The purpose of this section is to provide an assessment of the recent historical per- formance of instruments traded on Japanese markets that an individual investor could reasonably invest in. The aim is to establish a foundation as to what the lived experi- ence of participating in the market would have been like in Japan and how the results of this participation differed from other markets. This is done under the assumption that the lived experience of recent years directly affects the individuals’ perception of opportunities and risks associated with an activity through word-of-mouth perpetuation or personal experiences. When considering an individual investor, this research assumes a modal resident of a nation without any particular degree of training or enthusiasm for personal finance, the markets or investing in general.
The section lays out the performance of the Japanese markets in the past three decades. This period closely corresponds to the Heisei Era, excluding the first half of 2019. The markets examined herein are limited to equities, the fixed-income and real estate investment trusts (REITs). While this excludes the real estate as residential prop- erties owned by the retail investors themselves, it is the contention of this author that these are best omitted at this point due not typically being considered under the general label of financial assets as well as due to the multifaceted nature of buying or owning residential real estate, which may be either used as a primary residence of the proprietor and thus non-investment related, or a quasi-variable annuity providing a reliable source of income. This raises the question as to whether such properties are viewed as first and foremost as investments or spending by the proprietors themselves. Furthermore, were
75. Ujiie, Japanese Financial Markets. Toyama University, Kin’yū bigguban to kongo no keizai [Finan- cial Big Bang and Future Economy], Web Page, 1998, accessed 1 May 2019, http://www3.u-toyama.a c.jp/furuta/bigbang.html.
these only seen as investments, any analysis of residential real estate performance is diffi- cult not only due to the illiquid nature of the market but also due to the obscure returns calculation that would have to include ongoing and one-off costs, work hours on behalf of the landlord and asset depreciation. Assuming that an average property owner would indeed correctly know what their returns were on a property they owned over several decades is in the opinion of this author gratuitously optimistic. Nonetheless, the matter will be briefly addressed in the later chapters within a different context.
As with the other parts of this research, the United States is used as the main com- parator due to its position as the largest market and its dominance of the financial sector.
Equities
Although the bubble famously burst in 1992 when land and real estate prices began to drop, the stock market had reached its highest valuation already by December 29, 1989, when the Nikkei 225 peaked at ¥38,915.87. This valuation has not been reached to this day almost thirty years later. Not only had the valuation not been surpassed, it has not even come close. As of the first trading day of the year, 4 January, 2019, the index opened at ¥19,655.31 or just barely above half of its peak price. Conversely, the United States’ S&P 500 returned approximately six times and the German DAX roughly five times the performance. This raises certain questions as to how far the same investment principles that may have been fostered in the American and German retail investors could have taken root in Japan.
Typically the performance of an index or any other publicly traded instrument is displayed as a chart above. A more neutral representation method will be given later on in this section. In nominal terms, the Nikkei has severely underperformed the three other comparator indices used in this section.
The aforementioned stale performance of the Japanese market and the precipitous drop from the bubble should be of no surprise to anyone even remotely interested in Japanese economic history. However, one may argue that it is worth not only looking at the performance from some selectively picked point but rather examine what the investment experience would have been like for someone investing in the past thirty years. Assuming that the age at which one would begin having a disposable income and thus may be interested in investing a part of it is around mid-twenties, one can simulate the possible investment outcomes for a presently mid-fifties retail investor curious in participating in the stock market. This particular age group is important not only because they were the generation that entered the workforce at the peak of the asset price bubble, but also because they are presently at the peak of their earning capacity and are likely to
0 2 4 6 8 10 12
Nominal returns for 1989-2019
Nikkei 225 S&P 500 DAX KOSPI (DAX Price)
Figure 3.5 Note that DAX is a total return index that accounts for dividends being reinvested rather than distributed to the investors. Its direct companion is the comparatively rarely discussed variant called the DAX Price index76, which does not account for dividend reinvestment and thus bears a closer resemblance to the other three indices. It is displayed in this chart for clarity, but will be omitted henceforth as it is not the default go-to national index that the vastest majority of unsophisticated retail investors—the group investigated in this research—would be looking at.
