Commenter "jm" to a Brad Setser post remarks...

Because such a large fraction of any rise in the prices we pay the oil producers is immediately lent back to us — and in dollar-denominated loans — the effective drag on the US economy of higher oil prices is less by that fraction.

If we pay $10 more per barrel for oil, but $8 of that is immediately recycled back to us through purchases of US securities...

His analysis is right on, but perhaps it doesn't go far enough. As a thought experiment, let's make a bunch of assumptions, and see what happens:

1. Following JM, we posit that 80% of oil revenue is neither consumed nor used to fund domestic investment by oil producing countries, and is instead used to purchase debt securities on international markets.
2. We presume that the US has an 80% market share in the international debt securities market, and that this will remain constant over the short term. (Even if Iran, for example, buys only Euro debt, we assume that indirectly 80% of all international lending finds its way to the US.)
3. We guesstimate that the US accounts for 25% of the world's petroleum consumption, and that oil demand is perfectly inelastic (short term).
4. We assume that US current economic activity is cash constrained. That is, US agents will consume or invest all the cash that they borrow, without regard to the total debt load they are racking up in the process.¹
5. We assume for simplicitly that cashflows are instantaneous, that effects are not at all lagged.

Now, when a US consumer purchases that a dollar's worth of oil, the US economy sees an outward cashflow of $1. Oil producers see an incoming cash flow of $4 (since the US represents 25% of the world market by assumption). Oil producers lend $3.20 of that revenue to international markets. US borrowers take on $2.56 worth of debt. What's the net cashflow to the US economy associated with a purchase of oil? For ever dollar spent on oil, Americans get their dollar right back in lendings, plus an extra $1.56 of new lending to purchase homes, vacations, and SUVs with. In other words, an increase in the price of fuel is cash flow positive to the US economy! By assumption 4, this means that despite the reduction of net exports represented by high fuel prices, an increase in the price of oil leads to an increase in US GDP! It's a great deal, until something changes.

The assumptions above are not quite right. Probably something less that 64% of every dollar in the world spent on oil ends up repackaged as loans to the US. World oil demand is likely to be price elastic, especially outside of the US where funds spent on oil are not offset by incoming debt-driven cash flows. But I think the overall point stands. So long as the US collectively is not worried about its increasing debt load to the rest of the world, high oil prices may well be a current stimulus, and not a drag at all, on the US economy.

(For more on this topic, see Brad's post, the Nouriel Roubini post Brad was responding to, this post by "Calculated Risk" on his own blog, another post by CR on Angry Bear.)

Update: I think I should add that my personal opinion is that there is some truth to this story, in describing the recent past, but that the world is changing, the US is getting skittish about taking on more debt and the world is getting skittish about lending so cheaply, so that relying on high oil prices as a "stimulus" going forward would be a bad idea.

¹ Assumptions 4 and 2 are all mixed up, as the US "market share" of debt sales, which I've assumed constant, has something to do with Americans' willingness to increase their indebtedness, and with the price Americans can get for securitized debt, that is with the low yield Americans pay. For the past while, US long-term debt prices have been nearly constant, and the US share of sales to international debt markets has been consistently high. But as the price of Americans can command for taking on long-term debt is finally beginning to subside, assumptions 4 and 2 lose even the very modest contact with reality that they may once have had.