Counterfactuals are tricky

A commenter named “tpeach” recently asked the following:

My question is, what would have happened if the Fed hadn’t cut rates between Dec 07 and Apr 08? What would have happened to the base and velocity if the fed kept the rate stable while the Wicksellian or market rate plummetted during that time? Would the base shrink? If so, what are the mechanics behind that process? Also, how can the fed adjust the rate without changing the base? And why didn’t velocity drop when they cut rates during this time?

I wasn’t able to provide much of an answer.  Here I’d like to explain why.

At first glance, the obvious counterfactual would seem to be a smaller monetary base and a higher path of interest rates.  But that is a very fragile equilibrium, which could easily spiral off in one direction or another.  For instance, suppose the Fed had reduced the monetary base in late 2007 in order to prevent any fall in the fed funds rate.  What might have happened next?  One possibility is that the economy would have gone into a deep depression in early 2008.  Most likely, the Fed would have responded to that deep depression with a big rate cut and a big increase in the monetary base.  Thus in this case the counterfactual path of the base would have been a bit lower in late 2007, and much higher in early 2008. Indeed what I just described is roughly what did happen between early and late 2008—I am simply contemplating that scenario playing out 6 months earlier.

Monetary equilibrium often has “knife edge” qualities.  Imagine climbing along a mountain ridge with steep drop-offs on both sides.  If you are not at the peak of the ridge, you have the option of walking a bit further up the slope.  But if you go too far, you risk plunging down the other side.  Monetary economics is kind of like that.  Small changes are often “Keynesian” in character, meaning slightly tighter money means slightly higher nominal interest rates.  But larger changes can easily be “Neo-Fisherian” in character, meaning tighter money leads to lower nominal interest rates.  And it’s not just a question of more or less tight money, it’s more about expectations regarding the future path of policy.

Screen Shot 2017-05-12 at 12.04.59 PM

Yip Cloud recently pointed me to the latest in his excellent series of interviews of macroeconomists, this one of Atif Mian:

Some people have the 5-year adjustable rate mortgages (ARMs), others have the 7-year ARMs. Let’s say that the mortgages started in 2005. When 2010 comes, in the middle of the slowdown, those with the 5-year ARMs would get the interest rate reduction because the mortgages reset to a lower rate automatically. They get this reduction in the interest rate that the Fed was trying to pass through to the individual households. But those individuals who have a 7-year ARMs still have to wait for 2 additional years before they get a lower interest rate.

By taking advantage of this kind of variation in the cross-section, they can actually show the impact of the reduction in interest rate for the 5-year ARMs owners, by comparing them to the 7-years ARMs owners who didn’t receive the same reduction in interest rate just because they have a different kind of financial contract. What they’ve shown with this kind of analysis is that a reduction interest rate is actually beneficial. It actually allows the lenders to boost their spending and improves local economic outcome, in term of employment and aggregate demand. That’s just one example that actually shows monetary policy can be effective.

At the same time, that same work also shows why the monetary policy was ineffective. If you think about it, you need to be able to pass through the action of the Fed to the ultimate households. However, if people are struck in the 30-year fixed rate mortgages, they would not be able to take advantage of this lower interest rate environment. As a result, monetary policy is not able to pass through to the ultimate households. It is going to be constrained in the effectiveness. That’s a very important insight that has come about because of this kind of work that I emphasized. That’s a very interesting and useful development.

If people have borrowing capacity and willing to borrow, the same monetary policy shock can have more impact on the real economy. When you lower interest rate, for people who are prone to borrow more, they can borrow aggressively against a lower interest rate and that boosts the economy.

But if the same individuals have already borrowed a lot in the down-cycle, you can lower the interest rate but those individuals are underwater. They can’t borrow any more. Then the same reduction in interest rate is not going to have much of an impact on the macroeconomy. This kind of logic also suggests that monetary policy itself is going to be insufficient in dealing with the downturn. You need to focus on something that Sufi and I have to try to emphasize in our book.

