The effect on prices of share dealing by directors
Distribution 1 of the size of the transactions by directors 2 compared FIGURE 2
5 Empirical modelling and results
5.2 Multivariate approach
As indicated above, the multivariate statistical analysis makes it possible to test the ro-bustness of the above analysis. In addition, with this new analysis we may obtain more detailed information on certain aspects of the effects of share trading by directors.
This methodology allows us to use a greater number of observations as it is not nec-essary to delete those trades which overlap in the period subsequent to others. Fur-thermore, it also makes it possible to study as a separate phenomenon the effect on the price both on the day of the purchase and on the day on which the CNMV dis-closes this information to the market through its website.
The chosen method allows the effect of the transactions and their reporting to be different for each of the companies analysed. However, it imposes as a restriction that for one company, this effect must be similar throughout the sample period, with it being only sensitive to differences in the size of the transaction.
The model also uses the daily return of the share as a dependent variable although in this case it is not restricted to a time window, but uses the period between Janu-ary 2007 and June 2012. By construction, the error variance is considered constant during the sample period for each company.
This approach makes it possible to use more data than the univariate method.
Consequently, it is applied to the data of 117 companies, considering a total of 4,977 purchases and 1,914 sales by directors.
As control variables, we have used those included in the CAPM model (free of risk return and return of the Ibex 35). In order to analyse the effect of trading by direc-tors on share prices, the following variables are entered into the model:
– Variable which takes the value 1 on the date that the director buys shares and zero on other dates.
– Variable which takes the value 1 on the date that the director sells shares and zero on the other dates.
– Variable which takes the value 1 on the date the share purchase is registered with the CNMV and published and zero on the other dates.
– Variable which takes the value 1 on the date the share sale is registered with the CNMV and published and zero on the other dates.
– Size of the purchase, measured as the Napierian logarithm of the percentage of the capital acquired in the transaction.
– Size of the sale, measured as the Napierian logarithm of the percentage of the capital disposed of in the transaction.
The statistical testing has shown that both purchases and sales of shares affect the share price on the day on which they take place and they do so with the expected sign. In addition, the effect on the same day as the transaction is sensitive to the size of the transaction performed by the director so that larger purchase (sale) transac-tions produce a greater rise (fall) in the share price.
Multivariate approach. Effect on share prices TABLE 3
Ratio (mean)1 Wald Test 99% test Prob.
Purchases with an effect on the day of the
transaction 0.00205 116063 239.4 0
Purchases with an effect on the day of
registration -0.00001 598 133.5 0
Sales with an effect on the day of the transaction -0.01496 4644951 181.8 0 Sales with an effect on the day of registration -0.00092 35336 111.1 0
log(ACC/NOSH) Purchases 0.00010 77621 127.6 0
log(ACC/NOSH) Sales -0.00299 4280164 90.8 0
1 The table shows the mean of the estimated ratios for each one of the shares analysed in the sample, the mean of the dummy variable created for each one of the events (purchases on the day of the transaction or registration) and the mean of the variable of the size of the purchase in relation to total capitalisation. The ratios arise from a CAPM model and should therefore be interpreted as the average effect of the purchase or sale on the return of the share adjusted by the general evolution of the market. The comparisons of signifi-cance are made by means of the Wald test and, therefore, they apply to the set of estimates of the sample.
However, in contrast to the expectations of the theory, the day on which the pur-chase transactions are registered with the CNMV there is a fall in share prices. Al-though the effect is relatively small, it is significant and might indicate that, for the sample set, the market has already processed the information on the purchases by the directors at the time of the reporting. Nevertheless, the authors of this article do not have a clear explanation for this phenomenon. Notifications of sales do have the expected effect on prices.
Securities market legislation allows the directors of a company to trade in the shares of their own company providing said trading is not related to material information, which would be subject to a penalty for market abuse. Excluding these cases, we can assume that the directors of a listed company will have a greater capacity to assess their company than other investors, either through their education, training or ac-cess to non-material information which is not known by the rest of the market.
This fact means that other market participants pay special attention to trading by directors as these transactions may provide a signal as regards future changes in the company’s value or at least its market price. In a simple behavioural model in which a director buys (sells) shares simply to take advantage of the company being under-valued (overunder-valued), we can expect that requiring directors to report their transac-tions will increase informational efficiency and market liquidity and reduce the possible profits which the director may obtain from this trading.
European legislation, and with it Spanish legislation, follows this transparency mod-el on purchases and sales of shares and other rmod-elated financial contracts. It requires that they be reported in a period of five days, detailing the reasons, instruments, amounts and price, and requires that this information be disclosed to other partici-pants.
Using the data available in the CNMV’s Public Registry on the notifications of trans-actions by directors and executives for the period 2007-2012, we have studied the transactions and their short-term impact on the price of the affected shares.
Using two different approaches, we have found that on average purchases by direc-tors are usually related to increases in the share price, and these increases are con-centrated on the purchase days. With regard to sales, the results have been mixed as a conclusive result has only been obtained in one of the approaches.
In addition, the variance in the share prices is sensitive to the size of the transac-tions performed by the directors of the company (both individually and on an ag-gregate basis), which is again in line with the theory that, as the director’s economic risk increases, the signal provided by his/her transaction for other participants in the market is more powerful.
In addition, we have found that the effect of this trading on the price is concentrated at the time of the trade and not at the time it is registered with the CNMV and made
available to other market participants. In accordance with the theoretical model used, it could be expected that the price effect should be concentrated specifically at the time that the notification is disclosed.
At any event, with the data from the Spanish market, we can conclude that the im-pact on the price as a result of trading by directors is very low. This is largely due to the small size of the typical transaction in the sample used. However, the decision on the size of the trade by the director is not separate from securities market legisla-tion or the market institulegisla-tions. We could expect that regulalegisla-tion on the reporting of these transactions, the sensitivity of market participants to these transactions and the regulation of market abuse will encourage directors and managers to perform relatively small transactions and that the benefit obtained would normally be low.