Data: Bloomberg, DAX Price Index: Deutsche Bundesbank. Resolution: One month.
have children entering or at early stages of employment.77 In other words, rather than
picking a specific date at which an investment was made and another date at which it was subsequently liquidated, one can provide a more neutral analysis by assuming both a random entry and a random exit. While this does not in practice reflect the retail investor behavior because it eliminates the behavioral patterns of the market as a whole, it provides the most systematic way of looking at actual market performance for a given time frame. To do so, one can calculate all possible trades within a given time frame. While this does not remove the bias of the data set being temporally, it removes the selection bias. Thus, on the one hand, taking a dataset from 1989 to 2019 will still show the market as performing worse than by taking a dataset from 1993 to 2019 or even better from 2009 to 2019. On the other hand, the resulting performance is still more evened out than if one merely selected 1992 as a starting year, 2019 as the ending year and ignored all the losses the investors would have incurred by buying and selling between those two points.
The S&P 500 for the said period provides a look at a well-performing market when examined through this lens and thus constitutes a good entry point. The index performs
1989
1994
1999
2004
2009
2014
2019
Year of purchase
1989
1991
1994
1996
1999
2001
2004
2006
2009
2011
2014
2016
Year of sale
11.82% loss transactions
7620 of 64440 observations
1.19% mean return
for 1 to 360 month sample
Purchase diagonal
Return multiple
All S&P 500 trades for 1989 2019
2
4
6
8
10
Figure 3.6 The triangular area displays all possible returns within a given period. To see a performance of a single trade, one selects the year when hypothetical security mirroring the market is purchased on the x-axis and the year it is sold on the y-axis. The blank white areas signify that the trade is made at a loss and no further distinction is made for the purposes of readability. The black to gray to white gradient areas signify that the trade would have been profitable. The brighter the gray, the higher the returns. The right-hand scale provides a legend for the return multiple.
The chart more easily understood when examining from the perspective of the “purchase diagonal”. When a security is purchased at any point on the diagonal, it can be sold at any point directly beneath it. The earlier the security has been purchased, the more points there are at which the security can be sold until the end of the sample time frame, thus resulting in the diagonal reducing the number of sell opportunities by one unit for every one unit moved further in time when buying.
Gray diagonal lines beneath the purchase diagonal mark the points at which five years have passed since the purchase. Each following line suggests another five years. The dotted lines thus correspond to 5, 15 and 25 years since purchase; the dashed lines correspond to 10 and 20 years since purchase. The data ends at 30 years and no line is plotted at this point.
Finally, the figures in the top-right corner of the chart indicate which percentage of all possible transactions would have lead to a loss regardless of the severity of the loss. The number beneath indicates the mean return for the sample. This and all following charts of this nature are calculated based on a 1-month resolution and a 360 month period equivalent to 30 years. This means that a trade can only be entered into or exited from at the closing price at the end of a month and no trades could be executed on either daily, intraday resolution or otherwise.
Note that this specific chart does not account for inflation, tracking errors of investible index funds or any of the associated expenses (12b-1 fees, management fees and so forth). Furthermore, note that every second tick of the y-axis is half of a five-year period and thus refers to the June of the corresponding year. These are the years 1991, 1996, 2001, 2006, 2011 and 2016. The tick method applies to all subsequent graphs. Data: Bloomberg. Resolution: One month.
much like one would expect of an index that lends itself to the popular buy-and-hold investment style that serves as a basis of many defined contribution retirement funds. The chart is a nearly uniform gradient of black to light gray except for three large “holes” in the middle. These are the Dot-Com bubble of the late 1990s and the 2007 United States housing bubble, the implosion of both of which has lead to a temporary decline in equity prices. The closer to the peak of the bubble the security is purchased, the longer does it take until the first black dots appear signifying a positive return. Purchasing the security during these months would have lead to the worst outcome both in the short and long term.
