In his recent post commenting on the op-ed piece in the Wall Street Journal by Michael Woodford and Frederic Mishkin on nominal GDP level targeting (hereinafter NGDPLT), Scott Sumner made the following observation.
I would add that Woodford’s preferred interest rate policy instrument is also obsolete. In the next recession, and probably the one after that, interest rates will again fall to zero. Indeed the only real suspense is whether they’ll be able to rise significantly above zero before the next recession hits. In the US in 1937, Japan in 2001, and the eurozone in 2011, rates had barely nudged above zero before the next recession hit. Ryan Avent has an excellent post discussing this issue.
Perhaps I am misinterpreting him, but Scott seems to think that the decline in real interest rates reflects some fundamental change in the economy since approximately the start of the 21st century. Current low real rates, below zero on US Treasuries well up the yield curve. The real rate is unobservable, but it is related to (but not identical with) the yield on TIPS which are now negative up to 10-year maturities. The fall in real rates partly reflects the cyclical tendency for the expected rate of return on new investment to fall in recessions, but real interest rates were falling even before the downturn started in 2007.
In this post, at any rate, Scott doesn’t explain why the real rate of return on investment is falling. In the General Theory, Keynes speculated about the possibility that after the great industrialization of the 19th and early 20th centuries, new opportunities for investment were becoming exhausted. Alvin Hansen, an early American convert to Keynesianism, developed this idea into what he called the secular-stagnation hypothesis, a hypothesis suggesting that, after World War II, even with very low interest rates, the US economy was likely to relapse into depression. The postwar boom seemed to disprove Hansen’s idea, which became a kind of historical curiosity, if not an embarrassment. I wonder if Scott thinks that Keynes and Hansen were just about a half-century ahead of their time, or does he have some other reason in mind for why he thinks that real interest rates are destined to be very low?
One possibility, which, in a sense, is the optimistic take on our current predicament, is that low real interest rates are the result of bad monetary policy, the obstacle to an economic expansion that, in the usual course of events, would raise real interest rates back to more “normal” levels. There are two problems with this interpretation. First, the decline in real interest rates began in the last decade well before the 2007-09 downturn. Second, why does Scott, evidently accepting Ryan Avent’s pessimistic assessment of the life-expectancy of the current recovery notwithstanding rapidly increasing support for NGDPLT, anticipate a relapse into recession before the recovery raises real interest rates above their current near-zero levels? Whatever the explanation, I look forward to hearing more from Scott about all this.
But in the meantime, here are some thoughts of my own about our low real interest rates.
First, it can’t be emphasized too strongly that low real interest rates are not caused by Fed “intervention” in the market. The Fed can buy up all the Treasuries it wants to, but doing so could not force down interest rates if those low interest rates were inconsistent with expected rates of return on investment and the marginal rate of time preference of households. Despite low real interest rates, consumers are not rushing to borrow money at low rates to increase present consumption, nor are businesses rushing to take advantage of low real interest rates to undertake shiny new investment projects. Current low interest rates are a reflection of the expectations of the public about their opportunities for trade-offs between current and future consumption and between current and future production and their expectations about future price levels and interest rates. It is not the Fed that is punishing savers, as the editorial page of the Wall Street Journal constantly alleges. Rather, it is the distilled wisdom of market participants that is determining how much any individual should be rewarded for the act of abstaining from current consumption. Unfortunately, there is so little demand for resources to be used to increase future output, the act of abstaining from current consumption contributes essentially nothing, at the margin, to the increase of future output, which is why the market is now offering next to no reward for a marginal abstention from current consumption.
Second, interest rates reflect the expectations of businesses and investors about the profitability of investing in new capital, and the expectations of households about their future incomes (largely dependent on expectations about future employment). These expectations – about profitability and about future incomes — are distinct, but they are clearly interdependent. If businesses are optimistic about the profitability of future investment, households are likely to be optimistic about future incomes. If households are pessimistic about future incomes, businesses are unlikely to expect investments in new capital to be profitable. If real interest rates are stuck at zero, it suggests that businesses and households are stuck in a mutually reinforcing cycle of pessimistic expectations — households about future income and employment and businesses about the profitability of investing in new capital. Expectations, as I have said before, are fundamental. Low interest rates and secular stagnation need not be the result of an inevitable drying up of investment opportunities; they may be the result of a vicious cycle of mutually reinforcing pessimism by households and businesses.
The simple Keynesian model — at least the Keynesian-cross version of intro textbooks or even the IS-LM version of intermediate textbooks – generally holds expectations constant. But in fact, it is through the adjustment of expectations that full-employment equilibrium is reached. For fiscal or monetary policy to work, they must alter expectations. Conventional calculations of spending or tax multipliers, which implicitly hold expectations constant, miss the point, which is to alter expectations.
Similarly, as I have tried to suggest in my previous two posts, what Friedman called the natural rate of unemployment may itself depend on expectations. A change in monetary policy may alter expectations in a manner that reduces the natural rate. A straightforward application of the natural-rate model leads some to dismiss a reduction in unemployment associated with a small increase in the rate of inflation as inefficient, because the increase in employment results from workers being misled into accepting jobs that will turn out to pay workers a lower real wage than they had expected. But even if that is so, the increase in employment may still be welfare-increasing, because the employment of each worker improves the chances that another worker will become employed. The social benefit of employment may be greater than the private benefit. In that case, the apparent anomaly (from the standpoint of the natural-rate hypothesis) that measurements of social well-being seem to be greatest when employment is maximized actually make perfectly good sense.
In an upcoming post, I hope to explore some other possible explanations for low real interest rates.