Final Activity Report Summary - FINPROCESS (Financial Services reflecting the underlying Decision Making Process) In a time of aging demographics and increased self-responsibility for retirement, it is important to achieve a thorough understanding of the factors underlying individual investors' decision-making and the impact of such decisions on overall markets. Recent times are replete with instances of investor sentiment and how they may affect individual investor decision-making and subsequently overall stock markets. Investor sentiment can be related to hypes and crashes in financial markets that have little to do with the underlying fundamental properties of securities and other financial products. Investor sentiment may drive market prices up or down depending on whether negative or positive feelings dominate. Nevertheless, existing research neglects the important role of individual investor sentiment and its impact on, for example, the buy and sell decisions of financial market participants and its potential effect on overall market stability. Rather, current studies treat sentiment as an aggregate, overarching 'bin' of constructs that is held responsible for market ups and downs inexplicable with fundamentals without trying to understand its essential underlying properties and dynamics. Ignoring micro-level investor heterogeneity and sentiment, however, severely limits our understanding of (the drivers of) overall stock market developments such as hypes and crashes. The project had three main objectives. The first objective was to redefine the financial concept called 'investor sentiment' to bring it closer to the actual decision-making of investors. The second objective was to provide a robust empirical framework to validate the new definition of investor sentiment and explain the actual decision-making of individual investors and understand overall market price developments. The third objective was to propose a new approach to theoretical research in finance that builds upon combining 'hard' accounting data such as individual's investment account records which reflect their decision-making with 'soft' survey data which can be used to estimate individual investors' sentiment. Behavioural finance, which is the application of social science findings to the domain of financial research, emerged to explain why securities prices often diverge from 'efficient' theoretical prices. It relies on two fundamental assumptions; firstly, that arbitrage in securities markets is limited, and secondly, that the so-called concept of 'investor sentiment' exists. Arbitrage, the process of exploiting price differences in different markets such that similar products have similar prices, causes securities prices to revert to their efficient values. Prices may still diverge from efficient values if the force that drives prices away from efficient values is greater than the supply of arbitrage. 'Investor sentiment' is what finance has come to call the process that drives securities prices away from their efficient values. When prices diverge greatly from their efficient values, bubbles tend to occur. This speaks volumes about the importance of investor sentiment given the current state of global markets. Exactly what investor sentiment is remains nonetheless the subject of great controversy and to date no unambiguous definitions of this concept exist. As a result of investor sentiment's vague definition in current financial research, the term is used to refer to a plethora of different theories and measures. In addition, words like 'optimism' and 'confidence' are used interchangeably with investor sentiment, even though they do not represent the same underlying psychological concept. The consequent confusion between financial practitioners, regulators, journalists, and researchers prevents constructive use of the investor sentiment concept in any profession and a solid relation between sentiment, individual investor decision-making and the resulting overall market prices. The current project draws its motivation from the fact that by developing a specific, unambiguous, and theoretically sound definition of investor sentiment, the usefulness, comparability, and measurability of this concept become palpable. The first objective of the project is thus to propose a detailed theoretical framework that will serve as a specific definition of investor sentiment. The theoretical framework is rooted in psychology and fulfils our goals of usefulness, comparability, and measurability. We propose that investor sentiment is: ...an affective process in the minds' of market participants, driving their judgment and decision processes which can be related to overall market price developments. Our theory extends the standard view that market participants are divided into only 'rational' (those that drive prices towards efficient values) and 'irrational' participants (those that drive prices away from efficient values). Instead, markets are composed of market participants that are partly rational and irrational at the same time, because their investment decisions are driven by a cognitive, 'rational' process, moderated by an affective, 'irrational' process. In order to test our theory of investor sentiment as an affective process, it is necessary to introduce new methodologies and concepts to the field of finance. In order to validate our definition of investor sentiment as affect, it is necessary to use an approach that is innovative in three respects: (1) in terms of the way we approach the problem, (2) in terms of its interdisciplinarity, and (3) in terms of its data requirements. Our approach to the problem is unique, as we employ a 'bottom-up' methodology that has not been previously introduced to this field of research. Most research in investor sentiment uses a 'top-down' approach, attempting to use aggregate market data to deduce the behaviour of the 'average' investor. The latter approach is flawed as aggregate market data only reflects the outcome of decisions, from which it is not possible to deduce individuals' latent drivers of sentiment and hence their decision-making process. Our research has made a significant step forward by measuring the latent psychological drivers of sentiment of investors at the individual level. Structural Equation Modelling (SEM) frameworks allow us to link the latent variables of each investor with their actual trading decisions. As such, this is a technique requiring a very detailed database (which we possess), combining investor's 'soft' survey data, and their related 'hard' accounting or trading data. The validity of this 'bottom-up' approach has been proven in social sciences disciplines such as psychology and marketing where this method is extensively used, but until now has seldom been found in the finance literature or applied to financial decision-making contexts. The second objective of this project is therefore to identify the relevant empirical issues that arise in this instance of