1
What You Need to Know: ETFs Answered Liquidity Challenges in 2020
The COVID-19 related sell-off in corporate bonds in the first quarter 2020 caused some commentators to express concern about the liquidity and functionality of fixed income exchange traded funds (ETFs) after a financial or economic shock – in particular, corporate bond ETFs.1
Widening discounts between corporate bond ETFs and net asset value (NAV) pricing after COVID-19 was cited as further evidence of ETF illiquidity. But ETF NAVs use estimated prices for underlying cash bonds, not market trades, and may quickly become stale. In contrast, ETFs trade on exchange daily, and tend to rapidly move to new equilibrium pricing in the face of a shock such as COVID-19. This makes ETF pricing timelier than NAV calculations, as the Bank of England pointed out.2
There is scant evidence the liquidity infrastructure around ETFs, as driven by authorized participants (APs}, failed during the March/April 2020 period - and trading spreads were much narrower in some fixed income ETFs than the underlying bonds.3
Fears around bond ETFs deepened after COVID crisis.
The chart below shows the extent of significant corporate bond spread widening and market dislocation in Q1 2020, in both investment grade and high yield corporate bonds, in the U.S. and Eurozone.
Concerns about liquidity and functionality of bond ETFs are driven by the perceived decline in “market liquidity” in corporate bonds since the global financial crisis (GFC), combined with the growth in assets in fixed income ETFs. The main argument cited has been that corporate bond ETFs are built from corporate cash bonds, which trade OTC in a fragmented market with severely diminished liquidity since the GFC. Hence, it is argued, after a credit event, or market shock, ETFs will become "untradeable" with no liquidity, since if all investors seek to sell at the same time, the underlying securities that compose them will be untradeable. The widening discount between ETF pricing and NAV is also cited as a failure of the arbitrage mechanism designed to keep the ETF price in line with its NAV.
Evidence for this claim is cited as the discount that can appear in bond ETF prices to the NAV of these funds, based on the price of the underlying cash bonds, during periods of market stress. The related assumption is that discounts to NAV on this scale question the validity of bond ETFs as investment vehicles, since ETFs in other, more liquid, asset classes (such as government bonds or equities) have tended to trade much closer to NAV. This discount reached about 5% in some corporate bond ETFs during the March COVID-19 crisis.4
Insight: ETF NAVs are not based on intra-day trading prices.
It is noteworthy that NAV pricing for bond ETFs is often based on estimated prices for underlying cash bonds, as trading data may not always be available. Underlying cash bonds may not trade at all on some days, reflecting the fragmented market and diminished market-making since the GFC (it has been estimated the inventory now held by dealer-brokers in corporate bonds is currently less than $60 billion, compared with $418 billion of the inventory held pre-GFC).5 Nor is there an “official” price for these corporate bonds, with no exchange trading. In the absence of trading data, NAV prices are based on estimates from pricing services.
ETF liquidity is not determined solely by the underlying instruments.
Bond ETF “liquidity” is distinct from the “market liquidity” of underlying instruments. The supply of bond ETF shares can be varied by “authorized participants” or APs – often banks, or institutional investors – who can create or redeem ETF shares in a bond ETF in response to changes in demand for the ETF. This is known as primary liquidity in the bond ETF. Secondary, or on-screen, liquidity is the trading of ETF units that already exist, which drives pricing data, volumes, etc. Overall, liquidity in the ETF will be driven by primary and secondary market liquidity.
Some bond ETFs, such as high yield, have underlying assets with poor primary liquidity, but this does not mean the ETFs become impossibly illiquid in the secondary market after a market shock. The purpose of an AP is to act as an arbitrageur and liquidity buffer between investors and the ETF provider, creating extra ETF units when demand is strong, and redeeming units when demand is weak. In a study conducted before the COVID-19 crisis, and focused on previous periods of credit market stress, the UK’s Financial Conduct Authority found fixed income ETF liquidity in Europe to be resilient6 and that lower activity APs acted as alternative liquidity providers, when large ETF discounts appeared relative to NAV.
Authorized participants act as arbitrageurs and market-makers, but there has been no stepping away.
