Wednesday, November 30, 2016

Two Doubts About the Liquidity Effect

There is this story of how the Federal Reserve goosed the economy and the housing market with excessive monetary expansion, and then when the inevitable bust came, they fought the bust again by turning on the money spigots and lowering interest rates to save us from further decline.

The way that raising and lowering interest rates works is through the liquidity effect.  The Fed buys treasuries with new money.  That money gets deposited at the banks.  And, since the banks have more reserves as a result, there is less demand for cash in the overnight bank borrowing market, and short term rates decline.

The broader effect of a cash injection should be to raise rates, because the new money should raise inflation.  So, the liquidity effect is a temporary effect.  Over longer periods of time, interest rates should rise and fall with inflation rates.


Source
From May 2007 until March 2009, however, the Federal Reserve didn't buy any Treasuries.  In fact, they sold hundreds of billions of dollars worth of them.  Whether one describes Fed policy at the time as "tight" or "loose", how could the liquidity effect have brought down interest rates?  They weren't buying anything.  They weren't injecting money into any markets.

It is true that they were making emergency loans to banks, but they were purposely selling a dollar of Treasuries for each dollar of lending in order to prevent interest rates from declining.  In April 2007, the Fed held $902 billion in assets supplying reserve funds.  From April 2007 to August 2008, the total balance sheet of the Fed, including emergency loans and treasuries, grew by only about $2 billion per month.

In a falling rate environment, demand for cash should have been increasing, so, I think, if anything, it seems like the Fed would have needed to increase their holdings by more than usual in order to maintain a neutral stance.  But, in any case, how can anyone describe this period as a period where the Fed was dropping rates through the liquidity effect?

Furthermore, if the liquidity effect had anything to do with rates at the time, why were 3 month treasury rates well below the Fed Funds rate?  The Fed was injecting cash into the banking system through loans and it was selling hundreds of billions of dollars worth of short term treasuries.  If they were lowering rates through the liquidity effect, wouldn't treasury rates have been higher while the Fed Funds rate declined?

In the same graph, moved forward a few years, it looks like there is a similar problem with the QEs.  One of the Fed's stated goals with the QEs was to push down long term rates.  Yet, with each round, what we see is rates falling in the period before the round of QE and then rising during the implementation.

One might argue that rates were anticipatory.  But, the liquidity effect can't be anticipatory.  The liquidity effect happens because of frictions in the market that allow real-time changes in supply to temporarily overwhelm other factors that determine price, including expectations.  If markets can anticipate the liquidity effect, then they should anticipate all temporary movements.

It seems to me that the normal explanations for Fed policy during this period don't hold up.  Am I missing something?





9 comments:

  1. Hi Kevin,

    A few reactions.

    First, the standard narrative (how new money injection lowers rates) is flawed from the get-go by failing to distinguish between the Fed Funds rate and broader market rates. As you well know, Fed policy changes affect the shape of the yield curve in complicated ways. Specifically, when you write:

    "The broader effect of a cash injection should be to raise rates, because the new money should raise inflation"

    At the risk of taking a rhetorical point literally, my response is, "no no no!" The long-term effect of new money on inflation depends critically on future fed OMO's. If the Fed mops up the new money at the first whiff of inflation, then there won't be any inflation. The immediate effect of new money on long-term rates depend on market expectations. If markets expect the Fed to prevent inflation that way, then the new money doesn't cause any increase in expected inflation either, which means there is no Fischer effect, i.e., there is no increase in (long term) rates due to the new money's expected effect on inflation. And of course that's exactly what we saw. The Fed went way out of its way to signal that QE was temporary and the Fed would not allow inflation under any circumstances. And, the markets responded by fully believing the Fed, and parking all the new money in bank reserves, waiting for it to get mopped up again. Which totally undermined what the Fed was trying to accomplish.

    NGDPLT totally solves this problem of course.

    I won't respond point by point to the rest of your post. I presume you're making rhetorical points.

    > It seems to me that the normal explanations for Fed policy during
    > this period don't hold up. Am I missing something?

    Not that I can see.

    I don't see how there can be any liquidity effect when the banking system is full of excess reserves, and short-term Treasurys yield nothing. Bonds and cash are interchangeable in that environment. The EMH prevents any liquidity effect. For example, if the Fed withdrawing cash caused short-term rates to rise even an iota, you could make free money by buying Treasurys and selling them a moment later. I assume that's why the Fed has been using IOR instead of OMO to hit the Fed Funds target since 2016.

    I think https://fred.stlouisfed.org/graph/?g=bXHe is kind of amazing. You can see the market anticipating the Fed's Jan 2016 IOR increase so clearly, quickly and accurately. "Liquidity effect" is completely impossible because the 3-month Treasury yield change pre-dates the actual change in the Fed Funds target.


    ReplyDelete
    Replies
    1. Thanks, Ken. Interesting graph.

      You would think that my questions here are rhetorical. That's what's so weird. So many people treat these oddball claims as so obviously true that they don't need to be confirmed or questioned.

      Delete
  2. Among its many mistakes, I don't think the Fed realizes how powerful IOR is. They'd have to sell a couple trillion from their balance sheet to raise the Fed funds rate by 25bps, yet act as if raising IOR another 25 bps is "modest". And successful QE should result in an increase in the 10 year Treasury rate.

    ReplyDelete
    Replies
    1. Interesting. On the other hand, rising rates should lower the demand for holding currency.

      I'm a little undecided right now about how things will proceed. A month ago, I would have expected the next rate hike to trigger a yield curve flattening and contraction. But, since then, the long end of the curve has moved up about 1%, for legitimate reasons, I think, in hindsight. Now, I'm not sure. Raising rates still seems wrong, but to what scale?

      Delete
  3. Normally I'd say "don't reason from a price change", but in this case it's the right thing to do. An increase in the Fed funds rate precipitated by an increase in the rate of IOR is a price that is dictated (so we have to reason from it), and the Fed increases it to make banks want to hold more reserves. I'm not sure what that does to the demand for currency though.
    A month or two ago, I had the same thoughts as you did re the yield curve. If I were on the Fed, I would start letting my balance sheet slowly run down instead (call it negative QE?). But assuming they raise IOR at the next meeting, I currently expect little change on the rest of the yield curve as it appears that an increase is almost 100% expected.

    ReplyDelete
    Replies
    1. If the yield curve starts flattening after the rate hike, I would take that as a bad sign.

      Delete
    2. Me too. The Dec 2015 rate hike reduced the 10 year T from 2.25% to about 1.5%. And there was lots of bond market turmoil for a few months. Frankly, I'm very pleasantly surprised at how the economy kept turning out jobs this year.

      Delete
    3. Me too. But employment is a lagging indicator. Corporate profit is a leading indicator. It has signaled some recovery, so maybe there is hope. But, it hardly seems like a place in the business cycle to be in a hurry for tightening.

      https://fred.stlouisfed.org/graph/?g=bZOj

      Delete
    4. Wait. I take that back. The unemployment rate lags, but employment growth isn't a bad leading indicator.

      Delete