Where do new firms locate? The effects of agglomeration on the formation and scale of operations of new firms in Punjab

Haroon, Maryiam; Chaudhry, Azam Working Paper Where do new firms locate? The effects of agglomeration on the formation and scale of operations of new firms in Punjab Economics Discussion Papers, No. 2014-21 Provided in Cooperation with: Kiel Institute for the World Economy – Leibniz Center for Research on Global Economic Challenges Suggested Citation: Haroon, Maryiam; Chaudhry, Azam (2014) : Where do new firms locate? The effects of agglomeration on the formation and scale of operations of new firms in Punjab, Economics Discussion Papers, No. 2014-21, Kiel Institute for the World Economy (IfW), Kiel This Version is available at: https://hdl.handle.net/10419/97111 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/3.0/ Received May 4, 2014 Accepted as Economics Discussion Paper May 7, 2014 Published May 12, 2014 © Author(s) 2014. Licensed under the Creative Commons License - Attribution 3.0 Discussion Paper No. 2014-21 | May 12, 2014 | http://www.economics-ejournal.org/economics/discussionpapers/2014-21 Where Do New Firms Locate? The Effects of Agglomeration on the Formation and Scale of Operations of New Firms in Punjab Maryiam Haroon and Azam Chaudhry Abstract The formation of new firms is an important determinant of economic development and the industrial organization literature highlights agglomeration as one of the main factors affecting the formation and scale of operations of new firms. This paper is one of the first to use developing country data to estimate the impact of localization (the benefits accruing to firms that choose to locate in a specific region within a specific industry) and urbanization (the benefits accruing to firms located close to each other regardless of the type of industry to which they belong) on new firms’ formation and scale of operations. Our findings reveal that agglomeration measured through density of employment has a significant impact on the formation of new firms and on their scale of operations in Punjab. JEL L1 L2 Keywords Agglomeration; firms; localization; urbanization Authors Maryiam Haroon, Department of Economics, Lahore School of Economics. Intersection Main Boulevard DHA Phase VI and Burki Road, Burki, Lahore, 53200, Pakistan, [email protected] Azam Chaudhry, Department of Economics, Lahore School of Economics. Intersection Main Boulevard DHA Phase VI and Burki Road, Burki, Lahore, 53200, Pakistan Citation Maryiam Haroon and Azam Chaudhry (2014). Where Do New Firms Locate? The Effects of Agglomeration on the Formation and Scale of Operations of New Firms in Punjab. Economics Discussion Papers, No 2014-21, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2014-21 2 1 Introduction The formation of new firms is a key characteristic of a growing economy and the location of these newly established firms can have a profound impact on regional development. Besides the well established socioeconomic benefits and employment opportunities for the inhabitants of areas experiencing new business growth, a number of positive externalities can also accrue to the firms which decide to locate in an agglomerated region. Early work by Marshall (1920) classified the three principal benefits available to firms that choose to locate in a geographically concentrated area into: (i) labor pooling, (ii) knowledge spillovers, and (iii) availability of specialized inputs, all of which give entrepreneurs the incentive to locate in concentrated areas. Jacobs (1969) also emphasized the benefits accruing to firms in an agglomerated area from the presence of a diverse labor force. Firms located close to each other may also be able to lower costs through input sharing and accessing maintenance services through mutual contracts, all of which lead to the more effective use of resources. More recent empirical work (Rosenthal and Strange, 2010 and Otsuka, 2008) has found that localization (the benefits accruing to firms that choose to locate in a specific region within a specific industry) and urbanization (the benefits firms located close to each other regardless of the type of industry to which they belong) are the two main forces of agglomeration affecting the formation of new firms as well as their scale of operations. This paper presents the results of a unique developing country analysis of the relationship between agglomeration and the formation of new firms as well as their scale of operations at the district level in Punjab, Pakistan. We analyze whether the presence of similar manufacturing activity in a district fosters new firm formation and also whether a concentration of firms from different