A related concern about bond ETFs is “step-away” risk, where all APs in an ETF step away simultaneously from their arbitrage role in a highly stressed market. The U.S. Securities and Exchange Commission found no evidence in the GFC or subsequent stressed market events of this occurring. Indeed, there is evidence in a recent BlackRock report that over 60 asset owners and managers entered the market for fixed income ETFs for the first time in the first half of 2020.7
Bond ETFs with illiquid underlying securities, such as high yield ETFs, are more likely to trade at wider discounts and premiums to NAV than ETFs that hold government bonds during severe market stress because liquidity in the underlying securities is generally more reliable. This is often described as a “liquidity mismatch” problem. It arises because it may take APs longer to buy the underlying high yield cash bonds to create new ETF units in a rising market, and to sell the underlying cash bonds to redeem the ETF units sold in a falling market, after a shock like COVID-19. There is no evidence that the arbitrage mechanism failed in March 2020. On the contrary, there is evidence of net creation of units by APs during this period, and not redemptions, in some corporate ETFs.8
ETF pricing can be a useful guide to new equilibrium pricing on underlying bonds.
Sizeable discounts and premiums in ETF pricing relative to NAV can be a useful guide to the value of the underlying bonds, as the Bank of England concluded after the Q1 2020 sell-off.9 Faster price discovery in the ETF should not be confused with mispricing. Instead, this may be evidence ETF pricing has moved to the new, and lower, equilibrium pricing more rapidly, increasing price efficiency.
There are two main ways volatility can lead to NAVs that are less representative of “real” prices: One, yesterday’s closing price will be farther from today’s price if vol is higher. So stale prices are more inaccurate than normal; and two, in highly volatile markets, even fewer of the underlying bonds trade on any given day, so more of the NAV price is made up of estimates vs. hard trading data.
Surge in ETF trading volumes suggests bond ETF liquidity held up well, and trading spreads in many flagship ETFs widened far less than underlying cash bonds.
Trading volumes in both investment grade and high yield bond ETFs confirm there was no suggestion of a freeze in the secondary ETF market in Q1 2020, as the chart below shows.
Some investment grade and high yield ETFs showed volumes more than doubling on a daily basis over March. Similarly, although trading spreads in ETFs widened, iShares flagship ETFs moved far less than spreads in underlying cash bonds, as seen in the chart below.
This is further evidence the liquidity infrastructure around bond ETFs held up well during the crisis, and confirms the diminished liquidity and wide trading spreads in underlying cash bonds. A useful metric for gauging liquidity costs is price liquidity ratio (PLR), which looks at market impact and measures the movement in price of a security for an executed trade of a given size. A higher PLR represents a larger movement in price for a given trade size and therefore shows lower liquidity. The FTSE Russell market price/liquidity ratio in corporate bonds10 shows a pronounced deterioration during the March/April 2020 period, both in the March sell-off, and during the rally in April/May, after the Fed announced the broadening of its QE program to include corporate bonds and high yield ETFs.
Widening discounts between corporate bond ETFs and net asset value (NAV) pricing after COVID-19 was cited as further evidence of ETF illiquidity. But ETF NAVs use estimated prices for underlying cash bonds, not market trades, and may quickly become stale. In contrast, ETFs trade on exchange daily, and tend to rapidly move to new equilibrium pricing in the face of a shock such as COVID-19. This makes ETF pricing timelier than NAV calculations, as the Bank of England pointed out.2
There is scant evidence the liquidity infrastructure around ETFs, as driven by authorized participants (APs}, failed during the March/April 2020 period - and trading spreads were much narrower in some fixed income ETFs than the underlying bonds.3
Fears around bond ETFs deepened after COVID crisis.
The chart below shows the extent of significant corporate bond spread widening and market dislocation in Q1 2020, in both investment grade and high yield corporate bonds, in the U.S. and Eurozone.
Concerns about liquidity and functionality of bond ETFs are driven by the perceived decline in “market liquidity” in corporate bonds since the global financial crisis (GFC), combined with the growth in assets in fixed income ETFs. The main argument cited has been that corporate bond ETFs are built from corporate cash bonds, which trade OTC in a fragmented market with severely diminished liquidity since the GFC. Hence, it is argued, after a credit event, or market shock, ETFs will become "untradeable" with no liquidity, since if all investors seek to sell at the same time, the underlying securities that compose them will be untradeable. The widening discount between ETF pricing and NAV is also cited as a failure of the arbitrage mechanism designed to keep the ETF price in line with its NAV.
Evidence for this claim is cited as the discount that can appear in bond ETF prices to the NAV of these funds, based on the price of the underlying cash bonds, during periods of market stress. The related assumption is that discounts to NAV on this scale question the validity of bond ETFs as investment vehicles, since ETFs in other, more liquid, asset classes (such as government bonds or equities) have tended to trade much closer to NAV. This discount reached about 5% in some corporate bond ETFs during the March COVID-19 crisis.4
Insight: ETF NAVs are not based on intra-day trading prices.
