Export Upgrading and the Extent of Structural Change in Tax Revenue in Developing Countries

Abstract

This article examines the relevance of export-upgrading strategy (export quality improvement and export diversification) in developing countries for the structural change in tax revenue (trade tax revenue versus domestic tax revenue). The empirical analysis suggests that the lower the degree of export upgrading (higher export concentration or low quality of export products) the higher the extent of structural change in tax revenue, that is, a tax transition reform. In the meantime, the effect of export upgrading on the extent of structural change in tax revenue appears to be conditioned on the degree of countries’ openness to international trade.

JEL Classification: H1, F14, O1

Keywords

Export upgrading structural change in tax revenue developing countries

Introduction

The issue of domestic revenue mobilization is more important than ever in a context of implementation of new development goals, namely, the sustainable development goals adopted by the international community.¹ The implementation of these goals is also taking place in a context of renewed interest of economic diversification in developing countries, especially poorest countries.

Trade liberalization measures undertaken by developing countries, in particular, since the creation of the World Trade Organization (WTO) in 1995, have led to a decline of their trade tax revenues and consequently their total tax revenue, as these countries exhibit high dependence on trade tax revenues for government revenue. This situation has prompted many countries to engage in tax transition reforms, that is, policies aiming at progressively substituting trade tax revenues by domestic tax revenues in order to reduce their reliance on trade tax revenue, which would likely further erode if trade liberalization were pursued. A number of researchers and scholars had called for tax transition policies as a means for developing countries to compensate for the loss of tariff revenues and stabilize the overall level of their public revenue. For example, Berg and Krueger (2003) noted that tax transition policies aim at reducing the contribution to public finance of taxes on foreign trade at the origin of major distortions. Ebrill et al. (2001) recommended that tax transition reforms be based in particular on reinforcing the indirect domestic taxation, especially value added tax (VAT). In the same vein, several researchers have documented the importance of tax transition reforms based upon VAT, tax administration reforms as well as direct tax reforms in developing countries, notably African countries (see Chambas, 2005, 2011). Many international institutions, including the International Monetary Fund (IMF), the World Bank and regional banks, have been assisting developing countries, in particular, the least developed countries, to undertake these reforms by providing them with the required technical assistance and capacity building as well as the needed financial assistance.

The purpose of this study is to examine the issue of tax revenue mobilization within the export-upgrading lens, the latter being at the heart of the trade agenda of national policymakers, development institutions and international financial institutions. In particular, the article aims to investigate the impact of export upgrading on the structural change in tax revenue in developing countries. Export upgrading entails here export diversification and export quality improvement. Structural change in tax revenue means the extent of the shift between trade tax revenues and domestic tax revenues, within the total tax revenue and is assessed through structural change indicators. As we will see later on, while these indicators do not provide the direction of the structural change but only the extent of this change, data analysis indicates that over the period considered, the evolution of these two indices reflects a tax transition in developing countries, that is, a movement from less reliance on trade tax revenue to an increasing dependence on domestic tax revenue.

Indeed, as noted earlier, tax transition reforms have been carried out in a context where economic diversification, in particular, export-upgrading strategy in developing countries, has experienced a strong comeback in both policy and academic circles (notably after the 2008 financial crisis). While there has not yet been a convergence in the academic circle or even among policymakers, on what type of export-upgrading strategy could work as an one-size-fits-all approach for all developing countries, experience of several East Asian countries (Taiwan, South Korea and later Hong Kong and Singapore) has shown that changing the structure of economy by shifting from primary production to manufacturing and, within the manufacturing sector, from natural-resource-based to more sophisticated, skill- and technology-intensive activities, is extremely important in securing sustainable growth and development. Moreover, the recent experience of the so-called emerging countries (such as Brazil, India, China and even South Africa) has shown that not only could manufacturing products be conducive to sustainable growth paths, but services could also be an important engine for economic growth.

Export concentration and low quality of export products could potentially expose a country to external shocks. In that respect, the literature has put forth several reasons for countries to diversify their exports. The structural models of economic development (see, e.g., Chenery, 1979; Syrquin, 1989) argue that countries should diversify their exports from primary products into manufactured products. This is in line with the Prebisch–Singer thesis that vertical export diversification (which takes place by moving up the value chain to produce manufactured products) reduces declining terms of trade for commodity-dependent countries. Export diversification also contributes to mitigating countries’ vulnerability to economic shocks, including the volatility and instability in export earnings (see, e.g., Bleaney & Greenaway, 2001; Samen, 2010; Strobl 2005). Furthermore, trade openness reduces economic growth volatility in countries with relatively diversified export baskets (see Haddad, Lim, Pancoro, & Saborowski, 2013).

Against this backdrop, one could question whether or not a low level of export upgrading would not lead governments of concerned countries to diversify their revenue, particularly in a context of ongoing trade liberalization with a view to addressing the adverse effects of future shocks. This question is particularly relevant for developing country members of the WTO that would not have any other choice in the near future than to engage in further trade liberalization, either in the context of bilateral and regional trade agreements (given the proliferation of the latter) or in the context of the multilateral trading system. For example, the conclusion of the Economic Partnership Agreement between the European Union (EU) and the African, Caribbean and Pacific (ACP) countries may expose these countries to further trade tax revenue losses. Similarly, the conclusion of the mega-deals (mega regional trade agreements)² currently under negotiation would induce substantial erosion of trade preferences currently enjoyed by many developing countries, including least developed countries (LDCs), with adverse consequences on their trade tax revenues.

By providing a prima facie evidence on the extent to which low level of export upgrading in developing countries matters for the structural change in their tax revenue (i.e., the progressive movement from trade tax revenue to domestic tax revenue), this study is, to the best of our knowledge, the first that examines the issue of tax revenue mobilization within the export-upgrading lens. From a policy viewpoint, this analysis would provide policymakers with another perspective on the link between the international trade sector and revenue mobilization.

