The results are presented sequentially. We begin with the discussion on baseline results obtained by estimating a static model. After that, we move to robustness checks by discussing results of dynamic model estimated through system-GMM considering the evolution inertia effect in the dependent variable. As a second step of robustness checks, we also analyze the short and long term effects and the importance of the legal origin in the relationship between shadow economy and VCC.
3.1. Baseline findings
Table 2 presents results obtained by estimating the static specification of equation (1). The first column corresponds to the individual effect of the shadow economy on VCC, in which we control with dommy variables of the level revenue, the drgree of fragility, the size of the economy and the level of development. For this table of result, data on shadow economy are from Elgin and Oztunali (2012). The last four columns gradually incorporate a wide range of control variables to account for the traditional determinants of VCC.
Table 2: The effect of shadow economy on VCC
Dependent variable: Vulnerability to climate change
|
Technical estimation: Random Effects
|
Variables
|
(1)
|
(2)
|
(3)
|
(4)
|
(5)
|
Shadow Economy
|
0.073***
|
0.043***
|
0.045**
|
0.067***
|
0.048***
|
|
(0.017)
|
(0.016)
|
(0.018)
|
(0.018)
|
(0.018)
|
Social Readeness
|
|
-0.151***
|
-0.151***
|
|
-0.150***
|
|
|
(0.032)
|
(0.032)
|
|
(0.033)
|
Governance Readiness
|
|
|
0.007
|
|
0.015
|
|
|
|
(0.019)
|
|
(0.020)
|
Economic Readiness
|
|
|
|
|
-0.010
|
|
|
|
|
|
(0.025)
|
Global Readiness
|
|
|
|
-0.048
|
|
|
|
|
|
(0.034)
|
|
Income
|
-0.039**
|
-0.037**
|
-0.037**
|
-0.038**
|
-0.013
|
|
(0.018)
|
(0.019)
|
(0.019)
|
(0.019)
|
(0.014)
|
Fragility
|
0.006
|
0.010
|
0.011
|
0.004
|
0.024*
|
|
(0.017)
|
(0.017)
|
(0.018)
|
(0.017)
|
(0.014)
|
Country size
|
-0.001
|
-0.006
|
-0.006
|
0.001
|
-0.037*
|
|
(0.023)
|
(0.024)
|
(0.024)
|
(0.023)
|
(0.021)
|
Less advanced country
|
0.049**
|
0.052**
|
0.052**
|
0.048**
|
0.061***
|
|
(0.022)
|
(0.022)
|
(0.022)
|
(0.022)
|
(0.021)
|
Constant
|
0.537***
|
0.578***
|
0.575***
|
0.553***
|
0.526***
|
|
(0.048)
|
(0.049)
|
(0.050)
|
(0.049)
|
(0.041)
|
Observations
|
840
|
840
|
820
|
840
|
780
|
Number of countries
|
42
|
42
|
41
|
39
|
42
|
|
Hausman Test
|
0.5644
|
0.8474
|
0.5460
|
0.7943
|
0.6078
|
Note: Robust standard errors are reported in brackets. (***, **, *) indicate statistical significance at 1%, 5% and 10%.
Source: Authors
The results obtained show that, overall, the increase in informal sector activities increases the vulnerability of African economies to climate change. Referring to column 5 of the table which includes almost all the explanatory variables retained in this study, it appears that an increase in the size of the informal sector of 1% leads to an increase of about 0.048% in the vulnerability index to climate change. Negligible variations of this coefficient are observed for the different regressions performed. The effect of the informal sector on vulnerability to climate change in African countries is higher when the adaptive capacity variables of economies are omitted from the model (column 1) than when these variables are included (columns 2, 3, 4 and 5). There are two main arguments for the positive effect of the size of the informal sector on vulnerability to climate change in Africa. The economic literature shows that vulnerability to climate change is higher in countries with low adaptive capacity in social, economic, or governance terms than in countries with high adaptive capacity (Klöck and Nunn, 2019; Sarkodie and Strezov, 2019). Moreover, the adaptive capacity of economies depends on public policies to reduce the sensitivity of economies to climate change. Indeed, an increase in drought due to climate change induces instability in temperature and precipitation, which in turn reduces agricultural productivity and yields (Miguel et al., 2004; Miguel and Satyanath, 2011; Fjelde and Uexkull, 2012; Crost et al., 2018). These adverse effects that can be controlled by implementing adaptive measures in the form of investments in irrigation systems. Thus, the volume of informal economic activities through tax evasion deprives governments of the resources needed to finance public climate change adaptation policies (Dreher and Schneider, 2010).
