![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
Abstract
Stock market efficiency is associated with news being spread immediately in the market. The literature, however, offers two competing theories to explain this phenomenon. One theory, the mixture of distributions hypothesis (MDH) claims immediate dissemination, while the other, the sequential information arrival hypothesis (SIAH) argues for sequential dissemination, or effectively market inefficiency. The present paper provides a critical test of the two theories using emerging market data, specifically from Egypt, and finds evidence to validate both hypotheses, conditional on the regulatory regime (price limit versus circuit breaker). Using generalized method of moments estimation on 10 years of daily data on the EXG 30 market index, our results show that within the price limit window, news proxied by trading volume, spreads instantaneously to all market participants, consistently with the MDH. Within subsequent circuit breaker window, however, new information leaks out to all market participants only over a period of several days, consistently with the SIAH. We find this switch is moreover associated with an increase in price volatility. Thus, not only is the market less-efficient after the switch, it is also more volatile.
JEL Classification :
Notes
This potential problem was first identified by Lamoureux and Lastrapes Citation(1990), though they did not attempt to resolve it empirically.
Epps and Epps Citation(1976), however, derived a version of the model that takes the form . But where
is the proportional change in prices:
,
.
Karpoff Citation(1987) surveying the 19 existing empirical studies concluded that while the finding of a positive correlation between was universal, the correlations were often weak, especially when transaction data were used.
In their model, CBs are not intended to prevent a rapid, fundamental price adjustment but rather are intended to facilitate it by reducing transactional risks, thereby encouraging value buyers (who may take advantage of, and help reverse, price falls) to bring their demands to market.
Important news events play a role in determining both post-halt volume and volatility by increasing the divergence in interpretations of a common signal. However, empirically, news coverage does not completely explain increased post-halt volume and volatility.
This suggests that the absence of recent transactions prices may make potential traders less willing or able to reveal their demands. This reluctance leads to a noisier reopening price followed by higher volume and volatility in the post-halt period.
The turnover velocity of domestic shares is 66.4% and the number of traders is 13.45 million in 2008.
The trading days of the EGX are from Sunday to Thursday.
The EGX30 index (formerly known as CASE30) is a free-floated market capitalisation-weighted index, with a base value of 1000 points.
We also estimated the volume–volatility relationship for the index constituent (30 firms); however, we did not present it for the sake of space. Results are available from authors upon request.
A GARCH formulation, GARCH (1, 1), was initially estimated with implied symmetric effects for good and bad news. However, significant Engle and Ng Citation(1993) sign and size bias tests indicated that asymmetric GARCH models were likely to be a better fit to stock return volatility on our data.
A value of ˆ u
t−i
>0 is then used as a proxy for good news and a value of ˆ u
t−i
<0 for bad news. In this model, good news, therefore, has an impact α
i
and bad news has an impact . If λ
k
>0, as has been found in other studies, bad news increases volatility more than good news (leverage effect). Non-negativity condition: α0>0, α
i
>0, β
j
≥0,
, and if λ<0,
.
We also estimated alternative IDV measure as , with
, the high, low and closing prices on day t, following Alsubaie and Najad Citation(2009). We found similar results.
We experimented in the econometrics detrended trading volume, which is obtained by regressing raw volume against a quadratic function of time, and we also used the turnover ratio (trading volume/market cap) as a detrended trading volume measure and found very similar results.
In fact, we shall assume in the estimation, along with most of the literature, that p=q=1.
The test implies a lagged rather than simultaneous relationship between the two variables. The null hypotheses are that stock return volatility does not Granger-cause trading volume and trading volume does not Granger-cause stock return volatility.
V
t−1, V
t−2 are uncorrelated with the residuals of the OLS single equation estimates. However, h
t
is correlated with the residuals, and so (Foster Citation1995). Hence, the OLS estimates are inconsistent.
It is well known that an IV estimator must satisfy two conditions. First, it must be correlated with other endogenous variable(s) included in the system, and secondly, it must be orthogonal to the error term. The first condition was tested by examining the fit of the equation at the first stage, or by using the reduced form regressions, and regressing the endogenous variables (volume and volatility) on the full set of IVs (Bound, Jaeger, and Baker, Citation1995). The second condition is related to the correlation between the IVs and errors. This was tested using the J-statistic and the Sargan statistic (value of the GMM objective function at estimated parameters). The J-statistic test is a diagnostic test in any overidentified IV estimation (Newey and West Citation1987). A simple application of the J-statistic is to test the validity of overidentifying restrictions when we have more instruments than parameters to estimate. The null hypothesis is that the overidentifying restrictions are satisfied or that all instruments are orthogonal to the error. The J-statistic times the number of observations in the regression is distributed as χ2 with degrees of freedom equal to the number of overidentifying restrictions (Zarraga Citation2003).
We estimated TARCH model using Akaike and Schwarz information criterion.
We estimated the diagnostic tests (ADF, KPSS, Box and Pierce, and LMACH) for IDV within price limits and CBs windows. Results show that IDV is serially correlated and follows a stationary process. Results are available from the authors upon request.
We found similar results when IDV is used instead of ln vol.
The table is reprinted from Farag, H., and Cressy, R., 2011. Do regulatory policies affect the flow of information in emerging markets?, Research in International Business and Finance 25, no. 3: 251, Copyright (2011), with permission from Elsevier.
EquationEquations (9) | |||||