shinohara-xiao/industry_classification

industry_classification

行业分类

% % Background % A substantial body of literature investigates the influence of group homogeneity and industry classifications on cross-sectional return predictability. Numerous studies have demonstrated that stocks sharing the same industry or presenting similar characteristics exhibit group homogeneity, which has implications for asset pricing (Barberis and Shleifer, 2003; Hoberg and Phillips, 2010). Gaining a deeper understanding of the factors contributing to group homogeneity and its impact on asset pricing is crucial for both financial market participants and researchers.

% One line of thought explaining group homogeneity is based on traditional asset pricing theories. These theories argue that individual stock premia play a more significant role, while industry effects are treated as secondary factors or noise (Fama and French, 1993; Carhart, 1997). According to this perspective, group homogeneity arises from firm-specific factors or risk factors influencing individual stock returns, thus diminishing the importance of industry classifications in asset pricing.

% Alternatively, a behavioral finance approach emphasizes the role of investor cognition in shaping group homogeneity and asset pricing outcomes (Shleifer and Vishny, 1997; Barberis and Thaler, 2003). Proponents of this view suggest that investors' cognitive limitations and biases lead them to categorize stocks based on industry classifications or other observable similarities, causing industry effects and group homogeneity to be driven by investor behavior.

% Empirical evidence supporting these differing explanations is inconclusive. While some research indicates that industry classifications have a limited impact on asset pricing dynamics (Bhojraj et al., 2003; Cohen and Polk, 2005), supporting traditional asset pricing theories, recent findings reveal that multiple industry classifications result in comparable group homogeneity levels and minor differences in premia (Chen and Knez, 1995; Fama and French, 1997). These intriguing results imply that investor cognition may play a more significant role in group homogeneity and asset pricing than previously thought.

% --------------------------------------------------------------------- changed % Background

In the vast field of financial research, the centrality of group homogeneity and industry classifications in asset pricing and return predictability commands significant attention. Securities within similar industry categorizations or sharing common attributes demonstrate a distinct propensity for correlated behavior, thereby intricately influencing asset valuation practices (Barberis and Shleifer, 2003; Hoberg and Phillips, 2010). The utilization of industry-centric analysis significantly refines the decision-making processes of investment managers. This leads to enhanced precision in return forecasts and an optimization of asset valuation methodologies (Zhang, 2010; Huang and Zhang, 2012). Nevertheless, it is paramount to acknowledge the potential limitations posed by an overreliance on industry classifications. Beyond a certain point of granularity, the incremental contribution of such classifications to improved return prediction begins to diminish (Industry Classifications and Return Comovement).

% (1)A substantial body of literature investigates the influence of group homogeneity and industry classifications on cross-sectional return predictability. Numerous studies have demonstrated that stocks sharing the same industry or presenting similar characteristics exhibit group homogeneity, which has implications for asset pricing (Barberis and Shleifer, 2003; Hoberg and Phillips, 2010). % An effective industry-based analysis can offer a more appropriate measure and can be a valuable predictor of future industry returns，which is helpful in making reasonable asset pricing(Analysts, Industries, and Price Momentum). Also, stock groupings based on industry can exhibit stronger out-of-sample homogeneity than groups formed from statistical cluster analysis. (Industry Classifications and Return Comovement). % Most notably, when investment managers consider industry, they typically take into consideration the industry classification and industry affiliations of the component stocks (Industry classification schemes: An analysis and review). % (Industry Classifications and Return Comovement) has shown that that increasingly fine levels of disaggregation improve discrimination up to six-digit GICS codes, after which the benefits tail off, which means that asset pricing whould be differ across different level of industry classifications。 % Gaining a deeper understanding of the factors contributing to group homogeneity and its impact on asset pricing is crucial for both financial market participants and researchers.

Traditional asset pricing theories place primacy on individual stock premia, often minimizing the significance of industry effects or dismissing them as noise (Fama and French, 1993; Carhart, 1997; Campbell et al., 2001). Reinforcing this, the Capital Asset Pricing Model (CAPM) asserts that stock prices reflect all available asset information, suggesting that the division of stocks should not influence price volatility, given constant underlying information (Sharpe, 1964; Lintner, 1965; Fama, 1970). Hence, group homogeneity might arise more from firm-specific or risk factors affecting individual stock returns, casting doubt on the criticality of industry classifications in asset pricing (Banz, 1981; Fama and French, 1992).

