Asymmetrical Information

While SS benefits aren't actuarily pegged to expected lifespan, the age for collecting benefits is scheduled to increase over time. At least there's some modest recognition of increasing lifespans and a deterioration in the worker:beneficiary numbers.

People who work until 65 for full SS benefits under the present scheme are outnumbered by those who retire at 62 and collect reduced benefits. There is some evidence, however, that average retirement ages are starting to tick up after decades of decline.

Jane Galt:

"It encourages them to retire earlier, because the payouts are set politically, rather than actuarily pegged to expected lifespan, as normal annuities are."

You want to elaborate on this? I was under the impression that the benefit levels were set so that the expected cost to the trust fund is the same regardless of when you retire.

People are retiring early because they are tired of working and can afford to do so not because there is some big financial advantage in doing so. Note planning to retire early gives you a cushion, if benefits are cut you can work a few more years.

Jane Galt:

"But doing so would be a big problem for that government, whichever party controls it; that's more than doubling the payroll tax."

Reference please. What I have seen says that if nothing is done eventually payroll taxes will only cover 70% of promised benefits which implies they would have to be raised by less than 50% to cover all promised benefits not more than doubled.

I was under the impression that there was a subsidy for people retiring early since the actuarial early retirement reduction factors haven't been revised in quite a while but aarp has a different take
http://www.aarp.org/research/socialsecurity/benefits/aresearch-import-331-DD56.html

Whilst you're 'splainin' things, you might add the funding problems of medicare.

Social Security is also not actuarially fair for those who have paid the same amount in. Funny thing I posted a short post about this earlier at greyswan.net

GS

Jane_Galt: Could you please explain what you mean by this:

In point of fact, the problem with social security is not that its unfunded structure somehow means that the future economy "can't afford" to support all those old people via Social Security. The economy doesn't care whether it supports old people through dividend payouts or social security checks.

I've seen a lot of economic illiterates say this, but it's odd coming from you. I can't think of a sense in which it's true. Are you really equating dividends with taxation? Are you really claiming that I should feel the same way about the income tax as I do about my employer's Board declaring a dividend? Are you really claiming that drawing a dividend from an investment is economically the same as taxing someone?

In any case, I don't think it's just the labor force and productivity we need to worry about. First of all, benefits are (implicitly) indexed to productivity. Second, there's the nasty possibility of Atlas shrugging. People like to draw parallels between old industrial pensions and Social Security, so how about this one: Just as GM couldn't force people to work for it and buy its products, and in fact has a harder time doing so because of those nasty "legacy costs", the US government will have a harder time convincing newer workers to perform labor that falls under the US SS system as younger workers' taxes are increasingly diverted to something that doesn't benefit them. Remember, GM's plan was actuarially sound too, under the tenuous assumption they'd never get competition that didn't have their costs.

Thought experiment: tomorrow Congress declares that at any time you can opt out of Social Security. You never have to pay into it again, but you forfeit whatever you've paid in so far. Whatever age group is the "event horizon" for dropping SS is the group that will get the shaft anyway when the system inevitably implodes.

Individuals may be economically illiterate, but as a group the under 40 crowd knows that we're getting stuck with the bill big time.

Thanks mom and dad.

How many people are going to reckon on not saving much because, when they are in their mid-60s, Mom and Dad will have died and left them a tidy sum?

You might also want to consider the impact of immigration policy and climate change response policy on the Social Security issue. What happens to the solvency of the Trust Fund if we back fill with new immigrants all of those job vacancies that will be created by baby boomer retirement and all of those additional job openings in health care created by the needs of those baby boomers? The current Trustee actuarial assumptions seem to assume that the current legal limits of immigration determine how many of those jobs will be filled here and how many will be further pushed off shore. Is that a reasonable assumption?

And what about a response to global warming and climate change? Isn't it likely our current levels of carbon consumption and our current levels of efficiency in using that carbon create the potential for comparative advantage for the United States as we respond? May we not get further with less pain than others?

Person: What Jane was saying is that old people don't eat their pensions or savings, live in them, or take them as medicine. Everything they consume has to be provided by the work of younger people. By the time I reach retirement age, that's going to be a huge piece of our productivity that doesn't go to the benefit of the people actually doing the work. No matter how it's paid for, it seems like younger workers will be losing out.

