Phillip Hammond said yesterday that the public finances "continue to improve" "thanks to the difficult decisions we have taken in the last nine years." This is not quite correct.

To see why, imagine that people and companies were to save a constant fraction of their disposable income, come what may, in the face of fiscal austerity. As their incomes fall in the face of that austerity their spending would therefore fall one-for-one. That would further reduce aggregate demand which would in turn cut the government's tax receipts and add to government borrowing. We'd then suffer a severe paradox of thrift, in which everybody's efforts to save more would lead to weaker incomes for all and hence to lower savings.

This tells us that governments can borrow less if and only if somebody else borrows more or saves less. This is in fact trivially true. Every pound borrowed is a pound lent. One person's net borrowing can therefore fall if and only if somebody else's net lending falls.

My chart shows the point for the four main sectors of the economy. It shows that government borrowing has steadily declined since 2009. The counterpart to this, however, has been a sharp swing in household net lending. For example, in the 12 months to September 2010 they were net lenders of 4.8 per cent of GDP: they were saving a lot and borrowing little. In the 12 months to September 2018 (the latest period for which we have data) they were net borrowers of 1.3 per cent of GDP. This swing of 6.1 percentage points is by far the biggest counterpart to the 8.2 percentage points fall in government net borrowing. It is mainly due to a decline in the household savings ratio.

We can therefore tell a different story about why government borrowing has increased. It's because, in response to austerity, households saved less and borrowed more in an attempt to maintain their spending: companies have also borrowed a bit more too. This supported aggregate demand and hence employment and wages - which allowed tax revenues to hold up as spending was squeezed.

In this sense, the government has not cut borrowing. It has only privatized it - shifted it from its own account onto households.

Herein lies the reason why the OBR foresees only a small further reduction in PSNB - from 1.1% of GDP this year to 0.5% by 2023-24. It's because it foresees no great change in households or corporates net borrowing: see chart 3.28 on page 59 of this pdf.

My point here is that the public finances are never fully "under control", despite the claims of many people who should know better. Government borrowing depends upon the actions of the private sector - actions over which policy-makers have only limited control. If the private sector chooses to borrow a lot (as it did at the turn of the century) the government will be a net saver. And if the private sector borrows a lot - as it did after the 2008 crash - government will be a net borrower.

One Big Fact tells us this. It is that official forecasts for government borrowing are usually wrong. Since 2000 the average error (regardless of sign) in forecasts for PSNB made in the spring for the following fiscal year has been 1.1 percentage points of GDP - £24bn in today's money*. Even excluding 2008-09 (when borrowing was 4.4 percentage points of GDP more than forecast) the average error has been 0.9 percentage points of GDP. Such big errors are inconsistent with borrowing being fully under control: if it were, it'd be perfectly forecastable. But they are consistent with borrowing responding to the private sector's decisions to save or borrow.

So, Mr Hammond is wrong. The fall in government borrowing is not due to his government's "difficult decisions" - difficult for other people of course. It is due to a decline in private sector savings.

* Historical forecasts are here. Out-turns are in table PSA5A of this pdf.