Manufacturing in a Natural Resource Based Economy: Evidence from Canadian Plants.

• Author: Moshiri, Saeed

1. INTRODUCTION

   Since the 1980s, myriad studies have investigated the oil price shock impacts on the economic performance in oil-exporting countries. However, the theoretical and empirical research on understanding the responses of firms to oil price shocks is limited. While most studies have focused on aggregate or sectoral analysis, only recently have researchers begun to examine the firm level impacts of the oil price shocks, thanks to the availability of large-scale micro data (see Feyrer et al., 2017; Arezki et al., 2017; Allcott and Keniston, 2018; Karaki, 2018; Cust et al., 2019; James and Smith, 2020). The firm level studies allow us to examine the heterogeneous responses of firms to the oil price shocks based on their individual characteristics, which can aid to understand how regional economies perform generally after the boom. They will also provide better understanding of the diverse evidence on total employment and trade effects (e.g. Weber, 2012; Marchand, 2012; Hartley et al., 2015; Komarek, 2016; Maniloff and Mastromonaco, 2017).

   In this paper, we contribute to the literature of oil price shock effects on economic performance by first developing a theoretical model to analyze firm level outcomes in economies affected by a resource sector, and second, by conducting an empirical analysis using a large firm level data-set. Our general equilibrium model with heterogeneous firms and a natural resource sector allows us to understand how different firms respond to oil price shocks based on their given productivity. Relative to other studies setting out heterogeneous or productivity differentiated firms (Kuralbayeva and Stefanski, 2013; Allcott and Keniston, 2018), we therefore leave behind the classical dichotomy of a traded and non-traded sector suggested by Dutch disease model (Corden and Neary, 1982; van Wijnbergen, 1984); all firms trade endogenously. The model assumes a resource boom as an increase in the oil price or an increase in total factor productivity in the oil sector, which prompts a reallocation of labor from manufacturing firms to the oil sector. It further allows a geographical factor to interact with these dynamics. Specifically, the oil sector pays a wage premium over the wage of the manufacturing sector, which is largest for firms that are in the same locality as the oil sector and diminishes with distance.

   We derive a number of predictions that reflect the interaction between the resource sector and the firm's performance including the export decisions, wages, and employment. For instance, the model predicts how a firm's choice on exporting responds to a resource boom, which after aggregating to a sectoral level gives an endogenous expansion of non-trading firms. Specifically, our model suggests that while wage rate and domestic sales increase for a given firm, only most productive firms continue to export. Also, employment increases in a given non-exporting firm, but the effect is ambiguous for exporting firms. All these effects are attenuated by a distance parameter, allowing for a differentiation between geographies.

   As a case study, we take the predictions of the model to a rich Canadian data source. Canada is a developed country with well-diversified economic activities across its provinces. Specifically, major oil extraction facilities are located in the western province of Alberta, while the eastern provinces of Ontario and Quebec are the major producers of manufacturing products. However, the oil and gas produced in Alberta are almost exclusively exported, mainly to the United States. In this sense, the Alberta oil industry is exemplary case of a booming sector that generates a local foreign exchange windfall (Dissou, 2010; Naim and Tombe, 2013; Beine et al., 2015; Carbone and McKenzie, 2016; Moshiri and Bakhshi-Moghaddam, 2018). From the Annual Survey of Manufacturers (ASM) of Statistics Canada, we obtained yearly plant level data of all manufacturing plants in Canada from 2000 to 2010. We estimate the relationship between economic performance of each plant and time-varying revenues in the natural resource sector and cross-section varying distance of each plant relative to Ft. McMurray, Alberta, as the main hub of the oil extraction activity. We find that on average plants tend to be negatively affected by a boom in the natural resource sector in terms of employment, total revenue, productivity and exports. However, there exists a great heterogeneity in plants' responses, indicating that some plants actually do rather well. We cannot attribute this effect exclusively to the tradability of the produced output or industry linkages. Instead, we find that plants that have above average levels of productivity at the beginning of the sample period do relatively better and increase their exports in response to the resource boom.

   The organization of the paper is as follows: Section 2 presents a theory on firms' responses to the oil price shocks and the derived propositions. Section 3 reviews the data and presents the econometric model and estimation strategy, Section 4 presents and discusses the results followed by concluding remarks in Section 5.

