Poking around

I didn't check the date of every turning point on every graph. Consider the dates approximate.

Krugman provides a graph of private saving as a share of GDP. I duplicated it:

Graph #1: Private Saving as a share of GDP

Uptrend into the early '80s. Downtrend to the crisis.

Of course I remember hearing it said, since the '80s, that people were saving less. So the downtrend doesn't surprise me. But I also know that accumulated savings is very high as compared to money in circulation.

Roughly, M2 is money in circulation and in savings, and M1 is money in circulation only. So (M2-M1) is savings. (This is how I figured it, before I found FRED.) And (M2-M1)/M1 compares money in savings to money in circulation. Despite a significant "whoops" in the trend, the accumulation of savings generally tended to increase rapidly:

Graph #2: Dollars of Savings per Circulating Dollar

In 1960 there was about a dollar in savings for every dollar circulating. By 1984 there was three times as much in savings. Between 1984 and the mid-'90s, downtrend with a couple bumps in it, bumps I think I've seen elsewhere. Then from the mid-1990s, nothing but increase to the crisis.

At peak, there were four and a half dollars of money in savings, for every circulating dollar. That's what I call a monetary imbalance.

I say "roughly" because FRED has a series named SAVINGS which is described as "Total Savings Deposits at all Depository Institutions". That's no doubt a better estimate of savings than what I used. Divide that series by M1 and you don't get $4.50 at the peak. You get about $3.

That FRED series only goes back to 1980, so you don't get a comparison like I showed above. But the ratio does trend down from the early 1980s to the mid-1990s, with a couple bumps along the way (like Graph #2). It bottoms out below $1, then skyrockets up to $3.

For the record, FRED says:

M2 consists of M1 plus: (1) savings deposits (which include money market deposit accounts, or MMDAs); (2) small-denomination time deposits (time deposits in amounts of less than $100,000); and (3) balances in retail money market mutual funds (MMMFs).

What I called "savings".


So the saving rate was going down for 30 years. But the accumulation of savings increased rapidly. Calls for a little poking around, don't you think?

Krugman uses GPSAVE, and I don't know what that looks like. So:

Graph #3: Gross Private Saving

Your typical variegated exponential.

Looks like there is a bit of a break, mid-80s, with a slower exponential increase thereafter. The log look at GPSAVE confirms this:

Graph #4: Natural Log of Gross Private Saving

A bit of a kink around 1984, and a slower rate of savings thereafter.

There is also some hint of faster trend from 1969 or 1970 to the kink. That faster growth has probably something to do with the high inflation of those years. (My assumption is that people had more money, so they saved more, despite the inflation. That would seem to contradict what is often said, that inflation discourages saving.)

I divided inflation out of Gross Private Saving, and the variegated exponential straightened out a lot. Then I took the Log view again:

Graph #5: Natural Log of Inflation-Adjusted Gross Private Saving

The straight-line increase now appears to last to the mid-1970s, when a flattening begins to occur. So maybe the inflation *did* discourage saving, after all. Why the discouragement remained in place until 1993 or so, I cannot say.

Interesting that the wiggly line flattens at 2.0, but then from 1993 makes a sudden jump that takes us to the 2001 recession. And then another jump that takes us to the 2008 recession. And then what looks to be the start of another jump. One wonders what change in tax law may have been established around 1993.


So that's GPSAVE.

Immediately following his GPSAVE/GDP graph, Krugman says:

Obviously it jumped up after the housing bust, but until then it was actually declining, and even now it’s below historic highs.

When I read that, I took it to mean that people suddenly started saving more after the housing bust. But my Graph #4 doesn't seem to show that. Then it occurred to me: the housing bust, and the recession, and the decline of GDP. A big decline in GDP would make Krugman's ratio spike upward. That would make it look like there was a big spike in the savings rate, even if there was not. This seems to be the case:

Graph #6: Gross Private Saving and GDP since 2000

A second look...

Graph #7: Percent Change in GPSAVE and GDP

...does show a big spike in the savings rate, from minus five to plus twenty. But it's a big spike in what Graph #6 shows to be the smaller number.

The big blue spike in saving on Graph #7 was rapid and short-lived, and came right at the start of the recession, if not before. The leisurely decline of GDP (in red on Graph #7) lasts for the full term of the recession and does not display what can be called a "spike".

I'm thinking, both the blue spike and the red slump must have played a role in creating the "jump after the housing bust" that Krugman's graph shows. I took Graph #7 and added Krugman's ratio to it in green, again just for the years since 2000. Krugman's uses the vertical axis on the right:

Graph #8: Components of the "jump" in private saving

That big blue spike, say 2007-2009, is matched by a little green spike in the Krugman ratio. The relative size, the relative smallness of the green spike shows that GDP is a much bigger number than Gross Private Saving. The increase in saving did have some effect on Krugman's results, but not as much as the blue spike suggests.

The red line on Graph #8 shows the growth rate of GDP. There is a general decline from shortly before the recession started until shortly before it ended. Corresponding exactly to that timing, and even to the size of that decline, is a general uptrend in the green line. (The green line goes up when the red line goes down, because the red represents the denominator of Krugman's ratio. That inverts the relation.)

That general uptrend, caused by GDP decline, contributed to the little green spike that we thought we could attribute to the big blue spike in the saving rate. So the effect of saving on Krugman's "jump" is even less than we thought.


Above, I described the FRED series called SAVINGS, which goes back to 1980. Now I found SAVINGSL, which goes back to 1959. That's better.

The longer series (SAVINGSL) offers monthly values. The one that starts at 1980 offers weekly numbers. Because of that difference, the most recent value for the monthly series may be a week or three older than that of the weekly series; but otherwise, the two series are for all practical purposes identical:

Graph #9: Monthly and Weekly views of Accumulated Savings

The red line completely hides the blue for all values after 1980.

Krugman's graph uses GPSAVE, Gross Private Saving. Saving is the periodic (annual, say) addition to savings with a trailing "s", which is the accumulation. (Similarly, deficits are the annual addition to debt, and debt is the accumulation.)

Krugman's graph (see Graph #1 above) looks at (annual) saving relative to GDP. Flow relative to flow. I want to look at (accumulated) savings relative to GDP. Stock relative to flow. (Comparable to looking at total debt relative to GDP, for what it's worth.)

Graph #10: Accumulated Savings relative to GDP

Accumulated savings increased relative to GDP until the near-recession of 1966. Then there was a long, sweeping decline until around 1995. After that, savings again increased relative to GDP. There is a big low in the middle there, apparently associated with the recessions of 1980 and 1982. But that low began substantially before the first of those two recessions. And when it ended, the numbers bounced back quickly to the trend of the sweeping decline.

Peak around 1966. Trough around 1995. Anomaly around 1978-1983. None of these events seem to match up well with Krugman's 1982-1984 twin peaks.

My first reaction to this discrepancy: How much I *add* to my savings is one thing; how much I *have* in savings is another. Obviously there is a relation between additions to savings and the accumulation that results. But perhaps the graphs show that this relation is not straightforward. Not linear.

Obviously other factors are involved, such as GDP. But then GDP is used in both Graph #1 and #10, and used the same way in both. So now, I want to eliminate other factors, and compare saving to savings directly.

Graph #11: Additions to Savings relative to Accumulated Savings

Who knew?


Graphs #9 and #10 both appear to show a change around 1995: decline in the one case turning to increase, and increase in the other case suddenly becoming more rapid. Once again, I wonder about changes in tax policy that may be the cause of this change in savings.