It is commonplace among economists to emphasize that helping the poor with government welfare programs involves a trade-off: we get greater equality (or equity), but we sacrifice some economic efficiency. How to quantify this trade-off in a way meaningful for non-economists has always been a challenge, but the late economist Arthur Okun introduced a popular metaphor that helps clarify the role of efficiency in evaluating welfare programs.

When income is redistributed from rich to poor, Okun suggested imagining that “. . . the money must be carried from the rich to the poor in a leaky bucket. Some of it will simply disappear in transit, so the poor will not receive all the money that is taken from the rich.” Money does not literally disappear, of course, but inefficiencies produce results that can often be accurately characterized in this way. When there is no inefficiency, there is no leak in the bucket, and a dollar less for the rich means a dollar more for the poor. With inefficient policies, the bucket leaks, and the size of the leakage measures the magnitude of the inefficiency.

So, exactly how porous is the leaky bucket? When Okun wrote (1975), he thought the leakages were small, but today economists believe they are significantly larger than he supposed. Although there are too many variables to give a precise figure, we can get a rough idea that is consistent with the research literature by following a dollar on its journey from taxpayers to low-income recipients.

When the government acquires a dollar from taxpayers, it imposes a cost of greater than a dollar on them due to the inefficiencies produced by the tax policy. According to the President’s Council of Economic Advisers, taxpayers bear a cost of approximately $1.50 when the government collects a dollar in tax revenue. Of this fifty cent additional cost, ten cents is due to taxpayer compliance costs (record keeping, time spent filling out tax forms, etc.) and the remainder is attributable to distortions in economic behavior (effects on work, saving, and spending).

Armed with a dollar from taxpayers, the government spends it on a welfare program. The administrative cost of welfare programs absorbs part of this dollar, probably about ten cents on average. Thus, ninety cents worth of resources actually goes to recipients.

But recipients of welfare do not receive benefits they value at ninety cents because of the inefficiencies produced by welfare programs. Welfare programs undermine work incentives, affect living arrangements, and distort consumption decisions. The overall size of this leakage is less well established in the research literature than on the tax side of the transaction, but I estimate that a ninety-cent welfare benefit is worth only about sixty cents to recipients—a leakage of thirty cents.

One more leakage borne by low-income recipients of welfare is their compliance costs—providing evidence that they qualify for welfare benefits. I do not know of evidence regarding how large these costs are but will just assume that they are ten cents, the same as for taxpayers.

Thus, a transfer that places a cost of $1.50 on taxpayers provides a benefit worth fifty cents to recipients. In terms of the leaky bucket, two-thirds of the contents have leaked out due to the inefficiencies in the tax and transfer programs. We cannot provide a dollar in benefits to the poor at a cost of a dollar to the well off. Instead, taxpayers bear a three-dollar cost for each dollar of benefit to the poor.

Actually, this is likely to understate the size of the relevant leakage because the estimates are based on how the policies affect taxpayers and recipients at one specific time. But people do not stay in the same income classes over time—those who receive welfare benefits this year may be taxpayers next year. A recent Treasury study, for example, finds that more than half of the households in the poorest fifth of the population (likely to be receiving welfare) in 1996 were in a higher income class only nine years later (and likely to be paying taxes). What obviously matters is how the tax and transfer system affects people over their lifetimes, not just in one year.

To see why this means that the lifetime leakage will be greater than two-thirds, consider a scenario in which person A spends four years with a low income (receiving a fifty cent benefit) and one year with a high income (bearing a $1.50 cost), whereas person B has a high income for four years and a low income for one year. Person A has a significantly lower average income over the five-year period, but the 67 percent annual leak in the transfer bucket implies that A gains only fifty cents for the five-year period as a whole, while person B loses $5.50. When the leaky bucket is evaluated for the five-year period as a whole, more than 90 percent of the bucket has leaked out.

Recognizing that government policies other than welfare programs often harm the poor (such as the way immigration policy depresses wage rates for unskilled workers and ethanol subsidies increase food costs) makes it entirely possible that the welfare state as a whole ends up hurting those it is trying to help.

Each year high-income Americans transfer more than a trillion dollars to low-income Americans through a bewildering array of policies. The leaky bucket helps explain why the results have been disappointing. We should consider the possibility that the redistributive bucket is actually a sieve before we embark on a further expansion in welfare state policies.