On 31 July 2013 the Bureau of Economic Analysis of the USA released the initial results of the 14th comprehensive revision of the National Income and Product Accounts.

Many of the revisions were marginal. For example, the average annual GDP growth rate for 1929-2012 was 3.3 per cent, which was just 0.1 percentage point higher than in previous published estimates. Similarly, the average annual increase in the price index for gross domestic purchases for the period 1929-2012 was lowered from 3 per cent to 2.9 percent.

However, the estimates for personal income, disposable personal income and personal saving have undergone huge revisions. These revisions are mainly the result of using an "accrual approach for measuring defined benefit pension plans".

Under the new system, the sum of employers' actual and imputed contributions is the accrual-basis measure of the compensation income that employees receive from their participation in defined benefit pension plans. According to the BEA, "accrual accounting is preferred over cash accounting for compiling national accounts because it aligns production with the incomes earned from that production and records both in the same period; cash accounting, on the other hand, reflects incomes when paid, regardless of when they were earned". If you don't understand this, don't worry.

The Figure below shows the estimates for personal saving in June 2013 and then again in December 2013 for the period from 2001 to 2013. The new estimates are higher than the older estimates for 2001 to 2007, lower for 2008, and higher again from 2009 onwards. For October 2001, the new estimate is nearly 200% higher than the old one. For November 2001 it is nearly 100% higher. For April 2005 it is again nearly 100% higher.

When a mere accounting change results in such gargantuan revisions it is of course necessary to take a closer look.

On juxtaposing the old and new estimates for personal saving with the S&P 500 for the same period, as in the figure below, some patterns emerge. It is only during 2008 that the old estimates are higher than the new estimates. This also happens to be a period when the S&P 500 was falling. What this suggests is that much of the change is the result of the stocks being held by pension funds. It is true that in 2001 and 2002 when the S&P 500 was falling the new estimates were higher than the old ones, but again this can be explained by the fact that pension funds also held mortgage and mortgage-related bonds that were rising during the period.

During periods when the S&P 500 or the real estate market was rising rapidly employers needed to make little or no contribution to defined benefit pension funds. The rising value of the funds' assets accounted for the employers's "imputed contribution". The BEA's error is of course in adding the rise in value of DB pension fund assets to employees' income, disposable income, and personal saving, although the increase in no way adds to employees' current income and of course it adds nothing to employees' disposable income because they never get to lay their hands on it. What the BEA likes to call higher personal income is contributed neither by production during the period nor by employers but is simply a reflection of higher markets. After 2008 one could say without much exaggeration that it is a direct result of QE.

The BEA justifies its new definition of personal income by saying it is consistent with business accounting. But a commonplace of business accounting is that the difference between the recorded value of available-for-sale securities and their fair market value is added to the equity section (or comprehensive income) of the balance sheet, and not to the current year's net income. The latter is of course what the BEA's change in accounting policy amounts to.

Whatever the faults of the old cash-accounting method it related sensibly to reality unlike the new accrual method.