To analyze the data for the study, data for gross national savings (S) and level of income (Y) were converted to their natural logs to address the problem of large values and eliminate heteroscedasticity.
Descriptive statistics
The result of the descriptive statistics shows that LN_S stood at an average of 29.1911, for the period 1990 to 2019. The maximum value of LN_S was observed at 31.1802 while the minimum value was observed at 26.3998. The difference between the maximum and the minimum values informed the range of data. The standard deviation for LN_S was 1.4219. LN_Y has a mean value of 9.6013 and a standard deviation of 1.7289 with its minimum and maximum of 6.2139 and 11.8790 respectively. Further analysis of the descriptive statistics revealed that the average value of deposit rate, financial deepening, and inflation rate were 11.8407, 16.4019, and 18.9480 respectively. Their positive values of mean showed that deposit rate, financial deepening, and inflation rate were prevalent in Nigeria. The standard deviation for DR was 5.0130, FD stood at 5.3878 while that of INFR was 17.1863. Finally, the analysis indicated that the measurement of skewness showed that LN_S, and LN_Y were found to be leftly skewed (negatively skewed) while DR, FD and INFR were positively skewed. The coefficient of the kurtosis of DR and INFR indicated that the variable were found to be peaked (3.00 and above) (Leptokurtic) relative to the normal distribution while LN_S, LN_Y and FD were found to be below 3.00 (Platykurtic). The Jarque-Bera (JB) test measures the difference of skewness and kurtosis of the series with those from the normal distribution. The JB value of 41.3182 for INFR with corresponding probability of less than or equals to 0.05 percent confirms the normality of the data.
Table 4.1: Result of descriptive statistics
| LN_S | LN_Y | DR | FD | INFR |
Mean | 29.19111 | 9.601308 | 11.84070 | 16.40196 | 18.94800 |
Median | 29.54269 | 9.885343 | 10.76276 | 13.98554 | 12.21000 |
Maximum | 31.1802 | 11.87903 | 23.99000 | 25.15527 | 76.76000 |
Minimum | 26.39989 | 6.213962 | 4.704871 | 9.151674 | 3.610000 |
Std. Dev. | 1.421938 | 1.728963 | 5.013002 | 5.387899 | 17.18635 |
Skewness | -0.518664 | -0.432603 | 0.915876 | 0.276870 | 2.188879 |
Kurtosis | 2.034225 | 1.991031 | 3.230696 | 1.455677 | 6.726910 |
Jarque-Bera | 2.510965 | 2.208252 | 4.260667 | 3.364451 | 41.31828 |
Probability | 0.284938 | 0.331501 | 0.118798 | 0.185960 | 0.000000 |
Source: E-views 11.0 statistical software |
Unit Root Test
Unit root test were conducted to ensure that the series were stationary and check the problem of having a spurious regression. For this study, test of stationarity of the variables was conducted using the augmented Dickey-Fuller (ADF) unit-root test.
Table 4.2
Augmented Dickey-Fuller – Unit Root Test
Variables | ADF at Level | ADF at First Difference | Order of integration |
| Test Stat | Prob | Test Stat | Prob | |
LN_S | -1.605812 | 0.4667 | -6.604429 | 0.0000* | I(1) |
LN_Y | -2.448133 | 0.1392 | -3.798721 | 0.0513* | I(1) |
DR | -2.586069 | 0.1072 | -5.418179 | 0.0001* | I(1) |
FD | -0.909208 | 0.7708 | -5.087876 | 0.0003* | I(1) |
INFR | -1.993941 | 0.2877 | -3.709359 | 0.0095* | I(1) |
Source: E-views 11.0 statistical software |
The Augmented Dickey Fuller (ADF) test presented in table 4 reveal that all variables are stationary at first differences. This means that the hypothesis of unit root is not rejected for all variables at the 5% level of significance in level. Hence, gross national savings and the selected independent variables are integrated of the same 1(1) order.
Co-integration Test
The results of stationarity analysis presented in the Table 4.3 show that all the modeled variables are integrated of same order. Therefore, the study then applies the Johansen cointegration tests to explore the long-run relationships among the variables. The results for Trace statistic tests are reported in Table 4.5.
