Asset pricing anomalies refer to the evidence that cannot be explained or captured by an asset pricing model, i.e. the return is inconsistent with the estimated return from an asset pricing model. Anomalies are often used as the evidence against the efficient market hypothesis because it is abnormal compared with a normal return from the rational model. Recently, Cooper, Gullen and Schill (2008) show an important new anomaly -- the asset growth anomaly -- which reveals the negative relation between firm asset growth and subsequent stock returns on the US market after controlling for the book-to-market ratio and firm size. In addition, international studies of the asset growth anomaly show that it is more apparent in developed markets than emerging markets (see Titman, Wei and Xie, 2013; Watanabe, Xu, Yao and Yu, 2013). The explanations for the asset growth anomaly can be grouped into two broad categories: rational explanations (Li and Zhang, 2009; Watanabe, Xu, Yao and Yu, 2013; Hou, Xue and Zhang 2015) and behavioural explanations (Cooper, Gullen and Schill, 2008; Lipson, Mortal and Schill, 2011). In the first empirical chapter, the asset growth anomaly is tested across a long period and in different industries (Fama-French industry classification) on the US market. By using US data from 1963 to 2011, I show that 13 out of 44 industries feature the asset growth anomaly. Motivated by the different asset structure in different industries, I examine whether industry characteristics have influence on the asset growth anomaly. According to the empirical results, existing explanations (i.e. Q-theory with investment frictions and mispricing with limits-to-arbitrage) cannot fully explain the variations of the anomaly at the industry level. After controlling for existing explanations, I find the anomaly is a function of industry characteristics which proxy for industry competition and to a lesser degree the growth opportunities within an industry. The findings suggest that the asset growth anomaly can be at least partly explained by the reaction of investors to the growth opportunities within less competitive industries. In the second empirical chapter, I directly investigate if overreaction is the source of the asset growth anomaly. Investors have been warned not to pay too much for growth, yet empirically there is a strong negative relationship between asset growth and subsequent stock returns - the asset growth anomaly. It may suggest overreaction to firm asset growth. Previously, Cooper, Gullen and Schill (2008) show the reversal pattern of margin profit by comparing before and after the asset growth portfolio formation date. However, it does not test how the asset growth anomaly interacts with the degree of overreaction. In addition, some studies test how limits-to-arbitrage affect the asset growth anomaly. However, limits-to-arbitrage cannot state the source of mispricing. High limits-toarbitrage cause high risk (or high transaction cost) for the arbitrage activity and hence anomalies cannot be traded away easily; but it does not tell us how investors' behavioural biases move price away from fundamental value. Overall, there is a lack of direct evidence of overreaction as the source of the asset growth anomaly. I propose that investors' expectation error on the trend and profitability of asset growth is the reason behind paying too much for growth and hence the anomaly. The empirical analyses provide strong evidence that investors use the historic growth trend to extrapolate future growth. Specifically, the asset growth effect is stronger when the consecutive growth trend is longer. The finding is robust to controls for existing explanations of the asset growth anomaly (Q-theory with investment frictions and limits-to-arbitrage) and traditional risk measures. To control all the proxies of investment frictions and limits-to-arbitrage but avoiding multicollinearity, I also conduct factor analysis to extract common factors. Prior literature compares Q-theory with investment frictions and limits-to-arbitrage by one on one comparison rather than controlling for all the proxies to avoid high correlation among these proxies (Li and Zhang, 2010; Lam and Wei, 2011; Watanabe, Xu, Yao and Yu, 2013). In the third empirical chapter, I examine if other anomalies, like the asset growth anomaly, are more prominent in developed markets than emerging markets. Therefore, I study 16 extensively documented anomalies in 45 countries across the globe for the period between 1980 and 2013. The results show clear evidence that developed markets have more anomalies than emerging markets. And most importantly, I provide news watcher efficiency as an explanation of this phenomenon. Developed markets are considered more efficient than emerging markets and more efficient markets should have fewer anomalies based on the efficient market hypothesis. However, previous literature documents a puzzle that developed markets have more asset pricing anomalies than emerging markets. To understand the puzzle, I first apply the latest q-factor model (Hou, Xue and Zhang, 2015) and 5-factor model (Fama and French, 2015). Although these models provide explanatory power for some of the anomalies, the puzzle - a difference between developed and emerging markets - still remains. This test also provides an out-of-sample check for the new asset pricing model. Furthermore, the difference is more profound in equal-weighted than value-weighted anomaly returns. Building on Hong and Stein (1999) I hypothesize and show that very slow information diffusion in emerging market stocks and especially for small stocks provides an explanation of the puzzle. News watcher efficiency determines information diffusion speed and there is a nonlinear relation between news watcher efficiency and anomalies. When information diffusion is slow which is the first stage, there is no apparent price change caused by news watchers. And, therefore, momentum traders have no trend to follow; namely, there should be no momentum activities. As a result, anomalies cannot be observed even if the price does not reflect information in the market. This is the case for emerging markets. As information diffusion speeds up (which is the second stage), price will gradually incorporate information but not immediately. In this situation, there should be more momentum activities because momentum traders have a clear trend to follow. High momentum intensity, therefore, is more likely to overshoot the price and cause overreaction or price continuation. When investors realize the fundamentals, there is reversal in the long run. This information diffusion phase is the situation in developed markets. Consistent with the prediction, the empirical results show the nonlinear relation (the inverted U shape) between anomalies and news watcher efficiency proxies (higher education, investor sophistication and accounting standards). Therefore, in summary, this thesis considers three aspects of anomalies. First, asset growth anomalies, to some extent, relates to industry characteristics. Second, overreaction is the source of the asset growth anomaly. Third, there are more anomalies in developed markets than emerging markets, and this is due to the slow information diffusion in emerging markets.