Available online at www.sciencedirect.

com

Borsa _Istanbul Review

_
Borsa Istanbul Review 23-6 (2023) 1458–1473
http://www.elsevier.com/journals/borsa-istanbul-review/2214-8450

Full Length Article

A model validation of robo-advisers for stock investment

Atul Shiva a, Bijay Prasad Kushwaha b, Bikramjit Rishi c,*
a
Jaipuria Institute of Management, Noida, Uttar Pradesh, India
b
VIT Business School, Vellore Institute of Technology, Vellore, India
c
School of Management and Entrepreneurship, Shiv Nadar Institution of Eminence Deemed to be University, Delhi - NCR, India
Received 19 May 2023; revised 22 September 2023; accepted 22 September 2023
Available online 28 September 2023

Abstract

The study examines the intention of stock investors to adopt robo-advisers (also known as automated investing services) in financial in-
vestment decisions. Using an adapted questionnaire, we analyze data from 637 useable surveys. The study uses variance-based partial least
squares structural equation modeling to test our hypotheses. The study reveals that the critical drivers in determining the attitude of stock investors
toward the use of robo-advisers for decisions about investment in stocks are data security and perceived vulnerability. Further, prediction and,
judgment and behavioral biases are the performance constructs that are considered necessary by stock investors towards the use of robo advisers.
Our study has implications for financial advisers, brokerages, and practitioners in understanding the behavior of retail investors in adopting robo-
advisers.
Copyright © 2023 Borsa İstanbul Anonim Şirketi. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).

JEL classification: G23; G41; M15

Keywords: Financial services; Individual investor behavior; Partial least squares structural equation modeling; Robo-advisers

1. Introduction the client does not understand the advice given by the algo-
rithm or the advice given is not suitable (Mezzanotte, 2020;
Because of innovation and digitization, financial services Van Doorn et al., 2017). Robo-advisers have gaining legiti-
have experienced rapid technological disruption in recent years macy and are disrupting the financial advice and asset man-
(Grewal et al., 2020; Hashimy et al., 2022). The introduction of agement industries (Day et al., 2018; Fan & Chatterjee, 2020).
intelligent systems and algorithm-based investment decision- In emerging financial markets, such as India, it is a novel
making has opened new avenues for financial services in concept, hence, more research is needed to help stock investors
assisting and advising clients (De Gauquier et al., 2023; Hoyer adopt robo-advisers for decisions on financial investment.
et al., 2020; Puntoni et al., 2021). Robo-advisers enable the Furthermore, the diverse role of robo-advisers can be
extension of investment advice to investors without human designed to cater to the financial markets of various countries
contact (Huang & Rust, 2018; Phoon & Koh, 2017). In (Abraham et al., 2019; Hohenberger et al., 2019). The risk
financial services, this technology can replace human interac- associated with technological development is unavoidable for
tion in investment advice and consulting (Burton, 2014; Jung retail investors. The risks associated with fintech are misuse of
et al., 2018). In some cases, human interaction is required if data, price discrimination, and environmental and social risks.
Technological advances, such as robo-advisers, may help to
reduce such risks (Horn et al., 2020, pp. 309–327). One
* Corresponding author. drawback of robo-advisers is their inability to understand in-
E-mail addresses: [email protected] (A. Shiva), bijayrsm@gmail. vestors emotional and economic concerns (D'Acunto et al.,
com (B.P. Kushwaha), [email protected] (B. Rishi). 2019). They can provide rational advice without considering
Peer review under responsibility of Borsa İstanbul Anonim Şirketi.

https://doi.org/10.1016/j.bir.2023.09.005
2214-8450/Copyright © 2023 Borsa İstanbul Anonim Şirketi. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

the level of investors' ability to invest. The most crucial AI-enabled robo-advisers in investors' financial decisions is
advantage of a robo-adviser is that it offers advice on how to primarily confined to qualitative research or experimental
maximize wealth (D'Acunto et al., 2020; Rossi & Utkus, 2020). research designs (Bhatia et al., 2021; Pal et al., 2021). This
Although robo-advisers are in their infancy in developing aspect needs to be explored with a comprehensive quantitative
countries, such as India, several studies support expansion in structural model. In order to address these gaps, we provide a
the scope of robo-advisers in the financial sector (Blanche conceptual framework on the use of robo-advisers in stock
et al., 2019; Hohenberger et al., 2019; Zheng et al., 2022). investors' investment decisions. Our proposed model takes into
Few papers have covered the Indian context with respect to the account retail investors' perceived threat of adopting AI-
use of robo-advisers by retail investors (Rasiwala & Kohli, enabled technologies in emerging financial markets like
2021). According to a report by Statista (2019), the total India. The present study thus intends to focus on the following
amount of assets under management in the robo-adviser research question.
segment was US$1 trillion in 2021 and was projected to be
RQ1. What drives stock investors' attitudes and intentions to
$2.67 trillion in 2023. Numerous studies have been conducted
use robot advisory services for investment decisions?
to determine the scope, opportunities, acceptance, and chal-
This paper is structured as follows: Section two explains the
lenges of robo-advisers enabled by artificial intelligence (AI)
literature review and hypotheses development. Section three
involved in harnessing the power of computing to enhance
describes the research methodology, followed by results and
human insights (Arikan et al., 2023; Huang & Rust, 2018; Luo
analysis in Section four4. In Section 5, we discuss the theo-
et al., 2019; Ma & Sun, 2020; Marchand & Marx, 2021; Van
retical and managerial implications. The study concludes in
Doorn et al., 2017). Further, investors tend to rely on
Section 6 with limitations and suggestions for future research.
recognition-based heuristics-driven biases, resulting in irratio-
nal financial decisions (Ahmad, 2021; Dzyabura & Hauser,
2011). Hence, replacing traditional financial advice with on-
line services is a highly fruitful area of research (World Bank, 2. Literature review and hypotheses development
2015).
Many different AI and financial management-oriented ap- 2.1. Theoretical underpinning
plications are available, but prior studies indicate that AI-
enabled technologies focus primarily on behavioral and The literature on information systems (IS) on measuring
cognitive dimensions of the customer experience (Liu- individual attitudes and acceptance level of a specific appli-
Thompkins et al., 2022; Wang et al., 2022). Thus, more cation is extensive. In this regard, various theories have been
attention should be paid to the emotional and social compo- developed to predict the acceptance and use of newer tech-
nents of AI-enabled technologies for enriching the customer nologies. Theories such as the technology acceptance model
experience (Lemon & Verhoef, 2016). For example, with re- (TAM), the unified theory of acceptance and use of technology
gard to AI-enabled agents and human interactions, customers (UTAUT), and the extended unified theory of acceptance and
purchased less when they knew they were interacting with a use of technology (UTAUT2) have been integrated with other
chatbot (Luo et al., 2019). Moreover, service robots cause theories to offer adequate predictive power for adopting
higher psychological discomfort than human service providers emerging technologies (Venkatesh et al., 2003). The main
(Mende et al., 2019). Yet other studies report that the use of AI- findings from this integration of theories suggest new exoge-
enabled robots had a positive impact on shopping (Song et al., nous and endogenous mechanisms (Venkatesh et al., 2016)
2022; Song & Kim, 2022). with new outcome mechanisms. Our study considers user in-
Oehler et al. (2022) and Eren (2023) focus on the intention tentions to adopt AI in investment decisions in financial mar-
to use robo-advisers for investment decisions and the charac- kets. Thus, IT adoption in terms of AI in financial markets is
teristics of investors. Oehler et al. (2022) study undergraduate based on the tenets of systemic literature review studies by
students with minimal experience and knowledge about real Venkatesh et al. (2016) and Williams et al. (2015). Research on
investment, with investment funds provided by someone else. AI adoption stems from decision support systems, which as-
Therefore, their risk-taking ability increases, as they will not sumes that an effective man-machine symbiotic relationship
lose their money. Eren (2023) examines the intention to use can result in more intelligent actions (Syam & Courtney, 1994,
robo-advisers in private pension investment among experi- p. 450). However, models such as UTAUT and TAM focus
enced investors. Unlike these studies, the present study con- primarily on functional technologies and do not cater to the
centrates on the intention to use robo-advisers for investment in complex decision-making process involved in AI adoption
stocks, which are very volatile. Hence, this study answers our (Gursoy et al., 2019).
research question (RQ1) and contributes to the literature by Further, factors such as risk, trust issues, data security, and
examining the antecedents of the intention to use robo-advisers prediction errors are involved in adopting AI in financial
in stock investment decisions. decision-making by investors (Balakrishnan & Dwivedi, 2021;
These observations indicate a clear research gap and call for Vimalkumar et al., 2021). The technology threat avoidance
a study on the adoption behavior of prospective stock investors theory (TTAT) is the most widely used model for addressing
interested in using AI-enabled robo-advisers (Hohenberger these factors (Liang & Xue, 2009). Therefore, we focus on new
et al., 2019; Zheng et al., 2022). Furthermore, research on exogenous variables related to investor adoption of AI-based
1459
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

