Many economists encourage governments to abolish import taxes and have complete

6 The Effects of Trade Policy on Aggregate Growth and Poverty

Much of the literature on the effects of trade policy on the outcomes discussed in Sections 3 and 4 focuses on its short-run and static effects. The distinction between short- and long-run effects is not unique to trade policy. But in the case of trade policy, there are good reasons to believe that the long-run effects are orders of magnitude larger than the short-run effects. While the latter can be potentially identified through careful empirical work, the long-run effects are substantially harder to pin down empirically. As a result, arguments about dynamic long-run effects are often made based solely on principles and theoretical models, with little formal empirical support. In this section, we consider the [scant] evidence on long-run effects of trade policy on aggregate growth and poverty.

The relationship between a country's trade policy and aggregate economic growth is of key policy interest, and the empirical literature on the topic is one of the oldest areas of empirical inquiry in international economics. Although many economists believe, based on economic theory, that reductions in trade barriers promote economic growth, robust evidence on this relationship at the aggregate level has been elusive. Rodrik and Rodriguez [2001] and Hanson and Harrison [1999] review the issues that affect the estimation of the effect of trade policy on aggregate growth and conclude that the estimates of the effect are not robust. Most of the literature on the topic has examined the relationship between trade policy and growth in a cross section of countries. The issues that influence the inference include weak links between the empirical work to the underlying predictions from the theoretical literature, selective samples of countries with available data, measurement of trade policy at the aggregate level, consistency of measurement of key variables across countries and time, and endogeneity of trade liberalizations. To the extent that there is a positive relationship between trade policy and economic growth, it is not clear whether trade policy leads or lags. Does trade policy lead to higher economic growth or do countries at a certain level of development choose to implement more liberalized trade policy? Alternatively, do countries with less restrictive trade policy in general have economic institutions in these economies that are associated with higher growth?

While the robustness of findings in this area of research continues to be debated, researchers have recently used microlevel data on trade policy from trade liberalizations during the 1980s and 1990s and empirical frameworks guided by economic theory to make progress on the effects of trade policy on aggregate growth. Estevadeordal and Taylor [2013] find that countries that liberalized trade policy during the 1980s and 1990s [in part driven by the Uruguay round of the WTO negotiations] observed higher growth rates in GDP per capita over this period relative to countries that did not liberalize. According to a version of Solow model they develop, decline in import tariffs on capital goods increases incentives for firms to invest, which in turn increases steady state growth. Lower tariffs on intermediate inputs increase productivity, and subsequently steady state growth. Further analysis, which distinguishes between liberalized trade in production inputs and final consumption goods, finds that the positive relationship between trade liberalization and economic growth is driven by declines in tariffs on intermediate inputs and capital goods. Consistent with these channels, they show that countries that lowered tariffs on intermediate inputs and capital goods observed increased imports of intermediate and capital goods. On the other hand, there is no relationship between lower tariffs on consumer goods and economic growth. These findings provide country-level support for the effects of liberalized trade on improved efficiency of production through imported inputs and technology, channels that have been emphasized in studies of firm performance [Amiti and Konings, 2007; Goldberg et al., 2010; Khandelwal and Topalova, 2011].

The effects of trade policy [via economic growth] on poverty are even more difficult to quantify empirically than the relationship between trade policy and growth. In addition to establishing that trade policy affects growth, one needs to determine both whether and how trade policy-induced growth affects the poor. This is a challenging task to accomplish with aggregate data. Lack of availability of household survey data with information on consumption and income from many low-income countries affects measurement of poverty and average incomes of the poor [Deaton, 2005; Ravallion, 2001]. In the absence of reliable survey data, average incomes of the poor, which are often measured by the average income of the households in the bottom fifth of income distribution, are imputed from very noisy measures of income distribution within a country. With noisy measures of income inequality, this imputation makes is likely that measures of income of the poor simply follow changes in average incomes [Banerjee et al., 2006]. This biases the results in favor of pro-poor effects of growth.x In recent years, household surveys are increasingly available and the World Bank Research Department has made substantial progress on measurement of poverty across time and space. Nonetheless, the poverty measures do not span periods of trade liberalization for a large share of countries, so the relationship between trade policy and poverty across countries remains empirically elusive.