I can’t emphasize enough that (as Friedman, Bernanke and Mishkin pointed out) changes in interest rates are not the same as changes in the stance of monetary policy, for standard “never reason from a price change” reasons.  Thus it’s not possible to draw any conclusions about the effectiveness of monetary policy by looking at the impact of changes in interest rates.  To take the most obvious reductio ad absurdum example, a Mexican currency reform exchanging 100 old pesos for one new peso will immediately reduce the price level by 99%, without any significant change in interest rates.


Tags:

 
 
 

12 Responses to “Counterfactuals are tricky”

  1. Gravatar of tpeach tpeach
    12. May 2017 at 09:08

    Hi scott

    I understand that interest rates are not a good indicator of the stance of monetary policy. I have read this blog for many years.

    My question maybe relates more to the mechanics of fed policy. How would the base shrink if the fed kept the fed funds rate stable while market intetest rates fall? I know this is thinking in “concrete steppes” but it is just a process I am trying to wrap my head around.

  2. Gravatar of Plucky Plucky
    12. May 2017 at 09:31

    A bit off-topic-

    One of your long-time bugaboos is the excessive attention paid to the manufacturing sector relative to its share of the economy. I think I’ve come up with a good analogy for why that is the case.

    Imagine a 1-factory town (let’s call it Rust, OH), except instead of being a part of Ohio it’s an independent statelet with its own currency (the “Rusty”) and is part of NAFTA. Since the USA is far and away its biggest trading partner, the Central Bank of Rust (CBR) decides the best monetary policy is to maintain a currency peg to USD and just accept whatever inflation or disinflation goes along with it. The government of Rust has debts in both Rustys and USD, as does the First Bank of Rust (FBR), which owns most of the governments Rusty-denominated debt.

    One day the factory closes. This presents a large problem for Rust because the factory’s products were the biggest item on the positive side or Rust’s current account ledger. Without the factory, Rust starts running massive current account deficit, which in short order wipes out CBR’s currency reserves and leaves it unable to defend the peg.

    What we have is a standard issue currency/debt crisis, which has no good solution, only bad, worse, and terrible solutions:

    Bad: Abandon the peg, allow the currency to collapse enough to attract someone to buy the old factory and put it to some productive use, default on the USD-denominated debt, and bring in the IMF to loan enough money to bail out the government and FBR. Rust’s middle class sees its standard of living collapse along with the purchasing power of the Rusty, but at least it can move forward and start growing again.

    Worse: Keep the peg, which will force Rust’s economy to adjust nominal wages down until, and which requires the IMF to loan ever-increasing amounts of USD to the CBR to accomplish. The IMF also arranges a “voluntary restructuring” of Rust’s sovereign debt that is a default in practice but not in theory (can’t let those damned CDS contracts hit!)

    Terrible: Keep the peg, no default, and an endless stream of growing debt to the IMF that will never, ever get repaid even when Rust’s nominal wages adjust to lower the standard of living enough and attract enough investment to stabilize the current account.

    In this fantasy world, both Rust and the IMF have learned from the example of Greece and choose option “bad” instead of options “worse” or “terrible.”

    In the real world, though Rust has only option “terrible” because it’s not actually an independent state, does not have its own currency, and only with extreme difficulty can it default on its debts. All its adjustment has to be nominal, which likely means a decade plus of deflation-depression. Furthermore, Rust might be stuck in option “worse than terrible” because a) all of its private debts are USD-denominated as well, so nominal deflation puts not just the government but individuals as well into semi-insolvency and b) as part of the US it is subject to US minimum wage regulation which (potentially) legally forbids its wages to fall sufficiently to attract the needed investment.

    This is the stylized story of manufacturing in the rust belt. Sure, most of any town’s GDP consists of services, but all of the services serve local demand and don’t bring in any currency from “abroad”. Without the factory, the town has a current account crisis. If you’re an auto-mechanic or a florist or own the diner, you get hit by it too.

