Curve Balls

from Dimensional Fund AdvisorsPredicting interest rate movements correctly is hard. Predicting them for a living is harder still. But getting it wrong is nowhere near as painful as the experience of those who lose their own money based on someone's forecast.

A year ago, the Reuters news agency polled a group of people closer than just about any other community to those who actually decide rate movements. These were 16 money market dealers who do business directly with the US Federal Reserve.1

The so-called primary dealers — banks or broker-dealers — are market makers for government securities. They consult directly with the US central bank and Treasury about funding the budget deficit and implementing monetary policy.

So if you wanted an informed view about the interest rate outlook, these might be the people you would call on first, which is what Reuters did when it asked the dealers for their forecasts for Treasury bond yields three, six and 12 months ahead.

Back in late September 2010, the dealers came up with a consensus forecast for US 10-year Treasury note yields rising from 2.50 per cent to 2.70 per cent in three months, 2.80 per cent in six months and to 3.20 per cent by September 2011.

So how did those forecasts turn out? Well, after three months, the yields had already surpassed the 12-month forecast at around 3.3 per cent. Another three months on, yields had topped 3.4 per cent, again well above forecasts. But then they started coming down again and by September 2011, were close to 2 per cent.

So the expert panel misjudged the trajectory for bond yields in terms of the magnitude of the increase in the first six months and then completely got the direction itself wrong in the subsequent six months.

But it wasn’t just the sell side that misjudged the market. In February, the world's biggest bond fund PIMCO announced it had reduced its US government-related debt holdings from 22 per cent in December 2010 to just 12 per cent in January 2011, the lowest in two years.

In March, PIMCO announced it had eliminated government related debt entirely from its flagship fund, saying that bond yields had reached unsustainably low levels given the scale of government debt obligations and the chance of a correction when the Federal Reserve ended its quantitative easing program.

But by August, PIMCO manager Bill Gross admitted he had made a mistake, telling the UK Financial Times that he felt like "crying in his beer", so badly had he misjudged the movement in bonds in 2011.

"Do I wish I had more Treasuries? Yeah, that’s pretty obvious," Mr Gross told the FT. "I get that it was my/our mistake in thinking that the US economy can chug along at 2 per cent real growth rates. It doesn’t look like it can."

None of this is to impugn Mr Gross' logic earlier this year in saying that the term risk of investing in government bonds was not worth the meager return.

But as tends to happen with forecasts, events intervene and those who maintained an exposure to Treasuries in 2011 have enjoyed solid returns in the intervening months.

The chart above compares the relative yields of US Treasuries at various maturities in January this year versus more recently in September. You can see that the curve was relatively steeper earlier this year than it is now.

The yield spread between the 10-year bond and the 1-year bonds was just over three percentage points in January. By September, this term premium had contracted to two percentage points. The change reflects news in the intervening period. Sentiment about the US economy has deteriorated in that time and investors have become more averse to taking term risk.

Put another way, when yields fall, prices rise. So those whose net exposure was relatively longer earlier in the year have enjoyed a capital gain that was not available to those who took a bet against Treasuries early this year.

Now, Dimensional's own research has shown there is a reliable relationship between current term spreads and future term premiums. So wider yield spreads predict larger term premiums, while narrower yield spreads predict smaller term premiums.

This is why we employ a variable maturity approach, varying the allocation towards short-term and intermediate bonds depending on the shape of the yield curve.

The advantage of this approach is we are only using information available in the market at the present time. There is no need for forecasts, which no matter how rigorous the underlying analysis can come undone as events and circumstances change.

The bad news is that financial markets have a tendency of sending even the most well informed and respected forecasters a curve ball. The good news is that you don’t have to take those sorts of risks if you don’t want to.

1. POLL: Rising Bond Yields Constrained by QE, Reuters survey, Sept 28, 2011