Fixed income ETFs: when market structure affects pricing
ICE's Fair Value methodology helps asset managers meaningfully reduce their index tracking error.
Published
February 2024
Chief Product Officer
Senior Director, Quantitative Engineering
- Bond ETFs have emerged as a conduit for bringing the benefits of equity characteristics to fixed income markets - but with this hybrid structure, comes the challenge of accurate pricing
- On 5-6 days a year, bond markets close early while equities trade until 4 p.m. The early close for bonds translates to decreased trading activity after 1 p.m.
- Back-testing shows that utilizing ICE’s 4 p.m. Fair Value service for fixed income indices reduces tracking error by ~5 bps on average across all early close days for bond ETFs versus the traditional early close snapshot at 1 p.m.
- For the final trading day of 2023, the realized improvement is ~13bps for a large US Corporate ETF
Fixed income ETFs (exchange traded funds) have flourished in popularity, as instruments which provide transparent, liquid access to an asset class once traded only by appointment over the counter. By providing broad access to fixed income, ETFs are also influencing bond market structure: they stimulate higher trading volumes and support greater sophistication. Yet straddling markets with distinct characteristics - as bond funds which trade on equity markets - can challenge one of the basic premises of an ETF: to accurately track an underlying index.
A key difference in bond and equity market structure is their hours of operation. Bond market holidays and early closes are recommended by SIFMA (The Securities Industry and Financial Markets Association) while equity market holidays and early closes are established by the exchanges. For the most part, there is a high degree of overlap, however, on ~5-6 days a year, the bond market closes early while equities continue to trade until 4 p.m. The early close for bonds translates to decreased trading activity after 1 p.m., as demonstrated by trade counts reported to FINRA TRACE in Figure 1.
Figure 1: Trade counts by hour
Figure 1: Count of trades reported to FINRA TRACE on the last trading day of the year. Only trades with size above 50k are included (Source: FINRA TRACE)
By incorporating ICE’s fair value methodology* in fixed income indices – which utilizes proprietary models to estimate bond market movement between 1 p.m. and 4 p.m. - offers a way to address some of this discrepancy, helping asset managers to meaningfully reduce their tracking error.
As ETFs gain popularity, their tracking error is more notable on days where the bond market closes early, but the equity markets close at 4 p.m. ET, with tracking errors in December notably higher on early close days versus other months.
The industry has traditionally applied an early close snapshot of bond prices at 1 p.m ET. The problem with this approach is that on the last trading day of the year, markets were more volatile after 1 p.m. than before, this directly impacts fixed income ETFs as they trade until the equity market close at 4 p.m. As indicated in figure 3, the magnitude of the differences is only getting larger over time.
Figure 2 shows TLT (iShares Trust 20+ year Treasury ETF) compared to ZB H24 (the 30yr Treasury bond Future) on December 29, 2023. Note the drop at 1 p.m. and the subsequent volume seen by the ETF.
Figure 2: iShares Trust Barclays 20+ Treasury (TLT) vs 30yr Treasury Futures
Figure 2: Market movements in TLT (iShares 20+ Year Treasury Bond ETF) and ZB H24 (Future on the 30 yr Treasury bond deliverable in Mar 2024). Traded volumes (for TLT) are also shown on the lower panel (Source: ICE Connect and the CME)
To address this discrepancy, ICE Data Indices has applied ICE’s Fair Value Pricing service across its indices since December 22, 2023. Calculated by ICE Data Pricing & Reference Data LLC (PRD) this service uses proprietary models to estimate bond market movement between 1 p.m. and 4 p.m. based on a multifactor regression process. This process seeks to determine the relationship between historical movements in certain financial instruments (including Treasury futures) or “factors”, available during U.S. market hours and the underlying security. Figure 3 shows the market movements for each of the last trading days of the year for the past 5 years. Note the relative absence of market moves for 2019 and 2020, the larger move in 2021 and the largest ones in 2022 and 2023.
