February 09, 2004
Why Oh Why Are We Ruled by These Idiots? (Special Troika Forecast Edition)
White House Forecasts 2.6 Million New Jobs (washingtonpost.com): In the annual Economic Report of the President, the White House said the number of workers on U.S. non-farm payrolls was likely to rise to an average of 132.7 million this year from a 2003 average it thought would come in at 130.1 million.
Has Reuters completely misread the Troika forecast--the forecast by the Treasury Department, the Office of Management and Budget, and the Council of Economic Advisers? Or has the Troika completely lost its collective mind? Or have the High Politicians of the Bush administration reached down (as I believe they did last year) and imposed phony job growth forecasts on the Troika?
UPDATE: Sources believed to be reliable point out that the forecast was frozen as of last December 2, and that I have made additional errors in my calculations below. The effect is to lower the administration's expected employment growth rate to 320,000 per month (and the shortfall in job growth since October lowers the target to 131.9 million).
At the moment we stand at a January 2004 estimate of 130.115 million Americans on nonfarm payrolls. The 2004 average is just that: an average, an average of the January 2004 number and the February number and the March number and so on up to the December number. That means that--if you believe that employment is going to grow steadily over the year--the only way to reach an average this year of 132.7 million payroll jobs from the administration's starting point two moths ago is if come December 2004 you hit
135.3 134.5 [assuming steady job growth since last October when the forecast was frozen] million payroll jobs. The administration's forecast of payroll job growth from December 2003 to December 2004 was thus 5.3 3.8 million.
The U.S. economy has never added 5.3 million jobs in any twelve-month period. The closest it ever came was between July 1940 and July 1941, as armament for World War II began and employment grew by 5.2 million. Otherwise--unless I've made some stupid math error--3.9 million from December 1993 to December 1994 was the best we did in the 1990s, 4.5 million from June 1983 and June 1984.
132.7 131.9 (adjusted for employment shortfalls to date) million payroll jobs this year, the administration Troika has implicitly forecast that during the rest of 2004 the U.S. economy will gain 470,000 320,000 payroll jobs a month. The U.S. economy has only gained 200,000 payroll jobs since last June.
Informed sources defend the forecast 320,000 a month payroll job gains as reasonable: "slightly stronger (in percentage terms) than the Bush-Clinton recovery and much lower than the Reagan recovery." I'm not convinced. A 4% forecast annual rate of growth of real GDP, 3% growth in employment, 0.5% (say) growth in hours, and 0.5% rate of growth of productivity? In an economy where labor productivity growth has averaged 3.4% for the past eight years. This forecast is not credible.
But, given this administration, why is that surprising?
UPDATE: Yep. There it is. Table 3-1. Page 98. Calendar year average payroll employment level: 132.7 million.
Posted by DeLong at February 9, 2004 11:04 AM
| | Other weblogs commenting on this post
Speaking of which.. how's the search for Bush supporters going?
I think it is an error, because I have heard other reports say the expectations is to add 2.6 million jobs. That would be 132.7-130.1 (whatever the bad math is). Even if the number is only 2.6 million I am not sure they will reach it. That is a very optomistic forecast. No wonder the deficit keeps overshooting the estimates.
Hey, I don't know how to get a hold of you other than the comments, but I wanted to let you know that some time in the last week to ten days your rss feed has gone from including excerpts to having only the title.
Thought you should know...
The RSS feed is corrupted. You can see where by using an RSS validator such as http://feedvalidator.org/. Editing the posts flagged as having the problems to remove the causes should fix the feed.
President Proposes to Make Tax Benefits of Health Savings Accounts More Lucrative for Higher-Income Individuals - 2/9/04
The President's budget proposal to permit individuals with Health Savings Accounts to deduct their premium costs for health insurance purchased in the individual market would do little to help low- and moderate-income uninsured families obtain health insurance, would primarily benefit the higher-income individuals who can already use HSAs as a lucrative tax shelter, and would further weaken comprehensive employer-based coverage.
The president's economics staff, the Council of Economic Advisers, projected that the plan would result in the creation of 5.5 million jobs by the end of 2004—306,000 new jobs each month, starting in July 2003. Although jobs increased by 112,000 in the month of January 2004, the "Jobs and Growth Plan" still fell 194,000 jobs short of the administration's projection. (Actually, job gains in the month of January were less than 30,000 once seasonal adjustment problems for retail hiring and strike effects are removed.) The administration projected that a total of 2,142,000 jobs would be created in the first seven months after the tax cuts took effect. In fact, only 296,000 jobs were created over that period for a cumulative shortfall of 1,846,000 jobs.
