PaperlayerView exact revision on arXiv

Source-linked AI summary

Consumer Autonomy or Illusion? Rethinking Consumer Agency in the Age of Algorithms

Pegah Nokhiz, Aravinda Kanchana Ruwanpathirana

arXiv:2508.13440v1cs.CYcs.AI

TL;DR

The paper asks how systemic barriers and algorithmic manipulation constrain consumers’ ability to make independent spending and saving choices. It formally analyzes intertemporal consumption models with limited agency, finding adverse effects including early financial ruin and diminished utility, and considers value deliberation as an intervention.

  • Problem

    The paper examines whether consumers can make independent saving and spending choices when systemic barriers and algorithmic manipulation constrain consumption agency.

    Source: §1 · p. 1

  • Method

    The authors use formal analyses of intertemporal consumption models with limited agency to examine agency-related questions across several scenarios.

    Source: §6 · p. 15

  • Results

    The analyses illustrate early financial ruin and diminished consumer utility when systemic barriers, algorithmic manipulation, and obligatory consumption erode agency.

    Source: §6 · p. 15

  • Takeaways & Limitations

    The paper argues for recognizing consumer agency as a value and using value deliberation, including regulatory oversight and consumer education, as interventions.

    Source: §1 · p. 3, §6 · p. 15

    The optimization models do not yet incorporate the ethical dimensions of consumption, which the authors identify as an area for future research.

    Source: §5 · p. 13

Abstract

from arXiv · show

Consumer agency in the digital age is increasingly constrained by systemic barriers and algorithmic manipulation, raising concerns about the authenticity of consumption choices. Nowadays, financial decisions are shaped by external pressures like obligatory consumption, algorithmic persuasion, and unstable work schedules that erode financial autonomy. Obligatory consumption (like hidden fees) is intensified by digital ecosystems. Algorithmic tactics like personalized recommendations lead to impulsive purchases. Unstable work schedules also undermine financial planning. Thus, it is important to study how these factors impact consumption agency. To do so, we examine formal models grounded in discounted consumption with constraints that bound agency. We construct analytical scenarios in which consumers face obligatory payments, algorithm-influenced impulsive expenses, or unpredictable income due to temporal instability. Using this framework, we demonstrate that even rational, utility-maximizing agents can experience early financial ruin when agency is limited across structural, behavioral, or temporal dimensions and how diminished autonomy impacts long-term financial well-being. Our central argument is that consumer agency must be treated as a value (not a given) requiring active cultivation, especially in digital ecosystems. The connection between our formal modeling and this argument allows us to indicate that limitations on agency (whether structural, behavioral, or temporal) can be rigorously linked to measurable risks like financial instability. This connection is also a basis for normative claims about consumption as a value, by anchoring them in a formally grounded analysis of consumer behavior. As solutions, we study systemic interventions and consumer education to support value deliberation and informed choices. We formally demonstrate how these measures strengthen agency.

1 Introduction

The introduction frames consumption agency as shaped by digital persuasion, systemic constraints, and unstable work, rather than individual preference alone. It then motivates formal analysis of these constraints and interventions that cultivate agency through value deliberation.

Motivation

  • Digital persuasion and systemic barriers can narrow consumers’ choices, raising questions about whether spending and saving decisions reflect independent agency.The paper highlights filter bubbles, persuasive advertising, socioeconomic inequality, algorithmic bias, and automated scheduling as pressures on agency.Source: §1 · pp. 1–2

Constraints on agency

  • Obligatory expenses can accumulate through subscriptions, platform dependencies, and hidden fees, while algorithmic tactics can prompt impulsive purchases.Unpredictable schedules and earnings also complicate financial planning and informed saving and spending decisions.Source: §1 · pp. 1–2

Approach

  • The study uses discounted utility models to examine limited agency across obligatory consumption, algorithm-driven impulse purchases, and unstable work schedules.Its stated objectives are to analyze effects on financial well-being and stability and identify ways to mitigate them.Source: §1 · pp. 2–3

Contributions

  • The authors formally analyze agency constraints across three use cases: obligatory consumption, algorithm-driven impulse purchases, and dynamic work scheduling.They frame consumption as a value to be cultivated rather than a given.Source: §1 · p. 3
  • The paper argues that empowering consumers to deliberate about consumption values can help them move toward true agency and avoid financial distress.It presents consumer education and policy-driven regulatory measures as ways to encourage deliberation and informed choices.Source: §1 · pp. 2–3

2 Literature Review

The literature review situates the paper among work on consumption models, agency in algorithmic settings, and the consequences of financial and scheduling constraints. It also distinguishes structurally imposed expenses from persuasive tactics that prompt voluntary spending.

