An Assessment of the Impact of Trade Liberalization: The Relationship Between Mexico and the Manufacturing Industry of the Rust Belt

(1)

An Assessment of the Impact of Trade Liberalization:

The Relationship Between Mexico and the Manufacturing Industry of the Rust Belt

Master’s Thesis

MSc Public Administration: International and European Governance

Leiden University

Faculty of Governance and Global Affairs

Author: Raymond Durham (s1990683)

Supervisor: Dr. Alexandre Afonso

(2)

1

Abstract

Embedded in our current era of Trump politics is a consensus in the realm of public discourse that trade liberalization agreements like NAFTA have had a negative impact on portions of the United States economy. A broad survey of the literature examining trade liberalization policy indicates that there may be some truth to this rhetoric. The current study investigates if there is a significant association between trade with Mexico and any changes experienced in the labor market of the manufacturing industry in the Rust Belt states. Despite the large amount of literature dealing with trade liberalization and its effects, there has not been much research directed at examining sectors within regional economies. Therefore, this study seeks address this research gap by measuring the effects of trade with Mexico (considering the trade deficit, amount of import trade, primary trade partner and size of economy) on the labor market of manufacturing industry in Rust Belt states. To test the hypotheses formulated in this study, several multiple linear regression analyses were conducted. Each multiple linear regression analyses were used to gain insight into whether trade with Mexico has any association with the number of manufacturing firms, number of manufacturing jobs, economic output of the manufacturing industry and average annual wages of manufacturing workers. The results show that trade with Mexico was able to predict variation in the number of manufacturing firms and jobs in the Rust Belt states.

(3)

2

CHAPTER 1: INTRODUCTION

In the bid for the United States Presidency, Donald Trump ran on a platform against establishment politics and defying the norms of political correctness. When Donald Trump declared NAFTA “one of the worst trade agreements in history” he rejected the norms of traditional economic policy in the United States. For the previous two decades, NAFTA had represented the central policy in the embrace of capitalism and trade liberalization between North America’s three major economic powerhouses. Although the American economy has in many ways benefitted from free trade and elimination of tariffs, many have felt left behind from these agreements. That is why, on the campaign trail and in the run up to the 2016 Presidential Election, Donald Trump repeatedly railed against NAFTA. This message was particularly felt in the industrial heartland, the Rust Belt, where many factories have closed, and jobs disappeared.

During the campaign, the signatory members of NAFTA, Mexico and Canada, also became swept up in the storm of criticism created by then candidate Donald Trump. Trump infamously opened the campaign by declaring his intent to eliminate NAFTA saying, “I love the Mexican people, but Mexico is not our friend. They're killing us at the border and they're killing us on jobs and trade. FIGHT!”. The criticism continued through the campaign, with Mexico becoming the primary scapegoat for the deficiencies in the manufacturing industry. Even several months before the election, Trump was still maligning Mexico for the economic strife felt by so many in the American economy. “Businesses are leaving our country like they’ve never left before. They’re going all over. They’re going to Mexico, that I can tell you. Mexico is like the eighth wonder of the world” he said. This rhetoric, accurate or not, resonated and would eventually propel Donald Trump into the White House.

Over the past several decades, academic research has focused intensely on the impact of international trade on United States labor markets (Feenstra, 2010). During the 1990s, and after the implementation of NAFTA, economist began examining the forces behind rising inequality in the periods following economic liberalization (Autor et al., 2013). With free trade and economic liberalization came increased levels of trade,

(6)

5 competition and specialization (Sachs and Warner 1995). This phenomenon could be demonstrated in the effects on individual industries throughout the United State economy. With tariff-free trade a newly minted possibility, the threat of firms to leave for greener pastures with lower production costs became a reality. Outsourcing to foreign countries has become a consideration in relationship between changes of the labor market and trade (Lawrence and Slaughter, 1993; Berman, Bound and Griliches, 1994; Feenstra and Hanson, 1999). However, prior to trade agreements like NAFTA, scholars had considered trade with low-wage countries as having a low impact due to small amounts of imports (Krugman, 2000).

Logically any form of free trade between countries, despite income level, could have a significant impact on wages and employment (Autor et al., 2013). Within this the divide between rich and poor countries, low-wage and high-wage, there are demonstrated more significant impacts on labor markets. In fact, parts of trade theory point to low-wage countries as potential disruptors of labor markets (Krugman, 2008). Additionally, research indicates that low-skill workers will be disadvantaged by free trade agreements due to the potential to lower the cost of production resulting from lower wages (Freeman, 1995). Furthermore, some studies have found that free trade agreements tend to enlarge trade imbalance at industry level (Song and Zhao, 2012). There is a robust argument to be made that trade can significantly change substantial portions of an economy, particularly when countries enter into free trade agreements. For these reasons, amongst others, this thesis will examine the effects of trade on the manufacturing industry of the United States.

