• No results found

FORECASTING RESULTS

5.6 Causality tests

Since variance decomposition is an exogeneity test it implies that restrictions are placed on both the current and the lagged shocks of the explanatory variable to determine its effect on

the dependent variable. While this is important in determining the speed at which the inflation rate adjusts to changes in the considered variables, there might be need to establish how the variable’s current and lagged shocks affect inflation; thus a test of causality is carried out. The causality test helps ascertain if the inflation rate Granger causes any of the economic and financial variables or the causality flows towards inflation instead. The importance of this exercise is also to aid forecasting, since a variable that Granger causes inflation would be having important information about inflation movements, and this complements the variance decomposition results presented above as well as the findings of the cointegration test. Table 5.28 below presents the results of the causality tests on Botswana using all the variables considered for cointegration above.

Null Hypothesis: Obs F-Statistic Probability

DE does not Granger Cause DC 236 3.64115 0.01350

DC does not Granger Cause DE 0.69339 0.55693

DM does not Granger Cause DC 236 1.04579 0.37307

DC does not Granger Cause DM 1.34475 0.26062

DSA does not Granger Cause DC 236 2.65718 0.04915

DC does not Granger Cause DSA 4.28603 0.00575

DM above represents money supply; DE, exchange rate; DC, the inflation rate; DSA, South African inflation the main trading partner for Botswana.

Table 5.28: Granger causality test results for Botswana

The above findings show that the exchange rate in Botswana and the South African inflation rate Granger cause inflation, thus they have useful information about movements in inflation in Botswana. Interestingly, inflation in Botswana also Granger causes inflation in South Africa, which is surprising since the latter does not import significant goods from the former. There is no evidence that money supply has useful information about inflation movements in

Botswana. Table 5.29 presents the Granger causality results for Ghana; these show no evidence of the exchange rate having useful information about inflation movements.

Null Hypothesis: Obs F-Statistic Probability

DE does not Granger Cause DC 236 1.37890 0.24996

DC does not Granger Cause DE 0.41138 0.74498

DI does not Granger Cause DC 236 2.97433 0.03247

DC does not Granger Cause DI 1.01778 0.38554

DN does not Granger Cause DC 236 2.59673 0.05317

DC does not Granger Cause DN 0.71829 0.54197

DI above represents interest rates; DE, exchange rate; DC, the inflation rate; DN, Nigerian inflation the main trading partner for Ghana.

Table 5.29: Granger causality test results for Ghana

Interest rates and imported inflation Granger cause inflation in Ghana, with the causality non- bilateral as it moves only towards inflation. The role of interest rates, exchange rates and imported inflation on the movement of inflation is also evident in Kenya, as outlined below in Table 5.30. The F-statistic for the test of causality with a Null Hypothesis that interest rates do not Granger cause inflation is significant at 5% level of significance, which reflects the usefulness of interest rates on inflation rate movements.

Null Hypothesis: Obs F-Statistic Probability

DE does not Granger Cause DC 234 5.46756 9.1E-05

DC does not Granger Cause DE 2.20213 0.05508

DI does not Granger Cause DC 234 2.90622 0.01456

DC does not Granger Cause DI 0.78065 0.56457

DU does not Granger Cause DC 234 2.38389 0.03930

DC does not Granger Cause DU 1.54314 0.17747

DI above represents interest rates; DE, exchange rate; DC, inflation rate; DU, UK inflation, the main trading partner for Kenya.

The exchange rate results on the same hypothesis as above show a statistically significant F- statistic at 1% level of significant, and the reverse Null that inflation has useful information about movements in exchange rate is significant at 5% LOS, an indication of a bilateral causality. As in the Ghana case, there is evidence to show that imported inflation has useful information about movements in the inflation rate. The Ghana and Kenya results are not, however, reflected in the South African case in this regard.

Null Hypothesis: Obs F-Statistic Probability

DI does not Granger Cause DC 230 4.97003 4.5E-06

DC does not Granger Cause DI 2.14207 0.02740

DE does not Granger Cause DC 230 3.94107 0.00012

DC does not Granger Cause DE 0.75845 0.65495

DM does not Granger Cause DC 230 1.44880 0.16904

DC does not Granger Cause DM 0.51063 0.86599

DU does not Granger Cause DC 230 1.53220 0.13817

DC does not Granger Cause DU 0.73305 0.67833

DM above represents money supply; DE, exchange rate; DC, inflation rate; DU, German inflation, the main trading partner for South Africa.

Table 5.31: Granger causality test results for South Africa

The results of the causality test in South Africa are shown in Table 5.31 above, and the role of imported inflation can only be confirmed at 14% LOS, which is outside the standard significance levels of 1, 5 and 10%. Bilateral causality existed between interest rates and inflation in South Africa during the 20 years covered by this study. This is not surprising because South Africa practises inflation targeting and uses the interest rates to regulate inflation movements, which is a monetarist approach, as indicated in the previous chapter. Also seen in the results is the influence of inflation targeting on movements in the rate, owing to the usefulness of the exchange rate. South Africa manages the exchange rate to boost exports, and such a strategy makes imports expensive. Goods that depend on imported raw