Cluster Based Industrialization and its Effect on Firm Productivity in Pakistan

This paper examines the effect of externalities associated with agglomeration on firm level productivity in Pakistan. We begin our analysis by calculating six measures of agglomeration which give us the pattern of spatial concentration across the country. To calculate these six measures we use data at the country level and for Punjab between 1988-89 and 2005-06. The data is derived from the Census of Establishments for 1988-89, and the Census of Manufacturing Industries for the years 1995-96, 2000-01 and 2005-06. Our agglomeration measures show that although 59.5% of the industries in the country are highly agglomerated, the results are varied, with no one industry consistently emerging as a concentrated industry in either the country or in Punjab for the several years examined. Similarly, patterns of regional specialization tend to vary considerably over the length of time under examination with areas identified as specialized in the first round of data not ranking as most specialized in the next wave. Finally, the data shows that localization economies are typically stronger than urbanization economies in the Pakistani context, indicating that firms tend to benefit from intra industry externalities rather than inter industry externalities. Our analysis also considers development indicators which may be affecting the firm location decision. Specifically, we examine human capital development, crime, road density, and agricultural output. Overall, we find that human capital development is fairly low in the country, with large urban districts such as Karachi staying at the top of the ranking regardless of the year examined, and lower ranked districts failing to catch up. At the same time, crime in these urban districts remains high while road density is high in most of the large urban districts, with the exception of Karachi district, whose lower rank may be explained by its large size, and by considering that district includes less developed surrounding areas around Karachi city proper. Finally, districts in the agricultural heartland of the country, namely Multan, Rahim Yar Khan and Bahawalpur remain the highest in terms of agricultural produce, with Karachi and Karak ranked among the lowest. Having established the patterns of agglomeration and development across the country we next examine the impact of these location and industry concentration indexes on productivity in manufacturing firms through a plant level production function across all districts of Pakistan, while also identifying the impact of firm inputs such as capital and labor. Our regression specification includes controls for connectivity through a road density variable, human capital levels through an education index, and working environment through crime statistics. Our results indicate a strong role for agglomeration economies and the associated externalities in firm output over a long time horizon, and a significant role for development indicators like road density and human capital over a shorter time frame.