends on expected return, correlations with other components, and volatilities.
Dave

[stlutz]

It is clear to me that the premiums for small and value stocks could remain constant, shrink to nothing, or reverse entirely, but my question is, is there any likelihood at all that these premiums could actually grow? This seems completely counter intuitive, but its hard to say the market is "too efficient" when such a distinct premium appears to have existed over such a long period.

From the size side, there is a large amount of doubt as to whether there has ever been a size premium that is independent of valuation and liquidity factors. So, if the size premium has historically been close to zero, it's entirely possible that it could be higher in the future. It really depends upon economic developments that are unknown to us at present. If economic changes favor smaller companies, then such stocks will indeed outperform.

1. the start of the period coincides with a wide spread between each side (small/large or value/growth) of the factor
2. a significant compression in the spread between each side
3. a permanent elimination of the return premium as prices converge to a one factor world where all risk is the same (a function of the market) which would require a large one-time capital gain to value and small stocks relative to the market so they all trade at the same cash-flow adjusted multiples.

For value, #1 probably applies today. If we see a very strong period of factor returns, it will be because #1 and #2 occur. I doubt #3 will ever happen.

I don't think there is really a way to make such a determination. As Larry has pointed out many times in the past, the relative valuation of value stocks to growth stocks really doesn't forecast future performance very well. Most of the time, the market pretty much has it right--if value stocks are priced lower relative to growth than they have been on average, it's because they should be. Past performance is not a good predictor of future performance. That is, if value has underperformed recently, that is not an automatic forecast of future outperformance.

... the expected return for a company...

Finally, always remember that investments don't have expectations; only investors do. Everyone is free to "expect" what they want, and obviously some expectations are more reasonable than others. But there is no objective reality that states that investment A has an expected return of 5% while investment B has an expected return of 7%.

[asset_chaos]

Anything is possible in the markets. The data seem relatively clear---to my untutored eye---that small and value represent real economic risk factors. But I'm not aware of any good data or theory that suggests what the size of those factor premiums should be or why in the past they have realized a particular premium, instead of the factor premiums being 10% or 0.10%. So I think it's perfectly reasonable that the small and value factor premiums could go up or down in future by unknown amounts. I don't know why they were that size in the past. Is there theory that predicts the size of factor premiums?

[EDN]

From the size side, there is a large amount of doubt as to whether there has ever been a size premium that is independent of valuation and liquidity factors. So, if the size premium has historically been close to zero, it's entirely possible that it could be higher in the future. It really depends upon economic developments that are unknown to us at present. If economic changes favor smaller companies, then such stocks will indeed outperform.

-----------------------------

I don't think there is really a way to make such a determination. As Larry has pointed out many times in the past, the relative valuation of value stocks to growth stocks really doesn't forecast future performance very well. Most of the time, the market pretty much has it right--if value stocks are priced lower relative to growth than they have been on average, it's because they should be. Past performance is not a good predictor of future performance. That is, if value has underperformed recently, that is not an automatic forecast of future outperformance.

-----------------------------

... the expected return for a company...

Finally, always remember that investments don't have expectations; only investors do. Everyone is free to "expect" what they want, and obviously some expectations are more reasonable than others. But there is no objective reality that states that investment A has an expected return of 5% while investment B has an expected return of 7%.

On the size premium
It has been large and statistically significant over the longest periods of time we can measure in the US and foreign markets. Remember, the "size premium" is calculated as SG+SB+SV minus LG+LV+LV, so it teases out the effect of valuation. Has size been larger in "value" and "blend" than "growth"? Yes, but measuring for size while controlling for price also reveals a robust relationship. From 1927-1990, FF SmB averaged +3.1% per year. From 1991-2012 (post FF research on 3F model), FF SmB averaged +3.0% per year. Is some of the size premium associate with illiquidity risks? Probably, but that is a symptom and not a replacement for the more general "size" dimension.

People are fond of saying the size effect has been small since 1984, which happens to coincide with a period beginning after a huge run in small companies and an ensuing bear market for them that was unique to small stocks. This time period doesn't disprove the size effect anymore than the S&P 500 failing to outperform LT Treasury Bonds from 1982-2012 proves that bonds have higher expected returns than stocks.

On expected returns for value stocks
There is pretty robust evidence that the value premium does tend to be very large (above average) and persistent in periods after value has underperformed growth and the spread in valuations has widened. From a Jim Davis paper on the subject, he found that HmL returned +16%, +11.2%, and +6.5% annualized over the ensuing 1, 2, and 3 years after HmL had an extremely poor return (growth beat value by 10% or more in the preceding year).

