In August 2023, the Annual Review of Financial Economics published the article "Robo-Advice: Transforming Households into Rational Economic Agents", which discusses the distributional implications of robo-advice and its potential role in improving the effectiveness of economic policies. the role of provider incentives, and several questions that are still widely open to researchers in finance, economics, social psychology, and related fields. Robo-advisors use large and open data to provide consumers with a baseline of fully informed and reasonable expectations in all areas of household finance, including choices for consumption, savings, investment, and debt management. It also minimizes the monetary, cognitive, and psychological costs that families face in economic transactions. We reviewed recent research on the characteristics and impact of individual and overall economic outcomes through robo-counseling, which differs from traditional human advice. The Institute of Financial Technology of Chinese Min University (WeChat ID: ruc fintech) compiled the core part of the research. **annual review of financial economics
Author | francesco d’acunto and alberto g. rossi
Compile |Rune source.
Introduction
Every household in the world, regardless of their income, wealth, or complexity, needs to make complex financial decisions on a daily basis. First, households need to decide how much they want to spend on consumption or save to increase future consumption – they need to address an intertemporal consumption problem. Once they have made a choice to save on consumption, they need to allocate resources to durable and non-durable goods and choose how to finance their long-lasting consumption (e.g., income flows, credit card debt, mortgages, personal loans, student loans). Financing options vary depending on the multiple dimensions that need to be compared – term, term, interest rate, and amortization schedule. Households also need to manage their assets by deciding how to invest their savings to build wealth, maximize future consumption, and/or inheritance. For example, in addition to allocating resources among the investment assets in their financial portfolios, they must also choose how much to contribute to a retirement savings plan – a decision that has become very prominent around the world given the global distance away from the fixed-income pension system.
Ultimately, families, like business managers, need to make multiple financial decisions frequently without avoiding choices, and in this sense, avoiding decisions itself has an associated financial impact on the family (Madrian & Shea 2001). However, as opposed to business managers, families can't specialize in just one type of the many financial choices they need to make. Moreover, even though the vast majority of households have never been exposed to the most basic financial principles, such as diversification and compound interest, or have access to any financial education (Lusardi & Mitchell 2014), they need to make these choices.
Given these premises, it's not surprising that most households make decisions that differ from those of the standard economic model. The Standard Model assumes that there are utility-maximizing **, that they have access to all the information they need to make financial decisions, use it to form rational expectations, and have standard economic preferences. A large and growing body of financial and economics literature labels deviations from the standard model of household financial choices as household financial puzzles (Zinman 2015).
In addition to human financial advice and nudges, machine counseling is the third and most recent solution to the family's financial conundrum (D'Acunto & Rossi 2021). At first, machine counseling may seem like nothing more than traditional human financial advice and is automatically delivered to decision-makers through a personal device. Even though this is the only relevant feature of robo-counseling, it will make it more scalable than human counseling by reducing costs and allowing more families to access counseling. Therefore, this property will improve the proposed distributional consequences by adjusting the choices of vulnerable families to those of complex families (Philippon 2019). However, robo-consultations are not just traditional recommendations that are automatically generated and delivered through personal devices. A key feature of robo-counseling is that it reduces the difference between families and the characteristics of standard neoclassical rational decision-makers, thus allowing families, including the least sophisticated, to make choices that are closer to what standard utility maximization will make.
Machine consulting transforms biased decision-makers into standard neoclassical decision-makers through three unique characteristics. First of all, machine consulting uses big and open data to make families fully aware of their own characteristics and past behaviorFamilies (usually human advisors) would otherwise not have access due to their limited ability to recall past actions and collect and complex large data sets (d'acunto &rossi 2022). The big data captured by machine consulting is a variety of transactional areas that arise when households make economic decisions, such as the purchase of durable and non-durable goods, investment options, and financing options. Machine consulting has access to this wealth of information because since the digital revolution, electronic transactions can be stored and this data is open; That is, families can share them by linking their trading accounts to the robot consulting app.
SecondRobots consult the beliefs that manage the family, bringing them closer to the rational expectations assumed in the Standard Model。The economic literature documents the systematic deviation of some families' beliefs from those of standard economic agents. The size of these departures correlates with demographics and changes throughout the business cycle. Robo-advisors provide families with confidence in their future financial situation and the general macroeconomic situation that is closer to the norm**.
Third, robo-advisory reduces the transaction cost of executing an economic choice, which includes not only monetary costs, but also the cognitive and psychological costs that families face when trying to make economic decisions.
What to study
Reduce information friction
The standard neoclassical economic agents are considered to be well aware of their characteristics, the characteristics of other subjects, and macroeconomic conditions. With the exception of some special cases, the assumptions of this standard model of intertemporal consumption optimization are grossly violated in the data, such as highly complex institutional investors in financial markets, who are actively involved in expensive access to information (Gargano, Rossi & Wermers 2017).
