Paper: arXiv 2403.17127

Authors: David Ardia, Sébastien Laurent, Rosnel Sessinou

Abstract

We introduce a new framework for the mean-variance spanning (MVS) hypothesis testing. The procedure can be applied to any test-asset dimension and only requires stationary asset returns and the number of benchmark assets to be smaller than the number of time periods. It involves individually testing moment conditions using a robust Student-t statistic based on the batch-mean method and combining the p-values using the Cauchy combination test. Simulations demonstrate the superior performance of the test compared to state-of-the-art approaches. For the empirical application, we look at the problem of domestic versus international diversification in equities. We find that the advantages of diversification are influenced by economic conditions and exhibit cross-country variation. We also highlight that the rejection of the MVS hypothesis originates from the potential to reduce variance within the domestic global minimum-variance portfolio.

Complexity vs Empirical Score

  • Math Complexity: 8.5/10
  • Empirical Rigor: 7.5/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper employs advanced statistical methods including robust Student-t tests, batch-mean procedures, and the Cauchy combination test for high-dimensional spanning, indicating high mathematical density. It demonstrates strong empirical rigor through Monte Carlo simulations, application to real equity data for international diversification, and detailed discussion of implementation challenges.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Test Mean-Variance Spanning<br>in High Dimensions"] --> B["Methodology<br>Robust Student-t + Cauchy Combination"]
  B --> C["Input Data<br>Stationary Asset Returns<br>Test Assets > Time Periods"]
  C --> D["Computational Process<br>1. Individual Moment Tests<br>2. P-value Aggregation"]
  D --> E["Key Findings<br>Superior Power vs. State-of-Art<br>Diversification Benefits Vary by Economy<br>Rejection Driven by Variance Reduction"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes a new framework for mean-variance spanning (MVS) hypothesis testing that works for any test‑asset dimension and only requires stationary asset returns.

Source sentence
We introduce a new framework for the mean-variance spanning (MVS) hypothesis testing.

Assumes 6

  • Asset returns are stationary. statistical
    Source sentence
    The procedure can be applied to any test-asset dimension and only requires stationary asset returns and the number of benchmark assets to be smaller than the number of time periods.
  • The covariance matrix of returns is positive definite (nonsingular). statistical
    Source sentence
    We assume V to be nonsingular. Let Q ≡[0N×K ...
  • The estimator ˆθ is √T‑consistent for the true parameter vector θ. model
    Source sentence
    Assumption 1. ˆθ ≡(ˆθ1, . . . , ˆθd)′ is a √ T-consistent estimator of θ, the unique vector satisfying E[ ˆm(θ)] = 02N×1.
  • The process g(rt,θ) is stationary and ergodic. statistical
    Source sentence
    Assumption 2. g(rt, θ) is the t-th observation of a stationary and ergodic process.
Show 2 more
  • The batch‑mean method can be applied to the moment conditions without refitting the model. other
    Source sentence
    This allows one to avoid refitting the model when using the batch-mean method.
  • The number of benchmark assets K is smaller than the number of time periods T. data
    Source sentence
    The procedure can be applied to any test-asset dimension and only requires stationary asset returns and the number of benchmark assets to be smaller than the number of time periods.

Data

  • Daily returns of blue‑chip stocks market-prices unknown
    2007‑2022 · S&P 100 (U.S.), Euronext 100 (Eurozone), SMI (Switzerland)
    Source sentence
    Our empirical illustration uses daily returns from 2007 to 2022 on blue-chip stocks traded in North America and Europe. Specifically, at the begining of each year, we use stock returns of the historical constituents in the following equity indices: (i) S&P 100 (U.S.), (ii) Euronext 100 (Eurozone), and (iii) SMI (Switzerland).

Costs

Not applicable The paper reports no trading or execution result.

Finds 5

  • Monte Carlo simulations show the proposed BCS tests have correct size and higher power than state‑of‑the‑art methods. statistical
    Source sentence
    Simulations demonstrate the superior performance of the test compared to state-of-the-art approaches. Monte Carlo simulations show the proposed MVS tests have correct sizes and high power in most setups.
  • The benefits of international diversification depend on economic conditions and vary across countries. performance
    Source sentence
    We find that the advantages of diversification are influenced by economic conditions and exhibit cross-country variation. We find that the benefits of international diversification depend on economic conditions and vary across countries.
  • Rejection of the MVS hypothesis is driven by the potential to reduce variance in the domestic global minimum‑variance portfolio. theoretical
    Source sentence
    We also highlight that the rejection of the MVS hypothesis originates from the potential to reduce variance within the domestic global minimum-variance portfolio. The MVS hypothesis is rejected by the variance reduction potential in the domestic global minimum-variance portfolio.
  • The BCS tests work well on skewed, heteroscedastic, and fat‑tailed data, unlike many existing methods. statistical
    Source sentence
    Unlike state-of-the-art methods, they also work well on skewed, heteroscedastic, and fat-tailed data.
  • When N > T, the BCS tests remain feasible and retain decent size and power, whereas traditional tests become infeasible. statistical
    Source sentence
    Indeed, unlike the GRS, F1, and BJ, this test is still applicable when N > T and in this case, it has a decent size (unlike PY) and it has power against the null hypothesis Hα 0 (unlike GL). For N = 400, the GRS, F1, and BJ tests are not feasible.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 16 related papers). Coverage is partial, so this is neither replication nor refutation.

Cited by 1 paper in the archive (1 by independent authors)

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.