Friday, November 24, 2017

Grab a Barf Bag!

Here's a quote that would make Lars Syll retch:

Because DSGE models start from microeconomic principles of constrained decision-making, rather than relying on historical correlations, they are more difficult to solve and analyze. However, because they are also based on the preferences of economic agents, DSGE models offer a natural benchmark for evaluating the effects of policy change.

"... based on the preferences of economic agents, DSGE models offer a natural benchmark for evaluating the effects of policy change.

I think this is one of Syll's pet peeves! DSGE models are not "based on the preferences of" actual economic agents, but on simplified agents arising from "deductivist" assumptions. And, as Syll puts it:

When addressing real economies, the idealizations and abstractions necessary for the deductivist machinery to work simply don’t hold.

I like the MathWorks quote because it is a little bit of evidence that Syll is right. I have no idea who might be right about such things until I see a few bits of evidence like that.

I also like the MathWorks quote because it describes the alternative to DSGE models as "relying on historical correlations", which is the thing that makes sense to me.

// see also: Lucas critique?
// see also: Rethinking the same damn thought

Thursday, November 23, 2017

Sources of Income 1929-2016

Stacked FRED Graph showing Relative Shares of Income since 1929
Top to bottom: Interest income
Rental Income
Proprietor's income
Compensation of Employees

Source: FRED Table 2.1. Personal Income and Its Disposition: Annual

Wednesday, November 22, 2017

Hey Tom

Remember when I said

I predict a boom of "golden age" vigor, beginning in 2016 and lasting eight to ten years. It has already begun. In two years everyone will be predicting it.

Tom, you said

Art goes out on the limb.

Remember? That was 19, almost 20 months ago. Have you noticed that the mood is more upbeat now? Rather than predicting recession, people seem to think we can get 3% growth.

We'll beat 3%.

Tuesday, November 21, 2017

Six years ago at Asymptosis...

Concise, clear, and correct:
Ultimately, of course, private debt and interest charges on that debt needs to be paid off from real-economy income. Asset prices can’t rise forever.

Economists will tell you that gross debt levels don’t matter because one person’s debt is another’s holdings. (Net: zero.) They ignore it.

But if the gross private debt is too large, the real assets in the real economy can’t generate enough income to pay it off. Not really complicated, conceptually.

Monday, November 20, 2017

The return and maintenance of economic vigor

I started predicting vigor in March of 2016. Then in July 2016 I showed a graph of Household Debt Service Payments at a low point and ready to rise:
I called vigor because financial costs are down:
Graph #4: The Fall and Rise of Household Debt Service Payments
Financial costs have fallen from 13% to 10% of Disposable Personal Income. And now consumers have 3% of DPI to play with. Look at the extreme right end of the blue line on the graph: The line is starting to go up again. Already.

In March I said "We are at the bottom now, ready to go up." Four months later I showed it going up. Debt service is starting to go up again already, I said.

But then, a year after my original prediction of vigor, commenting in March 2017 on the graph shown above, I said

The uptick at the end of Graph #4 seems to have fizzled out. The link shows TDSP thru the end of 2016:

Unfortunately, I wasn't clever enough to capture the graph as an image, and TDSP has since changed. The link in my comment doesn't show what I was looking at in March of 2017, which was something like the red line here:

Graph #1: Household Debt Service as of March 2016 (blue) and March 2017 (red)
The tiny uptick at the right end of the blue line was gone (or perhaps just slightly lower, now that I see this comparison!) and the newer data was flat. But I was still confident. "Vigor will come," I said.

When I looked just now at the graph linked in my March comment, it was the current version. The data had been revised, and the flat end was gone. Debt service is rising again. It is struggling to rise (as you might expect following a debt crisis, while people remain cautious) but it is rising. I marked up an interesting similarity:

Graph #2: Household Debt Service as of October 6 2017
Two very brief, very tiny maximums in the data, one after each major decline.

I expect what happens after the second tiny maximum will look a lot like what happened after the first: uptrend, a sign of economic vigor. And since the more recent major decline is much bigger, I expect the uptrend will also be bigger this time. We will see economic vigor again, for perhaps a decade.