The results in Table 2 also show that the ability of governments to be able to finance both social and economic adaptation policies reduces vulnerability to climate change in African countries. A result that is consistent with the work of Adger (2006), Füssel and Klein (2006) and Sarkodie and Strezov (2019). Furthermore, the results indicate that standard of living and level of development are important determinants of vulnerability to climate change in Africa. The negative coefficient related to the income variable shows that middle-income African countries are less vulnerable to climate change than low-income countries. This is because richer countries with greater resources are likely to be better able to control the effects of climate change than less wealthy countries. Similarly, the results show that Africa's least developed countries are simultaneously the most vulnerable to climate change.
3.2. Robustness checks
The robustness analyses are conducted in two steps. First, we evaluate the effects of potential endogeneity biases that may be due on the one hand to the presence of the lagged dependent variable among the explanatory variables (dynamic specification of the model), and on the other hand to the double causality between our variables of interest and the CCV. In a second step, we analyze the heterogeneity by taking into account the influence of the linguistic origin on the results obtained.
3.2.1 Dynamic model specification
Following the static analysis conducted in the previous section, this section is devoted to analyzing the dynamic relationship of the effect of the informal sector on CCV. This analysis is based on the one hand, on the work of Sarkodie and strezov (2019), which shows that vulnerability to climate change can be self-sustaining. On the other hand, if we refer to the economic literature, a double causal relationship can be established between climate change vulnerability and the informal economy. In this perspective, Dreher and Schneider (2010) and Nkengfack et al. (2020) show that the informal economy influences vulnerability to climate change. It can increase climate change vulnerability because it reduces the resources available to fight climate change through tax evasion. It can also reduce CCV because it promotes the reduction of greenhouse gas emissions. Furthermore, the work of Elgin and Mazhar (2013) shows that an increase in CCV increases the size of the informal sector. Indeed, greater vulnerability to climate change leads to higher environmental standards, which tends to favor the development of the informal economy. Firms in the informal sector that do not have the capacity to comply with environmental standards tend to migrate from the formal to the informal sector.
To address this issue and correct the results for potential endogeneity biases, we estimate a dynamic panel model using the generalized method of moments. We use alternatively the informal sector index of Medina et Schneider (2018) and that of Elgin and Otzunali (2012) in an effort to ensure that the choice of measure of the informal economy does not significantly affect the results.
Table 3: Correcting endogeneity bias
Dependent variavble: Vulnerability to climate change
|
Variables
|
GMM
|
RE
|
GMM
|
VCC (-1)
|
1.061***
|
|
1.020***
|
|
(0.027)
|
|
(0.028)
|
Shadow Economy (MS)
|
0.018***
|
|
|
|
(0.006)
|
|
|
Shadow Economy (EO)
|
|
0.033**
|
0.035***
|
|
|
(0.015)
|
(0.011)
|
Social Readeness
|
0.007
|
-0.150***
|
0.017
|
|
(0.011)
|
(0.045)
|
(0.020)
|
Governance Readiness
|
-0.013
|
-0.000
|
0.009
|
|
(0.013)
|
(0.026)
|
(0.014)
|
Economic Readiness
|
0.024***
|
-0.000
|
0.024**
|
|
(0.008)
|
(0.045)
|
(0.010)
|
Income
|
0.002
|
-0.012
|
-0.014**
|
|
(0.002)
|
(0.014)
|
(0.006)
|
Fragility
|
0.001
|
0.025*
|
-0.003*
|
|
(0.002)
|
(0.013)
|
(0.002)
|
Country size
|
-0.000
|
-0.041*
|
0.009**
|
|
(0.002)
|
(0.022)
|
(0.004)
|
Less advanced country
|
-0.003**
|
0.064***
|
-0.011***
|
|
(0.001)
|
(0.020)
|
(0.004)
|
Constant
|
-0.050**
|
0.539***
|
-0.013
|
|
(0.018)
|
(0.040)
|
(0.020)
|
Observations
|
144
|
204
|
105
|
AR(1)
|
0.016
|
|
0.014
|
AR(2)
|
0.233
|
|
0.978
|
Hausman Test
|
|
0.6143
|
|
Hansen Test
|
0.449
|
|
0.416
|
Note: Robust standard errors are reported in brackets. (***, **, *) indicate statistical significance at 1%, 5% and 10%. MS and EO correspond to the measure of the shadow economy respectively from Medina and Schneider (2020) and Elgin and Öztunalı (2012).
