hard · Market Microstructure spread-econ
A stock has a true (efficient) price following a random walk with per-trade innovation variance σ_u^2. The observed transaction price is the efficient price plus a bid-ask bounce component, where each trade is buyer- or seller-initiated with equal probability and the half-spread is s. Under the Roll model, the first-order autocovariance of observed price changes is -s^2. An analyst estimates the implied spread from this autocovariance but the stock's order flow is actually positively autocorrelated (trade-direction indicator has serial correlation ρ > 0), violating Roll's independence assumption.
Relative to the true effective spread, how does the standard Roll estimator behave?
- It is biased downward, because positive order-flow autocorrelation makes the price-change autocovariance less negative (closer to zero), shrinking the spread implied by 2√(-Cov).
- It is biased upward, because persistent runs of same-side trades amplify the magnitude of the negative autocovariance term and thereby inflate the resulting 2√(-Cov) estimate.
- It stays unbiased, since the random-walk efficient price component is orthogonal to order flow and absorbs the autocorrelation, leaving the underlying bounce covariance fully intact.
- It is biased downward only once the autocovariance turns positive and the estimator becomes undefined; otherwise it still correctly recovers the true underlying spread.
Sign up free to see the explanation and track your rank →
More Market Microstructure spread-econ practice
- If the dealer uses a quote shading parameter of κ = 0.00004 to manage inventory, what is t
- In the Avellaneda-Stoikov model, a market maker who is currently 'long' a significant amou
- A stock has a daily price volatility of 1%. If a trader uses the Roll model and finds that
- Suppose the NBBO for GHI is $100.00 Bid × $100.05 Ask. A mar… — How will they likely 'shad
- If a stock's effective spread is $0.06 and the 5-minute realized spread for the same trade
- According to Roll's model, if the autocovariance of price changes for a stock is -0.0075
- If the market maker applies quote shading with a parameter κ = 0.0001, how will they adjus
- Using Roll's estimator, what is the estimated bid-ask spread?