Diversifying to low risk investments

Diarmuid

I have some money on € deposit. To this point I have been happy to leave the money in a low interest account while the world market is so volatile (I already have exposure to equities). However with the current political turmoil in the euro I am considering diversifying that money on deposit to low risk (mostly non-euro) investments.

So what I was considering was investing in foreign government bonds and US treasuries. I realise that this approach will leave me open to a substantial currency risk but I can't see the Euro gaining substantially on $/yen in the near future.

To achieve my goal I have been looking at Dublin domiciled ETF, specifically iShares Citigroup Global Government Bond (EUN3) and iShares Barclays Capital $ Treasury Bond 7-10 (IUSM). I was also considering the iShares eb.rexx® Government Germany 5.5-10.5.

Any thoughts or comments on this? Anyone recently done something similar?

Raskolnikov

The fact that you've titled this thread as diversifying into "low risk investments" suggests to me that you should be extremely careful about moving into US Treasuries. Take a look at the chart of the 10 Year US Treasury here. Despite the fact that people talked about there being a bubble in Treasuries in 2007, 2008, 2009, 2010 and today during 2011, you would have made money in 4 of those 5 years by being invested there; hence why the funds you've listed have scored spectacular 50% returns in that period.

Honestly, I think people buying long-term paper at those yields are going to get badly burned. If yields even go anywhere near the historical norm, you are going to get destroyed.

Diarmuid

Thanks for the the reply!
I was basing my "low risk" on the long term returns for bonds. Here is the summary of total returns from 1926 to 2001
Series|Arithmetic Mean|Std Dev
Intermediate term gvt bonds|5.3|5.6
US Tbills|3.8|3.4
Inflation|3.2|4.7

A lot of money has poured into bonds, true,

I think I'll have a good think/read about this over the weekend

soddy1979

Diarmuid wrote: »

Thanks for the the reply!
I was basing my "low risk" on the long term returns for bonds. Here is the summary of total returns from 1926 to 2001
Series|Arithmetic Mean|Std Dev
Intermediate term gvt bonds|5.3|5.6
US Tbills|3.8|3.4
Inflation|3.2|4.7
A lot of money has poured into bonds, true,

I think I'll have a good think/read about this over the weekend

You know in the short term you may be OK, because it's difficult to see a major increase in rates unless we see a major pick up in the global economy, or major inflation. Nevertheless, I am of the same opinion as Raskolnikov, and I only think long term rates are going one way.

Regardless of your circumstances, it makes sense for you to be in multiple asset classes anyway, to offset some of the idiosyncratic risk in the other.

Since you obviously like standard deviation as a measure of risk, you can calculate the standard deviation of a two asset class portfolio using the following formula:

s(p) = [(w1)^2(s1)^2 + (w2)^2(s2)^2 + 2(w1)(w2)(s1)(s2)(p1,2)]^0.5

But you should remember, that standard deviation is based on an assumption of normally distributed returns, and bond returns have negative skew - I like Historical VaR.

ROK ON

soddy1979 wrote: »

You know in the short term you may be OK, because it's difficult to see a major increase in rates unless we see a major pick up in the global economy, or major inflation. Nevertheless, I am of the same opinion as Raskolnikov, and I only think long term rates are going one way.

Regardless of your circumstances, it makes sense for you to be in multiple asset classes anyway, to offset some of the idiosyncratic risk in the other.

Since you obviously like standard deviation as a measure of risk, you can calculate the standard deviation of a two asset class portfolio using the following formula:

s(p) = [(w1)^2(s1)^2 + (w2)^2(s2)^2 + 2(w1)(w2)(s1)(s2)(p1,2)]^0.5

But you should remember, that standard deviation is based on an assumption of normally distributed returns, and bond returns have negative skew - I like Historical VaR.

Diarmuid wrote: »

Thanks for the the reply!
I was basing my "low risk" on the long term returns for bonds. Here is the summary of total returns from 1926 to 2001
Series|Arithmetic Mean|Std Dev
Intermediate term gvt bonds|5.3|5.6
US Tbills|3.8|3.4
Inflation|3.2|4.7

A lot of money has poured into bonds, true,

I think I'll have a good think/read about this over the weekend

Neither Std Dev or VaR are measures of risk. the are measures of deviation.
Risk is entirely different. Outside of risk mgt depts and academic circles, regard risk the the liklihood of a permanent impairment of ones capital. It cannot be measured, only gauged IMHO.

Back to the OP's question.
To invest in the type of Govt fixed income protfolio now that you metioned (specificallly the durations that are quoted in those etfs) you need to believe that inflation will remain pretty low if you intend holding those till maturity. If you are a deflationist or believer in low levels of inflation then these may well be astute investments. It really is that simple.
Now, balance sheet recessions and deleveraging tend to occur in tandem with low infltaion or deflation. However to offset that, fiscal problems that necessitate large amounts of printing money have tended to be followed by periods of strong inflation. By string think greater than 7-10% per annum. The instruments that you mention would suffer very badly if that was an outcome.

Personally I think we will get both. Low inflation/deflation now (next 2 years), followed by strong inflation over the medium term. Would you consider buying inflation linked bonds (TIPS)?

soddy1979

Eh.... standard deviation is a measure of risk, and so is VaR.