OTOH, there is one big difference between bonds allegedly owed by one government agency to another and private savings. Private savings get invested in capital items, research, etc., usually to buy things that (the managers hope) will improve productivity. This can result in a much bigger pie to split up in the future. Government spending occasionally has the same effect (infrastructure building and research), but much more of it is squandered on unwise capital investments ("bridges to nowhere"), or handed out to welfare cases (including corporations that ought to die and free up their resources for better run businesses). As for research funding, between the political earmarking of funds, politically pre-ordained results, and all the money poured into finding better ways to kill foreigners, a rather small percentage of government research funds goes into projects that will ever make any civilian's life better.

markm: Your second paragraph was exactly my point: the dividends someone receives is paid *out of* the gain in product created by their contribution. It is not a drain at all; if they didn't exist, you would't have to give them stuff, but that stuff wouldn't exist either.

Government taxes, as you note, are collected regardless of the productivity of the earlier "investment", and typically far exceed it. The individual being taxed is having his earnings diminished, while a dividend is paid strictly out of that which enhanced his earnings by a larger amount.

dearieme: Sadly, I have no numbers to back up this assertion. However, I think we can safely assume that those who receive such fortunate (?) largesse will be outnumbered by those whose parents will outlive their retirement savings.

Let me add one more point to why Social Security isn't likely to right itself with greater than expected economic growth:

Social Security benefits (projected at age 60) are indexed to average wages, not to prices (cost of living adjustments after age 60 are based on prices).

This makes a huge difference, because the most likely source of above average economic growth would be improvements in productivity. That figures to increase wages and therefore benefits in roughly the same proportion.

Telnar,

Wages do not nearly increase at the rate of productivity gains.

Social security is only partially indexed to wages. Once you initiate receiving your benefits, they are indexed to prices, not wages.

Don't overlook the benefit to the younger generation of being relieved of the full responsibility of supporting their retired parents. If they would save the equivalent, they could probably fund their own retirement.

The current system also encourages investment in the same country as the saver. A more private based system would probably have a lot more overseas investments.

You are right of course that gdp growth is roughly equal to labor force growth times productivity growth.

No one seriously questions the SS labor force projection although their immigration forecast are surprisingly low.The labor force projections are driven by the demographic projections that you have correctly described.

However, the major difference in the three projections that SS makes center on their productivity assumptions. Since WW II productivity growth as defined by the SS administration --
and even you will have to admit they have developed an unusual method of calculating productivity-- has averaged 1.9%.

The high cost and intermediate SS projections both assume that long run productivity will be significantly below the historic average at 1.6% and 1.3%, respectively. The 1.3% rate is what the US averaged in the 1975-85 productivity bust.

But if you use the low cost scenario that assumes that future productivity growth will average the same as in the past the results have the SS always in surplus -- or at least thru the next 75 years.

Of course, the projections for real wage growth and almost every other non-demographic projection are influenced strongly by the productivity projection.

So if you want to have a serious discussions of
prospects for SS maybe we ought to start with a debate as to why productivity will be much lower in the future then in the past.

If we are going to discuss the assumptions of the SSA with regards to projections for the future, it is necessary to use the correct numbers for the different scenarios. These can be found in the links provide below:

Link 1

Link 2

In these links the SSA discusses its assumptions. On balance, I find the middle projection, all in all, to be difficult to argue with.

OK Yancey, you caught me. I was quoting data off the top of my head without looking at the latest report.

But I want to thank you for the link.
the latest report provides an explanation of why they project slower productivity growth in the future that they had not provided in the past.

They say that much of the rapid productivity growth of the past 50 years stemmed from a shift of resources out of the low productivity farm sector to the high productivity nonfarm sector.
If they had made that claim for an earlier era, like 1980 to 1925, I would have no trouble with that explanation. but according to the USDA farm productivity since 1948 has averaged 1.76%, or almost exactly the same as the productivity calculation they use for the overall economy.
So I find it hard to explain this apparent conflict. Can you?

Moreover, I would also need to see if the shift of resources from the farm sector tot the nonfarm sector has really been all that big over the lat 50 years. I think I will look for the data to see. do you have it? To be honest this is the first time I have heard anyone use this explanation for US economic growth since I took
courses in US economic history in the 1960s.