In summary, the literature on the effects of trade policy on aggregate growth does not provide much robust evidence that trade policy affects growth. This is a very different conclusion from the message in the studies of the effects of trade on firm performance. This leaves one wondering whether the lack of robust aggregate evidence in part reflects the methodological challenges highlighted in Section 2, which are amplified in aggregate studies. Recent work by Estevadeordal and Taylor [2013] makes headway on overcoming some of these issues, while focusing on one particular channel of the link between trade policy and growth. Its findings are consistent with the evidence from firm-level studies that emphasize the role of trade policy in promoting efficiency and innovation through access to imported inputs and capital goods in less developed countries. However, more work is needed is this area.

2.3.2 Prisoners’ Dilemma

We now briefly consider the basic Prisoners’ Dilemma that arises in the monopolistic competition model. As with the models above, we characterize the Nash, efficient and politically optimal tariffs.

To characterize Nash trade policies, we begin by representing the best-response or optimal trade policies for the home and foreign governments. We assume in this discussion that the relevant second-order conditions hold. The home-country best-response import and export policies are then determined by the following two first-order conditions:

[36]Vpfdpfdι=0Vph*dph*dι*+Vp*wdp*wdτh*=0,

where the home-country import tariff τh affects pf via ι, the home-country export tariff affects ph*via ι*, and the indirect utility function V is given in [35]. Similarly, the foreign-country best-response import and export policies satisfy the first-order conditions

[37]Vph**dph*dι*=0Vpf*dpfdι+Vpw*dpwdτf=0.

We may now define the Nash trade policies, [τhN,τh*N,τf*N,τfN], as the tariffs that simultaneously satisfy [36] and [37].

Consistent with our discussion earlier, we note that terms-of-trade effects are absent from the conditions that determine the best-response import policies [the top expressions in [36] and [37]] but present in the conditions that determine the best-response export policies [the bottom expressions in [36] and [37]]. As Bagwell and Staiger [2015] discuss further, the optimal export policy for a given country thus represents a balance between the terms-of-trade gain that is associated with an export tariff and the firm-delocation benefit that is associated with an export subsidy.

Our next goal is to characterize efficient trade policies. Since this model also has sufficient trade-policy instruments with which to effect lump-sum transfers, efficient trade policies are those which maximize V + V*. Using [34], we observe that total income can be written as

[38]I[⋅]+I*[⋅]=L+L*+[pf−ϕpˆ−pˆ]M[pf,ph*]+[ph*−ϕpˆ−pˆ]E[pf,ph*]≡T[pf,ph*].

Notice in particular that world prices do not affect total income, which is to say that we can express total income as the function T[pf,ph*]. Using [35] and [38], we can now also represent joint welfare as a function J[pf,ph*]where

[39]V[⋅]+V*[⋅]=P[pf,ph*]−ϵθ[1ϵθ]+P*[pf,ph*]−ϵθ[1ϵθ]+T[pf,ph*]≡J[pf,ph*].

As this expression confirms, any trade-policy induced change in world prices corresponds simply to pure international rent shifting and does not affect efficiency.

Efficient trade policies thus maximize the joint welfare function J[pf,ph*]given in [39]. By [25] and [27], respectively, we also know that ph*=ph*[ι*]and pf = pf[ι]; thus, joint welfare is a function of the four tariffs only through ι and ι*. Assuming that the joint welfare function is strictly concave when treated as a function of ι and ι*, the set of efficient tariffs is thus characterized by the following two first-order conditions:

[40][Vph*+Vph**]dph*dι*=0[Vpf+Vpf*]dpfdι=0.

Bagwell and Staiger [2015] further explore the two conditions in [40] and show that efficiency requires a net subsidy to trade along each trade channel [ie, τf*+τh*