BlackRock has hit back at suggestions from the Bank for International Settlements that bond exchange traded funds might have short-changed market makers during the March 2020 market sell-off.

The BIS paper suggested that large price dislocations opened up last year between bond ETFs and the prices of their underlying securities because ETF managers deliberately handed authorised participants baskets of low-quality bonds. This would have discouraged underlying trading and jammed the arbitrage mechanism designed to bring the share price of an ETF back into line with its net asset value.

But in a report released on Monday, the world’s largest asset manager says the data it has compiled prove it did not do this, and that these so-called redemption baskets contained a mix of securities that fairly reflected the underlying holdings of each ETF.

APs and market makers are the engines for ETFs, not only providing liquidity but also performing an intermediary role — accepting baskets of securities from the ETF provider and returning them to the provider on a daily basis, trading in the ETF’s underlying securities and controlling the amount of ETF shares in existence by creating and redeeming them. This process is supposed to keep ETF prices close to the underlying net asset value of the securities they own, but in March last year there was a huge price dislocation.

BlackRock’s analysis shows that between February 19 and March 12, the worst day of the crisis, 95 per cent of the 1,000-plus securities held by the $21.8bn iShares iBoxx $ High Yield Corporate Bond ETF (HYG) appeared in the redemption baskets received by APs at some point.

Likewise, 67 per cent of the 2,000 or so credits held by the $41.6bn iShares iBoxx $ Investment Grade Corporate Bond ETF (LQD) appeared in the baskets. The figures for a suite of five European-domiciled fixed-income ETFs ranged from 80-89 per cent, except for one corporate bond fund that actually saw net inflows during the crisis period.

BlackRock argued that these figures show the baskets were highly diversified during the sell-off, leaving little wriggle room for it to have stuffed the redemption baskets with disproportionately lower quality securities, even if had been inclined to do so.

Unlike those for equity-based ETFs, redemption baskets for bond ETFs typically do not contain every security as the relatively high minimum trading size of bonds means it can be impractical to hand an AP a tiny sliver of potentially hundreds or even thousands of credits.

There has been scrutiny of how bond ETFs traded during the Covid-driven market sell-off in March-April 2020. Large parts of the $1.2tn fixed-income ETF universe plunged to unprecedented discounts to net asset value, with Vanguard’s $55bn Total Bond Market ETF (BND) and BlackRock’s $71bn iShares Core US Aggregate Bond ETF (AGG) closing at discounts of 6.2 per cent and 4.4 per cent respectively on March 12.

An array of other junk bond, municipal debt, bank loan and US government bond ETFs also traded at extraordinary discounts, even though ETFs are purposely designed to closely track the value of their underlying holdings.

Further data from BlackRock on the average duration, bid-offer spread and liquidity of the contents of the redemption basket, vis-à-vis the ETFs themselves, show relatively little difference between the two, although in some cases the spreads were a fraction wider for the basket securities.

“The [BIS] paper suggested some issuers would try to deliver more concentrated baskets. It was actually the opposite,” said Samara Cohen, co-head of iShares markets and investments, of a period she described as “the mother of all stressed bond markets”.

“In a volatile market with a lot of difficulty around price transparency, the best way to make sure we did deliver out a representative basket was to widen the basket,” said Cohen. “A more diversified basket was also what the market maker community wanted. It relieved them from having to sell any concentrated positions.”

Moreover, the higher the level of redemptions, the more necessary it is to broaden redemption baskets in order to avoid an ETF becoming increasingly skewed relative to its underlying index, she added.

The BIS analysis covered a longer time period than the BlackRock study, without an artificial March 12 cut-off. It also scrutinised a broader range of subsectors, including Treasury bond funds, and a wider range of ETF providers.

In contrast to the BIS study, Cohen said it was the discounted price of many bond ETFs that was truly reflective of the market’s perceptions of fair value during the crisis, not the prices of the underlying bonds used to calculate the fund’s NAV.

BlackRock said, for example, that on March 12 last year, at the height of the market storm, LQD traded almost 90,000 times, compared with an average of just 37 times each for the ETF’s five largest holdings.

Throughout March, more than half of LQD’s holdings traded between zero and five times a day, on average, it added.

“Price formation based on tens of thousands of trades is likely to be more informative than price formation based on dramatically fewer trades in the underlying bonds,” said Cohen.

“We have never published this data before,” Cohen added. “It’s a new dimension to what has become a really considerable body of literature around ETFs.”