industries leads to the entry of new firms into a particular district. This is done by analyzing the effects of agglomeration on the arrival and scale of operations at a district level in the manufacturing industry for 2008, incorporating socioeconomic characteristics at the district level and industrial controls using a combination of firm-level data and household survey data. Our results indicate that firms in Punjab significantly benefit from locating in agglomerated regions, which induces firm entry to gain the benefits of agglomeration. Localization has a significant and positive impact on new firm formation, and new firm formation is higher in areas of medium-scale urbanization. The scale of operations of new entrants increases where large or medium-scale firms belonging to the same industry are present. The scale of operations also tends to increase in areas of medium-scale urbanization. Interestingly we find that the presence of large firms across different industries in an area tends to negatively affect new firm entry. We also find that average income in a particular area has a significant and positive impact on arrival as well as on the scale of operations of new firms. The paper is organized as follows. Section 2 describes the related literature and Section 3 presents a theoretical framework. Section 4 describes the dataset used for in the empirical estimations. Section 5 presents the model to be estimated, while Section 6 discusses the results. The conclusions are presented in Section 7. 2 Literature Review The concentration of industrial activity has been a widely studied phenomenon and agglomeration - defined as the presence of different economic units within the same 3 geographical location, which allows them to extract some benefit from each other’s industries— occurs widely across economies (Krugman, 1995). Among the most well known examples is the computer hardware industry in Silicon Valley (Sorenson & Audia, 2000) and some of the major clusters in Pakistan include the surgical goods and sports goods industries in Sialkot (Nadvi, 1999, Chaudhry, 2010, and Nadvi, 2003). Marshall (1920) explained that agglomeration occurs as a result of three key factors: First, firms agglomerate near suppliers or customers to save on shipping costs. Second, labor-pooling benefits can accrue to firms when labor is used effectively and different firms share skills. Third, the rate of innovation can be increased through knowledge spillovers. Rosenthal and Strange (2001) show that all these factors play a role in inducing industries to agglomerate, varying from industry to industry and depending on the commodity being produced. In their paper the authors highlighted labor pooling as a particularly important variable in determining the geographic concentration of industries, and the study’s findings reveal that agglomeration is positively affected by labor pooling and input sharing. We examine agglomeration by analyzing two main factors: urbanization and localization. Localization refers to the benefits accruing to firms that choose to locate in a specific region within a specific industry. These benefits can also be described as benefits that are external to the firm, but internal to the industry, e.g., knowledge spillovers, input sharing, and labor pooling. Localization plays an important role in industrial growth because firms belonging to the same industry are more likely to use similar inputs and through localization, these inputs can be shared or contracts mutually formalized. The accessibility to suppliers provides significant input sharing advantages for a localized industry. Labor pooling allows firms to use specialized labor and avoid labor shortages. Knowledge spillovers are also a component of localization economies which take place across firms due to interaction of workers as they can share information about products in production, production process, innovations, existing and new technology, marketing agendas, and research and development (Parr, 2002). There are several examples of localization economies in the world, including the semiconductor industry in Silicon Valley in the US (Delgado, Porter and Stern, 2010). Urbanization can also impact industrial growth because of a variety of benefits to firms which include the presence of diversified suppliers, specialized labor and suppliers, and diversified production (Bosma, van Stel, & Suddle, 2006). Urbanization can also provide firms with market mechanisms, transportation facilities, infrastructure, and community facilities, which make certain areas more attractive for new firms to enter (Parr, 2002). The presence of different industries also facilitates production, since one firm