Wednesday, June 30, 2021

What I find disturbing

This is what I find disturbing: People who focus on "how to survive and thrive during an economic collapse", rather than on how to prevent that collapse.

If people have given up on the idea of prevention it is only because everything we have tried so far has failed. But as Thomas Edison said,

“When you have exhausted all possibilities, remember this - you haven't.”

Monday, June 28, 2021

Prices rise to absorb the wages paid

Last time, looking at "percent change from year ago", Graph #3 showed the Consumer Price Index. #4 compared two price indexes, one for output and the other for wages, for the Nonfarm Business Sector.

Checking my work, I noticed what appeared to be a strong similarity between the wage index on #4 and the CPI on #3. Following up, we look now at that similarity. First, for the 1948-1962 period:

Graph #1: Consumer Price Index 1948-1962 (blue) and a
Nonfarm Business Sector Wage Index 1948-1962 (red)

Next, for all years:

Graph #2: Consumer Price Index (blue) and NBS Wage Index (red)

There are some differences, as at the 1980 peak (where wages don't keep up with prices) and for much of the 1990s (again, wages don't quite keep up). But these two datasets are remarkably similar. Shockingly similar, I'd say. Almost as if the one is calculated from the other.

Remarkable similarity, nonetheless. If the two series are calculated independently, I'd be tempted to paraphrase Parkinson's law and say prices rise to absorb the wages paid.

Sunday, June 27, 2021

ULC, 1955-57, newsworthy inflation, and life as we know it

Wage-Price Dynamics: Are They Consistent with Cost Push? by Yash Mehra in the Federal Reserve Bank of Richmond ECONOMIC QUARTERLY, Volume 86, Number 3, Summer 2000.

I got distracted by ULC,  Unit Labor Cost. In the first paragraph, Mehra says:

Hence changes in productivity-adjusted wages are believed to be a leading indicator of future inflation.

Then, in the first footnote:

The term “productivity-adjusted wage growth” refers to wage growth in excess of productivity gains, measured here by growth in unit labor costs. The empirical work here focuses on this measure of growth.

Unit Labor Cost is calculated by leaving inflation in the numerator and removing it from the denominator. In effect they divide nominal labor compensation by nominal output, and then multiply inflation into the result. The deceptive arithmetic of ULC makes labor compensation appear to increase faster than it does in fact. It also skews the resulting values toward a path similar to that of inflation.

When the resulting values are compared to inflation and similarities are observed, people tell themselves that Unit Labor Cost is a useful tool and an excellent "indicator of future inflation."

People can tell themselves whatever they want, but ULC is fraudulent arithmetic and the comparison to inflation is bogus.

I go back now to the data, to Unit Labor Cost. As you will see, however, I do not compare ULC to inflation, find similarity, and use the similarity as evidence that wage growth has caused the inflation.

Graph #1: Unit Labor Cost in the Nonfarm Business Sector, 1948-1962

The context -- my reason for showing this graph and for focusing on the mid-1950s -- is the extremely rapid growth of the labor force in 1955 and the creeping inflation of 1955-57 which arose as a result of the labor force growth.

  1. The extremely rapid growth of the labor force:

    Graph #2: 1955, the Most Rapid One-Year Increase in Labor Force Size, Bar None

  2. The creeping inflation, which shows up in the CPI in 1956:

    Graph #3: The CPI, 1948-1962

Doublechecking inflation:

Graph #4: Price Index for Wages (blue) and Output (red) of the Nonfarm Business Sector, 1948-1962

From 1950 to 1952, wage inflation and output price inflation rise and fall together. Between 1954 and 1956, output price inflation rises to a plateau a year before wage inflation mimics that pattern.

The four dots on each line indicate the four quarters of 1955. Prices (red) start rising even before the dots, in the last quarter of 1954. The red dots are all above the zero line (meaning that prices were rising) and the red line is rising (meaning that the price increases were getting bigger) all through 1955.

For wages (blue) the first three dots are below the zero line: Wages were falling until the last quarter of 1955.

Again, the red line shows an upward sweep of prices from the last quarter of 1954 to the first quarter of 1956, and then runs between 3 and 4 percent for a year. The blue line shows an upward sweep of wages from the last quarter of 1955 to the first quarter of 1957, and then runs between 3 and 4 percent for a year. Wage inflation came a year after price inflation: the same pattern, but a year later.

Prices went up first. Wages followed. That's some kind of cost-push inflation. But it isn't wage-push.

During the inflation of 1950-1952, the Korean war-related inflation, prices and wages went up together. That was not any kind of cost-push inflation. It was demand-pull.

This topic keeps coming up because people say there is no such thing as cost-push inflation. But there is such a thing. I just showed it to you.

It keeps coming up because people say that even if there is cost-push inflation -- they call it "supply shock" inflation now -- the inflation doesn't last long because "shocks" are temporary. That came up in the news just the other day. Couple months in a row now we got newsworthy inflation. One of the news topics the other day was the question of whether this inflation now, in the months ahead, will be "temporary" or "sustained".

It matters because if the inflation is temporary the Federal Reserve won't do anything about it. They'll just ride it out. If the inflation is sustained, they will have to fight it. That is a change in thinking at the Fed. For many years, they thought they had to undermine inflation even before there was inflation, when inflation was only expected. Hm.

You might say the creeping inflation of 1955-57 was temporary. Compare it to the inflation of 1965-1984, which was sustained. In the process of slowing the economy to fight that sustained inflation, the Fed created recessions in 1969, 1973, 1980, and 1981 -- plus a "near-recession" in 1966. I'd guess they are hoping for temporary inflation this time around.