robo-advisers. Additionally, we cover factors that influence AI Hypothesis 1a. Trust positively influences attitudes toward
threat avoidance behavior to address individual responses to IT using robo-advisers.
threats in voluntary settings (Cao et al., 2016). TTAT empha-
sizes threat perceptions of negative consequences for users based
Hypothesis 1b. Trust positively influences the intention to use
on perceived severity and susceptibility. The theoretical frame-
robo-advisers.
work of this study covers both positive and negative impacts of
the adoption of AI (Breward et al., 2017) in financial decision-
making by investors in emerging financial markets. 2.2.2. Awareness, attitude, and intention to use robo-
In order to develop a detailed understanding of AI accep- advisers
tance and avoidance, we also use cybernetic theory. This theory In financial markets, the use of technology in various of-
explains that humans use feedback loops to regulate their own ferings has gradually increased (Gan et al., 2021). Consumers
behavior. Thus, when investors use robo-advisers for financial with a high level of financial acumen may be more aware of
advice on investment in financial markets, their behavior how robo-advisers operate with unbiased AI algorithms, which
comprises positive and negative feedback loops. The positive may increase their willingness to use financial robo-advisers
feedback loop of a cybernetic process refers to the distance (Bhatia et al., 2021). Further, robo-advisers may communi-
between the present state and the desired state, in which a large cate the reliability and accuracy of their advice to enhance
difference is significant in causing discontinuity behavior potential investors' knowledge and awareness of automated
(Carver, 2006) of AI. The negative feedback loop explains that financial robo-advisers (Bhatia et al., 2020). Fulk et al. (2018)
the behavior is disrupted if a discrepancy is discovered, and the state that investors with lower income and net worth, with less
current condition is not mapped to the anticipated end state impulsiveness financially, intended to use robo-advisers. Tech-
(Carver & Scheier, 1982) in financial markets. Thus, we inte- savvy investors see robo-advisers as a viable and cost-effective
grate cybernetic theory with TTAT to study investors’ intention way to obtain investment management services (Chandani
to adopt robo-advisers for investment in financial markets. et al., 2021). In light of these findings, we propose the
The theoretical framework comprises a human-centered following.
approach to explain the symbiotic AI-human relationship in
Hypothesis 2a. Awareness positively influences attitudes to-
the research model. The proposed model includes human per-
ward using robo-advisers.
ceptions such as trust, prediction, awareness levels, and secu-
rity issues. The positive and negative aspects can influence
investor attitudes and behavioral intentions to consider AI in Hypothesis 2b. Awareness positively influences the intention
financial decisions. Building on prior publications on tech- to use robo-advisers.
nology adoption and diffusion, this theoretical framework
shows a new dimension in AI adoption and threat perceptions
2.2.3. Behavioral bias, attitude, and intention to use robo-
by stock investors in IS and behavioral finance.
advisers
Robo-advisers employ preprogrammed algorithms. So, if
2.2. Conceptual framework
the algorithm fails, the developer has fed an incorrect algo-
rithm, or the presence of a bug or an erroneous update can
2.2.1. Trust, attitude, and intention to use robo-advisers
present a high level of risk for investors (Bhatia et al., 2021).
Trust is defined as the propensity to be susceptible to ex-
Aside from the programmer's competence, the program's
pectations and beliefs in an innovation (McKnight et al., 2011).
design and outcome could be affected by potential programmer
Some studies observe a positive relationship between trust in
bias and intent (D'Hondt et al., 2019; Uhl & Rohner, 2018). As
technology and usage behavior (Dimitriadis & Kyrezis, 2010),
a result, the behavioral biases of robot developers can be
which can be critical in adopting robo-advisers. The perception
passed on to programs (Baker & Dellaert, 2017). Additionally,
of trustworthiness and trustful communication is considered
potential human error by the programmer during initial coding
important in forming a positive attitude toward robo-advisers
or updates can lead to behavioral bias (Bhatia et al., 2020;
(Cheng et al., 2019; Jung et al., 2018). In one research study,
Creswell & Poth, 2017). At the same time, robo-advisers can
trust in institutions predicts trust in financial technologies
lead investors to take a passive investment approach and enable
(Kundu & Datta, 2015). Another research study observes trust
informed, less-biased investment decisions (Lisauskiene &
in financial institutions, structural assurance, and perceived
Darskuviene, 2021). Further, Capponi et al. (2022) show that
risks as significant predictors of adopting robo-advisers for
robo-advisers can facilitate timely information for investors
financial asset management (Jung et al., 2018). These robo-
and thus mitigate behavioral bias toward risk profiles in
advisers can establish a track record of consistency by
financial markets.
providing better advice and returns to investors to gain trust
Thus, we posit.
and eliminate the need for human intervention (Bhatia et al.,
2021; Woodyard & Grable, 2018). Based on these studies, Hypothesis 3a. Behavioral bias negatively influences attitudes
we present the following hypotheses. toward using robo-advisers.

1460
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

Hypothesis 3b. Behavioral bias negatively influences the Hypothesis 5a. Prediction and judgment positively influence
intention to use robo-advisers. attitudes toward using robo-advisers.

2.2.4. Cost effectiveness, attitude, and intention to use robo-

Hypothesis 5b. Prediction and judgment positively influence
advisers
the intention to use robo-advisers.
The main advantages of robo-advisers are cost effectiveness
and efficient wealth management (Mezzanotte, 2020). Finan-
cial advice based on AI can be provided to large clients at a 2.2.6. Data security, attitude, and intention to use robo-
lower cost (Rasiwala & Kohli, 2021). Financial advisers prefer advisers
to use robo-advisers to target large customers on limited bud- Today, the majority of financial transactions take place
gets (Jung et al., 2018). Robo-advisers can tap the passive online or digitally. Data security is a significant concern in
market at a lower cost and offer investors automated invest- financial transactions, and investors expect complete protection
ment advice (Uhl & Rohner, 2018). Investors choose robo- of their personal and financial information (Bhatia et al., 2020).
advisers because they are cost effective, and such digital so- In this regard, investors’ perspective on data security is
lutions also enable them to achieve financial goals (Bhatia multifaceted, based on familiarity, relationships, and customer
et al., 2021). Online automated advice can cost effectively experience, which contributes to high security standards
meet new and existing customer needs (Lopez et al., 2015; regarding robo-advisers (Kumar & Gupta, 2020; Sander,
Singh & Kaur, 2017). Robo-advisers can achieve returns to 2021). Hence, robo-adviser firms must also ensure unrivaled
scale by assessing quality information cost effectively to help data security technologies that protect their customers' data
investors select better investment (Baker & Dellaert, 2017). (Beck, 2021). Robo-advisers and investment advice solutions
Furthermore, robo-advisers demonstrate superior perfor- for stock investors are in their infancy in developing econo-
mance and facilitate robust, cost-effective solutions for diverse mies, where financial security and financial efficiency enable
specifications of risk-to-reward models (Tao et al., 2021). financial innovation (Liu, 2018). Financial technologies based
Additionally, the service cost is the critical success factor of on regulated algorithms with higher information disclosure can
automated portfolio rebalancing services by robo-advisers facilitate efficient investment decisions (Guo, 2020). Thus, we
(Almadi et al., 2014; Horn et al., 2020, pp. 309–327). In line propose the following hypotheses.
with these findings, we propose the following hypotheses.
Hypothesis 6a. Data security positively influences attitudes
Hypothesis 4a. Cost effectiveness positively influences atti- toward using robo-advisers.
tudes toward using robo-advisers.
Hypothesis 6b. Data security positively influences the inten-
Hypothesis 4b. Cost effectiveness positively influences the tion to use robo-advisers.
intention to use robo-advisers.
2.2.7. Perceived severity and perceived threats
2.2.5. Prediction and judgment, attitude, and intention to AI-based financial asset management advice and its harmful
use robo-advisers repercussions raise considerable concerns (Shrestha et al.,
Human financial advisers can account for investor spending 2019; Weld & Bansal, 2019). The severity of employing
and stable income status, whereas robo-advisers only provide robot services relates to the potential for undesirable conse-
advice and recommendations (Capponi et al., 2022; Tokic, quences (Liang & Xue, 2009). The perceived severity of using
2018). But recommendations and advice by robo-advisers are robot services alerts investors to take precautionary measures in
assumed to be objective and highly structured (Climescu, 2021; making financial decisions (Liang & Xue, 2009). Furthermore,
Huang et al., 2023). Therefore, advanced access modeling can the extent of perceived severity of using robo-advisers is
help robo-advisers meet clients’ needs (Chen et al., 2021; individual-specific and varies from individual to individual
Phoon & Koh, 2017). AI-based robo-advisers can be pro- (Cao et al., 2021). The adverse outcomes of using robot ser-
grammed to predict uncertainty and the potential worldwide vices positively influence the perceived threat of using robot-
impact of events on financial markets (Bhatia et al., 2020; Jung generated advice. Thus, we propose the following.
et al., 2018). Robo-advisers handle specific jobs, but in-
Hypothesis 7. Perceived severity of using robo-advisers
novations over the next few decades will enable them to predict
positively influences the perceived threat of using them.
and offer consultation on overall financial matters (Beck,
2021). In this regard, robo-advisers and their transactions can
predict the accuracy of models for investment in stocks. They 2.2.8. Perceived susceptibility and perceived threat
can serve as another indicator of technical analysis in the future Perceived susceptibility refers to using unsafe automated
(Tsai & Chen, 2022). Robo-advisers can forecast stock market online services that may harm an individual (Ali et al., 2022).
volatility, but it is up to investors to make the final call on According to Cao et al. (2021), AI-based financial services can
investment and judgments (Bhatia et al., 2021; Sander, 2021). offer harmful financial decision-making recommendations.
Based on these premises, we posit the following hypotheses. Additionally, perceived threats may lead to avoidance of robo-

