Paper: arXiv 2604.27186

Authors: Nilavra Pathak, Smriti Shyamal, Prasant Mhasker, Christopher Swartz

Abstract

We study finite-horizon budget allocation as a closed-loop economic control problem and evaluate receding-horizon Model Predictive Control (MPC) relative to reactive budgeting policies. Budgets are allocated periodically under execution noise and operational constraints, while return efficiency may evolve over time. Using a controlled simulation framework motivated by digital marketing, we compare reactive pacing to MPC across environments with increasing degrees of non-stationarity. Our results show that non-stationarity alone does not justify predictive control. When return dynamics are stationary or evolve through unpredictable stochastic drift, MPC offers no systematic advantage over reactive baselines. By contrast, when return efficiency exhibits predictable structure over the planning horizon, that is captured through an underlying model, MPC consistently outperforms reactive budgeting by exploiting intertemporal trade-offs.

Complexity vs Empirical Score

  • Math Complexity: 6.5/10
  • Empirical Rigor: 7.5/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper combines formal stochastic control theory and finite-horizon optimization (MPC) with a rigorous simulation-based framework that includes realistic noise terms, non-stationary market regimes, and specific operational constraints.

Research Flowchart

  flowchart TD
    A[<b>Goal:</b> Optimize Budget Allocation under Non-Stationarity] --> B[<b>Methodology:</b> MPC vs. Reactive Pacing]
    B --> C[<b>Inputs:</b> Execution Noise & Operational Constraints]
    C --> D[<b>Processes:</b> Simulate Digital Marketing Environments]
    D --> E{<b>Environment Type?</b>}
    E -- Stationary/Stochastic Drift --> F[<b>Finding:</b> No Advantage for MPC]
    E -- Predictable Structure --> G[<b>Finding:</b> MPC Outperforms via Intertemporal Trade-offs]

Cited References (112)