I started out looking at a paper by Yash Mehra, but I got distracted. So I jump now to the conclusion of Mehra's article to get the short version:

The cost-push view of the inflation process that is implicit in the expectations-augmented Phillips curve model assigns a key role to wage growth in determining inflation. In this article, I evaluate this role by investigating empirically both the presence and stability of the feedback between wage growth and inflation during the U.S. postwar period, 1952Q1 to 1999Q2. The results indicate that wage growth does help predict future inflation over the full sample period considered here.  However, this finding is very fragile, and it appears in the full sample because the estimation period includes the subperiod 1966Q1 to 1983Q4 during which inflation steadily accelerated.  Wage growth does not help predict inflation in two other subperiods, 1953Q1 to 1965Q4 and 1984Q1 to 1999Q2, during which inflation remained low to moderate. In contrast, inflation always helps predict wage growth, a finding that is both quantitatively significant and stable across subperiods. These results thus do not support the view that wage growth has been an independent source of inflation in the U.S. economy.

Pretty interesting. Not the easiest thing for a hobbyist to read, but interesting. Mehra is saying that the model is wrong, the model that "assigns a key role to wage growth" is wrong. That's definitely interesting.

Mehra's finding is that wage growth was driving inflation in the years of high inflation, the 1966-1983 period. And that wage growth was not driving inflation before or after the high-inflation years. This makes sense I think, because if wage demands are less vigorous, the inflation will be less vigorous.

Another point worth noting: If, as Mehra says, wage demands were not driving inflation in the 1953-1965 period, but there was inflation, then something else must have been driving it. And again in the 1984-1999 period, something other than wage demands must have been driving the inflation.

We need to know what that other cost is, so we can stop blaming workers and wages.

We need to know what that cost is, so we can reduce or eliminate the cost problem, as this is by far the best way to reduce or eliminate cost-push inflation.

And most of all we need to know what that cost is, so that we can reduce or eliminate the cost pressure that slows economic growth. Because civilization is like a shark. If it doesn't keep moving forward, it will die:


By the way, the "other cost" is the cost of finance. Excessive finance. (Just in case civilization still matters to you.)

Saturday, June 26, 2021

The "always and everywhere" rule makes no such distinction

Thomas Tooke, in An inquiry into the currency principle, page 133:

It cannot be too often repeated, that abundance of money in the hands of capitalists is in itself a cause of cheapness, while abundance of money in the hands of consumers is a cause of dearness.

Thursday, June 24, 2021

POOF! Our problems are gone

From "The Battle against Recession", the 22 April 1958 statement by Fed chairman William McChesney Martin:

The growth of business capital spending beginning in early 1955 was at a rate that was unsustainable. An economy with a long-run upward growth trend of about 3 or 4 per cent per year cannot sustain for long an increase in business investment of about 10 per cent per year ...

The Fed chairman seems almost careless, referencing "3 or 4 per cent" annual growth. He makes it sound like what, 3½ percent maybe? Pretty close to the 3.3% number that sticks in my mind. 

I took another look around the internet:

  • A page at ResearchGate gives the average US Real GDP growth rate at 3.04% for 1960-2015.

  • Google turns up this quote from TradingEconomics:
    GDP Annual Growth Rate in the United States averaged 3.09 percent from 1948 until 2021...

    but I don't find it at the link.

  • YCharts has a "long term average" of 3.18% for the US Real GDP growth rate.

  • I still remember Marcus Nunes, back in 2012, saying that
    It´s more or less recognized that US RGDP is trend stationary (maybe that´s changed now!), with real growth averaging about 3.3% from the early 50s to 2007.

These days it is difficult to find a long term average value for the real GDP growth rate, even on the internet. Maybe Marcus was right: The "trend stationary" concept has been purged from our thoughts. Or maybe it just got lost, like the Middle Ages was lost in the computer in the James Caan version of the Rollerball movie.

I think William McChesney Martin had a vested interest in making his listeners think growth was closer to 3% than 4%: He wanted to make the 10% growth rate of investment seem definitely unsustainable. Surely, 10% investment growth must be less sustainable if GDP growth is 3% than if it is 4% or more.

A 10% rate of investment growth is only about twice as much as 4% annual GDP growth. That isn't hopelessly unsustainable: Investment, after all, boosts the economy's ability to grow.

Statista shows total investment in China running between 30% and 40% of GDP -- easily twice the US GPDI -- between 1980 and 2020. Apparently that was sustainable.

"Average" growth is a number that applies to all the years you include in your calculation. Based on the average, the growth in any one year is just as good as the growth in any other. Based on the average for 1948-2021, for example, economic growth in 2020 was just as good as growth during President Reagan's "Morning in America", 1984.

It is more honest, a lot more honest, to consider the changes in the average as time goes by. For example, here's what I did find at TradingEconomics:

The United States is the world’s largest economy. Yet, in the last two decades, like in the case of many other developed nations, its growth rates have been decreasing. If in the 50’s and 60’s the average growth rate was above 4 percent, in the 70’s and 80’s dropped to around 3 percent. In the last ten years, the average rate has been below 2 percent...

On average, things are as good today as they ever were.

Wednesday, June 23, 2021

A plausible cause of the 1955 labor force increase

From "The Battle against Recession", the 22 April 1958 statement by Fed chairman William McChesney Martin:

The growth of business capital spending beginning in early 1955 was at a rate that was unsustainable. An economy with a long-run upward growth trend of about 3 or 4 per cent per year cannot sustain for long an increase in business investment of about 10 per cent per year in real terms, such as we experienced in 1955-56.

If, as we noted the other day, changes in labor force size are "procyclical" and are "induced by changes in demand" then if a massive increase in investment occurs, maybe we should expect to see a massive increase in the labor force as well.

