When David becomes Goliath: Repo dealer-driven bond mispricing

Using proprietary transaction-level data on gilt-backed repo trades, this paper demonstrates that dealer market power and the transmission of heterogeneous shocks within funding networks are significant drivers of bond mispricing, collectively accounting for up to 4 percentage points of yield deviation.

The Gist

Here is an explanation of the paper "When David becomes Goliath," using simple language and everyday analogies.

The Big Picture: The Invisible Handcuffs of the Money Market

Imagine the UK government needs to borrow money. It does this by selling special IOUs called Gilts (like bonds). To make these Gilts trade smoothly, there is a hidden "backstage" market called the Repo Market.

Think of the Repo Market as a giant pawnshop for the financial world.

• The Dealers (The "Goliaths"): A small group of big banks act as the pawnshop owners. They have the keys to the vault and the only way to get cash quickly.
• The Non-Dealers (The "Davids"): These are everyone else—pension funds, hedge funds, insurance companies. They need cash or they need to borrow Gilts to trade, but they can't go directly to the government. They have to go through the dealers.

The Problem: Because there are only a few "pawnshop owners" (dealers) and thousands of "customers" (non-dealers), the dealers have Market Power. They can act like a monopoly. They can charge higher fees or give worse deals because the customers have nowhere else to go.

This paper asks: How much does this "pawnshop power" mess up the price of the government's IOUs (Gilts) and make the whole financial system less liquid?

---

The Three Ways the Dealers Mess Things Up

The authors found that the dealers' power creates three specific types of "friction" (bumps in the road) that distort prices.

1. The "Toll Booth" Effect (Channel A: Individual Power)

Imagine you are driving on a highway, but there is only one toll booth, and the operator decides to charge you extra just because they can.

• What happens: The dealers charge a higher "interest rate" (fee) to lend money or a lower rate to borrow it, simply because they hold the power.
• The Result: This extra fee acts like a tax. It causes the price of the Gilts to drift away from what they should be worth. The paper found this alone causes a price error of about 0.5% to 1.3%.

2. The "Unfair Playground" Effect (Channel B: Uneven Power)

Now imagine a playground where some kids are huge and some are tiny. If the rules are different for everyone, the game gets messy.

• What happens: Not all dealers are equally powerful. Some are giants; others are smaller. When the "giants" charge different fees than the "smaller" dealers, it confuses the market. It's like if one gas station charges $5/gallon and the one next door charges $3/gallon, but you can't easily switch between them.
• The Result: This confusion creates a "misallocation" of money. The paper found that this difference in power between dealers causes even bigger price errors, adding another 2% to 4% to the distortion.

3. The "Ripple Effect" (Channel C: The Shockwave)

Imagine a giant boulder (a shock) falls on the head of the biggest dealer. Because everyone is connected to that dealer, the shock ripples through the whole network.

• What happens: If the biggest dealers get scared or run out of cash (a "persistent shock"), they stop lending or start hoarding. Because they are the center of the network, their panic spreads to everyone else, even if those other people are fine.
• The Result: This creates a "global dealer factor." When the big dealers are stressed, the whole market's liquidity dries up, and prices get weird. This is especially bad for long-term bonds (like 20-year Gilts).

---

The "Dash for Cash" Story (The Stress Test)

The paper looks at a real-life crisis: March 2020 (The "Dash for Cash").

• The Scenario: Everyone panicked. Hedge funds needed cash now. They tried to sell their Gilts to the dealers.
• The Twist: The dealers (the pawnshop owners) were also stressed. They didn't want to take on more risk.
• The Outcome: The dealers used their power to make it very expensive for hedge funds to get cash. This caused the price of Gilts to crash, even though the UK government wasn't in trouble. The paper shows that the dealers' power and their network connections amplified this crash significantly.

The "Two Sides of the Coin" (Repo vs. Reverse Repo)

The paper makes a crucial distinction between two types of transactions:

1. Repo (Dealers lend cash): This is mostly used by Pension Funds and Insurance Companies (long-term investors). When dealers are stressed here, it hurts these "savers."
2. Reverse Repo (Dealers borrow cash/Gilts): This is mostly used by Hedge Funds (short-term traders). When dealers are stressed here, it hurts the "gamblers."

