I think this is an excellent idea:

These considerations ought also to apply today. As
things stand, we have an anomalous system whereby county councils and
metropolitan authorities are responsible for the delivery of social
services, but have no say over the policies which they are charged with
implementing. They are obliged to pay benefits, but have no discretion
over who qualifies for them.

A simple administrative reform would allow local
authorities to determine entitlement levels (although, ideally, such a
reform would come as part of the much wider shift towards
self-financing councils set out elsewhere in the Localist Papers).
Several British counties and cities have larger populations than
several US states. And the most successful US states, notably Florida
and Wisconsin, were the ones that devolved responsibility further, to
local welfare boards and county authorities. British councils already
have the staff in place to run social security. All they lack is
concrete incentives to ameliorate the system.

If we’re goingto have the Boy Dave (M) ‘s "double devolution" then we should indeed actually devolve things.

However, in the first part of the article there is  great praise for the US welfare reforms.

It is true that the 1996 Act was passed at a time of strong economic
growth; but this alone does not explain the almost miraculous shift
from dependency to gainful work. The number of families on welfare has
fallen from 5 million to 2 million. There are 1.6 million fewer
children in poverty. And, perhaps most impressive, the reforms lifted
groups who had been untouched by every previous welfare initiative:
poverty among black children fell from 42 to 33 per cent; among single
mothers from 50 to 42 per cent.

Be very, very, careful when using US poverty statistics. They are not collected in anything like the manner they are over here (or in most other countries) and most certainly do not mean the same thing.

First, the Federal Poverty line is more like a measure of absolute poverty (all of this is well explained in "Poverty and Discrimination" by Kevin Lang. My thanks to the Princeton University Press for my copy.) than it is relative. Set in the 1960s and updated for the CPI since (and not earnings growth, note) it’s three times the cost of a minimal diet at that time. Thus it’s more like the secondary of the UK poverty numbers (which was set at 50% of median income adjusted for household size in 1999 I think, adjusted for consumer inflation) rather than the primary one (60% of median adjusted for household size adjusted yearly for income growth).

The second and much more important point is that we and just about everyone else calculate poverty after the effects of both the tax and the benefits systems. The US does not, they calculate the poverty line, then add both income and direct cash transfers (ie, welfare payments themselves) and then define anyone below the line at this point to be in poverty.

However, they do not include in these calculations the major anti-poverty programmes. The EITC (equivalent to our tax credits, indeed, where they actually come from) is not included, nor are housing vouchers (similar to our housing benefit) nor Medicaid (given the NHS we don’t have an equivalent) nor Food Stamps (ditto, we have no equivalent). So by our standards the US poverty figures are very screwy. They are not actually measuring those "in poverty" in the same way our statistics are. Rather, they’re measuring those in poverty before they get help. And note, very importantly, that it’s an absolute level, not a relative one as ours is.

(This has its implications for the John Edwards anti-poverty plan. He, quite sensibly, wants to increase the EITC, provide more housing vouchers and so on. However, he also depends on the Federal Poverty line for his definition of the numbers in poverty. So while his plan will indeed reduce poverty in a sensible manner, not one single person will, by the measurement he uses, move out of poverty.)

But we should go further as well. Over the decades (since at least the 70s) the major bipartisan move in the US has been from direct cash aid (welfare) to those programs that are not included in the calculations of poverty. We would actually expect, given this, the poverty rate there to have been increasing, not staying static as it has done for much of the time.

Further, we would have expected the welfare reform to have increased the poverty level, as measured, as the cash handouts were limited but the other programmes, the ones we don’t count, continued. As it happens, as above, even by these odd measures, poverty fell, meaning that it was a much greater success than the plain numbers show.

There are some calculations which show US poverty levels in a similar (but not the same) manner to the way we calculate ours. As far as I can remember (and I am running on memory here) the last such set were done in 2004.  There are some very large changes: using the usual method child poverty is vastly higher than (not all that much of a surprise) when it’s measured including all of the benefits that are actually on offer, rather than only those which are direct cash transfers. However, even here they’re not using the same definitions we do, 60% of median earnings.

Yes, of course, all of this is detail, but it’s just trying to point out that you cannot take the US poverty rate and apply it sensibly to anything done elsewhere. The calculations are so different that we can’t simply transpose one set of statistics to the other.

I have looked, in a desultory manner, to see if I can find truly comparable statistics and I certainly haven’t been able to. Perhaps somone else can find them? Do they even exist? Calculations of US poverty on exactly the same basis as the ones we use here? 60% of median income, adjusted for household size, after all taxes, tax credits and both direct and indirect handouts?