Again: If there is a massive increase in investment, maybe we should expect a massive increase in the demand for labor, in labor force size, and in employment.

Maybe we should:

Graph #1: Real GPDI (right scale, blue) and Total Employees
on Nonfarm Payrolls (left scale, red)  1950-1962

Compare the increase following the 1954 recession to the one following the 1958 recession.

For the blue line, the two "humps" are similar in size and shape. But the hump before 1958 is of longer duration than the one after.

The same is true for the two red humps: similar, but with longer duration before 1958.

The "after 1954" and "after 1958" humps are of comparable size.  By contrast, the red and blue humps after the 1961 recession are lower than those following the 1954 and 1958 recessions. The humps before the 1954 recession are higher, with low secondary humps.

What else?

  • For the "after 1954" hump, the longer duration of investment is matched by a longer duration of employment. For "after 1958", shorter occurs with shorter.

  • Beyond that, for the whole 1950-1962 period shown on the graph, employment seems to follow investment up and down with a small lag. The red and blue are quite obviously related.

  • The lag suggests that investment leads, and employment follows.

The longer durations after the 1954 recession seem to indicate greater levels of both investment and employment in 1955 and 1956. Building on Chairman Martin's observation, then, the investment boom of 1955-56 was closely followed by an employment boom: Greater investment led to greater employment.

It works for me.

I like this explanation of the unusual 1955 increase in labor force size. I like it, but I don't trust it. I need to find more ways to look at the data, for one thing, to see if it continues to show what I think I see on the graph above. And I need to look at other datasets that show the employment level, to see if they agree with the data used here. I need to poke around in it for a while.

Where I say the longer durations indicate greater levels of investment and employment, that has to be checked, too. The graph shows growth rates, not "levels".

The main thing, for the way I work, is that I just have to sit on this and wait until it isn't "new" to me anymore. Then I can look at it and see more clearly. One advantage of being a hobbyist: I don't have to have it done by 4 PM.

But I do like it. If business was investing more than usual, then I would expect them to be hiring more than usual, and that may just explain the unusual labor force growth.

It's certainly a better theory than "everyone wanted a '55 Chevy."

Tuesday, June 22, 2021

Finegan follow-up

Part of that quote again, from T. Aldrich Finegan:

Only one postwar recovery -- namely, 1954-55 -- was followed by a prolonged period of full employment when the country was not at war. And that period (1955-57) was very unusual in several respects. As table 3 reveals, real GNP rose only 1.5 percent a year -- less than half the average rate for the postwar period (3.7 percent). At the same time, growth in employment (1.2 percent per year) was abnormally high relative to the growth of real output but abnormally low for a full employment period.

1. Growth in employment was abnormally high relative to the growth of real output

Graph #1: Real GDP Growth (blue) and Employment Growth (red)

I'm looking at GDP, not GNP, but there is little difference between the two.

Real GDP (blue) growth shows a series of high peaks until the mid-1980s. These occur mostly as the "burst" of growth that follows a recession. After the mid-1980s, instead of the high peaks we had the "great moderation".

Employment growth (red) follows the same general pattern, usually at a lower level. 

It is easier to compare the red and blue if we subtract the one from the other and just look at the difference. I want to know about "growth in employment" so I'll start with employment growth and subtract the real GDP growth. 

What does that give me? Where the red line on Graph #1 is above the blue, on Graph #2 the line is above the zero level. In other words, where the line is above zero on #2, employment growth is faster than Real GDP growth:

Graph #2: Employment Growth less Real GDP Growth

On the first graph, the high points before the mid-80s were high points of growth. On the second graph, we subtracted Real GDP. So the high points on Graph #1 are low points on #2. And the lows of growth -- the recessions -- are high on Graph #2.

The gray vertical bars indicate times of recession. The blue line is high at every one of them, even after the mid-80s. (After the mid-80s the highs are lower, but that's a different matter.)

The blue line is also high between the 1954 and 1958 recessions. This high is not recession-related. Yet it is as high as any of the recession-related high points, except that of the 1974-75 recession. 

Between the 1954 and '58 recessions, this unusual peak runs high for the first three quarters of 1956. As T. Aldrich Finegan said, just one postwar recovery -- namely, 1954-55 -- was followed by a period when the growth in employment was abnormally high relative to the growth of real output.

2. Growth in employment was abnormally low for a full employment period. 

Full employment is "defined here as an unemployment rate of 4.5 percent", Finegan says in his 1972 article. In a footnote, he offers some historical context:

Throughout most of the postwar period, economists have typically used 4 percent as the benchmark for full employment. A somewhat higher rate seems appropriate now because a larger proportion of the current labor force consists of young persons (16 to 24 years) and married women, who have much higher unemployment rates than men 25 and over.

Here is a graph from the Monthly Labor Review of January 2018:

Graph #3

I'm going to figure that the times of full employment are the times on the graph where unemployment runs flat -- jiggy but flat -- when it reaches a bottom. Averages by eye:

  • just under 4% in the early 1950s
  • around 3% in the mid 1950s
  • a little over 4% in the latter 1950s
  • between 3 and 4% in the latter 1960s
Looks to me like unemployment in the mid-50s was abnormally low for a full employment period: the opposite of what Finegan said. But I'm not looking at "growth in employment". I have to look at that:

Graph #4: Employment Growth Decline in the Mid-1950s

Maybe. But it was a decline from a very high peak:

Graph #5: That 1955 increase was a big one!

Just looking at the employment level

Graph #6: The Employment Level, 1948-1962

I have to say that there was a large, rapid increase in 1955, and a slow increase in 1956 and '57. Growth in employment in 1956-57 was "abnormally low for a full employment period" because the 1955 increase was so big.

