 |
| Graph #1: Jacob Assa's Figure 2, US GDP (constant prices) and total employment, 1977=100 |
The graph shows Real GDP and employment running together until the early 1980s, when GDP accelerates and employment starts falling behind. Fascinated, I went to FRED to see if I could duplicate the graph using GDPC1 and PAYEMS.
Yes, I could duplicate the graph. Using data since 1977, indexed to 1977, the lines on my graph followed a pattern similar to Assa's. I even got similar index values for 2011 where his graph ends: 260 for GDP (versus Assa's 250) and about 160 for employment (versus 150).
Of course, when I created my duplicate graph at FRED, it didn't start with 1977. It showed all the data, as FRED does by default, going back to the late 1940s for Real GDP, and even earlier for employment. And when I looked at all the data, I didn't see GDP and employment "running together until the early 1980s". I saw them running together only for about ten years, beginning in the early 1970s.
 |
| Graph #2: Real GDP (blue) and Total Employment (red) |
"Under SNA 1953 the financial sector would have thus shown a negative £6.1 billion value added. “Adopting SNA 1968 had, in effect, made UK finance productive” (Christophers 130, emphasis in original)."Real GDP and employment do not run together in the early years, though it seems to me that by Assa's argument they should. And then they do run together for a decade mid-graph, after 1968, when by Assa's argument they should not. Assa's graph does not show the early years' mismatch, a mismatch that does not support the "financialization of GDP" argument. And then the graph shows about half of the "running together" decade in the middle, enough that we may imagine GDP and employment were running together from the start, as his argument implies. Thus, what remains is a graph which appears to support Assa's argument.
I felt deceived.
In a comment on mine of last March, Jerry reminded me that
it makes sense that RGDP should go up faster than the size of the workforce, even if there is no funny business involved with how finance is tracked -- from technology.Good point, but I still felt that Assa's argument was not really supported by the data. The only other option I can see is that I'm misinterpreting the argument. Maybe a year from now I'll read his stuff again...
Evaluating my graph showing all the data, Jerry added
Exponential graphs are hard to look at on a linear scale like this -- the right side always looks bigger than the left side, even if it's not really.Jerry thought it might be better to compare the growth rates of real GDP and employment. He presented a graph and said
It does look by eyeball there like there is a pre/post 1980 change. The blue is always above the red from 1980 to 2008.Yeah. But I think the change that strikes Jerry's eyeball there is called "the Great Moderation".
So I looked at the growth rates. Here's what I got:
 |
| Graph #3: The difference between the RGDP Growth Rate and the Employment Growth Rate |
That's about as far as I got, back in March, with my disappointment in Jacob Assa's argument. This time I want to stay on the case a little longer.
I downloaded the growth-rate-difference data to Excel and recreated the graph there. Looking at it, it seemed to me that the spikey data ended (and the "great moderation" began) with the data point at Q2 1984. So I figured two averages: one ending and another one starting at that date. This way I could compare the average performance of the period when the data seems to show the increasing financialization of GDP, to the period when it doesn't.
 |
| Graph #4: Same as #3, with Averages before and after Q2 1984 You can click the graph to see it bigger. |
The difference is less, in the late period. What does that mean? It means employment growth and RGDP growth are more similar in the late period than in the early period. It means I'm not finding the financialization of GDP that explains things for Jacob Assa. I'm not finding it.
I guess I could have done it differently. I could have looked at the original source data (my graph #2 above), picked data points for the time that the two data sets are running together, and excluded that period from my calculation. Compare the before to the after, in other words, leaving out the middle.
Looking at a close-up detail of Graph #2, I picked "arbitrary" but "best guess" start- and end-points for "two data sets running together":
 |
| Graph #5: Selecting (by eye) the Start and End Dates for the "running together" period. The dots indicate Q1 1976 and Q1 1982. (The lines are indexed to 1977.) |
 |
| Graph #6: Average Difference, RGDP Growth less Employment Growth, by Period |
The late period shows GDP, on average, growing about one and one third percentage points faster than employment. This is the period for which Jacob Assa shows his Figure 2 (my Graph #1 in this post) and says:
Recent research has suggested that the imputation of productive output for the FIRE sector in the national accounts has distorted GDP ... [It] drives a wedge between output (as measured by GDP) and employment trends, as visible in the phenomenon of ‘jobless growth’ observed after the three most recent recessions in the US (see figure 2).I have moved one closing parenthesis mark from this excerpt. The original reads
The late period on my graph is the period during which "the imputation of productive output for the FIRE sector" has driven a wedge between output and employment, according to Assa. The early and middle periods, I therefore presume, do not show the influence of the imputation of productive output for the FIRE sector.