may be another firm’s supplier and different firms can produce the same product together in different stages. Recent empirical works has found that localization and urbanization have an impact on a firm’s birth decision in a particular region due to the benefits arising from proximity. According to Sorenson and Audia (2000), new entrepreneurial activity is likely to take place in areas of geographic concentration since new firms to take advantage of the learning processes of old firms. Also new firms enter when they can visualize a developed market, existing suppliers, and the availability of factors of production at a lower cost (Bosma et al., 2006). They will use the existing specialized labor and inputs, which results in higher productivity and profits and there is also a greater likelihood of gaining the benefits of innovation. 4 Apart from the factors discussed above, the literature identifies a number of other determinants of new establishments, including regional characteristics. The presence of a large population in a region positively affects firm birth because it generates higher demand. According to Otsuka (2008), various location factors affect the formation of new establishments in a particular region, including market demand, agglomeration, market conditions, and factor cost. At the same time regional unemployment influences the creation of new establishments because an increase in unemployment is likely to positively affect future entrepreneurs. Workers who have lost their jobs may not want to move away from a particular area due to social ties and end up starting their own business. However, higher unemployment may also lead to a fall in regional income and, hence, to less demand for products, deterring firm entry. Firm entry is also affected by the concentration of personal or household wealth in an area, which affects the capital available to entrepreneurs. Finally, government policies attract new firms to a particular area through government spending on local infrastructure and the provision of direct assistance to firms (Reynolds, Storey, & Westhead, 1994). The agglomeration–organization relationship has been empirically analyzed for a few countries: Using US data, Glaeser and Kerr (2009) found that labor and suppliers have a strong impact on entrepreneurship and are the main drivers of new firm formation. Delgado, Porter, and Stern (2010) also used US data to examine the relationship between agglomeration and the growth rate of entrepreneurship and their results showed that agglomeration is associated with growth in new firm formation and scale of operations by reducing the barriers to entry. Rosenthal and Strange (2010) used data from the US manufacturing, wholesale trade, fire, and services industries and found that urbanization significantly affects firm arrival and scale of operations for small establishments in the manufacturing sector, while localization affects firm arrival and scale of operations for medium establishments in the manufacturing sector. Otsuka (2008) on the other hand used Japanese data to determine the impact of regional characteristics on new firm formation and determined that localization and urbanization positively and significantly affect firm birth in the manufacturing industry and also that firm birth was significantly affected by market access, road transportation availability, labor density, the presence of highways, and lower labor costs. In the context of Pakistan, Chaudhry, Nasir and Haroon (2012) analyzed the factors which affected new firm entry and the scale of new firms in the export clusters of Punjab. The results of their paper show that more export sector firms will enter highly concentrated industries and that firm entry increases significantly as a result of a depreciation in the trade-weighted real exchange rate, while the impact of changes in trade partner tariffs is not significant. Chaudhry (2005, 2010) illustrated the benefits of agglomeration in Pakistani clusters while Chaudhry (2011) looked at the drawbacks of agglomeration and showed that firms in clusters may be unwilling to try new suppliers when their existing ones are connected to them socially. Also Burki and Khan (2010) looked at district level industrial agglomeration and found that industrial concentration is a prominent characteristic of all districts of Punjab but is more apparent in the urbanized areas of the province. The discussion above shows that one of the limitations of the existing empirical analyses is that it primarily focuses on developed countries. This is an obvious shortcoming since the relationship between arrival and agglomeration is as likely to hold for developing countries where there is weak contract enforcement, markets are not as developed, firms face financial constraints, and locating in agglomerated areas is more likely to foster research and development. 