3.3.3. Trends in Employment

In this part, we focus on employment trends with and without GERD. Specifically, we examine the impact of GERD on employment in the agricultural sector by gender and how that translates into overall national unemployment numbers. Figure 7 presents projected trends in employment for both GERD and the baseline scenarios.

In the baseline scenario, the share of female workers engaged in agricultural activity is projected to trend upwards in the next few years, with noticeable swings in the short run (Figure 7a). A similar trend is projected for the share of male employment in the agricultural sector (Figure 7b). Not surprising, the share of the labor force engaged in the agricultural sector as a whole inherits the properties of these two series.

With the implementation of GERD, both male and female employment in the agricultural sector is projected to drop drastically, with female employment suffering significantly higher losses than that of male (Figure 7a,b). As shown in Table 6, the share of female workers in the agricultural sector is expected to decline by 25.48 ± 5.32 percentage points annually under the 3-year filling scenario relative to the baseline. For male workers, the losses are projected to be 5.00 ± 0.44 percentage points below the baseline. These translate into a total loss of 18.10 ± 1.85 percentage points in the share of the labor force engaged in the agricultural sector (Figure 7c), thereby contributing to the national unemployment rate of 11.24 ± 1.77 percentage points (Figure 7d). This is also consistent with the findings in Heggy et al. [9] amid the differences in methods.

**Figure 7.** Projected trends in (**a**) female employment in the agricultural sector, (**b**) male employment in the agricultural sector, (**c**) total employment in the agricultural sector (male and female), and (**d**) the national unemployment rate under the alternative filling scenarios.

**Table 6.** Average annual change (%) in agricultural sector employment and the national unemployment rate relative to the benchmark (no GERD) scenario.


Notes: The variable "employment in agric. sector (female)" represents the share of female labor force employed in the agricultural sector. Likewise, "employment in agric. sector (male)" is the share of male labor force engaged in the agricultural sector. "employment in agric. sector (total)" represents the share of Egypt's total labor force that is employed in the agricultural sector.

#### **4. Conclusions**

In this multidisciplinary study, we investigate the environmental and economic impact of the GERD. We do so by quantifying the volume of projected losses in Egypt's annual water allocation from the Blue Nile, focusing on the 3, 7, and 10-year reservoir filling scenarios that are part of the array of scenarios currently under consideration. We then estimate the resultant losses in Egypt's agricultural land and the corresponding impact on macroeconomic variables such as food production, food import and export, employment, the cost of living, real GDP per capita and general welfare.

Given the GERD reservoir volume of 74 km3, we estimate losses in Egypt's annual water allocation be 51.29 ± 2.62%, 24.75 ± 2.76% and 18.78 ± 2.76% for the 3, 7 and 10-year filling scenarios, respectively. These translate into projected annual losses in agricultural land of 52.75 ± 2.44%, 28.14 ± 2.56%, and 22.61 ± 2.58% relative to the baseline scenario. Under the 3-year filling scenario, these losses lead to an average annual decline in food production of 38.47 ± 2.18% relative to the baseline, leading to a rise in food import by 8.56 ± 0.49% and a corresponding decline in food export of 16.50 ± 0.94%. With regards to overall agricultural sector output, the losses are projected to be 17.51 ± 0.99% annually for the 3-year filling period relative to the baseline. Moreover, the decline in the supply of food and other agricultural raw materials leads to a rise in the overall cost of living (CPI inflation) by 9.38 ± 4.38 percentage points above the baseline.

Furthermore, with GERD, both male and female employment in the agricultural sector is projected to drop drastically, with female employment suffering significantly higher losses compared to male employment. Specifically, the share of female workers engaged in the agricultural sector is expected to decline by 25.48 ± 5.32 percentage points annually below the baseline under the 3-year filling scenario. For male workers, the losses are projected to be 5.00 ± 0.44 percentage points below the baseline. These translate into a total decline of 18.10 ± 1.85 percentage points in the share of labor force engaged in the agricultural sector, thereby contributing to a rise in the national unemployment rate of 11.24 ± 1.77 percentage points. Moreover, we estimate the projected annual losses in real GDP per capita to be 8.02 ± 0.45%, 4.28 ± 0.42%, and 3.44 ± 0.41% for the 3, 7, and 10-year scenarios, respectively. These translate into annual losses in Egypt's real GDP of \$26.30 ± 2.81 billion, \$15.70 ± 3.04 billion, and \$13.40 ± 3.11 billion, respectively, leading to overall welfare losses, defined as the decline in consumption per capita, by 12.83 ± 0.73%, 6.85 ± 0.67%, and 5.50 ± 0.66%, respectively.

There are a few caveats that we would like to reiterate. First, we carried out this study under the assumption that there are no mitigating strategy in place by the government of Egypt. Therefore, these estimates represent the worst-case scenario in terms of losses generated by GERD. In subsequent studies, we plan to extend the analysis to include various mitigation strategies that are likely to be implemented by the government of Egypt. We would also like to note that, while this study focuses on Egypt, there are other downstream countries (Sudan) that would be impacted in various ways. Furthermore, Ethiopia is expected to benefit tremendously from GERD when the project takes off. It would constitute a significant source of power for the Ethiopian economy, and is expected to bring employment and business opportunities. These benefits would go a long way to improving the overall living standards in the country. The coverage of this study is therefore limited in terms of the overall impact of GERD.