Figure 3: 30yr Treasury Futures change - last day of year
Figure 3: 30yr Treasury Futures changes, highlighting the volatility at 1 p.m. and movement to 4 p.m., on the last day of trading over the last five years. The lower panel shows the average transacted volume across the years (Source: CME)
Our analysis of bond ETF tracking error on December 29, 2023 (the last trading day of the year) includes five corporate investment grade ETFs and seven Treasury ETFs with significant assets under management that are benchmarked to ICE Indices. These results are presented in Figure 4.
Figure 4: ETF vs Index Comparison
| ETF | ETF Name | Index | Index Name |
|---|---|---|---|
| USIG | iShares Broad USD Investment Grade Corporate Bond ETF (USIG ETF) | C0A0 | ICE BofA US Corporate Index |
| CORP | PIMCO Investment Grade Corporate Bond Index ETF (CORP ETF) | C0A0 | ICE BofA US Corporate Index |
| IGSB | iShares 1-5 Year Investment Grade Corporate Bond ETF (IGSB ETF) | CVA0 | ICE BofA 1-5 Year US Corporate Index |
| IGIB | iShares 5-10 Year Investment Grade Corporate Bond ETF (IGIB ETF) | C6A0 | ICE BofA 5-10 Year US Corporate Index |
| IGLB | iShares 10+ Year US Investment Grade Corporate Bond ETF (IGLB ETF) | C9A0 | ICE BofA 10+ Year US Corporate Index |
| SHY | iShares 1-3 Year Treasury Bond ETF (SHY ETF) | IDCOT1 | ICE US Treasury 1-3 Year Bond Index |
| IEI | iShares 3-7 Year Treasury Bond ETF (IEI ETF) | IDCOT3 | ICE US Treasury 3-7 Year Bond Index |
| GOVT | iShares US Treasury Bond ETF (GOVT ETF) | IDCOTC | ICE US Treasury Core Bond Index |
| IEF | iShares 7-10 Year Treasury Bond ETF (IEF ETF) | IDCOT7 | ICE US Treasury 7-10 Year Bond Index |
| TLT | iShares 20+ Year Treasury Bond ETF (TLT ETF) | IDCOT20 | ICE US Treasury 20+ Year Bond Index |
Figure 4: Tracking error of ETFs benchmarked to ICE Indices. Blue bars represent tracking error of the ETF vs 4 p.m. FV index. Yellow bar represents tracking error of ETF vs equivalent 1 p.m. index. Source: ICE Data Services
As Figure 4 demonstrates, using the example of USIG (iShares US Investment Grade ETF) we see that utilizing the 4 p.m. Fair Value service aligns index pricing time with ETF net asset value time and reduces tracking error to the underlying ETF by 13bps for December 29, 2023. This difference is higher, on the order of 40 bps for TLT (iShares 20+ Year Treasury Bond ETF).
Figure 5: Performance comparison for three ETFs benchmarked to different indexes. Source: Factset
To further highlight the impact of benchmark tracking error, in figure 5, we compared TLT ETF (iShares 20+ Year Treasury Bond ETF) against two similar ETFs, SPTL (SPDR Portfolio Long Term Treasury ETF) and VGLT (Vanguard Long Term Treasury Index Fund). These ETFs are benchmarked to different indices – TLT is benchmarked against ICE US Treasury 20+ Year Bond Index, whereas SPTL and VGLT are benchmarked other index providers.
While all three of these ETFs have similar performance as their benchmarks for 2023 before the last day (through December 28, 2023), on the last day of the year, SPTL and VGLT underperform their benchmarks by about 40bps while TLT maintains its performance in-line with its benchmark.
As ICE continues to expand its global index offering, the application of Fair Value pricing across the relevant product suite is another tool which asset managers can use to retain investor confidence – and reduce tracking error and cost – amid a backdrop of stiff competition in the ETF sector.
* References to ICE’s Fair Value methodology are to ICE Data Pricing & Reference Data early bond market close days evaluated pricing methodology. The early bond market close days evaluated pricing methodology leverages ICE Data Pricing & Reference Data Fair Value Information Service methodology, to calculate evaluated pricing on early bond market close days.