Canada-India Joint Study Group Report: Exploring the Feasibility of a Comprehensive Economic Partnership Agreement

Chapter 6

Economic Modelling

The economic impacts of the CEPA between Canada and India can be empirically examined with a computable general equilibrium (CGE) model. In this Chapter, we use the Global Trade Analysis Project (GTAP) model for this purpose.

The modelling results should be considered in the context of both the advantages and limitations of the model, and of CGE models in general. The GTAP model integrates data on bilateral trade flows, trade protection and domestic support together with national input-output tables that describe the sale and purchase relationships between producers and consumers for 113 economies/regions. This allows the model to generate estimates of the impact of trade policy changes, such as preferential tariff elimination under free trade agreements, on trade flows, the level of national economic output and employment. However, the GTAP model can reflect only the expansion of trade in products already traded in the bilateral relationship, and cannot predict the creation of trade in new product areas, which is particularly important when the existing trade relationship is fairly narrow, as is the case between Canada and India. Further, increases in trade do not drive gains in productivity in the GTAP model, although it is possible to introduce a productivity increase in the model. Finally, the GTAP model only allows for analysis of gains from liberalisation in goods and services trade, and does not include gains from liberalisation and enhanced economic cooperation in other areas (investment being a key example).

Table 6.1: Summary of Simulation Results for Canada-India CEPA²⁶

Sources: Canadian simulation results from DFAIT, Indian simulation results from RIS.

Therefore, the results presented in this chapter should be seen as complementing the qualitative analysis of gains from a CEPA that are presented elsewhere in this study, and should also be considered as likely to underestimate the benefits to be had from a CEPA.

Simulations using the GTAP are carried out by both Canada and India. The simulations cover a wide range of liberalisation issues such as trade in goods and services and trade facilitation, and they examine the implications of productivity gains and increases in the supply of labour and capital, and their consequent impact on the economies of the contracting parties. Estimates of GDP gains range from US$6-15 billion for Canada and US$6-12 billion for India. These results show that the potential gains from liberalising trade between Canada and India are substantial. Further, the gains are fairly symmetric. Bilateral trade increases significantly with export gains for Canada ranging between 39% to 47% and for India, between 32% to 61%.

6.1 Technical Background on the Modelling Framework

The GTAP model version 6.2 is chosen for this study. This is a static version of the GTAP model family with an assumed market structure of perfect competition. On the production side, the model assumes that labour and capital are fully employed, mobile across all uses within a country but immobile internationally. On the demand side, there is a regional representative household whose expenditure is allocated across private consumption, government spending, and saving. Bilateral international trade flows are modelled based on the Armington hypothesis that goods and services are differentiated by region of origin and are imperfect substitutes.

In performing simulations, the modeller must make some choices with regard to which variables in the model are to be exogenous (i.e., fixed at predetermined values specified by the modeller) and which are to be endogenous (i.e., the values for which are solved by the model). Under the GTAP's default microeconomic closure, the supply of labour, capital, land and resources are fixed; factor prices (i.e. wages, and return to capital and land) adjust to restore full employment of the factors of production in the post shock equilibrium. Under alternative microeconomic closures that are sometimes used, the return to capital or to labour can be fixed and the supply of capital and labour then adjusts to restore equilibrium.

The economics of the welfare effects of tariff eliminations are modelled as allocative efficiency gains stemming from reallocation of productive resources across sectors. By contrast, modelling of regulatory barriers such as barriers to trade in services is less straightforward. Regulatory barriers require the use of real resources to meet regulatory requirements. They are not a necessary part of the production and delivery process, but rather are an added or discretionary burden of costs added by government. A reduction of regulatory barriers represents a saving of real resources. In general, the reduction of regulatory barriers would result in greater gains from liberalisation compared to tariff reductions on the price of traded products. This reflects the fact that tariffs represent a transfer of income. They contribute to national income in the country or region collecting the tariffs and, therefore, expand economic welfare commensurately with the reduction in economic welfare in the country/region paying the tariffs.

6.2. Canada's Approach

6.2.1 Framework for the Simulation

This section describes updates to the tariff baseline in the GTAP database, the assumptions underlying the simulation with respect to services liberalisation, and the closure rule adopted for the simulation.

Tariff updating

The database for the model is based on the GTAP database version 7, which is benchmarked to 2004. This version of the GTAP database includes the latest Canadian IO table, which has been updated to 2003 from 1990 in the previous version of the GTAP database.

The tariff data need to be updated to reflect the latest levels of protection in both countries. Because of unilateral liberalisation undertaken in each country since 2004, using the 2004 tariff data might inflate the scope of gains from bilateral trade liberalisation between Canada and India.