1.1 R

ESEARCH

Q

UESTION

The victory of Donald Trump in the 2016 United States Presidential Election sent shockwaves throughout body politic. Particularly due to the unexpected changes that took place since the last presidential cycle in voting preferences, which were most notable in the Rust Belt. In 2012, President Obama carried many of the states in the Rust Belt, including Illinois, Michigan, New York, Ohio, Pennsylvania and Wisconsin. In 2016, the United States observed a paradigm shift, something in the electorate had changed. Was Trump able to

(7)

6 sell himself as a transformative tour de force? Or was it discontent based on the policies of the Obama administration? Many point to Trump’s success in winning over voters in America’s manufacturing heartland with his anti-NAFTA policies (Cerrato, Ruggieri, and Ferrara, 2016). In many of the areas required to deliver the victory, manufacturing workers experienced loss of jobs to Mexico (Ibid).

This thesis will explore the relationship of trade between the United States and Mexico and changes in the manufacturing labor market of the Rust Belt. If we are to accept the theory that Donald Trump’s pro-manufacturing and anti-NAFTA rhetoric made a marked difference, then surely there is empirical data in the labor market of the manufacturing industry in support. More specifically, is there empirical data in the labor market of manufacturing industry that underscores the reason why Trump was so successful in the Rust Belt. From this line of inquiry, a more specific research question can be constructed:

RQ: “To what extent, between the years 2012 and 2016, does a relationship exist between trade conducted with Mexico and observable changes in the manufacturing industry of the Rust Belt states.”

From this research question, this thesis will seek establish a more direct link between the rhetoric of the Trump campaign, trade conducted with Mexico, and empirical data surrounding the manufacturing industry that will be analyzed.

1.2 R

ELEVANCE

Over the past half-century, public administration research has focused on examining changing global, political, and economic dynamics. Whether that be the application of classical theoretical approaches such as realism and liberalism, toneorealism (structural realism) and neoliberalism. Theory and the many research methods to explain global phenomena have evolved over time. Additionally, theory has become more targeted and specific. This explains why scholars have chosen to investigate refined areas of policy such as the overlap of

(8)

7 trade, immigration, and economic conditions (Ottaviano and Peri, 2006; Card, 2009; Peters, 2015; Peters, 2017).

The overlap of trade and economic conditions as an area of study has proven to be very important in understanding the results of policy and government action in a globalized world. This is particularly applicable when studying free trade policy, like NAFTA, and its effects. Due to this, researchers have focused on researching nations that enter into international trade agreements to explain changes in economic markets (Feenstra, 2010). There is a substantial link between economic liberalization, entering into free trade agreements and observable changes in a nation’s labor market (Peters, 2015; Peters, 2017). In addition, the governance framework and policies it has developed have an impact on international trade and business (Sanders and Carpenter, 1998). One of the prevailing hypotheses coming from these policies being that increased trade liberalization would increase economic growth and expansion, which results in differential gains within individual countries (Shatz, 2006).

An important area of examination relating to this study is the observed changes in the labor market, especially the number of jobs and firms within an industry. Trade is often viewed as creating net job gains or losses, but it doesn’t necessarily have that effect (Shatz, 2006). Ultimately, multi-lateral trade agreements can unintendedly target certain industries due to shifts in resources and between regions (Krugman and Livas, 1992; Rauch, 1993; Hanson, 1996) In fact, studies have shown that firm-level employment can be directly affected by trade liberalization (Revenga, 1997). Understanding the changes in number of jobs and movement of firms can be tied back to the shifts of production abroad (Feenstra and Hanson, 1996; Feenstra and Hanson, 1999). Moreover, the shift of production abroad can be influenced by the tangible and intangible resources of a firm (Dunning, 1988; Lu and Beamish, 2001). This phenomenon is known as outsourcing, or the shift of jobs out of one country and into (Lawrence and Slaughter, 1993; Feenstra and Hanson, 1996). Identifying and explaining the properties surrounding outsourcing is an essential component to understanding effects of trade liberalization on workers and firms.

(9)

8 Another area of examination related to the relevance of this study is the change in the composition of the labor market due to trade liberalization. One of the primary arguments surrounding these subjects claims that trade influences the types of jobs available and the industrial structure (Shatz, 2006). The research of the economic performance of less-skilled workers, like those in the manufacturing industry, is an important area of academic focus (Feenstra and Hanson, 1999). Trade theory and empirical data argues that when trade barriers are removed, industries that employ high-skill workers will receive gains, which will result in new jobs created with higher incomes (Lawrence and Slaugher, 1993; Shatz, 2006). Due to this occurrence, an important question many researchers have investigated is: what happens to those who don’t fall into the category of high-skill workers? Research has identified markedly negative effects on low-skill workers, or workers in declining industries (Feenstra and Hanson, 1999; Shatz, 2006). During the 1990s, empirical data and research indicated that the slow growth of middle-income jobs is rooted in the concurrent decline of manufacturing and growth of low-income jobs in the services industry (Wright and Dwyer, 2003). Therefore, it is important to examine shifts in the labor market to understand the effects of trade with Mexico on the manufacturing industry in the United States.