5 3 Theoretical Framework The theoretical framework we use to explain how agglomeration can have an impact on the entry of firms is the model of Soubeyran and Thisse (1998). The model assumes price to be equal in all districts (locales) and firms choose to maximize profit. Firms are attracted to areas that have a greater stock of knowledge. There are  D districts, with  dD{1,...i} . Each district has a fixed labor force represented by  Ld in district  d , earning positive wages. The district has an initial level of knowledge represented by  Kd0 , which workers have accumulated over the years. Entrepreneurs can start a new firm by acquiring capital  Kd at interest rate  r and hiring labor in a particular district, and can sell their product at price  p . Knowledge has been accumulated by labor, which makes districts attractive since firms acquire benefits from the existing knowledge base. The cost function of a firm in a particular district is given by )()(),,( dddddddd qrkqKwKwqC   (1) where  qd is output,  wd is the wage level,  Kd is the initial stock of knowledge in district  d and  represents workers’ knowledge base, which they have acquired through knowledge spillovers. The capital  k(qd) required by a new firm is constant across districts. The profit function of a firm locating in a particular district  d is defined as  d(qd,wd,Kd)pqdCd(qd,wd,Kd) (2) By locating in a particular district, a firm’s profits are affected by its initial stock of knowledge. Differentiating the profit function by  Kd shows how profit is affected by knowledge, which is given by 0)( ˆ ˆ     ddd d dKqw K   (3) The first-order condition indicates that profit increases with an increase in the stock of knowledge. Positive production by firms can be shown as ]}/)]({[)( ˆ1rKwpKq ddd    (4) Combining value functions (2) and (4) yields  ˆ  d d[ˆ q d(wd,Kd,r,p),wd,Kd]ˆ  d(wd,Kd,r,p) (5) which Soubeyan and Thisse (1998) have shown to equal  ˆ  dr  (ˆ q d) (6) Equation (6) represents the maximum profit a firm can derive by locating in a district  d . There are firms already located in district  d and with their presence workers have acquired skills 6 through knowledge spillovers. Districts with greater knowledge stocks have higher chances of firms entering. Assuming a positive production function and positive wages, full employment can be written as dddd LKqn )( ˆ (7) Manipulating equation (7) allows us to determine the number of firms in district  d , which can be represented as )( ˆ /dddd KqLn  (8) Equation (7) and the equality of profits between districts imply that  r  (ˆ q d)r  (ˆ q e) where  d,eI , and  I represents districts where new firms will be established. This indicates that the output produced by firms is the same across districts in equilibrium. Hence, equilibrium output can be stated as  Dd dd KLIq )()( ˆ  (9) where   is strictly increasing. Combining (8) and (9) gives the equilibrium distribution of firms:  nd(I)Ld  (Kd) Le  (Ke) eI ,  dI (10) Equation (10) is the key equation and states that the higher the number of workers or the greater the knowledge spillovers in a district, the higher the number of firms entering that particular district will be. The empirical analysis in this study analyzes how the density of employment both within a particular industry and overall employment across all industries in a district affect firm arrival and scale of operations. 4 Data Sources and Descriptive Statistics Our analysis focuses on the province of Punjab, Pakistan, and uses data from the Government of Punjab’s Directory of Industries (DOI) for 2006 and 2010. The DOI is a firm-level dataset which includes more than 16,000 firms in a particular year and includes information on firms’ years of establishment, employment levels, and districts. We have used the DOI 2010 to measure the arrival of firms and their scale of operations in 2008 and the DOI 2006 to measure local conditions (localization and urbanization) in 2006 and 20041. Socioeconomic characteristics at the district level are incorporated using the Government of Punjab’s Multiple Indicator Cluster Survey (MICS) dataset for 2003/04. In order to understand the nature of the data, it is useful to look at some basic summary statistics. Table 1 reports the number of new establishments (arrival) and their scale of 1 The reason for using 2006 and 2004 measures of local environment is to overcome the issue of reverse causality between the dependent variables (firm arrival and scale of operations) and the measures of local environment. 