Consistent with GTAP methodology for trade-weighted tariff averaging, the approach to update the Canadian and Indian tariffs to the 2008 level is described as follows:

1. We use the 2008 average Ad Valorem Canadian and Indian applied MFN duty data at the HS-6 level obtained from the WTO Integrated Trade Database²⁷. We use the 2006-08 three years' average Canadian and India imports to calculate the trade weights using the import data from the World Trade Atlas²⁸. We grouped the tariff and import data at the HS-6 level to the appropriate GTAP category according to the HS-GTAP concordance obtained from the GTAP.

2. For certain HS-6 categories for which there is a combination of ad valorem, mixed and special tariffs together, we chose the highest tariff in order to capture the full potential of trade liberalisation.

Modelling services trade liberalisation

While a comprehensive dataset is available for tariff barriers for goods trade, a similar database on the barriers to services trade does not exist. Moreover, given the fact that barriers to services trade typically are part of broader regulatory frameworks governing the delivery of services in a given jurisdiction, the extent to which liberalising measures can reduce the trade-inhibiting aspects of these regulations is at the present time effectively impossible to measure directly. Therefore, both the height of barriers to services trade and the extent to which their reduction is feasible must be inferred indirectly through econometric analysis.

The height of barriers to trade in services in this study is taken from the gravity model estimation of services imports by J. Francois et al²⁹. Regression analysis measures possible regulatory barriers that impede services trade across all countries. The estimation results are then applied to the country level to give rise to trade costs benchmarked against Hong Kong and Singapore (which are considered to be the closest to free trade countries). In Francois' calculation, the measured trade cost for all service sectors in Canada is 20.18% higher than Hong Kong and Singapore, while the corresponding figure for India is 18.03%. In other words, if trade in services in Canada and India were as free as Hong Kong and Singapore, trade costs in services in Canada and India would be 20.18% and 18.03% lower, respectively. Based on these estimates, we would expect equivalent direct efficiency gains in bilateral trade in services, or saving in real resources for both Canada and India.

The closure rule

Under the GTAP model's default microeconomic closure, the total supply of capital and labour are fixed. Factor prices adjust to restore full employment of the factors of production in the post shock equilibrium. In other words, the model assumes that there would be no extra capital and labour available in either country to meet the new demand stemming from trade liberalisation between Canada and India. Additional capital and labour have to come at the expenses of the existing production capacity that is used in the production for domestic consumers or consumers in third countries. The competing demand for labour and capital will bid up wages and returns to capital, resulting in strong trade diversion and terms of trade effects.

This is clearly not a very realistic presentation of the economy for the present case. In today's highly integrated global financial market, it is unlikely that a rise in the rate of return to capital would not be met with higher investment, even if this were not financed through increased domestic savings. Similarly, given the relatively small size of bilateral trade between Canada and India, as return to labour increases, both countries should have no difficulty to find extra labour to meet new demand. The extra labour could come as a result of population growth, increasing labour participation, or migration from rural areas to the industrial sector. Given the above considerations, in the following simulation, we choose the closure rule in which the supply of labour and of capital are endogenous, and their supply responds to higher returns due to freer trade, bringing factor prices back to their original level in equilibrium.

6.2.2 The Simulation results

Policy scenarios used in the simulation

The quantitative estimates obtained in this study are based on a simulation that involves:

1. full elimination of goods trade protection as captured in the GTAP database, updated to 2008 as described above, for all industrial and agricultural sectors (including elimination of all tariffs and tariff-rate quotas). Notably, no exception is made for "sensitive sectors", notwithstanding that trade and investment liberalisation initiatives often contain provisions that exempt certain such sectors from liberalisation or circumscribe the applicable extent of liberalisation;

2. a reduction of trade costs in Canada-India cross-border services trade (through the GTAP function for augmenting technological change through imports from the partner country) by an amount equivalent to that estimated by Francois et al as what would be required to make services trade as liberal as it is in Hong Kong and Singapore (which are considered to be the closest to free trade countries in services).

Impacts on the level of economic activity

Table 6.2 presents the results. As a percent of GDP, India's gains at 0.51% are slightly higher than Canada at 0.41%. In absolute terms, gains are essentially equivalent, about US$6.2 billion for Canada and US$6.1 billion for India.

Table 6.2 Canada's simulation results - Implications for GDP

Source: DFAIT Office of the Chief Economist estimates based on the GTAP model v.6.2

Trade impacts

The simulation results show that two-way bilateral trade as measured by exports of goods and services could potentially expand by over 50%, or nearly US$3 billion. While services trade expands by more than goods trade, its absolute contribution to gains is smaller, given its smaller base.

Table 6.3 Canada's simulation results - Implications for exports

Source: DFAIT Office of the Chief Economist estimates based on the GTAP model v. 6.2

Despite the overall higher level of tariff protection in India, and the higher initial level of Canadian goods exports to India, India's bilateral export gains in goods are similar in magnitude to Canada's, around US$1.2 billion. In percentage terms, India's bilateral export gains in goods are greater than that for Canada, an increase of 59%, compared to 45% for Canada.

