Market Microstructure Spread Decomposition Huang & Stoll (1997)

Huang-Stoll Bid-Ask Spread Decomposition Lab

Decompose the quoted bid-ask spread into adverse selection ($\alpha$), inventory holding ($\beta$), and order processing ($\gamma$) components using Roger Huang and Hans Stoll's (1997) structural econometric model.

Microstructure Regime Presets:
Adverse Selection ($\alpha$)
35.0%
$0.035 / share | Information cost
Inventory Holding ($\beta$)
25.0%
$0.025 / share | Carrying friction
Order Processing ($\gamma$)
40.0%
$0.040 / share | Net edge
Dealer Cumulative PnL
+$14.20
End Inv: -2 sh | Quoted: $0.20
Model Controls
Huang & Stoll (1997)
Initial Asset Price ($P_0$) $100.00
Quoted Bid-Ask Spread ($S$) $0.20
Total width between quoted ask $A_t$ and bid $B_t$. Half-spread is $S/2$.
Adverse Selection Share ($\alpha$) 35%
Proportion of half-spread compensating for informed order flow.
Inventory Holding Cost Share ($\beta$) 25%
Proportion of half-spread compensating for inventory holding risk.
Order Flow Autocorrelation ($\pi$) 0.30
$\text{Corr}(Q_t, Q_{t-1})$. Positive values model institutional order splitting.
Innovation Noise ($\sigma_e$) $0.08
Standard deviation of fundamental unobserved news shocks ($e_t$).
Real-Time Canvas

Microstructure Moments & Econometric Estimates

Analytical & Empirical
Parameter / Econometric Measure Theoretical Formula Calculated Equilibrium Value
Dealer Profit Sensitivity Matrix

Expected Dealer Net Margin per Trade ($) across Order Flow Autocorrelation ($\pi$) and Adverse Selection ($\alpha$).

Inventory Risk Friction Matrix

Inventory Holding Cost per Share ($) across Quoted Spread ($S$) and Inventory Holding Share ($\beta$).

Huang & Stoll (1997) Microstructure Masterclass

1. Economic Foundations of the Bid-Ask Spread

In financial markets, the quoted bid-ask spread is not pure monopolistic dealer profit. Instead, competitive market makers quote spreads to compensate for three distinct economic frictions identified in modern microstructure theory:

1. Adverse Selection ($\alpha$)

Pioneered by Bagehot (1971), Glosten & Milgrom (1985), and Kyle (1985). Liquidity providers regularly trade against informed investors who possess superior private signals. Because dealers lose money to informed traders, they must widen the spread on all incoming orders to break even.

2. Inventory Holding Risk ($\beta$)

Pioneered by Demsetz (1968), Ho & Stoll (1981), and Avellaneda & Stoikov (2008). Market makers must absorb order imbalances. Holding long or short inventory exposes the dealer to price volatility and capital financing costs, requiring quote skewing and a risk premium.

3. Order Processing & Profit ($\gamma$)

Covers exchange connection, colocation, clearing fees, regulatory compliance, technology infrastructure depreciation, and the dealer's competitive return on equity: $\gamma = 1 - \alpha - \beta$.

2. Structural Equations of the Huang-Stoll Model

Let $Q_t \in \{-1, +1\}$ denote the trade indicator (+1 for buyer-initiated trades, -1 for seller-initiated trades). Unobserved true fundamental value evolves with permanent information shocks:

$$V_t = V_{t-1} + \alpha \frac{S}{2} Q_{t-1} + e_t, \quad e_t \sim \text{i.i.d. } \mathcal{N}(0, \sigma_e^2)$$

The quoted midpoint $M_t$ reflects both fundamental value $V_t$ and the dealer's accumulated inventory adjustment. In increments, the midpoint equation is:

$$\Delta M_t = M_t - M_{t-1} = (\alpha + \beta) \frac{S}{2} Q_{t-1} - \beta \frac{S}{2} Q_{t-2} + e_t$$