% (2)One line of thought explaining group homogeneity is based on traditional asset pricing theories. These theories argue that individual stock premia play a more significant role, while industry effects are treated as secondary factors or noise (Fama and French, 1993; Carhart, 1997). % Also,he most famous asset pricing model (CAPM model) is based on the efficient market hypothesis, which holds that the stock price can fully reflect all available information of the asset, and when the information changes, the stock price will change accordingly. Based on this theory, we can think that no matter how we divide the stock, the price fluctuation of the stock will not change-because the information has not changed. % According to this perspective, group homogeneity arises from firm-specific factors or risk factors influencing individual stock returns, thus diminishing the importance of industry classifications in asset pricing.

In contrast to conventional asset pricing theories, the behavioral finance approach highlights the influence of investor cognition on group homogeneity and asset pricing (Shleifer and Vishny, 1997; Barberis and Thaler, 2003). This perspective asserts that industry-based stock categorizations facilitate a common understanding among stakeholders, aiding cognition in specific knowledge domains (The Classification and Taxonomy of Industries - Measuring the Right Thing). Further, the use of broad industry classifications to identify homogeneous groups of firms engaged in similar businesses underscores the importance of these classifications in predicting stock returns and asset pricing anomalies (Do Industries Matter in Explaining Stock Returns and Asset-Pricing Anomalies?). Therefore, the behavioral finance approach suggests that investor cognitive limitations and biases may significantly shape industry effects and group homogeneity.

% (3)Alternatively, a behavioral finance approach emphasizes the role of investor cognition in shaping group homogeneity and asset pricing outcomes (Shleifer and Vishny, 1997; Barberis and Thaler, 2003). % (The classification and taxonomy of industries-measuring the right thing) has demonstrated that dividing goups based on the industry classification can create a “common understanding” among stakeholders in a particular knowledge field, and “facilitate cognition of the knowledge field”,which is familiar with the consequences when we take industry as dividing scheme. Also, A broad industry classifications have been used widely to identify homogeneous groups of firms that engage in practice in “close” businesses(Do industries matter in explaining stock returns and asset-pricing anomalies?) . % Proponents of this view suggest that investors' cognitive limitations and biases lead them to categorize stocks based on industry classifications or other observable similarities, causing industry effects and group homogeneity to be driven by investor behavior.

% (4)Empirical evidence supporting these differing explanations is inconclusive. While some research indicates that industry classifications have a limited impact on asset pricing dynamics (Bhojraj et al., 2003; Cohen and Polk, 2005), supporting traditional asset pricing theories, recent findings reveal that multiple industry classifications result in comparable group homogeneity levels and minor differences in premia (Chen and Knez, 1995; Fama and French, 1997). These intriguing results imply that investor cognition may play a more significant role in group homogeneity and asset pricing than previously thought.

% Also,(market and industry factors in stock price behavior) research shows that in a smaller stock set, stock price changes tend to move as homogeneous groups, and show higher correlation and higher explanatory power; (industry classifications and return co movement) research also draws a similar conclusion: the finer the degree of industry division, the greater the impact on asset pricing and the more accurate the forecast.

% (Analysis of Variety of Returns to Deterministic Homogeneous Stock Groupings) found that the explanation of the regression equation of stock price regression after stock classification is stronger-is there any other classification method? The efficient market hypothesis regards this trend change as a kind of deviation error, and there is no way to explain the homogeneity of stock prices under different division methods, and the articles that put forward corresponding views have not given a reasonable explanation for the relationship between homogeneity and stock price fluctuations.