Spencer,

The change into manufacturing from farming involved a great change in total output for each worker that made the move, and it does not matter that farm output itself increased in about the same percentage as overall output in period you discussed- an equal percentage increase in lines of work do not necessarily mean equal increases in total value.

However, the most important point in Link 2 is this one:

The annual increase in total productivity averaged 1.6 percent over the last four complete economic cycles (measured from peak to peak), covering the 34-year period from 1966 to 2000. The annual increase in total productivity averaged 2.2, 1.2, 1.2, and 1.6 percent over the economic cycles 1966-73, 1973-78, 1978-89, 1989-2000, respectively.

In other words, over the last 4 economic cycles, productivity growth has averaged 1.6%, which is lower than the intermediate projection. To make the assumption that the low-cost scenario is the correct one would mean that one would have to assume that productivity growth has more or less permanently increased from the rate of 1966-2000. I see absolutely no reason to make such an assumption other than simple wishful thinking.

Yancey -- since 1965 farm employment fell 99,000 while nonfarm employment rose 7,218,000.
This means that from 1965 to 1995 the cumulative shift of labor from farm to nonfarm accounted for 0.1% at a maximum of the total increase in nonfarm employment.

This is what I mean when I said that if you were talking about the prewar era the shift from ag to nonag would have been an important cause of productivity growth . But over the last 50 years the shift out of farm and into nonfarm has not been large enough to make a significant difference no matter what you assume about the absolute level of productivity. If you assume that nonfarm productivity was double that of farm productivity this means that the shift from farm to nonfarm would have accounted for 0.2% of the cumulative productivity gains from 1965 to 2005. This is not enough to significantly impact trend productivity growth. The 0.2% cumulative total divided by 40 years gives a 0.005% annual impact

Spencer,

You will have give cites for those numbers. Just a quick scan of the BLS.gov website found the following numbers for agricultural and total labor force participation for the time frames of Dec 1994 and Dec 2006:

Dec 1994 Agricultural employment=3.5 million
Dec 1994 Total employment=124.6 million

Dec 2006 Agricultural employment=2.3 million
Dec 2006 Total employment=145 million.

I find it difficult to believe your assertion that non-farm employment only increased 7.2 million in those 30 years if the increase in the last twelve was 20 million by itself, and the numbers for the 12 year period of 1994 to 2006 for agricultural employment cast serious doubt on the assertion that agricultural employment only declined by 99,000 when 1994 to 2006 saw a decline of over a million by itself.

Here is the best place to find agricultural and total labor force employed in the years 1948-2007.

In the period of 1965 to 1995, the data show that, as a percentage of total employment, farm employment (agriculture) fell from 6% to 2.7% of total employment in the United States. If the 6% had been maintained, then employment in agriculture would have been 7.5 million workers in 1995 rather than 3.4 million. From the period of 1948 to 2006, agricultural employment fell from 13% of total employment to 1.5%. If the 13% had been maintained, total agricultural employment would be 18.8 million rather than the 2.2 million of today. This is a significant change in the composition of the labor force for both periods. The change in composition going forward will be smaller than the past because farm payrolls can't fall more than another 2.2 million.

Still waiting for your justification for your latent Marxism, Jane.

Yancey -- my source is at bls in the household survey,table t.a5 Employed by class of worker.
I'm citing data on civilian employment 16 years or older in agricultural industries and in nonagricultural industries.

It is the only place at BLS I could find data on agricultural workers. The link you provided only covers nonfarm workers. It provides no data on
agricultural employment. I could not find the data you reference.

You will have to give me a
better link.

Yancey_Ward: The change in composition going forward will be smaller than the past because farm payrolls can't fall more than another 2.2 million.

Sure they can -- what if there are "negative farmers", i.e., people whose job it is to destroy food? :-P

Spencer,

Good grief! The link of my comment at 1:40 p.m. contains all the information I described. I used Table A-1 for the total civilian employment and Table A-5 for agricultural employment, which also includes non-agricultural employment. Chose the options in the sublinks for non-seasonably adjusted which fills the last columns with annual data, and you can select in the buttons above these tables any year or years you wish between 1948 and today.

