D85 | Macro Paper Warehouse

Normal Approximation in Large Network Models

Mon, 01 Jan 0001 00:00:00 +0000

This paper proves a central limit theorem (CLT) for network formation models with strategic interactions and homophilous agents, addressing a foundational inferential gap in the econometrics of large networks. The setting is one where the econometrician observes a single large network — the asymptotic framework sends network size n to infinity — which is the empirically relevant case for most network datasets. The network moments of interest are averages of node-level statistics (1/n) Σ ψ_i, where ψ_i can capture degree, clustering coefficients, or subnetwork counts (triangles, k-stars) that have been used for structural inference in network formation games.

The model is a pairwise-stability network formation game augmented onto a latent-space/geometric-graph structure. Each node i has an i.i.d. type (X_i, Z_i), where X_i is a continuously distributed position vector capturing homophilous attributes. Two nodes i and j form a link if a joint-surplus function V(·) exceeds zero, where V depends on the scaled distance r_n^{-1}‖X_i − X_j‖ between positions, a vector of strategic interaction statistics S_{ij} (functions of neighboring links), node attributes Z_i, Z_j, and an i.i.d. utility shock ζ_{ij}. Homophily enters as a monotonicity requirement: V is decreasing in the distance component, so dissimilar nodes are less likely to link. Sparsity is ensured by setting r_n = (κ/n)^{1/d}, which keeps expected degree asymptotically bounded.

Strategic interactions enter through S_{ij}, which depends on links involving neighbors of i or j (local externalities), generating chains of cross-sectional dependence that are the central obstacle to the CLT. The paper identifies two distinct sources of dependence: (1) link interdependencies from best-response chains, where the realization of one link influences neighboring links; and (2) global coordination in equilibrium selection, where agents may condition on a common signal.

The main technical contribution is adapting “stabilization” conditions from the literature on geometric graphs (Penrose and Yukich 2003, 2008) to the strategic setting. Exponential stabilization (Assumption 5) requires that the radius of stabilization R_i — the smallest neighborhood of i such that ψ_i depends only on nodes within that neighborhood — has a distribution with exponential tails. This bounds the effective dependence neighborhood and provides the weak dependence structure needed for the CLT.

To verify stabilization from primitive conditions, the paper employs branching process theory. The key construct is the “strategic neighborhood” C_i^+, the component of i in the network of non-robust links D (pairs where strategic interactions can change the link outcome). The paper bounds |C_i^+| by a subcritical Galton-Watson branching process: if the mean offspring is below 1 (subcriticality, Assumption 7, stated as ‖h*‖_m < 1), the process is non-explosive and its size has exponential tails, yielding the required stabilization. The subcriticality condition directly restricts the strength of strategic interactions and is the network analog of the condition ‖β‖ < 1 in linear autoregressive models. A second condition (Assumption 8, decentralized selection) requires that equilibrium selection operates independently across disjoint strategic neighborhoods, ruling out global coordination; this holds under myopic best-response dynamics.

For inference, the paper proposes a network HAC variance estimator hat_Σ_n = (1/n) Σ_i Σ_j k(d_{ij}/b_n) hat_ψ_i hat_ψ_j^T, where k(·) is a kernel, d_{ij} is the path distance in A, and b_n is a bandwidth, and a network bootstrap that resamples nodes with replacement. Both are shown to be consistent (Theorem 3). Simulation results with n up to 500, varying strategic interaction strength θ_2 from 0 to 0.5, show that the network HAC estimator achieves nominal 5% rejection rates and 95% coverage for n ≥ 500, while the bootstrap slightly over-rejects in small samples and performance degrades as θ_2 increases.

The scope conditions are explicit: the CLT applies to sparse networks (expected degree bounded), undirected networks with local externalities, models admitting a pairwise-stability equilibrium, and equilibrium selection satisfying decentralization. Extensions to directed or denser networks are left for future work.

Q: What is the primary research question and why does it require new theory? A: The paper asks when sample averages of network statistics — degree, clustering, subnetwork counts — satisfy a CLT in strategic network formation models observed as a single large network. Standard CLT proofs require weakly dependent observations, but strategic interactions generate chains of link dependence of a priori unbounded length, and multiple equilibria allow global coordination, both of which can destroy asymptotic normality. Prior work (Leung 2019b; Menzel 2024) established laws of large numbers but not CLTs, which require stronger conditions.

