Harvesting Dividend Yield in China
Traditionally, in times of slower economic growth, when the central bank is being accommodative via a low interest rate policy, yield-seeking investors turn to dividend yield. The chart below shows the cumulative Dividend Yield factor return from Axioma’s newly updated China fundamental factor model, the AX-CN4. As of December 10 2018, the cumulative factor returns for Dividend Yield year-to-date was 4.6%. This compares very favorably to the CSI300’s -22% over the same period.
The attraction to dividend yield’s defensive qualities in times of bear markets is well documented in the literature, but does this still hold for an emerging market like China? And if so, how can we harvest it? Additionally, does this defensive quality cost us too much upside potential in times of bull markets?
We set out to answer these questions using the latest China fundamental factor model from Axioma. We created two long-only strategies to gain as much exposure to the dividend yield factor in our model while minimizing active risk to the CSI300 benchmark. The backtest ran from January 2015 through October 2018. This represents a period with multiple bull and bear markets. The first strategy is an unconstrained long-only and fully-invested portfolio seeking to maximize the active exposure to the dividend yield factor while minimizing active risk to the index. The second strategy is a constrained strategy which included neutralizing active constraints on non-target style factors, as well as a maximum active deviation of +/- 5% on sectors vis-à-vis the index. Both strategies are rebalanced monthly (at month-end).
Both strategies handsomely beat the index during that period with active returns of 4.91% for the unconstrained strategy and 3.62% for the constrained one. The chart below shows the contribution to active return from both strategies’ active exposure to dividend yield alongside the CSI300. We can see that a positive active exposure to dividend yield really helped the strategies during bear markets, both in the second half of 2015 and during all of 2018. Dividend yield’s contribution to active returns was definitively weaker during the bull markets of 2016-2017, but still contributed positively over that period with most of the under-performance coming in a single quarter (Q4 2016).
The table below compare the performance attribution summary for both strategies. Active risk of 4.15% was slightly higher for the unconstrained strategy than the constrained strategy at 3.96% - which is to be expected – but both achieved very respectable information ratios of 1.18 and 0.91 respectively. In both cases, dividend yield contributed the lions share of the active return. This style return was augmented with active bets on the profitability and value factors in the case of the unconstrained strategy. So, in China, it would seem that dividend yield seeks the company of profitability and value. No other style factor loadings were significant in the unconstrained strategy.
From a sector point of view, the unconstrained strategy ‘financed’ its overweights in consumer staples, energy, and real estate with large underweights in information technology, materials, industrials, and communication services. Unable to use active bets on non-target style factors to minimize active risk, the constrained strategy had to make more use of the sector correlations to achieve its active risk goal. We therefore see large deviations between the two strategies in their exposures to health care, financials, and energy. Both strategies, however, were in agreement that dividend yield isn’t to be found in the information technology, materials, industrials, or communication services sectors.
In summary, using Axioma’s updated China model, we were able to easily investigate the suitability of a dividend yield strategy for the Chinese market. In combination with Axioma’s portfolio construction and performance attribution platforms, we configured and ran two variants of a dividend yield harvesting strategy over the past four years. The evidence so far not only points to dividend yield as a strong systematic source of return in that market, but one that can easily be harvested with or without constraints on the other factors. It would also seem from this back-test that dividend yield might be an all-weather factor, contributing strongly in bear markets while not detracting much in bull ones. I look forward to using this updated model to test other fundamental systematic betas in the Chinese market and work with our clients to design strategies to harvest them.
 Factor Returns are estimated using often un-investible long-short portfolios (see model handbook for details)