T HE M EASUREMENT OF B EHAVIORAL F ACTORS ON C HOICE OF F UND IN U NIT T RUST F UND I NVESTMENT : A N E XPLORATORY F ACTOR A NALYSIS

The objective of this study is to develop items for the choice of the fund in unit trust fund investment in Malaysia by looking at the behavioural factors influencing the retail investors. The validity and reliability test have been applied to all the constructs. The respondents were from Kelantan, Terengganu, Penang and Kuala Lumpur. The data were analyzed using SPSS. Exploratory factor analysis was used to group the items in the process of measuring the following dimensions: (i) Snake Bite Effect; (ii) Past Success; (iii) Advice and information; (iv) Illusion of Knowledge and (vi) Judgment of fund. The results of the analysis highlighted the development of three primary underlying variables, namely considering the past, financial knowledge and miscalibration. These instruments can be used by unit trust companies, the authorities of unit trust and other banking’s product to understand the behaviour of their clients and later, can implement it to the expected new customers, and it will provide a glimpse on how behavioural factors influence the choice of fund.


Introduction
The unit trust fund is structured shared investments with investors that have the same objectives to contribute to the funds to invest in a portfolio of securities or assets (Gan, 2008).The fund is managed by professional fund managers and invests in a portfolio of a fund that may include cash, bonds and deposit, shares, properties and commodities.In Malaysia, the right to the fund is according to the units own as the fund is broken down into units (Gan, 2008).
Federation of Investment Managers Malaysia (2014) indicates that Malaysia has 42 unit trust management companies, 56,202 unit trust consultants together with 441 conventional funds and 190 Syariah funds to choose from as of December 2015.The net asset value for the conventional fund is RM 294,454 billion and Islamic based fund is RM 52,124 billion which is 20.45% of the net asset value of Bursa Saham Malaysia as of 31 December 2015 (Federation of Investment Managers Malaysia, 2016).The projection penetration rate for unit trust fund according to Securities Commission Malaysia ( 2014) is likely to be a double-digit growth from 18% in 2010 to 34% in 2020, which is almost the same rate observed in the developed countries.In investing unit trust fund, investors need to choose as there are choices.
According to the classical finance, the underlying assumption of individual behaviour which is an economic agent is to possess absolute criteria.Known as homo economicus, he should be able to maximize all the possible options, completely understand all the outcomes, and will only consider taking actions based on these results.Echoing the same idea, Simon (1997) mentions that homo economicus must have a super brain, and can process raw data, solve complex problems and have an exceptional mathematical mind.His brain is never full and is immune to cognitive and emotional error (Statman, 2017).Ackert and Deaves (2010) concurred that the behaviour of homo economicus is attributed to Mill (1874), who views the human as a person who wants to accumulate wealth and minimize cost and expenditure.Referring to Statman (1999), human in classical finance are rational, but in Behavioral Finance, humans are normal.Statman (2017) recognized that people are not irrational but mostly intelligent and "normal smart".Normal people do not intend to be ignorant or to commit a cognitive and emotional error.Instead, they act to get utilitarian, expressive and emotional benefits they want (Statman, 2017).Utilitarian benefits answer to the question "what does something do for me and my pocketbook?"While expressive benefits convey to us and others our values, taste and social status (Statman, 2017).Emotional benefits answer the question "how does something make me feel?" Sometimes, normal people are "normal foolish", misled by cognitive and emotional error.So, Behavioral Finance explains financial market anomalies by studying the actual behaviour of investors.
Considering the past, financial knowledge and miscalibration were used to look at how investors choose their fund.These three behavioural factors will be used as it incorporates the usual way of decision making concerning about choice (Thaler and Johnson, 1990).

