4.0 Data Analysis and Results Discussion

Table

1: Augmented Dickey-Fuller (ADF) test for Unit root

Variable

Level

First Diff.

Second Diff

ADF Critical values

test stat.

P-Value

test stat.

P-Value

test stat.

P-Value

1%

5%

10%

real GDP

-1.9271

0.6034

-3.0836

0.1377

-4.8618

0.006*

-4.49

-3.65

-3.26

Investment

-4.9801

0.003*

-8.1164

0.000*

-10.8814

0.000*

Export

-2.4299

0.3526

-5.0078

0.004*

-7.0237

0.000*

Tourism

-3.1441

0.1281

-2.5864

0.2891

-4.9556

0.004*

Note:

* indicates significance at 5% level

Presented

in table 1 are results of the ADF test which are obtained from the estimation

of equation 3.1. As it can be observed from the table, apart from Investment,

all other variables were not stationary at their level values. This mean that

they contain a unit root since their respective test statistic values are less

than their ADF critical values at 5% level.

This suggests that the variables must be differenced. After first

differencing, only Export, and Investment were stationary, but real GDP and Tourism

were not. Thus, there is need to take the second difference for all the

variables. At second, difference, all the variables become stationary, and

hence, there is no evidence of a unit root in all the variables after second

differencing. Therefore, all the variables shall be considered to be stationary

at second difference, and hence, be integrated of the order (2). This is

required because the Johansen test of cointegration demands all the variables

to be of the same order.

As

evident from the above, Ordinary Least Squares (OLS) method cannot be applied

directly in presence of a unit root. This will then require some adjustments be

done. However, pre-estimation tests proved the presence of autocorrelation in all

the variables. To mitigate the issue, the Generalized Least squares (GLS) was

applied as explained by Gujarati (2016). After application of the GLS method,

the regression results obtained are tabulated in table 2.

Table

2: Regression Results

Variable

Coefficient

Std. Error

t-Statistic

Prob.

Tour

0.400208

0.064009

6.252339

0.0000

Expo

0.633826

0.033902

18.69607

0.0000

Inves

-0.011240

0.016829

-0.667896

0.5132

R-squared

0.821333

Adjusted R-squared

0.800314

This

work seeks to investigate the impact of tourism on the economic growth of

Cameroon. The above estimated results were obtained from the estimation of

equation 3.2. The results are presented in the form of an equation as shown by

equation 4.1

. . . 4.1

Equation 4.1 provides some interesting

outcomes. At the onset, except for investment, the other two variables (tourism

and export) have correct expected sign (i.e. positively related with real GDP).

Starting with the coefficient of tourism which is 0.40. this indicates that,

keeping the other variables constant (i.e. investment and exports), an increase

in the number of tourists (by 1%) will, on the average increase the real GDP of

Cameroon by 0.40%. This relationship is statistically significant at even 1%

level as shown by the probability value.

Second, the coefficient of exports is

0.63, implying that if all the other variables (i.e. tourism and investment)

are to be kept constant, an increase in exports of goods and services in Cameroon

by 1%, will on the average, increase Cameroonian real GDP by 0.63%, and this

estimation is statistically significant at 5% level. But observe that the

coefficient of export is greater than the coefficient of tourism (i.e. 0.63

> 0.40). This clearly discloses that in Cameroon, exports have more

significant influence on real GDP than tourism.

Third, the coefficient of investment is –

0.01. this figure implies that, keeping other variables (tourism and exports),

an increase in the flow of foreign direct investment into the Cameroonian

economy by 1% will, on the average, decrease the real GDP of Cameroon by 0.01%.

But this estimation is statistically significant as probability values show.

However, it was not this work’s prior expectation that the coefficient of

investment has a negative sign. This is because an increase in the flow of

investment is always expected to have positive significant impact on the real

GDP of not only Cameroon, but other countries as well. Therefore, the negative

sign seen on the coefficient of investment is highly misleading.

From table 2, the value of 0.8213 has been

reported for R-squared, and 0.8003 for the adjusted R-squared. The value of

R-squared imply that about 82.13% of the total variation in the real GDP of

Cameroon is explained by the joint influence of the three variables in the

model. And, as for adjusted R-squared, it implies that after adjusting for

other variables that influence real GDP of Cameroon which are not included in

the model, the three explanatory variables still explain about 80% of the

variations in the real GDP of Cameroon.

The post-estimation of the results in

table 2 reveal that the model is free from serial correlation (using Breusch-Godfrey Serial Correlation LM Test) and

heteroscedasticity (using Breusch-Pagan-Godfrey

test)

problems. Using the cumulative sum (CUSUM) test, the model is found adequate,

and the residuals were also normally distributed as shown by the results of

normality test. Details of these tests can be found in the appendix.

However, to investigate whether there is

long-run relationship among the variables of this study, the Johansen test of

cointegration was conducted and the results obtained are tabulated in table 3.

