Friday, September 28, 2012

Facts for Jame Bullard

St. Louis Fed President James Bullard just delivered a speech where he claims that U.S. monetary policy has been stellar over the past four years.  In fact, he says monetary policy was "close to optimal." Yes, I about chocked too after reading that line.  His view is that (1) the Fed kept the price level on its long run trend and that (2) there was a reduction in U.S. potential output that undermines that case for looking at NGDP being below trend.   Consequently, there is nothing to the claims of insufficient aggregate demand.  It is all structural, end of the story.   

This is not the first time Bullard has made claim (2), but it is the first time he has combined it with the claim that the Fed has been doing a fine job since 2008.  Scott Sumner has already responded and I am sure others will too.  My response to Bullard is that your theory cannot explain an important development that have been ongoing since 2008: the elevated demand for liquidity.  If we are simply on a new growth path and the Fed has done a fine job with monetary conditions, then why is the demand for safe assets still so pronounced?  Below are five facts about this ongoing demand for liquidity:
Fact 1
Households and other retail investors have been increasing their holdings of FDIC-protected saving deposits at an usually rapid rate.  This surge in saving deposit growth starts during the crisis in 2008 and still is growing.  Since that time, households have acquired almost $2.4 trillion worth of saving deposits.

Fact 2
Households have been one of the biggest purchasers of U.S. treasuries over the past 4 years.  Through direct purchases, households have gone from holding $264 billion in 2007:Q4 to $1.3 trillion in 2012:Q1. If mutual funds purchases of treasuries reflect indirect household purchases, then household holdings have grown from $647 billion in 2007:Q4 to $2.2 trillion in 2012:Q1. This is more than double the Fed's increase in treasury holdings.  Only foreigners have bought more. (Update: these numbers were based on SIFMA data thru 2012:Q1.  The Q2 data shows major revisions such that direct household holdings went from $202 billion in 2007:Q4 to $878 billion in 2012:Q2.  Including mutual fund purchasers of treasuries put household holdings at $584 billion in 2007:Q4 and $1810 billion in 2012:Q2. These combined purchasers are still  more than the Fed's, but not double. Interestingly, these revision show the Fed's holdings at about 15% of total marketable treasuries.  So much for the Fed buying up all the deficit)

Fact 3
Household holdings of liquid assets as a percent of total assets soared in 2008 and have yet to come down.  Household porfolios, therefore, are still inordinately weighted toward safe, liquid assets and have yet to undergo the type of portfolio rebalancing associated with a robust recovery  (i.e. a rebalancing of portfolios away from low yielding, liquid assets to higher yielding, riskier assets will spawn indirect effects on aggregate nominal spending via balance sheet and wealth effects and directly through purchases on capital.  See here for evidence on this portfolio channel).

Fact 4
Interest rates on safe assets are at historical lows.  Given the facts 1-3 above,  it should be evident that these low interest rates are  the result of an elevated, ongoing demand for safe assets.   As noted above, the biggest purchasers of U.S. treasuries has not been the Fed over the past four years.    So don't blame the Fed.

Fact 5
Since the start of the crisis in 2008, movements in the stock market and expected inflation have been highly correlated as seen here.  This is an unusual relationship that only started during the crisis and continues to this day.  Thus, whatever caused it to happen in 2008 is still with us today.  The easiest interpretation that is consistent with facts 1-4 above is that spike in demand for safe, liquid assets that began in 2008 has yet to subside.  Here is why: any rise in expected inflation that would reduce this intense demand for liquid assets and thereby raise the expected future nominal income, would also raise expectations of higher future stock prices.  In anticipation of this development, investors buy stocks in the present (see David Glasner).  Thus, expected inflation and stock prices are currently related.  Since liquidity demand still remains elevated, there apparently hast not been a big enough increase in expected future nominal income to break this relationship.
Those are the facts.  They all indicate there is still an elevated demand for liquidity that is slowing the economy. Now here is the thing.  By changing expectations about the path of future nominal income, the Fed could reduce this demand for liquidity and spark a recovery in nominal expenditures.  Specifically, by setting a NGDP level target, the Fed could increase both the certainty and expected amount of future nominal income.  This would increase demand for credit today and kick start the private creation of safe assets.  It also would decrease the demand for safe assets.  That the Fed has not done this and, as a result, there is still elevated liquidity demand screams Fed failure, not success.  

 Now this is not to say there are no structural problems. Only that there still remains a sizable excess demand for liquidity that is constraining aggregate nominal spending.  My hope is that James Bullard and others who share his views will wrestle with these facts that point to this safe asset demand problem.

