World Development, Vol. 19, No. 4, pp. 341-347, 1991. 0305-750X/91 $3.00 + 0.00 Printed in Great Britain. © 1991 Pergamon Press plc Foreign Exchange and Industrial Development: A Frontier Production Function Analysis of Two Tanzanian Industries AMMON MBELLE University of Dar es Salaam, Tanzania and THOMAS STERNER* University of Gothenburg, Sweden Summary. - - A frontier production function is used to analyze the importance of foreign exchange for efficiency in two Tanzanian industries: textiles and beverages. In both industries, foreign exchange is found to be the most important factor of production. We also find, somewhat surprisingly, that there is some technical progress, at least for textiles, at the frontier (i.e. for the best-practice firms). The reason for the overall decline in industrial performance is instead found to be the poor individual performance of many plants due to low capacity utilization. 1. INTRODUCTION 2. DATA Tanzania is a country which, except for two years (1977 and 1978), has faced an acute foreign exchange crisis since the mid-1970s. This crisis, cited as the main cause of deterioration in economic performance in almost all sectors, has had several contributing factors, including neg- lect of the agricultural sector and deterioration in the terms of trade. In the 1960s, Tanzania appeared to be indus- trializing rapidly. Industrial share of GDP in- creased from 3% to 10%. During the 1970s, however, it stagnated, and during the 1980s the GDP share of industry even fell somewhat (see Mbelle, 1987). During 1974-86, total growth in industrial production was -29%. By contrast, corresponding figures for Kenya and Zambia were +133% and +117%, respectively. The most striking aspect of this sad perfor- mance is that investment has in no way been lacking. On the contrary, there has been a good deal of investment both in industry and in other sectors. In fact, one is tempted to conclude that much too much has been invested so that not enough foreign currency has been available for current maintenance and imported intermediate materials. This study looks at the importance of imported goods for production by examining two particular case studies. Our analysis covers two industries, textiles (ISIC Code 3211 -- spinning, weaving and finishing textiles) and beverages (ISIC Code 3134 - - soft drinks, beer and carbonated waters). For both industries, the analysis is based on a set of data for each plant and each year 1974-84. The plants produce fairly homogenous products (square meters of cloth and bottles respectively). Eleven firms are covered for the textile industry and 17 for the beverages in~tustry. The problem of poor data quality, common in developing countries, was also encountered in our study. There are several missing observa- tions, a fact which calls for caution when employing advanced and sensitive models. Con- siderable effort has, however, been put into controlling the quality of the data by crosscheck- ing between the following sources: the Ministry of Industries and Trade, TEXCO, the Central Bureau of Statistics, and the Bank of Tanzania, as well as our own field survey. We believe these are the best data available and that, in fact, access to such a disaggregated set of detailed and *This paper is based on research reported in chapters 5 and 6 of Ammon Mbelle's (1988) PhD dissertation. We would like to thank H. Bjurek, L. Hjalmarsson, F. F6rsund and two anonymous referees for valuable comments. 341