Output is Normally Determined by Effective Demand
An article came out today by Ricardo Caballero, Emmanuel Farhi and Pierre-Olivier Gourinchas, called On the global ZLB economy, where they talk about how a liquidity trap scenario can spread globally.They see a high demand for safe assets over their supply.
“The growing global shortage of safe assets imparted a strong downward secular trend to world real (safe) interest rates for more than two decades. Capital flows acted as the propagating mechanism by which the asset-scarce regions dragged asset-rich regions’ interest rates down.”
This is a consequence of growing income inequality in the advanced countries, combined with the extreme inequality of China becoming more influential in global economic flows. High concentrated wealth and income lead to a higher demand of safe assets and less consumption demand.
They do not see any change in this situation coming soon. The correcting mechanisms are being blocked.
I wrote yesterday about why I use labor share as the over-riding factor in determining the limit on output according to Keynes. Labor share is the best approximation of current income for consumption from current output.
They say in their article that global output is determined by demand, once demand for financial assets exceeds their supply.
“Given the nominal rigidities, output is aggregate-demand determined as soon as the global demand for financial assets exceeds their supply, at the ZLB.”
From my research, the limit on output is always demand determined by the effective demand limit.
They recognize that part of the re-balancing of the situation is for incomes to come down.
“In our model, once real interest rates cannot play their equilibrium role any longer, global output becomes the active margin: lower global output – by reducing income and therefore asset demand more than asset supply – rebalances global asset markets. In this world, liquidity traps emerge naturally and countries drag each other into them.”
But the the problem is that inequality is not decreasing, so high incomes are increasing relative to lower incomes which fuels more demand for safe assets, which keeps the interest rates low and keeps them from normalizing and re-balancing.
So in the end, the drop in labor share seen since 2000 was the signal that the global liquidity trap was bound to happen. And especially the precipitous drop in labor share after the crisis was an extremely important indicator to warn of a persistent liquidity trap. Economists are only understanding this now. Yet, I saw this and wrote about it over two years ago as I began my research of effective demand. (link) In that post, I predicted…
“If labor share of income does not rise, the Fed rate will stay dead no matter what they try to do.”
So the article on the global ZLB is revealing what I have been saying for a couple of years. But more will be discovered in time about how effective demand operates.
Edward ,
I think there’s a relatively simple change you could make to your model ( or simply add a second version of it ) that would greatly improve its diagnostic utility , but data availability may be a limiting constraint. The change would be to split labor share into two parts : top 5% and bottom 95%. Fazzari et al and Kumhof / Ranciere find that this split captures the essential differences between the high-income , low MPC group and the higher-MPC rest of the working population.
The danger in looking at aggregate labor share is that there could be changes in policy that would more strongly favor labor income vs capital income that could result in higher labor income flowing only to the top 5%. Thus aggregate labor share could rise without the beneficial economic impact that your current model would predict.
Sometime in the past I played around with trying to replicate the data in the papers mentioned above by manipulating the income data in Fred , with some success , as I recall. If I can dig it up I’ll post the method here and maybe you can try plugging it into your model.
Marko,
Your idea is awesome.
I hope you can find a way to get the “95% labor share” data. Then we can run it through the model.
Marko,
At the moment, I take the labor share index and convert it to the limit upon the utilization of labor and capital. The limit would stay the same, but the conversion factor would change a bit. What I would like to see is if the plot line lines up even closer to the limit through all business cycles. That would tell us that the share of income for the 95% really is the determining factor on output.
We can interpret the results once we get the 95% labor share data.
Edward ,
Here’s a page from Fazzari’s site that shows the basics of his work :
https://muddywatermacro.wustl.edu/node/93
I know I played around with trying to duplicate the top graph on that page by fiddling with the Fred data , but haven’t stumbled across it yet – poor organizational skills and all that. If I can’t find it I’ll give it a try again , but it may be a few days before I get back to you.
