Proxy Accounts and Behavioural Nudges: Investigating Support for Older Adults and their Financial Delegates

Banking tasks are listed as a component of Lawton's Instrumental Activities of Daily Living scale [24]. The exact portion of older adults engaged in online banking worldwide is unknown, but a recent Italian survey showed 79% of adults aged 60+ used the internet for banking [33], while a pre-pandemic 2019 study in Canada showed 42% of adults aged 55+ used online banking [32].

Management of finances becomes more difficult as older adults age [2, 25]. These difficulties, combined with limited comfort with technology [14], can lead older adults to seek help from close others with the use of banking technologies [22, 30]. Data from a 2018 Statistics Canada survey shows that 38% of caregiver respondents reported helping older adults with banking tasks [5]. This assistance is a valuable component of informal care work [3, 30] that can enabling aging in place [43].

Close others serving as financial delegates are making greater use of online banking: it is geographically convenient when the delegate does not live close to the older adult and temporally convenient when delegates cannot easily get to a bank during business hours [22]. Banking information is highly sensitive and contextually specific, which increases the privacy concerns older adults may consider [37, 39] when granting account access to close others [15, 16]. Previous work has shown that technologies sometimes acknowledge older adults’ desire for privacy while simultaneously assuming that they do not actually need it [27, 28]. When delegating financial tasks, older adults may share account credentials with close others [22] because the benefits outweigh perceived risks.

There are other possible mechanisms that financial institutions could offer to support older adults delegating banking tasks to close others. One-Time Passwords (OTPs) are generated passwords that are only valid once, and allow a delegate access to an account for a single session [31]. Proxy accounts allow an older adult to give a delegate limited access and permissions through a separate set of credentials [11]. Joint accounts (which may also support separate credentials) are distinct from proxy accounts in that joint accounts entail equal authority and ownership over accounts by both parties [50]. This is not typically appropriate except in cases where the close other is the spouse. In a proxy setup, the primary user maintains full authority and ownership of the assets, and can limit a delegate's access with fine-grained controls. Power of Attorney (PoA) is a legal framework that may be used to justify delegate access. Banks in Canada provide ‘trustee accounts’ to delegates who have invoked a PoA, but this removes account control from the older adult entirely.

When older adults delegate banking tasks to close others, it can expose them to a risk of misconduct by the close other [41]. While the vast majority of close others are well-intentioned and maintain financial propriety, statistics show that financial abuse by caregivers is a real problem, with at least 6.8% of older adults globally having experienced financial abuse [51]. It is important to differentiate financial misconduct from fraud: fraud is committed by individuals unknown to the victim, whereas misconduct is perpetrated by known and trusted individuals [9]. Financial misconduct occurs on a spectrum ranging from minor financial improprieties to serious financial abuse. Prior research shows that older adults tend to over-estimate the risk of fraud perpetrated by strangers and under-estimate the risk of financial abuse perpetrated by close others [34]. Given the sensitive and personal nature of financial abuse, and the fact that older adults may be dependent on the perpetrator for various types of care, it is likely that financial abuse goes under-reported [20].

Routine Activity Theory (RAT) posits that many criminal trends are born from non-malicious behaviour that an individual partakes in frequently [44]. The theory has been applied to describe the origin of financial and cyber crimes [42]. Setterlund et al. have proposed RAT as an explanation for the connection between financial misconduct and informal care [44]. They explain that routine access to an older adult's bank account could provide the repeated ease of access that might lead to close others engaging in initially minor financial improprieties. Close others might self-rationalize these improprieties as compensation for their care burden, and/or as familial entitlements. Over time, and if not detected, these improprieties could devolve into more serious and serial forms of financial abuse. The authors suggest that the participation of a “guardian” providing oversight and monitoring of the close other's behaviour can mitigate this abuse [44].