1461
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

advisers (Liang & Xue, 2009). For example, biometric-based Hypothesis 10. Attitude positively influences the intention to
SIM card issuance services involved great concern regarding use robo-advisers.
information security and safety, which caused a negative atti- The proposed conceptual model used in the present study is
tude toward their use. When perceived susceptibility to using illustrated in Fig. 1.
robo-advisers is minimal, individuals' perceived threat is
minimized and vice versa (Cao et al., 2021). Hence, perceived
3. Research methodology
susceptibility has a positive impact on perceived threats. Thus,
we propose the following hypothesis.
3.1. Sample statistics and survey
Hypothesis 8. Perceived susceptibility about using robo-
advisers positively influences the perceived threat from using This study is based on a survey conducted using a ques-
them. tionnaire (see Appendix Table A1) to examine the influence of
robo-advisers on investors’ stock investment intentions. The
research design is descriptive, and data were collected through
2.2.9. Perceived threat, attitude, and intention to use robo-
a cross-sectional survey using purposive sampling (Etikan,
advisers
2016; Tongco, 2007) in northern India. A total of 717 re-
Uncertainty about the outcome and fear of losing money or
sponses to the online survey were recorded, of which 42 were
private data are also threat concerns during the provision of
considered invalid due to missing data. Thus, 675 responses
services (Malaquias & Hwang, 2016). The greater the perceived
that were complete in all respects were used for further analysis
threat is, the less likely people are to use robo-advisers. Ac-
and interpretation. G*Power software was used to determine
cording to Kesharwani and Bisht (2012), perceived threats
the appropriate sample size, with a power of 0.80 (Faul et al.,
reduce people's willingness to use online and technology-based
2009). The minimum sample size requirement at a 5 percent
services. According to the cornerstone of modern portfolio the-
significance level was 160. Thus, our sample of 637 was found
ory, stock investors customize their risk-return preferences and
to be appropriate. In a pilot study, 30 investors validated the
diversify their investments accordingly (Chhabra, 2005). Indi-
reliability of the survey instrument used in the study.
vidual perceived threats negatively impact the intention to use
automated online financial advice (Duan et al., 2019).
3.2. Statistical methods
Furthermore, according to Pan et al. (2019), AI-based
innovative services negatively impact the individual attitudes
Nonparametric variance-based partial least squares struc-
and intention to use these services (Sharma et al., 2020). The
tural equation method (PLS-SEM) in SmartPLS 4.0 was
TTAT states that an individual's perceived threat from using
employed for statistical application and hypotheses testing
internet services is driven by perceived susceptibility and
(Ringle et al., 2015). PLS-SEM is considered a suitable
severity. The two perceived threat predictors are susceptibility
approach for prediction orientation in a theoretical framework
and severity (Liang & Xue, 2009). Thus, with respect to AI-
in the social and behavioral sciences (Hair et al., 2022). PLS-
based regulated algorithms of robo-advisers, we posit.
SEM is an appropriate multivariate data analysis method for
Hypothesis 9a. Perceived threat negatively influences atti- the present study as it involves a complex model (Hair et al.,
tudes toward using robo-advisers. 2019, 2022), with a prediction perspective of the intentions
of retail investors to adopt robo-advisers (Cheah et al., 2019;
Sarstedt et al., 2019) by. Further, PLS-SEM is applicable to the
Hypothesis 9b. Perceived threat negatively influences the
present study, which aims to establish theory, complex models
intention to use robo-advisers.
with many constructs and their associated indicators quickly
reach limits (Saari et al., 2021). Thus, our study is motivated by
2.2.10. Attitude and intention to use robo-advisers a pragmatic need to measure the phenomenon of interest by
The theory of planned behavior states that individual atti- employing PLS-SEM (Rigdon et al., 2017).
tudes positively affect behavioral intention. In the context of Our study aims to obtain prediction accuracy in order to
technology, TAM and UTAUT models also provide evidence offer recommendations for management practice (Hair et al.,
that attitudes predict the intention to use information technol- 2017), so CB-SEM is not the appropriate method. CB-SEM
ogy (Cao et al., 2021; Gursoy et al., 2019). The existing models the constructs as common factors based on covaria-
literature indicates that individual attitudes positively influence tion between their associated indicators. These common fac-
the intention to use technology and AI-based services (Duan tors are not known and do not have a finite range of values
et al., 2019; Dwivedi et al., 2021; Venkatesh et al., 2003). and thus are arbitrary quantities (Steiger, 1979). In contrast,
Research on service chatbots and robot interaction with peo- PLS-SEM covers determinate functions of composites as a
ple's adoption intention is abundant (Blanche et al., 2019; Fulk weighted sum of a specific subset of constructs. These com-
et al., 2018; Hohenburger et al., 2019; Huang & Rust, 2018; posites investigate a series of regressions to maximize the
Luo et al., 2019; Zheng et al., 2022). In line with these studies, explained variance of endogenous constructs (Shmueli et al.,
we propose the following hypothesis. 2016).

1462
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

Fig. 1. Proposed conceptual model with gender, age, trading experience, and risk aversion as control variables.

Further, the main objective of this study is to investigate the normality using Mardia's (1970) test (Zhang & Yuan, 2018). The
variance in investor attitudes and behavioral intentions. PLS- beta values of multivariate skewness and kurtosis values were
SEM with composites is more beneficial than CB-SEM for both determined to be statistically significant, demonstrating the
examining the data and evaluating different configurations, in nonnormality of the data. This multivariate nonnormality is
which the data need to conform to various measurement con- another reason for using PLS-SEM (Hair et al., 2022).
straints in a factor model (Jöreskog, 1969). Additionally, when
complex models are measured with less theoretical substantiation 4. Results and analysis
and lack comprehensive support of a measurement theory, PLS-
SEM is considered a more appropriate approach than CB-SEM 4.1. Descriptive statistics
(Rigdon et al., 2017, p. 13). Thus, PLS-SEM is considered a
more appropriate and suitable method for us to use than CB-SEM. The age of investors observed was from 23 to over 62; 55.6
percent of the participants in the survey were men and 44.4
3.3. Common method bias and multivariate normality percent were female. The majority of the respondents had family
checks income of as much as INR ten lakhs (USD 12055 Approxi-
mately) per annum. In the study, 57.9 percent of the investors are
The validity of the constructs is tested using Harman's single- risk takers, reflecting their tendency to accept advice on invest-
factor test and common method bias (CMB) (Podsakoff et al., ment decisions from robo-advisers. Most investors are experi-
2003). Harman's single-factor test reveals that the total varia- enced: 46.8 percent have 5–10 years of experience. Table 1
tion explained by a single factor is 28.623, well below the 50 indicated that the highest mean value of 4.14 was shown by
percent threshold (Podsakoff et al., 2003). As a result, CMB is prediction and accuracy, followed by data security with 3.91.
not a concern in this investigation. Furthermore, variance infla- The retail investors perceived the cost-effectiveness with a mean
tion factor (VIF) values for all latent variables in the investiga- value of 3.13. The variable under study represented the highest
tion are 1.116–2.520, indicating that no CMB concerns were association between data security and prediction, with .704
revealed by the complete collinearity test (Kock, 2015). Using a indicating retail investors’ interest in using robot services in
web-based calculator, the researchers evaluated multivariate stock investment decisions.
1463
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

Table 1
Descriptive statistics and correlations.
Mean SD Trust Awar. Beh_ Bias Cost Effect. Predict. Data Sec. Attit. Per_Suscep. Per_Sev. Per_ Theat IoU
Awareness 3.55 .872 .302** 1
Behavioral Bias 3.83 .779 .376** .393** 1
Cost Effectiveness 3.13 .844 .237** .372** .192** 1
Prediction 4.14 .764 .261** .187** .538** .041 1
Data Security 3.91 .690 .337** .291** .584** .103** .704** 1
Attitude 3.59 .648 .454** .348** .498** .197** .504** .582** 1
Perceived Susceptibility 3.84 .683 .372** .466** .640** .195** .541** .591** .526** 1
Perceived Severity 3.92 .750 .397** .302** .468** .347** .365** .422** .458** .441** 1
Perceived Threat 3.87 .638 .467** .271** .453** .227** .387** .459** .539** .444** .494** 1
Intentions to Use 3.69 .903 .346** .276** .370** .290** .374** .467** .452** .439** .463** .479** 1
Trust 3.55 .764
Note: Aware. = Awareness, Beh_Bias = Behavioral Bias, Cost Effect. = Cost Effectiveness, Predict. = Prediction, Data Sec. = Data Security, Attit. = Attitude,
Per_Sucep. = Perceived Susceptibility, Per_Sev. = Perceived Severity, Per_Threat = Perceived Threat, IoU = Intentions to Use.

Table 2
Reliability and validity of the constructs.
Construct Coding Factor Loadings Cronbach's Alpha RhoA Composite Reliability AVE
Trust TT1 0.774 0.766 0.776 0.865 0.681
TT2 0.87
TT3 0.829
Awareness Aware1 0.790 0.778 0.818 0.869 0.688
Aware2 0.889
Aware3 0.807
Behavioral Bias Bias1 0.871 0.805 0.809 0.885 0.720
Bias2 0.812
Bias3 0.861
Cost Effectiveness CostE1 0.757 0.606 0.672 0.778 0.545
CostE2 0.577
CostE3 0.853
Prediction and Judgment Pred1 0.836 0.854 0.856 0.902 0.697
Pred2 0.870
Pred3 0.872
Pred4 0.758
Data Security DSec1 0.808 0.651 0.651 0.812 0.592
DSec2 0.809
DSec3 0.685
Perceived Severity PSev1 0.373a 0.688 0.703 0.806 0.511
PSev2 0.450a
PSev3 0.633
PSev4 0.694
PSev5 0.681
PSev6 0.758
Perceived Susceptibility PSus1 0.790 0.769 0.770 0.867 0.685
PSus2 0.840
PSus3 0.852
Perceived Threat PerThreat1 0.558 0.651 0.714 0.793 0.501
PerThreat2 0.827
PerThreat3 0.842
PerThreat4 0.542
Attitude Att1 0.688 0.678 0.686 0.806 0.511
Att2 0.771
Att3 0.637
Att4 0.755
Intentions to Use Int1 0.889 0.881 0.883 0.926 0.807
Int2 0.892
Int3 0.914
a
Item deleted from analysis.