The rest of the article is structured as follows: the following section provides a brief literature review on the relationship between trade and tax revenue and discusses the effect of export upgrading on structural change in tax revenue. The third section lays out the empirical model, while the fourth section provides a glimpse on the statistical correlation between export-upgrading indices and structural change indicators. The fifth section discusses the appropriate econometric methodology to estimate this model. The sixth section interprets the estimations’ results obtained and the seventh section concludes.

Brief Literature Review and Discussion on the Effect of Export Upgrading on Structural Change in Tax Revenue

We discuss here the effect of export upgrading, namely, export diversification and improvement in export quality, on structural change in tax revenue in developing countries.

Before providing this discussion, we find useful to briefly review the literature on how the international trade sector affects tax revenue, particularly in developing countries.

Trade and Tax Revenue: A Brief Literature

The relationship between trade and tax revenues has been a long-standing issue in the empirical literature on the determinants of tax revenue performance in developing countries. In particular, several studies use various measures of trade openness/trade liberalization to assess the effect of trade policy reform on total tax revenues and trade tax revenues in developing countries.

Nashashibi and Bazoni (1994) find for Sub-Saharan Africa (SSA) countries that import liberalization undermines the tax base. Ebrill, Stotsky and Gropp (1999) provide evidence that tariff reforms have not resulted in lower trade tax revenue. Adam, Bevan and Chambas (2001) use a sample of SSA countries to provide amongst others evidence that trade openness raises overall tax revenue in CFA franc countries, but has little effect in non-CFA franc countries. However, the outcome on the disaggregated revenue suggests that trade openness improves trade tax revenues and lowers goods and services tax revenue.

Khattry and Rao (2002) investigate the argument that trade liberalization depresses tax revenue (as a share of GDP) in developing countries. Their analysis involves assessing whether structural characteristics of these countries limit their ability to make the transition from trade to domestic taxes. They find empirical evidence that trade liberalization is negatively associated with total tax revenue and international trade tax revenues. More specifically, their results point out that structural characteristics of low-income and upper-middle income countries—such as trade openness, the size of population, the age–dependency ratio and the degree of urbanization—have significantly contributed to explaining the decline in tax revenue as a result of falling income and trade tax revenues.

Agbeyegbe, Stotsky and WoldeMariam (2006) obtain evidence for a set of SSA economies that the relationship between trade liberalization and tax revenue depends on the measure used as a proxy of trade liberalization. However, their results point out that in general trade liberalization is not strongly linked to aggregate tax revenue or its components—though with one measure, it is linked to higher income tax revenue.

Baunsgaard and Keen (2010) argue that given the heavy reliance of many developing and emerging market countries on trade tax revenues, further trade liberalization may be hindered unless alternative sources of revenues are found by governments. The authors therefore test whether countries were able to recoup from domestic taxes the revenues they have lost from past episodes of trade liberalization. Their empirical study indicates that this was the case for high-income countries. For middle-income countries, results suggest robust signs of strong replacement both concurrently with the revenue loss and essentially dollar-for-dollar in the long run. However, for low-income countries, although signs of significant recovery are fragile, replacement has yet been (and become) higher than suggested by previous studies, but it is not sufficiently complete in many cases to give cause for concern.

Cagé and Gadenne (2014) compare the fiscal costs of trade liberalization in developing countries and in today’s rich countries at earlier stages of development. They observe evidence that trade liberalization seems to have come at a larger fiscal costs in today’s developing countries, possibly because they decreased taxes on trade before having developed tax administrations capable of taxing domestic transactions on a large scale. Moreover, they obtain that trade liberalization episodes led to larger and longer lived decreases in total tax revenues in developing countries since the 1970s than in rich countries in the nineteenth and early twentieth centuries. The authors conclude that the fall in total tax revenues lasts more than 10 years in half the developing countries in their sample.

Discussion on the Effects of Export Upgrading on Structural Change in Tax Revenue

The literature on the structural factors that determine the tax revenue share (to GDP) in developing countries highlights the prime importance of the foreign trade sector of the economy (i.e., the degree of international trade in the economy) as a tax handle. The size of the foreign trade sector of the economy is proxied in this literature by the degree of trade openness of the economy, the latter being usually measured by the ratio of the sum of exports and imports to the GDP. Indeed, as the international trade sector is the most monetized sector in the economy (given that imports and exports take place at specified locations), this makes the collection of trade taxes easier than income taxes for developing countries and particularly LICs (Low-Income Countries). In line with this, Stotsky and WoldeMariam (1997) point out that certain features of the international trade make it more amenable to taxation than domestic activities.

An increase in the international trade share, that is, a rise in imports and/or exports, as a share of GDP will, all things being equal, likely induce a higher level of trade tax revenues and, all in all, be associated with a rise in total tax revenue share of GDP. It needs to be noted that an increase in imports could contribute to increasing domestic tax revenue, including through VAT and excise revenue.

In particular, growth in export receipts in a country could potentially be associated with higher tax share, with the latter being driven by an increase in domestic tax revenues and/or trade tax revenues.

However, the expansion of export receipts in a country could take different forms: it can reflect an increase in the volume of (active) existing product lines (also referred to as growth of exports at the intensive margin); it can also reflect a rise in exports of new product lines or exports to new destinations (new trading partners) (also referred to as export growth at the extensive margins); finally, it can be associated with an improvement in the quality of existing products (export quality upgrading). We can, therefore, question how and whether export upgrading affects the extent of structural change in tax revenue, that is, a shift from trade tax revenue to domestic tax revenue within the total tax revenue. We posit here two arguments related to the effect of export upgrading on the extent of structural change in tax revenue in developing countries.

Argument 1: Export concentration and low export quality upgrading could be associated with higher extent of structural change in tax revenue.