Furthermore, the chart demonstrates—again reinforcing the validity of the buy-and- hold strategy—that, outside of the two bubbles, the only times one would be in the red is by selling the securities close to the purchase diagonal. This hacksaw-like pattern is caused by the normal short-term fluctuations of equity prices, which significantly increases the chances of incurring a loss by selling the security too early. This is further emphasized by the gradient of the chart, the gray becoming brighter and brighter as one moves further away from the purchase diagonal with a bright gray nexus in the bottom left quarter of the chart. At this point, one is the furthest away from the diagonal and has been holding the investment for the longest period. Occasional brighter vertical line-like patterns suggest that the valuations have been particularly low during those months of purchase, leading to higher returns for the fortunate buyer. Conversely, the dark horizontal line-like pattern suggests the periods immediately after which the prices had dropped rapidly.
Given this data, it is possible to compute the likelihood of a random position being closed at a profit or a loss as well as calculate the mean return of a transaction. Only 11.82% of the S&P 500 index transactions had led to a loss. Furthermore, when consid- ering both trades at a loss and trades at a profit, the mean return for random trade entry and random trade exit was a positive 1.19%. Note that this number is merely a mean of all possible transactions and is thus not annualized.
Importantly, these results would look considerably different were one to separate trades by the length of the trade duration. Shorter trades would result in a considerably lower ratio of wins to losses as well as lower mean returns. Trades held for 15 or more years would lead to no losses whatsoever again reinforcing the validity of the buy-and- hold strategy. Not even the Dot-Com Bubble or the Global Financial Crisis would have led to a nominal loss for investors.
An important theoretical limitation to this approach that must be emphasized is that the loss ratio and mean returns are susceptible to the length of the examined data. All else equal, a ten-year sample would have a greater ratio of losses and lower meaner returns than a fifty-year sample. In other words, the excellent performance of the long-term
trades compensates for the bad performance of the short-term trades. Conclusively, the calculated values in isolation in no way reflect the performance of the S&P 500 itself and are only useful as comparators to other indices or the same index with the sample taken from a different period. As the focus of this research is on current investor behavior, only the measures of the performance of the most recent period are provided.
Before examining the Nikkei itself, it is important to emphasize that the United States chart pattern is not usual for a healthy or even a moderately healthy economy. Below are the charts for the German DAX and the South Korean KOSPI general market indices for the same period. Note that both countries have undergone considerable turbulences such as the German Reunification in 1990 and the South Korean exposure to the Asian Financial Crisis in 1997, whereby latter lead to a large number of insolvencies and con- solidations in the financial sector as well as a $58.4 billion bailout package from the International Monetary Fund (IMF).78
1989 1994 1999 2004 2009 2014 2019 Year of purchase 1989 1991 1994 1996 1999 2001 2004 2006 2009 2011 2014 2016 Year of sale 11.43% loss transactions 7368 of 64440 observations 1.41% mean return for 1 to 360 month sample Purchase diagonal
Return multiple
All DAX (TR) trades for 1989 2019
2 4 6 8 10 1989 1994 1999 2004 2009 2014 2019 Year of purchase 1989 1991 1994 1996 1999 2001 2004 2006 2009 2011 2014 2016 Year of sale 18.92% loss transactions 12192 of 64440 observations 0.89% mean return for 1 to 360 month sample Purchase diagonal
Return multiple
All KOSPI trades for 1989 2019
1 2 3 4 5 6 7 8
Figure 3.7 Data: Bloomberg. Note that the DAX is a total return index as explained further below.
