FIP-to-NIP ratio ~ Real Interest Rate

Half as a joke, I started writing something that turned into the idea I call "industrial productivity":

To figure the labor productivity for our economy, we divide GDP by a cost factor, the hours of labor required to produce GDP. We want to do something similar when we ask how productive finance is.

But when we ask "How productive is finance?" we do not mean to ask how much the employees produce. We mean to ask how much the industry produces. The appropriate unit of cost is not the wage, but the profit. The appropriate ratio is not output-per-hour, but output-per-dollar-of-profit.

My natural inclination is to look at profit-relative-to-other-things, not other-things-relative-to-profit. But if the profit-per-dollar-of-output ratio is twice as high for financial corporate business as for nonfinancial, then output-per-dollar-of-profit must be half as high for finance. And if the productivity of finance is extraordinarily low, well, then the growth of finance is a big problem. And that is no joke.

 

I bring this up because I have one more graph to show. 

In Graph #1 here we compared profit per dollar of output for financial (FCB) and nonfinancial (NCB) corporate business. In Graph #2 we compared output per dollar of profit for financial and nonfinancial corporate business. And in Graph #3 we looked at the ratio for financial relative to nonfinancial, of the output per dollar of profit values. 

Table 1

The third graph shows that finance, on average, creates less than 40% percent as much output as nonfinancial corporate business, per dollar of profit. For example, if NCB produced $10 of output for every dollar of profit earned, FCB created less than $4 of output for every dollar it earned. Table 1 shows five dates when the NCB output-to-profit ratio was close to $10, and the FCB ratio for the same date. At the bottom it shows the five-value averages.

Reduced proportionally, as $10.18 falls to $10 even, $4.04 falls to $3.97. It's just dumb luck, really, that the FCB number comes out less than $4 for five sample values. For the whole dataset, however, it is not dumb luck at all. It is evidence. Per dollar of profit, the nonfinancial sector is far more productive than the financial sector. More yet, when you remove the "recasting" that exaggerates financial output.

 

But I have one more graph to show. We start with the third graph, the ratio of financial relative to nonfinancial industrial productivity. Then we add a second line, to show the real rate of interest.

Stephen Williamson shows "a crude measure of the real interest rate." I looked up his graph at FRED. He takes TB3MS, the 3-Month Treasury Bill Secondary Market Rate, and subtracts the percent-change-from-year-ago of PCEPI, the PCE price index. I used the same datasets and the same calculation for the second line on my graph.

PCEPI begins in 1959, so we miss some years at the start. And I end my graph early to eliminate the superhigh and superlow spikes of 2008, because they make everything else too small to see. But we still get to compare my ratio to the real interest rate for almost 50 years. 

To my eye, the comparison shows a lot of similarity:

Graph #4: Comparison of the FIP-to-NIP Ratio and the Real Interest Rate

The blue line is the ratio of industrial productivities, financial as a percent of nonfinancial, using the left scale.

The dark red line is Williamson's real interest rate, using the right scale.

The light red line that runs flat, just below the 40% level, left scale, shows the 37.9% average of all the blue values (Q4 1951 to Q3 2021).

The blue and dark red lines, to my eye, show similar trend paths, rising and falling together. It does look like blue leads and red lags behind; I'm looking at the blue lows of 1970, 1986, 1991, and 2001. And the high between 1986 and 1991, for example.

But red and blue show similarity all through the 1960s and into the 1970s. They separate sharply after September 1973, probably because of the oil embargo in October of that year. They are back on track, rising together, after 1977.

The only other separation that I see, comparable to the 1973 separation, occurs just at the end of the graph, after July 2007.

Abbreviations:

FCB = Financial Corporate Business
NCB = Nonfinancial Corporate Business
GVA = Gross Value Added
Industrial Productivity = GVA per dollar of Profit
FIP = Industrial Productivity of FCB
NIP = Industrial Productivity of NCB

The behavior of the FCB-to-NCB ratio is described as follows, other things equal:

Change in Financial Data:

Increasing GVA increases FIP as a percent of NIP
Increasing Profit reduces FIP as a percent of NIP
Increase in Industrial Productivity increases FIP as a percent of NIP

Change in Nonfinancial Data:

Increase in GVA reduces FIP as a percent of NIP
Increase in Profit increases FIP as a percent of NIP
Increase in Industrial Productivity reduces FIP as a percent of NIP

These changes appear to be related to changes in the real rate of interest.

Dis-inflating the GVA of Finance

The conclusion of my previous post was that financial corporate business (FCB) is less than 40% as productive as nonfinancial corporate business (NCB). That conclusion, however, rests on the assumption that everything counted in Gross Value Added (GVA) is output. 

The assumption is not well-founded. Dirk Bezemer and Michael Hudson, in Finance is Not the Economy, wrote

National accounts have been recast since the 1980s to present the financial and real estate sectors as “productive” (Christophers 2011).

