پیش بینی های مالی در طول بحران: آیا کارشناسان دقیق تر از مردم عادی بودند؟

The main goal of this paper was to examine the accuracy and confidence of financial forecasts during the 2009/2010 crisis. The study was carried out in February 2009 in Poland. The participants represented two groups: financial analysts and laypeople (people without knowledge or skills in finance). All participants were asked to forecast future stock market performance and foreign exchange rates. Additionally, they marked their confidence on a 100-point scale. The results showed that the forecasts significantly differed from the real values. In forecasting both the stock market and the currency exchange market, the prediction error significantly differed from zero. Even if the participants were optimistic in making the directional stock market forecasts, they were pessimistic when making point index predictions, which suggests a judgmental paradox. The experts were slightly better than the non-experts in predicting the stock market. However, their accuracy was generally not better in the exchange market forecasts. The next step of the analysis focused on the confidence factor. The results of this part of the research showed that the laypeople were less confident than the experts in all the judgments.

The financial crisis that struck the world markets in the first decade of this century is considered by many economists as the worst crisis since the Great Recession of the 1930s. It caused countless negative consequences for the global economic system, leaving its mark also on the psychology of financial markets, i.e. the way people make their financial judgments (Akerlof & Shiller, 2009). For example, analysts who compare economic and psychological features of the crisis argue that making predictions has become much more difficult (El-Erian, 2008 and Gärling et al., 2010). Kotler and Caslione (2009) claim that in ‘the new economic reality’, regular economic cycles are absent, and upturns (booms) or downturns (recessions) are unpredictable. Therefore the main goal of the research presented in this paper was to examine both the accuracy and the confidence of financial forecasts produced by experts (financial analysts) and laypeople during the recent crisis. The fact that the study was conducted at the peak of the depression (February 2009) makes it unique among previous studies on financial forecasts, because almost all of them were carried out at bull market. To my best knowledge, only one study tested directly how the stock market crash caused by the September 11 terrorist attacks influenced financial predictions (Glaser & Weber, 2005). This might mean that at least some psychological effects associated with the stock market judgments that were discovered in empirical studies are specific to propitious economic conditions. One of the examples can be the overconfidence fallacy considered as a universal feature of different financial judgments (see Shefrin, 2000, Wärneryd, 2001 and Zaleskiewicz, 2006, for a review). If people are aware that market predictability is lowered by the crisis they should forecast the market with limited confidence. Moreover, laypeople who have limited knowledge about the causes and consequences of the market downturn (as compared to experts) should be aware they are particularly prone to making erroneous financial forecasts. The study presented in this paper has not been a pioneer attempt to test the correctness of financial beliefs. Several authors examined the accuracy of both choices and predictions made by stock market experts. Ericsson, Andersson, and Cokley (2005), who reviewed a large body of research on financial expertise, conclude that some professionals are able to make superior portfolio choices by identifying underevaluated stocks. Likewise, Sundali and Atkins (1994) reported that security market experts did outperform both market averages and randomly thrown darts. On the other hand, classic and more recent analyses reveal that mutual funds employing skilled investment managers (experts) generally underperform the market ( Jensen, 1968 and Malkiel, 2003). Research on financial forecasting seems to be more consistent in revealing that financial analysts do not make more accurate probabilistic forecasts of stock prices than people without knowledge about stock market (Stael von Holstein, 1972, Tyszka and Zielonka, 2002, Yates et al., 1991 and Önkal and Muradoglu, 1994). In a relatively recent study, Törngren and Montgomery (2004) compared the accuracy of the stock performance predictions in two groups: financial professionals and laypeople. While the average accuracy in the group of students was 51%, the mean correctness in the group of professionals equaled 40%. In other words, experts scored worse than chance. The empirical background outlined above indicates that the quality of stock market forecasts produced by experts is rather low, which seems to support the assumptions of the efficient markets theory (Fama, 1991). However, this does not necessarily mean that judgments in all areas of finance are incorrect. Önkal, Yates, Sigma-Mugan, and Öztin (2003) compared the accuracy of forecasting exchange rates between expert traders and sophisticated amateurs. They found that professional accuracy usually surpassed non-expert accuracy. This result implies that differences between financial experts and laypeople may depend on the subject matter to be predicted and that professional judgments may be more successful for some markets than for others. Financial experts are not an exception in committing forecasting errors. Empirical findings in the field of the psychology of expertise suggest that professionals representing different disciplines rarely generate predictions whose quality exceeds the accuracy of judgments made by laypeople (Camerer & Johnson, 1991). A large body of research demonstrated a poor quality of expert judgments in such fields as: sport (Andersson et al., 2005 and Andersson et al., 2009), macroeconomics (Mills & Pepper, 1999), business (Aukutsionek & Belianin, 2001) or politics (Tetlock, 2006). Quoting the title of the paper by Camerer and Johnson (1991), one might ask: “How can experts know so much and predict so badly?” Two different approaches are offered as a solution of this paradox (Shanteau, 1992). The first of them, typical for the field of behavioral decision making (Kahneman, Slovic, & Tversky, 1982), assumes that experts commit the same cognitive errors as novices and this is why they do not produce better predictions. Examples of such errors are: anchoring (Tversky & Kahneman, 1974) and non-mean-reverting expectations (Kahneman & Tversky, 1972). The second approach proposes that the limited quality of expert judgments might be determined by specific task characteristics (Shanteau, 1992). Professionals representing certain disciplines act in a very dynamic environment, where stimuli involve human behavior, and conditions are highly changeable and barely predictable. This last point seems to be confirmed by results showing that experts who work in a more stable environment, e.g., weathermen (Murphy & Winkler, 1984), are able to make judgments of relatively high accuracy. According to Shanteau (1992), people professionally associated with the stock market, e.g., financial analysts, represent a poorly predictable area. If this is true, the quality of financial forecasts, especially at the economic crisis, should be low. The analysis presented in this paper focuses not only on the accuracy of predictions but also on the confidence of these predictions. Previous research showed that experts from different fields tend to make their judgments with too much confidence (Ayton, 1992 and Bolger and Wright, 1992). However, some experts were found to be well-calibrated, which means that in their case the correlation between confidence and accuracy was both positive and high (Keren, 1987 and Murphy and Winkler, 1984). This suggests that committing the error of overconfidence might be typical only for certain areas of expertise and depend either on the features of the expert discipline (Allwood & Granhag, 1999) or on the characteristics of the task that must be solved (Gigerenzer et al., 1991 and Klayman et al., 1999). Research on financial expertise (e.g., Tyszka and Zielonka, 2002 and Törngren and Montgomery, 2004) shows that low quality of financial forecasting is typically accompanied by overconfidence that leads to excessive trading and lower earnings (Barber & Odean, 2000). However, as mentioned above, these studies were carried out at the growth market, which might have increased the participants’ self-assurance. Consequently, one could assume that if both experts and laypeople participating in the present research were aware of the difficult economic conditions for making their predictions, they should not be too confident of their judgments.