Questions on ERP from a non-economist:
1. Is it true that a rise in temp jobs are a leading indicator of full-time employment? "Statistical analysis suggests that an increase of one job in temporary-help services corresponds to a subsequent rise of seven jobs in overall employment." (p.94, emphasis added)
2. Is it customary to include parables about apples in the EPR? (Legal incidence [of taxation] is unimportant, p. 106).
3. Do we need the Council of Economic Advisers to tell us How Markets Work (ch. 7)?
In 2003, real (inflation-adjusted) weekly wages fell for low- and middle-wage men and were stagnant or fell slightly for low- and middle-wage women. This trend is in sharp contrast to the significant and sustained real wage growth over the 1995-2002 period when unemployment was falling (see chart below). Despite the acceleration of gross domestic product (GDP) growth in late 2003, the wage growth of production, non-supervisory workers (over 80% of the workforce) actually slowed in this period.
It may be a case of subjective wish overriding objectivity. They cannot be seen to be endorsing a prediction that results in net job loss in the first term. It is commonly perceived that we have lost about 2.5 million under Bush, and the 2.6 is a nice number that would make things come out about even at election time.
Smoke and mirrors. They forecast that productivity will return to its more historic level of 2.1%. That is how they arrive at all the new jobs. Whether or not you trust the forecast in large part depends on whether you can swallow a huge drop in current rate of productivity increase.
Reuters butchers the numbers by subtracting averages, etc. The news you link does not accurately report the contents of the report.
The unadulterated numbers:
Mark -- we probably do not know what an increase in temporary workers means. There have been significant structural changes in the use of temporary workers over the past decade so it is virtually impossible to determine if a trend is cyclical or structural.
Until about a decade ago temp workers were relatively rare in the overall economy and
were virtually never used in manufacturing.
But over the past decade firms have been using temp workers on a semi-permanent basis--
especially in high tech. So historic experience prior to this cycle is unlikely to provide much of a guide as to what to expect this cycle.
Can you help a non-economist understand something in the ERP? On p. 94 the report says:
"if the labor force is growing at the same rate as the population (about 1 percent per year), employment would have to rise 110,000 a month just to keep the unemployment rate stable, and larger job gains would be necessary (and are expected) to induce a downward trend in the unemployment rate."
But I look at the current labor force data in the Household Survey and it seems that employment would need to grow by 122k/month.
Also, would I compare this employment increase to the change in unemployment in the Household Survey (down 74k in January), or the non-farm employment from the Establishment Survey (up 112k)?
And in a related vein, I've heard estimates all over the place about the number of jobs we need to create just to keep up with new entrants into the workforce. I've heard anywhere between 150k and 200k, which is a bit higher than what the ERP appears to be suggesting. Opinions as to the "right" number?
I'd like to understand so I can compare apples and oranges. Thanks.
Todd -- unemployment is derived from the household survey so you want to use that data.
The 122 K number you are quoting is the latest months data. But labor force employment growth has been weak over the past year so that is sort of a low-ball number. Labor force growth is cyclical. In good times more people enter the labor force and in bad times people drop out of the labor force. A more normal increase in the labor force would be more on the level of 150.
So you cqn look at the employment growth of under 150,000 -- what trend analysis would imply --
as a bearish indicator that the economy and employment prospects are sub-par.
Speaking of temp workers, last summer the Fed examined the impact of just-in-time employment on the joblessness of the recovery. PDF available at:
>>I think it is an error, because I have heard other reports say the expectations is to add 2.6 million jobs.<<
That's reporters who aren't clear on the distinction between year-over-year and December-to-December changes...
I just looked at the data on temp employment.
The data only goes back to 1982, so you only really have one full cycle -- the 1990 recession recovery to look at.
From 1982 to 1990 temp employment rose from 0.5%
of total employment to about 3.2% of employment.
Enough to support my earlier claim that there
had been secular trend change in use of temp employees.
I ran a couple of regressions. If you regress total employment lagged six months against temp employment you get the type of great regression
results you cited -- The R sq is 95 and the stats are good.
