Estimating the effects of containerisation on world trade

This thesis examines the effects that containerisation had on the growth in world trade between the years 1962 and 1990. Containerisation is a technological change that arises from shipping goods via containers rather than through the traditional break-bulk method which characterised international shipping since antiquity. This thesis makes many contributions to the literature. This is the first quantitative and econometric study into the effects of containerisation in economics. We collect data from a specialist business publication and construct container variables which are used for the first time in economics. We also use a scientific classification from 1968 to classify products as containerisable or non-containerisable. Another contribution is that the econometric models employed in this thesis allow for a "horse race" between the technology variable and the policy variables: free trade agreements, General Agreement on Tariffs and Trade membership and currency unions. We make use of the cross-sectional and time series variation available to us in the adoption of the technology across 157 countries to identify the effects of containerisation on world trade. We employ several specifications and try different trade flow dimensions to pin down the right way to model containerisation. In doing so, we deal with several econometric problems that arise in similar econometric studies such as omitted variable bias and endogeneity bias. The effects of containerisation are felt 10 to 15 years after bilateral adoption of the technology. We estimate that containerisation led to an increase of 380% in North-North containerisable trade 10 to 15 years later. We find no evidence for endogeneity in this specification and we can be confident to make a causal statement. We also find evidence that containerisation affected North-South trade the most, followed by North-North and then South-South containerisable trade although we cannot be as confident about making causal statements in the case of North-South and South-South trade. The evidence is however suggestive of strong effects on containerisable trade in the two subsamples. In all cases, the effects of containerisable are found to be multiple times the size of the effects of the individual policy variables - 2 to 10 times as large depending on the subsample and the variable in question.EThOS - Electronic Theses Online ServiceGBUnited Kingdo