Increasing returns to scale and urban characteristics: bridging firms and regions

Gustavo Britto, Land Economy Department, Darwin College, University of Cambridge

The role of the degree of returns to scale of economic activities has been subject of a long lasting controversy in the economic literature. The prevalence of increasing returns to scale, expressed by the connection between the rate of output growth and that of productivity, also known as the Verdoorn Law, is the centrepiece of cumulative causation models (Kaldor, 1966). The Verdoorn Law has been extensively estimated providing support for cumulative causation growth theories. However, the importance of increasing returns to regional growth and economic agglomeration has only recently found its way into mainstream economics.
This study has two main objectives. First, we aim to provide new estimates of the degree of returns to scale in the manufacturing industry. Unlike previous works, which tend to focus on aggregate or regional data, we estimate various specifications of the Verdoorn Law to seek for evidence of increasing returns to scale in the manufacturing industry using micro data. To this effect we used a large panel of Brazilian firms over the course of five years.
Secondly, the paper proposes a new methodology to assess the relevance of a spatial dimension to the cumulative causation growth literature. The objective is to measure the impact of urban attributes over the rate of productivity growth of firms. The proposed model must discern individual characteristics (level-1) from those that are eminently associated to the region where each city is located (level-2). Hence, hierarchical linear models are fitted to estimate an extended version of the Verdoorn Law using a cross section of firm-level data. Hierarchical models were specifically designed to treat data characterised by a clustered structure. Apart from statistical benefits associated to its use, hierarchical models also allow the introduction of cross-level interactions. The original equation is expanded to include variables such as schooling and technological gap, at the firm level, and variables to measure the urban scale, share of services sector and industrial concentration at the city level.
The model uses micro data from the Annual Industrial Survey (PIA) from the Brazilian manufacturing industry from 1996 to 2002. In the first part, the regressions showed widespread increasing returns to scale. In addition, regional dummies indicated significant regional differences in the estimated coefficients. In the second part, the model provides support for the thesis that there is an urban dimension that impacts significantly the rate of productivity growth of firms. In particular, the existence of economies of agglomeration (Marshallian and Jacobian) allows otherwise similar firms located in larger cities (or in cities where services are more readily available) to attain faster rates of productivity growth.

About the speaker
Gustavo Britto has a BA in Economics from the Federal University of Minas Gerais (Brazil), where he studied the localisation of high-tech industries in Brazil. After graduation in 1999, he started a Masters in Economics in Sao Paulo (Unicamp). During this period he joined the Centre of Industrial and Technology Economics (NEIT-IE-Unicamp), working under the supervision of Prof. Mariano Laplane. At NEIT, he took part in many of the centre's researches, including the full assessment of of Brazil's industrial chains (ECCIB), carried out in 2002, the analysis of the impact of the Common Market (Mercosur) on trade and developement carried out in partnership with multinational research centre Red Mercosur and in the international assessment of the impact of foreign direct investment on developing countries, led by CUTS International in partnership with the Department for International Development (DFDI-UK). His own research focused on the impact of trade liberalisation over Brazilian external accounts, using micro trade data and input-output matrices. After being awarded the masters degree, he started a PhD at the University of Cambridge in 2003. Under the supervision of Dr. John McCombie (Land Economy Department / Downing College), his current research focuses on productivity change and cumulative causation models.

Venue: Keizer Karelplein 19, Maastricht

Date: 12 April 2007

Time: 16:00