The main goal of the study presented in this paper was to examine the accuracy and confidence of forecasts concerning the stock market and the currency exchange market during the 2009/2010 financial crisis. The analysis compared the predictions produced by experts (financial analysts) and laypeople. Recent research on financial expertise, conducted typically at bull market, has revealed that professionals hardly differ from amateurs in the accuracy of their index changes predictions. However, accuracy is not correlated with confidence: experts are much more confident of their beliefs than non-experts. In general, the results of the presented research are consistent with these findings, but some results seem to be specific for making judgments at the market downturn. First of all, the research showed that people from both groups made erroneous predictions. In almost all cases the value of error commited by the participants differed significantly from zero. It means that both laypeople and experts were inaccurate in forecasting not only the stock market indices but also the foreign exchange rates. In fact, the stock market changes predictions were the only case where experts made slightly better judgments than non-experts. Moreover, there was hardly any difference in how inaccurate the two groups were in forecasting the currency market. This latter result seems to be inconsistent with the findings reported by Önkal et al. (2003), who showed that financial experts were able to predict the currency market correctly. The reason for this inconsistency might be that the participants of the present study were making their forecasts at the crisis when currency markets are generally less predictable (El-Erian, 2008 and Kotler and Caslione, 2009). As pointed out by Shanteau (1992), factors enhancing instability of the market account for low quality of judgments even if they are made by experts. The analysis of results concerning the stock-market predictions suggests a possibility of the judgmental paradox. The participants were optimistic in predicting the direction of the market change but at the same time pessimistic in forecasting the precise value of the growth. In other words, they were convinced that the market course was bound to reverse but did not expect such a strong increase of the index. The actual percentage change of the index was much greater compared to what the participants predicted. It seems that two explanations might be offered as possible interpretations of this paradox. The first suggests that the two tasks – to predict the general direction of the market change and to make precise forecast – vary considerably in their complexity. To forecast the directional development is a much easier task not only for laypeople but also for experts. Similar effects were found by Andersson et al. (2009), who examined forecasting results of sports competitions. They found that participants representing different levels of knowledge were good at making easy judgments, but only some of them performed better in more complex tasks. The second explanation proposes that two independent effects might have affected people’s judgments: mean reverting expectations on the one hand, and the perception of the nature of crisis on the other. As Glaser and Weber (2005) documented, people who made financial forecasts at the market crash after September 11 terrorist attacks, revealed mean reverting beliefs. According to both authors, this result suggests that investors interpreted the drop in share prices as temporary rather than permanent. The 2009 financial crisis did not result from an unexpected singular event but was related to a severe economic recession that began in 2007. The awareness of a more persistent nature of the present market downturn might have increased people’s pessimism or at least inhibit optimism. This suggests that people differentiate between various types of market crashes and adapt their expectations appropriately. The last result concerning the forecasting accuracy that seems worth discussing is that the dispersion of errors was similar for experts and laypeople (see standard deviations shown in Table 2 and Table 3). In other words, consistency of opinions among experts was not higher than among non-experts. As Einhorn (1974) suggested in his widely cited paper, between-expert agreement is necessary for expert competence. If we assume that crisis increases uncertainty and unpredictability of the market, this high dispersion of errors is not so surprising and does not necessarily suggest expert incompetence. This result indicates that future research on financial expertise should focus not only on how the changing market conditions affect the accuracy of forecasts but also the disagreement they cause in professional judgments. This study, similarly to other studies on expert judgment, tested not only the accuracy but also the confidence of market predictions. Recent research suggested that financial professionals tend to be overconfident in forecasting changes in the stock market (Tyszka and Zielonka, 2002, Törngren and Montgomery, 2004, Yates et al., 1991 and Önkal and Muradoglu, 1994). In this sense, they do not differ from experts representing other fields, whose judgments were found to be equally overconfident (Ayton, 1992 and Bolger and Wright, 1992). The results collected in this research showed that the financial analysts were more confident than the amateurs when making both the stock market and the exchange market forecasts. Moreover, the changes in confidence were much greater than changes in accuracy and even if the difference in accuracy was not significant, the confidence level among the experts was still higher. These outcomes are in contrast with the assumption that making judgments during a financial crisis decreases confidence. Stock market experts seem to be very confident of their knowledge and skills irrespective of positive or negative economic conditions.