Finding Philippon's FinEff.pdf

I found it, sort of. Here.

Better than nothing.

Here's Figure 1:

//

Here it is, as a PDF, from SSRN. Free. Thanks, guys.

Now I have to go and backfill a few of my posts with a new copy of the Figure 1 graph. (When I wrote those posts I trusted, foolishly, that Phillippon's paper would always be available at Stern where I first found it. It ain't.

//

ps, at SSRN the pdf filename is SSRN-id1972808

//

Edit, 18 May 2017.

But maybe the graph I want is Philippon's Figure 7 from that PDF:

Graph #2, from Has the U.S. Finance Industry Become Less Efficient?

Gross Domestic Spending: Documentation

This post provides links to two Excel files and eleven short PDFs. The "GDS Data Series" file is outputs, the result of my industry. The "Business Population" file is inputs, the source data that I used.

The GDS Data Series.xls file is an Excel file generated by OpenOffice Calc. The file contains one sheet. The first two columns of the sheet contain the year and GDS values. Columns farther to the right document the calculations used to come up with the numbers.

The Business Population.xls spreadsheet is an Excel file generated by OpenOffice Calc. The file contains two sheets. The Data Entry sheet lists the source files and data values obtained from them. Columns further to the right are notes on units and on the column headers, which evolved a bit over the years.

Note that there is some duplication of data on this sheet, as I was comparing the values provided by newer and older sources at the overlaps. This duplication is eliminated from the other sheet in the file.

The other sheet in this file is named ArtS Data Series. This sheet lists the source file, the year, and "Number", "Receipts" and "Net Profit" for total business enterprises -- that is, for sole proprietorships, partnerships, and corporations, combined -- for 1959 through 2008.

In order to document the source data for the GDS series, I extracted individual pages from various source PDFs. The extracted files are typically one-page PDFs. To extract the relevant pages I just printed them to doPDF.

The 12s0744.pdf isn't mine. I found it on the web. Somebody else was thinking along the same lines I was, apparently.

All files are stored on Google Drive, the old Google Docs.

PDF	PAGES	PERIOD	SOURCE
HS70 V 1-12	1	1959-1966	Historical Statistics (Bicentennial Ed.)
SA74 Table 775	1	1967-1968	Statistical Abstract (1974)
SA76 Table 823	1	1969-1971	Statistical Abstract (1976)
SA78 Table 915	1	1972-1974	Statistical Abstract (1978)
SA85 Table 868	1	1975-1979	Statistical Abstract (1985)
SA89 Table 846	1	1980-1984	Statistical Abstract (1989)
SA92 Table 827	1	1985-1987	Statistical Abstract (1992)
SA2000 Table 855	1	1988-1994	Statistical Abstract (2000)
SA0405 Table 716	1	1995-1999	Statistical Abstract (2004-05)
SA09 Table 722	1	2000-2003	Statistical Abstract (2009)
12s0744	1	2004-2008	Statistical Abstract (2012)

Table Style by TextFixer.com

The source PDFs (large files) are still available from census.gov, but the 2012 Abstract is the last one they issued.

You know, spending cuts.

Preliminary notes: What does it mean when GDS increases relative to GDP?

I showed this graph on 2 July:

Graph #1: Gross Domestic Spending as a Percent of Gross Domestic Product

It turned out that the raised area, in the 1970s and thereabouts,showed a fair similarity to the rate of inflation in those inflationary years.

Got me thinkin'

//

GDP is Gross Domestic Product, the the sum of final spending in our economy in a year. GDS is Gross Domestic Spending, the total of final and non-final spending in our economy in a year. The increases in GDS are measured in dollar volume, not "number of transactions" or any such thing. It's a dollar number. So if GDS increases it is because spending increased, plain and simple. Same with GDP.

One major effect of a change in the GDS/GDP ratio would seem to be a change in spending pressures -- that is, inflation pressures.

