Yonatan Ben-Shalom, Robert Moffitt, and John Karl Scholz observe that U.S. welfare state expenditures have been growing over the last two decades, a time when many have assumed that the welfare state has been in retreat.

Accounting for social insurance programmes and means-tested transfer programmes (including the Earned Income Tax Credit), real annual per-capita expenditure grew from $3,618 in 1990 to $5,584 in 2008 (2007 dollars), a growth rate of 54%. Considering only means-tested transfer programmes, which are more directly aimed at low-income families, expenditure grew even faster, at 78%, but this was largely because of the explosion of Medicaid expenditures. Excluding Medicaid, means-tested expenditure grew by 26%. Expenditure in every programme except two – the Temporary Assistance for Needy Families programme and Workers Compensation – had positive growth from 1990 to 2008.
The growth of expenditures has been particularly strong among “in-kind” programmes, i.e. those that provide benefits for specific consumption expenditure, such as medical care, food, and housing. In fact, there are only two major means-tested cash transfer programmes left in the US. These are the Supplemental Security Income programme, which provides cash to poor people who are age 65 or older or have a disability, and the Earned Income Tax Credit, which provides cash tax credits to those with low or moderate earnings (up to approximately $44,000 for those with three or more children). The US provides few open-ended cash payments to the poor, but only benefits for specific, “deserving” categories of expenditure, and only to specific groups that have a favoured status (older people, people with disabilities, and workers).
In contrast to the view of Murray (1986) and others, these expenditures had a major effect in reducing poverty (Figure 1). The poverty rate among US families in 2004 based on their non-transfer, private income alone was 29%; the transfers reduced it to 13.5% (this excludes Medicaid and Medicare from the transfers). About 21% of families had pre-transfer incomes below half the poverty line; that number was reduced to 6.6% by transfer-programme expenditures. Many of the programmes paid benefits to families above the poverty line as well.

Yet the authors also observe that there has been a divergence in the generosity of transfer payments:

For example, elderly families with pre-transfer income less than 50% of the poverty line received $1,106 per month in 1984 and $1,254 in 2004. Elderly families with pre-transfer incomes between 50% and 100% of the poverty line also saw increasing transfer expenditures over time. Families with a disabled member, likewise, received more support in 2004 than in 1984 regardless of their pre-transfer poverty status.
However, this was not the case for families that did not have an elderly or a disabled member. Single-parent families with pre-transfer income less than 50% of the poverty line, for example, saw a reduction in transfer receipt, from $1,231 in 1984 to $766 in 2004; while those with pre-transfer income between 50% and 100% of the poverty line saw an increase ($448 to $832). Among non-elderly families who do not have a disabled member, similar patterns occur whether we examine two-parent families, those with an employed member, or those without an employed member, though the reduction in transfers for the poorest of these families are largest for families without an employed member.
Partly as a result of these expenditure reductions, the percentage of these non-elderly families without a disabled member who were in “deep” poverty rose from 1984 to 2004 – at the same time that the percentage of the population below the poverty line fell.
Within the overall increase in transfers in the US, therefore, there have been two redistributions. First, transfers have shifted towards the elderly and the disabled, and away from the non-elderly and non-disabled, overall. Second, within the non-elderly, non-disabled groups, there has been redistribution from the lowest-income families to families with greater incomes.
The fact that benefits to “deserving” groups like older people and people with disabilities have increased, presumably reflects the support they generate from policymakers and voters, their political power, or both. Likewise, the desire to reward work by increasing transfers to workers both in traditional poverty programmes as well as by the Earned Income Tax Credit, is widely shared by many in the US. It is not widely recognised, however, that the worst-off families, those with the lowest incomes and the lowest employment rates, have been left behind by trends in the system. Their receipt of benefits has fallen not only relative to better-off families, but in absolute terms as well. While there may be many “undeserving” families in this group, the vast majority have low skills, health problems, or other barriers to work which prevent them from having significant employment (Blank 2007). Policymakers need to develop mechanisms to assist these families. [Emphasis added]

I think that welfare state expenditures should primarily flow towards work supports, and so the second redistribution the authors describe troubles me far less than the first. More specifically, I think the shift towards more transfers to the old rather than the non-old is a mistake.