Source: Authors
Table 3 shows that the results remain unchanged regarding the sign and significance of the estimated coefficient of the variable of interest. Table 3 shows that the effect of the informal sector on vulnerability to climate change remains positive and negative regardless of the indicator used. Indeed, the fundamental difference between the two approaches to measuring the informal economy lies in the fact that, the index of Medina and Schneider (2018) is obtained by a macroeconomic approach while that of d'Elgin and Öztunali (2012) is obtained by a general equilibrium microeconomic approach.
To complete this analysis, we then estimate our model using the Pooled Mean Group method. This method has the advantage of taking into account cross-sectional dependence effects and also allows us to distinguish between short and long term effects (Pesaran, 2007). Indeed, the lack of theoretical consensus on the relationship between the informal economy and CCV could reflect the fact that this relationship is non-linear. The short-run effect is therefore potentially different from the long-run effect. The results obtained are presented in Table 4.
Table 4: Short term and long term effect
Dependent variable: Vulnerability to climate change
|
Technical estimation: Pooled Mean Group
|
Variables
|
(1)
|
(2)
|
ECT
|
-0.225***
|
-0.184***
|
|
(0.032)
|
(0.021)
|
Long term effect
|
|
|
Shadow Economy (MS)
|
|
0.068***
|
|
|
(0.020)
|
Shadow Economy (EO)
|
0.060**
|
|
|
(0.027)
|
|
Short term effect
|
|
|
D. (Shadow Economy MS)
|
-0.034**
|
|
|
(0.016)
|
|
D. (Shadow Economy EO)
|
|
0.008
|
|
|
(0.006)
|
D. (Social Readiness)
|
0.018
|
0.000
|
|
(0.057)
|
(0.041)
|
D. (Gvernance Readeness)
|
0.005
|
0.006
|
|
(0.014)
|
(0.011)
|
D. (Economic Readiness)
|
0.037
|
0.012
|
|
(0.054)
|
(0.013)
|
Constant
|
0.127***
|
0.100***
|
|
(0.020)
|
(0.012)
|
Control variable
|
Yes
|
yes
|
Country characteristics
|
No
|
no
|
Note: Robust standard errors are reported in brackets. (***, **, *) indicate statistical significance at 1%, 5% and 10%. MS and EO correspond to the measure of the shadow economy respectively from Medina and Schneider (2020) and Elgin and Öztunalı (2012).
Source: Authors
Table 4 shows that, in the long run, the size of the informal sector positively and significantly affects CCV in Africa. Indeed, we have, in line with the work of Dreher et Schneider (2010)We have shown that the informal sector reduces the tax base of governments and therefore their stock of resources available to finance climate change adaptation policies. On the other hand, the results obtained indicate that the short-term effect of the informal sector on CCV is negative and significant according to the indicator of Medina and Schneider (2020) while that of Elgin and Otzunalii (2012) is insignificant.
In the short term, the informal sector has a direct effect on greenhouse gas emissions. As businesses in this sector are small, they contribute less to CO2 emissions than businesses in the formal sector. An increase in the size of the informal sector can therefore reduce the overall level of exposure of economies to climate change. However, this effect is either not sustainable in the long run or remains very small as many policies are implemented to limit the expansion of the informal sector. As a result, the effect of tax evasion is greater than the effect of pollution reduction, which justifies the positive long-term effect.
