Thursday, September 24, 2009 2 comments ++[ CLICK TO COMMENT ]++

Bond yields during three deflationary busts - 1990's Japan, 1930's USA, 2000's USA

(Sorry about the lack of posts. I wrote this post over several months so the text is a bit rambling and incoherent. Hopefully someone finds it useful :) )

Deflation is probably the last thing on anyone's mind. After all, gold has surpassed $1000, oil is up over 100% from its low this year, and, well, stocks are on fire. But then again, this is a contrarian blog and I don't quite think like the mainstream. This doesn't necessarily mean that I am right so it's up to you to figure out if my thinking is right.

I haven't altered my stance, which amounts to a tilt towards deflation. Apart from specific stocks, I have been investigating a macro bet on deflation. Apart from shorting select assets, you really don't have many choices if you are betting on deflation. The simplest bet is to overload on cash—cash is king during deflation—but the classic investment is long term government bonds (assuming the issuing government is solvent.)

As I mentioned a while ago, Hugh Hendry opted to enter into swap contracts (or maybe options contracts) on interest rates and that is one option worth looking into. Interestingly, this appears to be opposite what Francis Chu was contemplating (however Chu said he will only enter into the contracts when deflationary fears are high.) In any case, this is something that I have thought about but haven't done much homework on it yet. I have always said to myself that I won't purchase derivatives because they are a zero-sum game (I have zero confidence that I can outsmart Wall Street and professional investors.) But I think an exception can be made for macro bets. This might run counter to some people's thinking but I think it is probably "safer" to make a macro bet using derivatives than it is to use derivatives to place bets on specific stocks.


In this post, I will be discussing the behaviour of bond yields during the 1930's, 1990's Japan, and the last 10 years. The first two periods are major deflationary busts and the present period may also be one. The purpose of this analysis is to figure out how low bond yields may fall. That is, if one is going to bet on bonds due to deflationary expectations, they need to figure out how low bond yields can fall (because the big returns are in capital gains and not interest.) This is important because bond yields have already fallen quite a bit in the last two years (if 10 year bond yields were over 5%, the bond case is more obvious.) If they can't drop much further, there is little benefit to betting on bonds even if you expect deflation.


Notes About the Data

I will be painting with a broad brush and everything I say should be thought of as rough ideas rather than anything concrete. There are marked differences between the periods that one can't assume anything will repeat in the same way. For example, flow of information was more limited in the 1930's than now; for all intensive purposes, the US government took control of the Federal Reserve during the 1930's whereas that hasn't happened yet (from what I understand, the Japanese Central Bank is not very independent and can also be thought to be heavily influenced by the elected government); the poor demographics of Japan automatically results in a big deflationary pressure whereas that's not the case with present USA or 1930's USA; and so on.

All yields are for 10 years government bonds. The Japanese government bond yield is from Bank of Japan and I'm using the TSE-listed bond futures, which are very close to the actual government bond yield. I'm using the bond futures for cotinuity reasons (if you look at the BOJ source spreadsheet, you'll see that the official bond yield listings are broken up into two sections.)American data come from Robert Shiller's online data.

Bond Yield - 1930's USA

The chart below plots the 10 year US government bond yields during the early part of the 20th century.



What I am interested in is the deflationary late 20's to early 40's, but I'm plotting a bigger time period to illustrate how bonds were in a bull market long before the onset of deflation. The bond market started a bull market somewhere near 1920 and stayed in a bull market for a very long time. However, keep in mind that corporate bonds (not shown) were a complete disaster during the Great Depression and foreign soverign bonds of the weaker nations (not shown) were another disaster.

From the US government bond market perspective, the Great Depression wasn't really noticeable. There was a minor sell-off in late-1928/early-1929 but it was largely business as usual. (As a side note, this is probably what saved wealthy investors and insurance companies back then. Many more people owned government bonds back then and their wealth was largely preserved—in fact it went up—during that period. In contrast, deflation will be far more ruthless now since ownership of bonds is far lower now, with higher ownership of stocks.)

The whole purpose of this exercise is to figure out how low bond yields can fall. Unfortunately we can't draw any conclusion from the 1930's. We can tell that bond yields kept falling but it's hard to say what is a reasonable minimum. The reason is because the US government essentially seized control of the FedRes during World War II. Once the FedRes started monetizing bonds en masse, the yield basically fell to a ridiculously low value of 0.2%.

If I were to speculate, I would say a 2% yield, hit in 1940, is probably a good minimum to expect during normal times; a more conservative estimate might be around 2.5%, which is what it was from (roughly) 1942 to 1944. Beyond that it's hard to be confident with anything because the yield was driven down due to special circumstances.