1464
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

4.2. First-order measurement model assessments Table 3

Discriminant validity.
Hair et al. (2022) provide guidelines for first-order evalua- HTMT correlations between constructs HTMT Confidence
tion in measurement model assessments in the present study. Intervals
Table 2 lists the reliability of indicators of all latent constructs, 2.50% 97.50%
followed by internal consistency (Henseler's RhoA and com- Awareness <−> Attitude 0.487 0.373 0.593
posite reliability) and convergent validity (average variance Behavioral Bias <−> Attitude 0.674 0.567 0.769
extracted, AVE). The indicator loadings of all constructs are Behavioral Bias <−> Awareness 0.504 0.411 0.589
Cost Effectiveness <−> Attitude 0.397 0.289 0.486
above the critical value of 0.70 (Sarstedt et al., 2017), except Cost Effectiveness <−> Awareness 0.540 0.429 0.644
for four items that were retained in the model as the threshold Cost Effectiveness <−> Behavioral Bias 0.281 0.178 0.369
values of AVE were above 0.50 (Hair et al., 2022). These Data Security <−> Attitude 0.877 0.805 0.945
statements pertain to the construct of cost effectiveness Data Security <−> Awareness 0.421 0.327 0.521
(CostE2), perceived threat (PerThreat1 and PerThreat4), and Data Security <−> Behavioral Bias 0.807 0.735 0.876
Data Security <−> Cost Effectiveness 0.292 0.193 0.384
attitude (Att3). The AVE for these three constructs achieved Intentions_to Use <−> Attitude 0.585 0.503 0.658
the threshold value of 0.50, and the reliability of constructs was Intentions_to Use <−> Awareness 0.339 0.242 0.430
also well above the value of 0.70. Thus we retain them to avoid Intentions_to Use <−> Behavioral Bias 0.441 0.347 0.531
the deletion of essential dimensions of these constructs under Intentions_to Use <−> Cost Effectiveness 0.401 0.294 0.501
consideration (Hair et al., 2022; Saari et al., 2021). The Intentions_to Use <−> Data Security 0.617 0.529 0.699
Perceived_Severity <−> Attitude 0.731 0.628 0.821
construct of perceived severity is measured without the two Perceived_Severity <−> Awareness 0.403 0.301 0.503
indicators of Psev1 and Psev2, and the factor loadings are Perceived_Severity <−> Behavioral Bias 0.733 0.656 0.804
below the acceptable value, 0.60 (Hair et al., 2019; Saari et al., Perceived_Severity <−> Cost Effectiveness 0.440 0.333 0.543
2021). The insignificant statements of this construct are that Perceived_Severity <−> Data Security 0.720 0.611 0.816
robo-advisers may amplify discrimination in available data Perceived_Severity <−> Intentions_to Use 0.650 0.576 0.721
Perceived_Susceptibility <−> Attitude 0.727 0.617 0.823
(Psev1) and robo-advisers may be prone to reproducing insti- Perceived_Susceptibility <−> Awareness 0.610 0.515 0.697
tutional biases in available data (Psev2). Perceived_Susceptibility <−> Behavioral Bias 0.812 0.743 0.869
The heterotrait-monotrait (HTMT) ratio established the Perceived_Susceptibility <−> Cost Effectiveness 0.307 0.210 0.399
discriminant validity of constructs, and the values are below the Perceived_Susceptibility <−> Data Security 0.836 0.747 0.911
threshold value of 0.85. The heterotrait-monotrait ratio of cor- Perceived_Susceptibility <−> Intentions_to Use 0.536 0.454 0.613
Perceived_Susceptibility <−> Perceived_Severity 0.695 0.610 0.774
relations was used to examine the discriminant validity of first- Perceived_Threat <−> Attitude 0.817 0.738 0.890
order components. HTMT ratios were less than 0.85 (Henseler Perceived_Threat <−> Awareness 0.384 0.280 0.486
et al., 2015), indicating sufficient discriminant validity of all Perceived_Threat <−> Behavioral Bias 0.626 0.520 0.723
the latent variables. However, for prediction and judgment with Perceived_Threat <−> Cost Effectiveness 0.360 0.248 0.461
data security, HTMT was above 0.90 but within the lower bound Perceived_Threat <−> Data Security 0.709 0.599 0.811
Perceived_Threat <−> Intentions_to Use 0.640 0.553 0.720
and upper bound of confidence intervals at 95 percent, checked Perceived_Threat <−> Perceived_Severity 0.806 0.714 0.889
at 10,000 bootstrapping subsamples (Hair et al., 2022). The Perceived_Threat <−> Perceived_Susceptibility 0.633 0.541 0.722
appropriate range of the HTMT ratio can be tolerated at 1 as per Prediction <−> Attitude 0.663 0.578 0.738
the HTMT inference method (Hair et al., 2022). In this regard, Prediction <−> Awareness 0.241 0.167 0.323
the confidence interval approach is considered an appropriate Prediction <−> Behavioral Bias 0.651 0.573 0.721
Prediction <−> Cost Effectiveness 0.219 0.148 0.297
technique with 10,000 bootstraps. The lower bound of confi- Prediction <−> Data Security 0.944 0.879 1.005
dence intervals is 0.879, and the upper bound, 0.998. An HTMT Prediction <−> Intentions_to Use 0.433 0.354 0.510
ratio of 0.944 is observed between these two limits to justify the Prediction <−> Perceived_Severity 0.581 0.502 0.654
discriminant validity of the decision security and prediction and Prediction <−> Perceived_Susceptibility 0.669 0.591 0.737
judgment constructs. Table 3 lists the discriminant validity of the Prediction <−> Perceived_Threat 0.527 0.432 0.615
Trust <−> Attitude 0.631 0.532 0.718
constructs in the proposed model. Trust <−> Awareness 0.396 0.293 0.488
Trust <−> Behavioral Bias 0.480 0.383 0.572
4.3. Structural model assessment Trust <−> Cost Effectiveness 0.351 0.237 0.460
Trust <−> Data Security 0.478 0.374 0.576
We used the recommendations by Hair et al. (2019, 2022) to Trust <−> Intentions_to Use 0.423 0.328 0.514
Trust <−> Perceived_Severity 0.521 0.426 0.610
examine the structural model results for hypothesis testing and Trust <−> Perceived_Susceptibility 0.483 0.382 0.580
the model's explanatory and predictive capacity. The collin- Trust <−> Perceived_Threat 0.660 0.572 0.744
earity issues are diagnosed by VIF (Variance Inflation Factor), Trust <−> Prediction 0.323 0.229 0.416
which are less than the critical value of 3.33 (Hair et al., 2019), Note: ATT = Attitude; AWA = Awareness; BB = Behavioral Bias; CE = Cost
to demonstrate the absence of multicollinearity issues. The size Effectiveness; DS = Data Security; INT = Intention to Use; PS = Perceived
and significance of path coefficients in the structural model are Severity; PSUS = Perceived Susceptibility; PT = Perceived Threat; PRED =
tabulated in line with Saari et al. (2021) (Table 4). Prediction & Judgment; TT = Trust.

1465
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

Table 4
Hypotheses testing.
Hypotheses Вeta t-Statistics p-values CI [2.50:97.5] Significance? VIF f2
H1a TT - > ATT 0.159 4.220*** 0.000 [0.083:0.232] Yes 1.393 0.036
H1b TT - > INT 0.050 1.098 0.272 [-0.040:0.136] No 1.500 0.003
H2a AWA - > ATT 0.105 2.924** 0.003 [0.034:0.175] Yes 1.387 0.016
H2b AWA - > INT 0.031 0.786 0.432 [-0.046:0.110] No 1.427 0.001
H3a BB - > ATT 0.060 1.283 0.200 [-0.030:0.151] No 1.862 0.004
H3b BB - > INT −0.009 0.196 0.845 [-0.105:0.086] No 1.884 0.000
H4a CE - > ATT 0.035 1.098 0.272 [-0.031:0.097] No 1.217 0.002
H4b CE - > INT 0.165 4.209*** 0.000 [0.086:0.240] Yes 1.223 0.035
H5a PRED - > ATT 0.133 2.767** 0.006 [0.038:0.228] Yes 2.119 0.017
H5b PRED - > INT 0.044 0.965 0.335 [-0.047:0.130] No 2.164 0.001
H6a DS - > ATT 0.245 4.154*** 0.000 [0.132:0.360] Yes 2.369 0.050
H6b DS - > INT 0.219 4.315*** 0.000 [0.118:0.317] Yes 2.520 0.030
H7 PS - > PT 0.445 12.016*** 0.000 [0.366:0.511] Yes 1.322 0.235
H8 PSUS - > PT 0.240 6.260*** 0.000 [0.165:0.314] Yes 1.322 0.068
H9a PT - > ATT −0.248 6.940*** 0.000 [0.319: 0.178] Yes 1.544 0.079
H9b PT - > INT −0.262 5.874*** 0.000 [-0.348: 0.174] Yes 1.696 0.065
H10 ATT - > INT 0.092 2.014** 0.044 [0.003:0.183] Yes 2.001 0.006
CV Age - > INT −0.041 1.066 0.286 [-0.115:0.036] No 1.636
CV Gender - > INT −0.028 0.854 0.393 [-0.095:0.036] No 1.116
CV Risk Aversion - > INT 0.007 0.223 0.823 [-0.057:0.073] No 1.248
CV Trading Experience - > INT 0.021 0.616 0.538 [-0.044:0.088] No 1.281
Note: ATT = Attitude; AWA = Awareness; BB = Behavioral Bias; CE = Cost Effectiveness; DS = Data Security; INT = Intentions to Use; PS = Perceived
Severity; PSUS = Perceived Susceptibility; PT = Perceived Threat; PRED = Prediction & Judgment; TT = Trust. Hypo = Hypotheses; CI = Confidence Intervals at
95%; CV = Control Variables; β = Standardized Beta.

