Digital Financial Literacy and Investment Grip: A Study of Japanese Active Investors

Abstract

Investors’ ability to retain investments during bearish and uncertain market periods is a crucial behavioral trait for long-term wealth accumulation and reduces market instability. Nevertheless, little is understood about how digital financial literacy (DFL) shapes the capacity of increasingly digitalized financial environments. This study investigates the links between DFL and investment grip among Japanese active investors—defined here, following conventional Japanese regulatory and research practice, as individuals who maintain a securities account and have engaged with an online brokerage within the past year—building on several theoretical perspectives from behavioral science. Using survey data from 149,261 individuals with an active account at Rakuten Securities, we estimated ordered probit regression models as the main specification. The findings showed a strong positive association between DFL and investment grip, even after accounting for demographic, socioeconomic, as well as cognitive attributes. These results are supported by robustness tests employing a probit model with a binary outcome. The sample consists exclusively of digitally active retail investors; the findings are therefore most directly applicable to this subpopulation. Overall, the evidence suggests that DFL fosters investors’ capacity to endure market volatility by promoting rational decision-making and reducing panic-driven selloffs. This study offers new empirical findings that will help promote financial resilience in technology-driven markets.

1. Introduction

Investment decisions are shaped by uncertainty. An investor’s capacity to bear losses is crucial for long-term wealth accumulation (Siegel, 1998). However, empirical evidence indicates that many investors tend to liquidate their holdings prematurely when confronted with short-term losses, often sacrificing long-term benefits (Beers, 2022; Ben-David & Hirshleifer, 2012). For instance, during the 2007–2009 Financial shock, the value of the American S&P 500 fell by 56.8% and by 33.9% during the COVID-19 financial market shock (February–March 2020) (Yahoo Finance, 2025). Japan’s Nikkei 225 exhibited a similar pattern, falling by 61.4% during the period (July 2007–March 2009) and 31.3% amid the COVID-19 pandemic (January–March 2020) (Yahoo Finance, 2025) (see Figure 1). However, in both cases, the indices eventually recovered to their pre-crash levels, which is consistent with Siegel’s (1998) historical analysis that equity markets typically rebound in the long term. These episodes emphasize the importance of maintaining a stable threshold of investment grip, not only to foster long-term asset growth but also to reduce panic-driven sell-offs that can exacerbate market volatility (Nabeshima et al., 2025). Accordingly, investment grip, defined as the ability of an investor to stay committed to a financial strategy with emotional discipline, especially during market volatility and uncertainty, merits closer examination both as a determinant of individual investment behavior and as a factor contributing to broader financial stability.

Although risk tolerance and loss aversion are commonly measured through hypothetical scenarios or general attitudes toward paper losses (van Rooij et al., 2011; Gathergood & Weber, 2014), these constructs primarily reflect ex-ante preferences rather than the ability to maintain discipline once a loss has been experienced. Investment grip, while also elicited hypothetically, differs in a crucial way: it directly asks respondents to state the maximum realized drawdown (10%, 20%, 30%, or ≥40%) they would tolerate before abandoning their position. This hypothetical-yet-concrete framing places respondents in the psychological state of having already suffered a loss, thereby approximating the emotional and behavioral challenge that arises when losses move from “on paper” to “realized in the portfolio.” In behavioral terms, investment grip therefore captures the behavioral opposite of the disposition effect (Odean, 1998; Barberis & Xiong, 2009), the tendency to sell losing positions prematurely, making it a particularly relevant outcome for understanding resilience during actual market downturns.

The existing literature, including Nabeshima et al. (2025), Kuramoto et al. (2025), and Yamaguchi et al. (2025), has examined investment grip primarily through conventional psychological and behavioral frameworks, focusing on factors such as overconfidence, present bias, and financial knowledge, attitudes, and behaviors. Although these perspectives provide valuable insights, they are insufficient to explain investor behavior in today’s technology-driven financial markets. Today, more people are using mobile phones for internet banking, electronic trading, robotic advice services, and accessing financial information (Baker, 2024). While these digital advancements offer cost savings and effective financial services, they expose users to certain financial risks (Ahmed et al., 2024; Serrano, 2020). In Japan, there has been a sharp rise in illegal digital financial transactions through deception and manipulation, with recent attacks totaling over 700 million USD (The Japan Times, 2025; Japan Exchange Group, 2025; see Figure 2). Such evolving threats undermine the explanatory power of traditional financial literacy and behavioral models that are not designed to account for digital vulnerabilities or technology-mediated decision-making. Digital Financial Literacy (DFL) in this sense, which includes financial knowledge, digital proficiency, cybersecurity awareness, and behavioral adaptability (Lyons & Kass-Hanna, 2021), appears as a vital yet inadequately examined factor influencing investment grip. This study argues that incorporating DFL is key to explaining how individuals navigate financial risks in the digital era, resist emotionally driven reactions, and remain resilient in volatile markets.

Low investment grip tendency is a behavioral trait influenced by cognitive and emotional distortion that impair rational responses to market volatility, particularly in digital financial environments (Sahu et al., 2025). Several well-established theories, such as Human Capital Theory (HCT) (Becker, 1962), The Theory of Planned Behavior (TPB) (Ajzen, 1991), Technology Acceptance Model (TAM) (Davis & Granić, 2024) and Unified Theory of Acceptance and Use of Technology (UTAUT) (Venkatesh et al., 2016), Social Cognitive Theory (SCT) (Bandura, 2023) and Prospect Theory (Kahneman & Tversky, 1979) provide plausible theoretical pathways linking DFL to investment grip (motivational framework-not empirically tested- see Figure 3 for an overview).

Rather than treating the above theoretical pathways as competing explanations, we synthesize them into a coherent framework in which DFL positively affects investment grip through four distinct but interrelated mechanisms. First, DFL represents an accumulation of digital-specific human capital (Human Capital Theory) that improves investors’ ability to evaluate information quality, detect manipulation, and execute informed decisions on modern platforms. Second, it fosters self-efficacy and perceived behavioral control in digital environments (TPB and SCT), increasing investors’ confidence that they can manage volatility without panic. Third, it reduces perceived risk and uncertainty arising from phishing, algorithmic urgency cues, and platform dark patterns (TAM and UTAUT), thereby lowering the psychological threat of the trading environment. Fourth, it facilitates better cognitive reframing of realized losses (Prospect Theory), attenuating loss aversion and the disposition effect. Collectively, these mechanisms raise the psychological threshold for abandoning positions during drawdowns.

