**3. Results**

The present research investigated whether an investor can utilize this information and yield an abnormal return during the period following the publication of the announcement. In this section, we present the results of event study analyses for the entire sample and for different groups of announcements, as well as the GARCH(1,1) estimation, to examine how the exchange rate was influenced by subjective evaluations of investors.

#### *3.1. The E*ff*ect of Gas Discoveries on the Real Exchange Rate for the Entire Sample*

First, the research used a MAR to examine the influence of NG discovery announcements on the real exchange rate in general, for the entire sample of 296 pure announcements. Table 1 and Figure 3 describe the cumulative average abnormal exchange rate return CAAe−16,+16 during the 33 days surrounding the time of the announcement, beginning from day (−16) prior to publication, and until day (16) following publication.

Table 1 describes the effect of gas discoveries on the abnormal real exchange rate for the entire sample. In the table, the cumulative average abnormal exchange rate return (CAAe), median cumulative abnormal exchange rate return (CAe), percentage of positive abnormal real exchange rate return, t-statistics, and number of observations are reported for the six event windows.

**Table 1.** The effect of gas discoveries on the real exchange rate for the entire sample of 296 pure announcements. CAAe: cumulative average abnormal exchange rate return, CAe: median cumulative abnormal exchange rate return.


\* 90% significance level.

**Figure 3.** The cumulative average abnormal exchange rate return (CAAe) behavior during the 33-day event window surrounding the event day for the entire sample of 296 announcements.

Table 1 and Figure 3 show that during the period prior to publication of the announcement, the CAAe−16,−1 did not significantly decrease, which indicates that in general during this period, there was no leakage of information regarding the content of the announcement. Alternatively, the influence of publication of the announcement occurred a long time before the actual publication of the exchange rate. During this period, the Median CAe−16,−1 was almost unchanged, and 47% of the announcements had a positive CAe. On the day that the announcement was published, there was a significant decrease of 0.06% AAe0 in the exchange rate. This indicates that the announcements generally strengthened the ILS. The Median CAe0 decreased on this day by 0.09%, and 43.38% of the announcements had a positive CAe. Thus, it may be concluded that generally, the announcements provided new information to the investors on the day of publication and contributed to the appreciation of the shekel.

During the first three days (+1,+3) following the announcement, the abnormal appreciation increased and the CAAe+1,+3 decreased by an additional 0.11%. During this period, investors that were exposed to the first announcement were able to use a profit strategy by shorting the real exchange rate on the day of publication and closing the position at the end of the trading day on the third day following publication of the announcement. During this period, the Median CAe+1,+3 decreased by 0.24%, and 44.93% of the announcements had a positive CAe. In the subsequent period (+4,+16), the publication of announcements did not show a statistically significant impact on the exchange rate.

### *3.2. Analysis by Type of Announcement*

This section examines the impact of NG-related announcements on the real exchange rate according to five groups of announcements: Panel A (Export), Panel B (Exploration and Drilling), Panel C (Companies and Development), Panel D (Electricity and Infrastructure), and Panel E (Import and Regulation), as presented in Section 2.1.

Table 2 and Figure 4 describe the CAAe−16,+16 by group during the 33 days surrounding the announcement, from day (−16) prior to publication and until day (16) following publication.


**Table 2.** The e ffect of gas discoveries on the exchange rate by announcement type.

\*\*\* 99% significance level; \*\* 95% significance level; \* 90% significance level.

The following table describes the e ffect of gas discoveries on the abnormal exchange rate by five announcement types: Panel A (Export), Panel B (Exploration and Drilling), Panel C (Companies and Development), Panel D (Electricity and Infrastructure), and Panel E (Import and Regulation).

In each panel, the cumulative average abnormal exchange rate (CAAe), median cumulative abnormal real exchange rate (CAe), percentage of positive abnormal real exchange rate, t-statistics, and number of observations are reported for the six event windows.

Figure 4 and Table 2 show that there was no information leakage in any of the groups except for the Electricity and Infrastructure group. In that group, the information leakage began three days before the publication of the announcement and continued until one day before publication of the announcement, which was accompanied by a decrease in CAAe−3,−1 by 0.22% (t-statistics = −2). This provides an indication that investors, who had inside information concerning the decisions in the area of electricity and infrastructure, were able to short the exchange rate three days before the information was exposed to the rest of the investors in the market. They potentially could close the position at the end of the trading day prior to publication of the announcement.

**Figure 4.** The CAAe behavior during the 33-day event window surrounding the event day for the entire sample of 296 announcements by five announcement types.

Also of interest in this period is that the Percent Positive shows the tendencies of the groups in Figure 4. In the Exploration and Drilling group and the Import and Regulation group, approximately 47% of the announcements yielded an increase in the real exchange rate. Therefore, the two graphs remained stable and without change in the real exchange rate. In the Companies and Development and Electricity and Infrastructure groups, approximately 39% of the announcements yielded devaluation; thus, there was a negative trend in the exchange rate in those two groups.

