Friday, December 19, 2014

Feynman on reading

Do not read so much, look about you and think of what you see there.

Thursday, December 18, 2014

Feynman on assumptions

Whenever you see a sweeping statement that a tremendous amount can come from a very small number of assumptions, you always find that it is false. There are usually a large number of implied assumptions that are far from obvious if you think about them sufficiently carefully.

Wednesday, December 17, 2014

Feynman on discovery

The chance is high that the truth lies in the fashionable direction. But, on the off-chance that it is in another direction — a direction obvious from an unfashionable view of field theory — who will find it? Only someone who has sacrificed himself by teaching himself quantum electrodynamics from a peculiar and unfashionable point of view; one that he may have to invent for himself.

Tuesday, December 16, 2014

Feynman on uncertainty

People search for certainty. But there is no certainty. People are terrified — how can you live and not know? It is not odd at all. You only think you know, as a matter of fact. And most of your actions are based on incomplete knowledge and you really don't know what it is all about, or what the purpose of the world is, or know a great deal of other things.

Monday, December 15, 2014

Velocity: NBER 14144 and Satisfaction

This is one of those posts where I don't so much have something to say as I have something to see for myself. I didn't think it through before I wrote it. I just had an idea and started investigating, making notes along the way. These are those notes. So if the thing is unreadable, don't say I didn't warn you.

If you can't read it, skip to the end to check out the graphs. I'm happy how they worked out. My suspicions were not unfounded.

This is new to me: using NBER data to chase down an answer.

The most interesting of the FRED Velocity series is the one that goes back into the 1800s. From the Notes on that series:
Data Were Computed From Wartime Current Prices Divided By Money Which Had Been Centered To June 30 For Each Year. Source: Simon Kuznets, NBER

This NBER data series a14187 appears on the NBER website in Chapter 14 at

NBER Indicator: a14187
I clicked the "Chapter 14" link, and came to a whole page of NBER links on Money and Banking. Searched for "a14187". Found one:
db dat doc a14187U.S. Velocity of Money Stock 1869-1966

The three "d..." links look to me to be database, data, and documentation. I clicked doc and found this:
"c           VAR 0075      14187 VELO TY 1R 869-966               MD= 1E-37   "
"c           REF 0075         LOC  636 WIDTH  9             DK  10 COL 71-79  "
"c                                                 EXP DEC=  3                "
"c                                                                            "
"c              VELOCITY OF MONEY STOCK                                       "
"c              -----------------------                                       "
"c                                                                            "
"c              NBER SERIES:  14187                                           "
"c              AREA COVERED:  U.S.                                           "
"c              UNITS:  RATIO                                                 "
"c              ANNUAL COVERAGE:  1869-1966                                   "
"c              QUARTERLY COVERAGE:  NONE                                     "
"c              MONTHLY COVERAGE:  NONE                                       "
"c              SEASONAL ADJUSTMENT:  NONE                                    "
"c              SOURCE:  SIMON KUZNETS, NBER                                  "
"c                                                                            "
"c              YEAR.                                                         "
"c                                                                            "
"c              CHECKED MANUALLY; NO CORRECTIONS NECESSARY.                   "
"c                                                                            "
"c           ...............................................................  "



I backed out of that and did another search, this time for MONEY STOCK. I found "U.S. Velocity of Money Stock 1869-1966" again, and then "U.S. Money Stock" series m14144a. This time, the doc file captured my attention. Here's the body of it:

MONTHLY COVERAGE: 05/1907-12/1946
STATES (NBER, 1970).

DATA, 1955-1969 (4)--SEASONALLY ADJUSTED DATA, 1947-1969.

Currency plus Demand Deposits plus Time Deposits... Friedman and Schwartz... four variables... data back to 1867 and forward to 1969... I can use this. I backed out of the doc and clicked the dat. Three columns of data: year, month, and value. In a textfile. Should be importable into a spreadsheet.