Transaction prices occur at $P_t = M_t + \frac{S}{2} Q_t$. First differencing transaction prices yields the complete observable return series:

$$\Delta P_t = \frac{S}{2} Q_t + (\alpha + \beta - 1) \frac{S}{2} Q_{t-1} - \beta \frac{S}{2} Q_{t-2} + e_t$$
3. Econometric Identification & Roll (1984) Contrast

Richard Roll (1984) assumed market efficiency without adverse selection or inventory risk ($\alpha = 0, \beta = 0$). Under pure order processing bounce, serial price covariance is simply $\text{Cov}(\Delta P_t, \Delta P_{t-1}) = -(S/2)^2$, yielding the celebrated Roll spread formula $S_{\text{Roll}} = 2\sqrt{-\text{Cov}}$.

However, when adverse selection ($\alpha > 0$) and order flow persistence ($\pi > 0$) exist, trade covariance becomes positive, often causing the sample covariance of returns to be positive and the Roll model to fail (yielding imaginary spreads). The Huang-Stoll model resolves this fundamental failure by explicitly controlling for order flow autocorrelation $\pi$ and separating temporary inventory mean-reversion from permanent adverse selection shifts.

Interactive Spread Decomposition Mastery Quiz

Related Market Microstructure Simulators

Kyle's Lambda & Microstructure Lab

Model Pete Kyle's (1985) informed insider trading equilibrium, Glosten-Milgrom Bayesian pricing, and adverse selection spreads.

Avellaneda-Stoikov Market Making Lab

Model high-frequency market making, reservation price skewing, inventory penalty control, and optimal limit quote schedules.

Hasbrouck Information Share Lab

Decompose multi-venue price discovery into permanent common random-walk variance and Gonzalo-Granger Component Share.

Yang-Zhang Range Volatility Lab

Estimate historical volatility with 8x to 14x greater efficiency using Dennis Yang and Qiang Zhang's (2000) jump- and drift-adjusted OHLC model.

Frequently Asked Questions

The Huang-Stoll (1997) model, published by Roger Huang and Hans Stoll in the Review of Financial Studies, is the foundational structural econometric framework for decomposing the quoted bid-ask spread into three distinct economic components: adverse selection (alpha) representing private information risk, inventory holding costs (beta) representing dealer inventory rebalancing friction, and order processing costs (gamma = 1 - alpha - beta) representing administrative fees and normal dealer profit.

Adverse selection captures the risk of trading with counterparties possessing superior private information. In the model, when a buyer-initiated trade occurs (Q_t = +1), the fundamental efficient value of the stock permanently shifts upward by alpha * (S / 2). This permanent price impact reflects information revelation, meaning that liquidity providers lose value on the trade unless compensated by a wider adverse selection spread.

The inventory holding parameter beta reflects the market maker's desire to maintain a balanced inventory. When a dealer absorbs consecutive buy orders and becomes heavily short, inventory holding costs induce an asymmetric quote revision: quotes are raised to discourage further buyer-initiated trades and attract seller-initiated orders. This creates temporary, mean-reverting price movements in the quoted midpoint governed by the lagged trade indicator term -beta * (S / 2) * Q_{t-2}.

Order flow persistence pi represents the first-order serial correlation of trade direction: Corr(Q_t, Q_{t-1}) = pi, where P(Q_t = Q_{t-1}) = (1 + pi) / 2. Positive persistence (pi > 0) indicates institutional order splitting or herd momentum, where buys tend to follow buys. Negative persistence (pi < 0) indicates mean-reverting retail order flow. Higher flow persistence magnifies dealer inventory accumulation and increases the likelihood of adverse selection losses.

The effective half-spread measures the immediate transaction revenue collected by the liquidity provider at execution: S^e = Q_t * (P_t - M_t) = S / 2. The realized half-spread measures actual dealer profitability after allowing the market to incorporate private information over horizon tau: S^r_tau = Q_t * (P_t - M_{t+tau}). As tau increases, the price impact Q_t * (M_{t+tau} - M_t) permanently absorbs the adverse selection component alpha * (S / 2), leaving the dealer with only the net realized order processing profit gamma * (S / 2).