A higher degree of export concentration or a lower quality of export products would likely contribute to increasing countries’ exposure to external shocks, with potential negative effects on employment, export earnings and GDP volatility, and thereby induce an increase in the need of additional financial resources to address the adverse consequences of these shocks. As a result, government authorities of these countries would likely find ways for diversifying the sources of their tax revenue in order to be able to have at their disposal the required financial means/fiscal capacity to cushion the adverse effects of the shocks. Hence, the need to engage in a tax transition reform in a context where unavoidable further trade liberalization may erode the tax base (in particular trade tax revenue). In the meantime, it is arguable that export diversification or a higher export quality could make governments less incentivized to diversify their tax revenue. Similarly, export upgrading, in particular export product diversification could be associated with higher economic growth (see, e.g., Agosin, 2007; Hesse, 2008; Lederman & Maloney, 2007) as well as lower economic growth volatility (see, e.g., Haddad et al., 2013), thereby contributing to stabilizing or expanding the tax base. As a result, export upgrading may reduce governments’ incentives to engage in tax transition reform in order to diversify their sources of tax revenues towards domestic tax revenues. Overall, we expect higher export concentration or lower quality of export products to be related with the tax transition reform and particularly, higher extent of structural change in tax revenue.

Argument 2: Higher export concentration or low export quality, in particular in natural resources products, could undermine the process of structural change in tax revenue and result in lower extent of structural change in tax revenue.

This negative effect of low level of export upgrading on the extent of structural change in tax revenue would translate through the ‘Dutch disease’ phenomenon.

The Dutch disease hypothesis of resource curse highlights that access to natural resources drives up the domestic price level, crowds out the tradeable manufacturing sector and results in lower rates of productivity improvement and economic growth (see, e.g., Wiig & Kolstad, 2012). Hence, by affecting inflation with the associated appreciation effects of the real effective exchange rate,³ the ‘Dutch disease’ phenomenon could reduce domestic tax revenue (including through the Olivera–Tanzi effect [see Olivera, 1967; Tanzi, 1977, 1978]). In addition, the negative effect on manufacturing sector, which represents an important element of the tax base, would contribute to further reducing domestic tax revenue. Accordingly, concentration of export products in natural resources sector would potentially expose the country to adverse effects of the Dutch disease, and result in lower domestic tax revenue, notably through direct taxes (personal income and corporate tax revenues). This would, in turn, undermine the process of structural change in tax revenue.

With respect to export quality improvement, the scope for upgrading the quality of natural resources is lower compared to other products such as agriculture and manufacturing products. This also indicates that low export quality could undermine the process of structural change in tax revenue through the Dutch disease phenomenon.

Model Specification

The conventional literature on tax effort provides that not only is the international trade sector an important tax handle, but also other structural factors are relevant. These factors include:

Trade Openness: Given the difficulty to measure the extent of trade liberalization (reduction of tariff and non-tariff measures), the literature on the determinants of tax revenue has usually used the share of the sum of exports and imports, both in percentage of GDP as a proxy of the degree of trade openness.

Results of empirical studies on the impact of trade on tax revenue, including in developing countries are at best mixed as it depends on several factors, including the structure of liberalization and the effect of the latter on each component of tax revenue. These factors could include the extent of replacement of quantitative restrictions with tariffs, how tariff reduction affects imports, the price elasticity of demand for imports, the price elasticity of supply of import substitutes and how exports respond to the trade liberalization measures (see, e.g., Agbeyegbe et al., 2006 for a detailed discussion). Given that developing countries will not have any other choice than to further liberalize their trade, they could anticipate a loss in tax revenue, including trade tax revenue and engage in tax transition reform in order to secure in the medium to long term a more stable source of tax revenue, namely, domestic tax revenue. Overall, whether trade openness would result in a positive or negative extent of structural change in tax revenue would depend on the magnitude of trade liberalization effect on trade tax revenue (i.e., positive effect as higher imports would likely generate higher trade tax revenue) and domestic tax revenue (also positive effect through notably VAT and excise tax revenue). Nonetheless, we could expect here a positive effect of trade openness on the extent of structural change in tax revenue.

The overall level of development, proxied by the country’s real GDP per capita: As countries develop, they rely less on trade tax revenues and more on domestic tax revenue for their overall tax revenue. We could expect the rise in the level of development of a country to be associated with higher reliance on domestic tax revenue at the detriment of trade tax revenues.

Demographic characteristics (the population size): Bahl (2003, p. 13) points out that in countries experiencing faster growing populations, tax systems may lag behind in the ability to capture new taxpayers. Accordingly, a rise in the population rate is expected to be negatively associated with domestic tax revenue. At the same time, one could argue that the rise in the population could lead to higher level of imports and higher domestic consumption if the income of such a population experiences an increase. In this scenario, faster growing populations, even if making it difficult for the government to capture new taxpayers, may positively affect both trade tax revenue and domestic tax revenue. As a result, the overall effect on the structural change in tax revenue is unknown here.

In addition to these structural factors, we consider two other factors that could potentially affect the extent of structural change in developing countries: inflation and development aid flows.

Inflation: On one hand, inflation would lead to an appreciation of the real exchange rate and encourage imports, thereby generating higher trade tax revenue. On the other hand, inflation would depress domestic revenue (including through the Olivera–Tanzi effect, see Olivera, 1967; Tanzi, 1977, 1978). All in all, higher inflation would likely be associated with lower domestic tax revenue and higher trade tax revenue. As a result, it may discourage countries to engage in the change in the structure of their tax revenue. However, the crisis induced by higher levels of inflation (with the associated macroeconomic instability) could offer opportunity for governments to engage in tax transition reform.

In light of the foregoing and drawing from the empirical literature, in particular Baunsgaard and Keen (2010), we postulate the following model:

\begin{matrix} S C I_{i t} = α_{0} + α_{1} S C I_{i t - 1} + α_{2} E x p U p g r_{i t} + \\ α_{3} O p e n_{i t} + α_{4} \log {(G D P C)}_{i t} + \\ α_{5} G R P O P_{i t} + α_{6} I N F_{i t - 1} + μ_{i} + ε_{i t} \end{matrix},

(1)

where the subscript i represents the country index and t denotes the time period.