One of the areas of examination related to this study, that is perhaps the most relevant in our current economy, is wages. This is because wages are becoming increasingly polarized and directly affect an individual’s ability to earn a living. Research has demonstrated a direct tie of trade to wages and wage growth, particularly income distribution (Shatz, 2006). Further, research has also shown a direct tie between economic integration and economic convergence, which can affect wages (Sachs and Warner, 1995; Feenstra and Hanson, 2001). This is because economic convergence theory states poorer countries tend to grow faster than rich countries, thus so will their wages and the proportionate income gap would close over time (Sachs and Warner, 1995). However, this is not always the case. Since the implementation of NAFTA real wages and growth have remained far lower than in the 1980s, when many free trade agreements began being negotiated (Krugman, 1993). Perhaps, wage changes could also be explained by the shifting down of industry production and lack of labor demand (Revenga, 1997). Some scholars argue this circumstance has disproportionately affected workers who can’t risk losing their jobs and are willing to accept less pay (Freeman, 1995; Shatz,

(10)

9 2006). In sum, an argument that should be explored is whether economic integration, or trade liberalization with Mexico, had any effect on wages.

This study provides value-added by examining more in-depth the effects of trade with Mexico on the manufacturing industry, specifically in the Rust Belt. The refined nature of this study provides the ability to observe more individualized effects of economic liberalization and free trade with Mexico on an industry level. By analyzing labor market figures related to manufacturing like changes in the number of firms, jobs, economic output and wages, a more specific narrative can be drawn. Moreover, the examination of regionality is important because some areas of the United States could be disproportionately affected by trade due to the size of industries within their respective economies. Additionally, this study provides the ability to observe effects of trade with Mexico on a group of individual states. From this study, inferences can be potentially drawn on whether any observed effects provide any empirical backing to political rhetoric that arguably changed the course of a Presidential campaign and the face of American politics.

1.3 T

HESIS

S

TRUCTURE

The trade relationship between Mexico and the Rust Belt is not a frequently studied area within the field of public administration. Thus, to introduce the reader of this thesis to the topic the first chapter will include an introduction with context surrounding the aim of the study, the research question and the relevance of this study. In chapter two, the theoretical backing of this study will be detailed, beginning with definitions, followed by the rationale for the hypotheses generated and a summary of the hypotheses this study will explore. In the third chapter of this study the design of the research conducted will be explained. Subsequently, the results and analysis of the data will be detailed in the fourth chapter. The fifth chapter will provide a discussion of potential areas of further inquiry related to this study. Finally, this thesis will conclude with a summary of findings and potential inferences that could be drawn. This concluding chapter will be followed by the

(11)

10 references used in this study. Any further data or related information to this study will be located in the Appendix section at the end of this thesis.

(12)

11

CHAPTER 2: THEORY

This chapter begins by defining the manufacturing industry, the Rust Belt and their significance to the American economy. The background and definitions provided in these two sections should clarify some of the concepts central to this thesis. In the subsequent sections, a general literature review will be provided to demonstrate the research and studies underpinning the influence of trade between the United States and Mexico and the labor market of the manufacturing industry in the Rust Belt. Each one of these factors present information to generate convincing empirical arguments related to the effects on the manufacturing industry. After the research for these factors is presented, that information will presented as hypotheses that will be further explored. Finally, the last section of this chapter will summarize the hypotheses addressed by this thesis.

2.1 D

EFINITIONS

This section serves to better detail terms and concepts central to this thesis. These concepts include defining the manufacturing industry and the Rust Belt. Establishing concrete definitions of these terms and concepts is vital for the success of this study and the understanding of the reader.

2.1.1 The Manufacturing Industry

In 1790, when President George Washington addressed the first session of Congress he noted that the “safety and interest” of a free people “require that they should promote such manufacturers as to render them independent of others for essential, particularly military supplies” (CRS, 2017, p.1). During that time, a young nation was attempting to establish itself on the global stage. The means to that end, as Washington indicated, was self-sufficiency through the production of goods. Since, the manufacturing industry has been the center

(13)

12 of much economic focus over the course of the history of the United States. In fact, the Congressional Research Service indicates that the “scope and scale of manufacturing have changed considerably in the intervening centuries”, as the industry has established itself as a central component of the United States economy (Ibid).