Building on the motivational framework above, we hypothesize that higher DFL is positively associated with greater investment grip, which could operate via mechanisms such as accumulation of digital-specific human capital, enhanced self-efficacy and behavioral control, reduced perceived digital risk, and improved reframing of realized losses.

2. Review of the Literature and the Study Novelty

Previous literature on finance and investment behavior provides valuable insights into how individuals manage risks in volatile markets. For instance, much of the studies has linked financial literacy to favorable financial outcomes, such as the ability to assess investment risks, achieve greater portfolio diversification, and make more forward-looking time-preference choices (Bianchi, 2017; Chu et al., 2017; Li et al., 2020). Conversely, individuals possessing limited financial knowledge are more inclined to avoid risky assets and liquidate depreciated holdings before they mature, reflecting heightened loss aversion and missed opportunities (Cupák et al., 2020; Yeh & Ling, 2022; Khan et al., 2021). While these findings suggest that financial literacy is robustly linked with greater resilience in volatile markets, most of the existing literature conceptualizes financial literacy in traditional terms, overlooking the impact of digital transformation on financial services. Moreover, most studies focus on general risk tolerance rather than investment grip, which captures investors’ responses to realized financial losses more directly. Recent studies by Nabeshima et al. (2025), Kuramoto et al. (2025), and Yamaguchi et al. (2025) address this gap by explicitly examining investment grip and demonstrating that financial literacy overconfidence and present bias reduce it, whereas financial knowledge, behavior, and attitude strengthen it. However, these studies conceptualize investor decision-making within pre-digital frameworks. Importantly, traditional financial literacy does not capture digital transaction competency, cybersecurity awareness, digital self-efficacy, interaction with algorithmic systems, or technology-mediated risk perception. These dimensions are central to modern investor behavior but remain entirely absent in established behavioral-finance models.

In response, an expanding literature has begun to explore DFL, which advances conventional financial literacy to include skills like conducting online transactions, managing digital security and identity management, and effectively using financial technologies in the digital environments (Lyons & Kass-Hanna, 2021; OECD, 2024). This emerging strand of research primarily investigated predictors of DFL acquisition. For example, Jose and Ghosh (2025), the OECD (2024), and Lal et al. (2025) identify social and economic status, psychological factors, and infrastructural elements as key predictors of individuals’ capacity to attain digital financial competencies. Additional research emphasizes how DFL facilitates meaningful participation in technology-driven financial ecosystems (Koskelainen et al., 2023; Liu et al., 2021; Soldatos & Kyriazis, 2022; Subburayan et al., 2024). Recently, scholars have begun linking DFL to broader aspects of well-being. For instance, Choung et al. (2023, 2025) linked DFL acquisition with improved life satisfaction and enhanced financial well-being in South Korea. Similarly, Amarsanaa et al. (2025) showed that DFL could potentially mitigate anxiety about life after retirement (after 65 years) among Japanese investors. Recent international evidence further underscores the broader economic and behavioral benefits of DFL and inclusion. In ASEAN countries, Bajwa et al. (2025) reveal that digital financial inclusion significantly boosts economic growth while simultaneously highlighting potential environmental trade-offs, emphasizing the need for balanced regulatory approaches in rapidly digitizing markets. Similarly, using Chinese household data, Du and Lv (2025) demonstrate that digital finance helps bridge the gap between income and consumption classes, thereby unleashing household consumption potential and reducing consumption volatility. These findings align closely with the resilience-enhancing role of DFL observed in the present study.

Although these studies advance the understanding of DFL in many ways, they fall short of explaining how DFL influences behavioral financial decisions, such as investment grip. Specifically, no existing study has examined whether DFL, a construct theoretically broader and behaviorally more complex than conventional financial literacy, affects investment grip. DFL introduces new psychological mechanisms, such as confidence in digital platforms, trust in algorithmic tools, resilience to digital fraud risks, and the ability to manage information overload in digital markets. These mechanisms are theoretically independent from classical behavioral traits and offer a new pathway through which investor composure can be strengthened during market downturns. By integrating DFL into the study of investment grip, we extend behavioral finance beyond its traditional cognitive and emotional foundations to incorporate digital-era behavioral processes that have not been previously theorized.

In summary, while existing research highlights the indispensable role of both financial literacy and digital financial literacy in shaping people’s financial decisions particularly during market volatility and growing digital financial intermediaries, it remains unclear whether DFL itself is positively associated with improved behavioral finance outcomes such as investment grip. This question is particularly pressing in Japan, where widespread use of financial digitalization is accompanied by increased cyber risks, reliance on algorithmic platforms, and the behavioral vulnerabilities of a growing retail investor base (The Japan Times, 2025; Japan Exchange Group, 2025).

This study offers several further contributions to the existing body of literature. First, it advances behavioral-finance theory by introducing DFL as a previously unexamined determinant of investment grip. Prior studies have focused on psychological biases (such as overconfidence and present bias) or traditional financial literacy, but none have incorporated digital competencies, cybersecurity awareness, platform-navigation skills, and algorithmic trust, skills that shape investor behavior in modern digital markets. Positioning DFL as a behavioral capability therefore represents a substantial conceptual extension to existing models of investment grip. Second, it integrates multiple theoretical perspectives to explain how DFL supports sustained tolerance during market volatility and digital risk. Third, this study offers the first large-scale empirical test of the DFL–investment grip relationship. Using 149,261 Japanese investors and ordered probit estimation, the analysis provides robust evidence that digital competencies significantly enhance investors’ resilience to financial losses. Finally, by examining Japan’s rapidly digitalized financial environment, the study provides timely insights into how digital vulnerabilities, cyber risks, and technology-mediated decision-making influence investor behavior, producing implications for economies experiencing similar digital transitions.