In services trade, India's bilateral cross-border exports in services expand by US$280 million (or 67%) and Canada's by US$193 million (or 60%).

Table 6.4 details the individual bilateral sectoral trade impacts resulting from liberalisation of trade between Canada and India. India's leading export gains are concentrated in manufactured products and services. Among manufacturing sectors, the sector that leads India's gains are textile and apparel products³⁰, which accounts for more than 70% of total India's gains in goods exports. For Canada, the gains in exports are widely spread, ranging from primary agricultural and resources-related products, to chemical products, transport equipment, machinery and equipment, and services.

Table 6.4 Canada's simulation results - Implications for trade at sector level

Source: DFAIT Office of the Chief Economist estimates based on the GTAP model v. 6.2

Output impacts

Table 6.5 below provides the impacts of trade liberalisation between Canada and India on output in both Canada and India. In percentage terms, the impact of liberalisation on the total Canadian production output is larger than that of India. This is because the endogenous response of labour and capital to trade liberalisation under the closure rule of flexible supply of labour and capital is more significant in Canada than in India, which translates into a greater increase in production capacity in Canada than in India.

The impacts that emerge from the simulation at the sector level conform in broad outline to a priori expectations based on comparative advantage. Indian gains are most significant in textiles, wearing apparel and services and Canada's in industrial goods, machinery and equipment, and services.

Table 6.5 Canada's simulation results - Implications for output

Source: DFAIT Office of the Chief Economist estimates based on the GTAP model v. 6.2

6.3 India's Approach

6.3.1 Framework for the simulation

In this section, a comprehensive CEPA between India and Canada has been modeled, using a modified GTAP version 7 database that takes into account recent changes trade policy settings in both countries. The modeling also incorporates gains from trade liberalisation, trade facilitation and productivity gains. Sectors and regions are chosen for the model based on the economic and regional interest of both countries.

Tariff Updating

For the modelling exercise, the GTAP database version 7 is used where the database is benchmarked to 2004. In order to examine the impact of Canada-India CEPA on the overall economic activities of both the countries and also other regions of the world, there is a need for updating the tariff data for both the countries.

We have used TRAINS WITS online database to collect MFN tariff data at 6-digit HS for both Canada and India. Latest data available for both the countries is for the year 2008. For updating MFN tariff data, single year data is used. As discussed in chapter 2, tariff rates in India are declining very fast on account of unilateral liberalisation and similar as is the case for Canada. Updating of CGE database with latest tariff data would show the current state of liberalisation in both the countries. This effect is better captured by the current year data. We have also sourced data from national sources to update certain categories of tariffs such as mixed and specific tariffs. Using various methods, tariff equivalence of these tariff lines is estimated. Ad valorem tariffs are combined with these estimated tariff equivalence of the mixed and specific tariff lines to form comprehensive tariff variables for both the countries. Tariff lines are concorded with GTAP sectors using HS2007-GTAP concordance from the TRAINS WITS online database.

For the estimation of CGE model, we have used 18 regions by 27 sectors aggregated GTAP data. These sectors include seven from agriculture, fifteen from manufacturing and five from services. The HS2007-GTAP concordance is used to arrive at average tariffs for 22 sectors for the model in the goods sector (no tariffs for services sectors).

For the estimation of sector-level MFN tariffs, we used simple average tariff instead of imported weighted tariff. This approach is adopted on account of two reasons- (a) import-weighted tariffs have the tendency to suppress value of peak tariffs and (b) actual negotiation takes place on the basis of actual tariffs rather than import-weighted tariffs.

The above Figure 6.2 indicates the relative position of tariffs in specific aggregate sectors in both the countries and these sectors are used in the CGE model. Barring a few sectors, India's sectoral tariff rates are higher than those of Canada. Tariff liberalisation in both the countries may have some positive impact on trade and other macroeconomic variables.

Modelling Liberalisation with Trade Facilitation and Productivity

The current literature on regional integration suggests that deeper integration is essential for realising potential gains from any forms of regional cooperation like a Canada-India CEPA. Among various schemes of liberalisation suggested in the literature, trade facilitation is one of these important schemes which are often examined in the CGE framework³¹. Taking in account the distance factor into consideration, we have taken all modes of shipping cost as the proxy for the trade facilitation in the simulation.

It is expected that the proposed CEPA may create conditions for deeper integration of two knowledge economies. With various channels of linkages including investment flows, technology transfer, movement of natural persons, joint R&D, etc, there are greater possibilities of spillover of knowledge, leading to productivity gains for member countries.