Many consider manufacturing to have a central role in the economy of the United States, even in a more competitive globalized world with cheaper production costs. This is due to observed heightened levels of productivity relative to other sectors of the economy (CRS, 2017). This productivity growth, in theory, fuels economic growth within the entire American economy (Ibid). However, despite the stable share of GDP manufacturing has occupied over the last 50 years, if not for spectacular growth in several subsectors of the industry a different story could be told (Ibid). In roughly 90 percent of the manufacturing industry productivity and growth have been slow, thus, the share of GDP has fallen (Ibid). Over the last half-century, the manufacturing industry has demonstrated two competing components: 1. The US manufacturing industry has had positive growth in real output; and 2. There has been a steady decline over several decades in the size and number of job opportunities within the industry. (Baily and Bosworth, 2014).

Despite evidence demonstrating China as the largest manufacturer in the world and the American manufacturing industry shrinking in size, the United States remains the world’s largest manufacturing in terms of value-added (Baily and Bosworth, 2014). One of the largest causes in the reduction of the manufacturing industry are trade deficits, driven by low exports as much, if not more than high levels of imports (Ibid). Point of fact, the United States manufacturing industry has accumulated a massive trade deficit, growing from $316 billion in 2000, to upwards of $460 billion in 2012 (Ibid). Additionally, many of the largest US corporations continue to move their production facilities overseas (Ibid). It is no coincidence that ballooning trade deficits have been correlated with a large drop in the level of manufacturing employment (Ibid). This intertwined nature of trade deficits, trade liberalization, and other factors have changed the way American consume, which has not helped the state of the manufacturing industry either (Baily and Bosworth, 2014). Large amounts of imports driving trade deficits has demonstrated that if the United States manufacturers brought jobs back,

(14)

13 exports would increase, and it may be more profitable to produce in the United States (Ibid). For purposes of this these we will examine the confluence of these factors and effects, intended or unintended, as they relate to trade deficits with Mexico.

Defining what manufacturing is and the industry it encompasses is a difficult task. As economies have become more complex and production processes more involved, pinpointing an exact definition of manufacturing has proven difficult. The United States government defines manufacturing within federal law as the process of physically transforming goods (CRS, 2017). This could be perceived as a catch all phrase, but actual physical transformation is rather specific as it is encompassed by manufacturing activities such as molding, cutting and assembly (Ibid). For defining the entire manufacturing industry, the United States government is a bit more specific defining it as comprising “establishments engaged in the mechanical, physical, or chemical transformation of materials, substances, or components into new products,” in addition to the “assembling of component parts of manufactured products” for purposes other than construction (Ibid, p. 2). Fort et al. (2017) has a much more concise explanation of manufacturing and its firms, defining them as including all firms ever observed to have a manufacturing. There is some nuance to definition within literature.

For purposes of this paper we will define the manufacturing industry by utilizing all the previously mentioned criteria and the North American Industry Classification System (NAICS) code 133. The NAICS is the standard used by federal statistical agencies like the Census Bureau for classifying industry. By combining the general definitions presented by the literature, the United States government and its agencies, a more accurate estimation of the firms and workers within the manufacturing industry can be assessed.

2.1.2 The Rust Belt

One of the most scrutinized geographic regions of the United States over the past half-century, in terms of economic study, can be found in what is known as the “Rust Belt”. Despite the large amount of study there

(15)

14 is some debate as to what states define the region. Crandall (1993) identifies the Rust Belt as a group of states in the Northeast and Midwest. Whereas, Feyrer (2007) considers the Rust Belt to be concentrated in Illinois, Indiana, Michigan, Ohio and Pennsylvania. In some circumstances, this area can extend to counties located in upstate New York (Feyrer, 2007). Some scholars hold a much stricter definition, including the traditional states of Ohio, Indiana, Illinois, and Michigan (Kenney and Florida, 1991). More recently, Trubek (2018) has clarified the definition of the Rust Belt by stating that “Michigan, Ohio, and Pennsylvania are central to the region, as well as parts of Illinois, Wisconsin, and New York” (p. 2).

It is important that this thesis provide a more exact definition than that of Crandall (1993) and others. For the purposes of this thesis the “Rust Belt” will be defined more broadly by including the states of Illinois, Indiana, Michigan, Ohio, Pennsylvania, New York and Wisconsin. This approach contradicts defining the region by a more traditional approach by only including Ohio, Indiana, Illinois and Michigan, such as in Kenny and Florida (1991). This thesis argues that a broader definition will enable more variation, generating a more comprehensive analysis, while maintaining the integrity of the Rust Belt. Thus, this thesis will include more states within the definition of the Rust Belt to provide more instructive findings as to the effects of the trade deficit with Mexico and its effects on the region.