Consumption Models and Ruin

  • Discounted utility research models intertemporal consumption and saving, while ruin analysis and minimum-subsistence work address insolvency risk and basic consumption needs.The review also describes income fluctuation models, which account for income uncertainty and constrain consumption to current assets.Source: §2 · pp. 3–4

AI, Consumption, and Agency

  • Prior work examines how algorithms can enhance or restrict consumer agency, and links agency, fairness, transparency, and informed decision-making.The review notes that algorithmic choice constraints may disproportionately affect vulnerable populations.Source: §2 · p. 3

Obligatory Consumption

  • Obligatory consumption includes recurring costs and digital dependencies, as well as work, education, and health expenses embedded in daily life.Examples include platform fees, remote-work equipment, e-textbooks, and subscriptions for health services or devices.Source: §2 · pp. 3–4

Algorithmic Persuasion

  • Personalized recommendations, scarcity tactics, dark patterns, gamification, and social-media prompts are among the reviewed mechanisms for encouraging impulse purchases.The review also identifies mobile prompts and recurring microtransactions as drivers of repeated or spontaneous spending.Source: §2 · p. 4

Distinguishing Expense Types

  • Obligatory expenses arise from legal, social, or infrastructural pressures, whereas algorithmic persuasion prompts voluntary, often impulsive, spending.Some recurring payments appear in both categories, but the review distinguishes expenses required for access from strategically engineered choices.Source: §2 · p. 4

Work Schedule Instability

  • Research connects unstable schedules to income volatility and financial hardship, with scheduling software identified as a source of unpredictability for low-wage service workers.The review also covers worker participation and preference elicitation as approaches to fairer scheduling.Source: §2 · p. 4

3 Fallacy of Agency: Case Studies

The case studies use discounted-utility models to examine how structural, behavioral, and temporal constraints limit consumer agency. They show that rational consumers can face financial ruin or lower utility under these constraints.

3 Fallacy of Agency: Case Studies

  • The case studies represent agency constraints as structural commitments, manipulative digital design, and temporal uncertainty, spanning distinct but interconnected dimensions.The authors frame these cases as a multidimensional view of how autonomy can be eroded.Source: §3 · p. 5

3.1 Basic Model and System Properties

  • The baseline model maximizes discounted utility over consumption across time, subject to asset dynamics and a constraint that consumption cannot exceed available assets.Ruin is defined as reaching a time when assets are nonpositive.Source: §3.1 · pp. 5–6

3.2 Obligatory Consumption

  • With fixed obligatory consumption, a rational utility-maximizing consumer with concave utility can choose a finite path to ruin, and risk aversion does not prevent this outcome.The theorem also applies when consumption must meet a minimum subsistence level.Source: §3.2 · pp. 6–7
  • The obligatory-consumption result extends to a fixed minimum subsistence amount: with concave utility, a rational utility-maximizing consumer would go to ruin.The model treats this minimum as a lower bound on consumption at every step.Source: §3.2 · p. 7

3.3 Algorithmic Persuasion and Impulsive Consumption

  • When income and minimum subsistence are bounded and mean income is below mean subsistence, assets reach zero with high probability in finite time, while the probability of avoiding ruin declines exponentially.The discussion links low average income and high income variance to a high likelihood of early ruin.Source: §3.3 · pp. 8–9

3.4 Work Scheduling

  • For bounded-income instances, utility with k-step lookahead exceeds utility without lookahead, and the gap increases linearly with k [109, 110, 111].The paper relates advance awareness of schedules to improved financial management and greater autonomy.Source: §3.4 · pp. 9–10

4 Suggested Solutions: Intervening by Consumer Value Deliberation

The section treats consumer agency as a value to cultivate amid value capture and systemic barriers, and proposes deliberation and regulation to strengthen autonomy. It presents individual and societal interventions as ways to support more informed consumption choices.