Considering the Past
Using past outcomes as a factor in evaluating the current uncertainty commonly used by many investors is known for considering the past (Nofsinger, 2005).The discovery by Thaler and Johnson (1990) was that snake bite effect, and past success was the behavioural factors to consider if an investor considers the past in making a financial decision.
 Snake Bite Effect (Past Investment Losses): According to Nofsinger (2005), following a financial loss, investors become averse to risk.It is known as the snake bite effect.After losing money and are faced with a gamble, investors will choose to decline to take the risk.Snakes seldom bite people, but when they bite the victims become more mindful.
After feeling hurt and as long as the feeling of unluckiness is there, they will avoid risky adventure; therefore, they herd (Nofsinger, 2005). Past Success: As claimed by Nofsinger (2005), past successes are where people have experienced a gain or profit from an investment.Investors will always try to minimize the impact of their poor return by always highlighting their excellent return and by taking this step, will end up overrating both their past return and their potential future performance of their account (Wright, 2000).Overestimating their investment return is a common occurrence.
In evaluating risky decision today, investors are using the past outcome as a yardstick.After a profitable investment, the profit they earn is never fully considered their own money (Nofsinger, 2005).In an experiment, Nofsinger (2005) found that 77% of the economic student would continue betting after winning $15.00.After losing $15.00,only 41% chose to gamble.

Financial Knowledge
Statman (2017) grouped knowledge into three types in the context of finance.Type one is about financial facts knowledge.This knowledge is about financial markets, bonds, stocks and other investment instruments.It is also about diversification, the drawbacks of investment fees and the difficulty of beating the market (Statman, 2017).Human behaviour knowledge is about our wants, ways the investors make a decision such as cognitive, emotional shortcut and the errors investors make.It about wants such as social status, adherence to value and for riches.According to Statman, (2017), there are errors in framing, hindsight and emotional shortcuts and errors such as hope, pride, regret and fear.
The third is information knowledge and divided into three types according to Statman (2017).Exclusively available information is private or inside information.It is only available to among the administrator of the company such as directors and managing director of the companies.Narrowly available information is only available to a few managers and executive of the company.It is about information before it is announced to the public.Widely available information is available to everyone, but it does imply that everyone knows that information (Statman, 2017).
 Advice and Information: Investors invest based on information or advice they received and observed in Prospect Theory where the reaction of an investor depends on a subjective reference point.Forbes (2009) stated that one of the reasons why investors tend to agree is because of the nature of the advice they received from stockbrokers and market commentators.From 1989-1994, Welch (2000) reviewed 50,000 recommendations issued by 226 brokerage house by using US Zacks database.There was substantial evidence that investors followed recommendations towards the current consensus regardless of the underlying distribution of advice given.It is a well-known phenomenon known as "white coat effect" according to Desteno (2015) where ones believe that the person who gives advice is an expert or person with authority. Illusion of Knowledge: Varying levels of actual knowledge versus perceived knowledge, combined with an unrealistic expectation, according to Suzanne ( 2012) is creating sizable barriers to healthy decision making.Nofsinger (2005) claimed that the illusion of knowledge referred to the inclination for people to believe that the precision of their forecasts increases with more information; with more information increases one's knowledge thus improves one's decision.Partially, overconfidence is the result of the illusion of knowledge (Nofsinger, 2005).
Based on Nofsinger (2005), investors who have access to the internet can access information at a faster speed.They have raw data but most retail investor's lack of training, knowledge and experience on how to interpret and analyze it.It needs specialized skills, and it may be very costly to process raw data and yet to be able to use this information will lead to a price change which is either profitable or loss (Nofsinger, 2005).It is beyond the reach of many non-professionals as discovered by Nofsinger (2005)

Miscalibration
Miscalibration is a strain of overconfidence.Ackert and Deaves (2010) define miscalibration as the tendency for an investor to overestimate the precision of their knowledge.Shiller (1997) said that overconfidence is associated with investors' judgment, underestimating the margin of error likely to be committed.Graham et al. (2009) argued that people are more willing to bet on their judgments when they feel skilful or knowledgeable.Hilton et al. (2011) showed that the core finding that supports the reality of judgmental overconfidence is that people are miscalibrated.They overestimate the probability of their judgment to be correct.Miscalibration depends on the way it is measured.In particular, higher confidence is observed in the interval production task where participants were asked to state an interval such that they were XX% (e.g., 50%) sure that the correct response to the questions fell in that interval.

