Cascades in Networks and Aggregate Volatility

Daron Acemoğlu, Asuman Ozdaglar, Alireza Tahbaz-Salehi2010 · DOI

Abstract

We provide a general framework for the study of cascade effects created by interconnections between sectors, firms, or financial institutions. Focusing on a multi-sector economy linked through a supply network, we show how structural properties of the supply network determine both whether aggregate volatility disappears as the number of sectors increases (i.e., whether the law of large numbers holds) and when it does, the rate at which this happens. Our main results characterize the relationship between first-order interconnections and aggregate volatility, and more importantly, the relationship between higher-order interconnections and aggregate volatility. These higher-order interconnections capture the cascade effects, whereby low productivity or the failure of a set of suppliers propagates through the rest of the economy as their downstream sectors/firms also suffer and transmit the negative shock to their downstream sectors/firms. We also link the probabilities of tail events to sector-specific volatility and to the structural properties of the supply network.

Introduction: The Limits of Diversification

In classical macroeconomic theory, a large economy is often viewed as a collection of independent parts. According to the standard Law of Large Numbers (LLN), if an economy is composed of $n$ independent sectors, idiosyncratic shocks—such as a localized crop failure or a specific factory fire—should cancel each other out as the economy grows. Specifically, aggregate volatility should vanish at a rate of $1/\sqrt{n}$.

However, real-world economies consistently exhibit higher volatility than this theory predicts. This paper addresses this discrepancy by analyzing the interconnections between sectors. In a modern economy, no sector is an island; the output of one industry serves as the input for many others. When these connections are accounted for, a shock to a single supplier does not remain localized—it cascades through the supply chain. This research demonstrates that the specific architecture of these supply networks determines whether micro-shocks average out or explode into macro-crises.

Modeling the Supply Network

The authors develop a mathematical framework representing a multi-sector economy where each sector’s production depends on labor and a combination of inputs from other sectors, modeled via Cobb-Douglas technologies.

The Influence Vector

The central metric in this framework is the "influence vector," which is mathematically equivalent to PageRank or Bonacich centrality. This vector captures a sector's importance not just by its direct customers (first-order connections), but by the importance of its customers' customers (higher-order connections). A sector is highly influential if it supplies other sectors that are themselves central to the economy.

Asymptotic Analysis

To test the Law of Large Numbers, the authors study a sequence of economies as they become increasingly disaggregated (increasing $n$). They find that if a network contains "dominant sectors"—hubs that supply a disproportionate share of the economy—aggregate volatility does not disappear, regardless of how many sectors exist. In these cases, the LLN fails because the economy remains dependent on a few critical nodes.

Empirical Evidence from the US Economy

The paper grounds its theoretical findings using two primary datasets: the 2002 US Input-Output Matrix from the Bureau of Economic Analysis (BEA) and the NBER Productivity Database.

Network Fragility

The analysis reveals that the US supply network is "heavy-tailed." While the first-order degree distribution (direct customers) follows a power law with a shape parameter of $\beta \approx 1.5$, the second-order distribution (cascading influence) is even more extreme, with a parameter of $\zeta \approx 1.3$. This indicates that the cascading effects of shocks are more significant than direct linkages alone would suggest.

Slower Convergence and the Volatility Gap

Because of this centralized structure, aggregate volatility in the US does not decay at the theoretical rate of $n^{-0.5}$. Instead, it decays at a much slower rate of roughly $n^{-0.23}$.

To illustrate the impact, the authors perform a back-of-the-envelope calculation for an economy with approximately 2,300 sectors:

  • Standard Theory Prediction: Aggregate volatility of 0.005.
  • Network-Adjusted Prediction: Aggregate volatility of 0.037.

This seven-fold increase in predicted macro-volatility highlights how network structures amplify small, sector-specific shocks into significant national fluctuations.

Tail Risk and "Black Swan" Events

Beyond general volatility, the paper links network structure to the probability of "tail events"—large, negative deviations in aggregate output. In networks with influential hubs, the risk of a catastrophic economic crash is determined primarily by the single most influential sector rather than the average volatility of all sectors. This suggests that "Black Swan" events in the macroeconomy are often the result of the inherent fragility of the supply network rather than a series of unrelated unfortunate events.

Conclusion and Limitations

This research provides a rigorous foundation for understanding how microeconomic shocks lead to macroeconomic fluctuations. It suggests that monitoring the health of specific, highly-connected sectors is more critical for economic stability than previously thought.

However, the model has limitations. It assumes fixed Cobb-Douglas technologies, meaning sectors cannot easily switch suppliers in the short run. While this is realistic for sudden shocks, firms may adapt over longer periods, potentially mitigating some cascade effects. Furthermore, the network is treated as exogenous; future research could explore how firms’ individual choices to optimize their supply chains might unintentionally increase the fragility of the entire economic system.

Summary updated

Abstract

We provide a general framework for the study of cascade effects created by interconnections between sectors, firms or financial institutions. Focusing on a multi sector economy linked through a supply network, we show how structural properties of the supply network determine both whether aggregate volatility disappears as the number of sectors increases (i.e., whether the law of large numbers holds) and when it does, the rate at which this happens. Our main results characterize the relationship between first order interconnections (captured by the weighted degree sequence in the graph induced by the input-output relations) and aggregate volatility, and more importantly, the relationship between higher-order interconnections and aggregate volatility. These higher-order interconnections capture the cascade effects, whereby low productivity or the failure of a set of suppliers propagates through the rest of the economy as their downstream sectors/firms also suffer and transmit the negative shock to their downstream sectors/firms. We also link the probabilities of tail events (large negative deviations of aggregate output from its mean) to sector-specific volatility and to the structural properties of the supply network.
BibTeX
@misc{acemoglu2010cascades,
  author = {Daron Acemoğlu and Asuman Ozdaglar and Alireza Tahbaz-Salehi},
  title = {{Cascades in Networks and Aggregate Volatility}},
  year = {2010},
  doi = {10.3386/w16516},
}
Figure