Results and Interpretation

Before engaging with the analysis, stationarity of the first differences of the logarithm of both, real GDP and the implicit GDP deflator was confirmed applying Augmented Dickey-Fuller and KPSS tests. For the estimation of the bivariate VAR models, a lag length of one was chosen for all countries. This was determined by considering the Akaike’s Information Criterion (AIC) in conjunction with a subsequent Breusch-Godfrey LM test for serial correlation, indicating a VAR order ofp= 1 in all cases. The robustness of the estimated models was validated by carrying out Johansen tests for cointegration. No meaningful cointegrating relationships were found, which is in line with expectations due to the relatively short sample period and the bivariate nature of the models.

Figure 8.2: Demand and Supply Shocks in East Africa (1997–2017)

(a)Demand Shocks (b)Supply Shocks

The underlying demand and supply shocks are graphed in Figure 8.2. The shocks vary largely in magnitude from country to country, with DR Congo having experienced the widest swings in demand shocks, while Ethiopia accounts for the widest range in supply shocks. Except for DR Congo, demand shocks appear to be relatively equally distributed with a slight tendency for positive shocks. While supply shocks show more frequent and more pronounced peaks, they seem to be more balanced between positive and negative disturbances. Generally, shocks appear to be rather volatile, which might lead to difficulties when determining coordinated policies among the countries (Buigut and Valev, 2005).

Tables 8.2 and 8.3 illustrate the correlation coefficients of the identified demand and supply shocks among the East African Community and candidate countries. As Buigut and Valev (2005) note, the more symmetric shocks are (indicated by high positive correlation coefficients),

Table 8.2: Correlations of Demand Shocks

Burundi Kenya Rwanda Tanzania Uganda DR Congo Ethiopia Sudan

Burundi 1

Kenya 0.025 1

Rwanda 0.609 0.380 1

Tanzania 0.046 0.274 0.347 1

Uganda 0.211 0.157 -0.022 0.117 1

DR Congo 0.432 0.209 0.207 0.161 0.072 1

Ethiopia 0.080 0.462 0.310 0.129 0.080 0.226 1

Sudan 0.685 0.401 0.648 0.052 0.231 0.256 0.063 1

Table 8.3: Correlations of Supply Shocks

Burundi Kenya Rwanda Tanzania Uganda DR Congo Ethiopia Sudan

Burundi 1

Kenya -0.034 1

Rwanda 0.168 -0.495 1

Tanzania 0.040 0.457 -0.290 1

Uganda 0.185 -0.184 0.154 0.239 1

DR Congo 0.519 0.345 -0.051 0.305 0.121 1

Ethiopia 0.068 0.326 -0.117 0.418 0.141 0.021 1

Sudan -0.314 0.066 0.129 0.068 -0.075 -0.048 -0.133 1

the more feasible it becomes to form a monetary union. Here, greater concern should be given to supply shocks as they are more likely to be invariant to demand management policies (Bayoumi and Eichengreen, 1993).

Overall, the countries exhibit widely asymmetric supply shocks. This holds within the EAC as Rwanda has negative coefficients with both, Kenya (-49.5 percent) and Tanzania (-29.0 percent), but also for candidate countries, especially for pairs involving Sudan. On a more positive note, many countries also show correlations of shocks between 10.0 and 25.0 percent, however, these only provide limited evidence for symmetric shocks. In addition to Kenya and Tanzania, the economic leaders of the region, exhibiting a high correlation of 45.7 percent, the two candidates DR Congo and Ethiopia also showcase highly correlated shocks with most countries in the region. Demand shock correlations generally look more promising, as only Uganda and Rwanda have a negative correlation coefficient of -2.2 percent, indicating slight asymmetry of shocks. All other countries are positively related, however, most coefficients are only marginally positive between 2.5 percent (Kenya and Burundi) and 20.9 percent (Kenya

and DR Congo). Only Burundi and Rwanda display a significant amount of positive correlation at 60.9 percent. Of the EAC candidates, Sudan exhibits the highest correlation coefficients with Burundi and Rwanda also beyond 60.0 percent, but also Kenya’s demand shocks are significantly correlated with Ethiopia and Sudan at 46.2 and 40.1 percent, respectively. Overall, it is striking that candidate countries display higher coefficients with EAC members, than EAC countries among themselves.