Table 4 show the final relationship between the constructs (1), age (0 for up to 30 years and 1 for above 30 years), risk
studied. The coefficient of determination R2 is 49.6 percent for aversion (0 for risk averse and 1 for risk taking), and trading
attitude, whereas R2 is 36.6 percent for behavioral intentions. experience (up to 5 years = 0; above 5 years = 1). The results
Standardized root mean square residuals (SRMR) are used to reveal that young male investors are more prone to use robo-
evaluate the goodness of fit index. The value of SRMR for the advisers. In addition, investors who follow investment advice
estimated model is 0.068, less than the critical value of 0.08 from robo-advisers were risktakers and had longer trading
(Hair et al., 2022). experience. However, none of the control variables are sig-
The significant predictors of the intention to use robo-advisers nificant, but they indicate the interests of stock investors.
are perceived threat (β = −0.262, p < 0.000, supports H9b)
followed by data security (β = 0.219, p < 0.000, supports H6b). 4.4. Out-of-sample predictive relevance
In addition, cost effectiveness is another significant predictor of
the intention to use robo-advisers (β = 0.165, p < 0.000, supports Using the PLSpredict procedure, the study assess out-of-
H4b). In terms of the developing attitude of stock investors, the sample predictive relevance (Danks & Ray, 2018; Shmueli
key predictors are observed to be data security (β = 0.245, et al., 2019) of critical dependent variables such as sustain-
p < 0.000, supports H6a) followed by trust in robo-advisers able institutions, admissions, and the brand value of higher
(β = 0.159, p < 0.000, supports H1a) and the level of predic- education institutes in India. The root mean squared errors
tion and judgments fed into the algorithm (β = 0.133, p < 0.05, (RMSE) are calculated for PLS-SEM (theoretical model) and
supports H5a). Additionally, the awareness level of stock in- analyzed against the linear model (LM) benchmarks. The
vestors also drives the attitude of stock investors (β = 0.105, prediction errors are symmetrically distributed. Table 5 shows
p < 0.05, supports H2a). The adverse effects of perceived threat that all Q2 predict values of more than zero. Most RMSEPLS
are also a significant predictor of the negative attitude of the stock values are lower than the RMSELM benchmarks, with high
investors (β = −0.248, p < 0.000, supports H9a). The critical predictive power over stock investors' intention to use robo-
driver of the intention to use robo-advisers is attitude, which is advisers in Indian financial markets.
significant only at 5 percent (β = 0.092, p < 0.05, supporting
H10) (Refer to Fig. 2). 5. Discussion and implications
Moderate f-square values were observed for perceived
severity on perceived threat and indicated the importance of In an attempt to investigate the intention of stock investors
determining the perceived threat (R2 = 36.1 percent). The f- to use AI-enabled financial advice, we find forces that create
square has a small to moderate effect for most predictors of perceived threats, shape attitudes, and form an intention to use
attitudes and intention. The PLS-SEM results are controlled by robo-advisers. The results indicate that perceived severity and
age, gender, risk aversion, and trading experience. Gender is susceptibility are sufficient for forming a perceived threat
used as a dummy variable and coded as male (0) and female (Liang & Xue, 2009) about robo-advisers among stock

1466
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

Fig. 2. Structural model assessments.

success of robo-advisers (Almadi et al., 2014; Horn et al.,

Table 5
2020, pp. 309–327). The study confirms the findings on the
PLS prediction results.
Construct PLS-SEM Results LM Results PLS - LM
digitization and automation of financial advice (Jung et al.,
Intention to Use RMSE MAE Q2_predict RMSE MAE 2018) and the trust-influencing mechanism of robo-advisers
−0.007
Int1
Int2
0.884
0.801
0.689
0.606
0.231
0.298
0.891
0.805
0.704
0.629 −0.004 (Cheng et al., 2019). It also supports the results on the
Int3 0.919 0.717 0.245 0.904 0.712
Note: LM = linear model benchmarks; PLS = partial least squares;
0.015
knowledge and awareness of automated financial robo-advisers
RMSE = root mean squared error.
(Bhatia et al., 2020), data security and familiarity with robo-
advisers (Kumar & Gupta, 2020), and perceived threats and
tech-based services (Kesharwani & Bisht, 2012; Pan et al.,
2019; Sharma et al., 2020).
investors. It negatively affects attitude formation and intention Additionally, factors such as attitude, data security, and cost
to use robo-advisers. The findings are in line with those on the effectiveness increase the intention to use robo-advisers, and
relation between the fear of loss and online service process perceived threat and behavioral bias reduce it. However, direct
(Malaquias & Hwang, 2016), willingness to use online and forces such as trust, awareness, prediction and judgment, and
technology-based services (Kesharwani & Bisht, 2012), and control variables such as age, gender, risk aversion, and trading
automated online financial advice (Duan et al., 2019; Sharma experience do not have an important role in building stock in-
et al., 2020). vestors' intention to use robo-advisers. These outcomes are
Further, we find that trust, awareness, prediction and judg- similar to those on whether attitude predicts the intention to use
ment, and data security are positive forces, and perceived threat information technology (Cao et al., 2021; Gursoy et al., 2019;
is an opposing force in shaping the attitude of stock investors to Venkatesh et al., 2013; Zheng et al., 2022), data security, atti-
use robo-advisers, in which behavioral bias and cost effec- tude, and the intention to use robo-advisers (Bhatia et al., 2020;
tiveness are less critical. Cost effectiveness is significant for the Sander, 2021), and behavioral bias and the intention to use

1467
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

Fig. 3. Importance–performance map.

robots (Bhatia et al., 2020; Uhl & Rohner, 2018). Further, they 5.2. Managerial implications
are reinforced by findings on financial security, financial effi-
ciency, and cost effectiveness (Horn et al., 2020, pp. 309–327; The paper has practical implications for financial advisers,
Liu, 2018; Singh & Kaur, 2017; Tao et al., 2021). However, companies, technology firms, and policymakers at regulatory
they differs from findings on trust and usage behavior agencies. In this regard, using a priority map analysis, we
(Dimitriadis & Kyrezis, 2010; Lee et al., 2018) and prediction create a matrix (Ringle & Sarstedt, 2016) at the indicator level
and robots’ advice (Jung et al., 2018; Tsai & Chen, 2022). to achieve deep insights into the determinants of the intention
to use robo-advisers. The critical predictor of attitudes and
5.1. Theoretical implications perceived threat are essential for shaping the target construct of
the intention to use robo-advisers (Ringle & Sarstedt, 2016).
This study makes several contributions to the theoretical The IPMA (Importance Performance Map Analysis) findings
understanding of investors' attitudes, risky behaviors, and the show that the critical determinant of the intention to use robo-
intention to use robo-advisers for financial asset management. advisers is data security features, followed by cost-
The extensive research on technology adoption models primarily effectiveness and the attitude of stock investors. Based on the
focuses on technology's functional aspects. However, this study findings of IPMA, the critical determinant of behavioral in-
addresses the complex aspects of the decision-making process tentions were data security features, followed by cost-
by investors in adopting AI-based robo-advisers in investment effectiveness and the attitude of stock investors. A one-
decisions. The symbiotic relationship between people and AI in percentage-point increase in data security, from 72.716 to
financial decision-making is a new dimension in IS research and 73.716, raises the intention to use robo-advisers from 67.349 to
behavioral finance. The study also makes a contribution to 64.663, which is higher than that for cost-effectiveness
positive and negative loops in cybernetic theory by using robot (64.526) and attitude (64.474). The perceived threat also
advisers (Carver, 2006; Carver & Scheier, 1982). A prior study plays an essential role in determining the attitudes and intention
investigates AI adoption identifying the negative consequences of stock investors to use AI in financial investment decisions
of AI when stock investors cater to a perceived threat raised by (see Fig. 3).
TTAT (Liang & Xue, 2009). Our study presents a novel Our results indicate that stock investors have a positive
framework for integrating cybernetic theory with TTAT to un- attitude toward using robo-advisers. However, technology
derstand investors' behavioral intention to use robo-advisers. firms and financial advisers must work on data security and
Integrated theories provide new dimensions in technology perceived threat carefully. Similarly, prediction and judgment
adoption, among which threat aspects have wide theoretical and behavioral biases are the performance constructs that are
implications for further research (Cao et al., 2021; Carver & not considered necessary by stock investors in emerging
Scheier, 1982; Liang & Xue, 2009). Future research on AI financial markets such as India. These practical implications
adoption by stock investors can introduce new dimensions for are essential for policymakers and regulatory bodies of
understanding behavioral bias in investment decision-making in financial markets to consider for AI to play a role in an
emerging financial markets such as India. economy.