4 Suggested Solutions: Intervening by Consumer Value Deliberation

  • The paper links value capture and systemic barriers—including mandatory consumption and faulty automated scheduling—to reduced consumer agency, which it argues must be nurtured.Externally imposed values may be adopted without critical reflection or personal adaptation.Source: §4 · p. 10

4.1 True Consumption Agency

  • True consumption agency is an aspirational ideal: it could allow indefinite ruin avoidance, but unavoidable essentials and legal obligations limit its attainability.The paper presents striving toward the ideal as a framework for practical autonomy-enhancing solutions.Source: §4.1 · p. 10
  • Consumption value deliberation is proposed as a guiding mindset for individual and societal interventions that evaluate competing needs and preferences.The approach aims to address unavoidable obligations while empowering more informed choices.Source: §4.1 · p. 10

4.2 Individual Consumer Value Deliberation

  • With R_t=1, β>1/2, and income covering fixed expenses, optimal utility-maximizing consumption avoids ruin by extending finite plans into better infinite-horizon sequences.The theorem assumes y_t≥b for all t, where b represents fixed non-negotiable expenses.Source: §4.2 · pp. 11–12
  • Individual deliberation interventions prioritize genuine needs before discretionary spending, with budgeting and self-discipline proposed to help consumers control finances and avoid early ruin.The model allocates funds first to essential and obligatory expenses, then to opt-in purchases that may be shaped by algorithms.Source: §4.2 · pp. 10–12

4.3 Ethical and Social Regulations

  • Societal proposals include ethical AI guidelines against manipulation and schedule regulations requiring advance notice and minimum hours.The paper cites the Schedules that Work Act of 2014 as an example of schedule regulation.Source: §4.3 · p. 12
  • Employees given advance notice of schedule changes experience improved financial well-being and greater autonomy in financial planning.The cited studies connect advance planning with more effective consumption and saving decisions.Source: §4.3 · p. 12

5 Limitations and Future Work

The framework deliberately uses a tractable rational-agent model, while leaving out behavioral, financial, normative, and social complexities that shape real consumption. Its proposed interventions are context-dependent and face implementation trade-offs, and the authors do not claim they will fully restore agency.

Limitations and Future Work

  • The framework omits debt and other liabilities, although credit can enable purchases while adding interest and long-term debt burdens.Consumers may rely on credit even without enough assets to cover expenses.Source: §5 · p. 13
  • The model excludes organizations and communities with multiple objectives, and future work could incorporate ethical commitments such as advertising effects and social comparison.The paper also identifies normative dimensions as an area for future optimization models.Source: §5 · p. 13

Intertemporal Consumption Models

  • The rational utility-maximizing framework isolates agency constraints, but does not represent real consumers as perfectly rational or model behavioral factors that may compound the risks.The authors describe the resulting risk estimate as tractable and suggest behavioral models could show stronger or more pervasive effects.Source: §5 · p. 13

Limitations and Future Work

  • Modeled risks may be amplified for lower-income people, gig workers, and marginalized communities, motivating targeted regulation, consumer protection, and context-aware financial education.The proposed education formats should account for limited time, attention, technological access, and erratic income.Source: §5 · pp. 13–14

Other Future Directions: Macro Socioeconomical Impacts

  • The study leaves macroeconomic shocks and demographic differences outside its scope, including inflation, emergencies, and variation by gender, disability, race, or health status.The authors note that these factors can alter income, expenses, consumption, and responses to policy.Source: §5 · p. 14

Limitations and Future Work

  • Automation has widened pre-existing limits on agency, and the proposed interventions aim for gradual improvement rather than full restoration in every context.The authors frame the interventions as creating friction in exploitative processes and empowering consumers within constraints.Source: §5 · pp. 14–15
  • Interventions are context-dependent proposals rather than finished solutions, with compliance costs, transparency pushback, and low education engagement among the implementation challenges.The authors suggest future simulations or field studies to assess policy outcomes and trade-offs.Source: §5 · pp. 14–15

6 Conclusion

The paper formally analyzes how limited consumption agency can produce early financial ruin and diminished utility, and presents education and policy interventions as ways to support value deliberation. In an ideal setting with unconstrained consumption variability, its theorem shows that a rational agent can avoid early ruin indefinitely.