//

Whatever cause may lie behind the increase in spending, then, one fact remains: the spending increased. The trend of GDS/GDP, then, in the 1970s, was that costs increased faster than income: GDS increased faster than GDP.

If this is correct, I must be able to see it in business profits -- perhaps profits relative to GDP, certainly in profits relative to expenses.

I'll take that challenge.

Graph #2:Total Business Deductions Peak in 1982 as a Share of Gross Income

Yeah. It stands up to the challenge. Spending increased faster than revenue until 1982, and more slowly after that year.

If deductions rose till 1982 and then fell as a share of revenue, then profit must have done the opposite:

Graph #3: Business Profits hit bottom in 1982, Trending Upward Thereafter
That should come as no surprise.

Profits tended down until 1982, and up thereafter.

In case you missed it yesterday

A comparison of gross output, the output consumed by production, and the net output that remains, as against Gross Domestic Product and my measure Gross Domestic Spending, which adds nonfinal business spending to GDP:

Here's the picture:

Graph #2

The blue line is U.S. Gross Output -- the measure of what we produce in a year.

The red line is intermediate consumption -- the measure of output consumed in the process of producing output. It is almost half Gross Output. The red is almost half the blue.

The green line is output left over after using up output (red) to produce output (blue). The green line is just a little over half the blue. The red line is just a little under half; the green is just a little over.

The gold line is GDP. It is approximately equal to the green. The gold line, we call it "gross" domestic product, but really it is the part of output left over after using output to produce output. Almost the same as the green line. It differs for reasons you can read at Wikipedia.

The brown line is Gross Domestic Spending. It is GDP plus the business spending that is written off as tax deductions in the process of producing output.


Data sources for Graph #2 --
Blue, Red, Green: UNSNA via UNdata.
Gold: FRED.
Brown: Arthurian.

Here's the Zoho.

UNSNA: US Gross Output 1987-2011

UNSNA is UN SNA, the United Nations System of National Accounts. Like NIPA. Except "NIPA" is pleasing to the ear, and "UNSNA" sounds like something you'd wipe off with a kleenex.

I googled unsna data historical and found System of National Accounts 2008 - United Nations ... so I went there looking for data. I clicked DATA on their blue menu bar. That brought me to their National Accounts Data page. Text and links. At the end of the second paragraph is a link to UNdata. That's the one you want.

I suppose I could eliminate the previous paragraph and just give you the UNdata link direct. But I'm trying to give you a sense of the struggle, perseverance in the search for data. (Plus, I write this stuff down so I can find my way back later.)

Anyway, the UNdata page has a search box, so now we're making some progress. I searched for gross output. I got two hits. Two is good. What the hell am I supposed to do with 13 million hits?

I take Table 2.6 Output, gross value added and fixed assets by industries at current prices (ISIC Rev. 4) because it has 91124 records. The other one (ISIC Rev. 3) must be older, and has half a million records. Maybe I'll look at that later.

Now we're getting to the meat. The Rev. 4 link gives me this:

The main body of the default page is a table of info about Armenia. It consists of several columns of info I don't want. But there are select filters scrollboxes on the left. The first one is countries. So I can look at U.S. data instead.

Their scrollbar is neat. You just hover over the little down-arrow, and the list of countries scrolls, slow enough you can read it but not so slow that you're bored.

In the second filter box you can pick years for data. I want all the years, naturally.

Now the important part that you must not miss:

Just below the second filter list there is a link that says More. You'll want to click that. It lets you zoom in closer to the data you want.

Click the MORE and you get two additional filter boxes. At the top of the third box is "total economy". That's the one I want. The other options are industry-specific. I want big-picture.

The last filter box lets you pick the data you want to see. That's a big deal. Basically, I wanted the first three that appear in the filter window, the ones I saw at Wikipedia yesterday: Gross output, intermediate consumption and gross value added (the last of these being approximately equal to GDP).

Oh, and most important, be sure to click the Apply Filters link. It's below the last filter box. That's the magic that makes UNdata do what you want.