3.2.2. Analysis of the effect of official languages
This subsection is inspired by the work of Lata and Nunn (2012) who show that cultural factors, including language and religion, can explain international divergence in CCV levels. These authors show that language, particularly local languages, distorts individuals' conceptions of climate change, making it difficult to implement adaptation measures. In addition, the literature indicates that countries with a British legal origin have better economic structures than countries with a French, Portuguese or other legal origin.
We analyze the effect of official languages on the relationship between the informal sector and CCV. To do so, we introduce a legal origin variable with three modalities: French, English and other languages. In this last modality we include countries with Portuguese, Spanish, Arabic or a local language as their official language. Estimates are made for each of the categories and the results obtained are presented in Table 5.
These results show that the informal sector increases the vulnerability to climate change of African countries regardless of their legal origin. Indeed, for each of the linguistic modalities, the estimated coefficients of the different informal sector indices remain positive. However, it should be noted that the effect is insignificant in English-speaking countries, whereas it is significant in French-, Portuguese- and Arabic-speaking countries. This is initially in line with the work of Lata and Nunn (2012). This result can be explained by the fact that economic structures are more developed in English-speaking countries than in French-speaking countries, allowing the latter to be more resilient to climate shocks. In addition, the tax burden in these countries tends to be lower than in French-speaking countries, thus reducing the incentive for economic agents to engage in tax evasion.
Table 5: The effect of official languages
Dependent variable: Vulnerability to climate change
|
Technical estimation: Generalized Least Squares
|
|
French
|
English
|
Others
|
French
|
English
|
Others
|
Variables
|
(1)
|
(2)
|
(3)
|
(4)
|
(5)
|
(6)
|
|
|
|
|
|
|
|
Shadow Economy (MS)
|
0.080***
|
0.042
|
0.093**
|
|
|
|
|
(0.027)
|
(0.032)
|
(0.040)
|
|
|
|
Shadow Economy (EO)
|
|
|
|
0.090**
|
0.008
|
0.040***
|
|
|
|
|
(0.040)
|
(0.049)
|
(0.009)
|
Global Readiness
|
-0.044
|
-0.031
|
-0.042
|
-0.0007
|
-0.101
|
-0.187***
|
|
(0.033)
|
(0.093)
|
(0.098)
|
(0.069)
|
(0.126)
|
(0.052)
|
Income
|
0.193**
|
0.170***
|
-0.025**
|
0.188***
|
0.185***
|
-0.043***
|
|
(0.010)
|
(0.014)
|
(0.012)
|
(0.008)
|
(0.017)
|
(0.006)
|
Fragility
|
0.112***
|
0.169***
|
-0.006***
|
0.111***
|
0.183***
|
-0.004***
|
|
(0.001)
|
(0.016)
|
(0.0004)
|
(0.003)
|
(0.022)
|
(0.001)
|
Country size
|
-0.175***
|
-0.235***
|
0.091***
|
-0.166***
|
-0.249***
|
0.077***
|
|
(0.002)
|
(0.011)
|
(0.006)
|
(0.007)
|
(0.021)
|
(0.002)
|
Less advanced country
|
0.136***
|
0.405***
|
0.119***
|
0.133***
|
0.428***
|
0.102***
|
|
(0.002)
|
(0.024)
|
(0.016)
|
(0.003)
|
(0.029)
|
(0.011)
|
Constant
|
|
|
0.421***
|
|
|
0.540***
|
|
|
|
(0.079)
|
|
|
(0.032)
|
Observations
|
360
|
320
|
160
|
247
|
209
|
117
|
Sargan test
|
0.22
|
0.43
|
0.12
|
0.25
|
0.44
|
0.12
|
Number of countries
|
18
|
16
|
8
|
19
|
16
|
9
|
Note: Robust standard errors are reported in brackets. (***, **, *) indicate statistical significance at 1%, 5% and 10%.
Source: Authors