Q: What is the stabilization condition and why is it the right formulation of weak dependence? A: Exponential stabilization (Assumption 5) requires that the radius of stabilization R_i — the smallest K such that ψ_i depends only on the K-neighborhood of i in the network — has a distribution with exponential tails: lim sup_{w→∞} w^{-η} max{log τ_{b,ε}(w), log τ_p(w)} < 0 for some η ∈ (0,1]. This implies that each node’s statistic depends effectively only on a bounded fraction of the network, making {ψ_i} weakly dependent. The condition is a modification of stabilization conditions from the geometric graph literature (Penrose and Yukich 2003, 2008) adapted to allow strategic interactions.

Q: How does the paper connect the abstract stabilization condition to primitive model conditions? A: The paper defines the strategic neighborhood C_i^+ as the union of one-step network neighborhoods of nodes in i’s component in the non-robust link network D (where D_{ij} = 1 iff the link A_{ij} can be switched by strategic interactions). The size |C_i^+| controls the radius of stabilization. By mapping exploration of C_i via breadth-first search onto a Galton-Watson branching process, subcriticality (mean offspring < 1, i.e., ‖h*‖_m < 1) implies that |C_i^+| has exponential tails, which yields exponential stabilization with η = 1 (Theorem 2).

Q: What is the subcriticality condition and what does it restrict? A: Subcriticality (Assumption 7) requires that the mean interaction-strength measure satisfies ‖h*‖_m < 1, where h* bounds the probability that a given link is non-robust as a function of node attributes. This restricts how strongly the existence of one link influences the probability of neighboring links. The authors explicitly analogize this to the condition ‖β‖ < 1 in linear autoregressive models: both bound the magnitude of “autoregressive” dependence below one to prevent explosive propagation of dependence.

Q: What is the decentralized selection condition and what does it rule out? A: Assumption 8 (decentralized selection) requires that the equilibrium selection mechanism operates independently across disjoint strategic neighborhoods: A_{H_l} = λ_{|H_l|}(r^{-1}T_{H_l}, ζ_{H_l}) for each disjoint strategic neighborhood H_l. This rules out global coordination where agents condition on a common signal (such as the type of a particular node) to jointly select an equilibrium. The condition is satisfied by myopic best-response dynamics and is described as the single-network analog of requiring equilibrium selection to be independent across networks under many-network asymptotics.

Q: What is the structure of the CLT proof? A: The proof has two steps. Step 1 proves a CLT for the Poissonized model where the number of nodes N_n ~ Poisson(n), leveraging results from Penrose and Yukich (2008) for geometric graphs extended to the strategic setting. Step 2 is a de-Poissonization argument that transfers the Poissonized CLT back to the fixed-n model. The abstract CLT (Theorem 1) requires Assumptions 5 and 6, and Theorem 2 establishes that Assumptions 1–8 imply Assumption 5 with η = 1.

Q: How does the network HAC estimator work and what are its consistency conditions? A: The estimator is hat_Σ_n = (1/n) Σ_i Σ_j k(d_{ij}/b_n) hat_ψ_i hat_ψ_j^T, where d_{ij} is the path distance between i and j in the observed network A, k(·) is a kernel function, b_n is a bandwidth, and hat_ψ_i = ψ_i(N_n) − (1/n) Σ_j ψ_j(N_n) is the demeaned statistic. Consistency (hat_Σ_n →^p Σ_n) is established under appropriate conditions on the bandwidth b_n (Theorem 3). The bandwidth plays the same role as in time-series HAC estimation, controlling the window over which covariances are summed.

Q: What do the simulations show about finite-sample performance? A: Using a DGP with X_i ~ U([0,1]^2), ζ_{ij} ~ N(0,1), and θ_2 varying from 0 to 0.5 to control strategic interaction strength, the network HAC estimator achieves nominal 5% rejection rates and 95% coverage at n ≥ 500 across all settings. The bootstrap slightly over-rejects in small samples. Performance of all procedures degrades as θ_2 increases (stronger strategic interactions), consistent with the theoretical condition that subcriticality must hold. These results support practical use of the inference procedures based on Theorem 1.

