EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

389. An explanation of under-diversification puzzles through ambiguity tastes and beliefs

Invited abstract in session MD-63: Optimization Model for Novel Risks in Finance and Climate, stream OR in Banking, Finance and Insurance: New Tools for Risk Management.

Monday, 14:30-16:00
Room: S14 (building: 101)

Authors (first author is the speaker)

1. Stavros A. Zenios
Durham University and University of Cyprus
2. Somayyeh Lotfi
Department of Accounting and Finance, University of Cyprus

Abstract

We developed portfolio selection models with heterogeneous ellipsoidal ambiguity sets of correlated returns and a performance ratio without a normality assumption and explained the equity home puzzle through the ambiguity channel. First, we develop a parsimonious model without risk or ambiguity aversion parameters and show it to be consistent with second stochastic dominance. We generalize the model for continuous ambiguity aversion. These models account for perceived ambiguity (beliefs) and ambiguity aversion (tastes) as potential puzzle explanations.

Taking the models to the data, we obtain optimal allocations matching those of international investors in 21 developed and 19 emerging markets. This is shown both for worst-case ambiguity aversion under perceived ambiguity well within the market estimat ambiguity sets or for mild ambiguity aversions given the market-estimated ambiguity sets. The global average ambiguity aversion is about 0.6, statistically different from the ambiguity-neutral 0.5. Our estimates of ambiguity aversion from the observed asset allocations for a large sample of countries closely match the scant experimental evidence from US, Dutch, and Chinese population samples. Our findings are robust to two fundamentally different methods for measuring ambiguity.

The worst-case model applied to a homogeneous ambiguity set predicts lower ambiguity with less diversified portfolios. We verify this on a large dataset of US household portfolios, and by running

Keywords

Status: accepted

Back to the list of papers