Right now, the 10 year US government bond yield is around 3.41%. So, we are probably looking at no more than another 1% decline in yield. If you assume a 1% move results in 10% change in price (rough estimate; actual result depends on duration and convexity of bond) then the upside to investing in 10 year bonds is around 10%, plus another 3% from interest. This is not too attractive IMO.

Bond Yield - 1990's Japan

If deflation materlizes in USA, I think it will resemble 1990's Japan and not 1930's USA. So, let's look at how Japanese 10 year government bonds behaved in the 90's.



Unlike 1930's USA, Japanese bonds were not in a bull market before the deflationary bust. So Japan presents a clean look at how bonds are profitable during deflation.

Bond yields peaked roughly one year after the stock market peak (for newbies, do keep in mind that a yield peak means a price trough i.e. falling yield is bullish for bonds). Yields have kept falling ever since, although they may have hit a bottom in 2002. (On a side note, rising bond yields is usually bearish for stocks, since you will discount stocks at a higher discount rate. But I believe if the Japanese bond market has entered a bear market, it will likely be bullish for Japanese stocks. We should start seeing a massive shift of capital from Japanese bonds to stocks if my guess is correct.)

The Japanese case resulted in bond yields hitting a trough of 1%. I would consider this low as a "clean" bottom because we had none of the crazy central bank interference as was the case during WWII in USA. There is nothing to say the present, let alone the interest rates in totally distinct country, has to match what happened in Japan. But if you do think that US bond yields will mirror Japan, then we are looking at another 2.4% decline in bonds, from the present 3.41%. This scenario is attractive to me. If I thought a 1% yield was a high probability outcome, I would have no problem tying up capital with 10 year bonds.

Bond Yield - 2000's USA

The verdict is still out on whether the last 10 years and perhaps the next decade will be considered as a deflationary period. Unlike many, I conside the stock market peak to have occurred in 2000 (in contrast, many consider 2007 as the major peak.) Price inflation (CPI), commodity prices, real estate prices, and so on, do not indicate deflation in the last 9 years but it's too early to say for sure. In real terms, many seemingly well-performing assets, like stocks or real estate, haven't done that well. Even at the stock market peak in 2007, stocks were nowhere near their 2000 peak in real terms. In fact, if you were a Canadian, you never saw US stocks come anywhere near the 2000 peak (because the US$ declined quite a lot.)



Bonds were in a long bull market that started in 1981 but there was a niceable sell-off in late the 1990's; there was a multi-year peak in yields that was set in 2000. Bond yields peaked approximately 3 months before the S&P 500 hit a very major peak (especially in real terms) on March 24th of 2000. It is possible that this stock market peak may not be bettered in real terms for 20 or 25 years.

Bond yields have continuously declined for the most part in the last 9 years. Yields started rising between 2003 and 2007 but it wasn't very large. The 2003 low made many think it was a multi-decade trough in yields (peak in bond prices) but it turned out to be incorrect. As we now know, the crash last year took yields all the way down to 2.44% in December of 2008 (based on monthly prices.)

Yields have risen this year and stand around 3.4%. The question is, can yields go much further down? Or if you are making a risky bet like Hugh Hendry, will yields stay close to the current levels?

Bond Yields of all Three Scenarios

What I am about to say may be a seriously flawed way of comparing different periods so use caution if you are deriving any conclusions. For instance, there is nothing to say that yields are comparable across countries or across distant time periods. Even with some risk, I think it's ok to compare yields because I liked to think that valuations sort of normalize over time and across countries. The numbers may not be identical but I like to think that they should be somewhat similar. This is similar to how macro investors generally assume that present P/E ratios should be similar to what they were 50 years ago. There is no reason present P/E ratios have to be anywhere near what they were 50 or 100 years ago but I like to assume it will be close. In a similar vein, I am assuming that present bond yields will resemble those during the Great Depression and Japan to a large degree.

Even with all these flaws, I'm pursuing this line of thinking because I don't know of any other way to determine how much further bond yields may fall (if at all.)

The chart below looks at the three scenarios covered above on one chart:



The starting point was picked arbitrarily as one year before the bond yields peaked. The selection of the starting point may make this analysis flawed but there is no way around it. In the next section I look at the same chart with equalized peaks (this captures percentage change) but that may not be any better.

Another way of comparing the scenarios is to simply plot the charts with the peaks aligned at the same point without adjusting the values (i.e. this would have the effect of simply moving up the low-value curves.) I think this is even more flawed and chose not to look at this.