We also assume that the GERD reservoir will be filled gradually under a constant filling rate (fixed water amount per year). This constitutes a departure from reality, since Ethiopia will be filling the reservoir using "phase" mechanisms. For example, in each filling phase, they would add a certain volume of water to the reservoir for which the filling rate might not be constant from year to year. Moreover, we assume full recovery in Egypt's water allocation following the end of reservoir filling. We assume the same for the subsequent loss in agricultural land. However, we acknowledge that recovery may take longer, and in the case of agricultural land, the loss in productivity may not be fully recoverable. These shortfalls likely limit the robustness of our results. In spite of this, the results provide some useful insights into the real-world consequences of GERD.

**Author Contributions:** Conceptualization, A.K. and M.A.; methodology, A.K. and M.A.; software, A.K., M.A., and A.B.; validation, A.K. and M.A.; formal analysis, A.K., M.A., and A.B.; investigation, A.K., M.A., and A.B.; resources, A.K. and M.A.; data curation, A.K., M.A., and A.B.; writing original draft preparation, A.K., M.A., and A.B.; writing—review and editing, A.K. and M.A.; visualization, A.K. and M.A.; supervision, A.K. and M.A.; project administration, A.K. and M.A.; funding acquisition, A.K. and M.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Division of Research and Innovation at Texas A&M University—Corpus Christi, grant number 282625-21010.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data is sourced from the World Development Indicator (WDI) database published by the World Bank (2020). https://databank.worldbank.org/reports.aspx?source=worlddevelopment-indicators (accessed on 3 November 2020).

**Acknowledgments:** We would like to thank the Division of Research and Innovation at Texas A&M University—Corpus Christi for providing funding through the Research Enhancement Program for this research.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A. The VAR System**

The time span for the series is from 1970 to 2019. However, some of the variable in the system have missing entries that are significant. For example, variables for agricultural sector employment including that for male, female, and combined are only available from 1991 through 2019. Likewise, the entries for unemployment rate are spotty, especially in the earlier part of the series. We therefore make adjustments by excluding these variables where we see appropriate.

**Model 1**: In order to obtain the projected values for agricultural sector output, consumption per capita, real GDP per capita, and real GDP, we use the following system of variables:

$$(variables)\equiv \begin{pmatrix} \textit{algorithm\ hand} \\ \textit{gravitational\ sector\ output} \\ \textit{food\ component} \\ \textit{food\ except} \\ \textit{consumption\ per\ capita} \\ \textit{investment\ per\ capita} \\ \textit{real\ GDP\ per\ capita} \\ \textit{CPI\ inflation} \end{pmatrix}$$

We carry out a fitness test based on the Akaike's Information Criterion (AIC), Schwarz's Bayesian Information Criterion (SBIC), and the Hannan and Quinn Information Criterion (HQIC). We set *p* = 5 based on the selection order criteria. We also test for stationarity by examining the eigenvalue stability conditions of the companion matrix. The results show that all the eigenvalues lie inside the unit circle. However, a Johansen tests for cointegration fails to support the absence of long run relationships between the series (with a reported rank = 4). We therefore estimate both the short run VAR model and a long run Vector Error Correction Model (VECM). Results generated by the VAR model are reported in Tables 2 and 4, with graphs in Figure 5 (in the main text). Predictions based on the VECM estimation are presented in Tables A1 and A2.

**Table A1.** Projected average annual growth (%) in selected variables in a baseline (no GERD) scenario over the next 3-year period based on VECM estimates.



**Table A2.** Losses/gains in agricultural sector output, real GDP per capita and consumption per capita relative to the baseline (no GERD) scenario based on VECM estimates.

**Model 2**: In projecting values for food production, food import, and food export, we substitute the variable "agricultural sector output" for the variable "food production", in which case we set *p* = 4 bases on the selection order criteria. The resultant system also satisfies stability conditions. However, like the preceding system, a Johansen test for cointegration fails to support the absence of cointegration (with a reported rank = 3). Estimating this system via VAR generates the values in Table 4 and Figure 6 (in the main text). Results of the VECM estimates are presented in Table A3.

**Table A3.** Average annual losses/gains (%) in food production, food imports, food exports, and CPI inflation relative to the baseline scenario based on VECM estimates.


**Model 3**: For agricultural sector employment variables (with a shorter time span from 1991 to 2019), the unemployment rate, and CPI inflation, we use the full set of variables, while acknowledging the limitations of the data. The resultant system is given as follows:


For this system, we set *p* = 3 based on the selection order criteria. We also confirm stationarity of this system by examining the eigenvalue stability conditions of the companion matrix. The Johansen test in this case supports the absence of cointegration. Estimating this system leads to the values that are reported in Table 6, with graphs plotted in Figure 7 (in the main text).

### **Appendix B. Elasticity**

In Section 2.3.3, we presented a framework for estimating elasticity coefficients in agricultural sector output and food production given changes in available agricultural land. We now present a framework for obtaining elasticity coefficients for the other macroeconomic variables. This is given in the general form as follows:

$$
\ln \dots \text{Y}\_t = \alpha + \beta \ln \dots \text{Agri} \text{Y}\_t + \gamma \text{Z}\_t + \varepsilon\_t \tag{A1}
$$

where *ln*−*Yt* is the log of the dependent variable of interest and *ln*−*AgriYt* is the log of agricultural sector output. *Zt* is a vector of other control variables, also in logs. The constant β is then the elasticity coefficient for the variable *Yt* with respect to changes in agricultural sector output. Table A4 summarizes the methods for estimating these coefficients, which are reported in Table 3 (in the main text).

**Table A4.** Summary of methods for obtaining elasticity coefficients for selected variables with respect to agricultural sector output.


#### **References**

