Paper: arXiv 2412.10329

Authors: Lutz Honvehlmann

Abstract

Weighted reciprocity between two agents can be defined as the minimum of sending and receiving value in their bilateral relationship. In financial networks, such reciprocity characterizes the importance of individual banks as both liquidity absorber and provider, a feature typically attributed to large, intermediating dealer banks. In this paper we develop an exponential random graph model that can account for reciprocal links of each node simultaneously on the topological as well as on the weighted level. We provide an exact expression for the normalizing constant and thus a closed-form solution for the graph probability distribution. Applying this statistical null model to Italian interbank data, we find that before the great financial crisis (i) banks displayed significantly more weighted reciprocity compared to what the lower-order network features (size and volume distributions) would predict (ii) with a disappearance of this deviation once the early periods of the crisis set in, (iii) a trend which can be attributed in particular to smaller banks (dis)engaging in bilateral high-value trading relationships. Moreover, we show that neglecting reciprocal links and weights can lead to spurious findings of triadic relationships. As the hierarchical structure in the network is found to be compatible with its transitive but not with its intransitive triadic sub-graphs, the interbank market seems to be well-characterized by a hierarchical core-periphery structure enhanced by non-hierarchical reciprocal trading relationships.

Complexity vs Empirical Score

  • Math Complexity: 8.5/10
  • Empirical Rigor: 7.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper develops a closed-form exponential random graph model with exact normalizing constants, representing advanced statistical mechanics and graph theory, while applying it to real-world Italian interbank data with detailed empirical analysis.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Model weighted reciprocity in interbank networks<br>and identify its structural role"] --> B["Methodology<br>Exponential Random Graph Model with weighted reciprocity<br>Exact expression for the normalizing constant"]
  B --> C["Data Input<br>Italian interbank data pre- and during<br>the Great Financial Crisis"]
  C --> D["Computational Process<br>Apply statistical null model to network topology,<br>weights, and triadic structures"]
  D --> E["Key Findings & Outcomes"]
  
  subgraph E
      F["Pre-crisis: Significantly higher weighted reciprocity<br>than predicted by size/volume"]
      G["Crisis onset: Reciprocity deviation disappears,<br>driven by smaller banks disengaging"]
      H["Hierarchical core-periphery structure<br>enhanced by non-hierarchical reciprocal links"]
  end
  
  E --> I["Implication<br>Neglecting reciprocity/weights<br>leads to spurious triadic findings"]
  
  style A fill:#e1f5e1
  style B fill:#fff3e0
  style C fill:#e3f2fd
  style D fill:#f3e5f5
  style E fill:#fce4ec
  style I fill:#fff9c4

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper develops a novel exponential random graph model called the Reciprocal Enhanced Configuration Model (RECM) that controls for degree, strength and reciprocity simultaneously in directed weighted networks.

Source sentence
In this paper we develop an exponential random graph model that can account for reciprocal links of each node simultaneously on the topological as well as on the weighted level. The main contribution however is methodological: We developed a novel exponential random graph model, the Reciprocal Enhanced Configuration Model (RECM), in closed form which controls for degree, strength and reciprocity simultaneously.

Assumes 2

  • Weighted reciprocity is defined as the minimum of sending and receiving value in a bilateral relationship. model
    Source sentence
    Weighted reciprocity between two agents can be defined as the minimum of sending and receiving value in their bilateral relationship.
  • Dyadic independence is assumed for the linear dyadic independent models, enabling a closed‑form solution for the partition function. model
    Source sentence
    The difficulty generally lies in the large configuration space G which makes an exact enumeration oftentimes infeasible, thus precluding a closed-form solution of Z. In the following we present linear dyadic independent models for which a closed-form solution exists and which furthermore allow for the design of an exact sampling scheme from the probability distribution of networks (Squartini et al., 2015). 22 - 24)

Data

  • Italian interbank e‑MID dataset (tick data on realized trades) · e‑MID platform market-prices unknown
    2005-01-01 to 2011-12-31 · Italian banks trading unsecured overnight deposits
    Source sentence
    4.2 Dataset Our dataset consists of tick data on realized trades on the e-MID platform from 2005-01-01 to 2011-12-31. follow this advice which leaves us with 28 consecutive quarterly, directed and weighted networks between Italian banks on the unsecured market for overnight deposits, Wptq :“ twijptqu, indicating the total lending volume (in Euro) from bank i to j in the respective quarter t.