7 operations. As the table shows there were 312 new firms that entered the manufacturing sector in 2008, employing 10,501 employees. The table shows that localization and urbanization are higher in large-scale firms, followed by medium-scale firms. Table 2 reports the number of new industrial firms and their scale of operations in 2008 and average localization in 2006. The data shows that the highest numbers of new entrants were in the food, textile, plastic, and metal industries. It also shows that new firms entered areas with a higher average level of localization. It is also useful to map the geographical proximity of firms in certain areas: Figures 1 to 4 below show the geographic distribution of manufacturers in selected districts of Punjab. Previously established firms are represented by the red markers and new firms are represented by the yellow markers. The maps show that new firms enter areas where there is already a significant degree of industrial concentration, such as in the case of the Lahore and Gujranwala districts (Figures 1 and 2) as well as in the sports and food industries (Figures 3 and 4), which supports the idea that agglomeration has a significant impact on new firm entry. 5 Empirical Specification In order to analyze how local environment—measured by the agglomeration factors of urbanization and localization—and the socioeconomic indicators of a district affect the arrival and scale of operations we estimate the following Tobit models:  Arrivalid Aid   0  1localizationid   2urbanizationd idaspid X,543   (11) ididid onlocalizatiEsoperationofScale 10   idespidd Xonurbanizati ,5432   (12) where a  and   e are error terms,   4i and   4i are industry fixed effects, sp5  and sp5  are subprovincial fixed effects and  Xd represents the socioeconomic characteristics of a particular district. Equation (11) explains firm arrival (  Aid ) in industry  i and district  d , which is affected by localization, urbanization, and the socioeconomic characteristics of the district with industry fixed effects and sub-provincial fixed effects. Equation (12) is similarly interpreted with the dependent variable taken as the scale of operations of arrival (  Eid ). The first dependant variable,  Arrival (Aid), is the total number of newly established firms in industry i and district d, where a new entrant is defined as a firm that has reported its year of establishment as 2008. The second dependent variable, Scale of Operations (Eid), is the total employment of firms regarded as new entrants (according to the definition above) in 2008 for a specific industry i and district d. The measure of localization is constructed by aggregating employment in each industry for every district, which allows us to examine how the presence of the same industry leads to new firm 8 formation in a specific area. Urbanization is measured by the level of employment in the existing establishments within a particular district. This allows us to assess how the presence of different kinds of industries has led to new firm formation in a specific area. Localization and urbanization are initially incorporated as aggregate measures and then disaggregated into three levels of establishment: small, medium, and large. Or in other words we first look at the impact of the total number of other firms in an area (both within the same industry and in other industries) on new firm entry and size. This is followed by separately looking at the how the number of small, medium and large firms in an area (again in the same industry and across industries) affects new firm entry and size. This disaggregation is done because the impact of small firms in an area on new firm entry and size can be significantly different than the impact of medium and large firms in an area. One reason for this (discussed by Rosenthal and Strange, 2010) is that smaller firms tend to support the entry of other smaller firms because they rely more on shared infrastructure and agglomeration economies. Another potential reason for this differential impact is that large firms may be able to erect formal and informal barriers to new firm entry. In our analysis we define small establishments as those firms with fewer than 10 workers, medium establishments are defined as those firms employing between 10 and 49 workers, and large establishments are those employing 50 or more workers. In order to account for the socioeconomic factors (  Xd ) that affect the birth of new firms, we incorporate district-level controls. These include the average age of the population, the male percentage of the population, average income, unemployment rate, the percentage of the population with primary education, the percentage of the population with secondary education, and the percentage of the population with tertiary education. We also incorporate industry and sub-provincial region fixed effects to account for specific industry and regional characteristics that might have an impact on new firm formation in a specific industry and district. Examples of industry level heterogeneity include different barriers to entry, levels of innovation and technological shifts. 