Monday, June 21, 2021


Graph #1: The Labor Force Participation Rate and the 1955

The 1955 increase in the Labor Force Participation Rate was two percentage points in just the one year. That's way more increase than any single year since that time. The long increase since the early 1960s? There's no comparison. The rapid growth of the Participation Rate occurred in 1955. It just didn't last very long.

But that makes 1955 all the more interesting: The trend was downhill consistently for the whole decade, except in 1955. So, what was the reason for the sudden, sharp increase in that one year?

Graph #2: Labor Force Participation in the 1950s: It's all downhill, except 1955

The reason? I don't know. I don't find anything about Labor Force Participation on the internet except baby boom and women in the workforce.

My gut says it was Korean war veterans coming home and entering or re-entering the workforce. But the timing seems off: The Korean war ended in July 1953. If the big, yearlong increase began late in 1953 or early in 1954, I'd say that's the explanation. But the big increase was a 1955 event, so I'm left hanging. 

The timing is wrong for demobilization after Korea. It would be a better explanation to say people just wanted to buy the '55 Chevys, and entered the labor force with that in mind. The biggest ever one-year increase in Labor Force Participation remains unexplained.

1954 Chevy
  1955 Chevy

I looked around some more, and finally found something. From History Today: Troop Withdrawals from Korea:

 U.S. soldiers in 
South Korea

Year Number

 Source: Wikipedia 

Two US divisions went home early in 1954, but it was not until a year after the armistice, in August, that the US Defence Department announced the withdrawal of four of the remaining six American divisions.

August of '54? So maybe the timing is not that far off from 1955. The connection is plausible. Let's look at the numbers.

Counting the 1954 and 1955 reductions both, troop strength dropped from 326,863 in 1953 to 75,328 in 1955:  251,535 troops returned from Korea. Call it 252 thousand.

The civilian labor force increased from 63,312,000 in December 1954 to 66,445,000 in December 1955, an increase of 3,133,000 people. But that's based on "seasonally adjusted" workforce numbers. The "not seasonally adjusted" workforce increased from 62,999,000 (Dec 1954) to 65,869,000 (Dec 1955). That's an increase of 2,870,000 people. That's probably the more accurate number.

You know what? Just call it three million: There was a workforce increase of about 3 million people in 1955. 

Where did these three million people come from? 252 thousand of them, at most, could have been recently returning Korean war veterans. That's a far cry from three million.

Returning Korean war veterans alone do not account for the whole 1955 increase in the U.S. labor force. But maybe it's not just U.S. military personnel returning from Korea. At History in Pieces, David Coleman provides totals for military personnel. For 1954 the number is 3,302,104. For 1955, 2,935,107. The difference is 366,997. Call it 370 thousand. (That includes the 252 thousand returning from Korea.) But at 370 thousand we are still far short of the 3 million we need to account for the increased size of the labor force. Changes in military personnel alone do not account for the 1955 increase.

Maybe it was unions? From America's Best History, in their U.S. Timeline - The 1950s:

December 5, 1955 - The two largest American labor unions, the American Federation of Labor and the Congress of Industrial Organizations, merge to form the AFL-CIO, boasting membership of fifteen million.

Fifteen million unionized workers. Maybe three million of them were entering the labor force for the first time? But it seems not. At the Hoover Institution, in The Decline of Unions Is Good News, they note "the 35 percent union membership high mark last seen in 1954."

It appears that unions contributed little to the 1955 increase in the labor force.

How many people were born in the US, all told, in 1955? According to InfoPlease, 4.1 million babies. Newborn babies, of course, cannot join the labor force. But people born, say, in 1935 or 1940 might join the labor force in 1955. 2.4 million babies in 1935; 2.6 million in 1940. With that bigger number, we are in the neighborhood of the three million we are looking for.

But every last one of them would have to join the labor force in 1955. That's not realistic. Labor force participation varies, but you never have everyone entering the labor market. And why would they wait until 1955 and then all decide to look for jobs at once?

There is no way. Forget it.

How about immigration, 1954-55? Wikipedia shows an average of about 250 thousand immigrants per year for the decade 1950-59.

They show 249,187 in 1950 and 265,398 in 1960 -- suggesting gradual increase, which we might expect to see. But they also show only 237,790 for 1955. The number was low in 1955. Less than average. So there would be no way that 1955 immigration provided enough workers to swell the ranks of the labor force by almost 3 million workers.

So far we've got 370 thousand returning veterans, and say 240 thousand immigrants. Total, 610 thousand possible new entrants into the labor force -- at most -- of the 3 million we must find.

I have not discovered much of an explanation for the massive 1955 increase in the labor force. Demobilization after the Korean war contributed at most 8% of that increase. Or perhaps 12%, if we include not just Korean war vets but all returning veterans, and if all of them enter the workforce. 

There was no baby boom in the late 1930s that could account for the unusual 1955 increase in the labor force. Nor was there any great burst of immigration that could account for it. What immigration there was in 1955 boosts our 12% to perhaps 20% of the three million we're looking for. And don't forget, we're using numbers that are on the high side, and we assume everyone we count joins the labor force.

There is no explanation that I can find, to account for the 1955 increase. We can, however, look at components of the labor force, by age group, by sex, by race, to see what happened there.

By age, the biggest age group, 25-to-54 years old, provided the biggest part of the 1955 increase: 1262 thousand people. The next largest part, 783 thousand new workers, came from the smallest group, age 16-to-19 years. Taking each increase as a percent of the age group's 1954 size shows an increase of more than 21% for the 16-to-19-year-olds:

Labor Force by Age Group

This increase appears to be the first of three large increases for the 16-to-19 age group, the last of which is surely an effect of the baby boom. But the next older group, 20-to-25 years, shows three large increases at the same times; and I'm not sure that is a baby-boom effect. So I still have no explanation.