All that remains is for us to take advantage of the good years: Use the time to prevent another decade of bad years. All we need to do is limit debt growth to something sustainable. Better yet, we should seek the debt-to-GDP ratio that's most conducive to growth. We want to be like a surfer riding a wave of debt, where the wave carries us forward at speed but does not grow to the size of a mountain.

It can be done. To have a vigorous economy, we need lots of borrowing (and that day is coming). But lots of borrowing creates lots of debt. The obvious solution is to pay down debt: to pay down debt faster than we have done in the past. We need policy to make that happen.

Accelerating the repayment of debt may slow the economy some. But so does raising interest rates. We can use accelerated repayment of debt as an anti-inflation policy, along with the interest rate hikes that policymakers already rely on. If accelerated repayment is used, rate hikes will be needed less. And debt will be lower.

But there is something else: Raising interest rates affects all new borrowing, and that curtails economic growth. That's not the best outcome. Accelerated repayment of debt, on the other hand, affects people who have already borrowed. It only slows the economy for those people. People planning to borrow are not hindered by this new approach to inflation control, so economic growth is less severely reduced.

The new policy has to allow for the fact that we're starting with a high level of debt. It must reduce debt gradually, so that people are not thrown into crisis by it. That may seem to defeat the purpose, but it does not. As long as debt is going down, we're on the right track. And almost immediately after the level of debt begins to fall, the economy will begin to improve.

We never had policy that accelerates the repayment of debt. That's the reason private debt got so big. We have lots of policies that encourage the use of credit, and none to encourage repayment. To make policy more balanced, more sensible, we need policies that accelerate the repayment of debt. This way we can keep policies that encourage credit use (and enjoy the growth that results from those policies) but avoid having debt grow to an unsustainable level.

We still do have a mountain of debt to deal with. Some people might worry that "accelerated repayment" would make life difficult for debtors. But there is no reason the policy should be punitive. At the start, the new policy must help us reduce our debt: It must not be punitive. Once debt falls back to a level that is compatible with a vigorous economy (and stabilizes there) the policy can be made neutral. It need never be punitive.

Sunday, November 19, 2017

Financial Cost (4): F over N

Table 1.14 shows "Gross value added of corporate business" along with a list of the component parts of GVA. The table also shows GVA of "financial corporate business" and of "nonfinancial corporate business". There's no list of components for financial corporate, but there is one for nonfinancial corporate.

Add financial and nonfinancial GVA together on a graph, and they completely cover over the "corporate business" GVA. It's a near perfect match. So I'm thinking I can take the "corporate business" component list and subtract the "nonfinancial corporate" component values, and this will give me the "financial" components. Then I can compare the financial and nonfinancial corporate data.

I'll remind you that what they call "nonfinancial" I call "productive" business, and that what they call "financial" I call "nonproductive". For some reason, they put financial ahead of nonfinancial in Table 1.14. I'd put them the other way around, and use my labels. I guess it depends what you think is more important.

Here's financial-relative-to-nonfinancial GVA (red) nicely centered on corporate-minus-nonfinancial-relative-to-nonfinancial GVA (blue):

Graph #1: Financial GVA relative to Nonfinancial GVA
Since using "corporate less nonfinancial" gives me a perfect "financial", I'll just call it "financial" when I do the "corporate minus nonfinancial" calculation. Less messy. And then, to make sense of the descriptions of what the graphs show, I want to use abbreviations. I'll use a prefix letter, either F for financial or N for nonfinancial. And to describe the data, I'll follow the prefix with one of these:

 CE  Compensation of Employees 
 OS  Operating Surplus 
 GVA  Gross Value Added 

So, for example, the above graph is abbreviated as FGVA/NGVA, the ratio of financial to nonfinancial Gross Value Added.

For Compensation of Employees, financial gained on nonfinancial from 1945 to 1980, and then gained on it even faster:

Graph #2: Compensation of Employees, Financial relative to Nonfinancial: FCE/NCE
The graph starts at 1929, so there is not much to go on regarding the run-up to the Great Depression. But it looks as if FCE was high and gaining on NCE before and during the Great Depression, fell during the "correction", then rose again to the crisis. Looks like a high level of finance (F relative to N) is a sign of trouble. This reminds me of my Debt-per-Dollar graph.