2. THEORY

   The model follows closely that of Haaland and Venables (2016), with a Melitz (2003) type manufacturing sector that is characterized by monopolistic competition and consists of heterogeneous firms producing differentiated goods using a single factor, l. (1) We assume that each firm is composed of a single plant. We add an oil sector that produces an undifferentiated good, oil, which is exported in its entirety at an exogenously given price determined in the international market. (2) For simplification, we assume that there are no input-output linkages between the oil sector and manufacturing firms. In the case of Canada this is justifiable given that in Canada a large share of the extracted oil is directly exported to the US. (3)

   2.1 Preferences

   Preferences over the differentiated manufacturing goods take the Constant Elasticity of Substitution (CES) form:

   [Please download the PDF to view the formula],

   where [SIGMA] is the set of available varieties, q([nu]) is consumption of variety v, and [sigma] is the elasticity of substitution. Consumers can consume both domestically produced and imported varieties and do not differentiate between products origins. Optimization yields the individual good demand q([nu]) = p[([nu]).sup.-[sigma]]E[P.sup.[sigma]-1], where E is the expenditure on manufactured goods and [Please download the PDF to view the formula].

   2.2 The manufacturing sector

   The manufacturing sector consists of heterogeneous firms that produce distinct varieties subject to monopolistic competition. While all firms sell in the domestic market D, only a selection of firms sell in the export market X. Production of each variety incurs fixed and variable costs. The fixed cost, [f.sub.j], is common to all firms selling in the same market j [member of] (D, X), but variable costs vary with firm productivity [psi] [member of] (1, [infinity]). The cost function is given by

   [c.sub.j] = ([f.sub.j] + q /[psi]) [w.sub.m],

   where [w.sub.m] is the sector wage. Maximizing profits, each firm sets a price with a constant markup over the marginal costs of selling in market j, [p.sub.j] ([psi]) = ([sigma] /([sigma] - 1))[[tau].sub.j][w.sub.m] / [psi] where [[tau].sub.j] is an iceberg trade cost. We have [[tau].sub.D] [equivalent to] 1 for the domestic market and [[tau].sub.X] [equivalent to] [tau] > 1 for the export market. The firms' revenue is then given by:

   [r.sub.D]([psi]) = [sigma][zeta][([w.sub.m] / [psi]).sup.1 - [sigma]]E[P.sup.[sigma] - 1], [r.sub.X]([psi]) = [sigma][zeta][([tau][w.sub.m] /[psi]).sup.1 - [sigma]]E[P.sup.[sigma] - 1],

   where [zeta] [equivalent to] [([sigma] - 1).sup.[sigma] - 1] [[sigma].sup.-[sigma]] and E and P are the fixed expenditure level and price index that exporting firms face in the world market. (4)

   Foreign firms producing for the domestic market face a similar cost function as the domestic firms. They face a fixed cost [f.sub.M] and an iceberg cost [[tau].sub.M] to enter the import market. The foreign wage is taken as given and normalized to 1. Serving the import market yields revenue given by:

   [r.sub.M] ([psi]) = [sigma][zeta][([tau]M / [psi]).sup.1 - [sigma]] E[P.sup.[sigma] - 1],

   As in Melitz (2003), firm productivity is determined by a lottery. After paying a sunk cost [f.sub.E] to develop a variety, each firm draws its productivity level [psi] from the cumulative Pareto distribution G([psi]) = 1 - [(1 / [psi]).sup.k]. We make the standard assumption that k > [sigma] - 1. Because of the fixed market entry costs, only firms that draw a sufficiently high productivity [psi] will decide to produce. The productivity cut-offs [[psi].sub.j] are the lowest levels of productivity at which the firm's profits from serving market are non-negative. Noting that the firm's operating profit is a fraction 1/ [sigma] of its revenue, we see that the productivity cut-offs are determined by:

   [r.sub.D] ([[psi].sub.D]) / [sigma] = [w.sub.m][f.sub.D], (1)

   [r.sub.X] ([[psi].sub.X])/[sigma] = [w.sub.m][f.sub.X], (2)

   [r.sub.M] ([[psi].sub.M])/[sigma] = [f.sub.M]. (3)

   We restrict the parameter values such that they generate selection into export markets, i.e. [[psi].sub.D]

   [Please download the PDF to view the formula]. (4)

   The total value of output sold by domestic firms in the domestic market D, in the export market X, and by foreign firms in the domestic market M are given by:

   [Please download the PDF to view the formula], (5)

   [Please download the PDF to view the formula], (6)

   [Please download the PDF to view the formula], (7)

   where N and N denote the mass of firms in the domestic and foreign economy. (5) Note that total expenditure must equal domestic and import sales and the wage bill must equal the total value of sales by domestic firms:

   E = D + M, (8)

   [w.sub.m]l = D + X. (9)

   2.3 The oil sector

   The oil sector is...