An analytical model of Disequilibrium and decentralized productive Exploration

Abstract: This paper studies the economic role of persistent dispersion in allocations across agents. We develop a tractable model in which firms allocate resources under imperfect information and behavioral updating, generating sustained heterogeneity in beliefs and actions. While dispersion induces static misallocation, it also fosters decentralized experimentation, allowing the economy to explore a broader set of productive opportunities. We show that the economy converges to a stationary equilibrium with strictly positive dispersion and that, under plausible conditions, such disequilibrium can dominate the perfectly coordinated benchmark. The model provides a novel interpretation of observed dispersion in productivity and returns as reflecting both inefficiency and productive exploration. It also yields testable predictions linking dispersion to growth and innovation dynamics.

• The paper shows that persistent dispersion induced by behavioral noise can enhance aggregate welfare through decentralized exploration.
• It employs a Gaussian AR(1) framework where agents update noisy private signals, resulting in a strictly positive steady-state variance of beliefs.
• The model uncovers a trade-off where the benefits of experimentation and innovation from misallocation can outweigh the static costs of inefficiency.

Analytical Model of Disequilibrium and Decentralized Productive Exploration

Theoretical Framework and Model Structure

The paper develops an analytically tractable model capturing the endogenous emergence and persistence of cross-sectional dispersion in firm-level allocations under imperfect information and behavioral updating. The main departure from canonical equilibrium models is the explicit incorporation of a trade-off between static efficiency losses due to allocation dispersion (misallocation) and the dynamic benefits of decentralized productive exploration.

Agents observe imperfect, noisy signals about an unobservable—yet time-varying—fundamental ($\theta_t$), interpreted as the location of productive opportunities (e.g., sectoral profitability, technological direction). Each agent updates her belief over time with a behavioral rule incorporating both signal responsiveness ($\alpha$) and persistent idiosyncratic noise ($\sigma_\eta$), which prevents convergence to common knowledge even in the presence of abundant information. As a result, in the stationary regime, the cross-sectional belief and action dispersion ($v^*$) remains strictly positive as long as behavioral noise is nonzero, even in the limit of perfect information ($\sigma_\nu \to 0$). Aggregate output is defined as the sum of realized payoffs net of misallocation costs (variance of allocations from the true fundamental) plus a benefit term ($\gamma \Omega(v_t)$) representing the aggregate gains from decentralized experimentation.

Welfare Implications and Benchmarking

A key result is the theoretical possibility of productive disequilibrium: the stationary regime with nonzero dispersion in beliefs and allocations can dominate the traditional equilibrium with perfect coordination in terms of aggregate welfare, provided the benefit function $\Omega(\cdot)$ is sufficiently steep. In formal terms, for some positive $\gamma$ and sufficiently high $\Omega'(0)$, there exists an interior optimum $v^{opt}>0$ maximizing welfare, implying that neither perfect coordination nor maximal heterogeneity are generically optimal.

The model delivers explicit analytic expressions for the steady-state dispersion, the cross-sectional evolution of means and variances, and the threshold conditions under which productive dispersion exceeds the welfare that could be achieved via perfect ex post allocation. Notably, the findings underscore that any attempt to reduce observed misallocation—typically interpreted as wasteful—can, under plausible conditions, diminish aggregate output by suppressing endogenous exploration and discovery.

Empirical and Policy Implications

The framework provides a distinct reinterpretation of empirical measures of misallocation, such as the cross-sectional variance of marginal products, commonly used to diagnose inefficiency and motivate reallocation policies [restuccia2008policy, hsieh2009misallocation, midrigan2014finance]. In this model, dispersion in realized returns reflects a combination of static inefficiency and experimentation-induced exploration, and the observed correlation between dispersion and future growth can be reconciled theoretically.

On the policy front, the analysis reveals a non-monotonic effect of information and behavioral interventions. Improving information precision or mandating greater coordination compresses cross-sectional heterogeneity but also diminishes aggregate exploration. Therefore, policies aimed solely at reducing belief heterogeneity or misallocation may be welfare-reducing in the long run, especially in innovation-intensive sectors or environments characterized by high uncertainty. Similarly, interventions to “debias” agents should be evaluated in terms of their aggregate impact on exploration rather than simply reducing cognitive noise.

The paper recommends caution regarding blanket policy prescriptions based on the assumption that dispersion is solely symptomatic of inefficiency. Development, industrial, and information policies should recognize their potential effect on both static allocative efficiency and dynamic exploration capacity.

Model Extensions and Theoretical Directions

The model offers a foundation for future work on the interaction between agent heterogeneity, endogenous learning dynamics, and aggregate technological adaptation. Extensions might incorporate richer behavioral updating, distinct sectors with varying exploration benefits, or endogenous entry/exit dynamics. Additionally, the framework is applicable for studying macro-fragility: higher cross-sectional heterogeneity can enhance resilience to shocks via diversification, while excessive coordination amplifies systemic risk.

The results also contribute to the literature on information frictions and rational inattention [angeletos2010noisy, mackowiak2009optimal] as well as recent work on endogenous belief heterogeneity and aggregate fluctuations [bordalo2018diagnostic, coibion2015information]. By formalizing the productive aspect of disequilibrium, the model offers new microfoundations for persistent aggregate heterogeneity and links it directly to the endogenous process of innovation and economic growth.

Conclusion

This paper establishes a novel analytical paradigm for understanding the coexistence of static misallocation and productive decentralized exploration in general equilibrium. Utilizing a behavioral updating mechanism that ensures persistent belief heterogeneity, it is shown that positive steady-state dispersion can be welfare-improving, overturning the standard identification of observed misallocation with pure inefficiency. The framework rationalizes empirical findings relating dispersion to both low contemporaneous and high future productivity, underscores the need for nuanced policy design that accounts for the dual role of heterogeneity, and opens significant avenues for further research on the aggregate implications of endogenous disequilibrium in dynamic economies.