|
Post by Oliver on Feb 19, 2016 15:08:08 GMT
Hi,
An EC2002 student asked a couple of questions following the lecture yesterday. Here they are (more or less):
1. I read about yields in the press. How do they relate to interest rates? 2. My general impression was that all yields rose during the crisis. I'm trying to reconcile that with your CC-LM narrative of the crisis in which a contraction in bank loan supply caused Y to fall, i^L to rise and i to **fall** (and thus credit spreads to rise).
I am going to give a (more detailed) response below (than I gave the student yesterday). But, before you read that, have a think if you could come up with an answer to either of these questions...
Best Oliver
|
|
|
Post by Oliver on Feb 19, 2016 16:08:48 GMT
Hi,
1. For our purposes, we can use the term interest rates and yield interchangeable. Both our assets - bonds and bank loans are very simple. Borrow x today, repay y tomorrow. The interest rate and the yield are the same: 100*((y/x)-1). So if you read the FT, and you read that "yields on government bonds rose last week", feel happy to substitute the word interest rate in there. For the subtleties, wikipedia is a great resource, or take one of the finance modules next year!
2. The student's second question is a very astute one.
The short answer is: Maybe I oversold the CC-LM model a little!
The long answer:
We started with the IS-LM model, which is a great little model, but makes many difficult to swallow assumptions. The one that Bernanke and Blinder picked up on was the assumption of a single asset which all borrowers used. So, they added a second, made the distinction between bonds and bank loans, and made the distinction between two different interest rates. Good so far. The problem is, this is still a gross oversimplification of the assets and interest rates in the real world.
Here is a slightly richer list of assets/interest rates that you might read the FT or The Economist discussing (the list in not exhaustive but will get somewhere where I can answer the original question!)
1. Short-term interest rate: This is the (risk-free) interest rate that a central bank (ECB, Fed, BoE) might target. It might be an overnight borrowing rate, for example. 2. Long-term government bond interest rate: The interest rate at which a government can borrow at for maybe 10 years. Lenders expect this to be risk-free (if for example its the US or UK government - maybe not so for Greece or Argentina). 3. Long-term AAA corporate bonds. These are the bonds being issued by private sector firms. AAA are the really good firms (although not quite risk-free like government bonds). 4. Long-term BBB corporate bonds. These are bonds issued by much riskier firms. 5. Bank loans. Here the loan interest rate is going to depend on the characteristics of the borrower and certain characteristics of the loan. It's actually quite difficult to get good data on this.
So, roughly, what happened to these assets/interest rates during the crisis?:
1. Short-term risk-free rates ***fell sharply*** (until they got stuck at the the zero lower bound) 2. Long-term risk-free government bond rates ***fell*** 3. AAA corporate bond rates ***rose*** 4. BBB corporate bond rates ***rose sharply*** 5. Lots of borrowers simply lost access to bank loans (which is a bit like saying that the bank loan interest rate for some borrowers went up to ***infinity***)
Remember in our IS-LM model, the bond interest rate was supposedly catching all these 5 effects combined!
In the CC-LM model, think of the bank loans as trying to capture 4 & 5 - the risky borrowers. Interest rates went up in the crisis. In the CC-LM model, think of the bond interest rate as trying to capture 1, 2, & 3. It supposed to be capturing the interest rate at which the government can borrow and the interest rate at which low risk firms can borrow.
And, the CC-LM model predicted: ***i^L rises, i falls*** when a contraction in bank loan supply is the trigger for the crisis.
So, now hopefully you see where the problem comes and why the original question was asked. If you think of the bond rate in CC-LM as 1 or 2, the model gets the sign right relative to the data; if you think of it as 3, the model gets the sign wrong relative to the data.
So - a great question. Thanks a lot. Thinking hard about how the models we learn relate to the real world is super important. The CC-LM model is still an oversimplification and still has main problems. Keep this in mind...
...and keep asking tough questions!!
Best Oliver
|
|