3. Data and Methods

3.1. Data Source

This study draws on data from the 2025 wave of the Survey on Life and Money, an online survey jointly administered by Rakuten Securities and the Kadoya Lab at Hiroshima University between January and February 2025. The survey targeted Rakuten online securities clients aged 18 and above who had accessed the firm’s website at least once during the preceding year. Following common practice in Japanese household finance research and the reporting conventions of major online brokerages, we refer to individuals who maintain a securities account and have engaged with (e.g., logged into) an online brokerage platform within the past year as ‘active investors. This operational definition aligns with the usage of ‘active accounts’ by firms such as SBI Securities and Rakuten Securities, which typically classify accounts with at least one login, inquiry, or transaction in the past 12 months as active (SBI Holdings, Inc., 2024; Rakuten Securities, Inc., 2024). While this definition encompasses a wide range of trading behaviors, from occasional to frequent, it is especially appropriate in the Japanese context, where the surge in retail equity participation since the early 2020s has been driven largely by occasional and long-term investors using online platforms (Japan Exchange Group, 2025; Financial Services Agency, 2025).

The resulting sample is not intended to be nationally representative of the entire Japanese adult population. Instead, it is highly representative of the subpopulation most relevant to our research question: active retail investors who conduct their stock-market trading through digital platforms. Rakuten Securities is Japan’s largest online brokerage, with over 12 million securities accounts and one of the highest market shares among individual investors (Japan Securities Dealers Association, 2025). As of 2024, a substantial majority of individual stock transactions in Japan are executed via online brokerages, with Rakuten Securities leading the market in terms of account volume and retail investor engagement. Compared with general household surveys (e.g., KHPS, JSTAR, or the Financial Literacy Survey by the Bank of Japan), our respondents are younger, wealthier, and far more digitally engaged, characteristics that closely mirror the profile of Japan’s current digitally active retail trading population rather than the broader public.

The questionnaire included items tailored to assess respondents’ DFL, investment grip, as well as various demographic, socioeconomic, and psychological characteristics. Given that some respondents had participated in earlier survey waves (2022 and 2023), information on variables such as educational attainment, financial literacy, and short-term future orientation was consolidated from these waves. After excluding the cases with missing data, the final sample included 149,261 individuals, representing approximately 66.8% of the original sample. The study assumed that the missing data follow a random pattern (missing at random, MAR) and, therefore, were unlikely to bias the results. To validate this assumption, descriptive statistics and the main model were estimated for both the full and restricted samples. Differences in coefficients, means, and standard deviations were negligible, supporting the MAR assumption. Detailed comparison tables are available upon request.

3.2. Variable Definitions

The dependent variable, “Investment grip,” refers to an investor’s ability to adhere to a predefined investment strategy or financial plan, maintaining emotional discipline and resisting impulsive decisions during periods of market volatility, uncertainty, or prolonged underperformance. To measure investment grip, respondents were asked to report how much of their investment they would retain despite a hypothetical market downturn. The survey question used to assess this is as follows.

Q1. Suppose you invest JPY 1 million in an investment trust and experience a loss. Up to what level of loss would you be willing to continue holding the investment? (Just one)

1. JPY 990,000 (JPY 10,000 loss or 1% loss);

2. JPY 900,000 (JPY 100,000 loss or 10% loss);

3. JPY 800,000 (JPY 200,000 loss or 20% loss);

4. JPY 700,000 (JPY 300,000 loss or 30% loss);

5. JPY 600,000 or less (JPY 400,000 loss or more, or 40% loss or more).

Drawing on these responses, we generated a categorical measure of investment grip with thresholds at 1%, 10%, 20%, 30%, and 40% or above. The use of hypothetical scenarios is common in behavioral research because they elicit preferences under the assumption that individuals can anticipate their likely real-world behavior with limited incentive to misrepresent their choices (Kahneman & Tversky, 1979). Similar measures have been applied in previous studies on investment grip (Braga & Fávero, 2017; Nabeshima et al., 2025), supporting the validity of our approach.

The DFL is the primary independent variable used in this study. The measurement followed the framework developed by Lal et al. (2025) and Lyons and Kass-Hanna (2021), which conceptualizes DFL across five and eight sub-dimensions:

• Understanding of Financial concepts and digital skills;
• Familiarity with digital financial services [DFS], including financial mindsets and habits
• Hands-on ability to use DFS platforms and tools;
• The ability to make sound financial decisions by applying positive financial attitudes and behaviors;
• Skills to protect oneself from internet scam and fraudulent activity.

To evaluate the financial knowledge subdimension, we employed three questions adapted from Lusardi and Mitchell (2008). These include simple interest rates, inflation, and risk diversification. Correct answers were scored as 1, while incorrect or “I do not know” responses received 0. Scores ranged from 0 to 3 and were normalized to a 0–1 index.

The other seven components of DFL were evaluated using the five Likert scale questions adopted from prior studies (Choung et al., 2023, 2025; Lal et al., 2025). Respondents rated their agreement with each statement using a five-point scale where 1 represents strongly disagree and represents strongly agree. The sub-dimension scores were determined by averaging the scores of the items within each category. The composite DFL index was derived by summing all eight sub-dimensions, yielding a value of 7 to 36, with higher scores indicating higher DFL. To facilitate meaningful comparison, all scores were standardized through Z-score normalization. To maintain consistency with prior work, the full set of survey items used to assess DFL is not reproduced in this paper. For transparency and reference, the complete list of DFL assessment questions can be found in our previously published MDPI paper (Lal et al., 2025), available at: https://doi.org/10.3390/risks13080149 (accessed on 12 October 2025).

Due to the complex nature of the DFL index, evaluating its internal consistency was crucial to ensure a reliable measurement framework. An internal consistency test was conducted for the DFL subdimensions, with the exception of financial knowledge, which was retained as a continuous measure. Cronbach’s alpha was employed for this evaluation, yielding a coefficient of 0.8972. The results indicate that the DFL index was satisfactory, supporting the validity of the instrument for comprehensively measuring DFL. More detailed reliability statistics are available upon request.

