Does Corporate Governance Matter? Evidence from the AGR Governance Rating

Do better governed firms perform better than their peers? In the academic literature, numerous studies have tackled this question with mixed success. On the one hand, there appears to be a robust statistical association between governance indices based on anti-takeover provisions and future operating performance, see e.g. Bebchuk, Cohen, and Wang (2013). On the other hand, these indices are characterized by limited time variation for a given firm, and their effects are harder to identify when focusing on finer industry definitions (Johnson, Moorman, and Sorescu (2009)). Altogether, it appears that the evidence that better governance is indeed a driving force behind a firm’s success is still far from conclusive.

Parallel to these academic studies, several commercial metrics have been developed with the purpose of rating the soundness of a firm’s governance. Daines, Gow, and Larcker (2010) examine the relation between several of these commercial scores and a firm’s future output working with a snapshot of 2005 data, and find mixed to weak empirical evidence. Among the ratings considered, they document somewhat more promising results for MSCI’s AGR governance rating. This rating strives to obtain a more accurate view of a firm’s governance by combining governance inputs with outputs. The idea behind AGR is that lax governance environments are less immune to management misbehavior and thus are more prone to financial misreporting and fraud.

In our paper, Does Corporate Governance Matter? Evidence from the AGR Governance Rating, we reexamine the importance of a firm’s corporate governance using a panel of MSCI’s AGR ratings available for about 8,300 public U.S. firms over the 1997 to 2011 period. The availability of such a large panel in the cross-section and over time gives us the advantage of being able to investigate the dynamic relation between governance and performance.

As a prelude to our analysis, we document two important features of AGR scores. First, there is substantial time variation in a firm’s AGR score. Indeed, of the poorly governed firms in the bottom decile of the AGR distribution in a given month, only 35% of them remain in this decile after 12 months. Second, we find little to no correlation between AGR scores and the widely used G-score of Gompers, Ishii, and Metrick (2003) and E-index of Bebchuk, Cohen, and Ferrell (2009). These findings confirm that AGR scores capture a dimension of governance that is distinct from that reflected by the number of anti-takeover provisions in place, and that can be identified separately from time-invariant firm characteristics such as a firm’s culture.

Armed with this evidence, we ask whether corporate governance as captured by AGR scores is associated with higher future operating performance. We find that higher AGR-rated firms are indeed characterized by better future operating performance as measured by the firm’s Return on Assets (ROA), even after controlling for firm fixed effects and time-varying firm characteristics. In numbers, a 10-point increase in AGR rating is accompanied by a 0.15% expected increase in that firm’s 2-year ahead industry-adjusted ROA. A similar positive relation is found when relating the AGR score to other measures of firm’s output previously used in the literature such as Tobin’s Q, net margin, and sales growth. In the cross-section, the effect is more pronounced for firms in the lower AGR decile, i.e. poorly managed firms. We also find that over time, the relation between AGR and operating performance is actually stronger in more recent years.

We next investigate whether the information contained in AGR scores is fully reflected by current market valuations. To this end, we rely on a standard portfolio-formation approach where we group stocks into thirds based on their end-of-month market capitalization and, separately, into four groups based on the most recent AGR score. We then sum up the value-weighted returns to the size portfolios of a given AGR group, as in e.g. Hirshleifer, Hou, and Teoh (2012). We look at the performance of each of these four AGR portfolios, as well as of a long-short AGR strategy that goes long better governed stocks (Conservative group, bottom 10%) while shorting the group of poorly governed stocks (Very Aggressive group, top 15%). We regress the returns to these portfolios on eight risk factors: the five factors in Fama and French (2014), which augment the traditional market, size, and book-to-market factors of Fama and French (1993) with profitability and investment; a momentum factor; the accrual factor of Hirshleifer, Hou, and Teoh (2012); and a factor based on standardized earnings surprises.

We find that the long-short AGR portfolio delivers a nearly 50 basis points monthly abnormal return, and is highly significant. Interestingly, this alpha mainly originates from the set of poorly managed low AGR firms, which generates a negative abnormal return of about 40 basis points. We obtain quantitatively similar numbers when focusing on industry-adjusted instead of excess returns, and when removing any single industry from the sample.

As an alternative to the portfolio approach, we also estimate monthly cross-sectional Fama and MacBeth (1973) regressions. This methodology allows us to control directly for a large array of firm-level characteristics that may be related to expected returns, such as a firm idiosyncratic volatility or its book-to-market ratio. Notwithstanding the inclusion of these variables, we find that future firm returns are positively and significantly related to its most recent AGR score.

Our findings beg interesting questions for further research. Early studies find a strong relation between governance indices and abnormal returns. Recently, Bebchuk, Cohen, and Wang (2013) document that such correlation has actually disappeared in recent years. They advocate a learning explanation whereby investors have gradually incorporated differences in governance into market prices. Whether the association between MSCI’s AGR ratings and abnormal returns will persist beyond our sample period will depend much on investors having access to and trading on this information.

The full paper is available for download here.