What you may be thinking is that there is not very good evidence that the value premium tends to underperform (have below average returns) after a period of above average returns relative to growth. From the same paper, Davis found that HmL returned +4.6%, +4.6%, and +4.9% in the 1, 2, and 3 years after HmL had an extremely good return (value beat growth by more than 10% in the preceding year).

On investment expected returns
Investments absolutely do have expected returns. With expected being the key word. This is just basic corporate finance. An investment is nothing more than a security that represents an ownership stake or a loan to a business. That company has a cost of capital they must pay when attaining new financing (either by selling ownership shares or borrowing money) which is the price which the stock is traded and the interest rate that loans can be made. The expected return of the investment to the owner is simply the opposite of the company's cost of capital. Now, the expected return is not always the realized return -- but the more diversified an investment portfolio, the more likely that the two will converge and the random noise associated with any one security apart from its underlying risk will prevail. And finally, the expected return varies over time inversely with its risk. In times like 2008, perceived risk is high so prices fall to offer higher expected future returns to reward investors for their assumption of greater uncertainty. In times like 1999, the opposite happens.

Eric

[Random Walker]

For individual investors considering a new investment strategy such as small, value, momentum, or any source of returns, I think they should take a fairly systematic approach to evaluating the data for themselves.

1. Use the longest historical time series available
2. Look for out of sample data that supports or refutes such as different time period or different geographical market
3. Ask themselves if there is a logical risk or behavioral story supporting the source of returns.

Having, understanding, and believing in this information will give the investor the strength to stick with a plan at times when it isn't working out so well.

Dave

[stlutz]

Remember, the "size premium" is calculated as SG+SB+SV minus LG+LV+LV, so it teases out the effect of valuation. Has size been larger in "value" and "blend" than "growth"? Yes, but measuring for size while controlling for price also reveals a robust relationship.

If stock A has a market cap of $100, and stock B's is $80, is stock B a "smallcap" stock or a "value" stock? Hard to say. I suppose you could base size on measures independent of market cap/price (revenues, employment, earnings etc.). That's what Arnott does with fundamental indexing. And he argues that there is no smallcap premium; only a value one.

[stlutz]

Investments absolutely do have expected returns. With expected being the key word. This is just basic corporate finance. An investment is nothing more than a security that represents an ownership stake or a loan to a business. That company has a cost of capital they must pay when attaining new financing (either by selling ownership shares or borrowing money) which is the price which the stock is traded and the interest rate that loans can be made. The expected return of the investment to the owner is simply the opposite of the company's cost of capital.

Okay, but the the "expected return" is simply the dollar-weighted average of market participants opinions.

A vast majority of stock/bond transactions don't involve the underlying company, only the investors. If you think that stock X is a bad investment at price Y relative to other opportunities while I think just the opposite, the market price is simply a meeting point that allows for an exchange between us to occur. The same goes for when a company issues stock or debt; the price is simply a meeting point where conflicting opinions can result in an exchange.

I know I'm annoying in always stressing this point--here's why I do: Some proponents of various "premiums" on this board like to use language to imply an almost scientific certainty that their strategy will outperform, and that any detached objective observor will agree. Hence instead of saying "I expect investment A to do better than investment B, and here's why", you'll see "Investment A is expected to do better than investment B." Or you'll see the word "rational" thrown in. If you think investment A is better, you're "rational"; if not, you're "irrational". And finally, you'll see highly debatable assertions thrown in as fact, such as "riskier investments generally outperform less risky ones", or that "value stocks are riskier than growth stocks."

So, if you put it all together you might have a statement like, "In a rational market, value stocks have higher expected returns than growth stocks because they are riskier." Sounds persuasive, but the entire statement is really just a manipulation of language more than being a description of reality--i.e. marketing.

I recall reading one DFA paper which claimed an 85% chance that value will beat growth over any 15 year period. I'm sorry, but nothing is that certain in investing. By all means, make a 60/40 claim, but going beyond that is marketing, not science.

[mlewis]

I recall reading one DFA paper which claimed an 85% chance that value will beat growth over any 15 year period. I'm sorry, but nothing is that certain in investing. By all means, make a 60/40 claim, but going beyond that is marketing, not science.

I'm not sure how you can be so certain that the line distinguishing marketing and science lies somewhere between 60 and 85 percent.
That sounds like a marketing claim to me.

[stlutz]

I'm not sure how you can be so certain that the line distinguishing marketing and science lies somewhere between 60 and 85 percent.
That sounds like a marketing claim to me.

Nice--and definitely in the spirit of my post!

I'll rephrase and simply say that with the 85%, I personally think "newsletter writer"; with the 60% probability, I personally think, "This guy is at least being realistic. I'll listen further." Obviously, everyone can and will make their own determinations as to when their BS meter goes off.