Combining objective financial knowledge questions with subjective self-assessments is now the prevailing approach in digital financial literacy measurement. Although subjective items can sometimes reflect overconfidence (Xin et al., 2024; Gignac, 2022), prior research consistently shows that self-assessed components, capturing self-efficacy and behavioral attitudes, explain actual financial decisions more strongly than objective knowledge alone (Lusardi & Mitchell, 2014; Allgood & Walstad, 2012; Lind et al., 2020). When subjective items are anchored by objective knowledge questions, the resulting composite typically reflects calibrated rather than inflated confidence (Bucher-Koenen et al., 2024). This hybrid methodology has been adopted in numerous recent DFL studies. In the Japanese context, Amarsanaa et al. (2025) combined objective financial knowledge with Likert-scale assessments of digital competence and cybersecurity awareness to examine retirement anxiety among Rakuten Securities clients, whereas Lal et al. (2025) employed an identical structure, which was objective knowledge plus seven subjective dimensions, to study the determinants of DFL in the same population. Internationally, the same strategy appears in Adnan et al. (2023) among Malaysian students, Susanti et al. (2024) among Indonesian university students, Ravikumar et al. (2022) among Indian adults, Setiawan et al. (2022) among Indonesian millennials, and Prasad et al. (2018) among Udaipur households.

To isolate the effects of DFL on investment grip, we accounted for a range of demographic, socioeconomic, and psychological factors, such as age, sex, educational level, marital status, whether the respondents have children, employment status, household income, and asset ownership. These factors are consistently linked to risk attitudes and investment behaviors (Grable et al., 2020). We also included measures of psychological attributes such as, a myopic view of the future, and risk aversion, as these factors have been linked to suboptimal investment choices (Khan et al., 2021).

Table 1. Variables Definition.

Variables	Definitions
Dependent Variables
Investment grip	Ordinal variable: How much loss respondents can withstand if they invest JPY 1 million in an investment trust (1% loss/10% loss/20% loss/30% loss/40% loss)
Investment grip binary (for robustness test)	Binary variable: 1 = respondents can withstand a loss of 30% or more if they invest JPY 1 million in an investment trust
Main Independent Variable
DFL Index	A continuous variable derived from the average scores across eight domains: digital knowledge, financial knowledge, awareness of DFS, awareness of positive financial attitudes and behaviors, practical knowledge of DFS, positive financial attitudes and behaviors, and self-protection against digital scams.
Control Variables
Male	A binary variable set to 1 if the respondent’s gender is male, and 0 otherwise
Age	Continuous variable representing the age of the respondents
Age Squared	Continuous variable representing the square of respondents age
University Degree	Binary variable set to 1 if the respondent has a university degree or higher, and 0 otherwise
Unemployment	Binary variable set to 1 if the respondent is unemployed, and 0 otherwise
Married	Binary variable set to 1 if the respondent is married, and 0 otherwise
Having a Child	Binary variable set to 1 if the respondent has at least one child, 0 = otherwise
Household Income	Continuous variable indicating the estimated annual household income in Japanese yen
Log of Household Income	The natural logarithm of the estimated annual household income in Japanese yen
Household Asset	Continuous variable representing the household financial asset balance in Japanese yen
Log of Household Asset	The natural logarithm of the household financial asset balance in Japanese yen
Risk Aversion	Continuous variable indicating risk preference (percentage score based on the question, “Usually when you go out, how high must the probability of rainfall be before you take an umbrella?”)
Myopic view of the future	A binary variable assigned a value of 1 if the respondent agrees with the statement “Since the future is uncertain, it is a waste to think about it.” and 0 if otherwise.

provides detailed definitions and measurements of the study variables.

3.3. Descriptive Statistics

Table 2. Descriptive Statistics.

Variable	Mean	Std. Dev.	Min	Max
DFL Index a	30.24	4.572	7	36
Investment grip	0.245	0.127	0.01	0.4
Investment grip (Binary)	0.472	0.499	0	1
Investment grip (Categorical) b	-	-	-	-
1%	0.048
10%	0.240
20%	0.240
30%	0.157
≥40%	0.315
Male	0.676	0.468	0	1
Age	46.356	12.219	18	90
Age Square	2298.163	1175.832	324	8100
Married	0.672	0.469	0	1
Having a Child	0.592	0.491	0	1
University Degree	0.644	0.479	0	1
Unemployed	0.069	0.254	0	1
Annual Income	7,721,102.5	4,288,951.6	1,000,000	20,000,000
Annual Income (Logarithm scale)	15.69	0.621	13.816	16.811
Household Financial Asset	21,667,869	25,580,218	2,500,000	1.000 × 108
Household Financial Asset (Logarithm scale)	16.279	1.111	14.732	18.421
Myopic View of the future	0.152	0.359	0	1
Risk Aversion	0.535	0.238	0	1
Number of Observations: 149,261

^a Raw score before standardization. ^b Investment grip (Categorical) is displayed as category shares (rows 1%, 10%, 20%, 30%, ≥40%), so no single Mean/SD/Min/Max is shown for that row.

presents the summary statistics for the main variables in this study. On average, respondents demonstrated an investment grip equivalent to tolerating losses of approximately 24.5%. The categorical distribution of investment grip showed that 4.8% of respondents could only keep investments after a loss of 1% loss, while 24.0% could keep investments after a loss of up to 10%, 24.0% up to 20%, 15.7% up to 30%, and the remaining 31.5% indicated an investment grip of 40% or more. A supplementary binary indicator showed that about 47.2% of respondents were willing to tolerate losses of 30% or more The primary independent variable, DFL index, had an average score of 30.2.

The demographic profiles of the respondents indicated that 67% were male, the average age was 46 years, 67% were married, and 59% had at least one child. Socioeconomic characteristics indicate that 64% hold at least a university degree, approximately 7% are unemployed, the average annual income is ¥7,721,103, and household financial assets averaged at ¥21,700,000. Regarding psychological characteristics, 15% of the respondents believed that the future was uncertain and considered it a waste to think about, while 53% were risk-averse.