Methodology
The respondents were sampled in the state of Kelantan, Terengganu, Penang and Kelantan.They are all Malaysian retail investors of a unit trust fund with at least a year of investment experience.600 questionnaires were distributed, and only 333 were deemed suitable.The questionnaire was developed through adaptation of the previous study.It is as shown in Table 1.
All the items were structured based on five items of Likert Scale ranging from 1 'strongly disagree' to 5 'strongly agree', except for Understand the Investor.After that, it was checked by the expert and discussed with selected respondents for its face validity test.The questions were developed and distributed to the respondents, and the data will be analyzed using exploratory factor analysis.

Data Analysis
Data analysis was carried out through SPSS.It helps to facilitate data cleansing, screening and checking logical inconsistencies.In order to achieve the objective of this study, Factor analysis is used as it is capable of decreasing a large number of underlying items into a smaller or convenient set of factors and it is used to study construct validity.Thus, factor analysis is used to identify the items, which measure the essential underlying variables.According to Sekaran (2000), it needs a sample size of at least 300 respondents to provide a stable factor solution.
There are some steps involved in factor analysis.The first step is to determine factorability of the data by using Measure of Sampling Adequacy (MSA) with a value of more than 0.6, Kaiser-Meyer-Olkin (KMO) with a value of more than 0.5 and Bartlett's test of sphericity (BTS) which is significant (Pallant, 2005).
The number of factors to be used will depend on the extraction factor, and the scree plot will help to determine the eigenvalue by looking at the natural bend in the data as the curve flattens out (Costello and Osborne, 2005).To ensure significant factors to be selected, the only eigenvalue over one is selected (Pallant, 2005).Costello and Osborne (2005) acknowledged that the goal of rotation is to simplify and clarify the data structure.It cannot improve the fundamental aspects of the analysis, such as the amount of variance extracted from the items.Rotation according to Vogt (1993), is methods in factor analysis by which the researcher attempts to relate the calculated factors to theoretical entities.It is done differently depending on whether the factors are believed to be correlated (oblique) or uncorrelated (orthogonal).The more helpful is Yaremko et al. (1986, pp. 86), who define factor rotation as follows: "In factor or principal-components analysis, rotation of factor axes identified as the initial extraction of factors, to obtain simple and interpretable factors."Tabachnick and Fiddell (2007) argue that perhaps the best way to decide between orthogonal and oblique rotation is to request rotation (e.g.direct oblimin or promax from SPSS) with the desired number of factors (Brown, 2009) and look at the correlation among factor.If the data do not drive factor correlations, the solution remains nearly orthogonal.If correlation exceeds 0.32, then is 10% (or more) overlap in variance among factors, enough variance to warrant oblique rotation unless there are compelling reasons for orthogonal rotation.Varimax rotation is by far the most common choice according to Costello and Osborne (2005).There is no widely preferred method of rotation; all tend to produce a similar result (Fabrigar et al. 1999).
The final step is the definition of the factors.An item with loading higher than 0.6 is chosen to represent a factor (Tabachnick and Fiddell, 2001).The following steps above will be followed.

Snake Bite Effect (Past Failure Experience)
Eight items were used to measure the respondents' behaviour on snake bite effect on the choices of funds.To test whether factor analysis was appropriate, Kaiser-Meyer-Olkin measure of sampling adequacy (KMO), Measures Sampling of Adequacy (MSA) and Bartlett's Test of Sphericity (BTS) was used.The KMO was 0.657 exceeding the required value of 0.5.The BTS indicated a statistically significant correlation between items (p < 0.001) and MSA was more than 0.6, allowing the factor analysis to proceed.
The items were chosen to identify each factor were those with loading higher than 0.6 (Tabachnick and Fiddell, 2001).As a result, four items were omitted (Unit trust consultant and Co-investors and internet review and market sentiment).Factor analysis is then tested based on four items.The scree plot suggested a single-factor-solution with the first factor accounted for 49.523% of the variance.Table 2 captures the items to be used in the questionnaire with the statement of: "in deciding the choice of fund, my investment losses experience enlightened me to…" The Cronbach Alpha is 0.695.