Firefox let me save each dat page easily and//... Oh, there are only three 14144 pages, and nothing going back into the 1800s. Hm.

I searched around a bit. No luck. Well that's disappointing. Okay. I'll go with what I found.

OpenOffice opened the DAT file in Writer. Crap. Maybe I can copy and paste it to the spreadsheet?

Nope, not in a useful way. I need import...

Maybe Insert sheet from file??? Fixed width... Detect special numbers... OK...


Rename the worksheet before importing the others...

Okay, got 'em all. That went quick. Now document the data. (There is no header info in the NBER DAT files.) Copy a few lines from the NBER page... Paste to the Blogger "Compose" editor...

db dat doc m14144aU.S. Money Stock, Commerical Banks Plus Currency Held By Public, Seasonally Adjusted 05/1907-12/1946
db dat doc m14144bU.S. Demand Deposits, Adjusted Time Deposits, All Commericial Banks, Plus Currency Held By the Public 01/1955-09/1969
db dat doc m14144cU.S. Demand Deposits, Adjusted Time Deposits, All Commercial Banks, Plus Currency Held By the Public, Seasonally Adjusted 01/1947-09/1969

... the thing comes out as a fully formatted table. Sweet. Sometimes stuff just works. I gave it the border and white background to dress it up.

I don't think I need the "m14144b" file; the other two are both seasonally adjusted, and "b" doesn't say it is. Anyway file "a" takes me from 1907 to the end of 1946, and file "c" picks up at the beginning of 1947 and goes to 1969. Hm, I started my senior year in college, 1969.

Damn. I wish I had the numbers back to 1867.

File m14144a. How to get annual data from these monthly numbers? Move the file to Excel and use VBA? No.

Why do I care? Can I use monthlies? No. The GDP data will be quarterly or annual.

Just do some calculation that pulls out annual numbers.

... Okay, that works. I figured the average of 12 monthly values to get an annual number. Then I just added 12 to the starting row-number to to figure the next year's annual value. I used the same method to calculate my year values. If I didn't get nice round year-numbers, I'd know I had something wrong. Everything came out okay.

Now do the same for file C...

... I did it for file B also. Hey, you never know. I might need it. Okay, now I want to copy the three sets of annual data all onto the same worksheet, and then go round up some GDP data so I can calculate Velocity and compare it to FRED's Velocity of Money Stock for United States.

Graph #1: The NBER MONEY STOCK Data
NBER's "A" file (blue) and "C" file (red) fit together and cover the whole 1908-1968 period. There's a gap between the two lines because the blue ends in 1946 and the red starts in 1947. That gap'll disappear when I treat the two datasets as one data series.

NBER's file "B" (dashed green) starts in 1955 and follows the same path as the red data. I don't need the "B" data.

I divided MeasuringWorth's US GDP numbers by NBER's money stock numbers to figure velocity in red on Graph #2. For comparison I took FRED's velocity that goes back to the 1800s, in blue:

Graph #2:FRED's A14187USA163NNBR (blue) and NBER's m14144 series (red) in Velocity Calculations
The red one looks like my gold one yesterday, the Historical Statistics M2 Velocity, the one where I said

To my eye the gold and blue are very similar, as if the blue line was produced in part by revising the gold line.

I think I was right about that.

On the page, down near the bottom there's a bunch of links for My Account and About and Services and more. Among the links for Research is Monetary Aggregates. If you follow that link you find these notes:

We often receive requests for monetary aggregates data that cover time periods prior to January 1959, the beginning date for the Board of Governors currently published monetary aggregates. Various data are available for these years, but not all data are consistent with the current definitions of the Board's monetary aggregates. This section discusses available data. The problem is not one of finding data; the problem is one of constructing monetary aggregates that are consistent with currently published definitions.

FRED probably revised the NBER data to make it "consistent".

The NBER numbers are older; FRED's were revised in 2012. So I wanted to see FRED's numbers relative to NBER's, to see how the numbers were changed.