Broken down by sex, women added more to the labor force than men in 1955: 1918 thousand versus 1215 thousand. The increase, shown as a percent of the group's 1954 size, is even more impressive:

Labor Force by Sex

For women, almost a 10% increase!

If you look at all the years, the count of women in the workforce was growing faster than that of men almost continuously from start-of-data to the 1990s; then off and on since the '90s.

By race, I can't say. The data starts in 1954 for "White", in 1972 for "Black or African American", and in 1973 for "Hispanic or Latino". I'm looking at 1954-55.

So where did people come from, to make the 1955 labor force increase possible? Out of the woodwork, apparently. It's not like there were "extra" people, as in the baby boom. Women, I guess, were "extra" in the sense that they were transitioning to "working" from "not working". I can check that.

Women in the workforce, December 1954: 17,973,000
Women in the workforce, December 1955: 19,460,000

That's an increase of 1,487,000: A million and a half women joined the workforce in 1955. Holy shit. That doesn't account for all of the 1955 increase, but it is half of the 3 million we're looking for.

So now perhaps we have some idea of the source of the 1955 increase in the labor force, some idea where those three million people came from: They came mostly from the "not in the labor force" population and, apparently, they were mostly women.

Now I know why people talk about "women in the workforce" all the time.

But a question remains: What happened to cause the 1955 increase? No idea. Was there some widely popular movie that changed the culture, or a popular book that changed our thinking? 

I dunno. Maybe it was just that everyone wanted a '55 Chevy.

Sunday, June 20, 2021

Data frequency, and aggravation type

Watch that high point, the 5% in 1955.

Monthly Frequency (FRED default)


Annual Frequency, FRED Aggregation Type "Average"


Annual Frequency, FRED Aggregation Type "End of Period"

Saturday, June 19, 2021

Finegan again

Finegan describes data I can use to look for the 1955 increase in labor force.

Given my focus on 1955-57, I found the most interesting bit of Finegan's article to be on page 31 where he writes:

... the 2.5-percent rise in the labor force between 1969 and 1970 was equaled only by the increase between 1968 and 1969; only four other year-to-year increases (1947-48, 1954-55, 1955-56, and 1966-67) were larger than 2.0 percent.

Take another look at the FRED data:

Graph #1: Labor Force Growth thru 1980

The 1956 increase is NOT larger than 2.0 percent. And the 1955 increase is way bigger than 2%. If that one was 2% I wouldn't have noticed it. It would have been like 1959-60 on the graph: No big deal.

Dunno. Maybe there was some revision in the data, after Finegan wrote his article. I have to track that down. I still want to find the explanation of the super-high 1955 increase in the labor force.

I don't want to go to FRED for this. I don't want current data. I want to see the data Finegan was looking at in 1971-72. He identifies the data source at the bottom of Table 1:

Yeah, I'm not finding the "1971 Manpower Report of the President" on the internet. The Economic Report of the President comes to mind. I'll go with Table C-22 of the 1971 ERP.

Finegan's numbers for the total civilian labor force: 80,733 for 1969 and 82,715 for 1970. "Numbers in thousands". I find exactly the same numbers in the "Total" column (to the right of the "Armed Forces" column) in Table C-22. I'll go with that -- for "persons 16 years of age and over".


It's not so bad, typing the data in, if the data stops in 1970. Less than 25 numbers to type. Come to think of it, that's how I started, paying a quarter apiece at the library to photocopy pages from the Historical Statistics, Bicentennial Edition, and entering data thru 1970.

Entering data? Into what?? Nah, I didn't have to enter the data back then. I did my graphs by hand.


Here is the "civilian labor force level" data for 1948-1970, a comparison of current data from FRED and 50-year-old data from the 1971 Economic Report of the President:

The ERP data generates the wide blue line. The FRED data gives us the yellow line, not perfectly centered atop the blue, but not far off. The numbers have changed little in 50 years. Hmm.

Okay, so next I need the "percent change" values to see annual labor force growth rates.

I did the figuring, and peeked before making the new graph. Finegan talks of increases "larger than 2.0 percent" for 1955 and 1956 growth. As soon as I figured the growth rates, I looked:

 ERP DataFRED Data
 1955  2.17% increase  2.03% increase 
 1956  2.35% increase  2.42% increase 

Larger than two percent all around, as Finegan says. And no 5% increase in 1955. WTF!

I'm looking at annual data, not monthly. I notice that the monthly data varies: there is a big increase in June for example, at the end of the school year. It's not like debt, where every number is bigger than the one before until you have a crisis.

The annual values -- at least in the ERP -- are "averages of monthly figures". I got the 5% increase for "percent change from year ago" of monthly data. In my case, the default was monthly data. For Finegan, the default was annual, it appears. Maybe this accounts for the difference.

So I still don't know what accounts for the 1955 increase of the labor force. I still kinda think it must have been Korean war veterans returning to a peacetime economy. But I didn't find anything on that, and anyway the Korean war ended in July 1953. What were those people doing for a year and a half ? Apparently not joining the labor force in droves.

Time to look at monthly data.


I checked every Economic Report of the President from 1953 to 1968, looking for Noninstitutioinal Population and the Labor Force or something like. All of them list data for "persons 14 years of age and over". The 1968 edition was the first where I also found data on "persons 16 years of age and over".

I guess I can just use the "age 14" data, as it covers the mid-1950s. But I didn't know the "age 16" stuff wasn't available. And FRED's labor force data is age 16 and up. So I spent the morning looking at old ERP files. Good thing my head don't spin easy.