Q: How does this paper relate to prior work on CLTs for network data? A: Kojevnikov et al. (2021) prove a CLT for node-level data conditional on the network, but this does not apply to network formation because the network is the outcome, not a conditioning variable. Leung (2019b) and Menzel (2024) prove laws of large numbers for strategic network formation but not CLTs. Kuersteiner (2019) takes a different approach using a conditional mixingale assumption. The paper’s abstract CLT extends Penrose and Yukich (2008) by modifying the stabilization condition to accommodate strategic interactions; the primitive conditions are new and use branching process tools that build on Leung (2019b).

Q: What network moments can the CLT be applied to? A: The CLT applies to any average of node statistics ψ_i that depends only on the K-neighborhood of i in the network (Assumption 4 with finite K). Explicit examples include average degree (ψ_i = Σ_j A_{ij}), average clustering coefficient, and counts of connected subnetworks such as triangles and k-stars. Subnetwork counts have been used as the basis for structural identification and estimation of network formation games (Sheng 2020), making the CLT directly applicable to inference in those models.

Q: What are the scope limitations and directions for future work? A: The CLT applies to sparse undirected networks with local externalities (Assumption 2), homophily in positions (Assumption 1), and equilibrium selection satisfying decentralization (Assumption 8). It does not cover directed networks, denser networks where expected degree grows with n, or models with global link externalities. The authors identify extending results to directed and denser networks and developing more powerful inference procedures exploiting network structure as priorities for future work.

Stabilization (exponential): The condition that the radius of stabilization R_i — the smallest neighborhood of i beyond which ψ_i does not depend on further nodes — has a distribution with exponential tails (lim sup_{w→∞} w^{-η} log τ(w) < 0 for η ∈ (0,1]). This is the paper’s operative formulation of weak dependence for network statistics and is adapted from geometric graph theory to the strategic setting.

Strategic neighborhood (C_i^+): The union of one-step neighborhoods of nodes in i’s component in the non-robust link network D. A link (i,j) is non-robust (D_{ij} = 1) if strategic interactions can change its realization — i.e., the surplus V can be positive under some interaction configurations and non-positive under others. The size of C_i^+ governs the radius of stabilization and hence the degree of cross-sectional dependence.

Subcriticality (‖h*‖_m < 1): The condition that the mean-field interaction strength measure satisfies ‖h*‖_m < 1, where h* bounds the conditional probability that a link is non-robust. Subcriticality ensures that breadth-first search of the strategic neighborhood is dominated by a subcritical Galton-Watson process (mean offspring < 1), preventing explosive growth of the dependence neighborhood. The paper explicitly frames this as the network analog of ‖β‖ < 1 in autoregressive models.

Decentralized selection (Assumption 8): The requirement that the equilibrium selection mechanism assigns outcomes independently across disjoint strategic neighborhoods: A_{H_l} = λ_{|H_l|}(r^{-1}T_{H_l}, ζ_{H_l}) for each disjoint H_l. This rules out global coordination — agents conditioning on a common signal to select among equilibria — while permitting local coordination within strategic neighborhoods. Satisfied by myopic best-response dynamics.

Pairwise stability: The solution concept underlying the model. A network A satisfies pairwise stability under transferable utility if A_{ij} = 1{V_{ij} > 0}, meaning a link forms exactly when the joint surplus is positive. This is the equilibrium condition from which the strategic interaction statistics S_{ij} and non-robustness indicators D_{ij} are derived.

Network HAC estimator: The variance estimator hat_Σ_n = (1/n) Σ_i Σ_j k(d_{ij}/b_n) hat_ψ_i hat_ψ_j^T, where d_{ij} is the path distance in the observed network, k(·) is a kernel, and b_n is a bandwidth. It is the network analog of heteroskedasticity- and autocorrelation-consistent (HAC) estimators in time series, using path distance in place of temporal lag distance.

Homophily (in this paper’s sense): The property that the joint-surplus function V is decreasing in the first argument r_n^{-1}‖X_i − X_j‖ (scaled positional distance), so nodes that are more dissimilar in position are strictly less likely to form links. Combined with the sparsity scaling r_n = (κ/n)^{1/d}, this ensures that links decay with distance in social space and that the network remains sparse as n grows.

