07 Jan 2016

Continuing With My Questioning of Sumner

Scott Sumner 4 Comments

Scott partially answered my question (which I posted here for your viewing convenience), but he didn’t elaborate on why he uses the formula (i-IOR) to get his desired result that IOR is contractionary, even though higher interest rates per se are expansionary. (It should go without saying that I’m just dabbling within Scott’s framework to make sure it’s internally consistent. Obviously I don’t think it’s useful to classify “boosting NGDP” as “expansionary” etc.)

In the comments here at Free Advice, Aaron gave the natural answer:


i–IOR is th relevant number because IOR is the floor at which banks will lend. No one in their right mind will lend below IOR.

Before IOR, the floor was zero, so i–IOR = i. Now is not. So to compare historic data, your have to subtract IOR.

I agree that this is probably the reasoning Scott is using (though he never spelled it out). But if so, there are some problems. Very quickly:

==> Banks aren’t the only actors in the economy. There are businesses and households that hold cash balances, and factor interest rates into their decisions. So if interest rates across the economy go up because the Fed hikes IOR, then doesn’t that make everybody in the economy less eager to hold cash balances? Yet only the commercial banks are getting the subsidy. To switch examples: If the federal government starts stockpiling wheat and thereby raises the market price, that would still cause consumers to buy less bread. We wouldn’t say, “Nah, this is a subsidy and so the relevant metric is b-w, where b is the market price of bread and w is the subsidy price the government pays farmers for wheat.”

==> Even if we focus just on the commercial banks, it’s not obvious to me that they only care about the spread, as opposed to the absolute value of the market interest rate. The reason is that the Fed is simply a very safe borrower, from the perspective of the commercial banks. For example, forget about IOR; imagine it’s back in 2004 before that policy. Now some major housing developers approach the commercial banks looking to borrow money, and that pushes up market interest rates from 3% to 4%. Would Scott say that this isn’t expansionary, because the banks really care about (i-d), which is the market interest rate minus how much they can earn if they lend to the developers? Of course not; that doesn’t even make sense. So by the same token, when the Fed comes along and says, “Hey, if you ‘lend’ your reserves to us, we’ll pay you 25 bps instead of 0 bps,” I don’t see why that is qualitatively different from increased demand for reserves from any other borrower.

==> Even if you think the above two points are wrong, I have a third, independent problem with Scott’s analysis. Let’s say the Fed raises IOR by 25 basis points. Scott wants to say, “That’s contractionary because the relevant metric is i-IOR, and the Fed just raised IOR.” But hold on. An immediate reaction is for i to rise in response. Without Scott telling us more, I think most people would assume market interest rates in turn would all rise by 25 bps as well. So it would seem to a first approximation–using Scott’s stipulated framework–IOR should have no effect. In other words, any increase in IOR will presumably cause i to increase by the same amount, and so the metric (i-IOR) is unaffected by the policy of IOR. And yet Scott has spent the last 7 and a half years arguing quite vigorously that IOR is contractionary.

* * *

If this were just some minor point, I wouldn’t be making such a big deal out of it. But since the issue of IOR (and the possibility of negative IOR) has been floated for years by Market Monetarists as the smoking gun against the liquidity trap, I think Scott should at some point clarify how all this works, when his most recent posts are arguing that lower interest rates are contractionary per se.

4 Responses to “Continuing With My Questioning of Sumner”

  1. Transformer says:

    If more people decide they wish to borrow money to buy stuff then interest rates rise. This will induce others to lend money they previously kept as cash balances (which the borrowers then spend) so velocity rises and this is expansionary.

    If some people decide for some weird reason to borrow money just to hold in their bank account, then this will still cause rates to rise , but will not be expansionary as V will not increase (cash balances are just transferred from one place to another).

    IOR is a bit like the second scenario. The CB is saying “we will borrow as much you are prepared to lend us at the IOR rate” but as they don’t spend the money it doesn’t increase V, so isn’t expansionary.

    Of course if commercial banks can make good loans above IOR then banks will not “lend” to the CB.

    When Sumner says “the demand for reserves is negatively related to the market interest rate minus IOR”. I take this to mean that as IOR reduces the opportunity cost of holding reserves, the higher IOR is, the higher bank reserves will be. In other words banks always need access to reserves to meet potential obligations. When the CB pays IOR they will hold more than if there is no IOR , where it will be cheaper for them to go the interbank window if they run short.

    • Major.Freedom says:

      “If more people decide they wish to borrow money to buy stuff then interest rates rise.”

      Not true. If more people lend, then it possible total lending will not rise, as per capita lending falls.

      But even if we assume the average size of a loan does not fall, but remains the same, such that all loans increased, then there is still the question of where/how the additional loanable funds were “financed”. If they were financed by additional inflation, then it is likely rates will rise, but in that case the reason rates rose was because of the inflation, not because of the additional loans. If on the other hand the additional loans were financed by something other than inflation, then we have to reduce either investment or spending in some other areas of the economy. In this case, we of course have to assume constant spending, but if we do assume that, then the rise in lending could have come from a reduction in consumption spending, in which case rates will likely fall, since productive expenditures will have risen relative to overall expenditures. But if total spending was not constant, then rates could either rise or fall depending on whether there was a rise or fall.

      Long story short, absolutely no way to conclude that rates will definitely rise or definitely fall on the basis of a single variable of “more people”. That is far too crude and betrays the complexity of real world markets.

      No, no excuses to this. None. There can be no ” generally speaking”. It is as stated not right.

      • Transformer says:

        I suppose I should have been clearer that I meant if everything else stayed the same except the overall demand for loans increased then this would raise interest rates.

        I think my main point though is that I believe Bob is wrong to state that IOR is expansionary.

        No matter what the market-clearing rate of interest IOR will always encourage banks, at the margin, to keep higher reserves than with no IOR and this will always be contractionary for a given level of reserves.

  2. guest says:

    Just a thought: Are we getting away from the fact that “expansionary” refers to the money-substitute supply, held in reserves or otherwise?

Leave a Reply