    The point of the story is not to support protectionism or some sort of nationalist industrial policy, but to highlight that manufacturing does have impacts to a lot of places outsized to its share of GDP (even local GDP), and that it shouldn’t be surprising that politics reflects that

  3. Gravatar of Doug M Doug M
    12. May 2017 at 10:26

    The quote from Atif Mian bothers me

    All mortgages are callable. The holder of the 7/1 ARM, and the 30 year fixed have nearly the exact same ability to capitalize on lower interest rates to lower their monthly payments, and they can do that nearly instantaneously.

    However, the reduction of monthly payments does nothing to aggregate demand. If the borrower is paying less, then the investor is receiving less income. It is a zero sum proposition. You could take the Keynesian position that the borrower has a higher MPC than the investor.

    But the Monetarist should reject that as unnecessarily complicated. New lending and new money creation drives AD directly.

  4. Gravatar of Philo Philo
    12. May 2017 at 11:32

    Mian says: “if people are struck in the 30-year fixed rate mortgages, they would not be able to take advantage of this lower interest rate environment. As a result, monetary policy is not able to pass through to the ultimate households.” The mortgage borrowers do not get to take advantage of lower interest rates. Instead the advantage goes to the mortgage lenders–that is, to the stockholders of the mortgage lenders; but these are simply *different households* (with a bit of overlap, no doubt). So why isn’t there a comparable stimulative effect?

  5. Gravatar of Christian List Christian List
    12. May 2017 at 12:31


    I can’t emphasize enough that (as Friedman, Bernanke and Mishkin pointed out) changes in interest rates are not the same as changes in the stance of monetary policy, for standard “never reason from a price change” reasons.

    Let’s assume there’s a NGDP futures market and that the rising and falling prices of those futures are used to determine monetary police better. Isn’t that reasoning from a price change either?

  6. Gravatar of ssumner ssumner
    12. May 2017 at 12:57

    tpeach, It would shrink if the demand for money fell. In that case, the supply would have to be reduced to keep rates stable.

    Plucky, Yes, even 10% of GDP is quite a lot.

    Doug, Yes, that’s right.

    Philo, Good question.

    Christian, Not if NGDP is the policy GOAL.

  7. Gravatar of Christian List Christian List
    12. May 2017 at 13:21

    Okay, I try to get that. So I assume you move away from unemployment and inflation as policy goals. I’m trying to wrap my head around the implications. So market monetarism makes it way easier to distinguish between lacking AD and structural unemployment. Wow, that sounds pretty ingenious.

  8. Gravatar of Cloud Cloud
    13. May 2017 at 03:43

    Thanks Prof. Sumner for the mention!

    I am disappointed to myself for not spotting the “reason from the price change” issues in Prof Mian’s comment and ask him a follow up question. 🙁

    I think his view is based on the assumption that interest rate channel is one of the major channels of how monetary policy pass through, or at least how it can change households’ behaviors, which in turn have implications on aggregate demand. But I agree that this might not be the major channel of how monetary policy has its effect, and it is too quick to generalize the effect of this natural experiment.

    Thanks you for pointing all these out. I have learn a important lesson from it. 🙂

  9. Gravatar of Christian List Christian List
    13. May 2017 at 11:52


    I think his view is based on the assumption that interest rate channel is one of the major channels of how monetary policy pass through

    From what I read most people working for the FED and the ECB hold that view.

  10. Gravatar of flow5 flow5
    14. May 2017 at 05:30

    “changes in interest rates are not the same as changes in the stance of monetary policy, for standard “never reason from a price change” reasons”

    Then why does the Fed use interest rates as their monetary transmission mechanism?

    Interest is the price of loan-funds (market clearing rates). So savings flowing through the non-banks increases the supply of loan-funds, but not the supply of money & vice versa (it is a velocity relationship). The price of money is the reciprocal of the price level (the Fed’s bailiwick). And base-line investment hurdle rates, MARR, are idiosyncratic (demand being incentivized by, e.g., MACRS).

  11. Gravatar of ssumner ssumner
    14. May 2017 at 07:29

    Thanks Cloud, Keep up the good interviews.

  12. Gravatar of Geoff Orwell Geoff Orwell
    15. May 2017 at 02:30

    Terry Peach!? You taught me economics at Manchester University in 2007/8. Best ever.

Leave a Reply