Empirical evidence provides mixed support for these contrasting theories. Certain studies lend credence to traditional asset pricing models, suggesting that industry classifications exert a limited impact on asset pricing dynamics (Bhojraj et al., 2003; Cohen and Polk, 2005). However, recent findings reveal that multiple industry classifications can yield comparable levels of group homogeneity and minor premia differences (Chen and Knez, 1995; Fama and French, 1997), hinting at the potential importance of investor cognition. Research further indicates that in smaller stock sets, price changes exhibit higher correlation and explanatory power, and finer industry divisions enhance asset pricing impact and forecast accuracy (Market and Industry Factors in Stock Price Behavior; Industry Classifications and Return Co-movement). However, the efficient market hypothesis and related literature fall short of explaining this observed homogeneity under different classification methods (Analysis of Variety of Returns to Deterministic Homogeneous Stock Groupings), leaving the role of industry classifications in asset pricing an unresolved issue.

% methods

In order to capture and predict the fluctuating price trend of stocks more accurately, we take China A-share market as the research object, take industries as different classification methods, and explore the changes of stock homogeneity under different classification methods and intensity, as well as the relationship between the fluctuations of stock price returns and the possible reasons.

Academics and practitioners frequently highlight that overall market and industry performance is an important aspect of a firm’s profitability. However, few studies allow for the decomposition of a firm’s profitability into the market, industry, and idiosyncratic components, in that case, how could we know which part of the fluctuation in stock returns is caused by the fluctuation of industry and market, and which part is by itself ?

Therefore, through decomposition, this paper divides stock price fluctuation into three levels: market, industry and firm-idosycartic. These three components represent different classification efforts, and the separation from market to firm level is gradually strengthened, while industry can divide various efforts according to industry classification.

In this process, to well understand the influence of stock return across intensity of division, we separate firm profitability into the market, industry, and firm-idiosyncratic components and test the sensitivity and predictive significance of these three different level components. Through this separating process, we can separate the three major volatility factors (market, industry, and idiosyncratic) of the stock price and study the impact of these three different factors on the stock price volatility separately, without considering the influence and correlations among the components. This process helps us understand the role that market, industry, and idiosyncratic components play in influencing stock price fluctuations without considering the mutual influence of components as well as the cross-sectional fluctuations of market and industry components. Also, we assess the sensiticity of components and the component's power of prediction. A higher beta shows that the component has a greater influence on asset pricing and similarly, a higher R-square value means the stronger the index's ability to predict the next period, which means it is more consistent using this component to forecast for the next period with the actual situation.

Based on this, we explore the changes and differences of stock homogeneity when the classification method of components under these three different classification dynamics changes, and how this homogeneity affects the fluctuation of stock return price.

We apply the process above under different industry classifications and different level of intensity under the same classification, We finally find that under different industry classifications and different classification intensity of the same classification and we find that industry component has the highest sensitivity, and firm’s sensitivity is higher than market’s, which is saying that industry has an influence on stock price, but this influence is not as great as we supposed, instead, market performe the best. As for the power of prediction, we can see that Shenyinwanguo Classificaiton,which is designed based on industry earnings, performe better at pricing assets than other country-designed industry groupings. However, our study found that industry-level component has little difference in performance across industry classifications, instead, the difference seems bigger among classification intensity, rather than different classifications, and the stronger the classification intensity, consistent and persistence under this intensity seem to have better outcomes.

Based on what is talked above, we try to find how the stock price fluctuate across different level of industry classifications

% conclusion------ remain to change % 结论 Based on what is talked above, we try to find how the stock price fluctuate across different level of industry classifications and we find that (1) higher level classifications have better performance; (2) the difference across classificaitons is not as big as we supposed; (3) The differences between different classification intesnsity under the same industry classification are more striking than the differences between different industry classifications, which is very different from American market.

%原因解释 There is an obvious industry rotation phenomenon in the Chinese market, and also, the definition of industry classification criteria is different, for example, industry classification criteria for revenue purposes: Shen Yin Wanguo Industry Classification and industry classification formulated by the government: SFC Industry Classification, which indicates different power and intensity of division. What’s more, the structure of investors in the Chinese market is very different from American market, among which retail investors account for the vast majority. Individual investors’ expectations can differ from those of professional investors, and the corresponding asset pricing in influencing asset prices likely differs by aggregation level(Review Article: Perspectives on the Future of Asset Pricing)