John McDonnell has committed his party to a version of the ‘golden rule’ for public finances. A government in which he was Chancellor would only borrow to invest. Except when monetary policy was hampered by the zero lower bound.
The reasoning behind the Gold Rule is that borrowing for investment is judged less likely to scare the bond markets because that investment promises to yield returns that ultimately generate tax revenue to pay back the debt issued.
However, put like this, the Golden Rule is somewhat problematic.
Some things that would constitute ‘investment’, like building better facilities for caring for people at the end of their lives, would probably not generate any more tax revenues in the future – even if it would make all of us happier when we reach the point where we can use them.
Other kinds of investment, like in road infrastructure, would have a tenuous connection with future revenues without introducing tolls.
And other things that would not be thought of as ‘investment’, but would instead count as current spending, might well raise tax revenues. [A point that Ethan Ilzetzki at the LSE beat me to on Twitter].
For example, money spent on better teachers might improve the capacities of pupils as they entered the labour market, expanding output, perhaps even raising growth, and for a given tax rate, the government would take more taxes to service the expenditure on salaries. Similarly, spending money on salaries for health professionals on ‘public health’ – health education to change lifestyles so that disease is prevented – may generate very high returns at the current margin. [A smaller long-run health budget for the same outputs].
So, to repeat, there are things that we might define as ‘investment’ that don’t generate future tax returns; and things that we would label current spending that do. Both rather undermine the thinking behind the golden rule.
Moreover, if the pivotal feature of the deficit-inducing policy is that it generates future revenues, then any counter-cyclical fiscal change that lessens the risk of hysteresis, or helps reduce the risk of a deflationary spiral, ought to be as exempt as ‘investment.’
As Simon Wren Lewis writes, there is another issue with the conventional interpretation of the golden rule. There not being tax returns accruing to a public investment good does not mean there is no case for borrowing to finance them. Depending on your political standpoint, there may be perfectly sound ethical reasons to back such borrowing. (For example, spreading the tax burden across all those to benefit from a welfare improving investment means financing most of it now via debt).
But the point of the golden rule was not to deal with ethical or distributional considerations. Rightly or wrongly, it is to try to discipline debt issuance.
How good it is at doing that is moot. Even if there are tax revenues to be harvested in future as a result of borrowing to finance some fiscal action, there remains the issue of ensuring that the revenues that start accruing at that point are not matched with higher current spending. This involves a promise about some future action. Not much different from simply borrowing to finance current spending now, [eg to combat the zero bound], and promising to levy taxes that are higher than current spending in the future [once you have escaped it, and monetary policy can compensate].
longandvariable 
The problem with this golden rule, and also SWL’s improvement, that it mismatches the balance sheet of the State with the collected balance sheets of all the beneficiaries of State investment, those beneficiaries comprising at least all the balance sheets in the jurisdiction.
The dispersed yield on state investments is in all likelihood large enough to fund new taxes. But even that is unnecessary if the yield on the investment is sufficiently larger than state borrowing costs. The state should be satisfied to make such a value creating investments, as the financing of them is a separate issue, including money financing.
John