interdisciplinary research, and introduce the adapted relevant methodologies into the finance research context. The innovative result is a new set of variables to consider, such as latent behavioural characteristics. As behavioural finance progresses, it will rely more and more on methodologies such as the one presented in our research. Our methodology has the unique feature of putting individuals' behaviour at the centre of behavioural finance research, instead of pricing anomalies. It is an empirical innovation of which we are confident that it will pave the way for a fundamental rethinking of behavioural finance research and increasing it explanatory power in terms of individual level decision-making and overall market developments. Finance is a fortunate discipline in the sense that most of the data involved are specific, easy-measurable numbers such as prices or volatilities that leave no doubt concerning their reliability. At the moment, finance does not often use latent or intangible variables that are difficult to measure, but may inherently carry higher validity in terms of explaining actual decision-making by financial market participants. Contrastingly, other social science disciplines such as social psychology or marketing have more frequent contact with 'soft' data attempting to measure intangible concepts such as mood or risk perception. These other social sciences therefore have a great deal to offer finance in terms of behavioural research. As a matter of fact, to date standard financial theory and methodology is poorly equipped to examine what is most important to understanding actual investors' decision-making: the latent behavioural characteristics of market participants. As a result of this lack of tools to study individual level behavioural characteristics of market participants, behavioural finance focuses on pricing anomalies as measured by easy-to-observe pricing data, whereas the true area of interest should be the drivers of the behaviour instead of the aggregate outcomes. As these drivers of behaviour steer investment decisions, which in turn drive market prices and price anomalies, it is of utmost importance to identify and understand them. Solely examining market anomalies using price level data does not elucidate the behaviour of market participants, while examining the (drivers of) behaviour of individual market participants does elucidate market anomalies. The third objective of this research therefore is to enrich existing finance research with an approach that builds explanations from the individual latent characteristics up to the market price behaviour instead of the 'top-down' approach currently used. To date, the project yielded very promising results in support of reorienting behavioural finance research from 'top-down' to 'bottom-up'. Our results confirm that the previously defined objectives are met. We provided a new theoretical contribution - investor sentiment as affect - and tested its empirical validity. The results confirm our model on investor sentiment and serve as the first successful example of the 'bottom-up' approach and its related methodologies in behavioural finance. The affective constructs used to determine investor sentiment were successfully measured from the battery of surveys we sent to retail investors. Investor sentiment is composed of affective constructs that influence risk-taking behaviour in financial markets. The empirical model supported the theoretical model, thus validating our definition of investor sentiment as affect. Investor sentiment is found to drive decision-making under risk, and hence financial decisions. The affective constructs measured are mood, optimism, confidence, and susceptibility to affect richness. These constructs constitute investor sentiment, which we theorize is one of the drivers of risk-taking behaviour. We measure risk-taking behaviour by gauging the risk attitudes and risk perceptions of investors. Financial decision variables, generally trade decisions, are found to be driven by risk-taking behaviour. Hence, investor sentiment drives risk-taking behaviour which, in turn, drives individuals' investment decisions, and eventually overall market prices. The findings support our conjecture that markets are composed of investors that balance rationality and irrationality. Contrastingly, standard finance theory divides markets into rational and irrational investors. Our findings regarding, for example, investors' selling behaviour support previous findings from standard finance but offer a more in-depth explanation and understanding of these phenomena. Notably, our findings support the disposition effect, which represents investors' behaviour to sell winners too early and hold on to losers too long. Additionally, we examined a selection of different dependent variables, such as investors' return performance, the riskiness of their portfolios, the diversification of their portfolios, specific choice of asset class, and investors' preference for cash versus stock dividends and related this to their individual measures of sentiment and risk-taking behaviour. Moreover, we examined the cross-sectional relationships of this study in a longitudinal fashion, as we performed monthly investor surveys. We find, among other things, that the drivers of individual investor sentiment, such as mood and optimism, change over time in correspondence with the overall market and impact investors' risk-taking behaviour and subsequent investment decisions. When investors become more optimistic, they also become more risk-seeking and trade more and more aggressively. Additionally, these relationships change over time and across different investment vehicles. During our sample period, which includes the current subprime crisis, investors moved back and forth from stocks to options to bonds and back, depending on overall market conditions and individual level sentiment as expressed by its main drivers. As detailed above, the project has contributed extensively to three types of related but distinguishable scientific achievements: (1) a new theoretical framework on investor sentiment, (2) a new empirical framework on investor sentiment and its impact on risk-taking behaviour and investment choices, and (3) a new approach for research in the field of behavioural finance. These achievements have been documented in several presentations for academic and professional audiences in marketing and finance and research reports currently under review at major academic journals. The results of the sentiment project will result in a further elaboration and development of finance research and help to understand the behaviour of individual investors, overall markets, and the relation between these entities. A better understanding of these issues can help to understand, predict and possibly prevent stock market anomalies such as hypes and crashes by recognising the key drivers of micro level investor behaviour and macro level stock market price developments.