21 comments:

  1. Add this to the list. Really great post.

    ReplyDelete
  2. David,

    Technically speaking, doesn't an excess demand for safe assets imply that interest rates are too low?

    JoeMac

    ReplyDelete
  3. Interesting post. But I figure the excess demand for cash equivalents implies a deficient demand for investment.

    Which always comes back around to lack of income, jobs and aggregate demand.

    All the monetary tools seem to fall into the 'maybe' category regarding job and income stimulus. We need guarantees right now.

    My preference is we hire several million folks ASAP. And pay them nicely.

    ReplyDelete
  4. Not that I agree with Bullard (in fact I don't think I could disagree more), but I don't find any of your money demand figures overly convincing, or in some cases, even relevant. Money flows are not a good measure of money demand because flows can change for reasons unrelated to expected monetary policy. Safe asset demand can do the same -- e.g. banking regulatory capital requirements have moved towards favoring safe assets like government bonds and collateral requirements between financial counterparties have undoubtedly done the same thing post-Lehman.

    Fact 1 is an incomplete measure of household savings trends. Savings deposits have grown rapidly, but time deposits have not. Looking at the Fed Flow of Funds data from 2005 to 2011, total growth in savings and time deposits was much more rapid from 2005-2007 than from 2008 onward. Why have savings deposits grown unusually fast while time deposits have done the opposite? Probably because of low real interest rates -- savers do not wish to lock up their savings in time deposits and choose to hold more in discretionary savings accounts. Take altogether, the data does not reveal that households are rapidly hording dollars in deposits -- only the composition has changed greatly.

    Fact 2 shows that households actually sold $268 bil. of Treasuries in 2011, the most recent year of Flow of Funds data. The Fed data also is a tad misleading because of the starting date of measurement(Q4 2007). Recall that in 2008 the Fed sold $264 bil. of Treasuries and used the proceeds to buy other credit market instruments in hopes of easing the financial crisis. Since 2009, the Fed has bought an average of $400 bil. annually in Treasuries.

    Fact 3 is correct in premise, but incorrect in reasoning. The ratio of liquid assets to total assets soars because the denominator (total assets) falls by roughly 16% from 2007 to 2008. Total assets (mostly due to real estate values) remain below 2007 peak values hence the still elevated ratio you describe.

    Fact 4 isn't very convincing. I'm not one of those who suggests that the Fed has caused interest rates on Treasuries to fall to historic real and nominal post-war lows (I favor market monetarist explanations such as low expected NGDP growth, falling natural interest rates, Great Stagnation effects etc.)However,it's easy to see how those who disagree with you and blame the Fed could do so via simple correlation -- the Fed has been very active buying Treasuries the past three (not four) years and most of the interest rate decline has taken place over those same three (not four)years.

    ReplyDelete
  5. Numeraire, you are missing the forest for the trees. The high demand for liquid assets, regardless of cause, is a constraint on aggregate nominal expenditures. And it seems you are trying so hard to disprove this point you are not really seeing the data for what it says.

    Thus this statement misses the point,

    "Money flows are not a good measure of money demand because flows can change for reasons unrelated to expected monetary policy."

    The whole point of the excess money demand or safe asset demand is exactly that many different factors over the past four years have caused it to increase. Monetary policy has nothing to do it. The only job of monetary policy is to respond to these money demand shocks, to offset. And this is where it has failed miserabl.

    Also, your saving critique misses the real reason households are moving into saving deposits: increased demand for liquidity. Saving deposits are more liquid than time deposits. And the timing of the stark shift into saving and out of other less liquid assets--not just time deposits but also money market funds, etc.--should makes this clear. You are trying so hard to ignore the obvious.

    You are also missing the bigger point on the treasury acquisition by households. It is true that household moved out of treasuries in 2011, but they also readily moved back into them in the first part of 2012. So much so that they purchased more in that quarter than the sold in 2011. So why focus on 2011? The bigger point is that over the past four years households have been one the biggest purchasers and thus have driven down interest rates. Again, missing the forest for the trees. (Plus, if you add mutual fund purchases of treasuries, it is a gain for 2011. Also, direct household sales in 2011 were closer to $92 billion, not $264.)

    Finally, though the Fed did purchase many treasuries in 2011, these purchases only put it back at the share of total treasuries it had prior to the crisis. That's about 16%. In fact, the 2011 purchases merely offset the large sale of treasuries it did in 2007 and 2008. In short, the Fed is not holding anymore treasuries as percent of total treasuries than it has held on average prior to 2008. And its only 16%! Again, missing the forest for the trees.