But in economics we have a major problem of multicollinearity, the tendency of all the economic variable to move in lock step. this creates all kinds of problems of demonstrating
One of the ways to test against this problem
is to look at first differences, or growth rates.
If you regress the change in employment lagged 6 months against the change in temp employment
ove the same period you get an r sq of about 13--
that implies the tight causal relationship impled by the first regression may be spurious.
If you just glance at the charts of the two series it looks like their may be a leading relationship. But when you look at turning points in the series there is no consistent pattern of
temp employment leading total employment at turning points in the series. It looks like temp employment leads actual employment at about 50% of turning points -- . I conclude that there is something of a leading relationship, but not one I would bet much money on.
I develop and sell leading indicators of stock market industries relative performance to professional money managers for a living, and the relationship is just not that good.
I suspect the authors you are quoting are ignoring some problems to reach a conclusion that supports the case they are trying to make.
The largest recent figure for a 12 month period is 4,886 million in the 12 months to August 1984. http://data.bls.gov/servlet/SurveyOutputServlet
At that time the labor force was 94.5 million so this represents a 5.17 % increase.
An increase of 5.3 million on a 130 million labor force is a 4 % increase.
So hardly an unprecedented increase.
In fact looking at post war figures the largest is whooping 10% change in Feb 1950 and a 10.9% change in 1947. Beyond that other years in which there are one or more 12 month periods when the payroll total has increased 5% or more are
What’s odd about these figures? Well that there was no 12 month period in the 90’s when non farm employment grew by more than. Perhaps the 90’s were the exception and we’re now returning to a norm.
Bush promises 5.3 million jobs.
Last time that happened was during global warfare.
On MTP, Bush says that he's a "war president".
Now, let me try to connect the dots....
Remember, you can't spell WWIII without dubya.
(I can just picture Rummy going on about
"precious bodily fluids")
In fact what was even more excpetional about the 90's was that there wasnt even a month when the figure was more than 4%. A wierd wierd decade hey?
"What’s odd about these figures? Well that there was no 12 month period in the 90’s when non farm employment grew by more than. Perhaps the 90’s were the exception and we’re now returning to a norm."
It's hard to see how we can return to a "norm" of 5%+ annual job growth if the labor force is only growing at about a 1% annual rate. We'd run out of unemployed warm bodies pretty damn quick, even given the current slack in the labor market.
Anyway, by my math, a 5.3 million December-to-December increase in payrolls would require monthly job creation to average about 470,000 over the remainder of the year -- which is roughly ten times the average gain seen in the last half of 2003.
I'd be curious to know if U.S. job growth has ever ACCELERATED at such a rate before.
I too thought at first that it meant end the year at 132.7, but after going to the source (ERP), it clearly says that it will average that number.
Personally, I'd be pleased if the year ended at 132.7, but I doubt it.
Spencer - thanks! I had a hunch, but a groundling like me likes confirmation.
"I too thought at first that it meant end the year at 132.7, but after going to the source (ERP), it clearly says that it will average that number."
But dollars to donuts if they hit 132.7, they'll claim they met their moving goalposts...
How are soldiers and the Guard calculated? If they return and seek discharge, how will that impact the labor pool?
Stealing from the text of the ERP, I find this:
"The unemployment rate increased in the first half of 2003, reaching a peak of 6.3 percent in June, before falling during the second half of the year. In the fourth quarter, the unemployment rate averaged 5.9 percent, the same as it had been a year earlier. Because the labor force is constantly expanding, employment must be growing moderately just to keep the unemployment rate steady. For example, if the labor force is growing at the same rate as the population (about 1 percent per year), employment would have to rise 110,000 a month just to keep the unemployment rate stable, and larger job gains would be necessary (and are expected) to induce a
downward trend in the unemployment rate.
Looking ahead, temporary-help services employment—a leading indicator for the labor market—suggests substantial further employment growth.
Average growth in temporary-help services unmployment over a six-month period has a striking positive correlation with growth in overall employment over the subsequent six months (Chart 3-3). Statistical analysis suggests that an increase of one job in temporary-help services corresponds to a subsequent rise of seven jobs in overall employment. Employment in temporary-help services has expanded 194,000 since last April, suggesting robust growth in overall employment this year. The unemployment rate is projected to fall to 5.5 percent by the fourth quarter of 2004.
