Optimal Provision-After-Wait in Healthcare

Mark Braverman, Jing Chen, Sampath KannanMath. Oper. Res 2016 · DOI

TL;DR

This paper tackles the complex problem of allocating scarce healthcare resources by using waiting times as a primary rationing mechanism, aiming to maximize social welfare under a fixed budget. The authors show that finding optimal waiting times is computationally hard, but propose efficient approximation algorithms and demonstrate that waiting times can emerge endogenously from patient behavior. They also highlight that in certain scenarios, a simple randomized assignment can achieve optimal social welfare, outperforming equilibrium-based waiting time mechanisms.

What Problem Does This Paper Address?

Healthcare systems worldwide grapple with allocating limited resources, especially for non-urgent treatments. While monetary pricing is common in many markets, healthcare often relies on other mechanisms due to ethical considerations and insurance structures. This paper focuses on "waiting times" as a key rationing tool, where patients choose hospitals based on their preferences and the associated waiting periods. The core challenge is to design a system where a central authority (the "payer" or government) can set optimal waiting times to maximize overall patient welfare, while staying within a fixed budget. This involves understanding how patient choices lead to an equilibrium, how to compute the best equilibrium, and whether alternative mechanisms like randomized assignment can achieve better outcomes.

Methods

The paper employs a combination of mechanism design, computational complexity theory, and dynamic systems modeling.

  • Mechanism Design: The authors model the problem as a mechanism design challenge where the payer sets waiting times to influence patient behavior and achieve a budget-constrained, welfare-maximizing outcome. They analyze equilibrium assignments and explore alternative mechanisms like randomized assignment.
  • Computational Complexity: They prove that finding the optimal equilibrium assignment is NP-hard, motivating the development of approximation algorithms.
  • Approximation Algorithms: An algorithm is proposed that can find an equilibrium assignment with a slight budget deficit (up to (1+ε)B) that achieves near-optimal social welfare, running in polynomial time with respect to the number of patients. This algorithm leverages connections to unit-demand auctions.
  • Dynamic Systems: The paper models the emergence of waiting times as a continuous-time dynamic process, showing that under certain conditions, waiting times will converge to their optimal equilibrium values endogenously.
  • Game Theory: Concepts like equilibrium assignments and bidder-optimal matchings (from auction theory) are used to analyze patient behavior and hospital interactions.

Datasets

Not discussed in the paper. The paper focuses on a theoretical model with parameters representing patient preferences and hospital costs, rather than empirical data.

Key Insights and Ideas

  • Waiting Times as a "Welfare Burner": Unlike monetary transfers, waiting time itself doesn't benefit anyone and represents a direct loss of social welfare. This makes minimizing waiting time a crucial objective.
  • NP-Hardness and Approximation: The core problem of finding optimal waiting times is computationally intractable in general. However, a small relaxation in the budget constraint allows for efficient approximation algorithms.
  • Endogenous Emergence of Waiting Times: The paper demonstrates that the government doesn't necessarily need to explicitly enforce waiting times. By controlling the "quotas" (or service rates) of hospitals, waiting times can naturally adjust based on patient demand and converge to the desired equilibrium.
  • Randomized Assignment as an Optimal Strategy: Surprisingly, in certain scenarios (specifically, when patient preferences are concave), a simple randomized assignment of patients to hospitals (without any waiting) can achieve the maximum possible social welfare, outperforming equilibrium-based waiting time mechanisms. This suggests that restricting patient choice can sometimes lead to better overall outcomes.
  • Information Disclosure Paradox: Revealing quality information about hospitals could paradoxically lead to worse outcomes if it concentrates demand on better hospitals, increasing waiting times and burning more social welfare, especially if patients were previously ignorant and effectively using a randomized assignment.

Key Results

  • Theorem 1: Finding optimal equilibrium waiting times and assignments is NP-hard.
  • Theorem 2: An algorithm can find an equilibrium assignment with an ε-deficit budget that achieves social welfare at least as high as the optimal budget-constrained solution, running in time polynomial in the number of patients and hospitals.
  • Theorems 3, 4, and 5: Under specific conditions (independent patients, generic preferences), there's a unique optimal equilibrium, and waiting times will endogenously emerge and converge to these optimal values in finite time.
  • Theorem 6: For concave patient valuation functions, randomized assignment achieves optimal social welfare, outperforming any lottery (including equilibrium assignments).

Limitations and Future Work

  • Budget Constraint Relaxation: The approximation algorithm requires relaxing the budget constraint. Finding an algorithm that strictly adheres to the budget while still approximating welfare is an open problem.
  • Polynomial Time in k: It's unknown if an approximation algorithm exists that is polynomial in the number of hospitals (k).
  • Generalizing Randomized Assignment Optimality: The condition for randomized assignment optimality (concave preferences) is specific. Finding broader conditions for optimality for three or more hospitals is an open question.
  • Information Disclosure Impact: The paper raises the intriguing point that information disclosure might not always improve welfare in waiting-time-rationed systems. Further study is needed on how to optimally structure information release.
  • Mechanism Design for Truthful Revelation: Designing mechanisms that elicit truthful patient valuations without the payer needing to know them beforehand is a key future direction.
  • Quality Information and Patient Incentives: Investigating the impact of quality reporting on patient incentives and overall welfare in healthcare allocation systems is highlighted as an important area.

Summary updated

BibTeX
@article{braverman2016optimal,
  author = {Mark Braverman and Jing Chen and Sampath Kannan},
  title = {{Optimal Provision-After-Wait in Healthcare}},
  journal = {Math. Oper. Res.},
  year = {2016},
  doi = {10.1287/moor.2015.0731},
}
Figure