Correlations are up, yes. But they’re also up between all asset classes, not just within large cap stocks. Are you suggesting ETFs are to blame for the increased correlations of Gold and Bonds too?

Finally, don’t forget that far more notional value is getting traded in the futures market than in ETFs OR the underlying, day in and day out, and that really drives the prices of the underlying. It was futures trading, not ETFs, for instance, that caused the infamous “flash crash.”

Models work until they come into broad use. Then they stop working. The consensus is thus always wrong.

Actually, it’s quite a bit more likely that no such arbs exist. Any time someone claims the market is mispricing an asset, you have to ask how they came up with their valuation. Most of the time, it’s the valuation model, not the market that’s wrong. I think these two guys from the Kauffman Foundation are just cranky that the foundation’s endowment is getting its face ripped off in this market. It can’t be their models, so they have to find something to blame.

Real life is heteroskedastic. Models work right up until the point they don’t. Correlations go through cycles, too. There’s nothing you can do, except adjust. I mean, anyone who trades knows this instinctively.

To restate earlier comments, to blame ETFs is to blame a purely technical factor when there are obvious huge fundamental uncertainties in play which affect the whole market. And even on the technical side, there are also enormous futures markets, which probably have more effect on the market as a whole than ETFs. It takes much more than finger-pointing at a couple of graphs to make a convincing case here.

There may be smaller sectors where ETFs do have an effect (Gold?), but even then that does not make it in any way wrong for correlations to rise when there are market-wide or sector-wide effects.

Finally, just because things are different now from the past does not mean that the past was right and the present is wrong. Maybe the markets were more inefficient in the past. Is the time axis on that graph meant to imply that the rigged market of the 1920s is something we should aspire to?

you’re talking about NAV arbs – and that’s another good point – you WANT arbs to keep prices in line.

(you want “Rebalancing arbs” also, in fact, to take the price risk for you – they buy the stocks and sell them back to the ETF, lessening the ETF trading impact)

But wait! Felix says it’s all due to the evils of ETF’s, and produces a graph of historical realized correlations, claiming that correlations today are higher “even than they were in the Great Depression.” A glance at this graph, though, shows the opposite: although the highest spike is recent, they area under the curve between 1931-1934 is clearly greater than between 2008-2011. And in fact, the JP Morgan paper linked through to by Felix shows that today’s realized 2Y (i.e. integrated) correlations are not much greater than those of 1987, never mind the depression.

What’s more the primary cause identified by the JP Morgan paper is the “macroeconomic environment” by which they mean exactly my initial guess.

How can you write so confidently about something you can’t read, Felix? You present with remarkable symptoms of alexia without agraphia.

Correlations may be high, but they aren’t nearly perfect. If you have invested in Oracle over the past year-plus, your experience is very different from somebody who has invested in HP over that span of time. Over the past four years, those investing in WFC have lost 30%. Those investing in BAC have lost 90%. Over the past year, PEP and KO have opened a 10% gap, then closed it, then opened a 15% gap. All reasonably similar companies with VERY different performance. Isn’t that what price discovery is about?

I wonder if the observed rise in correlations has to do with the fact that the market is being driven by larger macro-concerns than in past years? European default is legitimately bad for nearly every company (to varying degrees). QE2 supports asset values equally broadly. This isn’t necessarily a failure in price discovery, but a recognition that values depend on much more than individual company execution these days.

If the purpose of an ETF (as it is with an index) is to reduce volatility risk, then this should have been in place already.

I really don’t see what “baby” is with the dirty bathwater here, Felix. ETFs are and have been constructed to fool B&H investors, not attract them.

If you want the benefits of indexing, you buy an index (QQQ, MSCI, SPX, DJI, VIX, RUT, to name a few).

If you want more expenses–with the possibility that you’ll beat one of those–you buy iShares (IWM, IWN) or Vanguard (VTWx, VRTGx) or some other “tracking stock.”

If you want something that claims to be related to the above but gives you no good information about its balancing act, you buy an ETF. And then you look up in wonderment when you find out they overweighted NFLX…