Figure 4 presents the distribution of loss tolerance across different levels of the DFL Index. The results indicate a clear positive association, showing that higher DFL is associated with greater willingness to tolerate investment losses. More specifically, individuals with relatively low DFL scores (around 27–28) report the lowest loss tolerance (1%). As DFL increases, individuals display progressively higher tolerance for losses, with scores stabilizing around 29–31 for loss tolerances of 10% to 30%. Notably, those with the highest DFL scores (approximately 32) are concentrated among respondents who are willing to tolerate losses of 40% or more. This pattern suggests that DFL enhances investors’ ability to understand and manage risk within the digital financial environment, thereby enabling them to accept higher potential losses. In contrast, individuals with lower DFL are more risk-averse, reflecting limited capacity to cope with market volatility.

3.4. Methods

We employed ordered probit regression analysis to model investment grip as an ordinal outcome. All analyses were performed using Stata BE version 18. Following the hierarchical specification outlined in Equations (2)–(5), we first estimated the effect of DFL on investment grip. Then, we progressively introduced demographic, socioeconomic, and psychological constructs into subsequent models. Notably, Equation (5), which incorporates all predictors, serves as the primary focus of this study. This structure enabled us to assess the incremental contribution of each conceptual block to explain the variation in investment grip. To investigate how DFL and other control variables influence investment grip among Japanese investors, the study utilized the following functional equation:

$$Y_i=\left({DFL}_i,X_i,{\epsilon }_i\right)$$

(1)

where $Y_i$ is representing the ith respondent’s investment grip. ${DFL}_i$ represents the respondent’s DFL level, and $X_i$ is a vector capturing demographic, social and economic status, as well as the cognitive and emotional factors. ${\epsilon }_i$ is the error term accounting for omitted factors that influence investment grip.

Equations (2)–(5) provided the detailed specifications of Equation (1) with each one corresponding to a distinct explanatory variable.

$${InvestmentGrip}_i={\beta }_0+{\beta }_1{DFL}_i+{\epsilon }_i$$

(2)

$$\begin{array}{cc}\hfill Investment{Grip}_i& \\ & ={\beta }_0+{\beta }_1{DFL}_i+{\beta }_2{Male}_i+{\beta }_3{Age}_i+{\beta }_4{AgeSquare}_i\hfill \\ & +{\beta }_5{Married}_i+{\beta }_6{Havingchild}_i+{\epsilon }_i\hfill \end{array}$$

(3)

$$\begin{array}{cc}\hfill {InvestmentGrip}_i& \\ & ={\beta }_0+{\beta }_1{DFL}_i+{\beta }_2{Male}_i+{\beta }_3{Age}_i+{\beta }_4{AgeSquare}_i\hfill \\ & +{\beta }_5{Married}_i+{\beta }_6{Havingchild}_i+{\beta }_7{UniversityDegree}_i\hfill \\ & +{\beta }_8{Unemployed}_i+{\beta }_9{AnnualIncome}_i\hfill \\ & +{\beta }_{10}{HouseholdAssets}_i+{\epsilon }_i\hfill \end{array}$$

(4)

$$\begin{array}{cc}\hfill {InvestmentGrip}_i& \\ & ={\beta }_0+{\beta }_1{DFL}_i+{\beta }_2{Male}_i+{\beta }_3{Age}_i+{\beta }_4{AgeSquare}_i\hfill \\ & +{\beta }_5{Married}_i+{\beta }_6{Havingchild}_i+{\beta }_7{UniversityDegree}_i\hfill \\ & +{\beta }_8{Unemployed}_i+{\beta }_9{AnnualIncome}_i\hfill \\ & +{\beta }_{10}{HouseholdAssets}_i+{\beta }_{11}{MyopicViewoftheFuture}_i\hfill \\ & +{\beta }_{12}{RiskAversion}_i+{\epsilon }_i\hfill \end{array}$$

(5)

We checked the predictor multicollinearity using pairwise correlation coefficients and variance inflation factors (VIFs) to ensure the reliability of the estimates. The maximum VIF was 1.35, and the highest correlation coefficient was 0.6. Both values fall below conventional thresholds (VIF < 5; correlation < 0.8), indicating that multicollinearity is not a concern (Shrestha, 2020). To save space, the complete results are available upon request.

To verify the robustness of our empirical findings, we conducted a re-estimation of the model using an alternative binary measure of investment grip. This variable was coded as 1 if the respondent’s level of investment grip is 30% or higher, and 0 otherwise. A similar approach was used in prior research (Nabeshima et al., 2025). Probit regression using the binary specification confirmed that the main results are consistent with those from the ordered probit analysis.

4. Empirical Findings

Table 3. Ordered Probit Results: DFL Index and Investment Grip.

	Dependent Variable: Investment Grip
Variables	Model 1.1	Model 2.1	Model 3.1	Model 4.1
DFL Index	0.2196 ***	0.2148 ***	0.1775 ***	0.1768 ***
	(0.0031)	(0.0031)	(0.0032)	(0.0032)
Male		0.3804 ***	0.3803 ***	0.3816 ***
		(0.0061)	(0.0062)	(0.0062)
Age		0.0373 ***	0.0273 ***	0.0270 ***
		(0.0015)	(0.0017)	(0.0017)
Age Square		−0.0005 ***	−0.0004 ***	−0.0004 ***
		(0.000)	(0.000)	(0.000)
Married		−0.0337 ***	−0.0857 ***	−0.0865 ***
		(0.0075)	(0.008)	(0.008)
Having a Child		−0.0555 ***	−0.0218 ***	−0.0233 ***
		(0.0074)	(0.0074)	(0.0074)
University Degree			−0.0553 ***	−0.0521 ***
			(0.0062)	(0.0062)
Unemployed			0.0104	0.0099
			(0.013)	(0.013)
Annual Income			−0.0088	−0.0086
			(0.006)	(0.006)
Household Financial Asset			0.2313 ***	0.2327 ***
			(0.0031)	(0.0031)
Risk Aversion				−0.0939 ***
				(0.0124)
Myopic View of the future				0.0038
				(0.0079)
/cut1	−1.6864 ***	−0.8486 ***	2.2711 ***	2.2472 ***
	(0.0056)	(0.0377)	(0.0892)	(0.0894)
/cut2	−0.5551 ***	0.3045 ***	3.4525 ***	3.4289 ***
	(0.0035)	(0.0376)	(0.0894)	(0.0895)
/cut3	0.0873 ***	0.959 ***	4.126 ***	4.1025 ***
	(0.0033)	(0.0376)	(0.0895)	(0.0896)
/cut4	0.5084 ***	1.3883 ***	4.5668 ***	4.5434 ***
	(0.0034)	(0.0377)	(0.0896)	(0.0897)
Observations	149,261	149,261	149,261	149,261
Pseudo R2	0.0131	0.0245	0.0396	0.0398

