FROM THE OFFICE OF PUBLIC AFFAIRS

The Treasury issues marketable securities on a regular and predicable schedule in order to capture a liquidity premium from investors and, thereby, achieve lower borrowing costs over time.  To do this we need to issue enough but not too much in each of our auctions.  This is easy to say but difficult to accomplish.  How to issue “enough but not too much” in each auction across the yield curve is the greatest challenge we face in managing the government’s debt. 

 A year ago I spoke to you about the challenge we face in managing our marketable debt on a regular and predictable basis in a changing world.  My message was that a schedule of regular and predictable auctions has never – and could never – mean that debt management practices do not change.  Our mix of instruments, our borrowing requirements and our maturity structure are always changing. 

 Today I will focus on the challenge of issuing “enough but not too much” at each auction in the face of continuous change in our borrowing requirements.  But first, let me begin at the beginning.

 The objective of federal debt management is to meet the government’s financing needs at the lowest cost over time.  The dominant constraint we face is that we see the future imperfectly; we must always make decisions in conditions of uncertainty about the future path of our financing needs.

 Our MO – our modus operandi – is that we issue securities on a regular and predictable schedule and we try to limit changes in auction sizes.  Investors and dealers rely upon routine availability of Treasury securities both as a source for constant duration matching and as a liquid vehicle for financial intermediation.  As a consequence, they tend to pay a slight premium for our newly issued securities.  By capturing this premium, we lower our borrowing costs. 

The regularity and predictability of our auction cycle also supports the liquidity of our secondary market which, in turn, supports demand for our auctions.

 Changes in debt management are regularly and predictably announced at our quarterly refundings.  We strive to make our actions transparent and understandable.  Ex ante market participants may not be able to predict exactly what changes we will make but ex post market participants should be able to understand our actions in light of our objective (lowest cost over time) and our constraints (uncertain financing needs).

 For all this to work in practice, we need to issue “enough but not too much” at each auction.  Neither is a concrete limit.   Too little and we cannot sustain a deep and liquid secondary market for our securities.   Too much and we create concern among primary market participants that they may find it difficult to distribute their holdings in the secondary market.  Matching these market size constraints with our aggregate borrowing needs is surprisingly difficult.

 First, economic and financial conditions are always changing the demand for our instruments so that the apparent optimal auction size at any one maturity point, in any one year, may not be the same in another year.  Second, our forecasts are less than perfect so we must always somehow distribute the deviations from forecasts across our maturity points.  Finally, in adjusting auction sizes over time we must manage both levels and rates of change for each maturity point: we want to issue enough but not too much at each point and we want to limit the rate of change of auction amounts at each maturity point as well.

  So how do we know whether we are issuing “enough but not too much” at each maturity point?  How do we know that we are doing our best to capture the liquidity premium at each and at all of our auctions?  How do we know whether, in the face of changing financing needs, we are adjusting our maturity profile too quickly or too slowly? 

 Unfortunately, we don’t – or, more precisely, I am not yet satisfied that we do.  But we are working hard to articulate better measures of our performance.

 A clearer, more complete understanding of auction performance, and of the intersection of primary and secondary market liquidity, will be the most direct measure of whether we are offering enough but not too much in each auction.   A fuller picture of our maturity profile, and its changes across time, will provide us a better gauge of whether we are adjusting the number, range and size of our offerings too quickly or too slowly.

 At our last quarterly refunding, we released a number of new charts and we discussed these charts with the Treasury Borrowing Advisory Committee.  These included some unconventional measures of both auction outcomes and our maturity profile.

 For auction performance, I am convinced that we need to look beyond the traditional measures such as bid-to-cover.  Bid-to-cover reflects the primary dealers’ commitment to meet their role as counterparties to the Federal Reserve Bank in addition to the underlying demand for Treasuries.  We need to develop measures that tell us more about that underlying demand because this ultimately determines whether we are meeting our objective of lowest cost financing over time.  We also need to develop measures that will help us better understand the relationship between the primary market and the secondary market.  We need to measure the range of auction outcomes for each maturity point as they accumulate over time, so that we can observe in as many ways as possible the market’s assessment of whether we are issuing enough but not too much.

