BOOK VALUE | About Book Value | 8-year BackTests | BookValue Blog |

TEST OBJECTIVE

By replicating the comprehensive Brandes Institute Price/Book exercise for the JSE (go read about this in the "Value Power" menu), we were hoping to provide strong confirmation of trends they had discovered regarding value investing. We aim to discover just how much the Price/Book, or Price/NAV ratio of a JSE share has an impact on portfolio growth. We will show, as with research done on international markets, that portfolios created from low Price/Book shares significantly out-perform all other shares and the indices over 1, 3 and 5 year periods.

METHODOLOGY PRIMER

We will test multiple periods commencing in 2000 (we are busy working on additional analysis back to 1990 for our subscribers only.) In each period we will sort the JSE by Price/Book ratio as at 30 March, based on the most recent financials of each company at that date, be they finals or interims (this ensures fresh fundamental data is used in deriving the Price/Book ratio). We then divide the sorted list into 10 sets of equal deciles, with Decile-1 containing shares with the highest 10% of Price/Book ratios (expensive shares) and Decile-10 containing the lowest ("value shares".) We create portfolios from each decile and then measure each portfolios' 1, 3 and 5 year subsequent growth. Growth is measured using the last day in March of each year as start and end dates.

We created five sets of deciles to measure average 5 year growth, namely 2000-2005, 2001-2006, 2002-2007, 2003-2008 and 2004-2009. There are 2 additional sets of deciles for the 1 and 3 year growths, namely 2005-2008 and 2006-2009. There is also one extra decile set for the 1 year growth, namely 2007-2008. So there are 5 sets of 5-year deciles, 7 sets of 3 year deciles and 8 sets of 1 year deciles. Decile sizes ranged from 23-30 shares depending on year. The tests have thus exposed the 1,3 and 5 year periods to the latest 2009 crash as well as the 2000 meltdown. Since there are two full boom and bust cycles in this test period it is a good indicator of strategy performance.

Finally we averaged the results for each decile (5 sets of results averaged for 5 year growth, 7 sets of results averaged for 3 year and 8 sets averaged for 1 year growths), and produce summary findings together with analysis of decile portfolio characteristics such as risk adjusted return, volatility, base rates and total return. This ensures a holistic method of evaluating a strategies performance instead of chasing strategies with seemingly the highest growth rates.

This set of tests represents the entire JSE universe ("ALL STOCKS"). We will also shortly publish the same tests conducted on universes that for each period eliminates the bottom 25% (to create a "LIQUID STOCKS") and bottom 50% ( creates a "LARGE STOCKS" universe) of the JSE by market capitalisation, to look at how the strategy performs with the "micro-cap" shares excluded, and how it performs on larger issues more common with fund managers/unit trusts. This data will be for subscribers only.

SUMMARY FINDINGS FOR "ALL STOCKS"

We see breathtaking confirmation of international tests, for all of 1, 3 and 5 year portfolio holding periods. On average, during the period 2000 to 2008, creating a portfolio of the lowest Price/Book shares on the last day in March in any year would have yielded a 52% CAGR if you held those shares for 5 years, 57.6% if you held onto them for 3 years, and 58.7% if you just hung onto them for 12 months. This is emphatic confirmation that low price/book deciles (9 and 10) substantially outperformed high price/book deciles, with significant out-performance by decile-10 which contained shares with lowest price-to-book ratios of 0.38 and less. The results on the JSE are amplified enormously compared to other larger stock exchanges where similar research has been conducted.

STANDARD DEVIATION

It is one thing having impressive average growth rates, but quite another if volatility is through the roof. Here we will examine decile-portfolio growth, versus standard deviation of underlying shares together with an averaged risk adjusted return ratio (Sharpe ratio=growth divided by standard deviation).

We see that Decile-10 provides, on average, a superior risk adjusted return by deviation of portfolio constituency (as indicated by the Sharpe ratio). In the 5-year periods, Decile 10 provided the 3rd best Sharpe ratio, although at a respectable 0.74, it was greater than the 0.55 for the 1-year periods and 0.7 for the 3-year periods. Note how the Sharpe increases as the holding period increases.

These graphs dispel the myths that expensive shares are "safe", as the lower decile shares have much lower Sharpe ratios than the higher ones. In the 1-year example, Decile-1 not only underperformed but the expensive shares that it was holding gyrated wildly resulting in a gut-wrenching Sharpe of 0.25! This meant many shares in the portfolio were decimated whilst others went through the roof. Unfortunately more went through the floor than through the roof as the anaemic average growths show.