2.2 T

RADE

L

IBERALIZATION AND

O

FFSHORING

When examining the relationship of trade and changes to the structure of economies a term that repeatedly presents itself is offshoring. Within the literature there are varying definitions presented to describe this phenomenon, but many rely on the same core ethos. According to Feenstra and Hanson (1999) offshoring is considered the transfer of economic activity to another country. This often manifests in the moving of low-skilled jobs to low-wage countries, which increases the demand for low-skilled labor domestically (Feenstra and Hanson, 1999). However, it is not simply the moving of entire production facilities and jobs to other countries, the practice is much more nuanced. In recent decades, many companies have begun offshoring increased

(16)

15 amounts of their production process to foreign countries (Feenstra, 1998). For instance, the automobile industry purchases many parts that used to be made in the United States from other countries, then assembles the automobiles domestically (Feenstra and Hanson, 1999). Additionally, many do not think the common practice of importing steel for production as offshoring, but it has the same effects (Ibid).

Due to the impact of offshoring on economies it has become a subject of serious consideration by scholars. There is a rich debate concerning the causes of offshoring that has unfolded particularly within the last several decades. Throughout the literature one can find little to refute that offshoring should have a marked impact on economies (Feenstra and Hanson, 1996; Feenstra and Hanson, 1997). Furthermore, almost all the literature relates offshoring to changes in economic policy. In the United States, according to Feenstra and Hanson (1997) trade policy reform, in theory, has allowed the companies from within the United States to move more production to Mexico. However, offshoring from the United States economy to other lower-wage economies is a much more complicated process than can be attributed directly to policies like NAFTA.

Many scholars who study the impact of trade on economies indicate an increase in trade as a source of offshoring. Over the course of time, trade throughout the world has increased dramatically as economies become more integrated, similar in size and nature (Helpman, 1987). Research studying OECD and non-OECD countries alike has provided considerable support for this hypothesis (Hummels and Levinsohn, 1995). However, not all scholars in academia are in agreement that an increase trade has made a significant difference. The findings in Krugman (1995) dispute the theory central to Helpman (1987), arguing that “it would be hard to argue that the sheer volume of trade is now at a level that marks a qualitative difference from previous experience” (p. 331). In contrast, detracting scholars argue that it is the disintegration of production itself that can lead to more trade (Feenstra, 1998). There is little dispute of the effects of increased trade, however, when one examines the increase of trade between the United States and Mexico in the last several decades there is some debate (Hanson, 1996). Much of the increase in trade has resulted from firms in the United States establishing export assembly operations in Mexico (Ibid).

(17)

16 When determining what other factors account for offshoring, Feenstra (1998) argues that one possibility could be trade liberalization. For much of the 20th_{century trade was closed off, with much of}

production coming from domestic sources. With trade liberalization new economic opportunities emerge, which can include increased trade and offshoring (Feenstra, 1998). Not only that, but Pavcnik (2002) notes that new levels of trade liberalization not only increased trade but had a noticeable effect on the production process or “value chain”. That is because through increased trade liberalization, manufacturing has shifted across borders, exporting goods primarily in the intermediate stages of production (Ibid). Compounding this issue, trade liberalization has significantly reduced the price of imported materials and machinery, while exposing many producers to increased foreign competition (Ibid).

Correspondingly, much of the observable effects of offshoring can be felt in the immediate vicinity of the United States, more specifically with trade conducted in North America. Hanson (2003) argues that the elimination of trade barriers within the region could induce economies to specialize more in export production. This is because changes in policies, like the adoption of NAFTA, allowed Mexico to become more specialized in export production (Hanson, 2003). In addition, fewer restriction on the amount of foreign development and investment (FDI) allowed for Mexico to become a greater part of the production networks in North American (Ibid). Due to trade liberalization, investment and development of Mexico became more viable due reduced fears of spikes in tariffs, which had been a significant deterrent to integrating the two economies since the 1980s (Fort et al., 2017).

Indeed, the central cause of offshoring as it relates to trade liberalization policies like NAFTA is the reduction or elimination of tariffs. When examining trade bi-laterally with the United States and other trade partners in the region, about 40 percent of the growth of trade can be explained by falling tariffs and reduced transportation costs (Rose, 1991; Baier and Bergstrand, 1997). With less distance to travel and little to no tariffs, there is an incentive financially to conduct more trade with countries close to the United States. Hanson (1996) notes that prior to NAFTA, trade was primarily dictated by the tariff practices in bi-lateral trade agreements. Some contrarian studies argue that tariff reduction does not have a large effect, as many goods

(18)

17 would still be provided in the United States no matter where they are manufactured (Quinn and Guile, 1988). Feenstra (1998) argues that it should be noted that lowering tariffs has an important effect on trade, far more than other factors like transportation costs, but it doesn’t give the complete picture.