    Your critique on point 3 also falls flat for several reasons. First, while total assets fell, the stock of liquid assets grew. Total assets are now about 5% below their 2007:Q4 value. Liquid assets are now about 25% above their 2007:Q4. Why ignore this? Second, see the link the to paper in the post. It proves systematic evidence that links this ratio to movements in aggregate nominal spending, regardless of why the ratio changes.

    Your final critique ignores all this evidence and the fact that Fed is now holding roughly the same % of treasuries that it has over the past decade, 16%. It is not that big a player. People may want to stick their head in the sand and ignore these numbers, but that doesn't mean they are right. The Fed is not an important reason for the decline in treasury yields since 2007.

    It is hard to rationalize away something that is real.

    ReplyDelete
  6. Most of the interest rate decline had little to do with the FED and occurred before treasuries were even bought, which means little buying of treasuries. Don't tell me buying a penny is "buying". That is not.

    Operation Twist is not buying.

    Yes, the high excess demand for safe assets does tell us interest rates are to low, which is the point this blog has made over..........and over..........and over...........and over........AND OVER!!! again. Since the FED is not satisfying the demand, interest rates do not rise.

    The media is so confused right now, they can't even think straight.

    ReplyDelete
  7. JoeMac:

    Excess demand for safe assets actually means interest rates are not low enough relative to the natural interest rate. In fact, all of this discussion about excess demand for safe assets is equivalent to saying the actual and expected market interest rates are above the natural (market clearing) interest rate.

    Now your statement in one sense is correct. A full recovery where the demand for safe assets falls would be accompanied by a rise in interest rates. But the rise would occur as a consequence of the shift out of safe assets into riskier ones. One could not just raise rates and hope for the economy to follow. It has to be the other way around.

    ReplyDelete
  8. Let me also note, the ACCA acted like a nominal spending surge in 2009-10 which drove interest rates higher for awhile. Little surprise when that wore out, interest rates declined and declined and declined.

    Why can't people get it?

    ReplyDelete
  9. Prof. Beckworth,

    Saving deposits are more liquid than time deposits. And the timing of the stark shift into saving and out of other less liquid assets--not just time deposits but also money market funds, etc.--should makes this clear.

    The timing of the shift into savings deposits is present after every bear market with the magnitude of the shift largely in correlation to the magnitude of the bear market. In fact the growth rate of savings deposits is almost inversely correlated to the growth rate in retail money funds. You must be eyeballing the data because the savings deposit component of M2 as a percentage of M2 does not grow until after the trough in NGDP.

    As evidence of a point I made before regarding investor psychology, retail money funds have been declining since the bull market began. The same phenomenon occurred in the early years of the 2002-07 bull market and to a lesser extent after the less severe bear markets of the early 1980's and early 1990's. During these periods in which retail money funds declined, savings deposits grew at faster than average rates and later slowed as RMF's grew rapidly (hence the inverse correlation).

    For the record, time deposits very much appear to be interest-rate sensitive. Doesn't this chart look awfully identical to the FFR over the same measured period?

    http://research.stlouisfed.org/fred2/series/WSMTIME?cid=29

    ReplyDelete
  10. So much so that they purchased more in that quarter than the sold in 2011. So why focus on 2011?

    The data I'm using from the most recent Fed Flow of Funds Q2 2012 does not correlate with your figures. Peak holdings of Treasuries was Q3/Q4 2010.

    It also appears that much of the apparent rise in household Treasury holdings during 2009 was related to the decline in household holdings of GSE debt, post-Fannie/Freddie conservatorship. It looks like an asset swap from one asset thought to be safe (GSE debt) into Treasury debt. That would be more of a substitution than fresh demand for safe assets.

    In 2008, households owned a combined $897 bil. in Treasury and GSE debt. In Q2 2012, households owned a combined $884 bil. in Treasury and agency debt.

    ReplyDelete
  11. Numeraire,

    the data I use comes from SIFMA. It provides a cleaner breakdown of treasury holdings than the flow of funds. I did find, though, that the 2012:Q1 release I was using was significantly revised. I edited the post to reflect their Q2 release. Doesn't change the conclusion, though.

    Again, the main point here. The demand for safe assets remains elevated, interest rates are historical lows, and NGDP remains below trend. All of these point to monetary policy that has failed to do its job.

