ABSTRACT
Assessing the degree of openness of an economy is notoriously difficult. Empirical studies have used varied approaches in the attempt to capture, via a summary measure, the multifaceted nature of trade policy. David (2007) surveys the literature and identifies six types of measures of openness: Trade ratios, Adjusted trade flows, Price-based, Tariffs, Non-tariff barriers and Composite Indices. The first three categories focus on outcomes while the last three focus on policies. Ideally, one would want to measure trade restrictions directly to determine the level of protection of a country. However, in general, it is easier to measure flows and prices than barriers. Flows are observable and quantifiable and for many countries data are available extending back several decades (at least back to 1970 for a large number of the developing countries). Conversely, data based on the observation of trade restrictions themselves is much harder to collect and work with. Gathering data on tariffs can be challenging. Countries do not report their weighted average tariff rate or even their simple average tariff rate every year, so the most recent data may be several years old. The quantity of data required for calculating weighted tariffs and ERPs is daunting. The data for tariffs are measured with error and there are frequently problems with missing data due to activities outside the formal market such as smuggling. Quantifying and aggregating non-tariff restrictions suffer from the same problems to a greater degree, as the researcher must calculate and combine the effects of what are frequently fundamentally different types of instruments as well as problems arising from the use of qualitative data. Our measure follows Pritchett (1996) and Loayza and Soto (2005) and is based on the
notion that highly protectionist policies should reduce the amount of economic activity that is traded. To estimate the size of this reduction he proposed ‘structure adjusted trade intensity’ measures, which are the residuals from a regression of trade intensity on structural characteristics such as population, land area, level of per capita GDP, trade composition and a dummy for landlocked economies which usually face higher transport costs. The residual from the equation implies by how much a country’s openness differs from what would be expected of a country with the same characteristics. We use a panel of 164 countries in the period 1970-2010 to run a fixed-effects model using around 5500 observations. Table A1 presents the estimation results.