A pair of screenshots of our high-fidelity online banking prototype. In both there is a dark navigation bar along the top that contains, from left to right: the fake bank's name and icon, a selected “Home” link and unselected links for “Accounts”, “Transfer Funds”, “Manage Payments”, and “Add Services”. On the right side of the bar is a link for “Logout”. Below the nav bar is an orange bar containing the message “You are currently banking on behalf of Maggie Winters” and a button labelled “Switch Accounts”. Below that is the main body of each page, black text on a white background. In the left screenshot a heading says “Hi Leigh”. Below is a panel titled “Accounts Summary” that contains a table with two accounts: "Account A (SAVINGS)" with a balance of $2876.55“ and ”Account B (CHEQUING)" with a balance of $267.89. Below the accounts is another panel titled “Cards Summary”, containing a table with a single entry: “Visa-Cash Back” with a balance of $180.71 and $1819.29 available. To the right of the account and cards summaries is a panel titled “Recent Activity” that contains information about the number of logins and transactions above a panel titled “Quick Links” with 4 links in a list format. In the right screenshot a heading says “Transfer Funds” over a form with three input fields. The first, labeled “From”, has a savings account selected, the second, labeled “To”, has a cellphone company selected, and the third, labeled “Amount”, has $56.74 filled. All fields have a green border on the left edge and the amount field has a slightly pinkish background. Below the inputs is a large text box with a yellow background containing the following message: “The value of the payment to this payee is different from your usual amount. Would you like to reset this to the usual amount, or submit with this new amount?” Below the message is a wide blue button labeled “Submit”‘ next to a narrower green button labeled “Reset”.

Behavioural nudges [46] have been used in digital interfaces across many domains [47, 49]. The principle of behavioural nudges is to manipulate a user's decision-making context by changing the value proposition of certain actions to influence their choices without limiting their options or removing agency [29]. Caraban et al. have investigated the similarities in psychological heuristics that nudges leverage and grouped nudges into 6 higher level categories: facilitation, confrontation, deception, social influence, fear, and reinforcement [6]. They also placed the nudges along two different axes: one indicating how transparent the nudge mechanism was to the user, and another that specified whether the nudge engaged the user's reflective versus automatic thinking [6]. Research into the efficacy of behavioural nudges has identified some important overall design considerations. For example, there is evidence that nudges that attempt to dissuade individuals from specific behaviour have more impact when presented to the person before they exhibit the behaviour, such as in dissuading smoking [18]. This is of particular relevance in delegated online banking as it suggests that nudges related to financial impropriety will be more effective at preventing small initial acts of misconduct by delegates as opposed to stopping an existing pattern of financial abuse, an idea supported by Routine Activity Theory [44].

Some researchers consider the use of behavioural nudges to be completely inappropriate and unethical given that nudges inherently attempt to manipulate the choices of the user. In contrast, Thaler claims that nudges are ethical in that they are unavoidable (as designers are always influencing behaviour whether they are doing so intentionally or not) and in that they do not actually constrain behaviour but rather provide guidance [46]. Hansen and Jespersen discuss both of these positions, concluding that the ethics of a nudge are nuanced and a nudge is more ethical when it is transparent and/or targets reflective (as opposed to automatic) thought [17].

Research on the attitudes and behaviours of informal caregivers has led to the introduction of the ‘good carer’ concept, which has been used in interview-based studies as a lens to help understand the motivations for caregiver responses [4], or to help guide study methodology [19]. The ‘good carer’ describes the tendencies of informal caregivers to position themselves as ethically virtuous, even in the face of conflicting motivations [35]. Caregivers in these studies gave performative narratives to emphasize their virtuous position, and were reticent to share any negative emotions regarding the care they provided.

Our prototype included 11 nudges leveraging a variety of mechanisms. The following considerations related to the delegated banking context guided our design process:

• We determined that almost all banking tasks can be subtly influenced in the appropriate context. Nudges were spread out across tasks to avoid over-exposing close others [45].
  
• To support close others and promote financial propriety without causing harm to the relationship, we avoided nudge mechanisms that leverage negative psychological effects even if they could potentially deter misconduct.
  
• We wanted to explicitly take a supportive position towards close others, so the tone of messages in the nudges was positive and supportive in order to avoid alienating close others.
  

Three sections arranged horizontally, separated by vertical bars. The left section contains a silhouette of a person with smaller icons of a bank building, a laptop, and a group of people arranged above. Below are labels stating “Exploring Caregiving and Banking Contexts” and “30 minutes”. The center section contains a silhouete of a person in profile sitting in front of a computer with a solid speech bubble above their head. The labels for this section state “Think Aloud Interaction with Prototype” and “30 minutes”. The right section is a silhouette of a person with a thought bubble above their head that is labeled “Debrief and Discussion” and “30 minutes”.