    ReplyDelete
  12. Numeraire,

    One last thing on this:

    The timing of the shift into savings deposits is present after every bear market with the magnitude of the shift largely in correlation to the magnitude of the bear market. In fact the growth rate of savings deposits is almost inversely correlated to the growth rate in retail money funds.

    This is just another manifestation of the increase demand for safe and liquid assets. MMMF were considered safe until they broke the buck in late 2008. It's about at that time that investors start moving their funds into saving accounts. Nothing myseterious here, just part of the bigger excess money demand problem the Fed has yet to fully offset.

    ReplyDelete
  13. The demand for safe assets remains elevated, interest rates are historical lows, and NGDP remains below trend.

    I agree with the latter two points wholeheartedly and even agree somewhat with the idea that safe assets are at elevated levels. The problem is that your figures and ratios do not show the evidence of this very well.

    Changes in asset values are not the same as asset reallocations, and it is the former that has greater influence when observing the ratio of safe liquid assets to total assets.

    The timing of asset reallocations also does not fit with your assertions, a point I've made repeatedly in our back and forth. You haven't addressed whether or not the large increase in household Treasury holdings is at all related to the disappearance of previously-deemed safe GSE debt, which would not represent elevated demand but rather asset substitution.

    ReplyDelete
  14. Numeraire:

    Changes in asset values are not the same as asset reallocations

    Attempted asset allocations are why asset valuation changes occur in the first place. In the current context, as the demand for safe has increased their prices are being driven up (i.e. interest rates are being driven down) given the relative shortage of them. Treasury prices are going up for this very reason. The whole point of excess demand for safe, liquid assets is that attempted allocations are getting frustrated and thus prices are going up. So this distinction you make is moot here.

    I don’t know the answer to the GSE question, but it too is moot given the main point being made here. As I keep noting, the share of liquid assets for HH has grown inordinately—and remember the liquid portion has grown 25% over the past four years while the total assets has fallen 5%--and that this ratio is systematically related to NGDP. I have given you links to a paper that shows this. You also continue to ignore the obvious fact that U.S. marketable debt has grown dramatically and yet most of it is readily being bought up by parties other than the Fed. (See my most recent post.) In normal conditions such record growth of debt would lead to higher interest rates, lower prices not the opposite. That so much U.S. debt could be purchased screams excess liquidity/safe asset demand. And given the all the other things going on it baffles me that you continue to try to explain it away.

    Regarding switches from MMMFs into savings, note I said ‘about’ that time. I should have been more precise, but the point should have been clear. The increase in saving over the past four years is part of a broader shift out of a number of less liquid assets into more liquid ones. I am not certain, but what happened to switch funds out of Retail MMMFs into savings in the early 2000s probably had something to do with heightened liquidity/safety demand then too. Risk spreads were still high through the period Retail MMMFs were shifting into saving deposits. The figure below suggests there is a pattern to this, but I would want to look closer before drawing any conclusions.
    http://research.stlouisfed.org/fred2/graph/?g=beE

    ReplyDelete
  15. the share of liquid assets for HH has grown inordinately—and remember the liquid portion has grown 25% over the past four years while the total assets has fallen 5%--and that this ratio is systematically related to NGDP. I have given you links to a paper that shows this.

    You keep insisting that the tail wags the dog -- namely that the share of liquid assets, which is never more than 1/6th or 1/7th the total stock of household assets is mainly responsible for the swings in the ratio of liquid/total assets.

    And of course, the ratio would be linked to NGDP, but not for the reason you think it is. When NGDP is at or close to a stable trend that allows for sufficient RGDP growth and avoids nominal income shocks (for instance, the Great Moderation), nominal asset values will grow faster than the nominal money stock (which is made up of deposits which expand as credit and NGDP expand.

    The ratio that you hold to be of great importance actually measures how fast nominal money deposits grow in relation to nominal asset values. In the 1970's, the ratio of safe assets was high relative to total assets because total assets grew slower than the NGDP and money stock variables (mostly likely because high inflation and steep MTR's eroded the after-tax value of capital). The ratio was not high because households had elevated demand for safe assets, which is what your model infers. In the 1980's, the ratio does not fall persistently because safe assets decline but because total assets grow much more rapidly than safe assets (also faster than the nominal money stock that creates deposits).

    NGDP currently is about 9% above pre-Great recession levels which explains some of the money stock growth and therefore growth in households money stock. However, total household asset values (the much larger denominator)are 5% below pre-crisis levels). In a typical economic expansion, nominal asset values would grow faster than the nominal money stock and push down the ratio of safe assets to total assets. The dog wags its tail.