Table

3: Johansen test of cointegration

Hypothesized

No. of CE(s)

Eigenvalue

Trace

Statistic

0.05

Critical Value

Prob.**

None *

0.944435

98.29087

47.85613

0.0000

At most 1 *

0.806576

49.15756

29.79707

0.0001

At most 2 *

0.608985

21.22874

15.49471

0.0061

At most 3 *

0.266362

5.265575

3.841466

0.0217

Trace test

indicates 4 cointegrating eqn(s) at the 0.05 level

* denotes

rejection of the hypothesis at the 0.05 level

Results

of table 3 estimate the long-run cointegrating vectors for all the variables

using the Trace statistics. The decision rule is stated: Reject the null

hypothesis if the value of Trace statistics is greater than its critical values

at 5%. The decision here, with regard to results of table 3, is to reject the

respective null hypotheses at 5% level. This is because in all the hypothesized

number of cointegrating equations, respective values of the test statistics are

all greater than their corresponding critical values at 5% level. Therefore,

the all null hypotheses stating that there is no cointegrating equations will

be rejected. Rejection of these null hypotheses provide some evidences that

there is cointegration among the variables (real GDP, tourism, exports, and

investment). With this outcome, it thus clear that in the Cameroonian economy,

there is long-run relationship among all the variables controlled in this

study.

4.1 Result Discussion

One

principal factor to emphasize here is the coefficients of tourism, export, and

investments reported in equation 4.1 (or table 2). Because the variables were

initially converted into natural logs prior to the estimation, the resulting

coefficients are therefore known as elasticity (Gujarati, 2016). Impliedly,

therefore equation 4.1 says that the elasticity of impact of tourism, exports,

and investment on the real GDP of Cameroon are 0.40, 0.63, and – 0.01

respectively. According to Wooldridge (2013) the resultant elasticities can

also be interpreted as the multiplier effect. Thus, 0.40 is multiplier effect

of tourism on real GDP of Cameroon. 0.63 is the exports multiplier effect on

real GDP, while – 0.01 is the investment multiplier effect. This negative value

reported on investment is rather absurd and misleading as already explained.

The possible cause of this could be inconsistency in the way the variable is

measured (i.e. foreign direct investment as a percentage of GDP). Because

frequent fluctuations in the GDP level will also mean the same fluctuation in

the level of investment.

However, the percentage reported on the

R-squared and adjusted R-squared is unreasonably high, because in real world,

thousands of variables (like infrastructure, government policy, imports,

exchange rate, taxes, etc.) work together to influence the economic performance

of an economy not just those used in this work.

Also, the results of Johansen

cointegration test are in line with prior expectation because in a developing

economy like Cameroon, one would expect at least export, and investment to have

long-run relationship with real GDP. Nonetheless, the results also are in line

with the finding of Khalil, Khan and Waliullah (2007), Saayman & Saayman, (2012), Makochekanwa

(2013), Šimundi?, Kuliš, and Šeri? (2016) and Bezuidenhout & Grater, (2016).

From the results of table 2, we arrive at

the first objective of this research. Thus, Tourism has significant impact of about

0.4% on real of Cameroon. And from the results of table 3, we achieve the

second objective of this work. Therefore, there is long-run relationship

between tourism and real GDP Cameroon. Overall, the empirical results of this

work are most likely in support of the economic theory reviewed in the

literature review section.

Finally, there are certain limitations

that worth emphasizing from the results of this work. One, there is scanty of

data on the variables used in this research, especially on tourism. Thus,

because the data is small, the results of this work may not be robust. Two, the

econometric models used in this work need further extension. For example, as

the Johansen test of cointegration indicates the presence of long-run

relationship among the study variables, there is need to go further to identify

how can this long-run relationship be established and at what speed. This can

be done using the vector error correction model (VECM). Finally, there is the

need to go beyond the VECM to find out what will be response of Cameroon real

GDP to a one standard deviation shock in investment, tourism, or investment.

This can be done using the impulse response function, and/or variance

decomposition tests.

5.0 Summary and Conclusion

5.1 Summary

In this work, effort has been made to

investigate the impact of tourism on economic growth of Cameroon, while

controlling for other variables. Both theoretical and empirical literature were

reviewed on the impact of tourism on economic growth of countries. The

methodology used in the conduct of this work involves the application of the

Ordinary Least Squares (OLS), and the Johansen test of cointegration. First,

the times series behaviors of the variables were examined using the ADF test

for unit root. All the variables were considered to be stationary at second

difference. And with the detection of serial correlation problem, the GLS

method was developed before finally running the OLS regression. From the

regression results, it has reported in this work that tourism and exports have

significant influence on the real GDP of Cameroon, while investment has not.

From the result of Johansen test of cointegration, this work identifies

evidence of long-run relationship among the controlled variables in this work.

5.2

Conclusion

From the empirical analysis of this work,

this research draws two conclusions: Tourism has significant impact on real of

Cameroon. And, there is long-run relationship between tourism and real GDP

Cameroon. However, the overall empirical results of this work are most likely

in support of the economic theory reviewed in the literature review section.

Policy

implication

Based

on the conclusion drawn, it can be suggested here that since tourism has

significant impact on the economy of Cameroon, the Cameroonian economic policy

makers should try to make and implement policies that promote tourism.

Furthermore, the policy makers should also make sure that policies that

strengthen international trade and promote exports are at the fore front of their

agenda because, exports have significant positive impact on the economy of

Cameroon.