We intentionally chose to omit nudges we considered inappropriate in a context where the close others are likely to be trusted by the older adults. These included any mechanisms Caraban et al. categorized as leveraging deception-based or fear-based psychological effects [6]. We also chose to omit nudges that target users’ ‘automatic mind’ in a non-transparent way, as we agree with Hansen and Jesperson that these types of nudges are akin to deceit [17]. We prioritized nudges leveraging mechanisms such as raising visibility, creating friction, or public commitment. The banner at the top of the prototype, reminding the proxy that they are banking on behalf of an older adult, is an example of a transparent nudge targeting the proxy's ‘reflective mind’ (see Figure 1).

Participants were required to currently use online banking services in Canada. We recruited using posters posted on the University of Manitoba campus and throughout Winnipeg, along with mailing lists for gerontological research communities, university alumni, and credit union associations. For this study we aimed to gather feedback and insights from the following groups:

• Adults (18+) who already perform online banking tasks on behalf of an older adult (65+).
  
• Older adults (65+) who currently use online banking, with or without delegated assistance.
  
• Adults who work in client-facing roles in a Canadian financial institution.
  

Older adults are intended to be the beneficiaries of our designs, though not necessarily the direct users. Including older adults in our study, whether or not they delegate any banking tasks, allowed us to ask them how they would feel if banks used nudges to influence the behaviour of a current or future delegate. We recruited 21 participants; each was given an ID based on age group and status as close others. Of the 21 participants, 4 were close others who provided banking support to older adults (CO1-4), 6 were older adults who were not currently receiving any assistance with banking tasks (OA1-6), and 10 were older adults who also held the role of being a close other providing banking support to other older adults (OC1-10). We had one banking professional (BP1).

Though eligible, we did not have any older adult participants who currently delegate banking tasks to a close other. We expect that some older adults who delegate banking tasks may do so because they are not comfortable using online banking systems. Our study eligibility required familiarity with online banking to ensure participants would be comfortable working with our online prototype. We hoped some of the older adult participants we recruited would report that they occasionally delegated banking tasks to close others, but all of our older adult participants were very independent, quite tech savvy, and did not require any delegated assistance.

Participants took part in a single audio-recorded session lasting from 60 to 90 minutes, during which they were guided through 3 distinct phases (as illustrated in Figure 2): a pre-interaction interview, a think-aloud interaction with the high-fidelity prototype, and a post-interaction interview and debrief. Compensation for taking part was $30 delivered to the participant directly via e-transfer or donated to one of 5 pre-selected charities on their behalf. Participants could attend via Zoom or in-person on our campus.

The primary data sources for the study were the audio recordings of the sessions, manually transcribed by the first author. Log entries from the prototype were inserted into the transcripts to provide additional context. Responses to questions asked during the initial open-interview phase were analyzed independently of the topics explored during and after interaction with the prototype. A lightweight content analysis [21] was performed on responses to questions given to multiple participants, and the results were used to provide a broad understanding of our participant pool. Additionally, all of the responses from a single participant were used to build an understanding of each of their informal caregiving contexts that was in turn used to help characterize later responses.

We employed content analysis for the think-aloud interaction and semi-structured post-interaction interview data. Each participant's response to a question was quoted in a large table and, where applicable, was given a high-level tag characterizing the attitude of their response (such as “positive”, “skeptical”, or “disliked”). The responses were examined by both researchers for commonality of responses and for notable outliers, taking the participant's contexts and attitudes from the pre-interaction interview into account. From this we derived the prevailing attitude of our participant pool as well as the extreme perspectives. Our justification for this approach was to allow ourselves to provide an indication of the attitudes towards both proxy accounts and behavioural nudges that may be common within our participants’ demographics while also highlighting the varied perspectives that should be taken into account.