Graph #3: FRED's number as a share of NBER's number

High and dry until the Depression; then the FRED velocity starts falling away.

We're looking at velocity numbers, GDP divided by money stock. To make velocity fall (taking GDP as a given) we'd have to increase the money number. So Graph #3 shows FRED's money numbers are higher than NBER's.

Sunday, December 14, 2014

Velocity 1870-2013

Jim's graph, here,

Jim says, "M1 Velocity trends downward after its peak in 1981 if
you use M1 adjusted for sweeps as the money stock."
is much more satisfying than mine from yesterday's post. Jim refigured M1 Velocity using what FRED unfortunately calls "M1 Adjusted for Retail Sweeps".

FRED has that name wrong. The "adjustment" (which begins in 1994) was a change in the definition of M1 money. FRED's version of M1 Velocity (shown on my graph yesterday) is based on the measure of M1 that is defined differently before and after 1994. Jim's version of M1 Velocity uses a definition of M1 that is the same before and after the 1994 date. It's FRED that's using "adjusted" numbers, FRED and me. (For more on this, see mine of 14 November.) Jim's graph is conceptually better than the graph I showed yesterday. But it is also satisfying.

Satisfying, because Jim's red line trends downward from the 1981 peak, same as the green line does. That just makes sense. What I said yesterday wasn't wrong; the different lines measure different quantities of money. What I said wasn't wrong, but it wasn't satisfying.

Thank you, Jim.

I changed all the data series to "annual" on Jim's graph at FRED, and downloaded the data.

I went to my old Google Drive files, dug up M1 and M2 (annual) from the Historical Statistics (Bicentennial Edition, series X414 and X415) and Nominal GDP (annual) from MeasuringWorth (Samuel H. Williamson, 'What Was the U.S. GDP Then?', August 2013). Using these series I calculated M1 Velocity and M2 Velocity for the 1915-1970 period.

Chronologically by start date we have six different datasets:

Table Style by

I put it all together on one graph:

Graph #2, Velocity: Data Revision and Clear Trends
The red and brown lines are a good match. The gold and green are not. But then, surprisingly, the blue and green are a good match.

To my eye the gold and blue are very similar, as if the blue line was produced in part by revising the gold line. Could be. Data for the gold line is from the Historical Statistics published in 1975. Date for the blue line is from FRED and was last updated in 2012, some 46 years after the last data item in that series.

So anyway, what I'm looking at now is a repeating peak-and-decline, with peaks in 1880 and 1981. Not exactly a hundred-year cycle, but not far off.

// Source File: Velocity 1870-2013.xls at Google Drive.

Saturday, December 13, 2014

Velocity -- the Long Run, and What Counts

The Federal Reserve has a limitless supply of money just waiting to be issued into the economy. Why don't we count that money when we figure Velocity?

Because it's not in the economy, that's why.

The same is true of the money we in the private sector accumulate but do not spend. It's not in the economy. It ought not be counted when we figure Velocity.

This is what FRED has on Velocity, 1869 to 2014:

Look at the older data first. There is a peak a few years after 1875, followed by a long, sweeping downtrend that gradually flattens, then finally explodes in the colors of the newer data.

There is a small peak shortly before 1925, a small decline shortly after 1925, and an up-down-up wiggle just before 1950. Those three are World War One, the Great Depression, and World War Two. It is true that, compared to nearby tiny changes in the purple line, those three are significantly large. But in the big picture, they are not big events.

Compared to general trends, Depressions and World Wars are trifles. For Velocity at least, this is the case. For the economy as a whole, I think.

The newer data begins in 1959. We have a choice. We can focus on the blue line, which seems to continue the flattened trend, stretching it to the whole of the 20th century. Or we can focus on the green line, which seems to continue a repeating, wavelike pattern of peak-and-decline.

