Philip Hammond promised this week "to renew and expand Britain's infrastructure". This might represent a return to Keynesianism in more ways than one.

The obvious sense is that he'll be using fiscal policy counter-cyclically, supporting the economy during a time of Brexit-induced uncertainty; economists expect real GDP growth of only 0.7 per cent next year, well below long-term averages.

But there's another sense. Keynes also looked forward to "a somewhat comprehensive socialisation of investment." Hammond might push us in this direction.

My chart, taken from ONS and OBR data, puts this into historical context. It compares business investment to public sector investment. You can see that in the early 70s the two were similar. Then during the 1979-97 Tory government public sector investment declined markedly. From the late 90s until 2010 however the gap between public and business investment narrowed, until Osborne's austerity and a pick-up in business investment widened the gap again.

It's possible that the next few years will see a resocialization of investment, to the extent that Hammond raises public sector investment at a time when Brexit uncertainty, as well as secular stagnation, holds down private sector investment.

In truth, though, there's another force which might resocialize investment - relative prices. Infrastructure spending is prone to an element of Baumol's cost disease. Because construction tends to have relatively slow labour productivity growth, its relative cost rises over time. However, to the extent that business investment comprises spending on IT, it benefits from Moore's law and so sees a fall in relative prices. These trends alone would tend to raise the share of public investment in GDP over time, and depress that of business investment: my chart shows ratios of investment to GDP in current prices.

And herein lies a problem. It's widely agreed that better infrastructure would raise productivity: better roads and broadband would make us more efficient. But there's a downside here. A greater share of infrastructure spending in GDP, and smaller share of business investment, would tend to depress productivity growth because of brute maths - because a sector with low productivity growth accounts for a bigger share of GDP.

I like to think that the former effect will prevail - and it probably will initially. Over the very long-run, however, this might be more doubtful.