Or... Historical Statistics: Data thru 1970, and in particular Table D 11-25, which lists the labor force as age 16 and up, might be just the ticket. There is a note, however.

The introduction of data from the decennial censuses into the estimation procedure in 1953 and 1962 and the inclusion of Alaska and Hawaii beginning in 1960 have resulted in 3 periods of noncomparability; see text.
The "text" for Table D 11-25 says in part:
In 1953, population data from the 1950 census were introduced into the estimating procedure, affecting the comparability of the labor force figures with earlier years. Population levels were raised by 600,000; labor force, total employment, and agricultural employment levels were raised by 350,000, primarily in the figures for all persons and for males. Similarly, population data from the 1960 census were introduced in 1962, reducing the population totals by 50,000 and the labor force and employment totals by 200,000.

The inclusion of Alaska and Hawaii in 1960 resulted in an increase of about 600,000 in population and 300,000 in the labor force, four-fifths of which was in nonagricultural employment.

These adjustments are close in time to 1955 but seem to miss it. They are big numbers, big enough maybe to create a 5% increase in the labor force, but ...
anyway I have to look.

If data for 1953 is not comparable with data for 1952 and before, and data from 1960 is not comparable with data from 1959 and before, then we can safely look at data for the 1953-1959 period. That might be just what I need.

Friday, June 18, 2021

T. Aldrich Finegan on 1955-57

During the baby-boom upswing of the Labor Force Participation Rate, after the 1970 recession, the Monthly Labor Review published an article and noted that its

Alternative projections of labor force growth suggest difficulties in returning to a 4.5-percent unemployment rate

T. Aldrich Finegan: "Labor force growth and the return to full employment" at JSTOR. Monthly Labor Review, February 1972

Looking back in time, Finegan writes (page 35):

Only one postwar recovery -- namely, 1954-55 -- was followed by a prolonged period of full employment when the country was not at war. And that period (1955-57) was very unusual in several respects. As table 3 reveals, real GNP rose only 1.5 percent a year -- less than half the average rate for the postwar period (3.7 percent). At the same time, growth in employment (1.2 percent per year) was abnormally high relative to the growth of real output but abnormally low for a full employment period. Evidently, only a rare combination of exceptionally small growth in output per worker and in the size of the labor force kept unemployment from rising during these two years.

He sees

  • slow economic growth
  • relatively high employment growth
  • low labor productivity growth

That's half a confirmation of my answer to the question about the cause of the 1955-57 inflation, except Finegan draws no conclusions about inflation. His focus is the 1970s, the baby boom entering the workforce, and unemployment.

Matter of fact, a lot of the argument in the late 1950s and early '60s was about whether the 1955-57 inflation was cost-push or always-and-everywhere inflation. Because it was a hot topic, it was easy for people to see what they wanted to see in the data. Finegan, writing in 1971-72 and focused on unemployment rather than inflation, was able to evaluate the data with a cool head. I think so, anyway. I have more confidence in his evaluation of the '50s than I do even in Samuelson and Solow's. Hindsight, and all that.


Finegan sees full employment in 1955-57, but also unusually slow economic growth. That combination screams "low productivity!" 

We have already seen the low productivity and its consequences. But if Finegan's details are right, we should be able to see that low in a graph of real GDP relative to the number of workers in the economy. Come to think of it, yes, that is one way to measure productivity:

Graph #1: Real GDP per Worker, with HP showing low productivity
between the post-recession bounces of 1955 and 1959

I showed the data in Excel and put a Hodrick-Prescott on it to see the 1956-57 low. Then I set the background to "no fill", copied the plot window, and pasted it over the FRED graph, matching Excel's GDPC1/PAYEMS line with the line in the FRED graph to get the fit right. Where there's a will there's a way.

Unrelated: I thought Finegan's method of identifying "recoveries" and "expansions" was interesting and might be useful:

The historical periods were selected in the following way. In the case of previous recoveries, the criterion was that each period should begin with the quarter immediately preceding that in which real GNP resumed growing and should end with the quarter in which the unemployment rate stopped falling. Two exceptions to this rule were made ...

In the case of the three "other expansions," the criterion was a generally sustained growth of employment not accompanied by a pronounced change in the unemployment rate...

Thursday, June 17, 2021

Remember when some people were saying the post-2008 decline of Labor Force Participation had nothing to do with economic conditions being so bad? It was baby boomers retiring, they insisted.

T. Aldrich Finegan in Labor force growth and the return to full employment (at JSTOR) from Monthly Labor Review, February 1972, page 30. (Page 2 of 11 JSTOR pages):

 "... it is well established that, other things equal, the labor force tends to be larger when (and where) labor market conditions are more favorable ..."

Alternatively, see Requirements for a Return to Full Employment, 1971-73  (PDF, typed, preliminary draft, text not selectable) 


I thought I was done with this one. Nope. At JSTOR I just found The Great Labor Force Projection Debate: Implications for 1980 by Marc Rosenblum, from the Fall 1973 issue of The American Economist. Rosenblum writes:

... new projections being issued by the Labor Department for 1980 fail to reflect a major lesson learned from the great debate over 1970 forecasts: projection accuracy depends in part on the extent to which actual economic conditions during the target year match the projection's full-employment assumptions. Changes in labor force size are, to that extent, induced by changes in demand at given wage rates...

Only seven years ago Jacob Mincer wrote that the procyclical behavior of the labor supply had finally been accepted, after "three decades of research and occasionally animated controversy."

... The two main critics of the Labor Department's forecasts, Alfred Tella and Thomas Dernburg et al., both held (correctly, as it turns out) that Government figures would understate the 1970 labor force. Along with their reasoning both offered projections based on their own regression models, avoiding the supposed shortcomings of inflexibility to economic conditions they both found (independently) in the BLS forecasts...