The red line, odd man out, just doesn't seem to fit. Still, it shows a pretty definite increase (until the crisis) so maybe it speaks on behalf of the green line and the wavelike pattern. But then, after 1981, the green line falls while the red rises to a new and more extreme peak. It's not tidy.

It's not inexplicable. The lines show different things. They consider different pools of money. The blue and green lines measure much larger pools of money than the red. They include all the money measured by the red line, and other money besides. When you divide by these bigger numbers to get Velocity, the Velocity number comes out on the low side. Divide by the smaller number that the red line uses, and the Velocity number is on the high side.

But remember: We're measuring velocity, a measure of how often a dollar is spent. We spend the dollars in our pockets. We don't spend the dollars in our piggy banks. The velocity of piggy-bank money is zero. Does it really make sense to include the money we don't spend when we're figuring how often the average dollar is spent? When we figure average driving speed, do we include parked cars in the calculation?

The velocity of the money we actually spend is much higher than the number we get by including the money parked in savings in our calculations. Much higher, as the red line is much higher than the green and blue.

It matters, because some people have no savings.

Friday, December 12, 2014

This should come as no surprise

I'm starting to like Larry Summers. I went out of my way this morning to read one of his at VOX: Reflections on the new 'Secular Stagnation hypothesis' from this past October.

Summers writes of the FERIR -- the "full employment real interest rate". A decline in that rate, Summers says, "coupled with low inflation could indefinitely prevent the attainment of full employment."

John Bull can't stand 2 percent, I guess. Anyway, Summers writes:
Laubach and Williams (2003) have attempted to estimate the FERIR – using data on actual real interest rates and measures of where the economy is relative to its potential. While many issues can be raised with respect to their calculations, Figure 4 illustrates their estimate of a substantial long term decline in the FERIR.

Figure 4. US natural rate of interest

Sources: Thomas Laubach and John Williams “Measuring the Natural Rate of Interest”

And, Summers writes of the IMF:

They have reached conclusions similar to the ones I have reached here – that the FERIR has likely declined in recent years. This observation, together with the observation that lower US inflation – and in Europe declining rates of inflation – make it more difficult than previously to reduce real interest rates. This in turn suggests that the zero lower bound and secular stagnation are likely to be more important issues in the future than in the past.

Okay. A low FERIR is a problem because of the difficulty of lowering it further. (This sounds like an unsustainable strategy, no?) So, could something be done to push the FERIR to a higher level?

Well, look at the Figure 4 graph. It wanders, but it wanders on a downhill trend. The two largest increases occur, one in the early 1960s and one in the mid- to late-1990s. Gee, in both of those periods, the economy was strongly improving. Okay, well, this supports Summers' contention that a falling FERIR is a problem.

But look at that second increase. See when it occurs? It comes just after the downtrend in the debt-per-dollar ratio, and just when that ratio starts rising again:

Graph #1: Accumulated Total Debt per Dollar of Spending-Money
This should come as no surprise.

Thursday, December 11, 2014

Boiling it down a little too much

Image Source: Dr. James G. Wellborn

Marc Thoma, via Lars P. Syll:
It’s the time of year when analysts and pundits begin “marking their beliefs to market” – telling us what they got right or wrong in the previous year. In that spirit, here are some of the things I got wrong about the Great Recession...

The lesson for me is that if you want the inflation rate to increase, demand has to increase. That requires more than simply creating a bunch of reserves that sit idle in banks.


Listen carefully to Milton Friedman in Free to Choose:
When a country starts on an inflationary episode, the initial effects seem good. The increased quantity of money enables whoever has access to it -- nowadays, primarily governments -- to spend more without anybody else having to spend less. Jobs become plentiful, business is brisk, almost everybody is happy -- at first. Those are the good effects. But then the increased spending starts to raise prices...

Catch that last bit? It is spending that influences prices, not the existence of money. That's what Friedman said.

If anybody thought inflation was caused by printing money, I guess they boiled the idea down a bit too much.