In addition, they show the numbers you used in your comment at 11:42 a.m. to be completely off by an order of magnitude, at least.

OK, I see my error. I forgot to switch from seasonally adjusted to not seasonally adjusted data so I only pulled wage & salaried workers into my spreadsheet -- leaving out self employed and unpaid family workers.

So the correct data shows ag employment fell from
4.361M in 65 to 2.206 in 2006, a 2.1 M drop.

Meanwhile nonag employment grew from 66.727 m to 142.221 M, a 75.494 gain.

So over 40 years ag employment accounted for 2.8%
of the growth in nonag employment or a 0.07% annual contribution. So it still is not a major factor although my original estimate was much smaller. Ok, off by a factor of ten.

Spencer,

Others can weigh in if they wish, but I think the way you are looking at it is incorrect. To consider the real effect over the course of time, one must look at what farm employment was in 2006 vs what it would have been without change in employment composition. This still doesn't give a huge % labor "transfer", but the SSA wasn't making a big adjustment either, based on this factor.

OK, why don't we look at this in a different way.

The BLS publishes data on total business productivity and on nonfarm business productivity.
The difference to the two should represent the contribution of shifting labor out of ag and into nonag firms on productivity.

Since 1965 total business productivity growth -- including farms --has averaged 2.1% while nonfarm business productivity growth has averaged 2.0%.

So the growth in nonfarm business productivity has been less then total business productivity. This implies that the shift of resources from the farm to the nonfarm sector has dampened or reduced productivity growth by 0.1% annually.

If you move someone with x productivity from the farm to the nonfarm sector where he has 2x productivity it increases productivity by x ( 2x-x).

As you point out once farm employment reaches some low that source of productivity growth can no longer take place.

The reason your suggested approach is not the correct methodology is that you are assuming someone that was never in ag is shifting from ag to nonag.

That is the reason that you must calculate the contribution based on the actual movement, not some theoretical level as you are proposing.

Yancey, you will notice the growth on business and nonfarm business productivity growth that I cited from the BLS are significantly higher then the productivity growth data that is used in the social security report.

That is because SS calculates productivity differently then BLS. BLS, and other official sources do not calculate productivity for government. Government output is assumed to equal the inputs. There are both practical and theoretical reasons for this. Basically it comes down to the point that since government services do not have a market price you can not measure its value. For example, what is the value of the output of the Marine Corps? This means that government is assumed to have zero productivity growth.

What social security does is to include government employees in its estimate of productivity.

So SS takes x government employees with zero
productivity growth with (100-x) private employees
with some level of productivity growth to derive
its estimate of productivity.

For example if you assume business accounts for 80% of employment and government some 20% of employment what ss does is this:

P == (B*(0.8) + G *(0.2))/t

P= social security productivity estimate
B = business productivity as calculated by bls
G = assumed zero productivity in govt.
t = government plus business hours worked

So the social security estimate of productivity is actually more like an estimate of real gdp per employed person then what we normally think of as
productivity.

Spencer,

Oh, where to begin?

As people moved into things like manufacturing from agriculture, their total output increased. What you are doing is trying to say that a 2% increase in agricultural output for a worker is the same value as a 2% increase in output for a manufacturing worker. You are completely ignoring, in your analysis, the change in productivity that occurs when the composition of the work force changes. This is a completely invalid way of looking at it. The manufacturer is inherently more productive to begin with. As the composition changes to non-farm from farm, the total economic output gets a boost as does productivity.

As to the effect of a declining percentage of farm workers to the overall employed numbers, my approach to analyzing the effect is the valid one because it captures the actual magnitude of the change. If agricultural employment had continued as is, then the percentage would have remained the same, and more people would have been in agriculture than actually turned out at the end of the time period. For example, as I pointed out above, if agricultural employment in 1948 had not changed relative to the rest of the economy over time, then we would have 18.8 million employed in agriculture today rather than the 2.2 million we do. In other words, the rest of the economy today has an additional 16.6 million workers in higher value output occupations than it would have without the transition.

On the issue of how SSA measures productivity, I really don't understand your objection. Wage growth is correlated with total economy product growth. Now, unless government employees don't pay FICA or collect benefits (and since exempt state and local employees must still be paid and have retirement funded, they must be included as well), then they must be included in any calculations that attempt to predict future wage growth and benefit payouts. What you seem to be trying to do is to apply the private sector's higher productivity growth to everyone regardless of whether they work in the private sector or not. That type of analysis is just simply fraudulent.