What Assa's claims mean, as I read it, is that the difference between RGDP growth and employment growth should get bigger as time goes by. The difference in recent years should be bigger than the difference in more remote years. And the difference in the middle period should be larger than in the early years, but less than in the late years. These expectations arising from Assa's argument are not fulfilled. I find this most disappointing, as I really do like Assa's story.
8 comments:
Is there a way to get "FGDP" numbers, e.g. on fred? (presumably, by subtracting something from GDP ... I just am not sure what the something would be.)
I think there's a way to calculate the correlation between two datasets -- I bet that the FGDP will correlate better with employment than GDP does. I can see if I can figure that out.
(Of course, even if it correlates better, that might not mean anything.)
I still think that it might make more sense to compare it to something other than employment. Maybe it depends on what we're trying to do? But I think there must be better measures of productivity -- number of cars produced, or family income (some kind of quality-of-life measure), I don't know.
I would guess that if you had the data for the last 200 years you would find that almost always productivity has grown faster than the growth in the number of workers. The data for 1970's looks to me like an anomaly, probably caused by faster worker growth due to women and baby boomers entering the market and a slow down in productivity due to rapidly rising costs of natural resources.
For the period up to the 1970's i would guess that the productivity gains over employment came mostly from Manufacturing and Agriculture sectors.
Since 1980 with Manufacturing and Agriculture's share of GDP in decline the question arises - where are the productivity gains after 1980 coming from?
Jerry, Assa's data for the years 1977-2011 is shown on his blog. He says the supporting "methodology and rationale" can be found in
http://www.economicpolicyresearch.org/econ/2016/NSSR_WP_102016.pdf
but I don't remember looking at it.
2. Correlation of two datasets ... I think ... can be done with Excel using Data: Data Analysis: Regression. I touched on it back in April, gingerly. I am too long out of school to have confidence in my abilities with this stuff.
3. Okun's law relates employment (or, unemployment, I guess) with RGDP. Productivity is figured by grunt-level logic: total output divided by total hours worked.
Okay... My assumption was that the source of productivity is always the same. That does seem pretty dumb, now that you point it out. I do remember reading somewhere about manufacturing and agriculture as the source of productivity... maybe Dietrich Vollrath's Growth Econ blog.
So maybe the RGDP-to-Employment gap is growing more slowly now because mfg and ag make smaller contributions to the total, as you say, and Jacob Assa's "financialization of GDP" has increased the output number somewhat, but not enough to make up the difference.
Okay, I can see that possibility. But I don't see a way to convince myself this is the correct interpretation.
Jim -- I forgot! Look at Assa's 2015 PDF, page 7. It's full of graphs comparing "share of value added" to "share of employment" for agriculture, mining, manufacturing, and other categories of industry. Jacob Assa writes:
"As the charts in Figure 1 show, all sectors except finance and real estate have a close correlation between their shares of total value added and their shares of total employment. Only the latter two sectors display a serious disconnect..."
How this fits with what you said (or with what I said) remains to be seen.
That link is
https://ideas.repec.org/p/new/wpaper/1501.html
I haven't really been able to grasp what Assa is getting at. lets suppose he is correct that GDP has been inflated. So what? How would the world be different if those numbers were not inflated? One consequence he suggests, would be we would not see the numbers for employment and GDP diverge so much. But again I ask - so what?
That said, your critique doesn't tell me much either. Its not very meaningful to average percentages. 2 years of 10% and 0% growth is not the same as 2 years of 5% growth.
But even if your method said something meaningful about the growth of GDP and employment, it doesn't really say much in regard to his main point that GDP {if calculated correctly} should be much lower and has been actually growing much slower than we think.
Jim: "Its not very meaningful to average percentages. 2 years of 10% and 0% growth is not the same as 2 years of 5% growth."
You said something like that maybe 3 or 4 years ago. I think you said the result of 5% annual growth every year is greater than the result of an alternating 10%, 0% repeating pattern of growth. Does that sound right?
I wanted to look into the numbers but by the time I was ready to do it I couldn't find your comment.
//
"I haven't really been able to grasp what Assa is getting at. lets suppose he is correct that GDP has been inflated. So what? How would the world be different if those numbers were not inflated?"
The world would not be different if Assa is correct. But our understanding of the world might be better if Assa is correct and we accept his idea.
Finance is a cost that could be easily reduced by policy, by (a) reducing the ratio of total debt to M1 money, and at the same time (b) increasing M1 just enough to fill the void created by part (a) of this plan.
Jacob Assa wants us to recognize that finance is a cost and is non-productive, and I am 100% in agreement with him on that point.
Post a Comment