Whereas Germany has closely followed the pattern of the United States, Korea exhib- ited a similar pattern, albeit with a greater number of loss periods largely attributable to the turbulence of the 1990s. The Global Financial Crisis has had a visibly lower impact on the South Korean valuations than it did on the Euro-American equities. The bright gray vertical line indicates that a market timer would have outperformed the earliest buyers by entering into a trade at the 1998 bottom of the Asian Financial Crisis, but one can nonetheless see that an investor holding a KOSPI tracking product for more than fifteen years would not likely have suffered a loss. The same cannot be said for Japan.
As has been noted previously, the Nikkei 225 index of 2019 still has not reached its
78. Kim Kihwan, “The 1997-98 Korean Financial Crisis: Causes, Policy Response, and Lessons,” in
1989
1994
1999
2004
2009
2014
2019
Year of purchase
1989
1991
1994
1996
1999
2001
2004
2006
2009
2011
2014
2016
Year of sale
64.08% loss transactions
41294 of 64440 observations
-0.05% mean return
for 1 to 360 month sample
Purchase diagonal
Return multiple
All Nikkei 225 trades for 1989 2019
1.00
1.25
1.50
1.75
2.00
2.25
2.50
2.75
3.00
Figure 3.8 Data: Bloomberg.
previous peak nearly thirty years later. However, this is not merely the case of an unlucky investor entering the market at the very peak of the bubble as is popularly vocalized in the cautionary tale of the Japanese markets, but rather a persistent state that only began a turnaround following the enactment of Prime Minister Shinzo Abe’s Abenomics reforms. Signs of recovery can be seen in the early 2000s time frame, but are immediately cut short by the burst of the U.S. American subprime bubble. Also notable is the fact that the Lost Decade indeed appears to be offering a bare minimum of nominal returns on the profitable trades, which are few and far apart.
Contrary to the previous three markets, one must also note that the nominal loss rate for random entry and exit transactions is excruciatingly high. Whereas the S&P 500’s loss rate lay at below 12%, the DAX below 12%—although this had been largely due to how the index is constructed rather than the real performance the of the market— and the KOSPI at under 19%, the odds of entering into a losing trade in the case of Japan remained above 64%. Furthermore, not only did the index expose any prospective investors to a greater degree of risk, it is the only of the four examined indices that had a negative nominal mean return.
Risk It nonetheless needs to be stated that despite the poor performance, the Japanese equities did not exhibit a high degree of risk as expressed by 30-day volatility.
0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50%
30-day price volatility 2-year SMA
Nikkei 225 S&P 500 DAX KOSPI
Figure 3.9 The chart displays the Japanese two-year trailing average of the 30-day volatility for the corresponding indices. The time frame corresponds to the same 1989 to 2019 period as previously with each point corresponding to the mean 30-day volatility of the previous two years. The two-year moving average is used to visualize otherwise unreadable data.
Data: Bloomberg.
The Nikko Research Center research points out that Japanese equities have performed poorer than those in the United States while being associated with greater risk.79 This
is correct, but perhaps more so a statement about the volatility of equities in the United States than those of Japan. As seen in the graph above, although the United States indeed had lower volatility than Japan, it also had a considerably lower volatility than both Germany and South Korea.
All three countries excluding the United States were undergoing considerable transfor- mations during the said period. Korea, which is still typically included in the “emerging markets” indices rather than the “developed markets” ones that the other three coun- tries belong to, had just barely entered its Sixth Republic era of democratization and dismantling of authoritarianism. Not even a decade later, it had fallen as one of the main victims of the Asian financial crisis. Germany had found itself in the process of reunifi- cation and reintegration of a former socialist republic into the capitalist system. From 2009 onward, the country remained involved in the European sovereign debt crisis and
the subsequent intra-European transformations. Japan had seen the burst of the asset price bubble, followed by the Lost Decade and eventually the introduction of Abenomics. By contrast, the United States did not go through any particular process that had not also been transferred to a significant degree to the other three countries, whether it be the run towards the technology stocks bubble in the late 1990s, the outbreak of the wars in the Middle East following the attacks on the Twin Towers or the subprime crisis and