1468
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

6. Conclusion Appendix Table A1 (continued )

Constructs Statements Adapted from
The goal of the study is to understand investor attitudes and Behavioral Bias 1. Incorrect algorithms and Gordon (2011)
threatening behavior toward robo-advisers in making investment inappropriate coding by Brenner & Meyll
decisions. The key antecedents of attitude are data security and the developers create (2019)
perceived threat involved in stock investors’ use of AI-based fear of investment Muralidhar & Berlik
decisions. (2017)
robo-advisers. The key determinants that drive the intention to 2. Investors are skeptical of Edwards (2018)
use robot advice are cost effectiveness, data security, and the developers of robo-
perceived threat from using a robot for investment decisions. advisers for providing
The results suggest that investors have developed a positive correct investment
attitude toward using robo-advisers. However, the perceived information.
3. Programmer or robot
threat is high compared to the positive aspects of robo-advisers. developers' personal bias
This study is not free of limitations. First, the study focuses on affects my investment
investors with the knowledge and expertise to use AI in the form decisions in the financial
of robo-advisers when making investment decisions. Future market.
research could be conducted on potential investors who are Cost Effective 1. I know the cost-effective Uhl and Rohner
and viable source of (2018)
novices in implementing AI in investment decisions. Second, for robo-advisers in the Woodyard (2018)
the sake of the convenience, we conducted the research with market. Bhatia et al. (2021)
stock investors in north India. A similar study could also be 2. Robo-advisers serve Singh and Kaur
conducted in Mumbai, India's financial hub. Future research large clients at a lower (2017)
should try to understand the new dimensions of AI and cost in achieving in-
vestors' financial goals.
maximum likelihood in handling big data and perform a senti- 3. Online automated robot
ment analysis of investors to take the pulse of financial markets. advice serves existing
and prospective in-
Declaration of competing interest vestors as a cost-
effective option.
Prediction 1. Robo-adviser recom- Creswell and Poth
We hereby declare that to the best of our knowledge, we do and Judgment mendations are more (2017) Sander (2021)
not have any conflict of interest regarding this manuscript. objective and structured. D'Hondt et al. (2019)
2. Financial modeling and Uhl and Rohner
advice by robo-advisers (2018)
Appendix. meet clients' Tsai and Chen (2022)
requirements for
investment decisions.
3. Robo-advisers can fore-
Appendix Table A1 cast stock market vola-
Constructs Statements Adapted from tility in financial
markets.
Trust 1. I trust financial in- Silva (2019)
4. Robo-advisers can pre-
stitutions that offer robo- Lam (2016)
dict uncertainty and any
advisors for stock Nikiforova (2017)
upcoming financial
investors. Lynn et al. (2016)
shocks in financial
2. I trust the structural Pearson & Yee (2013)
markets.
assurance of robo-ad-
Data Security 1. Robo-advisers offer Beck (2021)
visers for financial asset
complete protection of Kumar and Gupta
management.
financial information for (2020) Sander (2021)
3. I trust the consistency of
accurate advice. Bhatia et al. (2020)
information provided by
2. Investors' data security Guo (2020)
robo-advisers for better
in robot advice offers
advice and returns.
high-security standards
Awareness 1. My financial acumen is Cantrell & Issa (2018)
and protective features.
suitable for robo- Edwards (2018)
3. Robo-advisers’ technol-
advisers. Bhatia et al. (2020)
ogy standards are very
2. Robo-advisers enhance
high to protect investor's
investors' knowledge and
data.
awareness of investment
decisions in financial (continued on next page)
markets.
3. I have less familiarity
and awareness about
robo-advisers for wealth
management (R)
(continued on next page)

1469
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

Appendix Table A1 (continued )

References
Constructs Statements Adapted from
Perceived 1. Robo-advisers are likely Atwal & Bryson Abraham, F., Schmukler, S. L., & Tessada, J. (2019). Robo-advisors: Investing
Susceptibility to make bad investment (2021) through machines. World Bank Policy Research Working Paper, Article
decisions in the future. Cao et al. (2021) 134881.
2. The chances of robo- Ahmad, M. (2021). The role of recognition-based heuristics in investment
advisers making bad management activities: Are expert investors immune? – a systematic
investment decisions are literature review. Qualitative Research in Financial Markets. https://
high. doi.org/10.1108/QRFM-07-2021-0109. ahead-of-print No. ahead-of-print.
3. Robo-advisers may Ali, Q., Parveen, S., Yaacob, H., Rani, A. N., & Zaini, Z. (2022). Environ-
make bad investment mental beliefs and the adoption of circular economy among bank managers:
decisions at some point. Do gender, age, and knowledge act as the moderators? Journal of Cleaner
Perceived Severity 1. Robo-advisers may Shrestha et al. (2019) Production, 361. https://doi.org/10.1016/j.jclepro.2022.132276
amplify discrimination in Weld and Bansal Arikan, E., Altinigne, N., Kuzgun, E., & Okan, M. (2023). Robots be held
available data. (2019) responsible for service failure and recovery? The role of robot service
2. Robo-advisers may be Cao et al. (2021) provider agents' human-likeness. Journal of Retailing and Consumer Ser-
prone to reproducing Atwal & Bryson vices, 70, Article 103175.
institutional biases in (2021) Baker, T., & Dellaert, B. (2017). Regulating Robo advice across the financial
available data. Atwal & Bryson services industry. Iowa Law Review, 103, 713.
3. Robo-advisers may have (2021); Cao et al. Balakrishnan, J., & Dwivedi, Y. K. (2021). Role of cognitive absorption in
a propensity for intensi- (2021) building user trust and experience. Psychology and Marketing, 38(4),
fying systemic bias in Shrestha et al. (2019) 643–668.
available data. Weld and Bansal Beck, A. D. (2021). The role of artificial intelligence in robo-advisory. In
4. Robo-advisers may have (2019) P. Scholz (Ed.), Robo-advisory. Palgrave studies in financial services
a different objective due technology. Cham: Palgrave Macmillan. https://doi.org/10.1007/978-3-030-
to the difficulty of spec- 40818-3_11.
ifying the objective Belanche, D., Casaló, L., & Flavián, C. (2019). Artificial intelligence in Fin-
explicitly. Tech: Understanding robo-advisors adoption among customers. Industrial
5. Robo-advisers may need Management & Data Systems, 119(7), 1411–1430.
more adequate struc- Bhatia, A., Chandani, A., Atiq, R., Mehta, M., & Divekar, R. (2021). Arti-
tures, such as problem- ficial intelligence in financial services: Qualitative research to discover
atic models. robo-advisory services. Qualitative Research in Financial Markets.
6. Robo-advisers may need https://doi.org/10.1108/QRFM-10-2020-0199. ahead-of-print No. ahead-
more training for in- of-print.
vestors to perform well. Bhatia, A., Chandani, A., & Chhateja, J. (2020). Robo advisory and its po-
Perceived Threat 1. My fear of exposure to Daun et al. (2019) tential in addressing the behavioural biases of investors – a qualitative study
robo-advisers is high. Sharma et al. (2020) in the Indian context. Journal of Behavioral and Experimental Finance, 25,
2. The extent of my worry Malaquias and Hwang Article 100281.
about robo-advisers risk (2016) Breward, M., Hassanein, K., & Head, M. (2017). Understanding consumers'
is low (R). Pan et al. (2019) attitudes toward controversial information technologies: A contextualiza-
3. The extent of my anxiety tion approach. Information Systems Research, 28(4), 760–774.
about potential loss due Burton, B. (2014). Launch of the centre for qualitative research in finance.
to robo-advisers risk is Qualitative Research in Financial Markets, 6(2). https://doi.org/10.1108/
high. QRFM-08-2014-003
4. The extent of my worry Cao, G., Duan, Y., Edwards, J. S., & Dwivedi, Y. K. (2021). Understanding
about robo-advisers risks managers' attitudes and behavioural intentions towards using artificial in-
due to misuse is high. telligence for organizational decision-making. Technovation, 106, Article
Attitude 1. Using robo-advisers is a Malaquias and Hwang 102312.
good idea. (2016) Cao, Y., Long, M., Wang, J., & Yu, P. S. (2016). Correlation hashing network
2. Using robo-advisers is a Daun et al. (2019) for efficient cross-modal retrieval. arXiv preprint arXiv:1602.06697.
foolish idea (R). Dwivedi et al. (2021) Capponi, A., Olafsson, S., & Zariphopoulou, T. (2022). Personalized Robo-
3. I like the idea of using advising: Enhancing investment through client interaction. Management
robo-advisers. Science, 68(4), 2485–2512.
4. Using robo-advisers Carver, C. S. (2006). Approach, avoidance, and the self-regulation of affect and
would be pleasant. action. Motivation and Emotion, 30, 105–110.
Intention to Use 1. I intend to use robo- Gursoy et al. (2019) Carver, C. S., & Scheier, M. F. (1982). Control theory: A useful conceptual
advisers for making Cao et al. (2021) framework for personality-social, clinical, and health psychology. Psy-
investment decisions in Daun et al. (2019) chological Bulletin, 92(1), 111–135.
the future. Dwivedi et al. (2021) Chandani, A., Sriharshitha, S., Bhatia, A., Atiq, R., & Mehta, M. (2021). Robo-
2. I will always use robo- advisory services in India: A study to analyse awareness and perception of
advisers in my millennials. International Journal of Cloud Applications and Computing,
investment decisions. 11(4), 152–173.
3. I plan to use robo- Cheah, J. H., Tang, H., Ramayah, T., Memon, M. A., Cham, T. H., &
advisers frequently in Ciavolino, E. (2019). A comparison of five reflective–formative estimation
making investment approaches: Reconsideration and recommendations for tourism research.
decisions. Quality and Quantity, 53(3), 1421–1458.