SCI is the variable representing the index of structural change in tax revenue; ExpUpgr stands for the export-upgrading variable. It could be either the variable representing the overall export product concentration index (ECI) or the overall quality of export products, denoted by ‘EXQUAL’. It is important to highlight here that due to the potential multicollinearity between these two variables, we do not include simultaneously these two variables in the model.

Open is the ratio of the sum of country’s imports and exports to GDP; NAGRI is the non-agricultural value added share of total value added; GDPCap is the real GDP per capita of country i; GRTHPOP is the rate of population growth and INF is the inflation rate (in %). Description and sources of these variables are presented in Appendix 1.

α₀, α₁, α₂, α₃, α₄, α₅, and α₆ are parameters to be estimated. μ_i are country-specific effects. The disturbance term ɛ_it is assumed to be independently and identically distributed (i.i.d.; 0, $σ_{ε}^{2}$ ).

We further examine whether the effect of export upgrading on the extent of structural change in tax revenue depends on the degree of countries’ trade openness. To do so, we estimate the following model:

\begin{matrix} S C I_{i t} = α_{0} + α_{1} S C I_{i t - 1} + α_{2} E x p U p g r_{i t} + α_{3} O p e n_{i t} + \\ α_{4} E x p U p g r_{i t} * O p e n_{i t} + α_{5} N A G R I_{i t} + \\ α_{6} \log {(G D P C)}_{i t} + α_{7} G R P O P_{i t} + α_{8} I N F_{i t - 1} + μ_{i} + ε_{i t}, \end{matrix}

(2)

where α₀, α₁, α₂, α₃, α₄, α₅, α₆ α₇ and α₈ are new parameters to be estimated. μ_i are country-specific effects. The disturbance term ɛ_it is also assumed to be independently and identically distributed (i.i.d.; 0, $σ_{ε}^{2}$ ).

Data Description and Analysis

The analysis is conducted on a sample of 76 developing countries, over the period 1975–2006. The choice of these countries and the time period is dictated by data availability. Data on tax revenue and trade tax revenue (in % of GDP) are obtained from the database developed by Cagé and Gadenne (2014)⁴ and used in their study. Domestic tax revenue (in % of GDP) has been calculated as the simple difference between total tax revenue and trade tax revenue. This allows us to subsequently compute the ratio of domestic tax revenue in total tax revenue and the ratio of trade tax revenue in total tax revenue, for the sake of description of statistics. Appendix 1 provides details on all variables used in model (1).

As noted above, the SCI variable represents the index of structural change in tax revenue (that is the shift from trade tax revenue to domestic tax revenue and vice versa). This variable is computed using Moore’s (1978) approach of structural change (the computed variable is referred to as ‘SCI’). This approach is based on the premise that the structure of any output (here the total tax revenue) can be described as a vector whose coordinates are the quantities of output. The angle between two vectors measured at different points in time is then a measure of structural change. The vectors are here the share of domestic tax revenue in total tax revenue and the share of trade tax revenue in total tax revenue. Appendix 2 provides details on the calculation of this index.

As mentioned earlier, the computation of these two indices relies upon the database developed by Cagé and Gadenne (2014). To capture medium-term dynamics in our index of structural change in tax revenue, we used five sub-periods of 5-year non-overlapping periods (1975–1979; 1980–1984; 1985–1989; 1990–1994; 1995–1999) and one sub-period of 6-year non-overlapping period (2000–2006). All explanatory variables used in models (1) and (2) are therefore computed on the basis of these non-overlapping sub-periods.

Table 1 provides descriptive statistics on the evolution of key indicators of interest to us in this study over the non-overlapping sub-periods mentioned above: the share of trade tax revenue in total tax revenue and share of domestic tax revenue in total tax revenue, the index of structural change in tax revenue (‘SCI’), the index of export concentration and the index of export quality. These descriptive statistics are displayed for the full sample of 76 countries as well as for sub-samples of LDCs (24 countries) and non-LDCs (52 countries).

Table 1.

Evolution of Key Indicators: Full Sample of 76 Countries and Sub-samples of LDCs and Non-LDCs

		Share of Trade Tax Revenue in Total Tax Revenue	Share of Domestic Tax Revenue in Total Tax Revenue	SCI	ECI	EXQUAL
		(1)	(2)	(3)	(5)	(6)
Entire Sample	1975–1979	28.2	71.8	4.4	3.91	0.78
	1980–1984	27.0	73.0	6.1	3.92	0.78
	1985–1989	25.4	74.6	5.4	3.86	0.78
	1990–1994	23.0	77.0	5.9	3.79	0.77
	1995–1999	20.5	79.5	5.6	3.76	0.77
	2000–2006	17.4	82.6	5.1	3.75	0.76
LDCs	1975–1979	45.3	54.7	4.0	4.37	0.77
	1980–1984	43.7	56.3	7.7	4.41	0.76
	1985–1989	40.3	59.7	5.8	4.42	0.74
	1990–1994	34.6	65.4	8.9	4.42	0.74
	1995–1999	29.4	70.6	8.7	4.39	0.74
	2000–2006	25.4	74.6	5.8	4.26	0.72
Non-LDCs	1975–1979	22.9	77.1	4.4	3.70	0.78
	1980–1984	20.7	79.3	5.4	3.69	0.78
	1985–1989	19.9	80.1	5.2	3.60	0.79
	1990–1994	19.1	80.9	4.4	3.50	0.79
	1995–1999	17.5	82.5	4.1	3.47	0.79
	2000–2006	14.5	85.5	4.7	3.51	0.78

Source: Author.

Columns [1] and [2] show that over the entire sample, the share of trade tax revenue in total tax revenue declines over the sub-periods considered while at the same time, the share of domestic taxes increases over these sub-periods. This tends to suggest a tax transition process. The evolution of these two ratios is confirmed over the sub-samples of LDCs and non-LDCs.

For the entire sample, the structural change index shows a rise from the sub-period 1975–1979 to the sub-period 1980–1984, but decline over the period 1985–1989. While they experience a rise from the sub-period 1984–1989 to the sub-period 1990–1994, they remain roughly stable over the two subsequent sub-periods, that is, 1995–1999 and 2000–2006.

