For equity funds, the basic univariate stylized facts are about the same as any equity portfolio:
Still, there are special effects that show up in CEFs be cause we have two time series to look at --- the market price and the net asset value (NAV).
- Weak Ljung-Box measure of signal in the returns
- Strong Ljung-Box measure of signal in absolute values of returns
- Significant "Leverage effect" as measured by TGARCH and EGARCH fits
- AR(p) in mean and GARCH(1,1) in noise does a decent job
- You may squeeze out a tad more with GARCH driven by t-noise, or with an EGARCH.
Well, this is logical from the fee analysis, but it comes back into anomaly land when you look at comparable CEFs and OEFs (same fees, same asset class, same manager quality --- as if we could really measure manager quality --- but you know what I mean.
A related empirical problem is that the fee does not predict the discount nearly as well as one would hope. The papers that address this are not entirely believable to me, but I do believe that the phenomenon is real. Some people have use this "empirical fact" to argue that "fees do not matter" to the purchasers of CEFs.
Well, this is pretty close to insane.
Since it mostly shows up with funds that have a "managed distribution dividend policy" and monthly dividends --- a Nuveen niche, this seems to be the consequence of naive investors chasing "yield" which sadly consists mostly of "return of funds." Financial engineering run amuck!
It is surely the case that these bandit CEFs are "hard to borrow" --- or else shorting them would solve the problem. Still, a persistent individual investors (or a very small firm) may find enough to justify the effort.
I don't like using the word "correlated" here because when I speak of correlation I like to be explicit about the "random variable." Here what one hopes to capture is the phenomenon that there if you look at a collection of funds, look at the average discount on a fund-by-fund basis, then you find periods where almost all the funds have larger than average discounts and periods where almost all of the funds have a narrower than average discount.
People who like behavioral explanations of financial phenomena have argued that this comes from there being some collective "optimism" or "pessimism".
This means that you don't get as much bounce right away with the CEF as you get with the twin OEF. Still, it may convert into more exploitable momentum. For example, if on some glorious day in the future you see that the SP500 is up 3% on the open, you may be able to catch a close SP500 CEF that is up much less. Over the day or the week, you have a good shot of picking up 100 of relative performance. Naturally this is evanescent if you have to suffer market impact costs on your purchases.
Marty Zweig (a good friend of Wharton) was one of the earliest analyst to suggest that the size of the aggregate CEF discount might provide a useful measure of "investor sentiment." De Long and Shleifer have a nice piece that addresses this theme in the context of the 1929 crash.
One observation from De Long and Shleifer: "... the median closed-end fund premium in the third quarter of 1929 was about 50 percent."
That of course is completely nuts --- and it also gives us some encouragement about today's market.
It could be "fun" to revisit their analysis in the current context --- though I typically shy away from historical analogizing in favor of the "what's changed theory."
The mystery here is that there are any buyers of the IPO!
From our cynical perspective, this contributes to the classic technique of looking for stupid counter parties.
"Many models with rational agents attribute the underpricing of industrial IPOs to information asymmetry between the issuer and the investing public.."--- WFI resource
To be more PC, perhaps we should say that there are no stupid people, just people who do stupid things. The only reason I hesitate is that a lot of people do stupid things over and over again.
Here the big benefit naturally accrues to the issuer, but there may be some residual benefit to knowing that the initial holders of the asset are from the clueless part of the planet.
In other words, something like the India Fund trades a bit like convex combination of the "Indian asset basket" and a "US domestic basket." Here one needs to look at weekly or longer periods or else "clock effects" might be part of the explanation.
Here again, the prevailing argument seems to be behavioral, but I have not chased down the literature.
There may be something here that is is like the phenomenon of Over/Under Reaction of Stock Prices to Earnings Announcement that has been examined in a number of papers.
You can find times when the bond market rallies and the stock market sinks --- and at least some bond ETFs will sink too. Naturally, this is insane.
I have not seen any systematic study of this phenomenon, but there may be some academic papers that comment on this.
This fits very well with the Stupid CPT, since CEFs are primarily the domain of retail investors because of the low volumes and high market impact costs. It would be interesting to explore this in the micro-project or the main project.