6 Results and Discussion The results of our models are presented in Tables 3 and 4 which report the marginal effects for the arrival and scale of operations models using the measures of local conditions. The coefficients of the various indicators of local activity measure the effect of adding 1,000 workers to the local environment with a given establishment size. The estimations analyze the impact of localization and urbanization at an aggregated level (where we look at the overall impact of all neighboring firms regardless of size) and at a disaggregated level (where we separately look at the impact of small, medium and large neighboring firms). Our first analysis looks at the impact of agglomeration on firm arrival and Table 3 reports the results for the arrival model estimated for the manufacturing industry in Punjab. The model incorporates the local environment for two time periods separately. So the results in columns (1), (2) and (3) measures the impact of the local environment in 2006 on firm arrival in 2008 and columns (4), (5) and (6) measure the impact of the local environment in 2004 on firm arrival in 20082. In columns (1) and (4) the model is estimated using aggregate measures of localization 2 As stated above, the reason for using 2006 and 2004 measures of local environment is to overcome the issue of reverse causality between the dependent variables (firm arrival and scale of operations) and the measures of local environment. 15 Table 2: Number of new establishments and scale of operations in 2008 and average localization in 2006 for each industry in the manufacturing sector in Punjab Industry New firms in 2008 Scale of operations in 2008 Average localization 2006 Meat, fruit, vegetables, oils/fats 15 425 358 Dairy products 1 200 158 Grain mill products and animal feed 52 919 383 Other food products incl. sugar and tea 75 2,724 2,033 Beverages 8 452 259 Tobacco products 0 0 44 Textile spinning, weaving, and finishing 19 519 9,613 Other textiles 11 358 2,002 Apparel 12 1,038 2,462 Tanning and leather dressing 1 15 301 Footwear 2 26 267 Wood products 2 27 111 Paper and paper products 1 45 178 Refined petroleum products 3 80 103 Basic chemicals 4 104 201 Other chemical products 10 506 358 Rubber products 1 14 43 Plastic products 21 341 295 Glass and glass products 1 200 115 Nonmetallic mineral products 5 447 518 Metal products 21 605 700 Special-purpose machinery 2 35 286 Domestic appliances 12 161 585 Electric motors, generators, transformers 0 0 222 Electricity distribution and control apparatus 5 264 509 Electric lamps and lighting equipment 0 0 113 Medical precision instruments 11 353 1,014 Bodies for motor vehicles and trailers 0 0 1 Parts and accessories for motor vehicles 13 538 423 Source: Punjab Directory of Industries, Government of Punjab, 2006 and 2010. 16 Table 3: Marginal effects of Tobit estimation: Impact of agglomeration on firm arrival for manufacturing industry in Punjab Arrival 2006 2004 (1) (2) (3) (4) (5) (6) Localization Aggregated localization 0.00083*** 0.00082*** Localization at small scale 0.01438*** 0.01029** 0.01435*** 0.01029** Localization at medium scale 0.00445* 0.00358* 0.00450* 0.00363* Localization at large scale 0.00036 0.00045* 0.00033 0.00043 Urbanization Aggregated urbanization -0.00013* -0.00014* -0.00013* -0.000141* Urbanization at small scale -0.00034 -0.00038 Urbanization at medium scale 0.00144* 0.00148* Urbanization at large scale -0.00041** -0.00041** Socioeconomic characteristics of a district Average age of population -0.0054** -0.00540** -0.00340* -0.00552** -0.00542** -0.00351* Percentage of male population -0.0035 -0.00381 -0.00144 -0.00354 -0.00374 -0.00135 Average income 0.00005*** 0.00005*** 0.00005*** 0.00005*** 0.00005*** 0.00005*** Unemployment rate 0.00117 0.00135 0.00141 0.00117 0.00136 0.00148 Percentage of population with primary education -0.00361*** -0.00284** -0.00149 -0.00362*** -0.00284** -0.00150 Percentage of population with secondary education 0.00410 0.00204 0.00036 0.00414 0.00206 0.00036 Percentage of population with higher education -0.00092 -0.00061 -0.00154 -0.00093 -0.00061 -0.00154 Constant 0.24267 0.24613 0.0857 0.24027 0.24304 0.08238 Industry fixed effects Yes Yes Yes Yes Yes Yes Sub-provincial regions Yes Yes Yes Yes Yes Yes Note: *** = statistically significant at 1% level, ** = statistically significant at 5% level, * = statistically significant at 10% level. Source: Authors’ calculations. 