What you believe depends on how you interpret bond yields. Should you look at raw bond yields? Yield peaked around 3.5% in 1929 whereas they peaked around 6.5% in 2000. If you believe 6.5% in 2000 is sort of "equivalent" to 3.5% in 1929 then you should look at the raw yields as plotted above. In contrast, if you believe that a 33% decline in yields from, say, 3% to 2% in the 1930's is equivalent to a 33% decline in yields from, say, 6% to 4% then you should look at some chart that equates the peaks.


Anyway, going back to the chart, which means we are looking at raw yields, we see that Japanese yields seem to have stabilized in a range between 1% and 2%. During the Great Depression, US government bond yields stabilized between 2% and 2.5% (ignoring the unbelievably low yield in 1945.) If we end up in a deflationary period and assume the present must be somewhat similar, I think a good guess is for yields to fluctuate between 1% and 2.5%. That's a huge range but the important point is that they are much loser than where we are today. This implies that if you believe in deflation, 10 year (or longer) Treasury bonds still look quite attractive.

Bond Yields of all Three Scenario (Normalized)

The analysis in this section could be flawed so use caution.

The following, final, chart plots the three scenarios with the peaks normalized. Note that this is not the same as simply aligning the 3 peaks! What you are seeing is percentage changes in their respective periods (except Japan, which is left as the base case.) The numbers that are shown on the chart, except for Japan, do not represent yields and should not be treated as such.



This chart produces different conclusions from the prior chart. In the prior chart dealing with raw yields, it appeared that bond yields can still fall much further if deflation sets in. In this chart, that is not obvious. Presently, bond yields look like they can still fall further if we follow the Japan scenario. However, we have actually declined more (in percentage terms) than during the Great Depression! As usual, I'm ignoring the crazy situation during WWII when the FedRes was literally buying all the bonds the US government was issuing (I believe this was the case in other countries involved in the war as well.)

It's worth reiterating what this chart is saying: bond yields have already fallen more than during the Great Depression. So, this chart introduces some doubt into any potential bond investment.

Overall, it's very complicated and one cannot blindly follow anything. As I said, both, during the Great Depression as well as the 2000's, the bond market was already in a massive bull market. It's always possible that yields already fell a lot more prior to 1929 whereas they may not have fallen as much prior to 2000. I never bothered analyzing what happened in prior periods and it's just too complicated to start looking at additional scenarios. For instance, if you start looking at pre-1929, you have to somehow factor in the gold standard (which is inherently deflationary and keeps nominal returns on financial assets low); similarly, if we start looking at pre-2000, the starting point can be tricky and one may need to consider the possibility that the 1981 peak in bond yields may be a spectacular 100-year bubble in yields (i.e. irrationally low bond prices.)

Don't Look For Solid Conclusions Here

I started writing this post a few months ago and I'm as confused as ever. Overall, there is nothing concrete to be drawn from any of this. Similar to how one cannot say whether the stock market deserves a P/E of 17 today, it's hard to say how far bond yields will fall if deflation materializes. It ultimately comes down what return market participants are willing to accept.

I personally think that raw yields are what matter. Therefore the 2nd last chart is what I would rely on. Based on that, long term bond yields can still fall. I would estimate a minimum yield between 1% and 2.5%, which is way below the current 3.4%. Even if they don't fall, they can easily go sideways, in a narrow band of 1%, for more than 5 years (which was the case during the Great Depression and Japan.)




Tags: ,

2 Response to Bond yields during three deflationary busts - 1990's Japan, 1930's USA, 2000's USA

Aki_Izayoi
September 28, 2009 at 1:51 AM

Do you think yield volatility has a role in determining treasury prices?

10 Year Treasuries at 2% might be justifiable in a deflationary environment and was justified by the fundamentals, but the technicals do not justify that because the 2% coupon does not compensate for upward yield volatility. I suppose one reason why Japanese government bonds have low yields because the volatility on that asset class is low . (although the JGBs bubble popped in 2003 causing a large 100 basis point rise in interest rates)

Sivaram Velauthapillai
September 28, 2009 at 2:14 PM

I am not too knowledgeable about bonds and maybe some readers can answer you question. My understanding from what you learn from basic finance courses is that volatility (in the sense of prices fluctuating) is not a factor in bond prices. Things like inflation expectations are a big factor but I don't recall any of my textbooks including volatility into their bond calculation.

But the above point is from a long-term point of view. Short-term traders and other investors may value bonds based on their volatility. Something that is fluctuating wildly may be perceived as being more risky. I don't know how traders behave.

I think the yields on Japanese bonds are so low because return on other assets are low as well. Return on stocks and real estate have been very low, and, in fact, negative at times. Investing in low-yielding Japanese bonds in the last 20 years has not been a mistake. It was a rational decision that outperformed practically all other assets.

Post a Comment