The Boring '90s

Antonio Fatas looks at the change in stock market valuations since the November election and does some economist-math that I can almost follow. He also provides a link to a useful page at CBO. They have budget projections, spending projections, economic projections, and more. They have their most recently-issued projections and older vintages in separate downloadable Excel files. A useful site.

I got looking at their "Potential GDP and Underlying Inputs", then went to FRED by reflex for a gander:

Graph #1: Potential GDP Growth Rate

The graph includes a 10-year projection out to 2027. The projections always make me laugh because they are so obviously different from the historical record. The smooth curve shown at the right end of this graph anticipates lousy growth, but no recession.

Probably smart, not pinpointing the date of our next recession.

The other thing that gets me about Potential GDP is the anomalous high of the latter 1990s. Here -- I downloaded the data and put a trend line on it in Excel:

Graph #2: Growth of Potential GDP and Linear Trend thru 2016

The blue line shows 1950 thru 2027, same as Graph #1. I used the same data for the red line, but stopped the data at the end of 2016. Then I put a linear trend line on the red one.

The 2017-2027 projection (blue) doesn't look so funny now. But that anomalous high of the late 1990s, that one still bothers me.

I put a green line on the graph, connecting the high point of the early 1950s, the high of the latter '60s, and the high of the mid-80s. Then I put a linear trend line (dashed line) on that. Now it's easy to see how high Potential GDP was in the latter 1990s.

Graph #3: Trend of Peaks in Potential GDP Growth

Also, maybe the 2017-2027 projection is overly optimistic.

Now I want to deflate the late-1990s high, to bring it down to the dashed trend line like the earlier highs. Then we can see what happens to the other trend line on the graph.

I removed the dates from the x-axis and went with the default sequential numbering instead. I displayed the trendline formula and formatted it to show a dozen decimal places. I used the trendline formula to calculate values that lie on the trend. Then, checking my work, I displayed the values (purple line) to make sure they do indeed lie on the dashed trend line.

Graph #4: Using Excel's Trendline Formula to Calculate Trendline Data Points

Now I could find when the high point of the latter 1990s occurs (first quarter 1999), what that high point value is (4.18646 percent) and what is the first quarter 1999 value of the purple line (3.02275 percent).

I want to take that whole red peak above the purple line (plus some of the points below it) and scale that all down so the red peak just touches the purple line in first quarter 1999. Get rid of the anomalous high and see how things look then.

(Who would be interested in such things? No wonder nobody reads me.)

It took some fiddling but I finally got the calculation right. The high point of the latter 1990s has been lowered and now falls directly on the dashed trend line.

Graph #5: Removing the Anomalous Peak

The red is the adjusted data. The blue shows the given data. Blue in the 1990s is substantially higher than the trend of peaks.

You may notice that the solid black trendline is lower here than on the previous graph. Excel relocated it automatically when I made that anomalous red peak "nomalous".

I cleaned up the graph, got rid of the green line and the dashed line, and put dates back on the x-axis.

Graph #6: Growth of Potential GDP Without the Boom of the 1990s (red)

Come to think of it, the 2017-2027 projection now runs about as much above trend as the other highs of the red line.

And the anomalous high of the 1990s really stands out.


Wouldn't it be nice to know what happened to make potential GDP go unusually high in the 1990s? I guess everybody has a story about that. The trouble is, the story has to explain not only the big increase in potential GDP, but also the big decrease that soon followed. And it has to explain similar changes in productivity.

A lot of people will tell you the IT revolution accounts for the increase -- and that it suddenly petered out. Well yeah, I can see that the petering out happened in productivity and all. But that doesn't mean IT was the cause.

Here's my story, and it has nothing to do with IT. It has to do with money.

1. There was a significant decrease in the growth of debt from 1986 thru 1991, and the growth of total debt remained low (below 7.5%) for most of the 1990s:

Graph #7: A Sudden slowdown of Total Debt Growth, 1986-1991

2. There was a significant increase in the growth of the "funds that are readily accessible for spending" from 1989 thru 1992, and money growth remained high for some years after 1992:

Graph #8: A Sustained Increase in the Growth of Circulating Money

3. Slow growth of total debt combined with rapid growth of circulating money to create slack in the demand for credit in the early 1990s:

Graph #9: Monetary Slack led to High Potential GDP and High Productivity in the 1990s

4. Since people had more funds readily accessible for spending, the growth of total debt could remain low thru most of the 1990s:

Graph #10: The Slow Increase of the Debt-to-GDP Ratio in the 1990s

5. Combined with lower interest rates, the slow growth of debt meant that interest costs in the 1990s were lower than in the 1980s -- lower by four or five percent of GDP:

Graph #11: Reduced Financial Costs Opened a Door to Greater Production and Consumption

6. The reduced financial costs of the 1990s meant lower debt service payments:

Graph #12: Household Debt Service was Low in the 1990s

7. The reduced cost of finance left more money available for production and consumption. All else aside, the reduced cost of finance can account for the temporary but significant increases in productivity and Potential GDP that occurred in the latter 1990s. Moreover, we can do it again.

Is IT responsible for the good years of the latter 1990s? Maybe. But it could not have happened without the reduction of financial cost.


// The Excel file