The Analogy:

• If the Repo side breaks, it's like a drought affecting farmers (Pension Funds).
• If the Reverse Repo side breaks, it's like a traffic jam affecting race car drivers (Hedge Funds).
• The paper found that the Reverse Repo side (Hedge Funds) is where the dealers' power is most dangerous and causes the biggest price distortions.

The Bottom Line

"When David becomes Goliath" means that even though individual investors (David) are small, the structure of the market allows the big banks (Goliath) to dominate them.

• The Cost: These frictions (fees, uneven power, and network shocks) cause government bond prices to be wrong by a total of 2.5% to 5.3%. That is a massive amount of money in the bond market.
• The Lesson: We can't just look at the government's debt; we have to look at the "plumbing" (the repo market) that connects it to the rest of the economy. If the plumbing is clogged by a few powerful banks, the whole house (the economy) gets wet.

In short: A few big banks acting like a monopoly in the background are secretly making government bonds more expensive and the financial system more fragile.

Technical Summary

Here is a detailed technical summary of the paper "When David becomes Goliath: Repo dealer-driven bond mispricing" by Cañón, Gerba, and Baruník.

1. Problem Statement and Motivation

The paper investigates the transmission mechanism of funding market frictions from the UK repo market to government bond (gilt) prices and overall market liquidity. While episodes like the March 2020 "dash for cash" demonstrated that repo stress leaks into sovereign bond markets, there is a lack of systematic evidence on how dealer behavior drives this mispricing.

The authors focus on the dealer-centric structure of the UK repo market, where a limited number of bank dealers facilitate funding between non-bank financial institutions (NBFIs) and central counterparties (CCPs) using gilts as collateral. The core research question is: To what extent do dealer market power, the dispersion of that power, and the transmission of shocks through the dealer network explain bond mispricing and liquidity constraints?

2. Data and Methodology

Data

• Source: Proprietary transaction-level data from the UK repo market (January 2016 – January 2022).
• Scope: All repo and reverse-repo trades involving gilts between dealers and non-dealers, and between dealers and CCPs.
• Bond Data: ISIN-level metrics for gilts, including yields, volatility, liquidity (bid-ask spread), and residual maturity.
• Key Variables:
  • Mispricing: Absolute deviation of a gilt's yield from a smooth term-structure benchmark (spline-based).
  • Market Liquidity: A "noise" index proposed by Hu et al. (2013), aggregating bond-level mispricing.
  • Dealer Factors: Global factors capturing transitory and persistent shocks transmitted through the dealer network.

Methodological Framework

The study employs a two-step approach combining structural microeconomics and dynamic network analysis:

Step 1: Estimating Dealer Market Power (Structural Model)

• Approach: A structural supply-and-demand framework treating each dealer as supplying a differentiated funding product.
• Identification: The authors estimate dealer-level markups (in the repo segment, where dealers lend cash) and markdowns (in the reverse-repo segment, where dealers borrow cash).
• Instrumental Variables (IV): To address endogeneity in rates, quantities, and relationship trading, they construct Granular Instrumental Variables (GIVs) following Gabaix and Koijen (2024). This involves extracting idiosyncratic shocks at the dealer-non-dealer dyad level using principal component analysis to isolate variation uncorrelated with aggregate demand shocks.
• Output: Lerner indices (markup/markdown) derived from estimated demand/supply elasticities.

Step 2: Network Analysis and Shock Transmission

• Approach: A Time-Varying Parameter Vector Autoregression (TVP-VAR) model combined with generalized forecast error variance decomposition (GFEVD).
• Mechanism: The model tracks how shocks (transitory vs. persistent) originating from a subset of major dealers propagate through the network to other dealers and non-dealers.
• Metric: Global Dealer Factors are constructed to measure the proportion of market-wide variation attributable to shocks from specific influential dealers, distinguishing between transitory and persistent shocks.

Step 3: Econometric Linkage

• The authors regress bond-level mispricing and market-wide liquidity on:
  1. Dealer markups/markdowns (Channel A).
  2. Dispersion of markups/markdowns across dealers (Channel B).
  3. Global dealer factors (Channel C).
• They control for bond characteristics, monetary policy, and macro-financial covariates.