TOTAL RETURNS

We now evaluate total return delivered by a 1-year holding period strategy over the 2008-2008 period. We start in March 2000 and for each Price/Book Decile, we take a hypothetical R1.00 and use it to buy equal rand amounts among the underlying shares of the portfolio. We hold the portfolio for a year, after which we rebalance it (sell everything) and repeat the process (8 times) with the funds we have over. We exclude broker fees etc. (which will be about 2% each year per portfolio, 1% for the buying and 1% for the selling.)

The returns for Decile-10 are rather remarkable, with your R1.00 having grown some 2,700% over the 8 year period (55% CAGR). We did not assume average growths in this exercise, we used the exact achievements for each share for each of the 8 years when simulating the investment period. Deciles 7 through 10 performed the best, whilst also outperforming GRP which is an equal weighted JSE index (as opposed to the ALSI which is a market cap weighted index.) Note in the right hand picture how the Decile 10 returns were not due to a few years here or there of stellar growth but steady, consistent out-performance over all the other deciles.

SUCCESS RATES (BASE RATES)

Here we examine another important characteristic of the above 1-year holding strategy, namely its historical tendency to outperform the market as a whole. This is very important as very few investors can stand by and be patient with a strategy whilst it is under-performing the market for extended periods (even if its known to be a superior strategy over time). For our base rates, for each of the 8 years under comparison in the previous section, we created two portfolios called GRP (an equal weighted index of the JSE universe) and ALSI which is the standard JSE index with market-cap weighting. We feel it is important for a strategy to beat BOTH these peer groups at least 66% of the time to not deter the average investor. The below table shows what % of the time during the whole 8 year test period, each decile portfolio beat these two benchmark indices:

We see that the Decile-10 strategy out-performed the ALSI 88% of the time (7 years out of 8) and the GRP index for every one of the 8 years (100%). Expensive "glamour" shares constituted by the Decile-1 strategy failed to beat the ALSI 50% of the time and failed to beat GRP 75% of the time. Although Decile-8 and 9 were also strong performers in the total return tests, we can see their base rates would probably have led to investors ditching the strategy after a few years of under performance.

The extent of the out-performance of the benchmark indices by the Decile-10 Price/Book portfolios is shown below:

RISK ADJUSTED ANNUAL CONSISTENCY OF GROWTH

Here we look at the average per-annum growth achieved by each decile over the 8 years, compared to the standard deviation of said growth over the 8 year test period to derive a risk-adjusted score. This examines volatility of the per-annum growth of the respective portfolios as opposed to the volatility of their underlying shares. It gives us a sense of the "consistency" of the strategy over the 8 year period. Did it fluctuate wildly year-to-year, providing an uneasy ride for the investor, or did it provide smooth, consistent returns?

We see that Decile-10 was the clear winner here with Sharpe ratio of 1.7. This means the average annual growth for Decile-10 portfolios exceeded the standard deviation (volatility or fluctuations) of annual growth by 1.7 times. Every 17% of growth of Decile-10 portfolios only experienced a 10% fluctation on an annual basis. Generally figures above 1 are acceptable to the investor.

CONCLUSION

Whilst 8 years is by no means an exhaustive back-test, the performance of the low price/book indexing strategy is still impressive. We continue to extend our back-tests all the way to 1990 for this strategy, but felt it prudent to publish our findings to date, since it is merely a confirmation of extensive international research anyway.

As with the international research, we conclude that the price you pay for a share has a significant bearing on ultimate returns. Cheap, out of favour stocks eventually come back in favour and reward the investor. Bear markets are the best times to execute this strategy as the under-pricing by the market on low price/book stocks is even more extreme on the down and the overpricing more extreme on the up (see our 2003 Crash trough-to-peak back-tests)

You can do a lot worse in the market than by buying the 10 or 20 lowest Price/Book stocks on the JSE. You are likely to beat the market in the medium to long term and not experience too wild a ride on the way to the top. Later on, we publish in the subscribers section how combining Price/Book with other factors significantly improves performance whilst increasing base rates, decreasing volatility and decreasing required portfolio sizes. A classic example of this is the combination of a Piotroski score with low price-book stocks, which yields tiny portfolios that deliver stunning long term performance. We have completed an extensive 14-year back-test on a Piotroski coupled with Price/Book which you can go see over here.