Oh -- funny thing. While writing this post I was going through the motions to verify the steps. But I didn't pick the country and the years before I clicked MORE. Then after I told you about all four filters, I tried to pick the U.S. and follow through the process. But the thing wouldn't let me pick the U.S. The list of countries scrolled down as far as the S's and no further.

I clicked LESS to go back to the two-filter screen and tried again. No good. But I refreshed the page, and then it worked okay. (In case you run into trouble using the thing.)


When I was done selecting my filter options I clicked APPLY FILTERS and the screen changed to show the data I wanted to see.

And if you don't know about CSV...

As the first of the UNdata screen-captures shows, across the top above the table of data are options including DOWNLOAD and SELECT COLUMNS. Both look useful.

I used DOWNLOAD, which gives some format options. First time, I tried XML. But when I loaded it into OpenOffice Calc I got a blank spreadsheet. So then I tried the "value separated" option identified as "comma" (Sheesh! What an awkward way to say CSV) and it worked fine. Gave me a ZIP file.

Had to extract the data file from the zip before I could use it. (I just put it on my desktop along with everything else.) Then I right-clicked the data file and picked "open with" OpenOffice Calc. A "text import" window opened. I checked COMMA and unchecked TAB (or something like that) to make the sample import look right, then clicked OK.

OpenOffice Calc opened the file right away. Excel would be as easy, maybe easier.

Now we're getting to the meat.

I saved the spreadsheet right away, in spreadsheet format. Otherwise, it stays as a CSV file. And CSV files can't save the formatting that spreadsheets do, so the next time you open the thing most of your changes are gone. Save it in spreadsheet format, and you solve that problem.

I bolded the column labels in row 1, set column widths, and made the text in column C wrap automatically. (That's the kind of formatting you lose if you keep the file as CSV.) Then, because I wasn't careful with the import, I had to go into column F and delete a single-quote from in front of the number, on every row.

Also, I deleted some columns that UNSNA thinks important and I don't. Then I uploaded it to Zoho so you can see what the thing looks like. Here's the Zoho:



The filename was set by the UNdata exporter.

The first row identifies the columns. Look at rows 2, 3, and 4, column C. There you have output, intermediate consumption, and gross value added (all for 2011, as column D shows). The next three rows have the same three items of data for 2010. The rows continue like that, back to 1987. Then, from 1986 back to 1970 the only data item is gross value added. Before 1970, nothing.

Still, we've got three items of data for the years 1987 to 2011. Not a bad start.

The biggest problem I have now is that all the numbers are in column F. That means I can't copy a chunk of numbers and get just gross output or just intermediate consumption or something. I get a jumble. In order for the data to be useful, I have to separate it into different columns by type (so I get three columns of data rather than one) and also combine the three rows for each year into one row. Then I get ONE row for 2011 with three columns of data, and ONE row for 2010 with three columns of data, and ONE row for 2009 with three columns of data... I could go on, but I think you get my drift. It's obvious, isn't it?

So why doesn't UNdata know this? I think they DO know it. They're not stupid. Maybe they just want to make it difficult. Hey, I've never seen this data used anywhere. Why not? I think people are discouraged from using the data because it is so much work to undo the stupidities built into it by people who surely should know better. I'm beating around the bush here, trying to avoid being explicitly rude.

Oh, well.

If you scrolled the spreadsheet enough to see column F, you saw that the numbers are given in dollars. I mean, if the value is 27,170.5 billion, UNdata gives it as 27,170,500,000,000. That's not easy to read; you have to count digits. So I want to convert their numbers to billions. The numbers will be easier to read that way, and compatible with FRED's numbers.

That, and I want to convert the three-row-one-column format to one-row-three-columns, so the data is useful.

Now we're getting to the meat.

I copied the original data (Sheet1) to a new sheet to make the changes.

