Friday, September 9, 2016

Putting numbers on vigor (#3: Choosing Trends for CMDEBT)

Let me ask you something: Do you think you could make a pretty good guess about the future path of this graph?

Graph #1: Household Debt Service relative to Household Debt
I can: straight line, almost flat, forever. Now... invert the ratio and look again:

Graph #2: Household Debt relative to Household Debt Service Payments
I'd guess an ever-gentler curve, always rising, but always rising less than before. That's how I imagine it. But I put a trend line on it, on the concave-down part since 2002, and the thing peaks around 2020. Peaks and starts to fall:

Graph #3: Household Debt relative to Debt Service & 2002Q1-2016Q1 Trend
I have trouble picturing it. The fall doesn't seem realistic. What to do: Go with the meticulous calculation, or the gut feel?

I decided to take another look:

Graph #4: Household Debt relative to Debt Service & 2013Q1-2016Q1 Trend
It still peaks around 2020. But when I look at Graph #1 or #2 I just don't see that kind of change happening in the next few years.

Now this is more like what I had in mind:

Graph #5: Debt Service relative to Household Debt & Power Trend since 2010Q1
Changed from the Graph #2 ratio to the Graph #1 ratio and changed the trendline type from "polynomial" to "power". I don't know if this represents the true course of future events but it does represent my gut feel.

I think I'll go with Graph #4 and Graph #5, both of them, and compare the results when I'm done. I'll predict data out to 2025 for both ratios, then factor debt service out of the ratio. That will leave me with numbers for future household debt.

Come to think of it, I can take the household debt numbers and put a trend on them directly:

Graph #6: Household Debt in billions, with Trends Since 2010 and Since 2013
Yeah. I can use the trends from Graphs #4 and #5 to predict the relation between household debt and debt service, and pull out debt in billions from those relations. Then I can compare those two outcomes to the trends on Graph #6 and see if anything at all looks copacetic.

// the Excel file

No comments: