The small-firms effect is the theory that firms with a smaller capitalisation will outperform those firms with a larger market capitalisation. This effect was one of the three factors use in Eugene Fama and Kenneth French’s Three-Factor Model that was made in 1993.
To test this theory, we ran a backtest using the Bloomberg back-testing tool on the Singapore markets. In this test, we compared the performance of an equally-weighted portfolio of stocks against half the number of stocks with a different market capitalization.
Back-tests were done over the following:
- 10 years – Corresponding to the decade after the Great Recession. This is, generally speaking, a bullish period.
- 5 years – Corresponding approximately to the period where markets experienced the taper tantrum. This is a marginally bearish or sideways market.
- 3-years – Corresponding to the first three years of the Trump Presidency. This is, generally speaking, a bullish period.
Singapore Blue-chips
For Singapore blue-chips, we suspected that larger companies tend to do better because of its ability to dominate an entire market locally and power of Temasek as a sponsor. As such, we tested the strategy of holding 15 of the largest blue-chip counters against an equal-weighted portfolio of all 30 STI stocks. Our results are as follows in November 2019:
| Returns | 3 Years | 5 Years | 10 Years |
| 30 Equal-weighted stocks | 7.06% | 0.77% | 6.61% |
| 15 largest market cap stocks | 8.24% | 2.9% | 6.58% |
Our intuition that smallest blue-chips stocks in Singapore do not out perform the largest ones held true. Larger blue-chip stocks did almost as well as the smaller stocks post-Great Recession. It outperformed during the taper tantrum era, like because local banks benefitted from rising interest rates. It also did better during the first 3 years of the Trump Presidency.
Singapore REITs
For S-REITs, we always believed that the larger REITs would underperform because we felt that older retirees are attracted to the larger REITs because it had more predictable cash flows and would be willing to settle for lower dividend yields. Recently, we noticed that after screening for smaller REITs, the performance has been dipping below the baseline to such a large extent that we can no longer resisting testing a strategy that hypothesizes that the larger REITs would have been better investments in the more recent years. Our results are as follows in November 2019:
| Returns | 3 Years | 5 Years | 10 Years |
| All S-REITs in equal weights | 12.67% | 8.25% | 13.31% |
| 20 REITs with the largest market cap | 15.93% | 9.44% | 13.52% |
The results shattered our expectations that the smaller and more ignored REITs would do better than the rest. The past 10 years was driven by a “flight to quality” towards REITs that are larger with better sponsors and quality management teams. One possible explanation is that larger REITs get a bigger chance to be included in major indices and stand to gain more liquidity from exchange-traded funds.
All Singapore stocks
This leaves the rest of the Singapore stock universe, a rich universe of 700+ stocks that can be purchased by anyone with a brokerage account. Before we started on our screening, we excluded China-domiciled counters, REITs and stocks below $50M in size. This eliminates the stocks that do not have sufficient liquidity in the markets. The new universe consisted of only 303 stocks.
This time around, we continued to actively hunt for the smaller counters to see if the small firms effect held for at least the larger universe of stocks that do not belong to the blue-chip or REITs universe.
Our results were as follows:
| Returns | 3 Years | 5 Years | 10 Years |
| All 303 stocks in equal weights | -0.90% | -4.39% | 3.00% |
| Half of the 303 stocks with the lower market capitalization | 1.99% | -3.45% | 3.76% |
From the results shown above, we can at least conclude that the small firms effect still held sway in the Singapore markets for a subset of liquid equity counters that are not REITs, China-domiciled or smaller than $50 million in size. The outperformance, however, was quite weak compared to many other value-based factors that were used in back-testing.
Conclusion
Academic theories should not be imported lock, stock and barrel into the Singapore Stock Market.
When investing in the more popular blue-chip and REIT counters, it would be dangerous to harbour the misconception that smaller firms outperform larger ones in Singapore. Instead, big is often better. Note that this is not the same as saying deep value strategies do not work. The evidence is clear that they still do. Just that they cannot be juxtaposed onto the popular blue-chip and REIT counters.
We have evidence that the small-firms effect holds sway when we broaden our screening criteria to cover a larger stock universe. Even then, using market capitalization as a criterion for picking stocks would not result in a strategy that would have led to substantial outperformance.
A careful study of what has worked in the past combined with the current state of the markets as well as an educated guess as to what comes next is necessary for investors to develop a more rigorous, time-tested approach for investing in the stock markets and extracting a superior performance with the added bonus of yearly dividend cashflows.
This is what is offered to students of the Early Retirement Masterclass. You can register for a seat here if you wish to learn our methods of approaching the markets with dividends in mind.
Chris, I noticed you like to use 10 years for backtesting. IMO 20-25 years will be the minimum required for backtesting.
Of course many things don’t have 20 years of history for backtesting. In that case, backtesting using shorter histories will simply be interesting as talking points, but not as major justifications for investing.
The other thing about backtesting on factors e.g. size or value or quality or momentum, is that it needs to be applied to a large number of entities. 15-20 stocks of one factor just doesn’t cut it. Again …. if the market doesn’t have the numbers of stocks, then this backtesting should only be for chitchat & not as investing inputs.
Hi Sinkie,
Your opinion appears to be misinformed. The economic conditions 20 years ago are vastly different vs now. Where applicable, we must take into account interest rates, economic cycles, and debt cycles in our backtesting. That means restricting time periods to that which makes sense – such as only testing within a period of time where the market conditions have not changed significantly or rather testing for similar market conditions (we will put aside the complications involved with modelling changing economies and testing for future articles). We have morphed from a hub of manufacturing to a hub of human capital. Our greatest asset is our ability to provide an educated, intelligent workforce capable of talking English to anyone all over the world and Chinese when China rises to its feet.
Backtesting here was applied to 700 stocks. Not merely 15-20 entities. Even accounting for stocks not domiciled in CHina, less REITs, and above $50M valuation mkt caps, we still have abt 500 stocks. You may wish to fact check this through a bloomberg terminal.
Hi, I share similar concerns on the duration of back testing. Without looking back too far, a 10 year window for backtest now will miss the financial crisis in 2008. What does that say about the robustness of the results from such a backtest between 2009 till 2019?
Secondly, isn’t the whole point of back testing to stress test the performance of the portfolio under different market conditions? Referring to Irving’s justification that backtesting be restricted to time periods ‘that make sense’ – we will only know on hindsight whether the time periods make sense… after the events have taken place. How then does one reconcile the usefulness of backtesting within a limited period for constructing a portfolio going forward?
1. How meaningful is the 10 year window backtesting if it misses the crisis in 2008?
2. We only know on hindsight which time periods make sense – after they happen. So shouldn’t backtesting cover a longer period to stress test the portfolio performance over different market conditions? Otherwise, cherry-picking the time periods in retrospect for analysis has limited value for gauging the portfolio performance going forward.