Six microstructure measures, one composite — and the venue-cap puzzle

The microstructure literature has accumulated about a dozen liquidity measures, each capturing a different facet — quoted spread, no-trade frequency, price impact per unit traded, autocovariance-based implicit spread, high-low spread, return-on-flow regression. Six are usable on the Polish BondSpot panel without proprietary order-book data: bid-ask spread (BAS), zero-trading-day share (ZTD), Amihud (2002) ratio, Roll (1984) implicit spread, Corwin-Schultz (2012) high-low spread, and a Pastor-Stambaugh-style γ. We aggregate them into a single composite z-index for cross-section monitoring.

Two empirical regularities deserve attention. The first is a familiar one: liquidity measures are highly correlated within the cross-section but diverge sharply on stress-test dates such as the 2008-09 crisis, the May 2013 taper tantrum, the March 2020 COVID dislocation, and the 2022 inflation surge. The composite z-index spikes on each of those events, with magnitudes ranging from +1.2 (taper) to +3.1 (COVID).

The second regularity is genuinely new and is the venue-cap puzzle. Aggregate Polish bid-ask spreads do not respond to changes in total outstanding sovereign debt or to changes in the foreign-investor share, even when both are large. The same shocks load strongly and significantly on Adrian-Crump-Moench term premia at the 2-, 5-, and 10-year horizon. The mechanism is institutional: BondSpot Annex G caps the maximum bid-ask spread that a market-maker can quote, so quoted BAS is mechanically bounded above and cannot widen further when the marginal holder bears more inventory risk. The term premium is free.

The venue-cap finding has a direct policy implication: the Polish Ministry of Finance and Narodowy Bank Polski cannot use BondSpot bid-ask spreads as a real-time stress signal for the long end of the curve, because the cap binds. A composite z-index that puts weight on Amihud, Roll, ZTD and γ is informative; the headline BAS in isolation is not. The aggregate term premium is the cleaner signal for supply-side stress, but it requires a curve-fit and a no-arbitrage model and is therefore only available with a one-day lag and a non-trivial methodology.

For practitioners and policy users, the practical recommendation is to monitor the composite z-index daily and to cross-reference against the BRW 5-year and 10-year term premia at month-ends. The two together cover the venue-cap blind spot.