Today’s “Chart of the Day” from Business Insider’s Clusterstock blog presents an alarming picture of the US economy viewed through the prism of bank business lending. The chart, which I have reproduced below, shows a precipitous collapse in lending*, described in dramatic language as “falling like a knife”. There is no doubt that the US economy remains in very poor health, but should we be getting as excited as Clusterstock?
Annual Change in US Commercial and Industrial Loans
Closer examination of the chart reveals that it is in fact quite misleading.
For a start, it makes the very common mistake of plotting a long series of data without adjusting for the fact that over time the value of the dollar has declined through inflation and the US economy has grown. As a result, more recent movements in the data take on an exaggerated scale.
Also, the chart shows annual changes without providing any sense of the base level of lending. Not only that, while attention is drawn to the US $300 billion annual decline in lending, the increase of close to US $300 billion just over a year earlier is ignored, when in fact the two largely offset one another. Certainly lending has declined, but rather than taking us into historically unprecedented territory, as the Clusterstock chart suggests, it actually means loan volumes are back to where they were in late 2007.
Both shortcomings are addressed in the chart below, which shows the history of loan volumes themselves rather than annual changes and overlays a series scaled by the gross domestic product (GDP) of the US to represent lending in “2010 equivalent” dollars.
US Commercial and Industrial Loans
Changes in lending do provide a useful reading of an economy’s health. But, it is important to be careful when using annual changes to read its current state. The change from January 2009 to January 2010 is affected just as much by what happened a year ago as by what happened last month. Since monthly data is available, we can in fact look at changes over a shorter period. The charts below show monthly changes, which are probably a little too volatile, and quarterly changes which are probably the best compromise. Since these charts extend only over a five year period, it is not as important to adjust for changes in the value of the dollar and the size of the economy.
Monthly Changes in US Commercial and Industrial Loans
Quarterly Changes in US Commercial and Industrial Loans
Both of these charts reveal an economy that certainly remains unhealthy and lending volumes are still declining. However, the declines of the last couple of years evidently reflect an unwinding of the enormous increases of a few years earlier. So rather than fretting that lending is “falling like a knife”, we can take some comfort from the fact that the rate of decline is diminishing from the worst point of the third quarter of 2009. The moral of the story is that charts can mislead as easily as words and should always be treated with caution.
* The data is sourced from the St Louis Fed “FRED” economic database.
Possibly Related Posts (automatically generated):
- Retail Sales in New South Wales (4 June 2008)
- ngramr – an R package for Google Ngrams (16 July 2013)
- The Mule trips up (21 April 2010)
- More spreads (17 November 2011)
Thanks for this one, Sean. You know, I saw that chart and my very first reaction was, “That doesn’t look like it’s been indexed for inflation.” I’m learning things from you, Sir!
Since quarterly changes are the sum of monthly changes over one quarter the quarterly graph above, on a monthly scale, is a (scaled x 4) moving average of the monthly changes graph. This explains why it looks a lot smoother and gives the impression of some underlying smooth nonlinear trend. Moving averages of even completely uncorrelated random series can give this impression and are an example of a Junk Chart (possibly type #4).
Sorry, that should read “scaled x 3”. Note to self: there are 3 months in a quarter.
Stilgherrian: very happy to hear my persistent rantings are having an effect!
JamesGlover: something along those lines is a good idea. I was thinking of giving an example where the moving average drops not because the latest month drops, but because a large increase a year ago drops out of the moving average window….next time!.
Hey, Stubborn,
“Both of these charts reveal an economy that certainly remains unhealthy and lending volumes are still declining.”
So, the Quarterly Changes in US Commercial and Industrial Loans chart is the variation of a net balance (gross loans minus loan repayment), right?
“we can take some comfort from the fact that the rate of decline is diminishing from the worst point of the third quarter of 2009”.
So, it has been improving in the last 3 quarters. In other words: the private sector has been deleveraging since early 2009. Not so much now.
That’s certainly interesting and promising. But it still doesn’t look too good.
But, let me ask you this: is there any reason to believe that now the private sector is starting to get new loans (as opposed to just repaying less loans)?
I suppose this data does not include mortgages and credit cards, does it?
Is there any similar data relating to the UK and Australia?
“The moral of the story is that charts can mislead as easily as words and should always be treated with caution”
Yes, they sure can. Funny that you should mention this, Stubborn. I am also interested in Lies, Damn Lies and Statistics. :)
Was the first chart actually published?
Cheers,
Marco
PS: Very clever JamesGlover’s observation. I hadn’t thought of that!
Marco: the data does not include retail debt (mortgages, cards, etc). The RBA publishes similar data for Australia. I’m not sure about the UK, but I expect it’s out there somewhere.
I would say that the private sector is certainly still deleveraging. Here in Australia for example, 2008 and 2009 saw a significant amount of corporate equity raising rather than borrowing.
Yes, the first chart was published here.
I checked the link and there 13 commenst posted in reply to that article.
12 comments just take the chart at face value; while 1 poster suspects that there is something fishy with the chart, but failed to perceive what it was. It came close, though: it was an undefined something, but related to the y axis.
Interestingly, the interpretation that seems to prevail is that banks are not lending, as opposed to borrowers re-paying their debts.
Marco: at least someone there questioned the chart.
From the data itself, you cannot get determine how much of the reduction is the result of borrowers choosing to repay their debt (after all, why would a business facing significant excess capacity want to increase their debt?) or banks tightening their lending and turning borrowers down (and surely they would all be more risk averse than they were a couple of years ago). I think that it is safe to say that it is in fact a combination of both, but many under-estimate the contribution of the former, in my view.
Companies might be increasing their debt if they were drawing down on existing revolving credit facilities to cover “short term” cash flow problems due to economic downturn. I know of one small business that considered doing exactly this so we have a potential nonzero sample size. I don’t know if undrawn credit facilities are already included in total debt numbers or this is classified as debt only when drawn. The latter sounds logical but from bank capital side I think undrawn facilities are also included times a discount factor (but not sure).
James: you are right that capital calculations include undrawn facilities, but the St Louis Fed data, as I understand it, only includes drawn assets.