EXcess Idle Time

Abstract

LIQUIDITY is an elusive concept with various dimensions. Our focus, in EXcess Idle Time, is on execution costs (i.e., on "tightness" in the language of Kyle (1985)) rather than on price impacts ("depth" or "resiliency"). This section further illustrates the ability of EXIT to operate as an (il)liquidity measure. We compare it to true execution costs. We do so by varying the sampling frequency , for a given threshold, and by varying the threshold ξ, for a given sampling frequency.Specifically, we conduct simulations in which the data generating process features a time-varying modification on the baseline model in Section 4: we allow for changing execution costs (c) across days. The costs evolve according to the AR(1) model:where the ε t 's are i.i.d. normal. The model parameters are set as being equal to those estimated in Section 7 and reported in Table II We compute EXIT using 1-minute returns ( = 1/420), 5-minute returns ( = 1/84), and 10-minute returns ( = 1/42). For each frequency, we choose three thresholds ξ, namely

Estimating Risk in Illiquid Markets: A Model of Market Friction with Stochastic Volatility

“…There is today a solid literature linking illiquidity to zero returns; see e.g. Lesmond et al (1999), Lesmond (2005), Bekaert et al (2007), Goyenko et al (2009), Naes et al (2011), Bandi et al (2017), Bandi et al (2020). Our analysis reveals that using zero returns alone, as a proxy for market illiquidity, can lead to misleading results in periods of market distress characterized by high volatility.…”

“…A viable proxy of market (il)liquidity should take into account not only the number of zero returns but even the volatility of the information signal. From this point of view, our approach is close to Bandi et al (2017), which estimate a similar microstructure model using a set of moment conditions involving the Excess Idle Time (EXIT) estimator. The latter represents a generalization of the FZR measure which formally characterizes the deviation of the observed price from the traditional assumption made in the continuous-time finance literature of a Brownian semimartingale process.…”

“…The threshold parameter in our model is an estimate of round-trip costs, as in Lesmond et al (1999). From this perspective, our paper also relates to Bandi et al (2017), who estimates a microstructure model with transaction costs and asymmetric information using moment conditions involving the daily fraction of zero returns. Second, in our approach zero returns affect not only the conditional distribution of returns but even the dynamics of volatility.…”

Sluggish news reactions: A combinatorial approach for synchronizing stock jumps

“…Statistical tests for these common jump events, known as "cojump tests", implicitly assume that jumps happen at the same time across all relevant assets but, in fact, jumps occur asynchronously in transaction data. Empirical studies support this view: stock prices can move sluggishly (Bandi et al, 2017), jumps may develop gradually (Barndorff-Nielsen et al, 2009), jumps of less-liquid individual assets typically lag those of the more-liquid market index , and at an ultra-high frequency, cojumps are spurious (Bajgrowicz et al, 2016). Most researchers have dealt with this problem by settling for a coarse sampling grid (Barndorff-Nielsen et al, 2009;Bollerslev et al, 2008;Lahaye et al, 2011;Li et al, 2019).…”

Jumps or Staleness?

“…Zeros are a nonnegligible feature of high-frequency data, even for the most liquid stocks. Their omnipresence is caused by price staleness, an economically justified phenomenon (Bandi, Pirino, and Renò, 2017), and price discreteness (Bandi, Kolokolov, Pirino, and Renò, 2020 discuss how to separate the two phenomena). Nevertheless, the vast majority of models used in finance postulate the mere impossibility of zero returns in the price dynamics.…”