Value at Risk

What do you think it calculates, if it doesn't calculate probable impairment of capital?

ROK ON

soddy1979 wrote: »

Eh.... standard deviation is a measure of risk, and so is VaR.

Value at Risk

What do you think it calculates, if it doesn't calculate probable impairment of capital?

It is a measure of how a price of any security or fund, deviates from some averaging of that price trend in the past over a predeterminded history. If it was of any use whatsoever, we would not have had LTCM or the globa. financial metdown. Every investment bank or fund that I have seen deny it is in trouble usually rushes out its VaR.
Reardless of what type of talent one has at statistical analysis, you cannot know what the future path of any price series is. You cannot know that because you cannot predict the future. That being the case, it serves of no use IMHO as to what the liklihood of a future event where capital is gained or lost will be.

The CDS of most periheral PIIG paper had a very very low Std Dev and VaR up until 3 years ago. That has changed now.
Likewise, most CDO's had a low std dev up until about mid 07, the same with muni bonds till about mid 08, the same with most irish equity funds between 02 and 07.
Having spent 12 years listening to morons working in risk management depts at various institutions climing they understood risk and could model it, my answer has always been the same. The are reasonably efficient instruments at telling us what we already understand, they are very poor at telling us what we do not know. Most risk is contained in what we do not know.
For example, if we take the OP's original question, I think the success or failure of that strategy depends in part on the future path of inflation. I have an opinion on where inflation will go, but is is only that - an opinion. I cannot know for any certainty where inflation will go. Furthermore, I cannot know if states will change their calculation of inflation. These represent risks in the context of the proposed investment. I know that they are risks but I cannot measure them. Not only cannot I not measure them, the fact that I know what the std dev has been does not help me ascertain what the future liklihood of success/failure of the proposed investment will be.

If you disagree then that is your choice. I simply have a very different undertanding and grasp of what risk is than most people that I have met that are wedded to VaR methodologies.

Diarmuid

I see where you are coming from but I don't fully agree with you.

Yes, std dev gives you a good grasp of the risk factor based on the previous history. The length of that history is quite important though. When I wake up tomorrow, based on the substantial history we have, there's a pretty low risk that sun will not have risen. However you can't predict the future so you can't say for sure, that it rise.

Similarly we have returns and std dev for the S&P500 that goes back almost 100 years, which gives us a good confidence in the risk of investing in it. (obviously not sun-rising confidence )The CDS's you mention are very new financial instruments with a very short history (10 years) so I would not have a lot of confidence in the std dev figures

ROK ON

Fair enough. No time series analysis is in any way meaningful without having a sufficiently lengthy time series with which to compare against. However, I still disagree that it measures risk - and I would be passionate about this. It measures volatility. Risk is not volatility - even though the vast majority of business schools the world over still teach that it is.
For example, the long run time series of home prices in the US exhibited a low level of volatility until the point that it didnt. I am a professional investor and I live from my investments. I never use stddev of VaR to measure the riskiness of my investment choices. Furthermore, I do not know of any successful investors that do. I do know that the vast majority of banks, mutual duns and insurance companies are deeply obsessive about measuring std dev. Best of luck to them - they need it.

My point is that it is a poor choice to measure the riskiness of an investment. It is a perfectly valid instrument to measure how a spot piece of data deviates from its norm.

The investments that you mentioned have certain risk characteristics that stem from the level of inflation that they imply. An equity has a risk that stems from its level of valuation, its margin/revenue volatility and the strength or otherwise of its balance sheet. Commodity investments have risk that can be related to supply/demand imbalances and other environmental factors.
The use of std dev does not help here.

The fact that you look at 100+ years of data should not give you much comfort either, in that it is unlikley that you will be able to replicate that time series with your own investment.

Finally with most financial securities that I have studied, volatility is mean reverting over time (adjusted for the securities survivorship ). But when volatility changes it step changes.

Anyway, I hope that some of us have provided some sort of a framework for you to consider the investments that you mentioned. In all honesty, I believe that bonds are dangerously overvalued here and now, but the backdrop of global deleveraging to my mind means that in the course of the next 1-3 years they may become even more overvalued. It reminds me of Sorfos, when he said that the first thing he does when he sees a bubble is to buy into it. Then worry about getting out.

I have a few nice articles on the actual structure of risk and an appropriate way to think about it. If you would like to see them PM me. I can't post them up here as they are from a website protected site.

soddy1979

VaR measures the minimum amount your portfolio is expected to lose in a given period of time. It doesn't measure the size of the expected loss, and this is where I think its benefit is.

I wholly agree that statistics aren't good at telling you what is in the negative tail, but this also applicable to returns, or any other element of investment. At the end of the day, you can only base an investment decision on what is in front of you, and how you perceive how that will play out over time. And no matter what element of the decision you are looking at, be it macro, company fundamentals, etc., your choice as to whether to invest or not will be based on what your experience has taught you.

Rok On makes a good point concerning the time series, IE you won't be able to replicate its length. You should be very aware of structural changes in the time series as well. The major structural change shown in your chart is the bull bond market that started around 1980. This was driven by a consistent reduction in interest rates globally. Interest rates are on the floor now, so there is no way you would capture that part of the time series by investing now. Also, as Rok On said, inflation will destroy you unless you are in Inflation indexed bonds.