Costs

Not applicable The paper reports no trading or execution result.

Finds 4

  • Before the crisis, banks displayed significantly more weighted reciprocity than predicted by a null model controlling for trading partners and volume. statistical
    Source sentence
    Applying this statistical null model to Italian interbank data, we find that before the great financial crisis (i) banks displayed significantly more weighted reciprocity compared to what the lower-order network features (size and volume distributions) would predict (ii) with a disappearance of this deviation once the early periods of the crisis set in, Compared to a statistical null model for directed and weighted
  • The excess weighted reciprocity vanished as early crisis signs appeared and remained aligned with model‑implied values throughout the crisis period. statistical
    Source sentence
    Applying this statistical null model to Italian interbank data, we find that before the great financial crisis (i) banks displayed significantly more weighted reciprocity compared to what the lower-order network features (size and volume distributions) would predict (ii) with a disappearance of this deviation once the early periods of the crisis set in, Once the markets showed early signs of heightened counterparty
  • The reduction in reciprocal trading during the crisis is mainly driven by peripheral banks engaging in less intensive reciprocal relationships. statistical
    Source sentence
    Decomposing the network into a set of large core and small periphery banks, we find that this effect is mostly driven by the periphery engaging in less intensive reciprocal trading relationships in times of crisis.
  • Most triadic patterns are explained by lower‑order dyadic effects, with the exception that non‑hierarchical triadic trading cycles are significantly under‑represented in the empirical data. statistical
    Source sentence
    Using a null model which can additionally control for these dyadic reciprocity effects, we found most triadic patterns to be a random outcome of these lower order effects. The only exception to this finding is the tendency to form non-hierarchical triadic trading cycles which are significantly less represented in the empirical data than the models suggest.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 11 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (104)