By replicating the comprehensive Brandes Institute Price/Book exercise for the JSE (go read about this in the "Value Power" menu), we were hoping to provide strong confirmation of trends they had discovered regarding value investing. We aim to discover just how much the Price/Book, or Price/NAV ratio of a JSE share has an impact on portfolio growth. We will show, as with research done on international markets, that portfolios created from low Price/Book shares significantly out-perform all other shares and the indices over 1, 3 and 5 year periods.

METHODOLOGY PRIMER

We will test multiple periods commencing in 2000 (we are busy working on additional analysis back to 1990 for our subscribers only.) In each period we will sort the JSE by Price/Book ratio as at 30 March, based on the most recent financials of each company at that date, be they finals or interims (this ensures fresh fundamental data is used in deriving the Price/Book ratio). We then divide the sorted list into 10 sets of equal deciles, with Decile-1 containing shares with the highest 10% of Price/Book ratios (expensive shares) and Decile-10 containing the lowest ("value shares".) We create portfolios from each decile and then measure each portfolios' 1, 3 and 5 year subsequent growth. Growth is measured using the last day in March of each year as start and end dates.

We created five sets of deciles to measure average 5 year growth, namely 2000-2005, 2001-2006, 2002-2007, 2003-2008 and 2004-2009. There are 2 additional sets of deciles for the 1 and 3 year growths, namely 2005-2008 and 2006-2009. There is also one extra decile set for the 1 year growth, namely 2007-2008. So there are 5 sets of 5-year deciles, 7 sets of 3 year deciles and 8 sets of 1 year deciles. Decile sizes ranged from 23-30 shares depending on year. The tests have thus exposed the 1,3 and 5 year periods to the latest 2009 crash as well as the 2000 meltdown. Since there are two full boom and bust cycles in this test period it is a good indicator of strategy performance.

Finally we averaged the results for each decile (5 sets of results averaged for 5 year growth, 7 sets of results averaged for 3 year and 8 sets averaged for 1 year growths), and produce summary findings together with analysis of decile portfolio characteristics such as risk adjusted return, volatility, base rates and total return. This ensures a holistic method of evaluating a strategies performance instead of chasing strategies with seemingly the highest growth rates.

This set of tests represents the entire JSE universe ("ALL STOCKS"). We will also shortly publish the same tests conducted on universes that for each period eliminates the bottom 25% (to create a "LIQUID STOCKS") and bottom 50% ( creates a "LARGE STOCKS" universe) of the JSE by market capitalisation, to look at how the strategy performs with the "micro-cap" shares excluded, and how it performs on larger issues more common with fund managers/unit trusts. This data will be for subscribers only.

SUMMARY FINDINGS FOR "ALL STOCKS"

We see breathtaking confirmation of international tests, for all of 1, 3 and 5 year portfolio holding periods. On average, during the period 2000 to 2008, creating a portfolio of the lowest Price/Book shares on the last day in March in any year would have yielded a 52% CAGR if you held those shares for 5 years, 57.6% if you held onto them for 3 years, and 58.7% if you just hung onto them for 12 months. This is emphatic confirmation that low price/book deciles (9 and 10) substantially outperformed high price/book deciles, with significant out-performance by decile-10 which contained shares with lowest price-to-book ratios of 0.38 and less. The results on the JSE are amplified enormously compared to other larger stock exchanges where similar research has been conducted.

STANDARD DEVIATION

It is one thing having impressive average growth rates, but quite another if volatility is through the roof. Here we will examine decile-portfolio growth, versus standard deviation of underlying shares together with an averaged risk adjusted return ratio (Sharpe ratio=growth divided by standard deviation).

We see that Decile-10 provides, on average, a superior risk adjusted return by deviation of portfolio constituency (as indicated by the Sharpe ratio). In the 5-year periods, Decile 10 provided the 3rd best Sharpe ratio, although at a respectable 0.74, it was greater than the 0.55 for the 1-year periods and 0.7 for the 3-year periods. Note how the Sharpe increases as the holding period increases.

These graphs dispel the myths that expensive shares are "safe", as the lower decile shares have much lower Sharpe ratios than the higher ones. In the 1-year example, Decile-1 not only underperformed but the expensive shares that it was holding gyrated wildly resulting in a gut-wrenching Sharpe of 0.25! This meant many shares in the portfolio were decimated whilst others went through the roof. Unfortunately more went through the floor than through the roof as the anaemic average growths show.

