A common, even standard, way of thinking about the effect of interest rates on household decisions is to suppose that consumption tomorrow is just another “good” that the household may choose to spend money on today, it’s price being 1/(1+r) where r is the real interest rate (so if r = 5%, then to buy $1 in tomorrow’s dollars, it will “cost” you $0.9523 today). In that framework, an increase in the (real) interest rate will lower the price of consumption tomorrow relative to the various goods on offer today and households will consequently shift some of their income to savings. An increase in interest rates increases savings and lowers consumption today. Obvious, right?
That brings me to this article  on Bloomberg today. Here are the first three paragraphs:
Peking University professor Michael Pettis was discussing declining bank-deposit returns when a student interrupted with a story about her aunt that may stymie China’s plan to boost consumer spending.
“To send her son to university in six years it means she must replace each yuan in lost income with one from her wages,” the student said, according to Pettis.
The government’s policy of keeping interest rates low to reduce the burden of soaring municipal debt is costing savers as much as 1.6 trillion yuan ($236 billion) a year in lost income on bank deposits, according to Pettis, former head of emerging markets at Bear Stearns Cos. To make up the shortfall, savers have to set aside a larger proportion of wages, undermining China’s efforts to counter slower export growth with consumer spending at home.
This is essentially saying that savings can act as a type of Giffen good  — one for which consumers increase the quantity demanded when it’s price increases — when households take a future spending constraint into account today. When you know that you must have at least $x set aside tomorrow, an increase in the interest rate lowers the minimum amount of savings required today to meet the target. If that lower bound on savings was binding (i.e. without the future constraint you’d have preferred to spend more today) before the change, then higher interest rates will lead to a decrease in savings and an increase in consumption today.
To get this, we need is some (edit: non-divisible) future spending commitment that for some reason can’t be paid for with future income — two simple examples would be saving for retirement or, in America, for the university tuition of your children when they will be credit-constrained — and sufficient forward planning on the part of households so as to take it into account today. Both seem plausible for a large fraction of households.