  1. Too big to fail: The hazards of bank bailouts (2004)
  2. Stylized facts in money markets: an empirical analysis of the eurozone data (2024) — link
  3. Giant strongly connected component of directed networks (2001)
  4. Relationship lending in the interbank market and the price of liquidity (2017)
  5. Generalized percolation in random directed networks (2005)
  6. The asymptotic number of labeled graphs with given degree sequences (1978)
  7. Random graphs with arbitrary degree distributions and their applications (2001)
  8. A contribution to the mathematical theory of epidemics (1927)
and 96 more
  1. Random graphs as models of networks (2003)
  2. Error and attack tolerance of complex networks (2000)
  3. Connectivity of random nets (1951)
  4. On random graphs I (1959)
  5. Random graphs (1959)
  6. Decoding the structure of the WWW: A comparative analysis of Web crawls (2007)
  7. Wikipedias: Collaborative web-based encyclopedias as complex networks (2006)
  8. Directed altruism and enforced reciprocity in social networks (2009)
  9. Economic theories of the firm: past, present, and future (1988)
  10. Wage rigidity in a competitive incomplete contract market (1999)
  11. Does fairness prevent market clearing? An experimental investigation (1993)
  12. Labor contracts as partial gift exchange (1982)
  13. Reciprocity: the supply of public goods through voluntary contributions (1984)
  14. Altruism and economics (1993)
  15. An exact algorithm for a core/periphery bipartitioning problem (2011)
  16. Delinquent networks (2010)
  17. The interbank market after August 2007: what has changed, and why? (2011)
  18. Fitness model for the Italian interbank money market (2006)
  19. A network analysis of the Italian overnight money market (2008)
  20. The topology of the federal funds market (2010)
  21. Emergence of scaling in random networks (1999)
  22. Collective dynamics of ‘small-world’networks (1998)
  23. Network topology of the interbank market (2004)
  24. Financial contagion (2000)
  25. Networks (2018)
  26. Who shall survive?: A new approach to the problem of human interrelations. (1934)
  27. Statistics of social configurations (1938)
  28. An exponential family of probability distributions for directed graphs (1981)
  29. A method for detecting structure in sociometric data (1970)
  30. Transitivity in structural models of small groups (1971)
  31. Network motifs: simple building blocks of complex networks (2002)
  32. Early-warning signals of topological collapse in interbank networks (2013)
  33. The multiplex structure of interbank networks (2015)
  34. Intensity and coherence of motifs in weighted complex networks (2005)
  35. Patterns of link reciprocity in directed networks (2004)
  36. Multispecies grand-canonical models for networks with reciprocity (2006)
  37. Reciprocity of mobile phone calls (2010) — link
  38. Quantifying reciprocity in large weighted communication networks (2012)
  39. Patterns of dominant flows in the world trade web (2007)
  40. Temporal patterns of reciprocity in communication networks (2022) — link
  41. Reciprocity of weighted networks (2013)
  42. Trust and social collateral (2009)
  43. A collection of Fortran codes for large scale scientific computation, 2002 (2002)
  44. Computing core/periphery structures and permutation tests for social relations data (2006)
  45. An efficient heuristic procedure for partitioning graphs (1970) — link
  46. A fast algorithm for the discrete core/periphery bipartitioning problem (2011) — link
  47. Core-periphery structure in networks: a statistical exposition (2022) — link
  48. The average distances in random graphs with given expected degrees (2002)
  49. Financial networks and contagion (2014)
  50. Power and centrality: A family of measures (1987)
  51. On the limiting behavior of parameter-dependent network centrality measures (2015)
  52. Information theory and statistical mechanics (1957)
  53. Information theory and statistical mechanics. II (1957)
  54. Statistical mechanics of networks (2004)
  55. Grand canonical ensemble of weighted networks (2019)
  56. Analytical maximum-likelihood method to detect patterns in real networks (2011)
  57. Enhanced reconstruction of weighted networks from strengths and degrees (2014)
  58. Generalized bose-fermi statistics and structural correlations in weighted networks (2009)
  59. A critical point for random graphs with a given degree sequence (1995)
  60. Entropy of network ensembles (2009)
  61. Unbiased sampling of network ensembles (2015)
  62. Interbank tiering and money center banks (2014)
  63. Finding the core: Network structure in interbank markets (2014)
  64. Core–periphery structure in the overnight money market: evidence from the e-mid trading platform (2015)
  65. On the distribution of links in the interbank network: evidence from the e-MID overnight money market (2015)
  66. On the distribution of links in financial networks: structural heterogeneity and functional form (2020)
  67. Bank size, collateral, and net purchase behavior in the federal funds market: empirical evidence (1989)
  68. Lending relationships in the interbank market (2009)
  69. A micro model of the federal funds market (1985)
  70. Models of core/periphery structures (2000)
  71. Trust and the Interbank Market Puzzle (2022)
  72. Quantifying preferential trading in the e- M I D interbank market (2015)
  73. Effects of central bank intervention on the interbank market during the subprime crisis (2011)
  74. Mind the gap: Disentangling credit and liquidity in risk spreads (2019)
  75. The impact of the financial crisis on inter-bank funding: evidence from italian balance sheet data (2011)
  76. Financial Stability Report N o. 1 (2010)
  77. Networked relationships in the e- M I D Interbank market: A trading model with memory (2015)
  78. Inferring trading dynamics for an O T C market: The case of the euro area overnight money market (2011)
  79. Structure in the italian overnight loan market (2014)
  80. Network analysis of the e- M I D overnight money market: the informational value of different aggregation levels for intrinsic dynamic processes (2013)
  81. When safe proved risky: Commercial paper during the financial crisis of 2007-2009 (2010)
  82. On assortative and disassortative mixing in scale-free networks: The case of interbank credit networks (2013)
  83. An algebraic identity leading to Wilson’s theorem (1996)
  84. Graph Theory, 1736-1936 (1986)
  85. Scale-free networks in cell biology (2005)
  86. Statistical mechanics of complex networks (2002)
  87. Systemic risk in financial networks: A survey (2021)
  88. Multilayer networks (2014)
  89. Mathematical formulation of multilayer networks (2013)
  90. Consistency under sampling of exponential random graph models (2013)
  91. Mixing time of exponential random graphs (2008)
  92. An exponential family of probability distributions for directed graphs: Comment (1981)
  93. The weighted random graph model (2009)
  94. Statistical mechanics of multiplex networks: Entropy and overlap (2013)
  95. Correlations between weights and overlap in ensembles of weighted multiplex networks (2014)
  96. Temporal networks (2012)

0 of 104 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.