The evolution of this index over the sub-sample of non-LDCs indicates similar patterns as the one observed over the entire sample. However, for LDCs, the trend of the evolution of the two indices seems to be same as the one observed over the entire sample and non-LDCs, with the exception that the decline in the sub-period 2000–2006 (compared to the sub-period 1995–1999) is much steeper.

In light of the foregoing, the index ‘SCI’ tends to indicate that any increase in the extent of structural change in tax revenue on countries of the entire sample could be linked with a tax transition process.

Concerning the export-upgrading indicators and irrespective of the sample considered, we note a declining trend in the ‘ECI’ over the sub-periods (see column [5] of Table A.1). However, the average score of the ‘ECI’ appears to be higher in LDCs compared to non-LDCs. These patterns indicate that while developing countries are diversifying their export products over time, LDCs exhibit higher export concentration than non-LDCs. Regarding the export quality index (QUAL), we note (see column [5]) for the entire sample that it has been roughly stable over the entire period. However, for LDCs, it has been declining while at the same time, it has been increasing in non-LDCs.

We go further by examining the correlation pattern between SCI and export upgrading by sketching in Figures 1, 2 and 3 a scatter plot of the correlation between these two indices, respectively, over the entire sample and the sub-samples of LDCs and non-LDCs. The split of the entire sample into two sub-samples reflects the discussion in the previous sections and would allow us to examine whether the structural change in tax revenue effect of the export-upgrading variables varies among these two sub-samples of countries.

The left-hand side of Figure 1 shows a positive correlation pattern between overall export concentration and the extent of structural change in tax revenue in the entire sample; in other words, overall export diversification appears to be negatively correlated with structural change in tax revenue. On the right-hand side of Figure 1, we observe that the distribution is skewed to the left, thereby suggesting that the countries under analysis exhibit a relatively low level of export quality (the value of the index is not higher than 1.05, see descriptive statistics reported in Appendix 4). Nonetheless, the fitted line indicates a positive correlation pattern between export quality improvement and the extent of structural change in tax revenue. However, this graph is left-skewed, suggesting that countries characterized by lower export quality improvement exhibit high index of structural change.

Figure 1.

Source: Author.

Figures 2 and 3 focus respectively on LDCs and non-LDCs and display the same correlation patterns as in Figure 1.

Figure 2.

Source: Author.

Figure 3.

Source: Author.

Note that descriptive statistics as well as pairwise correlation between the variables used in models (1) and (2) are respectively reported in Appendices 3 and 4. Appendix 5 presents the list of countries used in the analysis.

Estimation Strategy

The presence of the one-year lagged values of the dependent variable as a right-hand-side variable would potentially generate the Nickell bias (Nickell, 1981), which would disappear if T tends to infinity. As our T is small (T = 6), we will likely obtain inconsistent estimates if model (A.1) is estimated by means of ordinary least squares (OLS). The endogeneity generated by the presence of the lagged dependent variable as a regressor is further compounded by the possible endogeneity of other explanatory variables. To address these endogeneity problems, we use the system generalized methods of moments (henceforth denoted SGMM) developed by Blundell and Bond (1998) (see Appendix 6 for some details on this estimator).

We check the validity of the SGMM instrumentation strategy by performing two diagnostic tests: the first is the Sargan–Hansen (SH) test for over-identification and second is the Arellano–Bond (AB) procedure to test for first- and second-order serial correlation. We also report the ratio (R) of number of countries (N) to the number instruments used (INST) ( $R = \frac{N}{I N S T}$ ).

In the estimations, we consider the variables ‘GRPOP’ and the one-period lagged values of the variable ‘INF’ as exogenous. The variable ‘GDPC’ (in level) is considered as predetermined, while our variables of interest, that is, ‘Export Upgrading’ as well as the variable ‘Open’, are considered as endogenous.

The rationale for considering these two variables as endogenous goes as follows: while we could expect export upgrading to be associated with the index of structural change in tax revenue, it is also possible that the latter could exert an effect on the former. Indeed, by engaging in tax transition reform, governments would reduce tariffs and thereby mitigate the anti-export bias induced by higher tariffs on inputs needed to produce exportable goods. This would, in turn, lower production costs of firms, particularly for firms that are inclined to export internationally. Hence, tax transition reform, in particular towards an efficient and non-distortive tax system, could contribute to higher exports including export diversification.

Similarly, the reverse causality could also run from structural change in tax revenue to trade openness, thereby also raising the endogeneity issue. In fact, it could be argued that when governments undertake tax transition reform, they would likely increase their degree of openness to international trade by eliminating or at least lowering protectionist measures.

Estimation Results

Table 2 presents the results associated respectively with the estimations of specifications of models (1) and (2) by means of the two-step SGMM estimator when the export upgrading variable considered is ‘export product concentration’, while Table 3 displays the outcome of the same estimations, with the ‘export quality’ variable. At the bottom of each of these tables, we report the results of the tests on the validity of the two-step SGMM estimator, notably the ratio R of the number of countries to the number instruments, the p-value associated with the S–H statistic, as well as the p-value associated with the A–B tests (AR1 and AR2). Across all these two tables, we observe that the ratio R is always higher than 1, the p-value associated with the Sargan statistic is always higher than 0.10. Moreover, the p-values of the A–B tests are lower than 0.10 and higher than 0.10, respectively, for first- and second-order serial correlation. Overall, these results confirm the validity of the GMM procedure. It is worth highlighting that across all the three tables, the coefficient associated with the one-period lag of the dependent variable is statistically significant at 1 per cent level, suggesting the relevance of the inclusion of this variable in the model specifications.

Let us now examine the results reported in Table 2.

Table 2.

