recent developments in high frequency financial ... - Index of
recent developments in high frequency financial ... - Index of
recent developments in high frequency financial ... - Index of
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
How large is liquidity risk <strong>in</strong> an automated auction market? 113<br />
example), ex-ante measures <strong>of</strong> liquidity are limited to quoted <strong>in</strong>side spread and<br />
quoted <strong>in</strong>side depths. It is also important to note that liquidity measures such as<br />
def<strong>in</strong>ed <strong>in</strong> Eq. (1) characterize committed liquidity as given by the stand<strong>in</strong>g limit<br />
orders only. With hybrid trad<strong>in</strong>g systems which mix characteristics <strong>of</strong> order book<br />
and dealer systems, unit prices at(v) and b t(v) give an upper bound on the price to be<br />
paid for the trade as the additional participants can add liquidity prior to the<br />
execution <strong>of</strong> the trade, decreas<strong>in</strong>g a t(v) or <strong>in</strong>creas<strong>in</strong>g b t(v). 3 This is also the case for<br />
automated auction markets which allow so-called hidden or iceberg orders (see<br />
below). In order book markets which feature hidden orders, ex-ante costs <strong>of</strong> trad<strong>in</strong>g<br />
measures and liquidity risk measures such as computed <strong>in</strong> this paper give an upper<br />
bound on these trad<strong>in</strong>g and liquidity costs (see also Beltran et al. 2005a). 4<br />
In this paper we will show that, with suitable data at hand, it is possible to<br />
quantify the liquidity risk over short term time horizons <strong>in</strong> automated auction<br />
markets. More precisely, we <strong>in</strong>troduce liquidity risk measures that take <strong>in</strong>to account<br />
the potential price impact <strong>of</strong> liquidat<strong>in</strong>g a portfolio. This approach is particularly<br />
relevant for short term impatient traders who submit market orders. The core <strong>of</strong> our<br />
methodology relies on the comparison <strong>of</strong> risk measures for so-called frictionless<br />
returns (i.e. no-trade returns) and actual returns (which take <strong>in</strong>to account the actual<br />
trade price for a v-share trade). These actual returns are particularly relevant for<br />
short-term impatient traders who currently hold the stock and who are committed to<br />
shortly submit a marketable sell order. In contrast, frictionless returns refer to<br />
traders who hold the stock over the same time period, but do not <strong>in</strong>tend to sell their<br />
shares. We rely on measures that orig<strong>in</strong>ate from the Value-at-Risk methodology<br />
(see Section 3) to characterize the liquidity risk. It should however be stressed that<br />
our framework is not the usual 10-day VaR framework familiar to f<strong>in</strong>ancial<br />
regulators. Hence, we use the VaR methodology to def<strong>in</strong>e our <strong>in</strong>traday risk measures,<br />
but <strong>in</strong> our paper these risk measures are meant to assess the <strong>in</strong>traday<br />
immediate liquidation risk faced by impatient traders. Consequently, we do not<br />
derive implications for f<strong>in</strong>ancial regulators. In contrast, our approach is more<br />
similar to Andersen and Bollerslev (1997); Giot (2000, 2005) or Chanda et al.<br />
(2005) who characterize volatility on an <strong>in</strong>traday basis.<br />
In contrast to a standard (frictionless) VaR approach, <strong>in</strong> which one uses prices<br />
based on mid-quotes, the Actual VaR approach pursued <strong>in</strong> this paper uses as <strong>in</strong>puts<br />
volume-dependent transaction prices. This takes <strong>in</strong>to account the fact that buyer<br />
(seller) <strong>in</strong>itiated trades <strong>in</strong>cur <strong>in</strong>creas<strong>in</strong>gly <strong>high</strong>er (lower) prices per unit share as the<br />
trade volume <strong>in</strong>creases. The liquidity risk component naturally orig<strong>in</strong>ates from the<br />
volume dependent price impact <strong>in</strong>curred when the portfolio is liquidated. Our<br />
approach relies on the availability <strong>of</strong> <strong>in</strong>traday bid and ask prices valid for the<br />
immediate trade <strong>of</strong> any volume <strong>of</strong> <strong>in</strong>terest. Admittedly, procur<strong>in</strong>g such data from<br />
traditional market maker systems would be an extremely tedious task. However,<br />
the advent <strong>of</strong> modern automated auction systems <strong>of</strong>fers new possibilities for em-<br />
3 Examples are a comb<strong>in</strong>ation <strong>of</strong> a limit order book and market markets who br<strong>in</strong>g additional<br />
liquidity (Euronext or Xetra, for non-actively traded stocks), or a comb<strong>in</strong>ation <strong>of</strong> a limit order<br />
book, a specialist and floor traders (NYSE), see S<strong>of</strong>ianos and Werner (2000) or Venkatamaran<br />
(2001). Note that the German Stock Exchange <strong>recent</strong>ly adopted the price impact as def<strong>in</strong>ed <strong>in</strong><br />
Eq. (1) as the key liquidity <strong>in</strong>dicator for the automated auction system Xetra (see Gomber et al.<br />
2002).<br />
4 Nevertheless, they do provide mean<strong>in</strong>gful <strong>in</strong>formation as they characterize the worst-case<br />
scenarios.