  1. Learning to bid with AuctionGym (2022) — link
  2. Reinforcement learning: An introduction (2018)
  3. Openai gym (2016) — link
  4. The Impact of Control Technology (2014)
  5. The Impact of Control Technology (2011)
  6. Learning-based model predictive control: Toward safe learning in control (2020)
  7. Expedia Financial Statements
  8. Learning to bid without knowing your value (2018)
and 104 more
  1. Online learning in repeated auctions (2016)
  2. Real-time bidding algorithms for performance-based display ad allocation (2011)
  3. Budget constrained bidding in keyword auctions and online knapsack problems (2008)
  4. Advertising Incrementality Measurement using Controlled Geo-Experiments: The Universal App Campaign Case Study. (2020)
  5. Bidding and pricing in budget and ROI constrained markets (2021) — link
  6. Sponsored search: an overview of the concept, history, and technology (2008)
  7. Principles of economics: unabridged eighth edition (2009)
  8. Dynamic pricing with constant demand elasticity (2008)
  9. Dynamic pricing in a labor market: Surge pricing and flexible work on the Uber platform. (2016)
  10. Time-varying price elasticity of demand estimation for demand-side smart dynamic pricing (2022)
  11. Dynamic pricing in retail gasoline markets (1993)
  12. Mostly harmless econometrics: An empiricist’s companion (2009)
  13. An efficient deep distribution network for bid shading in first-price auctions (2021)
  14. Perspectives on system identification (2010)
  15. Dynamic mode decomposition: data-driven modeling of complex systems (2016)
  16. Parameter estimation in stochastic grey-box models (2004)
  17. Nonlinear black-box modeling in system identification: a unified overview (1995)
  18. Sparse identification of nonlinear dynamics with control (SINDYc) (2016)
  19. Multi-Task System Identification of Similar Linear Time-Invariant Dynamical Systems (2023) — link
  20. PID control (2005)
  21. Model predictive control (2013)
  22. A bayesian data analytics approach to buildings’ thermal parameter estimation (2019)
  23. Online ad auctions (2009)
  24. Multi-unit auctions with budget limits (2012)
  25. Optimal delivery of sponsored search advertisements subject to budget constraints (2007)
  26. Counterfactual Reasoning and Learning Systems: The Example of Computational Advertising. (2013)
  27. Conjugate Bayesian analysis of the Gaussian distribution (2007)
  28. Digital marketing: A framework, review and research agenda (2017)
  29. A broad overview of interactive digital marketing: A bibliometric network analysis (2021)
  30. An analytical model for comparing the profitability of competing online marketing channels: Search engine marketing versus e-commerce marketplace (2021)
  31. Digital marketing communication in global marketplaces: A review of extant research, future directions, and potential approaches (2022)
  32. The role of search engine optimization in search marketing (2013)
  33. Search engine marketing: Transforming search engines into hotel distribution channels (2011)
  34. Return on quality improvements in search engine marketing (2012)
  35. The impact of positioning on click-through-rates in hotel metasearch engines (2022)
  36. Vertical integration of platforms and product prominence (2022)
  37. Setting the future of digital and social media marketing research: Perspectives and research propositions (2021)
  38. Data-driven Budget Allocation Optimization for Digital Marketing (2024)
  39. Forecasting at scale (2018)
  40. Generalized additive models (2017)
  41. Forecasting gas usage for big buildings using generalized additive models and deep learning (2018)
  42. Multivariate bayesian structural time series model (2018)
  43. Forecasting the yield curve in a data-rich environment: A no-arbitrage factor-augmented VAR approach (2008)
  44. Forecasting the yield curve in a data-rich environment using the factor-augmented Nelson–Siegel model (2013)
  45. Forecasting: principles and practice (2018)
  46. Survey of multi-objective optimization methods for engineering (2004)
  47. Leveraging Elastic Demand for Forecasting (2019)
  48. Market Response Models: Econometric and Time Series Analysis (2003)
  49. Marketing Budget Allocation Using Lagrangian Relaxation (2009) — link
  50. Dynamic budget allocation in social media advertising campaigns (2022) — link
  51. A robust optimization approach to budget optimization in online marketing campaigns (2025)
  52. Marketing budget allocation with offline constrained deep reinforcement learning (2023)
  53. Model-based constrained MDP for budget allocation in sequential incentive marketing (2019)
  54. An end-to-end framework for marketing effectiveness optimization under budget constraint (2023) — link
  55. Lagrangian decomposition algorithm for allocating marketing channels (2015)
  56. Marketing budget allocation with offline constrained deep reinforcement learning (2023)
  57. Adaptive Budget Optimization for Multichannel Advertising Using Combinatorial Bandits (2025) — link
  58. Budget optimization for online campaigns with positive carryover effects (2012)
  59. Optimal Advertising Policy Under Dynamic Conditions (1962)
  60. Optimal Control of the Vidale–Wolfe Advertising Model (1973)
  61. Optimal advertising for a generalized Vidale–Wolfe response model (2021)
  62. An Optimal Control in the Vidale–Wolfe Marketing Model (2018)
  63. Model predictive control: past, present and future (1999)
  64. Model Predictive Control (2010) — link
  65. Multi-stage stochastic optimization applied to energy planning (1991)
  66. Nonlinear Model Predictive Control: Theory and Algorithms (2011)
  67. Smoothing and Differentiation of Data by Simplified Least Squares Procedures (1964) — link
  68. A Media Planning Calculus (1969) — link
  69. How Advertising Affects Sales: Meta-Analysis of Econometric Results (1984) — link
  70. Market Response Models: Econometric and Time Series Analysis (1990) — link
  71. The Persistence of Marketing Effects on Sales (1995) — link
  72. A Unified Framework for Advertising Bidding and Budget Allocation (2019)
  73. Real-Time Bidding by Reinforcement Learning in Display Advertising (2017)
  74. Budget-Constrained Marketing Optimization via Automated Bidding (2023)
  75. Practical Lessons from Predicting Clicks on Ads at Facebook (2014)
  76. Model Predictive Control: Theory, Computation, and Design (2017)
  77. Advertising and consumer awareness of new, differentiated products (2019)
  78. Aggregate advertising expenditure in the US economy: Measurement and growth issues in the digital era (2021)
  79. Computable Markov-perfect industry dynamics (2010)
  80. Deep learning (2015)
  81. Sparks of artificial general intelligence: Early experiments with gpt-4 (2023) — link
  82. Deep learning using linear support vector machines (2013) — link
  83. Budget management strategies in repeated auctions (2017)
  84. Budget optimization in search-based advertising auctions (2007)
  85. Real-time bidding algorithms for performance-based display ad allocation (2011) — link
  86. Improved pacing for advertising campaigns (2013)
  87. Real-time bid optimization with smooth budget delivery in online advertising (2013)
  88. Budget Pacing for Targeted Online Advertisements at LinkedIn (2014)
  89. Real time bidding by reinforcement learning in display advertising (2017) — link
  90. Optimizing budget constrained spend in search advertising (2013)
  91. Real time bid optimization with smooth budget delivery in online advertising (2013)
  92. A field guide for pacing budget and ros constraints (2023) — link
  93. Robust Process Control (1999)
  94. Safe model-based reinforcement learning with stability guarantees (2017)
  95. Gradient-based hyperparameter optimization through reversible learning (2015)
  96. Dynamic pricing and learning with uncertain demand (2016)
  97. A system perspective on online advertising: A review and future directions (2022)
  98. Optimal control: linear quadratic methods (2011)
  99. Expedia Group, Inc. Form 10-K Annual Report (2024)
  100. Expedia Group, Inc. Form 10-Q Quarterly Reports (2024)
  101. Online optimal control with linear dynamics and predictions: Algorithms and regret analysis (2019)
  102. The power of predictions in online control (2020)
  103. A flexible growth function for empirical use (1959)
  104. Optimal spend rate estimation and pacing for ad campaigns with budgets (2022) — link

0 of 112 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.