Estimation of the Effect of Export Concentration on Structural Change in Tax Revenue

Variables	Results of Model (A.1)	Results of Model (A.2)	Results of Model (A.2)
	SCI	SCI	SCI
	(1)	(2)	(3)
One-period Lag of the Dependent Variable	0.162***	0.102***	0.127***
	(0.0317)	(0.0241)	(0.0360)
ECI	1.235***	1.810***	0.869***
	(0.238)	(0.241)	(0.277)
ECI*LDC		– 0.0238
		(0.975)
LDC		0.576
		(4.653)
ECI*Open			0.00224
			(0.00297)
Open	0.0290***	0.0240***	0.0205
	(0.00445)	(0.00567)	(0.0134)
Log(GDPC)	–1.853***	–1.022**	–1.502***
	(0.355)	(0.437)	(0.286)
GRPOP	– 0.626***	– 0.494***	– 0.216*
	(0.179)	(0.110)	(0.121)
INF_t–1	–7.24e–05	– 5.54e–05	– 0.000710
	(0.00112)	(0.00115)	(0.000971)
Constant	12.25***	4.377	10.24***
	(2.904)	(3.316)	(2.711)
Observations-Countries	291–76	291–76	291–76
Number of Instruments	45	54	54
Ratio R	1.69	1.41	1.41
Sargan (p-Value)	0.6581	0.5786	0.6418
AR1 (p-Value)	0.0038	0.0063	0.0063
AR2 (p-Value)	0.2549	0.2218	0.2079

Estimator: Two-step SGMM

Source: Author.

Notes: *p-value < 0.1; **p-value < 0.05; ***p-value < 0.01. Robust standard errors are in parenthesis.

Column [1] indicates that ‘overall export concentration’ (overall export diversification) is associated with higher (lower) extent of structural change in tax revenue in developing countries, which confirms our expectations: a 1-point increase in the index of export product concentration is associated with a 1.235-point increase in the variable ‘SCI’. This tends to suggest that countries with higher index of export concentration are likely to engage in tax transition reform. Trade openness is significantly and positively associated with SCI. Regarding other control variables, results point out that structural change in tax revenue is driven by lower level of development (i.e., per capita income) and lower size of population. Inflation does not appear here to affect the extent of structural change in tax revenue.

Column [2] indicates that there is no significant difference between the effect of ‘ECI’ on ‘SCI’ in LDCs versus non-LDCs (the coefficient associated with the interaction between the variable ‘ECI’ and the ‘LDC’ dummy is not statistically significant). Results over controls are similar to those in column [1].

Column [3] suggests that the effect of export product concentration on the extent of structural change in tax revenue does not depend on the degree of trade openness: there is no significant short-term/long-term effect of export product concentration on the extent of structural change in tax revenue. Indeed, although we still find a positive and significant effect of the ‘ECI’ variable on ‘SCI’, the estimated coefficient α₄ is not statistically significant. However, this outcome does not indicate whether this non-significance of the interaction effect is valid over the entire sample. To check this, we plot confidence bands (at 95 per cent) around the marginal effect of ‘ECI’ on the ‘SCI’ variable as a function of the trade openness variable ‘Open’. Figure 4 indicates at the 95 per cent confidence intervals the evolution of the marginal effect of export product concentration on the extent of structural change in tax revenue, for varying levels of trade openness. The 95 per cent confidence intervals around the line allow us to determine the conditions under which export product concentration has a statistically significant effect on the extent of structural change in tax revenue. Such significant effect should occur whenever the upper and lower bounds of the confidence interval are either above (or below) the zero line. We could glean from Figure 4 that the index ‘ECI’ has yet a strong positive and significant, but reductive effect on the extent of structural change as long as the trade openness degree increases (the reductive effect declines as the level of trade openness diminishes). Once the degree of trade openness is above the threshold 190 per cent of GDP, the degree of export product concentration exerts no significant effect on the extent of structural change in tax revenue.

Figure 4.

Source: Author.

Results over controls are similar to those in columns [1] and [2] with the exception here of a non-significant effect of trade openness on ‘SCI’.

Let us now turn to the results reported in Table 3.

Table 3.

Estimation of the Effect of Export Quality on Structural Change in Tax Revenue

Variables	Results of Model (A.1)	Results of Model (A.2)	Results of Model (A.2)
	SCI	SCI	SCI
	(1)	(2)	(3)
One-period Lag of the Dependent Variable	0.160***	0.0978***	0.177***
	(0.0349)	(0.0332)	(0.0262)
EXQUAL	–13.12***	– 3.753	– 7.792***
	(3.274)	(2.380)	(2.932)
EXQUAL*LDC		– 24.18***
		(2.503)
LDC		21.51***
		(2.179)
EXQUAL*Open			– 0.0335***
			(0.0115)
Open	0.0378***	0.0103*	0.0596***
	(0.00356)	(0.00547)	(0.00817)
Log(GDPC)	– 2.650***	– 0.531	– 1.957***
	(0.326)	(0.449)	(0.222)
GRPOP	– 0.452**	– 0.295**	– 0.198
	(0.218)	(0.147)	(0.171)
INF_t-1	– 0.000196	0.000628	– 0.000140
	(0.00142)	(0.00116)	(0.00106)
Constant	31.60***	9.625***	22.07***
	(3.430)	(3.654)	(2.534)
Observations-Countries	291–76	291–76	291–76
Number of Instruments	45	54	54
Ratio R	1.69	1.41	1.41
Sargan (p-Value)	0.2296	0.5111	0.2900
AR1 (p-Value)	0.0051	0.0074	0.0049
AR2 (p-Value)	0.3092	0.5092	0.3141

Estimator: Two-step SGMM

Source: Author.

Notes: *p-value < 0.1; **p-value < 0.05; ***p-value < 0.01. Robust standard errors are in parenthesis.

Estimates of column [1] of Table 3 are quite interesting. They suggest a positive effect of lower degree of export quality on the extent of structural change in tax revenue. Put differently, countries experiencing a low degree of export quality tend to engage in tax transition reform with a view to diversifying their sources of tax revenue towards domestic revenue. Once again, we obtain evidence that higher openness to international trade reduces the extent of structural change in tax revenue. In other words, countries tend to engage in tax transition reform when they experience lower share of exports + imports as percentage of GDP.