OK, a fall from 5.9% to 5.5% is, on 130 MM workers, about 500,000 new jobs, or 40,000 per month. Add to that the 110,000 they claim is needed to accomodate population growth, and we get 150,000 new jobs per month, which totals 1.8 MM by year end.
Since they mention roughly 200,000 temp jobs and a 7-1 ratio, that would suggest 1.4 MM new jobs.
Neither estimate is anywhere near the 5.5 MM increase by year end that would lead to an annual average of 132.7 jobs. Shouldn't the different pieces of the report come together to form a unified crescendo?
It just says Level, Calendar Year ... Where does it say "average" ... Looking at Table 3.1, page 98, last column of table ... ???
Can one really expect net employment in the US to grow anyplace except in those jobs that require human intervention and no exceptional talent?
Perhaps my personal experience, age and consulting level for the last 15 years or so bias my view, but technology is rapidly catching up with opportunity.
China and India have 2 billion people who are biologiaclly equivalent to the 300 million in the US. Given the distribution of various talents, these societies have roughly 7 times the "raw" talent of the US, talent that is just beginning to become available. They also have 7 times the human energy available for jobs that do not require exceptional talent in a world where technology may place a premium on a relative few exceptional talents at the same time as it eliminates the need for a much larger multiple. All this happens in a very short period of time in a world suddenly interconnected by the magic with which we communicate.
Is it possible we are seeing the economic equivalent of global warming, where the famous "tipping point" is reached and wrenching change begins to cascade? Is the US the global "economic canary"?
This is what Mankiw told the House Budget Committee. He does not give a productivity # for 04 but thinks the average should be 2.1%.
The Administration expects nonfarm labor productivity to grow at a 2.1 percent average annual pace over the forecast period, virtually the same as that recorded during the 43 years since the business-cycle peak in 1960. The projection is notably more conservative than the 4.4 percent average annual rate of productivity growth since the output peak in the fourth quarter of 2000. After such an extraordinary surge, a period of more typical productivity growth is likely as firms shed their hesitancy to hire. In addition, the slower pace of productivity assumed in the forecast reflects the Administration’s view that in the absence of a good explanation for the recent acceleration, it is prudent to base the productivity forecast on longer-term averages.
In addition to productivity, growth of the labor force is projected to contribute 1.0 percentage point per year to growth of potential output on average through 2009.
The Administration expects real GDP to increase 4.0 percent during the four quarters of 2004. This projection is close to that in the latest Blue Chip consensus economic forecast (as of January 10, 2004). This compares with GDP growth of 4.3 percent during the four quarters of 2003 and 2.8 percent during the four quarters of 2002. Measured on a calendar year-over-year basis, the Administration projects GDP growth of 4.4 percent in 2004, compared with 3.1 percent in 2003 and 2.2 percent in 2002.
Does Reuters have its own wonks go at the report or are they cribbing from a one page? If there is a one page, it would be interesting to see the numbers it uses.
Here's my entry on this issue
A dog whistle seems to have gone up that the Economic Report Of The President has made absurdly wild projections regarding the forecasted changes in the non farm payroll numbers. Calpundit comes running:
"Brad DeLong, demonstrating once again why he's an economist and I'm not, explains that the latest job growth figures from the White House are actually more peculiar than I suggested this morning.
In particular, the White House is projecting that non-farm payrolls will rise from 130.1 million at the end of 2003 to an average of 132.7 million for 2004. To me (and to Reuters) that sounded like an increase of 2.6 million, but Brad notes that to get an average of 132.7 million jobs we need to end the year at about 135.3 million This is, to put it mildly, extremely unlikely?"
Ok, let's use Brad's figures and periods (from 1940-date) for now. An increase of 5.3 million in the payroll figures from 130.1 million is a 4 % increase.
How likely is this? Well suppose you not only know nothing about economics but also that you don't even know what year it is. In the period from 1940 to the present there have been 159 12 month periods when the payroll has increased by more than 4% out of a total of 769 (overlapping) periods. So that means just taking a wild guess, which lets face it, is all anyone in Bush "troika" is capable of, you'd have a 20% chance of getting at least the 4% growth required.
But we're not guessing that wildly. We know that in the previous 2 years there have been net falls in the payroll numbers.
Since 1940 there have been 9 periods when the payroll numbers have fallen. In 5 out of 9 times payroll growth has exceeded 4% within the next 18 months. If we lower the cutoff to 3.7%, then the answer is 8 out of 9 times - the exception being the 90's upswing when payroll growth didn't exceed 3.7% in the subsequent pick up - at all.