Figured out a calculation to move the data for me into the format that I want. It's in the first few columns of Sheet2 in this Zoho file. (It only applies to the data back to 1987, where there are the three different values for each year. For 1986 back to 1970 I just copied and pasted the range of cells.)

Turns out that, like my column F values, my year values had the single-quote mark that left-justifies them and makes them behave like text instead of numbers. Next time I'll have to be more careful with the import from CSV.

I copied Sheet2 to make a new sheet and called it Sheet3. On the new sheet I copied the first few columns and pasted them as values over top of the calculations. After that I could delete the source data that was formatted badly, clean up the sheet, and credit UNdata for the data.

And of course, they also present the data backwards, newest first, rather than chronologically. I sorted that. Here's what I have so far -- data now ready to use.

Now we're getting to the meat.

Now I was ready to look at the numbers.

Graph #1

I noticed Wikipedia said

In the US economy, total intermediate consumption represents about 45% of Gross Output.

That's about right, for the blue line, since 1997 or so. More like 42% before that. And it looks like a trend of increase (for the blue line) and decrease (for the red).

Also... when you say "average" it suggests up-and-down variation, but it also implies stability in the value. If intermediate consumption is trending upward as a share of gross output, it is deceptive to speak of its average value. Would be nice if the values went farther back in time, so one could be more certain about trends.

Now we're getting to the meat.

Another thing the Wikipedia article said was that "gross output" is what we produce in a year, but we use up "intermediate consumption" in the process, which leaves us with "net output" which his just about the same as "GDP". That's four different data series right there.

To those four I want to add my context variable "GDS", gross domestic spending. That gives us five series to look at on one graph. You with me? Usually if I have more than two lines on a graph I start to lose clarity. This time it's five. But we're only looking at their relative sizes. That's straightforward.

The source for this graph is the same file as the Shares of Gross Output graph above. It's on the next tab to the right, so you can look at the numbers if you want. Here's the picture:

Graph #2

The blue line is U.S. Gross Output -- the measure of what we produce in a year.

The red line is intermediate consumption --  the measure of output consumed in the process of producing output. It is almost half Gross Output. The red is almost half the blue.

The green line is output left over after using up output (red) to produce output (blue). The green line is just a little over half the blue. The red line is just a little under half; the green is just a little over.

The gold line is GDP. It is approximately equal to the green. The gold line, we call it "gross domestic product", but really it is the part of output left over after using output to produce output. Almost the same as the green line. It differs for reasons you can read at Wikipedia.

The brown line is Gross Domestic Spending. It is GDP plus the business spending that is written off as tax deductions in the process of producing output.


Data sources for Graph #2 --
Blue, Red, Green: UNSNA via UNdata.
Gold: FRED.
Brown: Arthurian.

"Gross Domestic Spending" is not "Gross Output"

So I was rummaging around the internet, looking for other people's use of the term "gross domestic spending". Among the first few hits I found something that confused me. I couldn't tell if it was relevant to my topic. I had to take another look.

The puzzling piece was in the Production approach section of the Gross domestic product article at Wikipedia. The confusion arose because they explained one thing that I didn't know about in terms of two other things I didn't know about:

Gross value added = gross value of output – value of intermediate consumption.
Reading it over (more than once) helped. Realizing that "gross value of output" was the same as "gross output" helped. Reading someone else's version of it helped, too:

Gross output is an economic concept used in national accounts such as the United Nations System of National Accounts (UNSNA) and the US National Income and Product Accounts (NIPA). It is equal to the value of net output or GDP (also known as gross value added) plus intermediate consumption.

I'd swear I read that the NIPAs don't use the "gross output" concept, but I can't find the reference now. No matter. "Net output" equals "gross value added" equals GDP, they say. That helps me tie things together.

So Gross Domestic Product is net output. Reminds me of Jim's question: So what do you call gross spending? Do you call it Gross Gross Domestic Product?