It is noteworthy that NAV pricing for bond ETFs is often based on estimated prices for underlying cash bonds, as trading data may not always be available. Underlying cash bonds may not trade at all on some days, reflecting the fragmented market and diminished market-making since the GFC (it has been estimated the inventory now held by dealer-brokers in corporate bonds is currently less than $60 billion, compared with $418 billion of the inventory held pre-GFC).5 Nor is there an “official” price for these corporate bonds, with no exchange trading. In the absence of trading data, NAV prices are based on estimates from pricing services.
ETF liquidity is not determined solely by the underlying instruments.
Bond ETF “liquidity” is distinct from the “market liquidity” of underlying instruments. The supply of bond ETF shares can be varied by “authorized participants” or APs – often banks, or institutional investors – who can create or redeem ETF shares in a bond ETF in response to changes in demand for the ETF. This is known as primary liquidity in the bond ETF. Secondary, or on-screen, liquidity is the trading of ETF units that already exist, which drives pricing data, volumes, etc. Overall, liquidity in the ETF will be driven by primary and secondary market liquidity.
Some bond ETFs, such as high yield, have underlying assets with poor primary liquidity, but this does not mean the ETFs become impossibly illiquid in the secondary market after a market shock. The purpose of an AP is to act as an arbitrageur and liquidity buffer between investors and the ETF provider, creating extra ETF units when demand is strong, and redeeming units when demand is weak. In a study conducted before the COVID-19 crisis, and focused on previous periods of credit market stress, the UK’s Financial Conduct Authority found fixed income ETF liquidity in Europe to be resilient6 and that lower activity APs acted as alternative liquidity providers, when large ETF discounts appeared relative to NAV.
Authorized participants act as arbitrageurs and market-makers, but there has been no stepping away.
A related concern about bond ETFs is “step-away” risk, where all APs in an ETF step away simultaneously from their arbitrage role in a highly stressed market. The U.S. Securities and Exchange Commission found no evidence in the GFC or subsequent stressed market events of this occurring. Indeed, there is evidence in a recent BlackRock report that over 60 asset owners and managers entered the market for fixed income ETFs for the first time in the first half of 2020.7
Bond ETFs with illiquid underlying securities, such as high yield ETFs, are more likely to trade at wider discounts and premiums to NAV than ETFs that hold government bonds during severe market stress because liquidity in the underlying securities is generally more reliable. This is often described as a “liquidity mismatch” problem. It arises because it may take APs longer to buy the underlying high yield cash bonds to create new ETF units in a rising market, and to sell the underlying cash bonds to redeem the ETF units sold in a falling market, after a shock like COVID-19. There is no evidence that the arbitrage mechanism failed in March 2020. On the contrary, there is evidence of net creation of units by APs during this period, and not redemptions, in some corporate ETFs.8
ETF pricing can be a useful guide to new equilibrium pricing on underlying bonds.
Sizeable discounts and premiums in ETF pricing relative to NAV can be a useful guide to the value of the underlying bonds, as the Bank of England concluded after the Q1 2020 sell-off.9 Faster price discovery in the ETF should not be confused with mispricing. Instead, this may be evidence ETF pricing has moved to the new, and lower, equilibrium pricing more rapidly, increasing price efficiency.
There are two main ways volatility can lead to NAVs that are less representative of “real” prices: One, yesterday’s closing price will be farther from today’s price if vol is higher. So stale prices are more inaccurate than normal; and two, in highly volatile markets, even fewer of the underlying bonds trade on any given day, so more of the NAV price is made up of estimates vs. hard trading data.
Surge in ETF trading volumes suggests bond ETF liquidity held up well, and trading spreads in many flagship ETFs widened far less than underlying cash bonds.
Trading volumes in both investment grade and high yield bond ETFs confirm there was no suggestion of a freeze in the secondary ETF market in Q1 2020, as the chart below shows.
Some investment grade and high yield ETFs showed volumes more than doubling on a daily basis over March. Similarly, although trading spreads in ETFs widened, iShares flagship ETFs moved far less than spreads in underlying cash bonds, as seen in the chart below.
This is further evidence the liquidity infrastructure around bond ETFs held up well during the crisis, and confirms the diminished liquidity and wide trading spreads in underlying cash bonds. A useful metric for gauging liquidity costs is price liquidity ratio (PLR), which looks at market impact and measures the movement in price of a security for an executed trade of a given size. A higher PLR represents a larger movement in price for a given trade size and therefore shows lower liquidity. The FTSE Russell market price/liquidity ratio in corporate bonds10 shows a pronounced deterioration during the March/April 2020 period, both in the March sell-off, and during the rally in April/May, after the Fed announced the broadening of its QE program to include corporate bonds and high yield ETFs.