Production and Financial Networks in Interplay

Mon, 01 Jan 0001 00:00:00 +0000

This paper provides the first integrated empirical analysis of how bank credit supply shocks propagate through both the production network and the financial network simultaneously, using the universe of firm-to-firm VAT transactions and bank-firm credit register data for Spain during the 2008-09 global financial crisis. The theoretical framework, following Bigio and La’O (2016), links credit supply shocks to price distortions in the real economy and derives network-mediated propagation effects. The central empirical finding is that propagation through the production network triples the impact of direct bank credit shocks: a negative bank shock induces a 0.98 percentage point reduction in the directly affected firm’s purchases and sales growth, while first-order network effects add another 0.91 pp and higher-order network effects add 1.07 pp, for a combined indirect effect equal to twice the direct effect. Both upstream and downstream propagation are economically significant and of similar magnitude at the first-order level. Market concentration amplifies all propagation effects, and firms that are simultaneously central in both the production and financial networks generate disproportionately large aggregate contractions.

Summary of a published paper based on the CREI working paper full text, AI-assisted, pending human review. See the linked original for the authoritative claims and full conditions.

Layer 1: Overview

Huremovic, Jimenez, Moral-Benito, Peydro, and Vega-Redondo study how financial shocks originating in the banking sector propagate through interlinked production and financial networks, exploiting Spain’s administrative registers covering essentially the complete production and credit networks of the Spanish economy during the 2008-09 crisis. The Spanish data are unique: approximately 4.3 million VAT firm-to-firm transactions (above a €3,005 threshold) covering 245,000 firms, matched with 1.68 million bank-firm loans from 206 active banks. Bank credit supply shocks are identified using the Khwaja-Mian (2008) / Amiti-Weinstein (2018) approach — isolating bank-level credit supply variation by conditioning on firm-time fixed effects across firms with multiple bank relationships — and cross-validated using banks’ pre-crisis interbank market exposure. The paper’s main contribution is to show that treating production and financial networks separately understates the real effects of financial shocks by a factor of three: the combined direct and indirect (network-mediated) effects are three times the direct bank shock effect alone. First-order and higher-order downstream effects are both quantitatively significant, while upstream propagation is strong at first order but attenuates at higher orders.

In depth

Q1. How does the paper identify bank credit supply shocks, and what makes Spain’s administrative data unusual?

Bank credit supply shocks are identified using within-firm variation across bank relationships — the Khwaja-Mian/Amiti-Weinstein approach — which partials out all firm-level credit demand variation by including firm-time fixed effects, isolating the supply component of each bank’s credit change during the 2008-09 crisis. Spain is particularly suited for this analysis for two reasons. First, it is a bank-dominated economy with minimal shadow banking, so bank credit is the primary external financing channel and the credit register is comprehensive (capturing all loans above €6,000). Second, around 75% of credit comes from firms with at least two banking relationships, enabling the within-firm identification. A complementary shock measure based on banks’ pre-crisis reliance on interbank funding — a market sharply disrupted by the Lehman failure — yields similar results and does not require multi-bank relationships. Crucially, both shock measures show effects that are significant during the 2008-09 crisis but not in the pre-crisis year 2007, consistent with the shocks being crisis-specific supply disruptions rather than pre-existing trends.

Q2. What are the direct effects of bank credit supply shocks on firm-level real outcomes?

At the link (firm-to-firm) level, a direct negative bank credit supply shock to a supplier reduces the purchasing firm’s growth in purchases from that supplier by 3.7 percentage points (29% of the median purchase growth), while a shock to a customer reduces the supplier’s sales growth to that customer by 5.1 percentage points (37% of median sales growth). At the firm level, aggregating across all suppliers and customers, direct bank shocks reduce employment growth by 0.41 percentage points (41% of the median) and investment growth by 0.55 percentage points (9% of the median), consistent with the existing bank lending channel literature. Negative bank shocks also affect total credit availability at the firm level, including trade credit, indicating that the transmission operates through multiple channels and not only through the reduction in direct bank credit.

Q3. How large are the first-order and higher-order network propagation effects, and how do they compare to direct effects?

The first-order indirect effects — propagation from direct customers and suppliers — are of comparable magnitude to the direct bank shock effects: a negative bank shock to all direct suppliers generates a 2.3 pp reduction in firm purchases, while a shock to all direct customers generates a 1.9 pp reduction in sales, both comparable to the 0.98 pp direct effect on purchases and sales combined. Higher-order downstream effects (shocks to suppliers of suppliers) are also quantitatively important at approximately 2.0 pp, similar in magnitude to first-order downstream effects. In contrast, higher-order upstream propagation is weak — only first-order customer shocks matter for upstream transmission. This asymmetry is consistent with the theoretical model’s prediction that upstream propagation is non-linear in shock magnitude, attenuating more rapidly at higher orders than downstream propagation. In aggregate, the combined direct plus first-order plus higher-order effects triple the direct effect: the overall reduction in purchases and sales growth is approximately three times the direct bank shock effect alone.

