Housing Prices and Monetary Policy: The Long View

A long 140 year history on housing prices and interest rates provides evidence that lower short term interest rates have significant positive effects on mortgage borrowing and housing prices. These findings by Oscar Jorda, Moritz Schularick, and Alan Taylor pose a dilemma for monetary policymakers: should they prioritize restoring nominal GDP or trying to prevent housing bubbles? This is well illustrated by the front page of the digital WSJ at the time of writing: while the Fed considers how long it should continue keeping rates low, there are signs that subprime lending is expanding again.

Should Monetary Policy Respond to Financial Stability?

I have written before on what I think about the role of monetary policy authorities and financial stability. I am skeptical of the direct role of monetary policy for three reasons:

First, there’s no reason to believe that the Fed can accurately identify bubbles in advance. Second, even if a bubble appears, it’s not clear that raising short term interest rates could pop it. Third, even if monetary policy ends up bringing asset prices down, it is likely to do so only through hurting the livelihoods of average Americans.

In the article, I walk through a variety of historical examples including the Great Recession, the recession of 1937, and the Great Depression.

I still stand by the arguments I made in the article and want to offer several extensions. Note that I mean to criticize the idea that monetary policy should lean against financial bubbles — not the more general idea that central banks can play a role through macroprudential policies.

  1. The Swedish experience with raising rates in order to bring down debt levels has been a disaster. Instead of lowering debt to gdp ratios, tighter monetary policy instead reduced nominal incomes and caused debt to rise as a percentage of GDP.
  2. Lower rates are usually not a sign that monetary policy has been easy, but rather that it has been too tight. Low rates are typically a sign that demand is low, and as such central banks should try to lower rates even further in order to stimulate demand. If instead you try to raise rates too soon, by cutting off demand you just extend the period of low rates.
  3. There may be structural reasons to believe that housing prices tend to be higher when aggregate demand is low. Capital equipment, which is much more sensitive to the business cycle, become less attractive at every interest rate relative to houses. As such if central banks want to head off a housing bubble they may be better off ensuring rates are low enough to stimulate demand.

It’s also not clear that Jorda et. al. can get much traction on the types of monetary policy decisions facing much of the developed world right now. In their statistical approach, they identify exogenous variations in interest rates with falling interest rates from abroad that get passed through as a result of a fixed exchange rate and free capital flows. In other words, their monetary policy does not come from a domestic central banker deciding to ease instead of tighten, instead it comes from a decision from a foreign central bank.

This opens up the possibility that what they’re really identifying is unwanted expansionary monetary shocks when the economy is already booming, or the potentially destabilizing effects of rapid foreign inflows of capital. Neither of these are quite appropriate for central banks right now:

  1. The United States, for example, is sluggishly chugging along — hardly a boom time in which bubbles are started. The authors tell a Eurozone parable and explain how easy ECB monetary policy for Spain caused a property bubble. But even here we see this timing issue in effect — even if low rates in a boom cause asset prices to rise while doing little for output, the same may not be true in a slump.
  2. If the negative rate shock comes from a domestic source, that would mean that capital flows out to chase yields elsewhere . This is the exact opposite of what would be going on to the countries in the Jorda et. al. dataset.

Leave a Reply