Robust standard errors are in parentheses: *** p < 0.01.

shows results for ordered probit regression. Across all models, the findings show a positive association between DFL and investment grip at the 1% significance level. This suggests that investors with higher DFL are more likely to stay calm and adhere to their investment strategies during market downturns.

From Model 2.1 onward, demographic controls were introduced. Subsequently, socioeconomic variables were added in Model 3.1 and psychological traits in Model 4.1. Among the demographic variables, male investors consistently exhibit higher investment grip than female investors at the 1% significance level. Investment grip also increases with age, as indicated by the positive and significant age coefficients at the 1% level. However, the negative coefficients of the squared age term suggest a nonlinear relationship, indicating a turning point beyond which age may reduce investment grip. Being married and having children were negatively associated with investment grip at the 1% significance level. Socioeconomic factors had mixed effects. Having at least a university education and a higher household income are negatively associated with investment grip at the 1% significance level. Contrastingly, greater household financial assets are strongly and positively associated with investment grip, indicating that wealthier investors are more likely to stay the course during a period of market underperformance. Unemployment did not exhibit a statistically significant relationship with investment grip. Finally, psychological traits exhibited alignment with theoretical expectations. Risk aversion is negatively associated with investment grip, and is significant at the 1% level; however, a myopic view of the future is not statistically significant, suggesting a counterintuitive relationship that may reflect behavioral or contextual factors.

Table 4. Average Marginal Effects on Investment Grip (Ordered Probit Model).

	Delta-Method
	dy/dx	Std. Err.	z	p > |z|	[95% Conf. Interval]
DFL Index	−0.017	0.000	−49.140	0.000	−0.017	−0.016
Male	−0.036	0.001	−53.190	0.000	−0.037	−0.034
Age	−0.003	0.000	−16.100	0.000	−0.003	−0.002
Age Square	0.000	0.000	23.680	0.000	0.000	0.000
Married	0.008	0.001	10.690	0.000	0.007	0.009
Child	0.002	0.001	3.080	0.002	0.001	0.004
University Degree	0.005	0.001	8.320	0.000	0.004	0.006
Unemployed	−0.001	0.001	−1.040	0.297	−0.004	0.001
Log of Income	0.001	0.001	1.710	0.088	−0.000	0.002
Log of Asset	−0.022	0.000	−63.350	0.000	−0.022	−0.021
Risk Aversion	0.009	0.001	7.510	0.000	0.006	0.011
Myopic View	−0.000	0.001	−0.380	0.707	−0.002	0.001

presents average marginal effects from the ordered probit model (Model 4.1), showing the change in the probability of being in the lowest investment grip category. A one-standard-deviation increase in the standardized DFL index is associated with a 1.7 percentage-point decrease in the probability of being in the lowest investment grip category (p < 0.001). Similar patterns appear for other key characteristics. Being male reduces the probability of falling into the lowest investment grip category by 3.6 percentage points, and each additional year of age lowers this probability by 0.3 percentage points (although the significant age-squared term suggests the magnitude of this effect diminishes at older ages). A one-log-point increase in household financial assets is associated with a 2.2 percentage-point reduction in the probability of being in the lowest category (p < 0.001).

To assess the robustness of our findings, we re-estimated the main models using a probit regression with the binary “Investment Grip Binary” variable. The probit estimates across all model specifications, presented in

Table 5. Robustness Test: Probit Results using Binary Investment Grip and DFL Index.

	Dependent Variable: Investment Grip Binary
Variables	Model 1.1	Model 2.1	Model 3.1	Model 4.1
DFL Index	0.2212 ***	0.2139 ***	0.1755 ***	0.1749 ***
	(0.0037)	(0.0037)	(0.0037)	(0.0037)
Male		0.3438 ***	0.3482 ***	0.3494 ***
		(0.0072)	(0.0074)	(0.0074)
Age		0.0428 ***	0.0312 ***	0.031 ***
		(0.0021)	(0.0023)	(0.0023)
Age Square		−0.0005 ***	−0.0005 ***	−0.0005 ***
		(0.000)	(0.000)	(0.000)
Married		−0.047 ***	−0.0874 ***	−0.0881 ***
		(0.0088)	(0.0095)	(0.0095)
Having a Child		−0.0666 ***	−0.0305 ***	−0.0319 ***
		(0.0087)	(0.0088)	(0.0088)
University Degree			−0.0803 ***	−0.0772 ***
			(0.0074)	(0.0074)
Unemployed			−0.0161	−0.0166
			(0.0161)	(0.0161)
Annual Income			−0.036 ***	−0.0358 ***
			(0.007)	(0.007)
Household Financial Asset			0.2463 ***	0.2475 ***
			(0.0037)	(0.0037)
Risk Aversion				−0.0887 ***
				(0.0143)
Myopic View of the future				0.0065
				(0.0093)
_cons	−0.0882 ***	−1.058 ***	−3.989 ***	−3.968 ***
	(0.0033)	(0.0467)	(0.105)	(0.1052)
Observations	149,261	149,261	149,261	149,261
Pseudo R2	0.0199	0.0363	0.0614	0.0616

Robust standard errors are in parentheses: *** p < 0.01.

, are broadly consistent with the results from the Ordered Probit regression models reported earlier. Specifically, the direction, magnitude, and statistical significance of the main independent variable, as well as other control variables, remain unchanged. This consistency across models enhances confidence in the reliability and stability of our results and suggests that the relationships are insensitive to the estimation technique used.