Estimates associated with controls indicate that a higher extent of structural change in tax revenue (movement from higher dependence on trade tax revenue to a heavy reliance on domestic tax revenue) is driven by lower per capita income and lower population growth rate. Inflation exerts no significant effect on the extent of structural change in tax revenue.

Column [2] suggests evidence that not only are LDCs exhibiting a higher extent of structural in tax revenue than non-LDCs (the coefficient of the ‘LDC’ dummy is positive and statistically significant at 1 per cent level), but low quality of existing export products leads to higher extent of structural change in tax revenue in LDCs than in non-LDCs (the coefficient of the interaction between ‘EXQUAL’ and ‘LDC’ is negative and statistically significant at the 1 per cent level). The net effect of export product quality on the extent of structural change in tax revenue is given by –24.18. In other words, in the short term, a point decrease in the index of export quality in LDCs is associated with a 24.18-point increase in the ‘SCI’ of these countries. Regarding control variables, we obtain similar results to those in column [1], with the exception of per capita income which does not significantly affect ‘SCI’ here.

We observe from the results in column [3] that the improvement in the quality of existing products (export quality upgrading) induces lower extent of structural change in tax revenue when countries open up further their economies to international trade: the effect of export quality improvement on the extent of structural change depends upon the degree of trade openness. Specifically, the estimated coefficients α ₂ and α₄ are both negative and statistically significant at the 1 per cent level, thereby indicating that whatever the level of trade openness, lower export product quality always induces higher extent of structural change in tax revenue. This result seems to contradict with the one observed for export product concentration above (in Table 3). This is in fact an apparent contradiction, as export product diversification and export quality upgrading reflects two different realities: export product diversification reflects here either an increase in the number of export products or trading partners (extensive export diversification) or a diversification of exports among active product lines (intensive⁵ export diversification), whilst export quality improvement entails an upgrade of existing export products. Hence, the improvement of the quality of a country’s existing exports does not entail a diversification of these export products from, for example, agricultural exports towards manufacturing goods.

Figure 5.

Source: Author.

We also present here a graph depicting how the marginal effect of QUAL on the extent of structural change in tax revenue (SCI) evolves for different values of the level of trade openness in order to check whether this marginal effect is statistically significant for different values of trade openness degree. Figure 5 plots at the 95 per cent confidence intervals the evolution of the marginal effect of export quality on the extent of structural change in tax revenue for different levels of trade openness. It could be observed from this figure that the marginal effect of export product quality improvement on the extent of structural change in tax revenue is negative and statistically significant for all levels of trade openness (the value ‘0’ does not belong to any of the 95 per cent confidence intervals around the line). In addition, this figure shows that as trade openness degree increases, the reductive effect of improvement in the quality of existing export products on SCI remains almost always the same (it increases very slightly for higher values of trade openness).

Conclusion

This article’s analyses provide a different perspective on the effect of the international trade sector on government revenue in developing countries by investigating the extent to which it affects the structural change in their tax revenue (i.e., the shifts between trade tax revenue and domestic tax revenue within the total tax revenue of these countries) in these countries. First, we examine how export upgrading in developing countries affects the structural change in their tax revenue (trade tax revenue versus domestic tax revenue). Second, we investigate whether the previous effect depends on the degree of countries’ trade openness.

The analysis is conducted over a panel data set of 76 developing countries (of which 24 LDCs), over the period 1975–2006, by the use of the two-step SGMM estimator. We use a Moore’s (1978) Index of structural change to compute our indicator of structural change in tax revenue.

The empirical analysis suggests that a lower degree of export upgrading (export product diversification and export quality improvement) induces governments of developing countries to engage in tax transition reform, that is, increase in the extent of structural change in tax revenue in favour of domestic revenue. Hence, export upgrading appears to be an important driver of tax transition reform in developing countries. At the same time, the effect of export upgrading on the extent of structural change in tax revenue could be conditioned on the degree of countries’ openness to international trade: while higher export product concentration does not consistently exert a positive and significant effect on the extent of structural change in tax revenue when countries open up further to international trade, lower export quality does.

Footnotes

Acknowledgements

The author would like to thank the reviewer and the editor for their helpful comments on an earlier version of this article. This article represents the personal opinions of individual staff members and is not meant to represent the position or opinions of the WTO or its members, nor the official position of any staff members. Any errors or omissions are the fault of the author.

Appendices

Notes

References

Adam

, Bevan

, & Chambas

(2001). Exchange rate regimes and revenue performance in sub-Saharan Africa. Journal of Development Economics, 64(1), 173–213.

Agbeyegbe

T. D.

, Stotsky

, & WoldeMariam

(2006). Trade liberalization, exchange rate changes, and tax revenue in sub-Saharan Africa. Journal of Asian Economics, 17(2), 261–284.

Agosin

M. R.

(2007). Export diversification and growth in emerging economies (Working Paper No. 233). Santiago, Chile: Departamento de Economía, Universidad de Chile.

Arellano

, & Bond

(1991). Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. Review of Economic Studies, 58(2), 277–297.

Bahl

R. W.

(2003). Reaching the hardest to tax: Consequences and possibilities. Paper presented at the ‘Hard to Tax: An International Perspective’ conference, Andrew Young School of Policy Studies, Georgia State University, 15–16 May, GA, US.

Baltagi

B. H.

(2005). Econometric analysis of panel data. Chichester: John Wiley.

Baunsgaard

, & Keen

(2010). Tax revenue and (or?) trade liberalization. Journal of Public Economics, 94(9–10), 563–577.

Berg

, & Krueger

A. O.

(2003). Trade, growth, and poverty: A selective survey (IMF Working Paper No. 03/30). Washington, DC: IMF.

Bleaney

, & Greenaway

(2001). The impact of terms of trade and real exchange volatility on investment and growth in sub-Saharan Africa. Journal of Development Economics, 65(2), 491–500.

10.

Blundell

, & Bond

(1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 87(1), 115–143.

11.

