RDP 2004-11: Trade Openness: An Australian Perspective 1. Introduction

Consumers in other countries may know Australia best for its beer, dairy and wool products, but Australia's economy relies very little on exports. Of the 30 members of the Organization for Economic Cooperation and Development, only 3 are less export-dependent – Greece, the United States and Japan (New York Times, 7 March 2003, p W1).

There is substantial interest among academics and policy-makers in the role of external trade in macroeconomic outcomes. For example, there is a vigorous debate among economists as to whether trade openness influences either the level or growth of output (see, for example, Frankel and Romer 1999, and Rodriguez and Rodrik 2001). In the Australian case, trade is considered an integral part of economic activity and, among Australians at least, it is widely perceived that Australia is a very open economy. Yet, as shown in Figure 1, Australian trade (exports plus imports, a broader measure than used in our opening quote) as a proportion of GDP (henceforth termed openness) is relatively small compared to other industrialised countries.[1] With exports and imports of goods and services each equivalent to around 21–22 per cent of GDP, Australia's 2002 openness ratio of 43 per cent was substantially below the median for OECD countries of 69 per cent. In addition, of the 136 countries and territories for which the Penn World Tables have data for 2000, Australia was the 20th least open economy.

Figure 1: Trade as a Proportion of GDP – OECD Countries
Figure 1: Trade as a Proportion of GDP – OECD Countries

Source: authors' calculations based on IMF Direction of Trade Statistics

The fact that Australia trades less than other countries does not necessarily imply that Australia trades less than would be expected. There could be good reasons for its relatively low openness. For example, it might reflect Australia's geographic isolation, especially from the economic mass of North America and western Europe.[2] Another plausible factor might be Australia's geographic size. Geographically larger countries will typically have more diversified endowments, allowing them to produce a wider range of agricultural and resource products, which means they may have less need to trade than smaller countries. Thus, although Australia trades much less than most other OECD countries, it might actually be the case that it trades more than would be expected given its geographic attributes.

There is relatively little theoretical or empirical analysis looking at the determinants of the openness of countries, or at least at the variables that can explain openness.[3] This paper takes up that task with an empirical study of openness and the economic and geographic variables that are able to explain cross-country differences in openness.

To provide greater perspective for our results, we first examine a gravity model of bilateral trade. This model has been used extensively to estimate the effect on bilateral trade of factors such as the use of a common currency or common membership in a free trade agreement (FTA). Our approach is somewhat different in that we take the estimates from a gravity model and use different methods to extract information about countries' total trade. We show that the different methods can yield conflicting results. For Australia, one approach suggests that total trade is about 50 per cent more than predicted whereas another suggests that Australia trades fully 2.8 times the prediction of the gravity model. We interpret these results as providing substantial justification for our analysis focusing on openness and the total trade of countries.[4] Given that models of bilateral trade do not seem to provide good predictions of aggregate trade, we estimate a model for total trade as a proportion of GDP.

The basic openness equation includes some relatively standard variables motivated by the empirical work on the gravity model such as population, area, GDP per capita, and a measure of economic location based on average distance to the rest of the world. However, a more significant addition is the inclusion of a comprehensive measure of trade-policy liberalisation. Also, our robustness tests examine a number of additional variables such as a measure of the efficiency of transport infrastructure, and some resource-related variables. We find some interesting results, including evidence that the widely accepted positive correlation between the level of economic development and the amount of trade disappears after controlling for countries' trade policies. Overall, our equations are able to explain a substantial proportion of the cross-country variation in openness. And returning to the question that sparked our study, the openness equation suggests that Australia's level of trade is relatively close to what would be expected for a country with its characteristics.

The remainder of the paper is structured as follows. Section 2 presents some analysis with the bilateral gravity model. Estimation with the openness equation is discussed in Section 3. Some Australian extensions of the openness equation are provided in Section 4, and Section 5 concludes.


We focus on trade outcomes as our measure of openness, rather than the presence or absence of barriers to trade, the degree of capital account liberalisation or the size of capital flows. [1]

In related work, Battersby and Ewing (2003) have examined the impact of remoteness on Australia's trade using the gravity model and conclude that Australia's level of total trade is low because of its relatively remote location. The analysis in the current paper differs from Battersby and Ewing in that our focus is mainly on equations for total trade rather than on the gravity model and bilateral trading patterns. In addition, our analysis includes a wider range of possible economic determinants of trade and a larger sample of countries. [2]

We are aware of only a few published papers that examine the determinants of openness. In considering the implications of openness for the size of government, Alesina and Wacziarg (1998) provide a brief empirical analysis of the determinants of openness, using a sample ending in 1989. Pritchett (1996) briefly considers openness in a wider examination of the outward orientation of countries' policies. Rogowski (1987) considers the relationship between openness and government/parliamentary structure. There is also a recent working paper by Jansen and Nordås (2004) that addresses some of the same issues covered in this paper. [3]

The finding of strong country or regional effects in the gravity model – countries in some regions have bilateral trade levels that are consistently higher or lower than predicted (see, for example, Frankel and Wei 1998) – would provide additional justification for our approach. [4]