The calculation of GVA has been "recast": It has been changed in ways that increase what counts as output. The more the output, the more productive a business is -- or the more productive it is reported to be, depending on whether the new inclusions are output in fact, or only in the calculation.

As for myself, I like to say "nonfinancial" businesses are the ones that produce the goods and services we buy, and "financial" businesses are the ones that produce the money we use to buy those goods and services. That's a bit of a simplification, but it provides clarity. It also invokes the image of financial business as entirely non-productive. 

I have not concluded that financial business is entirely non-productive. Nor have I concluded that it isn't. Thus I asked the question in the title of the previous post: How productive is finance? But my answer depends on the size of output as given by the GVA of finance. If the output of finance has been increased by changing what's counted to make output look bigger, then GVA overstates the output of finance, and my previous post overstates the productivity of finance.

The Recasting

It seems Bezemer and Hudson's "recast since the 1980s" remark is a simplified version of the story. In Financial Output as Economic Input: Resolving the Inconsistent Treatment of Financial Services in the National Accounts, Jacob Assa provides a brief history of the changes to the accounting of finance in the System of National Accounts (SNA):

At a first stage in the history of this question (SNA 53 and before), all financial intermediation activities were excluded from calculations of national output based on the value-added approach, since they were considered to be mere transfers of funds (similar to social security payments) and hence unproductive. An intermediate approach followed with the SNA 68, where the output of the financial sector was considered to be an input to a notional (i.e. imaginary) industry which has no output....
Finally, with the 1993 SNA, financial intermediation became an explicitly productive activity, for which value added is imputed based on the net interest received by financial institutions (the FISIM approach).

Assa points out that these additions to the GVA of finance are subtracted from other components of GVA: They are treated as costs to the industries using those financial services, "thus affecting only the relative size of the financial sector rather than the total GDP".

So, if I have it right, they are not using these accounting revisions to double-count finance and boost the reported size of GDP. But they do count financial costs as part of output, by including them in the Gross Value Added of finance. In this context, it is important to remember that with "SNA 53 and before ... all financial intermediation activities were excluded ... since they were considered to be ... unproductive."

The end result of this "recasting" of the national accounts has been to include more financial cost as financial industry output and to count it as part of the Gross Value Added of finance.

The Bezemer and Hudson article and the Jacob Assa paper both reference Brett Christophers's 2011 paper Making finance productive. In the Abstract of that paper, Christophers writes:

By placing different activities on different sides of a pivotal ‘production boundary’, national income statisticians effectively dictate what counts as productive – as adding value to the economy – and what does not.

His statement clearly describes the redistricting of data that increased the output attributed to financial business. In brief, then, Bezemer and Hudson were exactly on point: the national accounts have been "recast" to present finance as productive.

Calling something productive doesn't mean it is productive. Sure, we may be able to buy more cars because of finance. And we may be able to manufacture more cars because of finance. But finance didn't make the cars. It only made the money available. 

It made the money available, and sent us the bill. And yes, that may be fair. But "fair" is not the same as "productive".

How much output does finance create, really? Not much; this is clear from the graphs of the previous post. But those graphs use the "recast" Gross Value Added as the measure of financial output. The true measure of financial output is less than the graphs show. And that means the productivity of finance is less than those graphs show.

How much output does finance create, really?

Michael Sandel teaches political philosophy at Harvard University. In an interview with Sam Harris of Making Sense, Sandel said

It's been estimated by folks who know more about it than I do, that only about 15% of financial activity consists in investment in new productive assets for the economy. 85% consists of simply bidding up the price or betting on the future prices of already existing assets or, increasingly, synthetically created derivatives and other fancy financial instruments that have precious little to do with making the economy more productive.

The productivity of finance is about fifteen percent.

In an article at LinkedIn, Rana Foroohar wrote

Market capitalism, as envisioned by Adam Smith, was supposed to funnel our collective savings into productive investment, via the banking system. But today, deep academic research shows that only around 15% of the money flowing from financial institutions actually makes its way into business investment. The rest gets moved around a closed financial loop, via the buying and selling of existing assets, like real estate, stocks, and bonds. 

The productivity of finance is about fifteen percent.

Sandel and Foroohar agree on the number. Neither one documents it: no references, no links, no quotes, no sources, not even any detail. So I cannot accept the number. Sure, it's probably right, I will say that. But as presented, it is no better than a rumor.

But how much output does finance create, really?

 

The productivity calc is one thing, the GVA calc another

Using my calculation, GVA per dollar of profit as a measure of industry productivity, corporate finance is less than 40% as productive as the standard set by nonfinancial corporate business. On average, 37.9%. But if GVA overstates the output of finance, then 37.9% overstates the productivity of finance.