In the CGE literature, use of productivity shock is a common feature. This is often used in the context of bilateral or regional free trade agreements, food security, climate change, labour, etc. among others. Several studies have examined the impact of productivity gains on various sectors of an economy in a CGE framework³². For the Brazilian economy, Carneiro and Arbache (2003)³³ applied 10% shock to the productivity variable to examine its effects on output and employment. We have taken a moderate view in terms of expected gains from productivity rise. However, we have chosen to put a productivity shock of 1% in the model.

6.3.2 Simulation results

In the modelling analysis, four scenarios are postulated to examine the efficacy of the proposed Canada-India CEPA. These scenarios are the following:

1. Trade liberalisation with shallow integration (tariff liberalisation only)
2. Trade liberalisation with trade facilitation
3. Trade liberalisation with productivity gains
4. Trade liberalisation along with trade facilitation and productivity gains

The results indicate that both India and Canada experience small welfare gains as a result of trade (tariff) liberalisation alone. Bilateral merchandise exports increase as trade barriers are removed. The gains are asymmetric, as would be expected given the relative size of the two economies and each country's respective initial level of trade barriers.

Trade liberalisation along with trade facilitation measures may not create many opportunities to generate large welfare gains for both countries. Productivity gains from the CEPA may be effective in improving significant welfare gains for both countries (See Table 6.6).

Table 6.6 India's simulation results: Implications for GDP

Source: RIS estimates based on GTAP v 6.2 database 7.
Note: Full Trade Liberalisation, trade facilitation measures liberalised by 10% and productivity improvement by 1%

Bilateral exports are likely to gain from the CEPA. As we move towards more comprehensive liberalisation, incremental gains for the export sectors are likely to increase, but these gains may not be significantly large (See Table 6.7).

Table 6.7 India's simulation results: Implications for bilateral exports

Source: RIS estimates based on GTAP v 6.2 and database 7.
Note: Full Trade Liberalisation, trade facilitation measures liberalised by 10% and productivity improvement by 1%

Partial equilibrium analysis could be applied to provide further insight into the impact of trade liberalisation between Canada and India.

²⁶ Dollar value impacts are based on GDP and trade levels in 2008. As the GTAP database ends in 2004, results from all simulations have been scaled up to 2008 levels.

²⁷ The Canadian and India tariffs in the WTO Integrated Trade Database can be found in the following website: India and the WTO and Canada and the WTO

²⁸ Information on World Trade Atlas available on the Global Trade Information Services website

²⁹ J. Francois, B. Hoekman and J. Woerz, 2007 "Does Gravity Apply to Intangibles? Measuring Openness in Services", memo.

³⁰ With the tariff relief initiative in Budget 2010, Canada has committed to liberalising most textiles in HS chapters 50 through 60, other than 62 tariff items that are finished products for retail sale, by 2015. This will moderate the impact of an FTA upon Indian textiles exports to Canada but not significantly according to our sensitivity test. India’s export gains also occur in wheat and dairy products. On dairy products, the variance in the individual tariff levels applied to different dairy products requires a more disaggregated modeling approach than is possible using the GTAP model in order to more accurately assess the trade liberalising impact in this sector. Consequently, the trade impacts as generated under the liberalisation scenario are likely underestimated with respect to India's gains in this sector. On wheat, Canada's global comparative advantage is likely not adequately captured in the GTAP model and as such, the trade impacts as generated under the liberalisation scenario are likely underestimated with respect to Canada's gains, and overestimated with respect to India's gains in the sector.

³¹ Hertel, T. W., Walmsley, T., and Itakura, K., 2001, "Dynamic effects of the "new age" free trade agreement between Japan and Singapore". GTAP Technical Report.; Mirza, Tasneem and Thomas W. Hertel, 2007, "The Role of Trade Facilitation in South Asian Economic Integration". Asian Development Bank Project Report.; Narayan, Badri, and Terrie Walmsley, editors, 2008, "Global Trade, Assistance, and Production: The GTAP 7 Data Base." Center for Global Trade Analysis, Purdue University. ; Mirza, Tasneem 2008, "A Cost-Benefit Analysis of Trade Facilitation in an Applied General Equilibrium Model", Economics Department, Purdue University, Working Paper.

³² Robinson, S., Y_unez-Naude, A., Hinojosa-Ojeda, R., Lewis, J. D., and Devarajan, S.,1999, "From stylized to applied models: building multisector CGE models for policy analysis", North American Journal of Economics and Finance, Vol.10, pp.5–38.; Barros, R. P., Corseuil, C. H., Cury, S., & Leite, P. G., 2001, "Abertura econ^omica e distribuic_~ao de renda no Brasil. Proceedings of the Workshop on Trade Liberalisation and the Labor Market in Brazil, Bras_Ylia, UnB/IPEA.

³³ Carneiro, Francisco Galrao and Jorge Saba Arbache (2003), " The impact of trade on the Brazilian Labour Market: ACGE Model Approach", World Development, Vol. 31(9), pp.1581-1595.