And that's just on the first page.


In these articles from the early 1970s, references to "full employment" policy are glaringly obvious. You don't see that in recent thinking. So different!

Tuesday, June 15, 2021

Cleaning up the Desktop

Instead of just moving files off my desktop, I decided to look at a few of them. That slowed things down. But I did get a blog post out of it.

Peak Interest Costs


Looking at that one, of course I wanted to see also the continuation of the 1960-1972 trend:

Just Looking


M1 money: 

Big increase in 2020

Some terms were redefined.  Savings was excluded from M1 before the change, and is now included.

This next one goes back to mine of 13 April, Answering an old question:

A price increase in 1950, let's say related to the Korean war. Then some remarkable price stability.
Then a price increase that begins in the second half of the massive 1955 increase in the Labor Force,
as adding all that unskilled labor led to falling productivity and rising business costs.


Looking at that graph, and now being more familiar with this data

  • price per unit
  • labor cost per unit
  • nonlabor cost per unit and
  • profit per unit,

I jumped to look at labor cost per unit as a percent of price per unit:

Unit Labor Cost as a percent of Unit Price of NCB output, 1950-1962

The drop in labor cost that was inspired by the 1954 recession ends in Q1 1955. The four dots there indicate the four quarters of 1955. The labor costs of nonfinancial corporate business ran low in 1955, and only started picking up in 1956: The increase of Q3 1955 is more like noise than an increase. (Compare it to the first increase after the low point of Q4 1950, or the first increase after the low of Q2 1959.) And from Q1 1956 to the 1958 recession is less than a one percentage point increase.

I think it is reasonable to say that in 1955-56, prices started going up before labor costs. Therefore, I have to say it was not wage-push inflation that drove prices up.


Price per unit is typically highest of the four data values, with labor the highest cost, then non-labor cost, and profits the lowest number of the four. But if you look at them on a graph, you just see four somewhat wiggly lines running across the page. So I indexed the four data sets to make them all the same value where they start, in 1947-01-01. Now we can see how they move, relative to each other:

Relative Movement of Unit Labor Cost (blue), Unit Non-Labor Cost (red),
Unit Price (green) and Unit Profits (purple)

With everything indexed and starting out equal in early 1947, profit (purple) is clearly the most variable of the numbers. The labor cost number runs almost unfailingly lower than everything else. That is due to the indexing; but the graph does show less variability in labor than in non-labor cost, and certainly less than in profits. And it is easy to see that non-labor cost increased faster than labor cost.

Note also that purple profits drop to low points when the red and blue costs reach highs. Based on this observation, we might think that at start-of-data, where profits are at a definite low point, labor and non-labor costs were already at high points that we cannot see due to the indexing -- and because no earlier data is available.

Well this is way more fun than just moving files off my desktop!

Sunday, June 13, 2021

The whole paragraph from Smith

According to the glossary of the U.S. Bureau of Economic Analysis, "Value Added" is

The gross output of an industry or a sector less its intermediate inputs; the contribution of an industry or sector to gross domestic product (GDP).

So I'm thinking that Gross Value Added is the gross output of an industry or sector, including the intermediate inputs. And total GVA would be greater than GDP.

The rest of the glossary entry:

Value added by industry can also be measured as the sum of compensation of employees, taxes on production and imports less subsidies, and gross operating surplus.


Value added for business and government adds up to 100% of GDP:

The rest of us are just consumers who add no value to the world. (Now you know.) So much for the wisdom of Adam Smith:

Consumption is the sole end and purpose of all production; and the interest of the producer ought to be attended to only so far as it may be necessary for promoting that of the consumer. The maxim is so perfectly self-evident that it would be absurd to attempt to prove it.

and Maynard Keynes:

Consumption — to repeat the obvious — is the sole end and object of all economic activity.


The whole paragraph from Smith:

Consumption is the sole end and purpose of all production; and the interest of the producer ought to be attended to only so far as it may be necessary for promoting that of the consumer. The maxim is so perfectly self-evident that it would be absurd to attempt to prove it. But in the [modern]mercantile system the interest of the consumer is almost constantly sacrificed to that of the producer; and it seems to consider production, and not consumption, as the ultimate end and object of all industry and commerce.

Friday, June 11, 2021

Secular Cycles: Population

Secular Cycles, by Peter Turchin and Sergey A. Nefedov

From Chapter One (page 19):

Most broadly the cycle can be divided into two opposite trends. In the literature these are sometimes called the positive "A Phase" and the negative "B phase," but we prefer the more descriptive terms integrative and disintegrative trends.

The population tends to increase during the integrative phase and decline or stagnate during the disintegrative one.



ABC News:

22 December 2020

US population growth smallest in at least 120 years

The U.S. population grew by the smallest rate in at least 120 years from 2019 to 2020, according to figures released Tuesday by the U.S. Census Bureau — a trend that demographers say provides a glimpse of the coronavirus pandemic’s toll.

Population growth in the U.S. already was stagnant over the past several years due to immigration restrictions and a dip in fertility, but coronavirus-related deaths exacerbated that lethargic-growth trend, said William Frey, a senior fellow at The Brooking Institution’s Metropolitan Policy Program.

“I think it’s a first glimpse of where we may be heading as far as low population growth," Frey said. “It’s telling you that this is having an impact on population."

The U.S. population grew by 0.35% from July 2019 to July 2020, an increase of 1.1 million people in a nation whose estimated population in July was more than 329 million residents, according to Census Bureau estimates.


