"To the extent that spending is credit-financed..."

In The myth of the “Phoenix Miracle” at VOX, Michael Biggs, Thomas Mayer, and Andreas Pick write:

To the extent that spending is credit-financed, demand in a particular period should be a function of the new borrowing that takes place in that period. Demand (and consequently GDP) is therefore a function of the flow of credit, and growth of GDP should be related to growth in the flow of credit rather than growth in the credit stock.

First off, I was glad to discover an actual example of how stock-flow consistency can make a difference. The article is quite specific about this, referring to "inappropriate comparison between the flow of GDP and the stock of credit." Their Figure 1 offers a very clear picture of the difference it can make.

I was also glad to see the emphasis on growth and new uses of credit. For the new uses of credit provide economic boost, while the cost of existing debt is a drag on growth. This duality is what eventually makes credit-use ineffective, as ever-larger new uses of credit are required just to offset the drag created by the debt from prior credit use.

The article also provides an optimistic conclusion:

If the pace of de-leveraging slows gradually from current levels, the increase in the credit impulse would support private sector demand growth even as debt levels fall.

But it seems a very narrow row to hoe. A little too much drop in debt levels would undermine growth; a little too little, and there is no de-leveraging at all.

There is another option. Note the introductory phrase in the first excerpt: "To the extent that spending is credit-financed...."

To the extent that spending is credit-financed, demand is a function of new credit use. To the extent that spending is *not* credit-financed, demand is independent of credit use. So, let's look at the extents.

The graph shows dollars of GDP per dollar of new credit use each year:

Graph #1: GDP relative to the Change in Total Debt

This graph, actually, is what I was trying to look at when I ended up comparing changes in both datasets, in I got this one by accident.

The spikes on Graph #1 are high until 1960. Several rise to between $100 and $200 of GDP per dollar of new credit use. Then the spikes are much lower -- under $100, say -- until maybe 1972. After 1972, there are basically *no* spikes at all.

GDP was largely independent of credit use in the years before 1960. It was somewhat independent of credit use until the early 1970s. Since that time, GDP has been tightly bound to credit use.

So I want to say that "the extent that spending is credit-financed" has changed significantly, from very little in the early years to very much in recent decades. The "extent that spending is credit-financed" varies, and has increased over time.

But that's not all. Look at the bottoms of the spikes -- leaving out the two really big lows. Consider the trend suggested by the path of the ordinary lows. The path is relatively high until after 1970, then lower and downtrending with a bottom at 1985. Then it rises through the 1991 recession, remains high until the macroeconomic miracle is under way, and declines again.

The trend of these minimums shows that there was substantially more output per dollar of new credit use in the 1950s and 1960s than in later years. Either credit use was more "efficient" in the early years, or "the extent that spending is credit-financed" was much less. Or the smaller extent made the use of credit more efficient.

Or the smaller extent of credit use made the use of credit appear more efficient, and left people talking about the productivity of debt, certainly a flawed concept.

Graph #2: Minimums from Graph #1 (with 1954Q1 and 2009Q4 removed)

Graph #2 is from this Google Docs spreadsheet.

h/t Marko #57 at Asymptosis.