In the Export group and the Companies and Development group, there was a statistically significant decrease in AAe0 by 0.12% (t-statistics = −2.13) and 0.15% (t-statistics = −2.12) in the exchange rate. More importantly, the statistically significant abnormal appreciation of ILS on the day of announcements related to the prospects of exports and development of the reserves provides direct evidence for investors' strategy in view of the expected NG export and future currency appreciation.

In the Electricity and Infrastructure group during the first three days following the announcement (+1,+3), there was a statistically significant decline in the exchange rate, and the CAAe+1,+3 decreased by 0.27% (t-statistics = −2.43). During this period, investors who were exposed to the first announcement were able to use a profit strategy by shorting the exchange rate on the day of the announcement and closing the position at the end of the trading day on the third day following the publication of the announcement. During this period, the Median CAe+1,+3 decreased by 0.25%, and 34.72% of the announcements had a positive CAe.

In the Exploration and Drilling group, the statistically significant impact on the exchange rate occurred only during the first few days (+4,+6) after the announcement. Here, the exchange rate decreased, and the CAAe+4,+6 decreased by 0.21% (t-statistics = −1.70). During this period, the Median CAe+4,+6 decreased by 0.27%, and 32.86% of the announcements had a positive CAe.

In the Companies and Development group, the greatest influence was during the period following the announcement. During the 16 days following the announcement (+1, +16), the real exchange rate decreased and the CAAe+1,+16 decreased by 0.93% (t-statistics = −3.19). During this period, investors who were exposed to the first announcement were able to use a profit strategy by shorting the real exchange rate on the day of the announcement and closing the position at the end of the trading day on the 16th day following the publication. In this period, the Median CAe+1,+16 decreased by 0.93%, and 30.12% of the announcements had a positive CAe.

Figure 5 summarizes the results for the five groups surrounding the event day. The present research examined whether there was information leakage to investors who were insiders and associated with the decision makers during the period preceding the event. Figure 5 shows the information leakage only in the Electricity and Infrastructure group. The research further considers whether the announcement brought new information to investors on the event day. In the Export group and the Companies and Infrastructure group, the new information had a statistically significant impact and appreciated the ILS. During the period following the publication, investors sought a profit strategy. At that time, three groups had a statistically significant influence on the exchange rate. These were the Electricity and Infrastructure group, Exploration and Drilling group, and the group that had the most influence on profit strategy: Companies and Development.

**Figure 5.** Summary of results according to five groups of announcements.

### *3.3. Old vs. New Announcements*

This section examines whether announcements published in recent years influenced the exchange rate differently than announcements published earlier. For the purpose of this analysis, the sample was divided into two groups. The first group, referred to as Old, includes 146 old announcements that were published between 15 December 2008 and 31 December 2012. The second group, referred to as New, includes 150 announcements that were published between 1 January 2013 and 6 May 2017. In 2013, the Tamar gas field was connected and began delivering NG to Israel. Therefore, this group analyses whether the impact of the announcements became stronger after the supply of domestic NG began.

Figure 6 and Table 3 describe the CAAe−16,+16 during the 33 days surrounding the announcement, beginning from day (−16) prior to and until day (16) following the publication.

**Figure 6.** The CAAe behavior during the 33-day event window surrounding the event day for old and new announcements.


**Table 3.** The effect of gas discoveries on the exchange rate for old and new announcements.

\*\*\* 99% significance level; \*\* 95% significance level.

The following table describes the effect of gas discoveries on the abnormal exchange rate for two announcement types: Panel A (Old) and Panel B (New).

In each panel, the cumulative average abnormal real exchange rate (CAAe), median cumulative abnormal real exchange rate (CAe), percentage of positive abnormal real exchange rate, t-statistics, and number of observations are reported for the six event windows.

Figure 6 and Table 3 show that there was no information leakage in any of the old and the new groups. Therefore, it may be concluded that there was no information leakage and that investors who were corporate insiders reacted similarly to new information and to old information before it was published for the general public. On the day of publication, old announcements had no influence on the exchange rate. However, in the group of new announcements, the AAe0 decreased significantly by 0.09% (t-statistics = −3.67). This indicates that the new announcements created an appreciation of the ILS. The Median CAe0 decreased in this group by 0.12%, and 44% of the announcements had a positive CAe. It may be concluded that since the connection of the Tamar gas field in 2013 for the

supply of domestic NG to the Israeli economy, the potential for export became more realistic. Every announcement regarding NG strengthened the expectations for a future appreciation, and brought about a stronger reaction of the investors on the day of the publication. This influence also continued during the first six days following the announcement (+1,+6), so that the exchange rate decreased and the CAAe+1,+6 decreased by 0.20% (t-statistics = −2.41). During this period, investors that were exposed to the first announcement were able to use a profit strategy by shorting the exchange rate on the day of publication of the announcement and closing the position at the end of the trading day on the sixth day following publication. During this period, the Median CAe+1,+6 decreased by 0.29%, and 39.36% of the announcements had a positive CAe.

