This article is a follow-up to the “short study of the recent Japanese crisis” which was published on this blog a few days ago. The study was actively discussed on the Bogleheads forum. Various interesting questions were raised that weren’t directly addressed by the original write-up and triggered the author to do more research about domestic tilts (or lack thereof), variable withdrawals, and past performance of the stock market in Japan.
What about eliminating the domestic tilt?
A key conclusion of the original study was that a Japanese investor would have been much better off by diversifying his/her asset allocation towards international equities (we used the MSCI World index to capture the history of corresponding returns). The study retained a domestic tilt in all cases though, by always keeping a percentage of the allocation in domestic Japanese stocks. What if we push the logic further?
Let’s compare two 60/40 portfolios, the first identical to one previously studied (40% World, 20% Total-Japan, 40% Bonds), the second eliminating any domestic tilt (60% World, 0% Total-Japan, 40% Bonds). Remember that ‘World’ includes Japan, and that in the 80s and 90s, Japan’s market capitalization represented a significant (and variable) proportion of the world’s equities, so the second allocation eliminates the tilt towards Japan stocks, but doesn’t eliminate Japanese market exposure. In addition, remember that we’re taking the perspective of a Japanese investor, using the Japanese currency (Yen) and its vagaries against the world’s currencies, as well as the Japanese inflation (very mild in this timeframe).
The following graph illustrates the trajectory of the two portfolios for Japanese investors retiring by the end of 1988 (one year before the peak of the bubble) using a naive constant withdrawal method (taking 4% of the initial portfolio value every year, while adjusting for inflation). The graph is constructed in ‘real’ (inflation-adjusted) currency. Unsurprisingly, the fully diversified investor would have fared better, being less exposed to the severity of the Japanese crisis following the bubble. Note that the 2008 financial crisis would have hit equally hard though, with a vertiginous 60% drop in both cases.
Now let’s keep the same asset allocation (60% MSCI World, 40% Domestic Intermediate-Term Bonds), and compare the perspective of a Japanese retiree (same as before, 1980-2016 average inflation around 1%) and the perspective of a US retiree (hence using USD currency and inflation, 1980-2016 average inflation around 3%). The following chart illustrates what would have happened to the portfolio of our two retirees.
Overall, the US trajectory is less hectic (notably in 2008/09). The last few years (2012+) divergence is explained by the vagaries of currencies, the Yen significantly depreciating against the US dollar in this time period. Actually, a careful look at the exchange rates largely explains where portfolio trajectories diverged on the entire graph. As a reminder, here is how the JPY/USD exchange rate evolved during the same time period.
- Investing with the world instead of using a domestically skewed asset allocation would have helped Japanese investors during this troubled time period, although not providing any protection against a worldwide crisis like 2008/09.
- Local currency rates can have quite a significant impact on returns provided by very similar investment vehicles over an entire retirement period.
Note: a knowledgeable reader pointed out that Wade Pfau has addressed a similar topic in 2014, studying globally diversified portfolios in local currency for 20 countries, and analyzing corresponding Safe Withdrawal Rates vs. a purely domestic strategy. See here.
What about using a variable withdrawal method?
Multiple graphs in the original study and in the previous section were based on a retiree using a ‘naive’ withdrawal method, based on the so-called “4% rule” from the famous Trinity study and the concept of Safe Withdrawal Rate. The pitfalls of such withdrawal method have been abundantly documented and we have seen in the original study that Japanese retirees could have been in deep troubles if they had used it blindly.
Let’s revisit some of the asset allocations used in the original study while using a variable withdrawal method, the Bogleheads VPW. In order to make the scenarios fairly realistic, we will make the following assumptions:
- A good part (roughly half) of the retiree spending budget (first year) comes from fixed income (e.g. social security, pension), which is inflation-adjusted over time (say 5% of 1,000,000 Yen, hence 50,000 Yen annually).
- The other part of the spending budget comes from portfolio withdrawals, starting with a 1,000,000 Yen portfolio. Withdrawals varies over time based on the PMT formula underlying the VPW method.
- The retirement period being illustrated on the graphs will be the usual 30 years, but the time period used in the PMT formula includes a grace period of 10 more years, foreseeing the fact that the retiree may actually live longer. Remember, VPW fully depletes the portfolio at the end of the time period, so one has better be cautious here (there are a lot of centenarians in Japan!).
- In real life, those amounts of money would have to be scaled by a factor of 100 or more given the low value of a single Yen, but we’ll keep those to ease the comparison with previously published charts.
