Above is a total return graph for JPC, Nuveen's High yield convert ETF, from 2007 (ie, 1/1/2007) to present. Note the return is basically zero. I pulled 39 High Yield ETFs, and their average return from 2007 to present has been about 21%, and is the same whether I equal-weight or value weight them, and note these have a survivorship bias in them, in that I am not using all extant funds from 2007. The JPC benchmark, the Bloomberg Active Index for Corporate Preferred funds generates a similar 20% return over that period.
In contrast, the Merrill High Yield Master II shows a whopping 43% return over this period, twice what bonds with that moniker actually generated. The Merrill index is not tradeable, which is very important. The bid-ask spreads in junk bonds are very high, often 5 points, and indices don't internalize these. Also, if you are big, there's trade impact. Academics might look at the Merrill index to estimate the return to junk bonds, but there's a 22% difference here, and that excludes all the survivorship biases in my comparison to actual funds.
8 comments:
JPC is a closed-end fund trading currently at a 13 percent discount to NAV. More than just bid-ask tracking error here.
well, remember JPC is down 6% over this period, so that would merely make it more like the 20% average for this group...the puzzle remains.
How does one get a total return chart on Bloomberg without using COMP?
I've been using a Bloomberg (rather skillfully, or so I thought) for 10 years and I can't figure out how you did that.
Well, the graph is just the price, but use =BDP("JFP US Equity","CUST_TRR_RETURN_HOLDING_PER ",{"CUST_TRR_START_DT","CUST_TRR_END_DT"},{20061231,20110331}), for example, to get the total return within a spreadsheet.
I might be mistaken but that graph doesn't show the price. That graph shows historical price adjusted for cash dividends.
If you ask my Bloomberg (or Google finance, I guess), the early 2007 price is ca $14/sh.
I figured it out-- it's the "Normal Cash Dividends" toggle on the DPDF screen. Now if I can just find a way to run that from the command line I will be all set.
My biggest gripe with BBG is that the parameters critical to understanding a data series -- in this case, whether dividends are added back -- aren't exposed. Eric's chart can't be distinguished from the same chart with DPDF set differently - only the displayed data would differ. Similarly, the BBG API will pull different data for the same call if PDF settings in BBG itself are changed, which leads to many bugs, snafus and irreproducibilities in real-life data exploration.
I think there might be some sort of dilution effect going on here - most mutual funds reflect a large inflow (in proportion to fund size) by applying a levy to reflect the cost of rebalancing the portfolio and consequently to recompense other unit holders for your money diluting their cost base. I believe (please correct me if i am wrong) that etfs do not do this - the cost of trading and holding an etf is bid offer and management charge is reflected in the price. Now if you look at the ishares HY etf since 2009 the AUM has gone up enormously (this is common accross high yield etfs) so everybody is having their cost base constantly adjusted upwards as the fund managers buy more and more expensive high yield (as the market has rallied) as the new units are created. This would lower your returns - say you bought in March 09 when average bond prices were in the 40s and 50s ever since then the managers have been buying bonds at higher and higher prices so when you calcualte your average cost to do your return on an asset level your denominator is constantly moving up and not being held constant. other effects like bid-offer will definitely come into it. thoughts?
HYG is an ETF that's been trading for about 4 years, total return on it has been 23%. Just FYI.
HYG is the iShares iBoxx $ High Yield Corporate Bond ETF.
Post a Comment