I'm not the guy who can determine what is and what isn't financial output. Mine is only a hobbyist's eye. To my eye, finance produces no output, only cost. I could guess that finance is only half as productive as the "Gross Value Added" says. I think it must be much less than half, but I certainly don't know. So to be fair (yeah, that again) fifty-fifty is the only guess I can make.

If half the GVA of finance can legitimately be considered output, then, on average, half of our 37.9% number is the answer. That brings the productivity of finance down below 20%, meaning finance is less than 20% as productive as nonfinancial business, per dollar of profit. More accurately, less than 19%.

That is not far at all from the 15% number of Sandel and Foroohar.

But then, again, if "output per dollar of profit" is a valid way to figure the productivity of industry, and if it was not used in figuring the 15% number, then we should want to use it and apply the "less than 40%" rule, reducing the 15% to less than six percent. 

And if we are now saying that finance reaches less than 6 percent of the productivity achieved by nonfinancial corporate business, well, we are almost back down to zero. Nonfinancial businesses are the ones that produce the goods and services we buy, and financial businesses are the ones that produce the money we use to buy those goods and services. The narrative is clear: Finance is entirely nonproductive.

How productive is finance?

Labor productivity measures how much output workers produce per hour. It is the ratio of output to employee time. But employees are paid for their time, often at an hourly rate. Saying output per hour is not far removed from saying "output per dollar of labor cost". 

To figure the labor productivity for our economy, we divide GDP by a cost factor, the hours of labor required to produce GDP. We want to do something similar when we ask how productive finance is.

But when we ask "How productive is finance?" we do not mean to ask how much the employees produce. We mean to ask how much the industry produces. The appropriate unit of cost is not the wage, but the profit. The appropriate ratio is not output-per-hour, but output-per-dollar-of-profit.

You could figure the productivity of an industry like finance by using components of GDP, if GDP was reported by industry. GDP isn't reported that way, but "Gross Value Added" (GVA) is. To figure the productivity of finance, you can use the GVA of financial corporate business, divided by the profit of financial corporate business.

Today's graphs present an example.

This graph, which we saw the other day, uses GVA as a measure of output. The graph shows profit per dollar of output:

Graph #1: Corporate Profits as Percent of GVA (output) for
Financial (red) and Nonfinancial (blue) Corporate Business

Finance runs high.

If we take that ratio and turn it upside down, we'll be looking at output per dollar of profit. We will get to see how much output is produced (and how much income is created) per dollar of reward to some bunch of corporations. Now, that is a measure of industry productivity!

Profit-per-dollar-of-output measures the benefit to the business that generated the profit. Output-per-dollar-of-profit measures the benefit to the economy that generated the output.

Profit-per-dollar-of-output has micro-economic significance. Output-per-dollar-of-profit has macroeconomic significance. 

//

This graph, then, shows the macro view:

Graph #2: GVA (output) as Percent of Corporate Profits for
Financial (red) and Nonfinancial (blue) Corporate Business

Here finance (red) runs low. Red is approximately half, maybe less than half the blue. Output per dollar of profit, for financial corporations, is half or less than half the output per dollar of nonfinancial corporate profit. Per dollar of profit, the benefit to the economy produced by finance is half or less than half the benefit produced by nonfinancial corporate business.

How productive is finance? Half or less than half as productive as nonfinancial business. 

//

Less than half. The nonfinancial measure of "output per dollar of profit" runs consistently higher than the financial measure. The nonfinancial measure sets the standard. How does finance compare? The next graph shows the financial measure as a percent of the nonfinancial:

Graph #3: GVA-to-Profit for FCB as a Percent of GVA-to-Profit for NCB

40 percent, about. Less than half. Give or take, the financial measure is about 40 percent of the standard set by the nonfinancial measure of output per dollar of profit.

The average, for all the quarterly (Q4 1951 to Q3 2021) data shown on the graph is 37.9%. Less than 40% of the nonfinancial measure, on average.

If output per dollar of profit is a measure of how productive business is, financial business is less than forty percent as productive as nonfinancial business. That's based on corporate "Gross Value Added" data and corporate profits. I'm not making it up.

//

Now... The calculation of Gross Value Added has been tinkered with, to make GVA of Finance bigger. There is evidently some disagreement about which parts of financial business activity should count as "productive" and which parts shouldn't. We'll review the history of this tinkering tomorrow.

For now, suffice it to say that when we count every dollar of GVA reported for corporate finance, and count it all as productive, financial corporate business is still only 37.9% as productive as nonfinancial corporate business. If we undo the tinkering and take out the changes that made the GVA of finance bigger, we make it smaller again. If we do that, finance will be less than 37.9% as productive as nonfinancial corporate business.

This all is based on output per dollar of profit, which makes good sense to me as a measure of industry productivity.