Now let’s revisit some of our scenarios. Let’s take a 60/40 asset allocation, purely domestic, and show the (inflation-adjusted) trajectory of portfolio value and annual spending budget (combining fixed income with portfolio withdrawals), starting at the end of 1988. The portfolio value is shown as the left vertical axis (thin lines on the graph) while the annual spending budget is shown as the right vertical axis (thicker lines on the graph).
As expected, the constant withdrawal technique worked… until it failed (and then all that’s left is fixed income). While the variable withdrawal enforced enough discipline to provide a sustainable spending budget, albeit with a lot of belt tightening (the part coming from the portfolio dropped by roughly 50% over time).
Then let’s revisit the 60/40 scenarios with more international exposure, say 40% World, 20% Total-Japan, 40% Domestic Bonds. The “4% rule” would have been ok for this scenario, but this is plain luck. VPW would have adjusted the spending as the troubled times unfolded, allowing to splurge a bit more at times.
Finally, let’s get more aggressive with a 80/20 asset allocation, internationally diversified, while keeping 20% equities with Total-Japan. This would have provided for more spending ability with VPW, while possibly triggering severe heart burns in 2008/09…
Finally, note that we focused here on a very difficult starting point (end of 1988). The author ran a more comprehensive test with all starting years between 1980 and 1995, and VPW always displays an advantage on the average spending budget, often exceeding the “4% rule” by 20% to 30%.
Conclusion: variable withdrawals navigated the various difficult scenarios as expected, in a much more satisfactory manner than a constant withdrawal technique, adjusting the trajectory to the dynamics of the chosen portfolio. When accounting for a possibility of a true oddity like the Japan asset price bubble (in the midst of more rosy scenarios), there is just no question that a retiree should adopt a well-thought variable withdrawal plan.
What happened in Japan before the 80s?
The original study (as well as the previous sections) were based on detailed monthly data sets allowing to determine total returns (dividends included) for both Japanese stocks and bonds for 1980-onward.
Older historical data is harder to come with, notably for the dividends component. The DQYDJ Web site includes a Nikkei calculator which aims at addressing the challenge from a variety of sources, but the numbers seem odd (e.g. check 1952, and the hard-to-believe 23% dividend yield), and not quite consistent with other sources.
In order to get a sense of what happened, we can use two data series (Morningstar and FRED) which basically provide the same numbers, the *price* trajectory of the Nikkei 225 (the Japanese equivalent to the US S&P 500) since 1950, the inception date of the Nikkei 225. Global Financial Data reconstituted a price index extending the Nikkei series back to 1914, and the corresponding data series can be found on finfact.ie.
Let’s compare the price trajectory of the Japanese market with the US market (price only, from Prof. Shiller), starting from the end of 1926 (to focus on reasonably reliable historical data). This is a bit surprising, to say the least, with the Nikkei growing 7.5% a year overall, while the S&P 500 (again, no dividends included) grew roughly 6% a year.
The most glaring feature is that the US market displays a relatively steady growth since the 1930s, while the Japan market is just totally different from one period of time to another.
During the 1930s and 1940s, although much less impacted by the great depression which occurred in the US, returns in Japan went sideways then fell after the war, which is hardly surprising given how WW II left the country in shambles and starvation. World wars are truly horrific events, and no numbers could capture the corresponding devastation. This also lead to dramatic changes in the value of the Japanese Yen (see Wikipedia Yen).
Between 1950 and 1979, the compound growth (CAGR) of the Nikkei 225 (price only) approached a staggering 15%, only mitigated by the oil crisis, while the S&P 500 (price only) delivered a comparatively paltry 6.5%. Again, the Yen trajectory probably has a lot to do with it, with the Yen being apparently strongly undervalued until 1971, and its subsequent trajectory being a possible root cause for the 1989/90 asset price crisis.
Overall, Japan went though the horrors of WW II, rebuilt its economy, went on a tear, got hit by the oil crisis in the mid 70s, quickly recovered, went in another tear, then got struck by the Japan asset price bubble in the 80s, and finally faced the repercussions of the 2002 Internet crisis and the 2008 financial crisis in full force. More detailed charts from Nomura capturing all key events during such historical period can be found here.
This should put in question the ‘rosy lens’ that some US investors might be subject to while looking at their own stock market history. Maximizing the diversification of one’s asset allocation, while being flexible with portfolio withdrawals, seem to be strong requirements in order to be adaptive to an unpredictable future.