The Wikipedia page links to their Intermediate consumption page, which also helped:

Conceptually, the aggregate "intermediate consumption" is equal to the amount of the difference between Gross Output (roughly, the total sales value) and Net output (gross value added or GDP). In the US economy, total intermediate consumption represents about 45% of Gross Output.

Okay, I was at last mentally prepared to deal with the idea of "Gross gross domestic product" -- of a quantity bigger, more general, more "gross" than GDP. GDP is "net" output, after some other output is used up producing it. The part that gets used up is "intermediate consumption". The part that remains is GDP. And the total of the two is "gross output".

Forty-five percent, they say: "Total intermediate consumption represents about 45% of Gross Output." How can I make use of that fact?

Gross Output equals 100% of Gross Output. Intermediate Consumption is 45%. And GDP is what's left. So GDP is about 55% of Gross Output. Pretty remarkable, don't you think? Gross Domestic Product is barely half of Gross Output.

How does this compare with Gross Domestic Spending? Well, GDS is about three times GDP. Gross Output is about two times GDP. So Gross Domestic Spending is bigger. Let me see...

Gross Output is 100% of Gross Output. GDP is about half of that, or 55%. And Gross Domestic Spending (as we saw the other day) is a little less than three times GDP. Three times 55 is 165. A little less, so say 150. So Gross Domestic Spending is about 150% of Gross Output.

So if you were wondering what businesses spend all that money on, something like two times GDP, well, "intermediate consumption" explains half of it.

But don't forget: The part of GDS that is all tax deductions is twice the size of GDP.

"Gross Domestic Spending" is not "Gross National Expenditure"

After coming up with the very satisfying description "Gross Domestic Spending" it took me a few days to Google the term. I didn't want to discover that someone else was already using it to mean something else.

Gross Domestic Product is all the stuff produced here in a year, or all the final spending we do in a year. (That's what you'd think, anyway.) Gross Domestic Spending is all the spending it took to produce all that stuff, that year. It's a really good term. It has symmetry -- the "G" and the "D", I mean. And the product is the product, and the spending is the spending. The meaning is just very clear. That's what I think, anyway.

And if you repeat the phrase a few times, it starts to seem natural. "Gross Domestic Spending" seems a more natural term to me now than "Gross Domestic Product". I like that.

But Google it I did. I think I'm safe. The term doesn't seem widely used for anything else. But I did find a few results.

Actually, I Googled "gross domestic spending" in quotes like that. I should get exact matches. The first result was to Gross Domestic Product (GDP) - Bureau of Economic Analysis but the blurb says "A description for this result is not available because of this site's robots.txt..." That's funny, I think. What is BEA hiding and why are they hiding it?????? Anyway I went to the page and found nothing matching "gross domestic s".

I tried the third result next: BDLive, from Johannesburg apparently. The article talks of "gross domestic expenditure" as if it was GDP, forecasting growth rates and describing the consumer component and like that. Reminded me of something I read at Wikipedia the other day. I didn't read much of either, actually. But we'll come back to that.

The second result is Gross national expenditure (current US$) | Data | Table from the World Bank. Turned out to be worth the click. Under the heading Gross national expenditure (current US$) they offer a definition of sorts:

Gross national expenditure (formerly domestic absorption) is the sum of household final consumption expenditure (formerly private consumption), general government final consumption expenditure (formerly general government consumption), and gross capital formation (formerly gross domestic investment).

Hard to read with all those parenthetical phrases, but okay. It's the sum of final consumer spending and final government spending and investment. Sounds like GDP. Maybe they leave off "net exports". But...

Oh, maybe that's the difference, net exports.

The number they list for the U.S. for 2012 is $16,791.8 billion. The number FRED gives for GDP for the U.S. for 2012 is 16,244.6. FRED's number is 96.74% of the World Bank number. The difference is 3.26%. That sounds about right for loss due to net exports...

I went to FRED to check, and got net exports as 3.37% of GDP for 2012. Pretty close to 3.26%. Close enough. You can do more years and check my work, if you want. If you do, let me know what you find.