Increasingly, one of the areas that has felt the biggest impact from offshoring has been in manufacturing firms. Many of these firms have been affected by the changes in policy and have had to move abroad. Studies have shown that there is importance placed in producing internationally (Feenstra, 1998). This could be due to the greater competition United States firms are facing from low-wage countries, which makes them much more likely to change locations or upgrade (Bernard et al., 2006; Khandelwal, 2010); Bernard et al., 2011). A lot of the recent growth in trade can be accounted for from the breaking of industries apart by putting low-skill production in low-wage countries and high-skill production in high-wage countries (Feenstra and Hanson, 1997; Feenstra and Hanson, 2001). The fast shift by firms to international production has prompted the creation of phrases like: ‘‘kaleidoscope comparative advantage’’ (Bhagwati and Dehejia, 1994); ‘‘Slicing the value chain’’ (Krugman, 1996); ‘‘Delocalization” (Leamer, 1996); and ‘‘Intra-mediate trade’’ (Antweiler and Trefler, 1997).

Through many studies, labor supply has shown to dictate how multinational firms begin to choose their location of production (Markusen, 1995; Markusen and Venables, 1996). The firms that are especially vulnerable to the increased competition of low-wage countries could cease to exist if the do not adjust (Fort et al., 2017). The adjustment for import-competing firms producing goods is often a choice between relocation or “firm death” (Ibid). Haltiwanger et al. (2013) define a firm death as all of the departments within a firm ceasing to exist. This is particularly notable with firms that have plants that are tied directly to the production of specialized products (Fort et al., 2017). Many argue that to encourage firms to locate their production within the United States, tariffs must be increased (Fort et al., 2017). When considering the effect of NAFTA on firm changes, it is important to note the possibility of firms adjusting to the market the treaty was formally approved (Hanson, 2003). In fact, about 75 percent of the decline in manufacturing employment

(19)

18 between 1977 and 2012 can be attributed to “firm death”, which was particularly focused in firms that existed prior to 1977 (Fort et al., 2017).

It stands to reason, if one considers the body of literature discussing outsourcing in this section, that there could have been some effect on the number of firms from the trade conducted with Mexico. Amongst others, Feenstra and Hanson (1997) and Feenstra and Hanson (2001) argue that much of the breaking apart of industries, like the manufacturing, by putting low-skill production in low-wage countries and high-skill production in high-wage countries. Furthermore, research has demonstrated that certain industries are industries have more exposure than others from trade (Autor et al., 2014; Acemoglu et al., 2015). In theory, this exposure should extend to regions, or states, that a large portion of their economy is focused in industry that has a high degree of exposure like manufacturing. Thus, one could argue that trade with Mexico should have a measurable association with the number of manufacturing firms in the Rust Belt states. Thus, the first hypothesis in this thesis that will be tested is:

H1a: There is a significant association (α< 0.05) between the trade conducted with Mexico and

variation in the number of manufacturing firms (per 100,000 people) in a Rust Belt state, adjusting for trade deficit, amount of import trade, the primary trade partner, and size of the economy (GDP).

The corresponding null hypothesis that will be tested being:

H10: There is no significant association (α> 0.05) between the trade conducted with Mexico and

variation in the number of manufacturing firms (per 100,000 people) in a Rust Belt state, adjusting for trade deficit, import trade, the primary trade partner, and size of the economy (GDP).

2.3 T

RADE

L

IBERALIZATION AND

E

CONOMIC

S

HIFTS

There is extensive literature to suggest that offshoring and trade liberalization have devastating effects on the job market. As firms locate their production apparatus in other countries, globalization can be manifested in

(20)

19 the severe decline of employment in sectors (Fort et al., 2017). Empirical information backs this claim, with research in Fort et al. (2017) finding that over the period of 1984-1990, the manufacturing plants that were in industries with the largest reductions in tariffs had the highest reductions in employment. This could be due to the finding that firm death drives levels of employment (Fort et al., 2017). Of course, firms are not the only factor that dictate employment levels in individual economic sectors. However, there is an observable trend that has played out over time that indicates a decline in the share of total employment attributable to manufacturing (Hanson, 1996; Kakaomerlioglu and Carlsson, 2006; Baily and Bosworth, 2014).

The change in number of manufacturing jobs, according to many, can be reduced to the supply and demand of certain kinds of labor in the economy. Within trade between richer and poorer countries there can often be a divide in the availability of low-skill labor. For instance, compared to the United States, Mexico has much larger supply of low-skill labor and human capital (Hanson, 2003). Leamer (1993) argues that trade policies like NAFTA should theoretically benefit Mexico due to their presumed comparative advantage in supply of low-skill labor and manufacturing. Before the implementation of NAFTA, trade protections were in place that provided less exposure for the low-skill labor market of the United States (Hanson, 2003). Since the removal of trade barriers, investment and reform could be attributed to the demand for low-skill labor in Mexico (Ibid). Hanson (2003) argues that this is because Mexico appears to have a cost advantage in production. In addition, relative to the services in the United States economy Mexico has an upper hand for assembly services (Ibid). Simply put, the Mexican manufacturing industry has become a subcontractor for the rest of North America (Hanson, 2003).

