An extraordinary graph

Here is a truly amazing graph. Yes, I know, 'amazing' is relative and risks sounding over the top, and what might be amazing to an economist mightn't hit the spot for anyone else, but it's still amazing.

It's from a Treasury Working Paper published in February, 'Measuring Savings Rates in New Zealand: An Update'. I missed it at the time, so I should credit Jeff Cope and Steffi Schuster at Stats for pointing me to it.

It shows the estimated household savings rate (savings as a percentage of disposable income), as measured by Statistics New Zealand at various points in time.

So. What do you see?

In 2004, the household savings rate as measured at the time (blue line) indicated massive dissaving: in 2004 the average household was earning $100 after tax and spending about $112. And the conventional explanation, then and now, was that people were spending the wealth they'd accumulated from increases in the value of the family home.

This sounds feckless, and lots of people worried about it this 'home equity withdrawal'. Maybe you worry, too, that people ought to take care about spending today on the basis of uncrystallised capital gains that might not be there tomorrow.

On the other hand, 'wealth effects' on consumption are widely documented. I collect stamps: if in the morning my collection is worth $100K rather than the $30K it might normally fetch, and I'm reasonably sure that I can realise the $100K or thereabouts, I could very safely spend $20K today on the strength of the new valuation. I'm making a realistic reassessment view on my consumption profile based on more than just current income.

In any event, in 2007 (red line), same thing, except now it's more like earn $100, spend $114. And yet again in 2009 (purple line): household dissaving is still very high, and, you'd think, absolutely in synch with everything you've seen before. Earn $100, spend $114.

And then you hit the green line - Stats' best current take on what was happening, based on the best available data to hand today.

It's a complete break from the accepted wisdom. Yes, households still look like they are spending more than they earn, but the scale of it doesn't seem to be quite as bad as previously thought, and far from being a picture of ever-increasing fecklessness (as all the previous estimates had suggested) households, on this latest reading, were actually getting their act together. On the latest data, households are actually breaking even - spending roughly what they earn. It may not be the most thrifty behaviour in the world, but it's stopped being outright profligate.

There are two possible explanations (and they are not incompatible) for what is going on here.

The first is that the original data were inadequate, and that a more complete count of income and spending shows that families weren't engaged on a massive and unsustainable spending binge. The second is that households' behaviour changed, and they stopped being as profligate as before.

If I had to pick between these explanations, I'd give more weight to the first one. When you count things more accurately (e.g  by capturing more of the income of trusts, as Stats has) you find that households were earning more than previously thought, so their spending wasn't quite as large relative to their (newly updated) incomes, meaning that their savings must have been greater.

That said, the second explanation also makes some sense. I'm perfectly prepared to believe that the GFC, and maybe other cyclical influences, also altered people's behaviour, encouraging more precautionary savings.

I have no definitely knock-out way of judging whether better estimation or a change in behaviour is the bigger explanation of the changes in the measured data, and I admit it's a judgemental guess. But my guess, for what it's worth,  is that the original data were well wide of the mark, and that more recently we're getting a better sighting shot on what households have actually been doing. Or in other words, the data revisions are the big story, and any change in behaviour is a contributory but secondary factor.

You're led that way by a point that the Treasury paper makes. Dissaving rates, as apparently measured by the original sets of official stats, must have been substantially overstated. If households were really spending massively more than they earned, then absent more than compensating revaluations on their assets, their net wealth would have gone down. As the paper notes (p20), "the negative saving rates based on the Household Income and Outlay Account imply that households would have completely drawn down their stock of net wealth. However, this is inconsistent with the evidence from both the aggregate sector data on household wealth published by the Reserve Bank, and the micro stock data from SoFIE [Survey of Family Income and Employment]". In other words, dissaving at the rate supposedly estimated would have seen households' assets exhausted. But they weren't. So the original dissaving estimates can't have been right.

A  great deal of newsprint and policy analysis (not to mention actual policy initiatives such as KiwiSaver) were based on a view of household saving and spending behaviour that, with the benefit of better data later on, now looks to have been more mirage than reality. As the authors put it (p15), rather moderately in the circumstances, "These findings underscore the need for the policy debate to be grounded in solid evidence, and to be cognizant of any limitations of the data that underpins that evidence".

None of this necessarily means that we might not have a national savings issue - these revisions affect only what is being measured for households - and our persistent current account deficits suggest we might. But they absolutely do make you reconsider the supposed behaviour of the New Zealand household.