No -- you are assuming that when a person moves from ag to nonag that their productivity increases.

You do not know that. I know it was once true.
But I do not know that it is still true and neither do you.

That is my point that over the last 50 to 75 years that agriculture has had such strong productivity growth that ag productivity apparently is now higher then nonag productivity. So that when someone moves from ag to nonag they may no longer be moving from a low productivity job to a high productivity job.

If you look at the productivity data you see that in 1965 that total business productivity was 58.8 vs 61.7 for nonfarm productivity. In 2006 the total was 137.7 vs 136.7 for nonfarm. This means that in 1965 farm productivity was less then nonfarm productivity but in 2006 farm productivity was greater then nonfarm productivity. Now this may be simply an index number problem-- the indices are set equal to 100 in 1992 but this does not mean that they were equal in 1992.

Normally, the way to check this would be to check wages. If farm wages were less then nonfarm wages it would imply that farm productivity was less then nonfarm productiivty. But as we discovered yesterday, because I made a mistake, almost half of farm labor is self-employed and unpaid family workers that do not receive their compensation in the form of wages.

Spencer,

Well, this will be my last attempt.

Yes, I was giving SSA the benefit of the doubt when they claim agricultural work was inherently less productive than the alternatives for the time periods discussed. As you wrote, this has been traditionally true, and I have seen nothing, when looking at historical BEA.gov statistics (see my last paragraph), to doubt this was still true in the periods we have been discussing.

Where you are making your mistake is here: you are claiming that productivity increases in farm work that match, or exceed, productivity increases in non-farm work mean that a decrease in the relative percentage of workers doing farm work will mean that total productivity increase is equal to or less than it would have been if the percentages had remained unchanged. This is completely invalid analysis since it completely ignores the differing output of the two classes of work. Here is a numerical example:

Farmer Bob in 1975 produced $100 of output/day, and Welder Bill produced $150 of output/day.

In 1985, Farmer Bob produced $120 of output/day, and Welder Bill produced $180/day, or both increased their productivity by 20%, and, as one would expect, total productivity has changed from $250 to $300 per day, or an increase of 20%. However, if Farmer Bob had become a Welder Bob, then total product would have been $360/day in 1985 for an increase of 44%. One can extend this analysis in a general way by incorporating changes in the composition of the work force over time with regards to farmers and welders.

The real question is whether farm work is actually less productive. Using the BEA data (see BEA.gov) for gross output by industry for which I can easily find data for the period of 1987 to 2005 in tabular form, and using the numbers from the BLS for total and agricultural employment for those years, I find that total output per worker in 2005 was $160,869 while the output of agriculture per worker was $115,248. The ratio is even worse in earlier years indicating that productivity in agriculture was improving more rapidly than that of the total labor market which is what you keep pointing to, but that only contradicts SSA's explanation if farm output/worker is already higher in an absolute sense, which data show otherwise for any year since 1987 (and will certainly show for any earlier year as well).

Really, this should end the debate, don't you think?

Yancey -- I know your point about moving from a low productivity job to a high productivity job.
I have been using it all along. You are the one who tried to get away from it by trying to base the comparison on a scenario where you assumed that farm employment was the same in 2006 as in 1965 so you would have imaginary farm workers moving into nonfarm jobs.

did you read my point about wages in ag? the same point is true about comparing output. In nonfarm jobs the standard assumption that everyone works a 40 hour week for 50 weeks to have 2000 hours worked a year is reasonably accurate in almost every industry. But it is not a safe assumption about the farm. But I will let it go unless I find some better info on ag hours. but productivity is based on hours, not employees.
Ag has a massive share of its labor as seasonal that does not work anywhere near 2000 hours a year.

However, you have still provided no explanation of why total business productivity growth has been stronger then nonfarm business productivity for 20 years. Please explain how that can be if farm productivity is much weaker then nonfarm productivity.

I understand what you are saying and in theory I agree completely with you. All I am saying is that the data is not showing what theory says it should.

Why??

