Trader Talk

Deutsche Bank: ETF flows have a predictive value

rvlsoft | iStock | Getty Images

Watching ETF flow trends as an indicator of demand for stocks and bonds.

It's an ugly truth about the stock market that past performance is not an indicator of future trends. June was up, but it doesn't say anything about July.

There's been attempts for years to look at other indicators. In the last few years, a mini-industry has grown up around mining data on ETF flows.

Recently, Deutsche Bank (DB) began publishing reports around a new methodology it calls a Tactical Asset Allocation Relative Strength Signal (TAARSS--a mouthful, I know) that is based on an analysis of ETF flow trends.

The point: Fund flows may have some predictive value, but you need to look at more than just the magnitude of the flows; you need to look at what Sebastian Mercado, DB's VP and ETF Strategist, calls flow trend formation.

What's the difference? It's one thing to say that, for example, ETFs attracted $25 billion in inflows in June.

That's interesting, but it's not clear how predictive that is of anything.

DB says it's more important to see a pattern of, say, $1 billion in inflows every day in a particular asset class, on a regular basis, than to see a choppy pattern of big inflows, followed by outflows.

When you get consistent inflows, that's a sign there is an investment demand shift for that asset class.

The key, Mercado says, is to be able to look for activity that is related to asset allocation. You want to see real investors putting money to work in a specific asset class on a regular basis. That's commitment.

That seems obvious, but a lot of flow activity has nothing to do with asset allocation. Many times, flow activity is being driven by short positions or hedging positions.

So while you might see inflows in the (SPY), the largest ETF, it may be because market makers are lending out the stock to short.

That's a false signal. DB's methodology eliminates some of the largest ETFs because they are often used for non-asset allocation activities.

A report published by DB last week saw the following ETF patterns in the last few weeks:

  1. Strong equity flow across all regions, but particular strength in Europe and North America;
  2. in Europe, peripheral countries like Italy and Spain have attracted stronger flows;
  3. emerging markets have seen stronger flows overall (as a percentage of assets under management) than developed markets;
  4. U.S. treasury and corporate bond inflows were negative;
  5. gold: Despite a rally in gold prices, flows remained flat.

The point of the research is that trends tend to go for multiple periods.

Has any of this been back-tested? Mercado says following this kind of flow data generates larger and more stable returns than, for example, a straight 60/40 stock allocation, or price/momentum strategies.

I'm personally somewhat skeptical about this, but I find it intriguing because it represents an additional level of sophistication over simply watching fund flows on a dollar basis.