So from a purely probabilistic perspective the projection is, to put it mildly, extremely likely to be met. Unless you'd only lived in the 90's.
The second point is whether the linear average Brad uses to calculate the number of jobs required to increase the average by 2.6 million overstates the required increase. I think it does. The fall in payroll numbers is likely to be front loaded over the next year as jobs are created on the back of recent strong growth. So rather than having constant growth of 466,000 per month over the next year (5.3 million in total), its more likely that we'll get a convex trend say - 4 months where 500 thousand jobs are added, followed by 2 of 400,000 , 2 of 300,000 then tailing off to 4 of 100,000. These numbers would increase the yearly average by 2.6 million but would only require the addition of 3.8 million jobs - considerably less than the 5.3 million we got with the linear trend. This represents only a 2.9% increase in payroll numbers and over the last 50 years pay roll numbers have increased by more than 2.9% more often than they haven't.
So the only thing that's "extremely unlikely" is that "the integrity of the administration's staff economic forecast pattern is gone."
NB figures from here ? look for 12 Months Percent Change non farm pay roll seasonally adjusted.
the data links
look for 12 Months Percent Change non farm pay roll seasonally adjusted.
"So from a purely probabilistic perspective the projection is, to put it mildly, extremely likely to be met."
Are you seriously suggesting that one can predict the economy with Monte Carlo?
>>>>The fall in payroll numbers is likely to be front loaded over the next year as jobs are created on the back of recent strong growth. So rather than having constant growth of 466,000 per month over the next year (5.3 million in total), its more likely that we'll get a convex trend say - 4 months where 500 thousand jobs are added, followed by 2 of 400,000 , 2 of 300,000 then tailing off to 4 of 100,000.<<
The calculations underlying the estimates of Table C-1 show a constant number of payroll jobs (320,000) created in each month of 2004...
"over the last 50 years pay roll numbers have increased by more than 2.9% more often than they haven't."
If you double check your math, Giles, I think you'll find that payroll growth has exceeded 2.9% in only 16 of the past 50 years.
"its more likely that we'll get a convex trend say - 4 months where 500 thousand jobs are added, followed by 2 of 400,000"
(4 x 500k ) + (2 x 400k) = 2.8 million jobs. Using December's 130.16 million jobs as a benchmark, that would mean a 2.15% increase over the first six months of 2005. The U.S. economy hasn't managed to do that on ANY six month basis since the first half of 1984 -- when:
A.) Output grew at almost an 8% annualized rate -- twice the 4% growth the CEA is forecasting for next year,
B.) Nonfarm business productivity rose at a 1.41% rate -- only barely more than what the CEA is projecting for next year,
C.) the economy was still coming out of the deep, deep, deep 82-83 recession.
Any way you slice it, the CEA's numbers are fantasy.
THREE CHEERS FOR GOOGLE
In Austria, we have a finance minister whose so-calles "tax reform" is based on the same presumptions (and weird data and vodoo economics) that characterize your Prezident's. His name is Karl-Heinz Grasser. He is also famous for a personal website that officially cost something like 250.000 $ (the whole industry is hilariously amused) und was paid for entirely by a sponsorship grant from the Associations of Industrialists (in other countries, this would be calles corruption).
To-day, entering Google Austria (www.google.at) and typing in the key search word "voellige Inkompetenz" (utter incompetence) and then hitting the "I'm feeling lucky" button, you ended up on the finance minister's personal website. ROFL - at least for a few hours and up to now (4:45 p.m. local time).
Some are wondering if Google might have been hacked.
You wouldn't know anyboda who could do this to Goggle US and Dubya?
Best of luck, anyway.
oops, spelling. Your preview button is there for real, one should use it.
I'm posting, of course, about a "so-called" tax reform and the "voodoo economics" it is based upon, about a grant from the Association of Industrialists (there is only one, not several), and I was looking for "anybody" who could do this to "Google" US. Sorry.
Ain't no disgrace to be poor - but might as well be.
Virtue never stands alone. It is bound to have neighbors.
The world is a beautiful book for those who can read it.
Those whose paths are not the same do not consult one another.
Gratitude is the most exquisite form of courtesy.
Love can damage more than you can heal with drinking.