17 Table 4: Marginal effects of Tobit estimation: Impact of agglomeration on scale of operation for manufacturing industry in Punjab Scale of operations 2006 2004 (1) (2) (3) (4) (5) (6) Localization Aggregated localization 0.03193 *** 0.03118*** Localization at small scale 0.28418 0.21611 0.28383 0.21611 Localization at medium scale 0.17276** 0.14906** 0.17454** 0.15097** Localization at large scale 0.01924* 0.02398** 0.01809* 0.02285** Urbanization Aggregated urbanization -0.00141 -0.00164 -0.00141 -0.00163 Urbanization at small scale -0.03180 -0.03366 Urbanization at medium scale 0.06690** 0.06882** Urbanization at large scale -0.01405** -0.01420** Socioeconomic characteristics of a district Average age of population -0.14449* -0.14633* -0.09964 -0.14545* -0.14729* -0.10417 Percentage of male population -0.24568** -0.25051** -0.11673 -0.24458** -0.24874** -0.11289 Average income 0.00148*** 0.00146*** 0.00170*** 0.00148*** 0.00147*** 0.00172*** Unemployment rate 0.00496 0.01086 0.03089 0.00508 0.01117 0.03378 Percentage of population with primary education -0.06964 -0.05262 -0.03137 -0.07000 -0.05257 -0.03142 Percentage of population with secondary education -0.00251 -0.05089 -0.04094 -0.00160 -0.05074 -0.04143 Percentage of population with higher education 0.02118 0.02713 -0.03649 0.02115 0.02731 -0.03673 Constant 13.6650** 13.756** 5.71314 13.614** 13.670** 5.56554 Industry fixed effects Yes Yes Yes Yes Yes Yes Sub-provincial regions Yes Yes Yes Yes Yes Yes Note: *** = statistically significant at 1% level, ** = statistically significant at 5% level, * = statistically significant at 10% level. Source: Authors’ calculations. 18 Table 5: Marginal effects of Tobit estimation: Impact of agglomeration on firm arrival and scale of operation for manufacturing industry in Punjab with district fixed effects Arrival Scale of operations 2006 2004 2006 2004 (1) (2) (3) (4) Localization At small scale 0.00037*** 0.00038*** 0.00780 0.00798 At medium scale 0.00010* 0.00011* 0.00435* 0.00456* At large scale 0.00001* 0.00001* 0.00101*** 0.00098** Constant -0.00176*** -0.00180*** -0.06546*** -0.06735*** Industry fixed effects Yes Yes Yes Yes District fixed effects Yes Yes Yes Yes Note: *** = statistically significant at 1% level, ** = statistically significant at 5% level, * = statistically significant at 10% level. Source: Authors’ calculations. 19 Figures Figure 1: Location of manufacturing firms in Gujranwala, Punjab Note: Previously established firms are represented by the red markers and new firms are represented by yellow markers. The actual number of previously established firms and new firms is greater than the number of markers since multiple firms that are located close to each geographically can be represented by one marker. Source: Punjab Directory of Industries, Government of Punjab, 2010. Figure 2: Location of manufacturing firms in Lahore, Punjab Note: Previously established firms are represented by the red markers and new firms are represented by yellow markers. The actual number of previously established firms and new firms is greater than the number of markers since multiple firms that are located close to each geographically can be represented by one marker. Source: Punjab Directory of Industries, Government of Punjab, 2010. Figure 3: Location of manufacturing firms in Punjab’s food industry Note: Previously established firms are represented by the red markers and new firms are represented by yellow markers. The actual number of previously established firms and new firms is greater than the number of markers since multiple firms that are located close to each geographically can be represented by one marker. Source: Punjab Directory of Industries, Government of Punjab, 2010. Figure 4: Location of manufacturing firms in Punjab’s sports industry Note: Previously established firms are represented by the red markers and new firms are represented by yellow markers. The actual number of previously established firms and new firms is greater than the number of markers since multiple firms that are located close to each geographically can be represented by one marker. Source: Punjab Directory of Industries, Government of Punjab, 2010. 20 Please note: You are most sincerely encouraged to participate in the open assessment of this discussion paper. You can do so by either recommending the paper or by posting your comments. Please go to: http://www.economics-ejournal.org/economics/discussionpapers/2014-21 The Editor © Author(s) 2014. Licensed under the Creative Commons Attribution 3.0.