Most households have little understanding of macroeconomic conditions and often even ignore definitions of underlying economic aggregates, such as inflation rates or interest rates (Binder 2017). Households also show that, on average, there is an erroneous assessment of the economic situation of other economic agents, which is mainly due to the lack of complete information on the income flows, expenditures and debt positions of others and the striking and significant transactions that households interpret as representing the average transactions performed by others (Charles, Hurst & Roussanov 2009). Ultimately, and perhaps most surprisingly, households are often unable to recall or process information about their own income flows, consumption, and debt transactions, even though all of this information was generated by them, so they don't need to collect public or private signals from others. For example, households do not save information on all their transactions, but usually only follow them temporarily and for short periods of time (D'Acunto, Rossi & Weber 2019). Households also tend to have biased recollections of their consumption, due to their tendency to recall that they paid **less than they actually faced** (d'acunto & weber 2022).
The first unique feature of robo-counseling is that it pushes families closer to fully understanding themselves, others, and overall economic outcomes. Robo-advisors can achieve this by collecting and elaborating observations of large and open transaction-level data – household characteristics and past choices, characteristics of other decision-makers and past choices, macroeconomic conditions and information on economic decisions that may be relevant in other aggregate variables.
Reduce distorted beliefs
Standard neoclassical decision-makers are also thought to form expectations rationally; On average, their beliefs about their future own outcomes and overall outcomes are consistent with post-event realizations. Contrary to this benchmark, households tend to exhibit multiple biases about beliefs about themselves, others, and macroeconomic outcomes.
For example, households tend to misestimate their own survival probabilities in terms of outcomes, which leads them to make suboptimal choices about retirement savings and insurance policies (Heimer, Myrseth & Schoeenle 2019). In addition, many households systematically distort beliefs about future income mobility and income growth: they extrapolate recent income growth rather than systematically reduce and overestimate future income (Cutler, Poterba & Summers 1990; de long et al. 1990; greenwood & shleifer 2014; barberis et al. 2015; gennaioli, ma & shleifer 2016; barberis et al. 2018; carroll et al. 2020;d’acunto, weber & yin 2022).At the same time, more than a third of the representative population overestimated the likelihood of a negative shock and the need to save precautionarily and to spend insufficient relative to their ability to precautionary savings motivation declined when households received a credit line for insurance (d'acunto et al.). 2020）。
Households also exhibit biased beliefs in forming expectations of others – they tend to overestimate the consumption of others and underestimate their savings rates, as noted above (d'acunto, rossi &weber 2019) – as well as future macroeconomic variables. These system distortions correlate with demographic characteristics such as socioeconomic status (Das, Kuhnen, and Nagel 2020), cognitive ability (D'Acunto et al. 2019), age, and cohort (Malmendier and Nagel 2016). Stango & Zinman (2023) provides a categorization of multiple biases in family expectations and beliefs and documents commonalities and patterns of correlation between them, suggesting that many biases in beliefs may have something in common**. Ultimately, faith seems to be more biased against disadvantaged households, which may contribute to perpetuating economic inequality.
Reduce transaction costs
In order to optimize future consumption through wealth accumulation, the standard portfolio optimization model**, as long as the cost of participation is minimal, almost every household should participate in the market by investing a small portion of their savings. Conversely, the extensive literature in economics and finance has raised the cost of participation as a potential explanation for one of the most common findings in the global household finance market participation conundrum, i.e., a significant proportion of households do not participate in the market and many invest in a small fraction of the standard model of wealth ratio (Moskowitz &Vissing-J Rgensen 2002)). Costs can also be associated with consumption-saving transactions (Abel, Eberly & Panageas 2007), not just in the form of monetary costs.
Monetary transaction costs, such as spending transactions or **transactional**, have been declining over the past few decades, but may still remain a barrier to participation, especially for the most vulnerable households (Reher & Sokolinski 2021) Based on this argument, Philippon (2019) proposes a model in which a key feature of robo-advisors and other fintech platforms is that they reduce monetary and financial transaction costs. Philipp (2019) argues that fintech platforms, including robo-advisors, should increase their participation in the market and reduce wealth inequality within and between countries. Reher & Sokolinski (2021) provides empirical evidence to support this thesis. They studied an American robo-financial investment advisor who, during the sample they observed, cut the minimum amount required for access, which amounts to a reduction in the cost of participation, as households must maintain the minimum value of their accounts rather than otherwise using those resources. Reher & Sokolinski (2021) showed that after this intervention, the share of households with ** increased and their portfolio performance improved. Since this type of intervention does not affect wealthy households that invested more than the minimum before the change, the benefits are entirely concentrated in the middle class and poorer households.
