January 03, 2004

Note: Statistical Discrepancies

In the second quarter of 1994 some $39.7 billion more worth of goods and services were sold to customers than were earned as income--an amount equal to 2.4% of the then-ongoing flow of national income. This cannot be, of course: the National Income and Product Accounts are set up to enforce the identity that the money paid for everything sold becomes somebody's income, that income is equal to expenditure. But the NIPA numbers are estimates, and fuzzy estimates at that--this 2.4% of GDP in 1994:II was thus a statistical discrepancy (and a very unusually large number for this discrepancy.

By the first quarter of 2000 this statistical discrepancy had swung around to be -$42.9 billion: incomes were some $42.9 billion--some 2.0% of national income--higher that quarter than the value of goods and services sold. This large swing in the statistical discrepancy matters for our picture of growth in the late 1990s: according to income measures, the U.S. economic growth rate from 1994:II to 2000:I was some 0.8% per year higher than according to output or expenditure measures.

Since 2000, the statistical discrepancy has swung back. This latest quarter--the third quarter of 2003--the statistical discrepancy was $13.5 billion, or 0.6% of national income: some $13.5 billion more of goods and services were produced and sold than should have been given how much accrued in incomes. This means that over the past three and a half years, economic growth as measured by incomes has been at a rate 0.7% per year slower than economic growth as measured by product and expenditure.

Our standard measures of productivity are based on the "product" side of the NIPA. According to our standard measures, American nonfarm business labor productivity growth averaged 2.1% per year from 1994:II to 2000:I during the boom of the late 1990s, and then accelerated to a remarkable and astonishing rate of 4.0% per year from 2000:I to 2003:III--in spite of the fact that a century and a half of business cycle experience had taught us that times of recession and slack demand are times of slow, not rapid, productivity growth.

But suppose the "income" side is a more accurate measure than the "product" side. If we add the statistical discrepancy in and calculate adjusted productivity numbers, they show an average growth rate of 2.9% per year from 1994:II to 2000:I during the boom of the late 1990s and then 3.3% per year from 2000:I to 2003:III. Productivity growth doesn't slow down during the recession, but we are not left searching for a cause of an extraordinary productivity boom just at the time when firms are cutting back shifts and making less intensive use of capital either.

The problem is that I'm not qualified to say whether the "income" and the "product" side measures are superior. It does, however, seem to me that the income-side measures give a picture of American economic history over the last decade that is more sensible and less weird.

Posted by DeLong at January 3, 2004 05:13 PM | TrackBack

Comments

"In the second quarter of 1994 some $39.7 billion more worth of goods and services were sold to customers than were earned as income--an amount equal to 2.4% of the then-ongoing flow of national income. This cannot be, of course ..."

Why not? Maybe part of it has been purchased by debt, and part by drawing on savings?

Posted by: cm on January 3, 2004 05:58 PM

____

"Why not? Maybe part of it has been purchased by debt, and part by drawing on savings?"

Oops, presumably the income of the seller would be included in the income figure? Not sure, though ...

Posted by: cm on January 3, 2004 06:00 PM

____

I should have been less trigger-happy, but seriously, can some accounting effect be involved? Often goods and services are delivered with pretty long payment terms, up to a quarter or more, or with installment payments, and the income is recognized whenever it arrives (not sure when the sale is recorded).

By and large most of this should level out -- if I perform a service for you each quarter that you pay in the next quarter, then there are abberations only in the first and last quarters. 2% looks quite large in this regard.

Posted by: cm on January 3, 2004 06:09 PM

____

If I work at my uncle's resturaunt and get paid in cash wouldn't that drive up consumption, but not income?

Of course, there is no obvious reason for black market economic activity to be on the rise in the current tax climate.

Posted by: Jason on January 3, 2004 06:18 PM

____

Jason: I made the same interpretation mistake initially -- this is about national income, so when you purchase the good, the seller will report both a sale and income. For each individual seller, the sales and income numbers better match up. In the larger picture, your cash gets off your uncle's balance sheet somehow (or maybe he pays out of his pocket, in which case his bank account will show a lower balance than otherwise). In the end all of the numbers should somehow match up. All (well, by far the most) of the money in the economy comes from the Federal Reserve, and most transactions are recorded. Even if there is a shadow economy (and there always is), for most transaction pairs (sale/income, purchase/expense) either both parts are recorded or none. If not, it may become obvious that illegal things are going on.

So things should still more or less match up.

Posted by: cm on January 3, 2004 06:30 PM

____

"All (well, by far the most) of the money in the economy comes from the Federal Reserve, "

No, it doesn't. The Fed can't create money.

Posted by: Chuck Nolan on January 4, 2004 06:38 AM

____

"the income-side measures give a picture of American economic history over the last decade that is more sensible and less weird."
Indeed. Does the variation in the output data show some correlation with trade deficit data?

"Of course, there is no obvious reason for black market economic activity to be on the rise in the current tax climate."
This is about as wrong as it gets. How about the payroll tax? Payroll taxes are what created the German shadow economy. Itīs running pretty strong. In fact, itīs the payroll tax that Bush should have cut.

Posted by: Joerg Wenck on January 4, 2004 01:56 PM

____

Chuck: Please let me know where the money is coming from. My understanding is that money enters the economy through the Fed's puchase of Treasury notes, and through loans extended by banks which are matched in some or their Fed accounts. So the Fed may not _technically_ create all the money, but it does effectively, does it not?

Posted by: cm on January 4, 2004 02:51 PM

____

I'm not sure that I completely agree with your statement that both cyclically and secularly that low productivity is associated with a weak economy.

On a secular basis you are right.
from 1961 to 1974 productivity growth was 68%
of Real GDP growth. From 1975 to 1994 it was 55%
of Rgdp growth and since 1995 it has been 94%
of RGDP growth. On a secular basis I would argue that the causal relationship is the other way, that productivity sets a speed limit for the economy.

But on a cyclical basis productivity leads growth -- it leads RGDP by about 2 quarters.
But the biggest productivity gains occur in
a recession and early recovery stage of the cycle.
Cycically, the strongest productivity gains occur when the economy has substantial excess capacity
and is in the process of starting to use that excess capacity again. The weakest productivity occurs late in the cycle when growth is also slowing and the economy has little or no excess capacity. This early cycle swing in productivity explains the cyclical swing in earnings (profits).
The single greatest determinate of profits is the spread between unit labor costs and prices. Early in the cycle strong productivty leads to Unit labor costs increases falling below price increases and is the process by which firms rebuild margins and profits. late in the cycle this process reverses.

Posted by: spencer on January 5, 2004 06:21 AM

____

Interesting question. It deserves some serious number crunching to answer.

Posted by: Stirling Newberry on January 5, 2004 12:18 PM

____

Post a comment
















__