### *3.4. Single vs. Multiple Announcements*

When the market has expectations for additional information, it is likely that investors' reactions will strengthen when given continuous information rather than a single information item. This section examines the di fference in the behavior of investors toward announcements that appear only once, which are referred to as single announcements, versus those that appear a number of times, which are referred to as multiple announcements. The analysis was made on the first announcement in the series of multiple announcements. The group of single announcements includes 243 announcements, which were not republished during the period of three months from publication of the first announcement. In contrast, the group of multiple announcements includes 43 announcements, for which an additional announcement was issued within three months after the initial publication.

Table 4 and Figure 6 describe the CAAe−16,+16 during the 33 days surrounding the announcement, beginning from day (−16) before publication and continuing until day (16) following publication.


**Table 4.** The e ffect of gas discoveries on the real exchange rate for single and multiple announcements.

> \*\* 95% significance level.

The following table describes the e ffect of gas discoveries on the abnormal exchange rate for two announcement types: Panel A (Single) and Panel B (Multiple).

In each panel, the cumulative average abnormal exchange rate (CAAe), median cumulative abnormal exchange rate (CAe), percentage of positive abnormal exchange rate, t-statistics, and number of observations are reported for the six event windows.

Table 4 and Figure 7 show no influence of single announcements on the exchange rate, whereas for multiple announcements, information leakage began from the second day before the event and was accompanied by currency appreciation. The decrease continued on the day of publication of the announcement, as well as during the first three days following the event. It may be concluded that investors were able to identify initial single announcements and initial continuing announcements. Therefore, they were able to attain abnormal profits in the multiple announcements group. The profit strategy could be shorting the exchange rate either two days prior to the publication for investors with inside information, or on the day of the publication for the rest of the investors, and closing the position at the end of the third trading day following the event. A possible reason for this phenomenon is that the investors had expectations for future information that would be published in additional announcements.

**Figure 7.** The CAAe behavior during the 33-day event window surrounding the event day for single and multiple announcements.

### *3.5. GARCH Regression Results*

The validity of our results based is on GARCH modeling of exchange rate returns and the parametric test statistic. Table 5 describes the estimation of the parameters according to the expected normal return of exchange rate volatility (Equation (3)) during the period of the estimation window.

**Table 5.** Estimation of expected normal return for exchange rate and its volatility according to the GARCH(1,1) model during the period of the estimation window.


The results in Table 5 show that the model is significant (F = 6.35), as well as the explanatory variables: change in the square exchange rate in day t − 1 ε2*t*−<sup>1</sup>, and the variance in day t − 1 (σ2*t*−1).

Using the estimations from GARCH(1,1) regression, MAR was examined to reveal the influence of NG announcements on the exchange rate for the entire sample of 296 pure announcements. Table 6 and describes the CAAe−16,+16 during 33 days surrounding the time of the announcement, beginning from day (−16) prior to publication, and until day (16) following publication.


**Table 6.** The e ffect of gas discoveries on the real exchange rate for the entire sample of 296 pure announcements with the GARCH model.

\*\* 95% significance level.

The table describes the e ffect of gas discoveries on the abnormal real exchange rate for the entire sample with the GARCH model. In the table, the cumulative average abnormal real exchange rate (CAAe), median cumulative abnormal real exchange rate return (CAe), percentage of positive abnormal real exchange rate return, t-statistics, and number of observations are reported for the six event windows.

Table 6 reveals that during the period prior to publication of the announcement, the CAAe−16,−1 did not significantly decrease, indicating that in general during this period, there was no leakage of information regarding the content of the announcement. Alternatively, the influence of publication of the announcement occurred a long time before the actual publication of the real exchange rate. During this period, the Median CAe−16,−1 was almost unchanged, and 46.5% of the announcements had a positive CAe. On the day of the announcement, there was a statistically significant decrease of 0.08% in AAe0 (t-statistics = −2.31) in the exchange rate. This indicates that the announcements generally strengthened the ILS. The Median CAe0 decreased on this day by 0.11%, and 42.91% of the announcements had a positive CAe. Thus, it may be concluded that generally, the announcements provided new information to the investors on the day of publication and contributed to the appreciation of ILS.

During the first three days (+1,+3) following the announcement, the abnormal appreciation increased and the CAAe+1,+3 decreased by an additional 0.15% (t-statistics = −2.45). During this period, investors that were exposed to the first announcement were able to use a profit strategy by shorting the real exchange rate on the day of publication and closing the position at the end of the trading day on the third day following publication of the announcement. During this period, the Median CAe+1,+3 decreased by 0.10%, and 44.59% of the announcements had a positive CAe. To summarize, the GARCH regression results confirm the general outcome of event study evaluation and confirm the hypothesis that NG-related announcements strengthened ILS.