Spencer,

Imaginary farm workers? The boost in total productivity that arose from the declining employment in agriculture came not only from people who worked on the farms then moved into higher value work, but also from the fact that the declining percentage of workers in farming over time meant that more workers were freed to be employed in higher output work- even if they never worked on a farm. This is a real increase in overall productivity, not imaginary.

Going back to a previous example, I noted that had agriculture employed the same percentage of people today that were employed in 1948, we would have 18.8 million workers in agriculture and 16.6 million less in the non-farm sector than we actually have today. Just using the 2005 gross output numbers per worker from my last comment gives the following:

Agricultural output today = $253 billion
Total Economy gross output today= $22800 billion

If 1948 %s had continued then

Agricultural output today = $2162 billion
Total Economy output today = $22050 billion

In other words, the effect of the declining employment in agriculture over the last 67 years added at least $750 billion dollars to real output in 2005, and I must point out that I have completely ignored the negative cumulative effects on total output and productivity that would also have occurred over time if more people worked in farming the entire 67 year period. In other words, my example here makes the hypothetical decrease in total output much less than would have actually occurred if farming’s relative importance had remained the same.

As for the seaonality of farm work, it doesn't matter. I used annual employment numbers from the BLS and annual gross output from the BEA.

As for your last question, if I am reading it correctly, I would point out that it appears that product per worker in agriculture itself is growing more rapidly than the non-farm sector. However, this is a really, really unsurprising result since agriculture total product/worker is on a much lower level than the non-farm sector. This is why your focus on rates of change in productivity is leading you astray. The ratio of total output vs farm output grows every year, even though agricultural productivity is growing faster. In other words, agriculture is a smaller and smaller portion of the economy, year after year, which was the essence of the SSA's explanation to begin with.

That should have read "57 year" rather than "67".

I talked to the people who do the productivity numbers at BLS -- they are always glad to be of all the help they can be.

Nonfarm productivity is still about double that
of farm productivity,even after you make all the adjustments I could think of.

Farm productivity growth over the last quarter century has average 5.4% -- more then double
nonfarm productivity growth.

Your explanation of what happens when a low productivity farm workers moves into a high productivity nonfarm job is correct -- and I
never disagreed with that.

We both agree that farm employment output is such a small share of the economy it will not have a significant impact on productivity growth going forward -- what SS says.

Where we seem to be disagreeing is my conclusion that it has not had a significant impact on economy wide productivity over the last quarter century. I say you have to look at the actual data and calculate the changes that way. You say you need to look at how many people would be in ag if they still had the 1965 share of employment.
I think this calculation creates a category of workers that were never in ag -- the difference between the actual and what that number would be if ag still had a higher share of employment. But you are taking these workers who never existed or worked in ag and saying if they moved into nonfarm that would have generated a big increase in aggregate productivity. But that did not happen so it can not be responsible for a significant share of total productivity growth.

How would you react if I said that to calculate going forward assume that in 2050 ag workers accounted for 10% of the labor force?

Maybe I'm wrong in looking at growth rates rather then levels. But you should get the same results if you look at an issue from two different directions. When this happens it always raises a major caution flag that something is wrong somewhere. However, I'm still left with the point that total business productivity growth -- including farms -- is stronger then nonfarm productivity growth. But it is only 0.1% annually. the expert I talked to at BLS could not explain that either and agreed that it should be the other way. So we are left with an anomaly and will continue to disagree over how significant the shift out of farms has been over the last quarter century. But, if it has not been significant over the last quarter century it can not be a reason for slower productivity growth in the future..

Spencer,

You asked why total economy productivity growth is higher than that for non-farm alone. The only case in which this wouldn't be true is if farm productivity growth were a negative value since the total economy is the sum of the contributions of farm and nonfarm, or, rearranged, total economy minus farm = nonfarm. Also, if you do the calculations for the years 1987 and 2005, you find that agricultural output/worker was 59% of the Total Economy output/worker in 1987 and 71% in 2005- so productivity of the workers has actually been growing more rapidly in the farm sector between those two endpoints. And, even with this increased productivity of farm workers between the endpoints of 1987 and 2005, the gap between gross output/worker in farm vs total has grown from $31,142 to $45,621.