Okay, so I'm saying "Gross National Expenditure" is GDP before they figure net exports.

That's not the same as Gross Domestic Spending.

Next, a look at what I found at Wikipedia the other day.

Context vs Context

The measure of all the stuff we produce in a year is called Gross Domestic Product or GDP. The measure of all the spending it took to produce that stuff is called Gross Domestic Spending or GDS. That's my name for the context variable I've been developing in the last few posts.

GDP includes final spending. GDS includes final and non-final spending.


I want to compare GDP to the spending it takes to generate the GDP. I want to compare GDP to Gross Domestic Spending:

Graph #1: Gross Domestic Product as a Percent of Gross Domestic Spending

It's about a third. GDP is about one third of GDS. For every dollar of GDP, there was about $2 of preliminary or "intermediate" spending. And that seems fairly stable.

There is a bit of a low spot on the graph, though, in the 1970s. Oh... the 1970s. Let's turn the graph other side up:

Graph #2: Gross Domestic Spending as a Percent of Gross Domestic Product

There. The low spot is now a high spot. In the 1970s. Hm.

Well, what can we say about this graph? It's pretty high up. GDS is in the neighborhood of 300% of GDP, three times GDP. The only other thing I remember being that high is total (public and private) debt. Oh, and maybe private debt by itself, that was up there too. But debt was increasing to get up to that level. Gross Domestic Spending is just *at* that level.

You could finesse this, but I don't think GDS being three times GDP is the thing that created the problem. If it was, the problem would have surfaced long ago. On the other hand, Gross Domestic Spending doesn't vary very much. Yeah, there is a small high spot in the 1970s during the Great Inflation. And yeah, there has been a tendency for the ratio to increase since the 1990s. But only a tendency.

Let me look again. Let me get rid of a lot of the empty space below the blue line on the graph, so the line shows more variation. And let's see how that high spot in the 1970s compares to inflation. I'll add two measures of inflation, the CPI and the GDP Deflator to the graph, on the right hand scale:

Graph #3: GDS as a Percent of GDP (blue) and Two Measures of Inflation

Well, look at that. The spending that generates GDP rose and fell, relative to GDP, in a pattern that bears surprising resemblance to the Great Inflation.

I know what you're thinking: Well, sure. Inflation pushed spending up!

Well, okay. But inflation pushed GDP up, too. There is inflation in both top and bottom of the GDS/GDP relation. Inflation cancels itself out of the calculation. The blue line on the graph does not run similar to inflation because inflation pushed spending up. I think the opposite is true. I think the increase in spending relative to GDP is what pushed prices up.

Velocity and the Context Variable

The measure of all the stuff we produce in a year is called Gross Domestic Product, or GDP. The measure of all the spending it took to produce that stuff is called Gross Domestic Spending, GDS. That's my new name for the context variable I've been developing in the last few posts.

GDP includes final spending. GDS includes final and non-final spending.


When economists speak of "velocity" they mean the number of times the money supply must be spent in order to do all the final spending that happens in a year. But I wonder what use that is. Because if they only count final spending, they only count a small part of the spending that happens. It doesn't make sense.

Type velocity in the search box at FRED, and the first thing that comes up on the list is Velocity of M2 Money Stock. It looks like this:

Graph #1: M2 Velocity a la FRED

I want to duplicate it.

I downloaded annual numbers for M2 from FRED and used them to plot "velocity" two ways: using GDP for the numerator, as FRED does, and again using my Gross Domestic Spending number for the numerator. Velocity is higher my way:

Graph #2: The Blue Line of Graph #2 is the Same as the Blue Line on Graph #1
Blue Line: Velocity of Final Spending ... Red Line: Velocity of Final + NonFinal

My blue line shows velocity at about 2 or a little under, just like the FRED graph. My red line shows velocity around 5 or 6 -- that is, 2½ to 3 times higher than the blue. That's what you should expect, as the sum of final and non-final spending is 2½ to 3 times higher than final spending alone.