I looked at Fazzari’s site. The 5% have been increasing their share of income much more. This means that the labor share of the 95% would be continually decreasing over time. The result is that the 95% labor share would not line up with the limit on the utilization of labor and capital. and the article at the site talks about how consumption stayed the same. It was through lower saving rates. But what happens in the end, I suspect, is that current production still responds to overall labor share, as overall labor share sets an equilibrium on profit rates… the bottom line. Profit rates are based on all income, even that paid to the top 5%.
So selecting out the 95% would appear to not be an improvement in fine-tuning the effective demand limit of profit maximization.
But it would good to run through the numbers and see what happens.
EL – there is a very good source of data on this topic. It comes from Social Security. It tracks total labor compensation, it breaks down the data in % of total. The data is available back to 1990. The link:
https://www.socialsecurity.gov/cgi-bin/netcomp.cgi?year=2014
So go crunch some numbers – you might be surprised by the results. The % of total compensation for the top 5% has been very stable the past 15 years:
2000 = 31.0%
2008 = 29.1%
2009 = 28.3%
2014 = 30.0%
The SSA data shows that the top 5% were in their best position at the end of the Clinton years. It shows no significant changes over the past 15 years.
Two possibilities, either the SSA data is wrong, or the chart of data Marko provided is not correct. Me? I would use the SSA data.
BK it doesn’t track “total labor compensation” because it doesn’t track compensation granted in the form of returns on capital in the form of stock options or anything else of that sort. From the first lines of your link: “The national average wage index (AWI) is based on compensation (wages, tips, and the like) subject to Federal income taxes, as reported by employers on Forms W-2.”
Stock options and related compensation overwhelmingly confined and granted to the top 1% and a good part of the top 5% does not show up on a W-2. You can’t honestly just equate “total labor compensation” as measured by W-2 numbers and “total compensation for the top 5%”. Well you CAN. If your ideas of “crunch some numbers” is more like “Captain Crunch cereal after 15 minutes in milk”
Mush in, mush out.
Webb – Did you look at the report? It clearly includes stock compensation. From the 2014 report:
Net compensation components for 2014 Compensation subject to Federal income taxes $6,797,372,805,419.35
Deferred compensation plan
Contributionsa
Distributionsb
+ 255,336,805,992.78
– 2,450,397,767.58
Net compensation 7,050,259,213,644.55
$253B of stock compensation is included in the net compensation #.
No??
Key word is Labor.
Kumhof ( BOE / IMF ) shows similar results as Fazzari :
Both of them , as I recall , do an income / outlay decomposition ( resolving “imputations” and such ) to arrive at more real-world measures of cash incomes and spending. I don’t think they counted capital gains ( which would only have made the numbers look even worse. )
Webb – No reply?
Add up all of the fields for income. The sum comes to the exact penny of the 7,050,259,213,644.55. The data does include stock compensation.
Could it be that you’re wrong? Mush!
The discrepancy is due to the fact that W-2 income ( wages and salaries ) doesn’t come close to capturing all income , showing an aggregate total of only ~ $ 7 trillion for 2014. Personal consumption expenditures alone were almost $ 12 trillion in 2014.
Fig 18 in this paper calculates a comprehensive labor share , total and minus the top 10% , 1% , and .1% :
http://utip.gov.utexas.edu/papers/UTIP%2066.pdf
It shows a similar decline post-1980 for the bottom 90 and 99% as reflected in Fazzari’s graph for the bottom 95%.
Marko – There are many components of PCE that are paid for with income not included in W2/1099/stock compensation. A few examples:
-SS benefits and other retirement payments. Medicare, medicaid, CHIPS and food stamps are also included. Medical expenses covered by insurance.
-PCE includes the value of a car sold, but a significant amount of this is financed. Many other components of PCE are financed.
-Education costs are part of PCE, much of this comes from savings or student debt.
-PCE is supported by investment income and draw downs of savings.
-PCE includes expenditures from the ‘black’ economy. (think waiters, bartenders and cab drivers who live off un-taxed tips)
Bottom line – there are many reason why PCE is greater than the sum of all reported wages.
The link I provided is from Social Security. The data comes from Treasury and represents the totals compiled by the IRS. Do you doubt this data source?
And Webb – I’m still waiting for you……….