6 Conclusion

  • Formal analyses illustrate early financial ruin and diminished utility when systemic barriers, algorithmic manipulation, and obligatory consumption erode agency.The conclusion frames consumer agency as a value and points to regulatory oversight and consumer education as potential intervention measures.Source: §6 · p. 15
  • The conclusion identifies value deliberation and consumer education, alongside regulatory oversight, as measures for addressing the adverse effects of diminished agency.The paper presents these as potential measures, not as a guarantee that agency will be restored in every context.Source: §6 · p. 15

A True Agency in an Ideal Setting

  • With known-mean nonnegative income and consumption allowed to vary subject only to 0 ≤ ct ≤ at, a rational agent under isoelastic utility with λ > 0 and λ ≠ 1 can consume indefinitely without early ruin.The proof considers shifting a positive amount of consumption from the terminal period into the next period, avoiding asset ruin at that point.Source: A True Agency in an Ideal Setting · p. 21

B Work Scheduling: Formal Theorem and the Corresponding Proofs (Restated from [109, 110, 111])

The section restates a work-scheduling result: limited lookahead can create a linear expected utility gap relative to k-step lookahead, even for randomized algorithms. The proof uses a distribution over bounded incomes to establish that gap.

B Work Scheduling: Formal Theorem and the Corresponding Proofs (Restated from [109, 110, 111])

  • The comparison concerns instances with incomes bounded in [0, Y], where both individuals know the input distribution and x is uniform on [0, 1].The proof assumes both individuals start with zero assets and scales Y to 1 without loss of generality.Source: B Work Scheduling: Formal Theorem and… · p. 22
  • Prior work reports that without lookahead, even complex strategies cannot achieve high total utility [110] [111].The proof frames the comparison around whether the agent can predict x.Source: B Work Scheduling: Formal Theorem and… · p. 22
  • For deterministic algorithms satisfying condition (4), the no-lookahead algorithm is Ω(k) worse than the lookahead algorithm.The proof establishes that condition (4) holds with probability at least 1/3 when x is uniform on (0, 1), using c = 1/12.Source: B Work Scheduling: Formal Theorem and… · p. 23
  • The expected utility gap is Ω(k): with probability 1/3, no-lookahead performs Ω(k) worse than k-step lookahead, and the lower bound extends to randomized algorithms.The argument applies Yao’s min-max theorem.Source: B Work Scheduling: Formal Theorem and… · p. 23

C Modeling Debt and Behavioral Anomalies

The section discusses debt and hyperbolic discounting as relevant extensions to the paper’s model, which are omitted from the current formalism for tractability and interpretability.

C Modeling Debt and Behavioral Anomalies

  • Debt can relax the consumption constraint but may amplify financial fragility under limited agency, including through overborrowing and repayment burdens.The discussion identifies credit cards, Buy Now Pay Later schemes, and payday loans as ways of smoothing consumption when income or assets are insufficient.Source: C Modeling Debt and Behavioral Anomalies · pp. 23–24
  • Hyperbolic discounting gives greater weight to immediate rewards than exponential discounting predicts and can make saving plans time-inconsistent.The passage links this pattern to myopic spending and greater susceptibility to manipulative design features.Source: C Modeling Debt and Behavioral Anomalies · p. 24
  • The current formalism excludes debt and non-standard temporal discounting to preserve tractability and interpretability.The authors identify both as promising extensions.Source: C Modeling Debt and Behavioral Anomalies · pp. 23–24

D Additional Experimental Analysis

The additional experiments examine algorithmic persuasion and ruin, differences across income and education groups, and the effects of work scheduling on financial utility.

D Additional Experimental Analysis

D.1 Experimental Setup

The experiments use value iteration and simulated asset and consumption trajectories to examine ruin under algorithmic persuasion and impulsive consumption. In that scenario, many agents experience ruin early, though some remain solvent through the planning horizon.

