We examine the statistical properties of daily changes in emerging market bond spreads
over US treasuries, and simulate an agent-based model to attempt to replicate those properties.
The actual data indicate that changes in spreads: 1) are definitely not normally distributed,
exhibiting much fatter tails; 2) are serially correlated, suggesting deviation from market
efficiency; and 3) exhibit excessive co-movement, suggesting contagion. A simple model of
interacting traders produces alternating booms and crashes, as in reality, but is not capable of
producing fat-tailed distributions or contagion. We focus on an extended model with market
makers whose bid/asked spreads widen with increased volatility and the size of their inventory.
This model highlights the role of liquidity (or lack of it) in explaining large rate movements and