Figure 8.3: Correlation of Demand and Supply Shocks with Anchor Kenya

When represented graphically against Kenya as a possible anchor country as seen in Figure 8.3, the relationships become clearly visible.⁵¹ Burundi effectively shows no shock correlation with Kenya, which is less worrisome than the negatively correlated supply shocks with Rwanda and Uganda. While the symmetry of shocks with Tanzania looks promising, emphasising the economic relationship of the two countries, the strongest correlations are experienced with the candidate countries DR Congo, Ethiopia, and Sudan which are currently not considered in the East African monetary unification efforts.

The responses in real output and prices to a one standard deviation positive shock in demand and supply are shown as impulse response functions in Figures 8.4 and 8.5. Overall, the imposed restrictions for the effects of shocks on output hold for all countries; positive supply shocks lead to a level increase in output, while positive demand shocks only have temporary effects on output. A different picture is drawn for the over-identification restrictions on prices.

51Kenya would be the most obvious anchor candidate due to its predominant economic position in the region.

Figure 8.4: Impulse Response Functions for Output

(a)Burundi (b)Kenya

(c)Rwanda (d) Tanzania

(e) Uganda (f ) DR Congo

(g)Ethiopia (h)Sudan

Figure 8.5: Impulse Response Functions for Prices

(a)Burundi (b)Kenya

(c)Rwanda (d) Tanzania

(e) Uganda (f ) DR Congo

(g)Ethiopia (h)Sudan

Positive demand shocks generally result in a level increase of prices, which is in line with the AD-AS Model’s implications. In contrast, most countries respond to a positive supply shock with a level increase in prices, representing a conundrum. Bayoumi and Ostry (1997) provide a reasoning for these inconsistencies by pointing towards the generally high fluctuations and poor quality of macroeconomic data of developing countries in Africa.

Focusing on the results of individual countries, it must be noted that while the directions of responses to shocks in output are similar across all countries, the magnitudes vary significantly.

Burundi, Kenya, Rwanda, and Uganda respond to supply shocks with a long-term level increase in output of around 3.0 percent. In contrast, Tanzania is affected less at 1.7 percent, whereas DR Congo, Ethiopia, and Sudan are affected to a larger extent between 5.7 and 13.5 percent.

Large variation can also be observed in the periods of adjustment. Most countries stabilise after around six periods; however, it takes Burundi, Rwanda, and Kenya only two to three periods to stabilise, and DR Congo and Ethiopia much longer at ten to twelve periods.

In line with the implications by the AD-AS Model, demand shocks do not have long-term effects on output for all countries. However, the magnitude of short-term effects as well as the length of adjustment periods vary significantly. While Kenya, Rwanda, and DR Congo are almost not affected by demand disturbances, stabilising more or less immediately. Burundi, Tanzania, Uganda, and Ethiopia have instantaneous effects of up to 2.0 percent, requiring more time to stabilise between three and six periods. When considering the effects on output as a response to supply disruptions, EAC countries were affected more similarly than candidate countries. However, this pattern is not clearly observable for output responses to demand shocks.

As Tanzania and DR Congo are the only countries fulfilling the over-identifying restrictions for the effect of supply shocks on prices implied by the AD-AS Model, the results should be interpreted with care. Tanzania experiences a -5.4 percent level change in prices, whereas DR Congo is affected by -15.8 percent. The long-run impact for the remaining countries is rather volatile, ranging from 2.1 percent (Burundi) to 11.9 percent (Ethiopia). Adjustment periods are more similar than before, extending from six to nine years, with Burundi being the only outlier with stabilisation after only two periods. In contrast, all countries fulfil the over-identifying restrictions for the effect of demand shocks on prices, all facing level increases in the long-run.

Most long-term responses are clustered within an array of two percentage points around 10.0 percent. Exceptions are Rwanda at 15.2 percent and DR Congo at 19.2 percent. The adjustment

periods tend to be shorter than for supply shocks ranging from four to seven periods. Again, Burundi stabilises after only two periods.

When briefly comparing the results for East Africa with findings for the EMU by B¸ak and Maciejewski (2017), it must be noted that the over-identifying restriction generally holds for the EMU countries. Further, lower magnitudes of shocks and higher consistency in the duration of adjustment periods stand out, overall indicating a higher degree of integration.