Examining Bond Portfolio Trends: Fourth Quarter 2018
Beginning in May 2012 we started tracking portfolio trends of our bond accounting customers here at the Correspondent Division of CenterState Bank. At present, we account for over 140 client portfolios with a combined book value of $9.8 billion, or $71 million on average per portfolio. Twelve months earlier the average portfolio size was $61 million. Much of the increase in average size over the past year comes from a handful of larger portfolios added during 2018 rather than individual portfolio growth which has generally been static to slightly declining over the past year as loan demand bit into investment portfolio allocations.
For the eighth straight quarter yield behavior was characterized by curve flattening. The beginning of the quarter was marked by bearishness in October with the 10-year yield rising 8bps but in a fitful manner as questions started to arise about global growth, the impact of trade protectionism, and the cloud of a Fed rate hike on the horizon. The real action in the quarter, however, began in November as the bouts of equity selling and volatility accelerated and were fed by the Fed rate hike and policy statement on December 19th that was perceived as insufficiently concerned about the message being sent by financial markets. That led to 10-year yields falling from a quarterly high of 3.24% on November 7th to 2.69% as the year ended amidst heavy stock selling.
Meanwhile, short-end yields continued to move higher early in the quarter on the back of Fed rate-hiking expectations. The 2-year Treasury started the quarter yielding 2.82% and exhibited an upward bias but without strong conviction until November 8th. Then the aforementioned equity selling and overall market volatility drove the 2-year yield to 2.49% as the year ended. For the quarter, the two-year yield fell 33bps while the 10-year yield dipped 37bps. That action drove the 2yr-10yr Treasury spread from 24bps as the quarter began to 19bps as the year ended. The 2yr-10yr curve flattening reached a low of 11bps on December 19th, the day of the Fed rate hike. That spread represented an eleven and half-year low. The Fed now appears more cautious about causing an inverted curve and all that that implies vis a vis recession risks so a pause in hikes seems likely, at least for early 2019. With that backdrop, lets look at how portfolio allocations and performance have changed over the past year.
Let’s begin by revisiting allocations a year ago, December 31, 2017, as shown in the pie chart to the immediate right. The MBS/CMO sector comprised 39.4% of portfolios, municipal allocations stood at 28.1% and Agency/Treasury investments were at 24.1%, almost identical to year ago levels. Thus, the shift into municipal bonds that began in 2012 slowed to a standstill in 2017 as uncertainty over possible tax reform clouded the investment prospects for the sector.
Fast forward one year to December 31, 2018. The MBS/CMO sector comprised 38.3% of portfolios for a modest 1.1% decrease during the year as prepayments/maturities continued to be reinvested nearly dollar-for-dollar. Municipal allocations slipped to 26.7% (22.9% tax-free, 3.8% taxable) as the reduced corporate tax rate dimmed enthusiasm for the sector. Agency/Treasury investments picked up the slack in muni’s increasing from 24.1% a year ago to 27.5% at year-end. The “Other” category (corporates, CDs, and other floaters) dipped from 8.4% to 7.5% as CD investments and floaters dipped slightly after several years of growth as portfolio managers searched for short-duration yield.
In summary, portfolio allocations shifted slightly towards Agency/Treasury investments while the municipal and MBS/CMO sector saw slight outflows over the past twelve months. The “Other” category dipped as well with the drop occurring evenly between the CD and floater categories. In general, portfolio managers sought to maintain existing allocations with a slight decrease in enthusiasm for CD’s and muni’s. The decrease in muni investment was somewhat understandable given the corporate tax cut, even though the sector still provides some of the best yields for bank portfolios. The lack of a more dramatic change in allocation, however, indicates the moves were mostly tactical and not strategic changes in portfolio direction. With most banks experiencing strong loan demand in 2018, investment portfolio management seems to have been running mostly on autopilot during the year.
Now let’s look at portfolio performance trends. The graph below tracks average portfolio tax equivalent book yield, duration, unrealized gain/(loss) as a percent of book value, as well as 10-year Treasury yields and average portfolio size from March 2016 through December 2018.
As the above graph illustrates, tax equivalent book yields (red line) continue to be remarkably stable during the period shown with a generally slight uptick over the last three quarters. The average book yield was 2.67% at quarter-end which represents a negligible decrease of 1bp during the quarter but a 7bps increase in yield versus a year ago. With corporate tax cuts beginning in 2018,portfolios with moderate to heavy muni allocations saw tax-equivalent yields dip slightly in the first quarter but that drop of approximately 5bps was reversed in the second quarter. The third quarter built on second quarter yield gains as higher market rates were reflected in higher portfolio yields, especially in the MBS category. The fourth quarter book yield held up well despite the 37bps drop in market yields during the quarter.
Looking at portfolio duration (green line), the graph shows a decrease during the last three months dipping from 3.68 years in September to 3.28 years at year-end. As the entirety of the graph shows, durations have varied very little over the past twelve months with the December 2017 duration at 3.45 years. As previously illustrated in the pie charts, portfolio allocations have changed little in the past year so the decrease in duration during the quarter stems more from the decrease in market rates and subsequent increase in prepayment/call expectations given the lower rate environment. As shown, the 10-year Treasury began the quarter at 3.06% and finished the year at 2.69%. That 37bps decrease in yield increased prepayment and call expectations and that contributed as much as anything to the drop in duration from 3.68 years in September to 3.28 years at year-end.
Unrealized losses (blue line) improved during the quarter given the decrease in market rates during that time and the aforementioned decrease in portfolio duration. The year began with portfolios at an average unrealized loss of –0.76% of book value and the losses peaked at the end of the third quarter at –3.06% of book as market rates, driven by Fed rate hikes, inexorably increased through September. With yields moving decidedly lower in November and December unrealized losses improved from –3.06% when the quarter began to –1.86% at year-end, the best level since January.
Finally, purchases for 2018 focused on the MBS/CMO sector with 49% of total purchases (32% MBS, 17% CMO) and Agency/Treasury purchases totaling 32%. New muni purchases, however, were only 10% in 2018 compared to legacy muni allocations that comprised 27%. The average tax equivalent book yield on 2018 purchases was 3.10% with an average effective duration of 2.78 years. In summary, 2018 purchases ($1.8 billion in total par) were concentrated in the mortgage and Agency /Treasury sectors while purchased yields were 43bps more than existing portfolio averages. That yield advantage was also accomplished with less price risk as the average duration on 2018 purchases was a half-year shorter (2.78 years vs. 3.28 years) than the existing portfolio average.
We will update this data again in April to track how allocations and performance characteristics are trending in early 2019, a year that may represent something of a transition for the economy and hence for balance sheet management.
Thomas R. Fitzgerald
Director, Strategy & Research
400 Interstate North Parkway
Atlanta, GA 30339