Sunday, September 2, 2012

Good question!

I am looking at one graph from Windyanabasis: The Great Unravelling, where (at least according to the Blogger editor) rsj misspells his new name for the Great Moderation.

Which eras are those of moderation, and which are those of excess volatility? What did the central bank succeed in stabilizing as a result of relying on interest rate adjustments rather than income flow adjustments?

Leverage: Domestic Non-financial debt/GDP 1950-1980 versus 1980-2010:

Which era is “moderate”?

Which era is moderate? That's a good question.

rsj expects us to see very little debt growth in the 1950-1980 period, and very much in the 1980-2010 period.

We are looking at debt and output of two thirty-year periods, hoping to get a reliable answer. I know, I know: You think you have a reliable answer.

I disagree.

Let's start with something simple: When economists compare GDP from one year to GDP from another, they generally are looking for a size difference. But they generally want to see the "real" difference, as opposed to the difference due to rising prices. So they take inflation out of the numbers. After that, they can compare output to output to see which is bigger, and by how much.

When economists, hobbyists, and normal people want to compare the GDP of one 30-year period to that of another, we run into the same interference from rising prices. The problem is solved the same way: Take the inflation out of the numbers first.

It's the same story when you compare debt from two different 30-year periods. But it's a bit more complicated, because any one year's debt is itself a collection of debts accrued over many years. So you can't take the inflation out of the debt number the same way you take inflation out of GDP. You have to break up the debt into yearly additions to the total. You have to adjust each year's piece of debt separately.

For more on the inflation-adjustment of debt see Measuring the erosion of debt (PDF).

But rsj is looking at neither debt nor GDP. rsj is looking at the ratio of debt to GDP.

If it was correct to adjust both debt and GDP for inflation by the same calculation, then the ratio of reals would be identical to the ratio of nominals, and this discussion would be moot. But that is not the case.

The inflation-adjustment of debt differs significantly from the inflation-adjustment of GDP. So the ratio of reals differs significantly from the ratio of nominals. Therefore, if it is our intent to compare these values across inflationary time, it is essential to use the inflation-adjusted values.

I downloaded some numbers from FRED, annual data for TCMDODNS, GDP, GDPC1, GDPDEF, and a CPI series that I ended up not using. The original download is the "FRED Graph" page of the Google Docs spreadsheet linked at the bottom of this post.

First step was to duplicate rsj's graph using the FRED data:

Graph #2: Nominal Debt relative to Nominal GDP

You can see the same great flat spot on this graph in the 1950-1980 period, and the same great increase thereafter, just as rsj's graph shows. But this is the ratio of nominals. It is not useful for comparisons in an age of rising prices.

The appropriate graph shows a ratio of reals. For debt, I use incremental inflation adjustment, where each year's addition to debt is adjusted separately. For GDP I use the standard calculation, aggregate inflation adjustment, because GDP is a flow.

Graph #3: Incrementally Adjusted Real Debt relative to Aggregate Adjusted Real GDP

So now let me ask you the same question: Which era is moderate?

Very little debt growth in the 1950-1980 period? I don't see it.

Look. Sorry about the spelling-error thing. I think rsj's post is very interesting. The topic, too. I have no love for the so-called Great Moderation. But if you're gonna set your sights on something in order to take it down, then you need your best argument. You cannot do it with bad graphs.

You can access the Google Docs spreadsheet used for this post.


Clonal said...


You will like this lecture by Ann Pettifor.

Delusional economics and the economic consequences of Mr Osborne - Ann Pettifor

Jazzbumpa said...

Certainly, by this measure, 2000-2009 is not moderate, and 1950-1967 is. 1991 to 2000 slopes down

What you're concerned with here is rate of change. I'll recommend looking at slopes. As a first stab, over a five year increment. Plot that and see where the high and low points are.

My old feeble eyes can't make a valid comparison of the 70's to the noughts on your graph.


Joshua Wojnilower said...

What might explain the downward sloping period? Response to housing decline in '91? Strong growth during period suggests decline in debt from this perspective maybe not a strong factor on growth. Should we not worry if the level retraces for some time?

The Arthurian said...

For me, "slack" explains it.

There was a massive intermission in the increasing rate of debt accumulation, 1985-1991 or so, which created slack in the cost of finance and opened the door to the "macroeconomic miracle" of the Clinton years.