 The average maturity of our total outstanding debt may be of interest to historians and economists but it is not a particularly useful metric for the debt manager.  It moves so slowly it’s almost inert.  Having served as the Under Secretary for almost two years now, I am now responsible for roughly 2/28ths of the maturity structure of the federal marketable debt.  (Had we not suspended the long bond, I would only be responsible for 2/30ths.) 

 At the other end of the spectrum, there is much (understandable) market attention to the nominal size of each auction we hold, relative to recent previous auctions.  But this has a different kind of inertia.  We can easily see the nominal changes in each auction size from quarter to quarter so movements in these dollar amounts tends to dominate market attention and our refinancing decisions.  What is harder to see, but much more important, is the future sustainability of a given maturity profile in light of our (albeit imperfect) forecast financing requirements.

 In one of the charts released at the refunding, we introduced the concept of “constant issuance maturity” which depicts for each quarter what that quarter’s issuance pattern would have produced as an average maturity if it were to have been sustained for ten years.  Given the imprecision of our forecasts, this is not a realistic hypothesis.  But it is interesting to see whether the issuance pattern itself would have been sustainable. 

 The surprising result is that this measure of our issuance pattern has been highly volatile over time, abruptly swinging long and short.  This suggests that on a number of occasions our issuance pattern itself has not been sustainable, setting up the need for subsequent changes and, thus, the series’ volatility.

 It is entirely to be expected, and appropriate, that variance in revenues (positive or negative) will first be absorbed by changes in bill issuance.  We auction bills every week and coupon instruments only monthly and quarterly.  But eventually, decisions must be made to distribute changes in our financing needs across the curve – consistent with our aim to issue enough but not too much at each of our different maturities. 

 In order to assess whether we are adjusting our maturity profile too quickly or too slowly, we need a better understanding of the sustainability of our maturity distribution.  A concept like constant issuance maturity may provide just such a measure.

 I am frequently asked why we don’t renew issuance of the long-bond in order to “lock in” these currently “low” rates.  When I am asked this question, I know that I am talking to someone who does not understand the Treasury’s debt management strategy.  I also realize that I must also be talking to someone who has exceptional confidence in the accuracy of their ten-year forward forecast of ten and thirty year rates.  Does anyone actually think that a small group of Treasury officials is better than the market at forecasting long term interest rates?  Regular and predictable issuance assures the market that we will not make the mistake of trying. 

So whenever I hear the idea of our “locking in” low, long-term rates I know that it must be time to give another speech about our pattern of regular and predictable issuance. 

 We give up the opportunity to time the market so that we can capture the liquidity premium in all of our auctions.  If we were to try to time the market we would introduce much greater uncertainty to our borrowing pattern, raising the risk premium that we would have to pay investors and losing the liquidity premium we now capture as a cost saving.

 If the United States Treasury were actually a market timer we would do great damage to the efficiency of our capital markets.  We would raise our cost of borrowing, and potentially the cost of borrowing for others.  We do make changes in our borrowing pattern, but they are not a function of our expectation for shifts in the shape of the yield curve.  Rather, the changes we make are a consequence of the changes in our expected financing needs and changes in our understanding of how best to achieve the lowest borrowing costs over time.

 As changes in our borrowing pattern become necessary we will continue to announce them at our regular and predictable quarterly refundings.  We will hold our auctions on a regular and predictable schedule.  We will strive to issue in amounts that are enough to sustain the liquidity of our secondary market but not too much to eliminate the on-the-run premium.  We will also continue to look for better measures of our own performance and to learn from our past actions. 

 Setting debt management on a path of continuous improvement, while we strive to be regular, predictable and transparent in the implementation of changes to our practices, is the only way we will achieve our objective of lowest cost financing over time.