We disagree on the other point because you are just looking at it the wrong way. These workers did exist, and it does not matter whether or not they ever worked in agriculture and moved to non-farm work. Each year's new workers (and old workers) had a choice between farm and nonfarm work, and the fact that the percentage of farm workers was falling throughout the time period meant those new workers more and more went into the higher output non-farm sector. Each year, total output/worker was higher because fewer workers were needed for agriculture. These effects are cumulative, year to year, and the way to account for them is to add up the changes from year to year. The wrong way to look at it is to see 7 million agricultural workers in 1948, and see 2.2 million in 2006, and say the net change was 4.8 million from farm to non-farm. I will give you a numerical example, that is related to my previous example of Farmer Bob and Welder Bill, to illustrate:

Farmer Bob in 1975 produced $100 of output/day, and Welder Bill produced $150 of output/day. Total output was $250/day. In 1976, Bob's son Rob and Bill's son Will enter the work force and face two choices, farmer or welder. If the percentage of farmers does not change from 1975 to 1976, then we will get Rob as a farmer and Will as a welder and total output will be $500/day, or an increase of 100%. However, if, in 1976, the percentage of people in farm work falls to 25% of the work force, then Rob and Will both become welders and total output in 1976 will be $550/day, or an increase of 120%. You get this boost from 100% to 120% even though the number of farmers has not changed at all, and neither Rob nor Will has ever been a farm worker.

Yancey -- what you are postulating is what could be a theoretical possibility or an outer limit.
But it has nothing to do with what actually happens at any one point.

your example consist of a two person economy.
in 1975 it is bob and bill
...........farmer....welder....total

1975........100.......150.......250

in 1976 it is rob and will

1976 .........0.00....300.......300

in 1977 it is rob and will

1977..........0.00.....300.......300

You can only count rob shifting from farming to
being welding once. What you are doing is calculating what would happen if rob shifted from farming to welding over and over again and again year after year.

Spencer,

No, I am counting them once to demonstrate the effect on output due to the change in employment ratios. In 1977, the new baseline of total output is $550, and any new workers would be, by definition, new workers, not Rob and Will again. Any additional increases in output will have to come from higher worker productivity and the output of new workers. Again, the new workers would have the option of farming or welding. If farming declines again as a fraction of workers, then more of these new workers will choose welding than would have without the change, and, again, total productivity gets a boost because of the change in relative employment ratios. You will always get a boost in total output/worker when any line of employment with a below average productivity/worker declines as a fraction of total employment. Such a boost from the decline of agriculture in the United States is pretty much at an end because there are now so few such workers, and because the actual productivity of the remaining farm workers is getting closer to that of non-farm workers.

Yancey -- what are we trying to do? Estimating the impact of labor shifting from farm to nonfarm employment

..........farm.........nonfarm
1965.......4.0..........68.9
2005.......2.2.........142.2

difference..-1.8........73.3

From 1965 to 2005 farm employment fell 1.8 million
while nonfarm employment grew 73.3 million.

But out of that 73.3 million increase in nonfarm
employment only 1.8 million was a shift from farm to nonfarm. The other 70.1 million was the natural growth of the labor force that had nothing to do with what we are measuring.

the relevent numbers for what we are trying to calculate, the impact of labor shifting from farm to nonfarm are:

...........farm.........nonfarm...total

1965........4.0.........68.9......72.9
2005........2.2.........70.7......72.9

The 142.2 number does not enter into the calculation of how much labor shifted from
farm to nonfarm employment.

The shift from farm to nonfarm employment was only
1.8 million, not 73.3 million.

The shift from farm to nonfarm employment only accounted for 2.3% of the 73.3 million or 106%
increase in nonfarm employment from 1965 to 2005.

so if farm output is 50 and nonfarm output is 100
the increase in output is the number of farmers
leaving ag (1.8) times their nonag output (100)
minus what their output would be if they
stayed in ag (1.8 * 50).
increased output of farmers shifting to nonfarm:

(1.8*100)- (1.8*50) = 90

it is not the total increase in nonfarm employment (73.3) times their nonfarm output less what their output would have been if they had gone into farming (73.3 * 50)
(73.3*100) - (73.3 * 50) =3665

But this is what you are estimating, and as a consequence you are overestimating the impact of the 1.8 million individuals shifting from farm to nonfarm by a factor of 40 (3665/90).

Comments are Closed.