During initial discussion about their banking habits and/or support contexts, participants who assisted older adults were asked questions related to their methods of accessing the assets of the older adult they support, what kinds of tasks they performed, and how they managed the workload of providing banking assistance. When discussing access mechanisms, none of our participants reported using proxy accounts. Two of the 14 participants used the password set up by the older adult they help, while in five cases, they themselves set up the online banking credentials under the persona of the older adult (with their permission). In these cases, the banks were not aware that the close other was “impersonating” the older adult to set up the online banking access. OC7 opted to stay out from under the bank's eye by handling all communications through their aunt's email:

What I do is I am [aunt], and that's the easiest way for me. I have her email, I do everything. I don't want stuff about her coming through my email, so it's as if she's doing the banking herself. I mean that's what it seems like [to the bank], but it's me. - OC7

For some close others whose banks were aware of them as a close other, admonishment by the banks about password sharing made them feel uncomfortable, but it was still most convenient. CO2, who provided banking assistance to their mother-in-law who is in her late 90s and lives alone on the other side of the country, explains:

Yes, I have to lie to them. We went in… My wife and I got set up [to assist mother-in-law]... Anyway, the bank allows me to do financial things for them, but they don't let me use the internet. I can use the telephone, but they don't trust the internet. - CO2

This participant went on to explain the effort they expend switching mobile sim cards in order to impersonate their mother-in-law to complete her banking tasks online:

Except for now, they've got this two step verification using my cellphone number, but because I'm in the States for 3 months I got a new cell phone, we took out my old SIM card... so I'm gonna have to pull out the new SIM card, put in the old SIM card to get my verification and that's fine, I get it, it just slows things down a little. - CO2

This participant explained that careful attention was required to ensure legitimate transactions they made on behalf of their mother-in-law appeared legitimate:

... I wanna go out of my way to avoid the appearance that anybody might think I am [taking advantage of mother-in-law], which is why I put my wife's email on there, because it says [wife's last name, same as mother-in-law's last name]. You know, just little things that we're trying to do, that I'm maybe overly sensitive of, but am definitely sensitive of. - CO2

Seven of the 14 participants who act as close others provided banking help using joint accounts, an access mechanism that is more acceptable to, and supported by, the banks. Some of these were created under an invocation of a Power of Attorney, while others were set up in collaboration with the older adult. Four of these participants explained that joint accounts also make handling of estates easier in the event of the older adult's passing

Only one participant (CO1) indicated that they preferred to help the older adult they support by directing them through the process (in-person or over the phone) of how to do the transactions in the online banking system, without having any way of directly accessing the older adult's accounts. CO1 highlighted that they saw this as a way of fostering their mother's independence:

I want her to hold it, and I want her to be able to do it so she has the confidence later, so if she's not pushing the buttons herself she's not, uh, learning any of the technology and she's not equipped to work with it later. - CO1

Despite some hurdles, banking on behalf of an older adult was not considered too much of a burden. Most of the older adults being assisted by study participants had simplified financial situations and automated bill payments. Once an access mechanism was established, the close others reported the workload to be low: typically less than 1 hour of time spent on such tasks each week.

The majority of our older adult participants planned to rely on their adult children for banking task assistance in the event their abilities decline. The four that did not have children were unsure about how they would get help with financial tasks in the future. Most mentioned spouses or more distant relatives, but they did so with much less confidence in those hypothetical arrangements due to perceptions of ability (close others also being older adults) and/or levels of trust (for more distant relations). Only one participant said that they would consider delegating banking tasks to a professional caregiver in the future.

Both pre- and post-debrief, multiple participants shared personal stories and fears about financial abuse or misconduct perpetrated by people close to older adults, demonstrating that this is an issue our participants cared about. For example, OC6, an older adult who provides banking support to their parent, shared information about ongoing financial abuse of a different older adult relative:

You know I'm really worried about people who steal from older adults. That's a problem. I don't steal from mine, I never did, but I have a cousin who would help herself to her mother's money, and I think that's wrong, but obviously there's no regulation for that... I mean my cousin is not an evil person but she, you know, her mother [unknowingly] helped finance her house, which is not nice.... When her mother eventually passes away, that estate is going to be a whole lot less. - OC6

When asked about how they expected to be supported when they aged, one close other participant mentioned fears of financial misconduct happening to them:

This is more problematic, because I don't have children. And, although I have nieces and nephews, who knows where they will be at the time I need that assistance? So in fact, that is a bit of a concern for me, you know.... I'm very grateful that banks, if they see a big cheque go through, they'll call you, because I think elderly people, especially those with cognitive impairment, are horribly vulnerable to, um, mistakes by caregivers or worse! - OC4

Our banking professional said concerns around financial abuse is something that is salient in their job:

I've seen a situation where there was an older member who has, I believe, a couple of kids on their account, it wasn't just one. And there was a situation where funds were being transferred through online banking, right? One of them had access to their online banking just because they were joint. So, there was a situation like that where funds were being e-transferred and the member wasn't aware of that, the elderly member. And it caused a big problem... - BP1

Participants were unanimously positive about the idea of using proxy accounts, volunteering thoughts about this throughout the study. Participants noted potential legal or logistical aspects to consider, but felt that if banks offered them, proxy accounts would be helpful to close others and older adults. OC1 noted that the proxy settings provide agency and control to the older adult, giving them some control over the potential actions of the close other:

I do like the proxy options, and the permissions, you know the gradient of permissions that a person can authorize. So I think that that has a lot of potential. - OC1

One notable benefit identified was the legitimacy that proxy accounts could provide to close others. For example C02, who banks for his mother-in-law, explained:

I think the whole concept of a surrogate account is great, because I feel, guilty but annoyed that my bank...is making me do this nefarious route of pretending to be her. - CO2

The banking professional also saw value in this:

...our institution or other institutions could adapt something like this where they have a proxy system.... this is something that would work... there are a lot of situations where our elderly members don't want to use online banking but, you know, their caretakers do and it makes it easier for everyone, so I think that this is something that's important. - BP1

A popup box titled “Show Transactions”. The body contains the following message: “Are you sure you want to look at the details of Maggie's accounts? They will get a notification that you've elected to do so. In the bottom right corner is a red button labeled “Cancel” to the right of a blue button labeled “Show Transactions”.

In the following sections we summarize responses to the nudges that garnered the most reaction from participants during their interactions with the prototype. Not all participants saw every nudge; some nudges simply existed in the interface and may not have been encountered depending on how much the participant explored the interface, while other nudges were part of the study task interaction flows. A table showing participant nudge exposure can be found in Appendix B.

5.4.1 Hidden Transactions. OC9 had mentioned in the pre-interaction interview that they had concerns about seeing transactions they were not supposed to. As it happens, one of the first nudges that many participants encountered was the nudge that caused transactions to be hidden by default. When participants looked at the details of an account, the list of transactions was blurred out, and a button indicated that they could click it to show the transactions. When clicked, a confirmation box popped up, indicating that the older adult would be notified if they continued (see Figure 3).

Attitudes towards this nudge were lukewarm. While a few participants appreciated it conceptually, most found it unnecessary. Twelve of the 14 participants that encountered the nudge clicked “Show Transactions” and confirmed without any additional consideration. One participant initially cancelled, but returned and opted to show the transactions the second time. Only one participant made strong positive statements about this nudge:

Oh, I see, oh that's good... She will see that I'm snooping around in her account, that's good! I like that... - CO2

A notification panel with an orange background containing a message asking “If you are reimbursing yourself for a purchase, would you consider uploading a receipt?”. To the right of the message within the panel is a green button with white text labeled “Attach Receipt”.

5.4.2 Uploading Receipts. One nudge prompted participants to upload a receipt for a transfer reimbursing themselves for groceries (see Figure 4). This nudge was regarded positively, with 7 of 8 participants finding value in this. Participants saw this as a way of providing additional accountability and transparency for their banking tasks:

...then I'm leaving a paper trail, or at least an information trail, of what's happening. Which is a good idea in those situations. - OA4

Two screenshots combine to highlight an interaction flow, one atop of the other.The top screenshot is of a multi-select form titled “Today's Tasks” with checkboxes next to labels. Below the title is a prompt with the message: “Hello Sam! What tasks will you be performing today on behalf of Maggie?” The form has options for “Check Balance” (selected), “Transfer Funds”, “Pay Bills”, “Manage Payments” (selected), “Manage Services”, and an “Other” field with a text input field below. At the bottom of the form is a green “Select All” button and a blue “Submit” button. The bottoms screenshot is of a small pop up panel titled “Navigation Confirmation” that reads “When you selected the tasks you were going to perform today, you did not select balance checking. Would you like to continue or turn back?”. In the bottom-right of the panel are a red “Go Back” button and a blue “Continue” button.

