Oliver B. Linton,Haihan Tang,Jianbin Wu
Abstract:We propose a confirmatory dynamic factor model for a large number of daily returns across multiple time zones. The model has a global factor and three continental factors. We propose two estimators of the model: a quasi-maximum likelihood estimator (QML-just-identified), and an improved estimator (QML-all-res). Our estimators are consistent and asymptotically normal. In particular, the asymptotic distributions of QML-all-res are the same as those of the infeasible OLS estimators that treat factors as known and utilize all the restrictions of the parameters of the model. We apply the model to MSCI equity indices of 42 developed and emerging markets, and find that the market is more integrated when the US VIX is high.
What problem does this paper attempt to address?
### Problems the paper attempts to solve
This paper aims to develop a framework for modeling the correlations between daily returns in stock markets in different time zones. Specifically, the paper proposes a Confirmatory Dynamic Factor Model to analyze international stock returns as if they were measured at the same time and place, thereby eliminating significant biases that may occur in the research.
#### Main problems
1. **Asynchronous trading problem**:
- Since global stock markets are distributed in different time zones, trading hours do not overlap. For example, the US market and major Asian markets have no overlap, while the European market has partial overlap with the US and Asian markets.
- When using daily data to measure correlations, researchers need to deal with the asynchronous trading problem. For example, the closing price in the US market contains a lot of news information, which cannot be reflected in the closing price of the Asian market on the same day, and vice versa.
2. **Market integration**:
- Studying the correlations between stock returns in different countries is very important for strategic allocation decisions of international stocks and for understanding the degree of segmentation in international capital markets.
- The paper analyzes the degree of market integration in different time periods through the model, especially whether the market integration increases during high - VIX (Volatility Index) periods.
3. **Model estimation**:
- Two methods for estimating the model are proposed: the Quasi - Maximum Likelihood Estimator (QML - just - identified) and an improved estimation method (QML - all - res).
- The consistency and asymptotic normality of these estimation methods are proven, and the asymptotic distribution of QML - all - res is the same as that of the infeasible OLS estimator, which assumes that the factors are known and utilizes all the restrictions of the model parameters.
4. **Empirical application**:
- The model is applied to the MSCI stock indices of 42 developed and emerging markets, and it is found that when the US VIX is high, the market is more integrated.
- In particular, the global factor loadings in the Asian sub - period increase significantly during high - VIX periods, which means that the returns in the Asian market have stronger trading signals for other markets during high - VIX periods.
### Model overview
The paper proposes a dynamic factor model that includes a global factor and three continental factors (Asia, Europe, and America). The model assumes that the 24 - hour closing - price log - returns follow a dynamic factor model driven by global and continental factors. Markets in each continent are affected by global and local information in different sub - periods (Asian sub - period, European sub - period, and American sub - period).
### Estimation methods
1. **QML - just - identified**:
- Estimate the model parameters by the quasi - maximum likelihood method to ensure the identifiability of the model.
- This method has consistency and asymptotic normality as the sample size increases.
2. **QML - all - res**:
- Improve the QML - just - identified method by considering all the restrictions in the model.
- It is implemented by the EM algorithm and improves the accuracy of estimation.
### Empirical results
- **Market integration**:
- It is found that during high - VIX periods, the market integration increases significantly, especially the global factor loadings in the Asian sub - period increase.
- This means that during high - VIX periods, the returns in the Asian market have stronger trading signals for other markets.
- **Weak - form efficient market hypothesis**:
- The study finds that the weak - form efficient market hypothesis does not hold globally, and some international news takes more than 1/3 of a day to fully unfold or dissipate.
### Conclusion
Through the proposed dynamic factor model, the paper successfully solves the asynchronous trading problem and reveals the correlations and integration between stock markets in different time zones. In particular, the increase in market integration during high - VIX periods provides a new perspective for understanding the dynamics of international capital markets.