    In normal conditions such record growth of debt would lead to higher interest rates, lower prices not the opposite.

    In our entire back and forth, I've agreed that NGDP trend growth is too low relative to pre-crisis trends and that low interest rates are a sure sign of such a condition. I'm am merely pointing out that your models do not show what you claim and are of no utility to supporting the claims you make.

    Regarding switches from MMMFs into savings, note I said ‘about’ that time. I should have been more precise, but the point should have been clear.

    Retail MMF balances did not decline until after the NGDP contraction -- your model says that those balances should have declined most rapidly during the contraction, as households supposedly rushed to convert MMF's into more liquid savings deposits or safer Treasury securities. There is no correlation that supports your argument on this point.



    ReplyDelete
  16. If the growth of safe assets (which constitute approx. 1/7th of total assets) is the key driver of the ratio of safe assets to total assets, then I guess it is plausible to claim that the Federal Reserve is responsible for low interest rates on Treasuries as they own a similar ratio of 1/7th of all U.S. Treasury debt.

    ReplyDelete
  17. TheNumeaire, you are confusing structural changes with cyclical changes in the ratio of liquid assets to total assets. I am talking about cyclical changes in it. Portfolios readjust based on economic expectations that change over the cycle. And this readjustment--the portfolio channel of monetary policy--has a huge bearing on nominal spening. All this ratio shows is an indication of those changes. And all I have claimed about it is that it shows an elevated demand for liquid assets for households alone.

    "In a typical economic expansion, nominal asset values would grow faster than the nominal money stock and push down the ratio of safe assets to total assets. The dog wags its tail."

    Yes, this ratio declines in expansions. But the reason it does is because money demand (i.e. safe asset demand) is declining. Changes in money demand precede changes in spending. If money demand remains elevated, then money demand remains depressed. This measure is a proxy for the intensity of money demand.

    I have attempted to answer your question, but it seems we are at an impasse. Consequently, we have to agree to disagree.

    ReplyDelete
    Replies
    1. Oops, this sentence should read " If money demand remains elevated, then nominal spending remains depressed."

      Delete
  18. David

    T-bills and T-bonds are safe assets, the safest one can find. Why isn't the answer to a safe asset shortage a retaliatory explosion of safe assets?

    Why should the Fed buy up T-bills and T-bonds from the market, thereby draining it of collateral? Isn't a better way that the federal government deficit spends (or cuts taxes), borrows t-bills and the Fed buys up these new t-bills to ensure that the interest burden is the absolute minimum? And, to the extent that asymmetric irredeemability matters and 'money' matters, you also have 'new money'.

    A helicopter drop, basically. Why QE and not helicopter drop?

    Your graphs show a lot. Indeed, the Fed doesn't really own much of the debt. But the debt itself has expanded substantially. So, if I was to take a 'market fiscalist' standpoint, saying that the expansion of safe assets has mitigated the deflationary spike in demand, but it has not been enough to keep nominal income on track. In other words, the Treasury caused the NGDP to fall. Where in this argument would you disagree with me?

    You have already moved beyond base money to safe assets. Now let's move form the central bank to the Treasury? Or at least the coordinated F-F-M (Fiscal Financial Monetary) complex of the government, as Buiter calls it.

    ReplyDelete
  19. Ritwik, I agree that the creation of safe assets via fiscal policy would be a fix. I have even said as much on earlier posts (here for example).

    But this doesn't mean traditional monetary policy can't work. The reason QE has not been successful is that it isn't expected to lead to a permanent increase in the monetary base (something a helicopter drop would do). If the Fed could change expectations that it would permanently increase the monetary base, then expected nominal incomes and the expected price level would be higher too. That would catalyze private safe asset creation today.

    I believe the Fed could change expectations about the future path of the monetary base via a NGDP level target.

    ReplyDelete
  20. David

    If the expected price level rises, does that catalyze safe asset creation, or reduce safe asset demand? I think your fellow MMers would agree with me that if anything, it's the latter.

    Plus, I am still confused how you're shifting the focus so easily from safe assets to the monetary base. If the problem is an excess demand (or short supply) of safe assets, then how does a permanent increase of the monetary base alone matter. Was the fall in NGDP indicative of the public's expectation of a permanent decline in the trend path of the monetary base? If not, then how does the monetary base matter.

    Where is the crux of the issue, as per you? Monetary base vs near-moneys, or near-moneys vs. non-moneys?

    ReplyDelete