In part (3) of this series of posts I had in mind to subtract net interest from corporate profits, but I found out that net interest is not included in corporate profits, so I threw that idea away. For the graph below, I went the other way: I used Operating Surplus, which is profits plus net interest plus some other thing.

Graph #3: Operating Surplus, Financial relative to Nonfinancial: FOS/NOS
For Operating Surplus, financial gained on nonfinancial consistently, with an interruption every 20 years or so. Some kind of cycle, maybe? It looks like the interruptions are getting longer and longer. I wonder if that's a good thing or a bad thing. Maybe it depends on how high the line is getting. Maybe low and short and good go together; and then high and long and bad go together. That could be useful policy guidance (if true).

I noticed that coming out of WWII, compensation of employees (Graph #2) increases from 0.04 to something over 12 by the end of the graph. But the operating surplus (Graph #3) increases to maybe 25 by the end. I got the impression that the operating surplus (net interest and profit) is increasing quite a bit faster than employee compensation.

Is it? Look at the CE ratio relative to the OS ratio, Graph #2 divided by Graph #3:

Graph #4: Compensation of Employees F/N Ratio relative to Operating Surplus F/N Ratio
Looks pretty flat to me, bottoming repeatedly at the 0.5 level. So I have to think the employee compensation and operating surplus F-to-N ratios are increasing at about the same rate. The one big anomaly occurs at the Great Recession. And, possibly, there are cyclical highs where Graph #3 shows cyclical lows. But there is no reliable up- or down-trend that I can see. So it looks to me that the operating surplus is neither gaining ground nor losing it relative to employee compensation. At least not in the financial relative to nonfinancial corporate data.

So I'm left thinking that the problem, if one is visible in this data, the problem is not found in the increase of profit and interest relative to employee compensation. Rather, the problem would have to be in the growth of corporate finance in general, relative to nonfinancial corporate business. That problem is clearly visible in Graph #1.

// The Excel file

Saturday, November 18, 2017

Financial Cost (3), aborted

I want to take "net interest" out of corporate profits, and compare what's left to "gross" interest paid.

If you look at National income: Domestic business: Corporate business: Compensation of employees at FRED, down near the bottom of the page they show RELEASE TABLES. Clicking the link for Table 1.14 brings up a page that identifies the hierarchy of components of "Gross Value Added of Domestic Corporate Business in Current Dollars".

The table shows that net interest is not counted in corporate profits. So then, I don't have to subtract it out.

Friday, November 17, 2017

Financial Cost (2): Inspection

I want to use GVA (gross value added) as the context for profits and interest expense. Better than GDP in this case, I think, because the GVA I'm using is "Gross value added of corporate business". The Financial Times lexicon says

GDP is commonly estimated using one of three theoretical approaches: production, income or expenditure. When using production or income approaches, the contribution to an economy of a particular industry or sector is measured using GVA.

So the GVA number I'm using measures the part of GDP produced by corporate business. I think that's a good context number for corporate profits and corporate interest expense.

The data I downloaded for mine of the 12th (the data I'm using here) includes GDP and RGDP. (From these I can get price deflator values if needed.) Also: interest paid and compensation of employees for "domestic corporate business"; GVA for "corporate business"; and "corporate" profits. I'm pretty sure these data series (except GDP and RGDP) are all numbers for "domestic corporate business", but I don't know for sure. If you know something I don't, let me know.

First graph shows employee compensation, monetary interest paid, and corporate profit, each as a percent of corporate GVA:

Graph #1:
I should say right away that "monetary interest paid" is not a component of corporate GVA. It is a measure of all the interest expense of corporations, or gross interest. Only "net" interest is included in GVA. The reason for this, no doubt, is to avoid "double counting" of income. But cost is cost, so I'm looking at the bigger number. More on this later.

Employee compensation (blue) as a share of corporate GVA runs pretty stable (more stable than Labor Share, which seems to show decline since 1960), running above 60% until the year 2000, when it starts to drop.

Corporate profits (green) decline from about 25% of corporate GVA to just over 10%. They run low in the Reagan years (interesting!) but pick up a bit in the latter 1990s, then start rising just as labor share starts to drop. Profits are now running above 20% again.

