in the short interest, is the fact that growth in assets committed
to ETFs reflects an entirely different process than growth in assets committed
to conventional mutual funds. With trivial exceptions, it is not
common practice to sell shares in conventional mutual funds short.
If creation of ETF shares to lend for a short sale is fully replaced by borrowing
from traditional investors, each share sold short supplies an
additional long share that appears in some investor’s account but does
not increase the fund’s shares outstanding. ETF shares credited to investors’
accounts consist of the total fund shares outstanding on the fund’s
books plus the short interest. The short selling mechanism leads to more
ETF shares “owned” in shareholder accounts than there are shares outstanding.
This phenomenon merits careful consideration by all ETF
users. If you do not understand this, reread the prior three paragraphs
until you do.
The most widely circulated data on ETF assets focuses on the current
market value of each fund’s recorded outstanding shares.18 This
weekly ETF market report places no emphasis on changes in the number
of shares outstanding in each ETF. Investors looking at this report perceive
growth or decline in the value of ETF portfolios as more a function
of market price changes in underlying portfolios than of net
investment or disinvestment by fund share holders.
Aggregate ETF net investment and redemption data reflecting the
value of changes in shares outstanding is published monthly by the
Investment Company Institute (ICI). These reports translate share
changes into net purchases and sales at the prices of the actual purchases
(creations) and sales (redemptions). The ICI data compilation
shows and prices the changes in shares outstanding appropriately, but it
cannot take into account the fact that changes in an ETF’s short interest
substitute for shares purchased or redeemed with the fund.19 In Exhibit
4.3, we average the sums of the shares outstanding and the short interest
for each of the 10 largest equity funds for the middle of the months
December 2002 and January 2003 and compare that average with the
same data for the middle of December 2003 and January 2004.20 Three
of the larger ETFs, particularly the S&P 500 SPDRs and the QQQs,
experienced declines in shares outstanding from the end of 2002
through 2003. When the general increase in ETF short interests over
this period is added to shares outstanding, only the S&P MidCap SPDR
showed a decline in its net share position—a very small one—over that
interval. If the reduction in shares outstanding was due primarily to
market makers withdrawing from ETF share lending, substantial net
ETF purchases by the public have been accommodated by short sellers
and, hence, net long ETF investment has been much more robust in
2003 than some observers have suggested.
Exhibit 4.4 illustrates how changes in an ETF’s short interest can
distort interpretations of that ETF’s popularity with (long) investors. A
number of analysts have noted that the third largest U.S. ETF, the
iShares S&P 500 fund, enjoyed an increase in shares outstanding in
2003 while the older and massively larger S&P 500 SPDR had fewer
shares outstanding at the end of 2003 than at the beginning. Exhibit 4.4
shows what happens when we add the short interest to the outstanding
shares of each of the funds. The 2003 increase in shares held by long
investors in SPDRs was still less than the net increase in long positions
in the iShares S&P 500 fund. The SPDR, held long did increase, however.
From an analytical perspective, the large and fluctuating size of
many ETF short-interest positions, year-end tax motivated transactions
by dealers, and uncertainty about where the shares sold short are borrowed
make any statement about short-term changes in investor interest
in ETFs of dubious validity.
Read More : What Is The Significance Of The Short Interest For Growth In ETF Assets?
