Auto-hedging reaches 40 percent of Italian bond trades
Auto-hedging surged from minimal levels before 2016 to 40 percent of Italian sovereign bond trades on MTS by late 2024. A study by Onofrio Panzarino and Antonio Perrella shows dealer-specific technology and risk appetite drive this algorithmic adoption rather than market liquidity alone.
From five percent to forty
Using transaction data from MTS Italy between January 2010 and April 2026, the authors track auto-hedging trades executed within a one-second window to offset inventory shocks.
Usage expanded after mid-2020, with active dealers rising from five to over 22 by mid-2025 across bills and coupon bonds.
In the 2023–2026 sample, 43 percent of auto-hedging trades incurred zero execution cost, with most remaining trades costing under 2 cents.
Logit estimates show dealer fixed effects account for most explanatory power, with R² falling from 0.29 to 0.01 when dealer controls are removed.
Wider bid-ask spreads reduce auto-hedging probability by 1.76 percentage points per standard deviation increase.
Treasury scoring curbs dealer speed
Between 80 and 90 percent of auto-hedging occurs in the exact same bond rather than across correlated instruments.
Sell-initiated trades, where market makers buy bonds, are 1.69 percentage points more likely to trigger auto-hedging than buy orders.
Large orders above €20 million significantly reduce hedging probability due to market impact concerns.
Furthermore, auto-hedging propensity declined from early 2025 following the Italian Treasury's introduction of a new dealer evaluation criterion.
Algorithmic armor with regulatory limits
Automated hedging has become indispensable for sovereign bond dealers managing low-latency quoting risks.
Yet internal technological differences matter far more than market liquidity conditions.
The swift reaction to Treasury scoring rules shows that supervisory incentives can effectively discipline algorithmic behavior.