Bond Yields and Saving Behaviour in the US

What relationship, if any, consistently holds between interest rates and private savings behaviour? Do savings really go up when interest rates go up? And can excessively high savings ratios be brought down by keeping interest rates low? If so, what rates, and how low? And if savings rates stay high regardless does that mean we're in some sort of liquidity trap?

I should probably have had this down decades ago when I first encountered an ISLM graph, but the truth is that I've never met anyone in the market who actually uses ISLM analysis. Or mentions it.

Instead I have focussed on private cashflows and savings behaviour, usually looking at private sector savings surpluses and tracking their impact on bond yields. For emerging markets, this is often crucial: the most powerful financial dynamic bar none in emerging markets is what happens to government bond yields when a private sector savings deficit flips over into a surplus, or vice versa. In these cases, the cashflows are easy to trace: if an economy develops a savings surplus, then on a net basis, the private sector is dumping cash into a financial sector which, by definition, can use it to buy only government bonds or foreign assets. Hence bond prices rise and yields fall. Easy money.

Even in the massively-open and highly disintermediated financial system of the US, traces of the relationship remain.

Interesting though this is, it's hardly a complete theory linking bond yields with savings behaviour. Nor would one expect it to be: if private sector savings surpluses and deficits were the only determinant of bond yields, the world wouldn't have so many fixed income economists (and professional Fed watchers). And the financial world would never had heard of 'fair value models' for bond yields.

So let's look at those models. In my experience these fair value models regress and regularly recalibrate from three factors:

• policy rates
• inflation rates
• growth rates

A movement, or an expected/forecast movement in any one of these will change what the model signals to be the 'fair value' of a bond.

Anything that regresses and recalibrates enough will end up looking like it has useful explanatory power. Here's how my simple fair-value model of US bond yields compares with what actually happened over the last 21 years. It's not a superb fit, and even if it was, that would be testament simply to the power of serial recalibration rather than the theory it allegedly sets out to test.

Nonetheless, the conclusions which we can draw from this model are remarkably similar to those wrested from doubtless far more sophisticated models produced by our august Wall Street friends. And so are the conclusions are drawn by comparing actual bond yields with 'fair value' yields. What screams out in retrospect is that during 2004-2008 bond yields were far lower than 'fair value'. And from there it is but a step to conclude that the chief reason for that was that policy rates were set too low for too long, and, moreover, were expected to be kept too low for longer still. The graph serves as the charge-sheet against Alan Greenspan. And as it looks as if the same thing is happening again now (bond yields far lower than 'justified' by likely economic growth and inflation), we might eventually find it thrown into evidence against Ben Bernanke at some later date.

What the chart is saying right now is simple: bond yields are simply too low, making bonds an unattractive investment. Let's put it even more bluntly: who in their right minds would save to invest in bonds right now? At which point, we get to ask (and answer) an important question – regardless of the absolute nominal bond yield, does sufficiently 'bad value' in bond yields usually dissuade saving, and do sufficiently 'generous' bond yields usually encourage saving?

And I think we can answer than question empirically with a simple 'Yes'.

Consider the relationship between the deviation of bond yields from fair-value, and movements in the private sector savings surplus. The chart below illustrates it well, and that's not just because I've fiddled the axes. More importantly, the correlation between sequential movements in these two during the last 87 observations passes the 1% significance level quite easily.

For those of us interested in recent US economic history, and in global savings/investment imbalances, this chart is pretty irresistible, as it links the descent into major private sector savings deficit during 1997-2000 and again in 2005-2007 with bond yields being somehow maintained at levels which actively discouraged saving. When yields rose (relative to 'fair value') savings deficits were trimmed and reversed.

And now? Bonds represent absolutely rotten value, and as long as this is the case, the US private sector savings surplus will continue to decline, boosting US consumer demand at a pace slightly exceeding those of private sector income growth (widely defined).

One more thing: there is absolutely no sign of a Keynesian 'liquidity trap' anywhere on this chart.