D.1 Experimental Setup

  • The model uses value iteration to compute optimal consumption paths, with the discount factor β set to 0.95.The discount factor is based on what is commonly practiced in general settings.Source: §D.1 · p. 24
  • The baseline setup uses initial assets of $141,140 and monthly income drawn from a log-normal distribution with mean $5,957.25 and standard deviation $378.74.The asset and income statistics are based on 2019 sources.Source: §D.1 · p. 24
  • The setup also specifies low- and high-income groups, with monthly income means of $1,899.33 and $8,869.92, respectively.Their respective standard deviations are $77 and $199.60.Source: §D.1 · p. 24
  • Each experiment simulates 50,000 agents over a 100-month horizon and defines ruin as the first month assets fall below zero.Ruin-time histograms show the number of agents going to ruin at each month.Source: §D.1 · p. 24

D.2 Algorithmic Persuasion, Impulsive Consumption and Ruin

  • The algorithmic-persuasion experiment uses the general population without income classes, minimum subsistence conditions, and mean monthly expenditure of $5,253.Its model and parameters otherwise follow the standard setup.Source: §D.2 · p. 24
  • A significant fraction of agents under minimum subsistence constraints experience ruin within the first 10 months, while some survive to the end of the planning horizon.The frequency of ruin declines steeply over time in the 50,000-agent simulation.Source: §D.2 · p. 25, Figure 2 · p. 25

D.3 Diverse Subgroups: Low Income vs. High Income

This experiment examines how income relates to financial ruin under impulsive consumption. Low-income agents are markedly more vulnerable, while higher income provides some capacity to delay ruin.

D.3 Diverse Subgroups: Low Income vs. High Income

  • The experiment examines income as a factor in financial ruin while considering varied demographic segments rather than assuming social homogeneity.Source: §D.3 · p. 25
  • The low-income group uses mean expenditure of $2,850 and β around 0.5; the high-income group uses $7,082.83 and β around 0.9.The expenditure figures follow 2019 statistics; the discount factors reflect prior research on the income classes.Source: §D.3 · p. 25
  • Nearly all 50,000 low-income agents go to ruin in the first time step, while many high-income agents also face early ruin but some survive longer.The high-income distribution is more spread out, indicating a modest degree of financial resilience.Source: §D.3 · p. 25

D.4 Diverse Subgroups: High School vs. College Graduates

This experiment examines how educational attainment, through differences in β, relates to financial stability under impulsive consumption. High-school graduates experience ruin sooner, while college graduates show a more spread-out distribution.

D.4 Diverse Subgroups: High School vs. College Graduates

  • High-school graduates are assigned β around 0.5, while college graduates are assigned β around 0.83; higher education is associated with higher income and financial flexibility [110].Source: §D.4 · p. 26
  • The experiment's setup is similar to §D.1, and Figure 4 presents ruin-time distributions for high-school and college graduates.Source: §D.4 · p. 26
  • Nearly all 50,000 high-school diploma holders go to ruin within the first 20 time steps, while college graduates show a wider spread with more surviving longer.The analysis describes the high-school group as financially fragile and the college-group spread as reflecting potential differences in consumption bounds.Source: Figure 4 · p. 26, §D.4 · p. 26
  • The results indicate that subgroup factors such as β can affect time to ruin.Source: §D.4 · p. 27

D.5 Work Schedule Instability Analysis

Prior work links schedule predictability to financial utility and well-being. Advance notice of schedule changes is associated with improved financial planning and greater autonomy for workers with variable schedules.

D.5 Work Schedule Instability Analysis

  • Workers with limited foresight into future schedules have lower financial utility than those with greater lookahead, and advance knowledge supports consumption and savings planning [110] [111].Higher earnings also allow workers to meet consumption needs more comfortably; disadvantaged workers with rapidly depreciating assets have greater need for advance schedule information.Source: §D.5 · p. 27
  • Workers given advance notice of schedule changes experience improved financial well-being and can manage consumption and savings more effectively.The cited examples include the Schedules that Work Act of 2014 (H.R. 5159) and San Francisco's Retail Workers Bill of Rights.Source: §D.5 · p. 27
Loading 2508.13440v1…