Past Success
Seven items were used to measure past success.The MSA was more than 0.6, KMO was 0.838 with BTS chi-square of 699.139 (p<0.000),justifying a factor analysis.Two items were dropped (lowest cost fund, meet my religious need) due to factor loading (less than 0.6).All five items had strong loadings (exceeding 0.6) on only one underlying factor after rerunning the factor analysis.This factor explained that 47.45% of the variance in response.The scree plot also suggested a single factor solution.Table 3 captures the items to be used in the questionnaire by using the statement of: "in deciding the choice of fund, my past success experience will…" This shows that the snake bite effect and past success can be further analyzed and the items are appropriate in line with the objective of the study.

Advice and Information
Ten items were used to measure advice and information.Three items (friend, co-investors and internet review) were dropped because the factor loading was below 0.6.Factor analysis was rerun, and the MSA was more than 0.6, the KMO value was 0.803 with BTS of 712.164, df = 21 and (p<0.001).This information indicates that the variables are suitable for factor analysis.All seven items loaded into a single factor with an eigenvalue of more than 1.The single factor extracted 46.97% of the total variance in response.Table 4 captures the items to be used in the questionnaire by using this statement of: "in deciding the choice of fund, the advice and information come from:"

Illusion of Knowledge
For the illusion of knowledge, six items were prepared.The MSA was more than 0.6, KMO was 0.849 with a significance value for BTS (p<0.000)indicating that the data were suitable for factor analysis.An item (low-cost fund) was dropped due to the loading of less than 0.6.Factor analysis was run on the five items.One factor was extracted, and this factor accounted for 56.359% of the variance.The scree plot provides further support for the single factor solution, and it is appropriate.Table 5 captures the items to be used in the questionnaire by using this statement of: in deciding the choice of fund, the extra information will…" Financial knowledge that consists of advice and information and the illusion of knowledge can be further analyzed.Next, miscalibration will be factorized, and there is only one dimension.

Judgement of fund
Five items were used to measure the judgment of fund.The MSA was more than 0.6, KMO was 0.765 with a BTS chi-square of 448.842 (p<0.000)allowing a factor analysis.An item (low-cost fund) was dropped as it did not fulfil the factor loading of more than 0.6 and factor analysis was rerun on the five items.The extracted factors were with an eigenvalue of more than 1.Table 6 captures the items to be used in the questionnaire by using the statement of: "in deciding the choice of fund, my judgment of fund will…" Miscalibration which has only one item judgment of fund can be further analyzed.Choice of the fund will be factorized next.

Choice of Fund
Eight items were used to measure the choice of fund.The MSA was more than 0.5, KMO was 0.559 with a BTS chi-square of 262.184 (p<0.000)allowing a factor analysis.Two items (diversification of investment, choice from UTC) were dropped as it did not fulfil the factor loading of more than 0.6.Three factors were extracted with an eigenvalue of more than 1.Table 7 captures the items to be used in the questionnaire by using this statement: "my choice of the fund is:"

Conclusion
The items on the choice of the unit trust fund were validated by exploratory factor analysis and had achieved the requirement of the validity and reliability test.The items demonstrate how investors choose their fund which gives an impact on their decision making.
During decision making for choice of fund, investors will consider the above items as they are likely to be influenced by their own behavior.Their behavior will decide which fund to invest, reinvest and to sell.All these items attributed to the behavior of investors who are accumulating their wealth.These items can be used by unit trust management companies, The Federation of Investment Managers, Bank Negara, Securities Commission of Malaysia to understand the behaviour of investors in their choice of fund.By understanding the behaviour of the investor, it is much easier for the companies to market their products and help the government agencies to come out with the better legislature to protect the investors.

Table 2 . Summary of factor loading for snake bite effect
Snake bite effect has eigenvalue of 1.981, percentage of variance explained of 49.52% and KMO of 0.657.The Bartlett's Test of Sphericity Approx.Chi-Square 220.42, df =6 with Sig.= 0.

Table 7 . Summary of factor loading for the choice of fund
The Eignvalue is 1.068, the percentage of variance explained is 72.350% and the KMO is 0.559.The Barlett's Test of Spericity Approx, Chi-Square is 262.184,df = 15, Sig = 0.00.The Cronbach Alpha is 0.631.