A Prelude to Possible Armageddon

'Be forward-looking'. Right now, I'd rather not. And that's not only cowardice speaking, but a claim that in times like these, the blazing inferno of our forthcoming possible catastrophes is at the very least likely to blind us to our present actual condition.  Today, and tomorrow, if I have time, I intend to look back - to establish at least where we stand, even if several of our paths onwards lead over the precipice.

What present itself in the data - even up to May and June data - is the sheer buoyant health of the world economy, and its sheer normality. Let's remember what's actually happening in world trade, and what's happening to world demand.

First, the link between G3 demand and NE Asian exports is as tight as it has ever been, and, what's more, so far as we can tell, both have been interruptedly buoyant now since the middle of 2009.  Looking at G3 import data in dollar terms, by May US imports were rising 20.7% YoY, Eurozone imports were up 32.6% YoY, and Japan's imports were up 27.5% YoY. Overall, then, G3 imports were growing 26.9% YoY. What's more, sequential momentum remained positive, with the MoM growth in the three months to May rising 0.8 SDs faster than the historic seasonalized average, and the 6m of this momentum indicator riding at 0.92 SDs.  This is not immediately signalling the developed world economy is dangerously anaemic.

And, of course, NE Asia's exports machine is answer the call: by May NE Asia's exports were growing 14.8% YoY (China up 19.3%, Korea up 22.4%, Taiwan up 9.5%, and even Japan  managed a 1.8% YoY rise, despite the worst that earthquakes/tsunamis/nuclear accidents could throw at them. By June, my best guess is that NE Asian YoY export growth will come in again around the same (China up 17.9%, Korea up 13.6%, Taiwan up 10.8% and, based on the first 20 days, Japan up 9.2%). As the chart below shows, there's been a downward inflection in the 6m momentum trendline, but we're still ranging deep in positive territory.  And that's despite the multiple catastrophes suffered in the region's second-largest industrial economy.

Up to May and June, then, if you had to choose a description of the world economy judging from its trade data, it would be fair to use words like buoyant, or - if you're determined to be a grizzled pessimist - steady.

So it should be no surprise that you'd probably end up using much the same vocabulary if you looked at domestic demand data.  A word about how I do this. In each country, I look at what I consider to be the obvious monthly data-points for domestic demand: retail sales; auto sales; labour markets (employment and wages); construction and, where possible, service industries activity.  For each of these indicators, I develop the seasonalized historic expectation and measure monthly deviations from that pattern, expressed in terms of standard deviations.  For each country, I will then take a flat average of those deviations. When looking across countries (ie, for NE Asia, or for Europe, or for the world) I weight according to a 5yr average of dollar GDPs.

Here's how the situation looks up to May:

This ought to ring some bells: momentum accelerating throughout 4Q10 and 1Q11, followed by a sharp fall-off in April and May, with the beginnings of a recovery visible in June (probably - we've only partial data so far).  On a 6m trend basis, we have the same thing we saw in the trade data: an inflection point downwards in April, but still defiantly positive underlying sequential momentum. (Details: for 6m to June, the US is up 0.49; Europe is up 0.21; and NE Asia is up 0.05. For June itself, the relative strengths (on the data we have) are almost exactly reversed - ie NE Asia is up 0.78; Europe up 0.13; US down 0.26). 

Once again, one would conclude from this data that although the world economy's future may turn out to be catastrophic for several popularly identifiable reasons, its OK so far. (As the man said as he flew past the 30th floor window on the way down.)

Naturally enough, when an economist goes to the considerable trouble of constructing these indicators, he likes to use them to get a quick grab on likely GDP growth.  In the case of this demand model, the newly-normalized model (yes, shameless, I know) runs to an r-squared of 0.9 over the last 12 years when regressed against the IMF's global GDP result.  Last year, the IMF says the world economy grew by 5.01%, whilst this demand indicator suggested 5.2%. This year again, it's again telling us to expect the world will grow by around 5.2%. 

Unless, of course, the immediate future is dreadful.  Which it might be - it demands no highly-developed imaginative powers to envision it.  But as you ponder the possible trajectories, do at least remember that for all the analytical ink spilled, and for all the weaknesses already printed in the world's data, during 1H 2011 the world was growing quite nicely at a clip of around 5%.