Bond

(2002). Dynamic panel models: A guide to micro data methods and practice (Working Paper No. CWPO9/02). Institute for Fiscal Studies. Department of Economics, UCL, CEMMAP (Centre for Microdata Methods and Practice).

12.

Cadot

, Carrère

& Strauss-Kahn

(2011). Export diversification: What’s behind the Hump? Review of Economics and Statistics, 93(2), 590–605.

13.

Cagé

, & Gadenne

(2014). Tax revenues, development, and the fiscal cost of trade liberalization, 1792–2006 (Working Paper No. 1132). United Kingdom: University of Warwick.

14.

Chambas

(2005). TVA et transition fiscale en Afrique: les nouveaux enjeux [VAT and Tax Transition in Africa: New Stakes]. Afrique contemporaine, 3(215), 256.

15.

Chambas

(2011). Afrique au Sud du Sahara: quelle stratégie de transition fiscal [Sub-saharan Africa: What Tax Transition Strategy]? CERDI, Etudes et Documents, E 2005.01. France: Clermont Université, Université de Clermont I, CERDI-UMR CNRS.

16.

Chenery

(1979). Structural change and development policy. New York, NY: Oxford University Press.

17.

Davidson

, & MacKinnon

J. G.

(2004). Econometric theory and methods. Oxford, UK: Oxford University Press.

18.

Ebrill

, Stotsky

, & Gropp

(1999). Revenue implications of trade liberalization (Occasional Paper No. 180). Washington, DC: International Monetary Fund.

19.

Ebrill

L. P.

, Keen

, & Bodin

J. P.

(2001). The modern VAT. Washington, DC: International Monetary Fund.

20.

Haddad

, Lim

J. J.

, Pancaro

, & Saborowski

(2013). Trade openness reduces growth volatility when countries are well diversified. Canadian Journal of Economics, 46(2), 765–790.

21.

Henn

, Papageorgiou

, & Spatafora

(2013). Export quality in developing countries (IMF Working Paper No. 13/108). Washington, DC: International Monetary Fund.

22.

Henn

, Papageorgiou

, & Spatafora

(2015). Export quality in advanced and developing economies: Evidence from a new dataset (WTO Working Paper No. ERSD-2015–02). Washington, DC: WTO.

23.

Hesse

(2008). Export diversification and economic growth (Working Paper No. 21). Commission on Growth and Development.

24.

International Monetary Fund (IMF) (2014). Sustaining long-run growth and macroeconomic stability in low-income countries: the role of structural Transformation and diversification. Washington, DC: International Monetary Fund.

25.

Keen

, & Ligthart

J. E.

(2002). Coordinating tariff reduction and domestic tax reform. Journal of International Economics, 56(2), 489–507.

26.

Khattry

, & Rao

J. M.

(2002). Fiscal Faux Pas? An analysis of the revenue implications of trade liberalization. World Development, 30(8), 1431–1444.

27.

Lederman

, & Maloney

W. F.

(2007). Trade structure and growth. In Lederman

& Maloney

W. F.

(Eds), Natural resources: Neither curse nor destiny (pp. 1–12). Palo Alto, CA: Stanford University Press.

28.

Moore

J. H.

(1978). A measure of structural change in output. Review of Income and Wealth, 24(1), 105–118.

29.

Nashashibi

, & Bazoni

(1994). Exchange rate policies and fiscal performance in sub-Saharan Africa. IMF Staff Papers, 41(1), 76–122.

30.

Nickell

(1981). Biases in dynamic models with fixed effects. Econometrica, 49(6), 1417–1426.

31.

Olivera

Julio G.

H. (1967). Money, prices and fiscal lags: A note on the dynamics of inflation. Banca Nazionale del Lavoro Quarterly Review, 20(82), 258–267.

32.

Palit

(2015). Mega-RTAs and LDCs: Trade is not for the poor. Geoforum, 58, 23–26.

33.

Roodman

D. M.

(2009a). How to do xtabond2: An introduction to difference and system GMM in Stata. Stata Journal, 9(1), 86–136.

34.

Roodman

D. M.

(2009b). A note on the theme of too many instruments. Oxford Bulletin of Economic and Statistics, 71(1), 135–158.

35.

Samen

(2010). A primer on export diversification: Key concepts, theoretical underpinnings and empirical evidence. Washington, DC: World Bank.

36.

Sarafidis

, Yamagata

, & Robertson

(2009). A test of cross section dependence for a linear dynamic panel model with regressors. Journal of Econometrics, 148(2), 149–161.

37.

Stotsky

J. G.

, & Wolde-Mariam

(1997). Tax effort in sub-Saharan Africa (IMF Working Paper No. WP/97/107). Washington, DC: International Monetary Fund.

38.

Strobl

(2005). Export diversification and price uncertainty in developing countries: A portfolio theory approach. Paper presented at the UNU-WIDER Conference ‘The Future of Development Economics’, Helsinki, 18 June.

39.

Subramanian

(2007). Capital account convertibility: A neglected consideration. Peterson Institute for International Economics. Paper presented at the Centre for Policy Research, New Delhi. Retrieved 27 February 2017, from http://www.iie.com/publications/papers/subramanian04071.pdf

40.

Syrquin

(1989). Patterns of structural change. In Chenery

& Srinavasan

T. N.

(Eds), Handbook of economic development (pp. 203–274). Amsterdam: Elsevier Science Publishers.

41.

Tanzi

(1977). Inflation, lags in collection, and the real value of tax revenue. IMF Staff Papers, 24(1), 154–167.

42.

Tanzi

(1978). Inflation, real tax revenue, and the case for inflationary finance: Theory with an application to Argentina. IMF Staff Papers, 25(3), 417–451.

43.

Vikström

(2001). Long term patterns in Swedish growth and structural change: 1870–1990 (Research Memorandum GD-48). Netherlands: Groningen Growth and Development Centre, University of Groningen.

44.

Wiig

, & Kolstad

(2012). If diversification is good, why don’t countries diversify more? The political economy of diversification in resource-rich countries. Energy Policy, 40(C), 196–203.