3. Key Contributions

1. Micro-to-Macro Linkage: The paper explicitly links industrial organization features (dealer market power) in the funding market to individual bond mispricing and aggregate market liquidity, bridging the gap between microstructure and macro-financial stability.
2. Segmentation of Repo Markets: It distinguishes between the repo (liquidity provision) and reverse-repo (liquidity absorption) segments, showing they have asymmetric effects on pricing and different client specializations (e.g., Pension Funds/Insurers in repo vs. Hedge Funds/Asset Managers in reverse repo).
3. Dynamic Shock Transmission: It introduces a novel method to quantify how persistent vs. transitory shocks to core dealers transmit through the network, demonstrating that the persistence of the shock is a critical determinant of its impact on bond prices.
4. Dispersion of Power: It highlights that the distribution (dispersion) of market power among dealers is as critical as the average level of power in causing mispricing.

4. Key Results

A. Dealer Market Power and Mispricing (Channel A)

• Finding: Dealer-specific market power creates significant frictions.
• Quantification: Frictions related to markups and markdowns explain between 0.5 and 1.3 percentage points of bond yield deviations.
• Mechanism: Higher market power leads to wider spreads (higher costs for non-dealers), which translates directly into gilt mispricing. The effect is stronger in the reverse-repo segment.

B. Dispersion of Market Power (Channel B)

• Finding: The heterogeneity of market power across dealers is a major driver of inefficiency.
• Quantification: The dispersion of markdowns (specifically in the reverse-repo segment) explains a significant portion of mispricing. An increase in markdown dispersion by 1% increases bond mispricing by 1.3 percentage points.
• Non-linearity: The effect is non-linear; mispricing rises rapidly as dispersion increases, particularly for long-term gilts.

C. Global Dealer Factors and Shock Transmission (Channel C)

• Finding: Shocks to influential dealers transmit to the broader market, causing mispricing.
• Quantification: The transmission of heterogeneously persistent shocks accounts for 2 to 4 percentage points of yield deviation.
• Asymmetry:
  • Persistent shocks in the reverse-repo segment (affecting Hedge Funds/Asset Managers) significantly increase mispricing.
  • Persistent shocks in the repo segment (affecting Pension Funds/Insurers) actually reduce mispricing (improving alignment with fundamentals).
  • Transitory shocks generally worsen liquidity and mispricing.

D. Market Liquidity

• Higher markdowns and their dispersion, along with exposure to persistent dealer shocks in the reverse-repo segment, are associated with significantly worse market liquidity.
• The "dash-for-cash" period (March 2020) amplified these effects, with dealer factors showing much stronger impacts on medium-term gilts during stress.

E. Aggregate Impact

• Combined, these three frictions account for 2.5 to 5.3 percentage points of gilt yield deviation from the term-structure benchmark.
• The combined effects are sizeable but not purely additive; in some cases, the transmission from the two repo segments acts in offsetting directions.

5. Significance and Policy Implications

• Financial Stability: The paper provides empirical evidence that microstructural frictions in the repo market (dealer dominance and network shocks) are a primary source of sovereign bond mispricing. This suggests that stress in the "plumbing" of the financial system can rapidly destabilize government bond markets.
• Regulatory Focus: Regulators should monitor not just the level of dealer market power but also its dispersion and the connectivity of the dealer network.
• Segmentation Awareness: The distinct behavior of different non-dealer sectors (Hedge Funds vs. Pension Funds) implies that shocks to specific dealer segments have heterogeneous effects on the yield curve.
• Crisis Management: During stress events (like the dash for cash), the transmission of persistent shocks from core dealers becomes the dominant driver of market dysfunction, suggesting that interventions targeting dealer capacity or network connectivity could be more effective than broad liquidity injections.

In summary, the paper demonstrates that the "David" (individual non-dealer) is often at the mercy of the "Goliath" (dealer network), where the structure, power, and shock transmission of a few key dealers dictate the efficiency and pricing of the entire sovereign bond market.