From The Economics of Population Growth by Jonathan Rochford:

At the most basic level, population growth increases the total size of the economy including the demand for labour. There are more people purchasing goods and services so the economy grows to meet that demand. Politicians are positively disposed to population growth as it allows them to boast about economic growth and job growth...

However, this is a very shallow line of thinking ...

 Amen, brother.

From The Role of Population in Economic Growth by E. Wesley F. Peterson:

In his important book on inequality, Thomas Piketty (2014) observes that economic growth “ . . . always includes a purely demographic component and a purely economic component, and only the latter allows for an improvement in the standard of living” (p. 72). Economic growth is measured by changes in a country’s Gross Domestic Product (GDP) which can be decomposed into its population and economic elements by writing it as population times per capita GDP.

Right. That's a version of "growth accounting".

There is no greater hypocrisy than to proclaim one's desire to improve the economy when one's solution is to boost GDP by boosting the population, which of course does nothing to improve GDP per Capita.

Meanwhile, slow population growth itself is an indicator that society is entering, or has already entered its "B" phase, the disintegrative phase of the "secular cycle", the "The End Is Near" phase.

Toynbee of course says that our civilization does not have to follow in the footsteps of the 20-odd other civilizations he identifies, none of which survived. Dunno yet what Turchin and Nefedov say on that topic.

Thursday, June 10, 2021

Secular Cycles: "fiscal flows in real terms"

Secular Cycles by Peter Turchin and Sergey A. Nefedov

As I have noted, the authors use the language of our time to describe patterns that exist outside of time. They use economic and business cycle terminology to describe "secular" cycles that last perhaps 300 years: expansion and stagflation and crisis and depression. They also describe in modern terms problems that occur and recur over the past thousand years and more. For example:

 ... states have no choice but to seek to expand taxation, despite resistance from the elites and the general populace. Yet the amount of surplus production declines (as discussed in the previous section), and the state must compete for this shrinking surplus with increasingly desperate elites. As a result, attempts to increase revenues cannot offset the spiraling state expenses, and even though the state is rapidly raising taxes, it is still headed for fiscal crisis.
It is written as though it describes us and our time. It is not. The cases identified in the table of Contents include Russia since the 15th century, France and England since the 12th, and Rome before the time of Christ. The authors' style makes secular cycles more real for today's reader; and it suggests that the same things may be happening again, this time to us. It is almost deceptive.

But it is not deceptive. That the same patterns may be visible in our time is precisely the point of the book. Anyway, I think that if we were in the "expansion"phase of a secular cycle, the things they describe would sound great, and we wouldn't be at all troubled by their description. We are only troubled by it because Turchin and Nefedov seem to want to locate us in the midst of a troubled time.

Maybe the word "elites" doesn't seem quite right, but then maybe we're not quite as far along in the cycle as the description that I quoted.

As I see it, the book describes us and our time. We are indeed living through another of Turchin's secular cycles. History repeats itself. Or it doesn't repeat itself, but it rhymes -- as Mark Twain supposedly said. And the authors have found a perfect way to present their cycles in a way that makes us feel, for better or worse, it is now our turn at the wheel.


I was initially tempted to say that Secular Cycles is deceptively convincing because it is written in the language of our time, as an explanation of the problems of our time. But then I remembered a criticism of the great historian Rostovtzeff, by N. H. Baynes in "The Decline of the Roman Empire in Western Europe: Some Modern Explanations". Baynes summarized Rostovtzeff's argument and objected to it:

"The peasant of the army made common cause with the peasant of the countryside and both waged a war of extermination against their oppressors of the city. The explanation of the downfall of the aristocracy and with them of the ancient civilization is thus to be found in a class-conscious alliance between the soldier and the worker on the land. Professor Rostovtzeff, it must be remembered, has seen in his native country an aristocratic regime overthrown by a similar alliance."
 - In Decline and Fall of the Roman Empire, Donald Kagan, editor; page 83

Baynes, it seems, objected to Rostovtzeff's story of ancient Rome because the same story could be told of more recent times. But of course, that is the sort of thing one must describe, if one is describing a recurring cycle in history!

The next sentence, after the above quote from Turchin and Nefedov is this:

Note that declining real revenues may be masked by persistent price inflation, and it is therefore important to express all fiscal flows in real terms.

"In real terms" means "measured in prices that are unchanged by inflation" or measured in dollars of constant value, constant dollars we might say. Good idea. I hadn't thought of looking at federal spending in real terms.

Graph #1: Federal government Receipts 1929-2020
Blue = As Reported ... Red = Adjusted for Inflation

Note that the red line shows federal receipts "in real terms" running flat or slightly downhill since 2015. This is "masked by persistent price inflation" as the blue line shows.

Yes, indeed. Thumbs up for the Secular Cycles book.

Wednesday, June 9, 2021

Turchin and Nefedov: Secular Cycles

The book, a gift from my son Jerry, a covid-delayed christmas present.

From the Table of Contents -- partial and in brief. I added "subchapters include":

Chapter 1: Intro

Chapter 2: Medieval England; subchapters include:

  • The Expansion Phase
  • Stagflation
  • Crisis
  • Depression

Chapter 3: Early Modern England; subchapters include:

  • Expansion
  • Stagflation
  • Crisis
  • Depression

Chapter 4: Medieval France; subchapters include:

  • Expansion
  • Stagflation
  • Crisis
  • Depression


You get the idea.

I'm in. I like the authors' use of the word "stagflation" to make our time and these earlier times thoughtfully similar. I like the repeating, four-stage business cycle format. I like these Turchin cycles, uh secular cycles I guess, obviously far longer than a business cycle or a kondratieff, but shorter than a Toynbee (two thousand year) cycle. Turchin's secular cycles fill a hole in my picture of cycles within cycles within cycles.

I'm in.