I see that my colleague Tony Travers, writing with Simon Jenkins, is calling for greater devolution to local authorities as a way of effectively implementing cuts to public expenditure.
They suggest two principles. The first concerns true delegation of responsibility which they argue leads to experimentation at the margin. What they don't point out is that it also leads to greater competition amongst local authorities, something which economists generally argue is a good thing.
The second principle is that devolution must not disadvantage poor areas. They argue that this can be achieved through redistribution of local revenue. Unfortunately, this is a very narrow interpretation of "must not disadvantage poor areas". If, however, you take a broader definition this leads down a long slippery slope to opposing any differences across places because "postcode lotteries" are unfair.
Focusing instead on poor people helps avoid this problem. My local authority might decide to increase local taxes rather than, say, cut funding to my local art gallery. My love of art and my vast disposable income (this is a hypothetical example) mean I can live with this. In the neighbouring authority poorer households decide they would rather forgo the art and prefer to see their taxes cut.
More generally, when expenditure is devolved the overall pattern more closely represents local preferences. In turn, this closer link between preferences and local expenditure increases the incentives of people to live with others who share their preferences. As these preferences are likely highly correlated with income or education we get more segregation. We end up with some local authorities housing mainly poorer households and providing the kind of local services that they want. Other communities are richer and provide more (or maybe less) local services. These differences make all individuals better off, but you might never guess this by looking at the effect on poor areas.