Comparing Return Dynamics Across Major Asset Classes

Summary

The question asks whether daily returns across asset classes have measurable differences in time-series behavior, distributions, trader composition, and relationships with macroeconomic variables. It gives examples of possible contrasts, such as whether an ARMA-GARCH model describes one market better than another, or whether return distributions differ in skewness and kurtosis. It also asks whether an unidentified series could be classified by resemblance to a known asset class.

The answer recommends a broad reference on expected returns across major asset classes as a starting point for research. The exchange itself does not present comparative data, statistical tests, classification procedures, or evidence that any particular asset class follows a specified model. The reference is therefore a literature pointer rather than a direct resolution of the empirical questions. Any attempt to classify an unknown series would require selecting measurable features and evaluating them against representative data, which the discussion does not detail.

Key ideas

• Asset classes may differ in return distributions, serial dependence, volatility, and macroeconomic relationships.
• The question proposes comparing time-series models and distributional characteristics across markets.
• The exchange points to a broad survey of expected returns across major asset classes as a research starting point.
• It supplies no empirical comparison or method for assigning an unknown series to an asset class.

Tags

# Asset class dynamics differences


# Asset class dynamics differences












If we compare daily return dynamics of the main asset class time series (e.g. Stock indexes, bonds, precious commodities, etc) do we observe quantifiable differences? Are there some reference paper on that topic?

What do I mean is that, for example, maybe gold returns might be well described by ARMA-GARCH model (of some orders) while I don’t know maybe forex not. Or maybe they have difference returns distributions (more skew, positive vs negative kourtosis, etc), they are more traded by a certain type of trader so different properties can be observed, they have different correlation with main macroeconomic variables, and so on.

Therefore if I have an unknown time series is it possible (and how can it be possible) to say if it behave “like” or more similar to a particular asset class?

## Answer by vonjd (score 1, accepted)

https://quant.stackexchange.com/a/39427

The most authoritative source for those questions is the following book:

Expected Returns: An Investor's Guide to Harvesting Market Rewards by Antti Ilmanen

You can find a free shortened (but still exhaustive) version here:

Ilmanen, Antti, Expected Returns on Major Asset Classes (June 1, 2012). CFA Institute Research Foundation 2012 - 1. Available at SSRN: https://ssrn.com/abstract=2616228

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.