Q4. What is the symmetric finding on upstream versus downstream propagation, and why does it matter?

Upstream and downstream propagation at the first-order level are of similar magnitude — a negative bank shock induces a 3.7 pp contraction in purchases (downstream, from the shocked supplier to the buying firm) and a 5.1 pp contraction in sales (upstream, from the shocked customer to the selling firm) — challenging the prior literature’s assumption that production network propagation is predominantly downstream. The comparable magnitudes of upstream and downstream propagation imply that financial shocks hitting customers matter for suppliers almost as much as financial shocks hitting suppliers matter for customers. The model provides the analytical basis for this result: downstream propagation is linear in shock magnitude (input supply contraction is passed through proportionally), while upstream propagation is non-linear (demand shortfalls at the customer do not fully translate into supply contraction from the supplier if the customer can be substituted). The near-symmetry at first order, however, means that both channels must be modeled for accurate aggregate impact assessment.

Q5. How does market concentration amplify financial shock propagation?

Firms operating in more concentrated markets — proxied by sectoral market concentration — experience stronger propagation both upstream and downstream; firm-to-firm propagation is also amplified when the two connected firms are mutual trading partners (both buyer and seller of each other), and for downstream propagation specifically when firms are geographically distant and share no common bank. The market concentration amplification is consistent with the theory: concentrated markets have fewer substitution possibilities for inputs and outputs, so firms cannot easily re-route around a shocked partner, forcing the shock to transmit more fully along the existing network link. The amplification from mutual trading ties reflects that the combined demand-and-supply shock through a reciprocal link creates compound effects. The attenuation of downstream propagation when firms share a common bank is consistent with the bank internalizing the financial interdependence of borrowers connected in a supply chain.

Q6. What is the contribution of combining production and financial network analysis jointly, beyond studying either separately?

The joint analysis reveals that the real effects of financial shocks are massively understated when production and financial networks are studied in isolation: the overall impact triples the direct bank shock effect, a result that only emerges when both network structures are mapped and their interaction is quantified. The paper also shows that aggregating to the firm level — rather than analyzing only link-level effects — is essential: firms minimize shocks from individual connections by adjusting across multiple suppliers or customers, so link-level estimates do not translate directly to firm-level outcomes. The joint network analysis further reveals a “dual centrality” amplification: firms that are central both in the production network (high customer-supplier centrality) and in the financial network (large credit relationships with strongly-shocked banks) generate disproportionately large aggregate output contractions. A standard deviation increase in a firm’s customer centrality is associated with a 3 pp decrease in its purchase growth, while the same increase in supplier centrality is associated with a 0.6 pp decrease in sales growth.

Key Concepts

upstream propagation : the transmission of a bank credit supply shock from a directly shocked customer to that customer’s suppliers, operating through the demand channel — a customer facing tighter credit reduces its purchases, contracting the supplier’s sales; the paper shows first-order upstream effects (5.1 pp reduction in sales growth) are of similar magnitude to first-order downstream effects.

downstream propagation : the transmission of a bank credit supply shock from a directly shocked supplier to that supplier’s customers, operating through the supply channel — a supplier facing tighter credit reduces its output, contracting the availability of inputs to customers; both first-order (2.3 pp) and higher-order (2.0 pp) downstream effects are quantitatively large.

dual centrality amplification : the finding that firms simultaneously central in the production network (many supplier-customer relationships) and in the financial network (large credit from banks that receive large supply shocks) generate disproportionately large aggregate output contractions when hit by financial shocks, because the shock propagates through both network channels simultaneously.

Khwaja-Mian identification : the strategy of isolating bank credit supply shocks by exploiting within-firm variation across banks — conditional on firm-time fixed effects, changes in credit from different banks to the same firm reflect supply rather than demand — originally proposed by Khwaja and Mian (2008) and extended by Amiti and Weinstein (2018).

credit network shock : a bank-level credit supply shock derived from the Khwaja-Mian/Amiti-Weinstein methodology, capturing the component of each bank’s credit contraction attributable to bank-level supply factors rather than firm-level demand; the paper uses both this measure and an interbank-market-exposure measure to cross-validate identification.