CO4 was initially uncomfortable uploading receipts, worried about providing the banks with too much detailed spending information, but was okay uploading non-itemized receipts:

... because if we're using this bank and their credit card and all that, then they already have that information... and if they're just trying to confirm that, it seems alright. -C04

On the other hand, OC6 did not like this nudge at all as they were concerned by the possibility that close others would feel obligated to upload receipts, causing additional burden when performing tasks.

5.4.4 Transaction and Payee Notifications. Certain nudges indicated that a notification would be sent to the older adult, such as in response to a large transaction being completed or a new payee being added (see Figure 6). All participants who discussed these notifications were in favour of their inclusion as they felt that they improved transparency. For example, OC5 commented:

Adding a new payee and somebody would need to check? I think that's very useful because, especially if somebody else is helping you with your banking, you don't really want them to add more payees out of your account. So that's actually probably good that it would come back to the owner of the account. - OC5

The banking professional had a similar view:

I think adding that feature... some way to notify the member when something's going on with their account, I think that's really important. Because that member, the one in my situation, they had no clue until they came to the bank and I told her... So I think transparency with the member and what's going on with whoever's taking care of their accounts, that's extremely important... - BP1

Many participants had concerns about notification delivery given that not all older adults use online banking, cell phones, or other digital tools that would be the communication channel for such notifications. Participants suggested sending the notifications through the mail or to other proxies, if applicable, to ensure they are received and reviewed by a trusted and capable individual.

The large payment notification is an orange panel containing the message “This payment is large enough that a notification will be sent to your mother.”. The new payee notification is a red panel stating “Notification of this new payee will be sent to Maggie.”

5.4.5 Reminding of Consequences. Nudges that highlighted the prevalence of financial abuse of older adults (Figure 7) were somewhat well-regarded, with 6 of 11 participants who discussed these nudges expressing positive reactions. Participants appreciated the reminder of risks. This was the only passive informational nudge in our design that did not mention the older adult's name (as the "banking on behalf of" banner does). It was also the only nudge that participants thought might be ignored after repeated exposures.

A nudge that reminds the user of consequences. A red panel is titled "Did you know?" Below, text reads: "Did you know that financial misconduct affects 1 in 20 older adults, and impacts their ability to support themselves and maintain independence?" Below the message is a black button with a white label of "Get More Info" and an icon indicating clicking the button will open a link in a new tab.

In summary, reaction to nudges was mixed. Participants were generally in favour of nudges that provided transparency, tracking of information, and notifications to the older adult, but were less positive about nudges that were seen as ‘just extra clicks’ with no repercussions.

Participants were asked various questions about their thoughts on the use of behavioural nudges overall.

The feelings of trust our participants had in their informal care relationships defined their security postures and their attitudes towards financial misconduct. The only participants who discussed concerns about misconduct perpetrated by a close other were those who had personal or second-hand experiences with it. Participants that had not experienced or heard about misconduct affecting someone they knew were either dismissive of the possibility of such misconduct, or were significantly more concerned about fraud perpetrated by unknown parties. These results align with research that shows people are likely to underestimate risks of financial misconduct by family members or other people close to them [34].

Our older adult participants were comfortable with the possibility of sharing their credentials in order to delegate banking tasks, reinforcing previous findings [22]. This illustrates a potential gap in our participants’ knowledge about digital security: none of them spoke of any concerns about what might happen if they lost control of their passwords, likely because they had such high levels of trust in their potential future delegates. Given that password reuse across different systems is common [1, 8], the scope of access to their personal information granted by a single password may be much greater than our participants realize.

Given the trust most older adults have in their close others, our results suggest that security and privacy features in online banking interfaces should be incorporated in a way that makes them beneficial to both older adults and their delegates. For example, encouraging delegates to attach receipts for transactions not just as a method of validation or accountability, but also for easy reference. Similarly, a banking interface that encourages a logged in delegate to add a second delegate can both serve to enable oversight and share the financial task workload.