Table 4.3: Johansen Co-integration Test
Series: LN_S LN_Y DR FD INFR
Lags interval (in first diference): 1 to 1
Unrestricted Co-integration Rank Test (Trace)
Hypothesized No. of CE(s) | Eigenvalue | 0.05 Critical Value | t-statistics | Prob** |
None* | 0.790962 | 105.9721 | 69.81889 | 0.0000 |
At most 1* | 0.647656 | 62.14542 | 47.85613 | 0.0013 |
At most 2* | 0.493100 | 32.93726 | 29.79707 | 0.0210 |
At most 3 | 0.279274 | 13.91287 | 15.49471 | 0.0854 |
At most 4* | 0.155822 | 4.742964 | 3.841465 | 0.0294 |
Source: E-views 11.0 statistical software |
The results for trace rank tests indicate the presence of at least four co-integration vector at 5% level of significance. This result suggests that at least four co-integration vector exists among the dependent variables (S) and all the independent variables (Y, DR, FD and INFR). Therefore, the null hypothesis which states that there is no long run equilibrium relationship between the variables is rejected.
Regression analysis
The ECM regression analysis in table 4.5 presents the regression results for determinants of domestic savings in Nigeria.
Table 4.4: ECM regression result
Dependent Variable: D(LN_S)
Variable | Coefficient | Std. Error | t-Statistic | Prob. |
C | 21.32595 | 0.540119 | 39.48380 | 0.0000 |
D(LN_Y) | 0.873426 | 0.061278 | 14.25340 | 0.0002 |
D(DR) | -0.002470 | 0.012313 | -0.200571 | 0.8427 |
D(FD) | -0.027875 | 0.014235 | -1.958123 | 0.0415 |
D(INFR) | -0.001817 | 0.002529 | -0.718589 | 0.4791 |
ECM(-1) | -1.018432 | 0.189396 | -5.377261 | 0.0000 |
R-squared | 0.984886 | Mean dependent var | 29.19111 |
Adjusted R-squared | 0.982468 | Durbin-Watson stat | 1.836013 |
| | | | |
Source: E-views 11.0 statistical software |
Discussion of findings
This study examined the determinants of domestic savings in Nigeria using annual data for a period of 30 years (1999-2019). To achieve the stated objectives of the study, the data were analysed using the descriptive statistics, unit root test, Johansen co-integration test, and ECM regression technique. The descriptive result shows that the JB values for INFR with a corresponding probability of less than or equal to 0.05 percent confirms the normality of the data. This implies that the data are suitable for generalization.
The unit root test was carried out on the data collected using Augmented Dickey-Fuller (ADF) to find out the stationarity of the time series data use and avoids spurious results. The ADF test reveals that all the variables become stationary after the first difference, due to the low p-value of the test statistics. Similarly, the cointegration analysis reveals, evidence of a long-run relationship between gross national savings (S) and all the independent variables (LN_Y, DR, FD and INFR). This finding implies that LN_Y, DR, FD, and INFR have long-term policy implications on domestic savings in Nigeria. These findings are in line with the study of Musa and Ibrahim (2014).
The Error Correction Model (ECM) was also used to tie the short-run dynamics of the co-integrating equations to their long-run statics dispositions. The ECM was properly and appropriately signed with a coefficient of -1.018432 thus indicating that about 10.18% of disequilibrium is corrected yearly by changes in savings determinants. The ECM results, further revealed that the coefficient of LN_Y was 0.8734 with a probability value of 0.0000. This signified that the level of income positively and significantly affects domestic savings in Nigeria. The result demonstrated that changes in LN_Y will over time lead to a significant increase in domestic savings in Nigeria. These findings corroborate those of Friedman (1957), Modigliani (1970), Edwards (1996), Kavatiri (2005), and Koko Morou (2007) who support the hypothesis. Further the results revealed that the coefficient of DR was -0.002 with a probability value of 0.8427 signifying that deposit rate negatively affects domestic savings in Nigeria and is statistically insignificant. This indicates that a decrease in deposit rate leads to an increase in domestic savings in Nigeria. The result is contrary to Otiwu, Okere, and Uzowuru (2018) who found that a high deposit interest rate will encourage people to save more since they will earn a higher return in the future.
Furthermore, the coefficient of FD was − 0.0278 with a probability value of 0.0415. This implies that financial deepening affects domestic savings in Nigeria negatively but was found to be statistically significant. The results conform to Pagano (1993) who states that financial deepening reduces the rate of savings but contradicts the findings of McKinnon and Shaw (1973); Athukorala and Sen (2004) who confirm a positive relationship between the financial deepening and the domestic savings.
Finally, the study deduced that inflation rate (INFR) negatively and insignificantly affects domestic savings in Nigeria. That is, a percentage change in INFR will cause a corresponding percent decrease in gross national savings in Nigeria. The results agree with Imoughele and Ismaila (2014); Nwachukwu and Egwaikhide (2007) who found that the impact of inflation on domestic savings is negative and insignificant.