Interestingly, demand for different levels of skill through trade liberalization can make for a compelling lens to examine changes in wages. Feenstra and Hanson (2001) argues that there is evidence from observing the changes in trade during the 1980s and 1990s altered industry productivity, pricing, and in turn the demand for skilled labor. This could be in part because free trade has significant effects on labor demand in import-competing industries (Feenstra and Hanson, 2001). Feenstra (1998) indicates that from the perspective of observing the United States labor market through scarce factor, the competition from low-wage

(21)

20 countries lowers the demand for low-skill labor. Feenstra and Hanson (2001) also suggest that international trade, in part due to the ability to outsource, contributes to the divide between low and high-skill demand. The elastic nature of labor demand, as detailed by Slaughter (1997), is affected by increased trade. Perhaps even more cynical, Shatz (2006) argues that workers with less skills and in declining industries have trouble finding jobs and lack value to other employers.

Notably, much of academic research contradicts the rhetoric of those in the United States that supported NAFTA. The argument often made is that imports from low-wage countries like Mexico were good because they created jobs (Burfisher et al., 2001). However, now much of debate surrounding free trade concerns the impact on employment and wages of low-skill workers (Feenstra, 1998). Hanson (1996) suggests that NAFTA will contribute to moving firms in the United States to the border of Mexico, perhaps out of the country. The data demonstrates that borders have lower cost access to foreign markets, which make them natural production sites (Hanson, 1996). Thus, as NAFTA sought to better integrate the Mexican and United States markets, the border region became an ideal production area (Ibid). This phenomenon is not a unique problem experienced by the United States, many advanced economies have experienced a similar phenomenon (Kakaomerlioglu and Carlsson, 2006). The increased import penetration provided by NAFTA disproportionately affects certain industries, causing decreased employment and output (Fort et al., 2017). Furthermore, there is extensive research to suggest that firms in related industries will spatially agglomerate, as will the jobs in those firms (Henderson, 1982; Rauch, 1989; Venables, 1993).

When Mexico eliminated trade barriers and lifted restrictions on foreign investment and ownership, a large export manufacturing industry was established domestically (Hanson, 1996). This industry relied on the specialization of factories in the assembly of foreign-made parts and components (Ibid). This cut directly into the jobs of the Rust Belt as the breakup of the region’s manufacturing industry could have been caused by losses of employment not just to Mexico, but also the south and west of the United States (Blanchard and Katz, 1992). The liberalization of trade threatened the employ of low-skill workers in the United State, as they were not forced to compete with the abundance low-skill workers, resulting in lost jobs (Shatz, 2006). As

(22)

21 Feenstra (1998) argues the decision of many firms to move their production internationally will most certainly impact employment of such firms domestically. This in turn would impact the relative demand of low-skill labor in developed countries, in much the same way as replacing workers with automation (Feenstra, 1998).

More influential than automation, over the last several decades there has been a dramatic shift from a manufacturing-based economy to one that is more services oriented. In countries like the United Kingdom, Prime Minister Margaret Thatcher’s government hailed the rise of the services and the age of a “post-manufacturing economy” (Hirst and Zeitlin, 1991). Many studies suggest that this is due to the inverse relationship between service industry growth and manufacturing industry growth (Kakaomerlioglu and Carlsson, 2006). Kakaomerlioglu and Carlsson (2006) explains that as manufacturing in industrialized nations shrinks, service industry employment has almost always grown. This phenomenon can be explained by the empirical research in Feenstra (1998), which suggest slower productivity growth of the manufacturing industry relative to services. That, in conjunction with the growth in the relative price of services and the ability to substitute services and other goods, insinuates much faster growth in the services industry (Feenstra, 1998).

The shift from the manufacturing industry to the services industry can be observed in many areas of empirical evidence. For instance, between the years 1977 and 1990 the share of GDP occupied by the manufacturing industry decreased by about 5 percent, while services increased roughly 5 percent (Kakaomerlioglu and Carlsson, 2006). Contrary to the manufacturing industry, the services industry, and specifically producer services have had higher growth than the entire economy over the last several decades (Ibid). Shifts in priority from one industry to another as an economy grows are not uncommon and has been demonstrated through research conducted over the last century (Kindleberger, 1958). More specifically, research in Kindelberger (1958) suggests that as an economy grows demand shifts towards services. This argument is further supported by the suggestion of studies that the cause of the growth in services is related to increased demand of producer services (Ray, 1986; Martinelli, 1991).

Moreover, this transition from manufacturing to services within the economy could be directly attributed to the changes within firms as suggested by Coffey and Bailly (1991) and Fort et al. (2017). The

(23)

22 research in Coffey and Bailly (1991) suggests demand firms undertaking restructuring efforts spanning from downsizing to subcontracting, and even the reorganization of multinational corporations. Furthermore, as new technologies are developed, some argue producer services can equate to a direct substitute for manufacturing within an economy (Quinn and Guile, 1988). Perhaps the most reasonable argument made for the growth in the services industry lies in Marshall (1989), which argues that the industry does not possess as much exposure as manufacturing due to a lack of movement to other regions or countries. Further research also suggests that the services industry is less exposed due their ability to restructure themselves as changes in the economy occur (Marshall, 1989).