TOTAL RETURNS

We now evaluate total return delivered by a 1-year holding period strategy over the 2008-2008 period. We start in March 2000 and for each Price/Book Decile, we take a hypothetical R1.00 and use it to buy equal rand amounts among the underlying shares of the portfolio. We hold the portfolio for a year, after which we rebalance it (sell everything) and repeat the process (8 times) with the funds we have over. We exclude broker fees etc. (which will be about 2% each year per portfolio, 1% for the buying and 1% for the selling.)

The returns for Decile-10 are rather remarkable, with your R1.00 having grown some 2,700% over the 8 year period (55% CAGR). We did not assume average growths in this exercise, we used the exact achievements for each share for each of the 8 years when simulating the investment period. Deciles 7 through 10 performed the best, whilst also outperforming GRP which is an equal weighted JSE index (as opposed to the ALSI which is a market cap weighted index.) Note in the right hand picture how the Decile 10 returns were not due to a few years here or there of stellar growth but steady, consistent out-performance over all the other deciles.

SUCCESS RATES (BASE RATES)

Here we examine another important characteristic of the above 1-year holding strategy, namely its historical tendency to outperform the market as a whole. This is very important as very few investors can stand by and be patient with a strategy whilst it is under-performing the market for extended periods (even if its known to be a superior strategy over time). For our base rates, for each of the 8 years under comparison in the previous section, we created two portfolios called GRP (an equal weighted index of the JSE universe) and ALSI which is the standard JSE index with market-cap weighting. We feel it is important for a strategy to beat BOTH these peer groups at least 66% of the time to not deter the average investor. The below table shows what % of the time during the whole 8 year test period, each decile portfolio beat these two benchmark indices:

We see that the Decile-10 strategy out-performed the ALSI 88% of the time (7 years out of 8) and the GRP index for every one of the 8 years (100%). Expensive "glamour" shares constituted by the Decile-1 strategy failed to beat the ALSI 50% of the time and failed to beat GRP 75% of the time. Although Decile-8 and 9 were also strong performers in the total return tests, we can see their base rates would probably have led to investors ditching the strategy after a few years of under performance.

The extent of the out-performance of the benchmark indices by the Decile-10 Price/Book portfolios is shown below:

RISK ADJUSTED ANNUAL CONSISTENCY OF GROWTH

Here we look at the average per-annum growth achieved by each decile over the 8 years, compared to the standard deviation of said growth over the 8 year test period to derive a risk-adjusted score. This examines volatility of the per-annum growth of the respective portfolios as opposed to the volatility of their underlying shares. It gives us a sense of the "consistency" of the strategy over the 8 year period. Did it fluctuate wildly year-to-year, providing an uneasy ride for the investor, or did it provide smooth, consistent returns?

We see that Decile-10 was the clear winner here with Sharpe ratio of 1.7. This means the average annual growth for Decile-10 portfolios exceeded the standard deviation (volatility or fluctuations) of annual growth by 1.7 times. Every 17% of growth of Decile-10 portfolios only experienced a 10% fluctation on an annual basis. Generally figures above 1 are acceptable to the investor.

CONCLUSION

Whilst 8 years is by no means an exhaustive back-test, the performance of the low price/book indexing strategy is still impressive. We continue to extend our back-tests all the way to 1990 for this strategy, but felt it prudent to publish our findings to date, since it is merely a confirmation of extensive international research anyway.

As with the international research, we conclude that the price you pay for a share has a significant bearing on ultimate returns. Cheap, out of favour stocks eventually come back in favour and reward the investor. Bear markets are the best times to execute this strategy as the under-pricing by the market on low price/book stocks is even more extreme on the down and the overpricing more extreme on the up (see our 2003 Crash trough-to-peak back-tests)

You can do a lot worse in the market than by buying the 10 or 20 lowest Price/Book stocks on the JSE. You are likely to beat the market in the medium to long term and not experience too wild a ride on the way to the top. Later on, we publish in the subscribers section how combining Price/Book with other factors significantly improves performance whilst increasing base rates, decreasing volatility and decreasing required portfolio sizes. A classic example of this is the combination of a Piotroski score with low price-book stocks, which yields tiny portfolios that deliver stunning long term performance. We have completed an extensive 14-year back-test on a Piotroski coupled with Price/Book which you can go see over here.