5. Discussion

Exploring how DFL influences investment grip offers a crucial yet underexplored perspective within behavioral finance. This is especially important owing to the rapid digitalization of financial markets. Utilizing data from 149,261 active Japanese investors, this study draws upon established behavioral, learning, and psychological pathway theories. Throughout this study, the term “active investors” follows the standard definition used in Japanese household finance research and by major brokerages. As far as we are aware, this study is among the first to conduct a comprehensive empirical analysis of the connection between DFL and investment grip. The results reveal a strong and positive association between higher DFL and greater investment grip (i.e., willingness to tolerate larger realized drawdowns). This association is consistent with the several theoretical perspectives presented in the theoretical section, whereby DFL could foster enhanced grip primarily through four complementary pathways: (i) operating through enhanced self-efficacy and perceived behavioral control in digital environments (TPB, SCT), (ii) reduced perceived risk from fraudulent or algorithmic cues (TAM, UTAUT), and (iii) improved cognitive reframing of realized losses (Prospect Theory) and (iv) accumulation of valuable digital-specific human capital (HCT). However, because we do not directly measure these intermediate constructs, the exact mediating channels remain an open and important question for future research. The relatively conservative average investment grip of approximately 24.5% observed in our sample is fully consistent with the risk profile of Japan’s broad population of active retail investors, the vast majority of whom are occasional or long-term participants rather than frequent traders (Japan Exchange Group, 2025; Financial Services Agency, 2025). Although direct prior evidence linking DFL specifically to investment grip remains scarce, the pattern is fully in line with broader behavioral finance research showing that superior knowledge, confidence, and behavioral control in digital contexts improve decision quality and psychological resilience under market stress (OECD, 2021; Song & Valencia, 2024).

For instance, higher levels of financial knowledge are positively associated with more accurate risk assessment and help reduce the likelihood of premature asset sales. Within the framework of the HCT, such knowledge represents an investment in skills that yield long-term resilience. The ability to appraise risk and endure short-term decline supports greater investment grip by promoting commitment to long-term strategies. Similarly, empirical evidence shows that financially literate investors display lower loss aversion and greater persistence under market turbulence, both hallmarks of higher investment grip (Bianchi, 2017, Cupák et al., 2020; Yeh & Ling, 2022). Confidence in using digital financial platforms also reinforces loss-tolerant behavior. According to the TAM and the UTAUT, digital competence enhances trust and perceived control, reducing anxiety during downturns and promoting compliance under pressure. Similarly, prior research indicates that proficiency in digital finance strengthens investment intentions (Sukumaran et al., 2023), forming a psychological basis for sustained engagement even amid losses. Furthermore, familiarity with DFS curbs impulsive trading and lengthens holding periods, behaviors that align with the TPB, which emphasizes deliberate informed intentions guiding consistent actions (Abdallah et al., 2025; Grable et al., 2024).

Applying sound financial behavior in digital contexts also mitigates herding and loss aversion. The SCT attributes such disciplined behavior to enhanced self-efficacy, operationalized through investors ability to view temporary losses as manageable rather than as triggers for panic selling (Liu et al., 2021; Peter & Mathew, 2025). Similarly, cybersecurity competence, a key component of DFL, reduces perceived vulnerability and fosters a focus on long-term financial objectives. From the Prospect Theory perspective, reframing potential losses within a broader temporal horizon diminishes overreactions to short-term fluctuations. Evidence linking sub-dimension of cybersecurity confidence to improved economic well-being and reduced risk aversion supports this interpretation (Makridis & Liu, 2025; Venkatesan, 2023). Overall, these theoretical and empirical insights suggest that DFL, as an integrated capability, may foster both cognitive and behavioral resources for investors to strengthen their investment grip and maintain composure amid digital market volatility. These mechanisms not only explain individual resilience but also have macroeconomic implications: as Bajwa et al. (2025) show, DFL-driven inclusion can boost economic growth in emerging markets, while Du and Lv (2025) illustrate how it stabilizes household consumption by bridging income-consumption gaps, suggesting that enhanced grip may contribute to broader financial stability in digital economies.

Regarding the control variables, demographic characteristics exert distinct influences on Japanese investors’ investment grip, as indicated by our ordered probit estimations (Table 3). Although women have traditionally managed household finances in post-war Japan (Iwao, 1993), our results indicate that male investors exhibit higher investment grip. This pattern aligns with gendered confidence in financial decision-making and socialized risk-taking tendencies, which may attenuate sensitivity to short-term losses (Barber & Odean, 2001; Giannikos & Korkou, 2025; Nabeshima et al., 2025). Such confidence appears to foster greater composure during market downturns, particularly on fast-paced digital platforms where decisiveness and perceived control are critical (OECD, 2021; Song & Valencia, 2024). Within behavioral finance frameworks, this composure reflects not only psychological readiness but also digital financial literacy, reinforcing how gendered confidence, particularly among male investors, translates into platform competence and loss-tolerant behavior.

Investment grip also follows an age-related pattern: it rises through early and middle adulthood but declines later in life. Younger and middle-aged investors tolerate greater losses in pursuit of higher returns, given their longer recovery horizons, whereas older investors, reflecting Japan’s aging demographic, tend to prioritize capital preservation due to shorter investment horizons (Bodie et al., 1992; Cocco et al., 2005; Horioka, 1990). Marital status similarly shapes risk preference, whereby being married is associated with lower investment grip, as household responsibilities in Japan’s high-cost urban setting encourage financial conservatism to ensure family stability (Ono, 2010; Nabeshima et al., 2025). The influence of having children present a more complex dynamic, whereby parenthood is negatively associated with higher investment grip, consistent with prioritizing short-term financial security (Zumbühl et al., 2013).