1470
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

Cheng, X., Guo, F., Chen, J., Li, K., Zhang, Y., & Gao, P. (2019). Exploring Guo, L. (2020). Regulating investment robo-advisors in China: Problems and
the trust influencing mechanism of robo-advisor service: A mixed method prospects. European Business Organization Law Review, 21(1), 69–99.
approach. Sustainability, 11(18), 4917. Gursoy, D., Chi, O. H., Lu, L., & Nunkoo, R. (2019). Consumer's acceptance
Chen, X., Ye, S., & Huang, C. (2021). Cluster-based mutual fund classification of artificially intelligent (AI) device use in service delivery. International
and price prediction using machine learning for robo-advisors. Computa- Journal of Information Management, 49, 157–169.
tional Intelligence and Neuroscience, 1–14. https://doi.org/10.1155/2021/ Hair, J. F., Hult, G. T. M., Ringle, C. M., & Sarstedt, M. (2022). A primer on
4984265 partial least squares structural equation modeling (PLS-SEM) (3 ed.).
Chhabra, A. B. (2005). Beyond markowitz: A comprehensive wealth allocation Thousand Oaks, CA: Sage. Forthcoming.
framework for individual investors. The Journal of Wealth Management, Hair, J. F., Jr., Matthews, L. M., Matthews, R. L., & Sarstedt, M. (2017). PLS-
7(4), 8–34. SEM or CB-SEM: Updated guidelines on which method to use. Interna-
Climescu, A. M. (2021). What role do social media play for robo-advisors? In tional Journal of Multivariate Data Analysis, 1(2), 107–123.
P. Scholz (Ed.), Robo-advisory. Palgrave studies in financial services Hair, J. F., Risher, J. J., Sarstedt, M., & Ringle, C. M. (2019). When to use and
technology. Cham: Palgrave Macmillan. https://doi.org/10.1007/978-3-030- how to report the results of PLS-SEM. European Business Review, 31(1),
40818-3_12. 2–24.
Creswell, J. W., & Poth, C. N. (2017). Qualitative inquiry and research design: Hashimy, L., Jain, G., & Grifell-Tatjé, E. (2022). Determinants of blockchain
Choosing among five approaches. Sage publications. adoption as decentralized business model by Spanish firms – an innovation
D'Acunto, F., Rossi, A. G., & Robo-Advising. (2020). CESifo working paper theory perspective. Industrial Management & Data Systems. https://doi.org/
No. 8225. https://doi.org/10.2139/ssrn.3578259. Available at: SSRN: 10.1108/IMDS-01-2022-0030. ahead-of-print No. ahead-of-print.
https://ssrn.com/abstract=3578259. Henseler, J., Ringle, C. M., & Sarstedt, M. (2015). A new criterion for
D'Acunto, F., Prabhala, N., & Rossi, A. G. (2019). The promises and pitfalls of assessing discriminant validity in variance-based structural equation
robo-advising. Review of Financial Studies, 32(5), 1983–2020. modeling. Journal of the Academy of Marketing Science, 43(1), 115–135.
Danks, N. P., & Ray, S. (2018). Predictions from partial least squares models. Hohenberger, C., Lee, C., & Coughlin, J. (2019). Acceptance of Robo-advisors:
Applying Partial Least Squares in Tourism and Hospitality Research, Effects of financial experience, affective reactions, and self-enhancement
35–52. https://doi.org/10.1108/978-1-78756-699-620181003 motives. Financial Planning Review, 2(2), e1047.
Day, M., Cheng, T., & Li, J. (2018). AI Robo-advisor with big data analytics Horn, M., Oehler, A., & Wendt, S. (2020). FinTech for consumers and retail
for financial services. In IEEE/ACM international conference on advances investors: Opportunities and risks of digital payment and investment ser-
in social networks analysis and mining (ASONAM), 1027–1031. Barcelona: vices. Ecological, societal, and technological risks and the financial sector.
Institute of Electrical and Electronics Engineers (IEEE). Hoyer, W. D., Kroschke, M., Schmitt, B., Kraume, K., & Shankar, V. (2020).
De Gauquier, L., Willems, K., Cao, H. L., Vanderborght, B., & Brengman, M. Transforming the customer experience through new technologies. Journal
(2023). Together or alone: Should service robots and frontline employees of Interactive Marketing, 51(1), 57–71.
collaborate in retail-customer interactions at the POS? Journal of Retailing Huang, W. C., Chen, C. T., Lee, C., Kuo, F. H., & Huang, S. H. (2023).
and Consumer Services, 70, Article 103176. Attentive gated graph sequence neural network-based time-series infor-
D'Hondt, C., De Winne, R., Ghysels, E., & Raymond, S. (2019). Artificial mation fusion for financial trading. Information Fusion, 91, 261–276.
intelligence alter egos: Who benefits from Robo-investing?. arXiv preprint Huang, M. H., & Rust, R. T. (2018). Artificial intelligence in service. Journal
arXiv:1907.03370. of Service Research, 21(2), 155–172.
Dimitriadis, S., & Kyrezis, N. (2010). Linking trust to use intention for tech- Jöreskog, K. G. (1969). A general approach to confirmatory maximum likeli-
nology-enabled bank channels: The role of trusting intentions. Psychology hood factor analysis. Psychometrika, 34(2), 183–202.
and Marketing, 27(8), 799–820. Jung, D., Erdfelder, E., & Glaser, F. (2018). Nudged to win: Designing Robo-
Duan, Y., Edwards, J. S., & Dwivedi, Y. (2019). Artificial intelligence for advisory to overcome decision inertia. In ECIS (p. 19).
decision making in the era of Big Data – evolution, challenges, and research Kesharwani, A., & Bisht, S. S. (2012). The impact of trust and perceived risk
agenda. International Journal of Information Management, 48, 63–71. on Internet banking adoption in India: An extension of technology accep-
Dwivedi, Y. K., Hughes, L., Ismagilova, E., Aarts, G., Coombs, C., Crick, T., tance model. International Journal of Bank Marketing, 30(4), 303–322.
& Williams, M. D. (2021). Artificial intelligence (AI): Multidisciplinary Kock, N. (2015). Common method bias in PLS-SEM: A full collinearity
perspectives on emerging challenges, opportunities, and agenda for assessment approach. International Journal of E-Collaboration, 11(4),
research, practice, and policy. International Journal of Information Man- 1–10.
agement, 57, Article 101994. Kumar, M., & Gupta, S. (2020). Security perception of e-banking users in India:
Dzyabura, D., & Hauser, J. R. (2011). Active machine learning for consider- An analytical hierarchy process. Banks and Bank Systems, 15(1), 11–20.
ation heuristics. Marketing Science, 30(5), 801–819. Kundu, S., & Datta, S. K. (2015). Impact of trust on the relationship of e-
Eren, B. A. (2023). Antecedents of robo-advisor use intention in private service quality and customer satisfaction. EuroMed Journal of Business,
pension investments: An emerging market country example. Journal of 10(1), 21–46.
Financial Services Marketing, 1–16. Lee, S., Choi, J., Ngo-Ye, T. L., & Cummings, M. (2018). Modeling trust in the
Etikan, I. (2016). Comparison of convenience sampling and purposive sam- adoption decision process of robo-advisors: An agent-based simulation
pling. American Journal of Theoretical and Applied Statistics, 5(1), 1–4. approach. In Proceedings of the tenth midwest association for information
Fan, L., & Chatterjee, S. (2020). The utilization of robo-advisors by individual systems conference (pp. 17–18). Saint Louis, MO, USA.
investors: An analysis using diffusion of innovation and information. Lemon, K. N., & Verhoef, P. C. (2016). Understanding customer experience
Journal of Financial Counseling and Planning, 130–145. throughout the customer journey. Journal of Marketing, 80(6), 69–96.
Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power Liang, H., & Xue, Y. (2009). Avoidance of information technology threats: A
analyses using G * Power 3.1: Tests for correlation and regression analyses. theoretical perspective. MIS Quarterly, 71–90.
Behavior Research Methods, 41, 1149–1160. Lisauskiene, N., & Darskuviene, V. (2021). Linking the robo-advisors phe-
Fulk, M., Grable, J. E., Watkins, K., & Kruger, M. (2018). Who uses Robo- nomenon and behavioural biases in investment management: An interdis-
advisory services, and who does not? Financial Services Review, 173–188. ciplinary literature review and research agenda. Organizations and Markets
Gan, L. Y., Khan, M. T. I., & Liew, T. W. (2021). Understanding consumer's in Emerging Economies, 12(2), 459–477.
adoption of financial Robo-advisors at the outbreak of the COVID-19 crisis Liu, C. Y. (2018). Legal risks and the countermeasures of developing intelli-
in Malaysia. Financial Planning Review, e1127. gent investment advisors in China. In International conference on intelli-
Grewal, D., Hulland, J., Kopalle, P. K., & Karahanna, E. (2020). The future of gent human systems integration (pp. 76–82). Cham: Springer.
technology and marketing: A multidisciplinary perspective. Journal of the Liu-Thompkins, Y., Okazaki, S., & Li, H. (2022). Artificial empathy in mar-
Academy of Marketing Science, 48(1), 1–8. keting interactions: Bridging the human-AI gap in affective and social