The costs faced by households in executing economic transactions can also be cognitive and psychological. Even if families are fully aware of the economic problems they face and are able to obtain near-optimal solutions, conceptualizing these problems and identifying their solutions requires the use of cognitive skills and time, both of which are scarce resources (Abel, Eberly & Panageas 2013). Families often lack the cognitive skills and economic or digital literacy needed to understand and solve economic problems, which increases the cognitive cost of making decisions (Lusardi & Mitchell 2014). Cognitive abilities can also ** family choices, economic variables, and criteria for making economic decisions (d'acunto et al.). 2022）。In addition, a large body of work in behavioral economics has demonstrated that the subtle complexity of choice is not fully grasped by consumers, leading to choice avoidance (Iyengar & Lepper 2000) and inertial decision-making (O'Donoghue & Rabin 1999, Madrian & Shea 2001).
By streamlining the decision-making process and making the potential future benefits of active choice clearer and more tangible, robo-counseling can reduce cognitive costs, psychological costs, and inertia in decision-making. For example, Bianchi & Briere (2021) studied a French robo-advisor for employee savings accounts that provides a significant target for portfolio allocation that investors can observe at any given time. They found that bot suggestions increased people's attention to their own accounts. In turn, a higher focus leads to higher risk-adjusted returns and lower volatility (Gargano & Rossi 2018). In addition, the robo-advisor's automated, graphically calculated objective descriptions reduce the cognitive cost of assessing whether and to what extent portfolio allocations deviate from the optimal benchmark. The cognitive cost of articulating the optimization problem and mapping the solution to an investment action decreases investor inertia in financial decisions and leads to more frequent portfolio rebalancing, resulting in higher risk-adjusted returns relative to unassisted portfolios. Bianchi & Briere (2021) found that the benefits of robo-advisors were higher for investors with smaller portfolios and lower pre-adoption** participation, which may represent less affluent and lower-income employees. Thus, reducing the cognitive cost through robo-counseling has the same potential distributional consequences as reducing the cost of monetary transactions – it encourages the participation and wealth accumulation of more vulnerable households.
Controversy
Whether these families understood their biases after taking the robot's advice and how to correct them is an open question and has important policy implications. If there is a learning Xi, access to robo-counseling will not need to be repeated over time, and the role of robo-advisors transforming families into standard economic decision-makers will be quite effective. If robo-consulting creates a spillover effect of learning Xi bias and how to correct it across economic decision-making domains, its effectiveness will increase. In this sense, robo-consulting can be interpreted as a means of providing financial literacy, which is inexpensive, as opposed to traditional financial literacy projects: it does not require the recruitment and payment of teachers, the time invested by users, or the cognitive cost of trying to apply abstract principles to real-world choices. Instead, users learn Xi by imitation that they will need to make economic choices (often flawed) economic choices.
The drivers for the adoption of robo-consulting are not well understood. The purpose of the study design on the effect of robotic counseling on selection was to obtain quasi-exogenous heterogeneity in families with similar observations. The marginal benefit of choosing whether or not to adopt is important to evaluate the overall effectiveness of the robo-consult. Most existing randomized studies based on robo-counseling** have found that less sophisticated families benefit the most, as their independent choices are often more deviant from the standard neoclassical**. However, due to lack of awareness, distrust of unknown applications, and other reasons, less mature households are less likely and slower to adopt new technologies (Foster & Rosenzweig 2010).If the same is true for the adoption of robo-advisory, then the heterogeneous effects of performance based on adoption conditions will be negated.
This paper focuses on the impact of robo-consulting on human decision-making from the perspective of users, but the extent to which the incentive mechanism of boto-consulting service providers may affect the characteristics and consequences of robo-consulting is also an important open question. In some applications, financial institutions face trade-offs that third-party providers don't. To date, despite the importance of these differences, there has been little research on whether and how the nature and incentives of robo-advisory providers shape their design and effectiveness. In addition, the influence of the market that controls the proliferation and effectiveness of fintech robo-advisory applications, i.e., that shareholders of third-party providers may have an incentive to exit the investment by giving decentralized ownership to financial institutions rather than the market (Wang 2018), is an interesting and open topic. Conclusions and Recommendations
Robotic counseling can provide a well-informed and rational benchmark of expectations by reducing costs and human intervention to transform families who are biased in decision-making and economic issues. In this article, we highlight that research on robo-consulting is still in its infancy, and we present some open-ended questions to shape this agenda in order to better understand whether, how, and why robo-consulting can achieve the above goals. We also believe that robo-counseling can reduce or widen wealth inequality across a range of demographic dimensions. Understanding this trade-off is an important area for future investigations, and the policy implications of all the outstanding issues we point out in this article. To make researchers aware that robo-counseling is not just a normal way of saying that fintech applications allocate households' portfolios, but a potentially transformative force for empirical and theoretical research on individual economic decision-making in most dimensions. The following is a screenshot of the article
Get the full report
Please reply to "Machine Inquiry" in the background
Get the ** link
endEdit the rune source.
Reviewed by Li Jinxuan.
Editor-in-charge: Yang Shiqi.
Extended reading] JFE |Machine learning Xi with ** Manager common skills jfe |Machine science Xi recognize the emotional colors of financial vocabulary RFS |Use machine Xi and big data to choose from optionsPredictable inequality? The impact of machine Xi on credit markets.