Socioeconomic factors further shape investment grip through differences in economic resources and financial priorities. University-educated investors negatively associate with higher investment grip, contrary to the common expectations that education encourages risk-taking (Grable, 2000). In Japan, higher education often cultivates prudent financial attitudes and a preference for conservative investment strategies that minimize losses (Fiel’ardh, 2024; Nabeshima et al., 2025). Higher household income is similarly associated with lower investment grip, as wealthier investors tend to emphasize capital preservation through low-risk portfolios to maintain social and financial status (Calvet et al., 2007). In contrast, greater household financial assets are positively related to investment grip: accumulated wealth provides a buffer that markets temporary losses less consequential and allows investors to endure market volatility more comfortably (Nabeshima et al., 2025). Psychological variables also influence investment grip through behavioral dispositions. Risk aversion is negatively associated with higher investment grip, indicating that loss-sensitive investors are prone to premature selling to avoid emotional distress (Kahneman & Tversky, 1979; Nabeshima et al., 2025).

There are, however, several limitations to this study that should be acknowledged. First, the measure of investment grip is based on a hypothetical loss scenario, which captures intended loss tolerance rather than revealed behavior during an actual market downturn. This discrepancy is a recognized limitation in behavioral finance research, as real-world stress can yield stronger emotional and psychological reactions than survey-based scenarios. Nevertheless, hypothetical loss vignettes remain widely used in large-scale investor research due to the practical and ethical impossibility of observing real-time investor reactions during market crashes. Moreover, prior studies show that stated loss tolerance is strongly predictive of actual selling behavior under stress, suggesting that the measure, while imperfect, provides meaningful insight into grip tendencies (Braga & Fávero, 2017; Kahneman & Tversky, 1979; Nabeshima et al., 2025). Second, a primary limitation of this study is its reliance on cross-sectional data, which precludes definitive causal inference. Although the analysis controls for an extensive array of observable covariates, including gender, risk aversion, myopic view, income, wealth, age, education, marital status and so on, unobserved time-invariant heterogeneity could still partly drive the observed association between DFL and investment grip. Establishing causality would require longitudinal data, randomized interventions (e.g., digital-literacy or cybersecurity training programs), or credible exogenous shocks to DFL. Notwithstanding this limitation, the large magnitude, high statistical and economic significance, and remarkable consistency of the association across numerous specifications, reinforced by robustness checks (binary investment grip) provide strong suggestive evidence that DFL plays a critical role in fostering investors’ behavioral resilience in contemporary digital markets. Third, the sample consists exclusively of active clients of Rakuten Securities. Because the sample comprises only digitally active retail investors who trade through online platforms, results may not generalize to offline investors, less digitally engaged households, or individuals who do not hold securities accounts. While Rakuten Securities is Japan’s largest online brokerage with over 12 million accounts, representing approximately 11% of the adult population and nearly one-quarter of the domestic online securities market, its clients are younger, wealthier, and more technologically engaged than the average Japanese citizen. This selection makes the sample highly relevant to the research question (behavioral resilience in digital trading environments) but limits broader population-level inferences. Finally, although the DFL index used here demonstrates high internal reliability, it may not encompass all facets of digital capability, as technologies and associated risks evolve rapidly. Nevertheless, the applied multidimensional framework aligns with established constructs in the existing literature (Lyons & Kass-Hanna, 2021; Lal et al., 2025), supporting the robustness of the present analysis.

Subsequent studies could overcome these limitations by employing longitudinal data to examine how DFL and investment grip co-evolve as investors gain more digital experience or encounter real market downturns. Given this study’s focus on Japan, a country with high digital penetration and aging demographics, comparative research across different economies or investor groups (such as younger traders or retirees) would provide fruitful evidence on whether the behavioral effects of DFL are context-specific or universal. Furthermore, as fintech continues to advance, refining the DFL index to include emerging domains, such as AI-based advisory tools, robot investing, and cybersecurity behavior, would help capture the dynamic nature of digital competence. Future research could also formally test the mediating roles of digital-specific human capital accumulation, self-efficacy, perceived digital risk, and loss reframing proposed in the theoretical section of the paper using longitudinal or experimental designs. Finally, using qualitative or mixed-method approaches could offer deeper insights into the psychological mechanisms, such as trust, digital confidence, and perceived control, that link DFL to sustained composure during financial losses. Such extensions would not only strengthen causal understanding but also deepen the theoretical integration of DFL within behavioral and psychological finance frameworks.

6. Conclusions

This study is a pioneering step in linking DFL to investment grip among Japanese investors. By adopting a theory-driven framework and leveraging a large-scale dataset, the analysis provides robust evidence that a higher DFL is positively associated with investors’ enhanced resilience to financial losses. However, the sample consists exclusively of digitally active retail investors; the findings are therefore most directly applicable to this subpopulation. These findings extend behavioral finance to the digital era by demonstrating how sub-dimensions of DFL, such as digital competence, cybersecurity awareness, and adaptive decision-making sustain investor confidence during volatility.

Demographic, socioeconomic, and psychological characteristics, including age, gender, marital and parental status, education, income, assets, and risk aversion, also significantly associates with investment grip. These results underscore the multifaceted nature of investor behavior and highlight the importance of comprehensive modeling when examining behavioral resilience in digital financial environments.

These findings suggest several strategies for strengthening investment grip through targeted DFL interventions.

• Personalized DFL coaching: Financial advisors at institutions, such as Rakuten Securities, can deliver individualized training to navigate trading platforms safely and recognize phishing attempts. Enhanced confidence in digital finance can reduce impulsive sell-offs triggered by loss aversion.
• Interactive learning tools: Investors may benefit from tailored online simulations that cultivate digital transactions and cybersecurity skills by directly addressing the competencies presented here to strengthen their investment grip.
• Targeted educational workshops: Programs designed for groups with lower investment grip, such as women, older adults, and married investors, can improve digital self-efficacy and resilience to digital risks, including fraud and misinformation.
• Peer-led online communities: Participating in investor forums that share DFL strategies can reinforce self-efficacy and foster adaptive investment grip.

By implementing these micro-level, DFL-focused initiatives, financial institutions and individuals can enhance behavioral resilience and maintain composure in increasingly digitalized markets. Thus, investors can better withstand temporary losses and contribute to a more stable and inclusive digital financial ecosystem.