1471
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

customer experience. Journal of the Academy of Marketing Science, 1–21. Sander, M. (2021). Success factors for robo-advisory: Now and then. In
https://doi.org/10.1007/s11747-022-00892-5 P. Scholz (Ed.), Robo-advisory. Palgrave studies in financial services
Lopez, C. F., Jeevarajan, J. A., & Mukherjee, P. P. (2015). Characterization of technology. Cham: Palgrave Macmillan. https://doi.org/10.1007/978-3-030-
lithium-ion battery thermal abuse behaviour using experimental and 40818-3_13.
computational analysis. Journal of the Electrochemical Society, 162(10), Sarstedt, M., Hair, J. F., Cheah, J.-H., Becker, J.-M., & Ringle, C. M. (2019).
A2163. How to specify, estimate, and validate higher-order constructs in PLS-SEM.
Luo, X., Tong, S., Fang, Z., & Qu, Z. (2019). Frontiers: Machine vs. humans: Australasian Marketing Journal, 27(3), 197–211.
The impact of artificial intelligence chatbot disclosure on customer pur- Sarstedt, M., Ringle, C. M., & Hair, J. F. (2017). Partial least squares structural
chases. Marketing Science, 38(6), 937–947. equation modelling. Handbook of Market Research, 26(1), 1–40.
Ma, W., & Downs, P. (2020). The hunt for unicorns: How sovereign funds are Sharma, R., Singh, G., & Sharma, S. (2020). Modelling internet banking
reshaping investment in the digital economy. John Wiley & Sons. adoption in Fiji: A developing country perspective. International Journal of
Malaquias, R. F., & Hwang, Y. (2016). An empirical study on trust in mobile Information Management, 53.
banking: A developing country perspective. Computers in Human Shmueli, G., Ray, S., Velasquez Estrada, J. M., & Chatla, S. B. (2016). The
Behavior, 54, 453–461. elephant in the room: Evaluating the predictive performance of PLS
Marchand, A., & Marx, P. (2021). Automated product recommendations with models. Journal of Business Research, 69(10), 4552–4564.
preference-based explanations. Journal of Stocking, 96(3), 328–343. Shmueli, G., Sarstedt, M., Hair, J. F., Cheah, J. H., Ting, H., Vaithilingam, S.,
Mardia, K. V. (1970). Measures of multivariate skewness and kurtosis with & Ringle, C. M. (2019). Predictive model assessment in PLS-SEM:
applications. Biometrika, 57(3), 519–530. Guidelines for using PLSpredict. European Journal of Marketing, 53(11),
Mcknight, D. H., Carter, M., Thatcher, J. B., & Clay, P. F. (2011). Trust in a 2322–2347.
specific technology: An investigation of its components and measures. ACM Shrestha, Y. R., Ben-Menahem, S. M., & Von Krogh, G. (2019). Organiza-
Transactions on Management Information Systems (TMIS), 2(2), 1–25. tional decision-making structures in the age of artificial intelligence. Cali-
Mende, M., Scott, M. L., van Doorn, J., Grewal, D., & Shanks, I. (2019). fornia Management Review, 61(4), 66–83.
Service robots rising: How humanoid robots influence service experiences Singh, I., & Kaur, N. (2017). Wealth management through Robo advisory.
and elicit compensatory consumer responses. Journal of Marketing International Journal of Research –Granthaalayah, 5(6), 33–43.
Research, 56(4), 535–556. Song, C. S., & Kim, Y. K. (2022). The role of the human-robot interaction in
Mezzanotte, F. E. (2020). An examination into the investor protection prop- consumers' acceptance of humanoid stock service robots. Journal of
erties of Robo-advisory services in Switzerland. Capital Markets Law Business Research, 146, 489–503.
Journal, 15(4), 489–508. Song, C. S., Kim, Y. K., Jo, B. W., & Park, S. H. (2022). Trust in humanoid
Oehler, A., Horn, M., & Wendt, S. (2022). Investor characteristics and their robots in footwear stores: A large-N crisp-set qualitative comparative
impact on the decision to use a robo-advisor. Journal of Financial Services analysis (csQCA) model. Journal of Business Research, 152, 251–264.
Research, 62(1–2), 91–125. Statista. (2019). Market report. available at: www.statista.com/outlook/337/
Pal, A., Sharma, S. S., & Gupta, K. P. (2021). The role of analytics and robo- 100/robo-advisors/worldwide.
advisory in investors' financial decisions and risk management: Review of Steiger, J. H. (1979). Factor indeterminacy in the 1930s and the 1970s: Some
literature post-global financial crisis. International Journal of Business interesting parallels. Psychometrika, 44(2), 157–167.
Analytics (IJBAN), 8(2), 46–62. Syam, S. S., & Courtney, J. F. (1994). The case for research in decision support
Pan, J., Ding, S., Wu, D., Yang, S., & Yang, J. (2019). Exploring behavioural systems. European Journal of Operational Research, 73(3), 450–457.
intentions toward smart healthcare services among medical practitioners: A Tao, R., Su, C. W., Xiao, Y., Dai, K., & Khalid, F. (2021). Robo advisors,
technology transfer perspective. International Journal of Production algorithmic trading and investment management: Wonders of the fourth
Research, 57(18), 5801–5820. industrial revolution in financial markets. Technological Forecasting and
Phoon, K., & Koh, F. (2017). Robo-advisors and wealth management. Journal Social Change, 163, Article 120421.
of Alternative Investments, 20(3), 79–94. Tokic, D. (2018). BlackRock Robo-Advisor 4.0: When artificial intelligence
Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). replaces human discretion. Strategic Change, 27(4), 285–290.
Common method biases in behavioural research: A critical review of the Tongco, M. D. C. (2007). Purposive sampling as a tool for informant selection.
literature and recommended remedies. Journal of Applied Psychology, Ethnobotany Research and Applications, 5, 147–158.
88(5), 879–903. Tsai, S. C., & Chen, C. H. (2022). Exploring the innovation diffusion of big
Puntoni, S., Reczek, R. W., Giesler, M., & Botti, S. (2021). Consumers and data robo-advisor. Applied System Innovation, 5(1), 15.
artificial intelligence: An experiential perspective. Journal of Marketing, Uhl, M. W., & Rohner, P. (2018). Robo-advisors versus traditional in-
85(1), 131–151. vestment advisors: An unequal game. Journal of Wealth Management,
Rasiwala, F. S., & Kohli, B. (2021). Artificial intelligence in FinTech: Un- 21(1), 44–50.
derstanding stakeholders perception on innovation, disruption, and trans- Van Doorn, J., Mende, M., Noble, S. M., Hulland, J., Ostrom, A. L.,
formation in finance. International Journal of Business Intelligence Grewal, D., & Petersen, J. A. (2017). Domo arigato Mr. Roboto: Emer-
Research, 12(1), 48–65. gence of automated social presence in organizational frontlines and cus-
Rigdon, E. E., Sarstedt, M., & Ringle, C. M. (2017). On comparing results from tomers' service experiences. Journal of Service Research, 20(1), 43–58.
CB-SEM and PLS-SEM: Five perspectives and five recommendations. Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User
Marketing: ZFP–Journal of Research and Management, 39(3), 4–16. acceptance of information technology: Toward a unified view. MIS Quar-
Ringle, C. M., & Sarstedt, M. (2016). Gain more insight from your PLS-SEM terly, 27(3), 425–478.
results on the importance-performance map analysis. Industrial Manage- Venkatesh, V., Thong, J. Y., & Xu, X. (2016). Unified theory of acceptance
ment and Data Systems, 116(9), 1865–1886. and use of technology: A synthesis and the road ahead. Journal of the
Ringle, C. M., Wende, S., & Becker, J.-M. (2015). SmartPLS 3. Bönningstedt: Association for Information Systems, 17(5), 328–376.
SmartPLS. Vimal Kumar, M., Sharma, S. K., Singh, J. B., & Dwivedi, Y. K. (2021).
Rossi, A. G., & Utkus, S. P. (2020). The needs and wants in financial advice: 'Okay, google, what about my privacy?': User's privacy perceptions and
Human versus robo-advising. Available at: SSRN 3759041. acceptance of voice-based digital assistants. Computers in Human
Saari, U. A., Damberg, S., Frömbling, L., & Ringle, C. M. (2021). Sustainable Behavior, 120, Article 106763.
consumption behaviour of Europeans: The influence of environmental Wang, Y., Kang, Q., Zhou, S., Dong, Y., & Liu, J. (2022). The impact of
knowledge and risk perception on environmental concern and behavioural service robots in retail: Exploring the effect of novelty priming on con-
intention. Ecological Economics, 189. https://doi.org/10.1016/ sumer behaviour. Journal of Retailing and Consumer Services, 68, Article
j.ecolecon.2021.107155 103002.

1472
A. Shiva, B.P. Kushwaha and B. Rishi _
Borsa Istanbul Review 23-6 (2023) 1458–1473

Weld, D. S., & Bansal, G. (2019). The challenge of crafting intelligible intel- documents.worldbank.org/curated/en/665641467998781808/pdf/97222-
ligence. Communications of the ACM, 62(6), 70–79. WP-PUBLIC-Box391472B-GCC-Engagement-Note-Final.pdf.
Williams, A. (2015). Move over, millennials, here comes Generation Z. New Zhang, Z., & Yuan, K.-H. (2018). Practical statistical power analysis using
York Times, 18, 1–7. web power and R. Granger: ISDSA Press.
Woodyard, A. S., & Grable, J. E. (2018). Insights into the users of robo- Zheng, K. W., Cheong, J. H., & Jafarian, M. (2022). Intention to adopt
advisory firms. Journal of Financial Service Professionals, 72(5), 56–66. robo-advisors among Malaysian stock investors: Using an extended
World Bank Group. (2015). Improving the quality of financial intermediation version of TAM model. Lecture Notes in Networks and Systems, 299,
in the gulf cooperation council (GCC) countries. Retrieved from http:// 658–672.

1473