Total corporate Interest paid (red) is shown as a percent of GVA for purposes of comparison. Interest starts out low, less than 2.5% of corporate GVA, but rises to 10% by 1970. It peaks at 27.4% of GVA in 1982, runs high while corporate profits run low (the Reagan years), and peaks at 28.5% in 1989. Thereafter, the trend is slowly downward, but with increasing volatility.

It appears that interest gained at the expense of profits in the 1980s, and that after 2000 profits gained on employee compensation ("labor"). For what that's worth.

To see the relative changes in the three series on the first graph, I divided each series by its 1947 value so all three series start with the value 1.0:

Graph #2:
Indexing the values that way really magnifies the "interest paid" numbers, because that data starts at such a low level. And because employee compensation was so high on the first graph and shows relatively little change, the variation in that series is reduced to almost nothing; here, the blue line is almost straight. The green line, not so high on the first graph and varying more than the blue, still shows some variation on the second graph.

The first graph emphasized the different sizes of the three data series. The second emphasizes the changes in size. The red line really stands out here, but the extreme flatness of the blue is equally noteworthy.

I took profits and employee compensation from Graph #1, and subtracted them from total corporate GVA to see how much of GVA remains for other corporate spending. That's the blue line here:

Graph #3:
I put the red line from Graph #1 on this graph for comparison. It starts significantly lower than the blue line, but catches up by around 1980, and doesn't drop off much after that.

"Interest paid" is a big number. Big, and mostly unnecessary.

Thursday, November 16, 2017

Financial Cost (1): Overview of the data

This graph shows "percent change" rates for the data downloaded for my post of 12 November. Annual data, so the graph shows "percent change from prior year".

The first tall spike we come to, above 40%, between 1948 and 1952, that yucky blue line shows corporate profits. Corporate profits are pretty volatile: The line peaks repeatedly, both upward and downward, and this behavior is consistent for the whole 1948-2016 period shown on the graph.

That volatility is the reason I'm doing this follow-up to mine of the 12th. In that post I looked at corporate data for 1955 through 1958, a brief period during which the growth of profits began by increasing 26% over the previous year, and ended by falling 12% below the previous year.

Those growth rates suggest that corporate profits were well above trend in '55 and well below trend in '58. I wasn't thinking about that when I wrote the post. I was working with those years because those were the years Samuelson and Solow considered in particular in their 1960 paper.

The differences from trend no doubt exaggerated the movements of my "interest cost relative to profits" ratio, giving me the ten-point increase in interest cost (from 15% to 25%) relative to profits. So I'm thinking my numbers exaggerated the situation, perhaps significantly. That's why I'm taking another look.

After the yucky blue line, the red line stands out as also highly volatile. That's the interest cost. This one shows more variation than do profits, across the whole 1948-2016 period. The red line is pretty tame before the mid-1960s, staying generally between 10% and 20%. Then from the mid-1960s to the early 1980s the red peaks grow increasingly higher, and the down-pointing peaks increasingly lower. The growing volatility in this period likely reflects the inflation of the time. Finally, since the mid-1980s the red line seems to show some increase over time, but I'd say the main feature is that the movements here are very much bigger than those before the mid-1960s. They are about equal in size to movements during the 1970s and early '80s, but run lower: During the Great Inflation the red line didn't want to go below zero; since the mid-80s it has been centered on the zero level.

The next  line that catches my eye is the gold one, Corporate Gross Value Added (GVA). This line shows little volatility, little vertical motion. It's chief function on this graph seems to be to hide the other lines that display little volatility: nominal GDP (purple) and Employee Compensation (dark blue).

The one line so far omitted from my review of the graph is the greenish line, which shows the growth of GDP with inflation removed from the numbers. This line is mostly hidden by the gold line, except during the years from the mid-1960s to the early 1980s. Again, this is the time of the so-called Great Inflation so we should expect to see the green line run low in these years. And the fact that it does run low supports the view that the growing volatility of the red line in those years "reflects the inflation of the time.

Notice that the gold line and the lines that it hides show a hint of uptrend during the years of inflation. This isn't news. But I want to mention it now so that maybe I won't be fooled by it when I'm lost in the details of my analysis of the numbers.