Our results uncover serious challenges around legitimacy of close others supporting older adults with financial tasks. There are clear benefits to the practice of sharing passwords for financial accounts that cause older adults and close others to leverage this practice. However, this causes additional friction because this access mechanism violates bank policies. Five of our close other participants had to impersonate the older adult they support in order to set up online banking accounts that they used to conduct banking on behalf of the older adult, which demonstrates a clear need for legitimate access mechanisms. Close others should not need to feel like criminals while supporting older adults with banking tasks. Close other participants spoke of these issues with bafflement and discomfort as they highlighted several ways that they would attempt to more accurately impersonate an older adult they supported in order to complete banking tasks and avoid the disapproving eye of the financial institutions. This negatively impacts close others, making them feel like they are acting inappropriately even while they are performing a valuable service and acting as ‘good carers’ [35].

Our prototype enabled participants to experience using proxy accounts to access an older adult's bank account and provide banking assistance, instead of logging in using the older adult's password. Our participants were in favour of this access mechanism because it would eliminate close others’ needs to impersonate an older adult for the sake of convenient access, enabling the close other to feel supported and acknowledged by the financial institution. The ability of the older adult to set limits on transactions was perceived to provide some level of control over the extent of damage that could be caused by a malicious actor (either an ill-intentioned close other, or an unknown criminal stealing the close other's credentials).

Proxy accounts would also allow the older adult to set up multiple close others as proxies. This was seen as positive by our participants and could help older adults remain independent and age in place [43]. This expands on results showing that older adult spouses examining bank transactions together helps create a shared understanding and added sense of security [26], and from Setterlund et al. who claim that the presence of ‘guardians’ can help deter financial misconduct by an informal caregiver arising from routine activities [44]. Additionally, some close other participants noted that having multiple delegates would allow sharing of the workload and flexibility in how and when they provide assistance, which could help to lessen the overall caregiving burden [3, 7, 40].

The implementation of proxy accounts would potentially require a non-trivial effort from financial institutions. This effort would be worthwhile, given the increasing use of online banking [10], the aging population [38], and the recognition that proxy accounts could be useful in informal care dynamics for older adults as suggested by Dunphy et al.[11], and supported by our results.

While adding more proxy accounts to support additional delegates does cause a slightly increased attack surface by virtue of creating more login credentials that can be lost or misappropriated, careful configuration of the associated accounts would greatly mitigate the impact of such events by ensuring that malicious actors can be easily detected through logging and notification controls and have limited ability to manipulate assets through restricted permissions.

Ultimately, proxy accounts would give older adults more control over their financial affairs as the current choice is very binary: do all the banking themselves, or give a close other unilateral access to their assets through joint accounts or shared passwords.

The behavioural nudges that we included allowed us to successfully gather a wide-ranging set of perspectives that effectively highlights some of the pros, cons, and considerations about their use within the context of delegated online banking. Given that the primary concerns of participants were related to fraud instead of misconduct, the strict limits enabled by proxy accounts were perceived as more likely to minimize financial damage in the case that an untrusted individual managed to gain access to the account. By comparison, the idea of mitigating misconduct via subtle suggestion was not considered as effective. The lukewarm opinions expressed by our participants in hypothesizing about the potential efficacy of nudges at deterring financial misconduct is unsurprising given the high level of trust our participants had in their relationships. Participants also raised concerns of over-burdening close others with more interactions and repeated exposure, potentially reducing the nudges’ effectiveness, which is a known issue with behavioural nudging [45].

Despite these concerns, the nudges were still seen to have merit. The nudges’ utility, from our participants’ perspectives, was in encouraging and supporting visibility and communication within the support dynamic and reminding the close others of their fiduciary responsibilities. For both older adults and close others alike, the ability to have greater awareness of the actions performed by a close other could increase the sense of security and trust between them. Nudges may also help support the close other's ‘good carer’ self-image by providing visible evidence of both diligence and accuracy. Using nudges to remind users that the information is being automatically shared, or to encourage them to volunteer more information, elicited positive responses from our participants for these reasons. Given that participants had no ethical concerns, including nudges in interfaces could have a positive impact on both the users’ experiences and their caregiving dynamics, so long as any burden they place on the close others is carefully balanced. Further investigation is needed to determine whether such nudges would actually be effective at deterring financial misconduct.