Often overlooked by much of the rhetoric in public discourse surrounding the services industry is how it compliments manufacturing (Kakaomerlioglu and Carlsson, 2006). Much of the discussion has been centered around observing these two industries irrespective of one another. However, empirical assessment suggest that these two industries are highly interrelated (Ibid). Thus, there is an intrinsic link between these two industries that could correspond to the changes observed in these two industries, in particular the shift from manufacturing to services. There are some studies that suggest that a lack of formal understanding of the services industry could account for such an occurrence (Ibid). Kakaomerlioglu and Carlsson (2006) argues that the rise in concern surrounding the decreases in the United State manufacturing industry and correlated increases in the services industry have resulted in part from misconceptions surrounding services. Whatever the reason, there is a mountain of empirical research to suggest undeniable link between the two industries.

It stands to reason, if one considers the body of literature discussing the factors effecting manufacturing firms and employment, that there could have been some effect on the number of jobs from the trade conducted with Mexico. Correspondingly, from the research presented in this section there is a body of evidence to suggest that trade liberalization specifically has influenced the number of jobs. The evidence backing this theory lies in literature like Hanson (2003), which argues that before the implementation of NAFTA, trade barriers like tariffs provided less exposure for the low-skill labor market of the United States. Additionally, research like that in Fort et al. (2017) has found that the manufacturing plants in

(24)

23 industries with the largest reductions in tariffs had the highest job losses. Furthermore, the research in Hanson (1996) suggests that since the elimination of trade barriers Mexico has developed a large export manufacturing industry due to the cost advantage for production. One could argue that trade liberalization, or the lifting of trade barriers, has affecting the job figures in industries with high exposure like manufacturing. Thus, this thesis argues that trade with Mexico should have a measurable association with the number of manufacturing jobs of the Rust Belt states. Therefore, the second hypothesis in this thesis that will be tested is:

variation in the number of manufacturing jobs (per 100,000 people) in a Rust Belt state, adjusting for trade deficit, amount of import trade, the primary trade partner, and size of the economy (GDP).

The corresponding null hypothesis that will be tested is

variation in the number of manufacturing jobs (per 100,000 people) in a Rust Belt state, adjusting for trade deficit, amount of import trade, the primary trade partner, and size of the economy (GDP).

Delving further into the literature surrounding trade liberalization, in addition to changes in the job numbers, economic output is also examined. Kakaomerlioglu and Carlsson (2006) notes within their study that between the years 1977 and 1990 the share of GDP that was occupied by the manufacturing industry shrank from nearly 24 percent to 18.5 percent, while services share increased. Rodrik (1998) supports this claim and ties it directly back to trade by arguing that the manufacturing component of GDP shrank due the openness of the economy increasing. Going further, Rodrik (1998) appears to make the claim that government expenditures are the solution to offset the external risks to manufacturing from trade. On the other hand, the services industry has less exposure from international trade agreements (Sondheimer and Bargas, 1993). This could be perhaps, as suggested in several studies, services are best supplied by firms located near their customer base (Kakaomerlioglu and Carlsson, 2006).

(25)

24 It stands to reason, from the literature touched upon in this section, that within the massive shift experienced in the United States labor market that employment was not the only areas of the manufacturing industry effected. Kakaomerlioglu and Carlsson (2006) that output of the industry has decreased as has its share of the economy. Amongst others, Feenstra and Hanson (1999) suggest that outsourcing causes a reduction in jobs and movement firms to low-wage countries, but in theory it could also affect economic output. From the body of literature surrounding trade liberalization we know that, as noted by Hanson (1996), outsourcing is likely to occur there is a cost advantage. Moreover, one could argue that the jobs and firms outsourced went to Mexico due to the cost advantage and large export industry established. From outsourcing, with less firms and production the economic output of the manufacturing industry decreased. Thus, this thesis argues that trad with Mexico should have a measurable association with the economic output of the manufacturing industry of the Rust Belt states. Therefore, the third hypothesis in this thesis that will be tested is:

variation in the economic output of the manufacturing industry (millions of USD) in a Rust Belt state, adjusting for trade deficit, amount of import trade, the primary trade partner, and size of the economy (GDP).

The corresponding null hypothesis that will be tested is:

(26)

25

2.4 T

RADE

L

IBERALIZATION AND

W

AGES

Within the debate surrounding the effect of trade policy on economic conditions, the relationship between trade and wages stands out. Particularly when examining free trade agreements like NAFTA, the literature suggest that regional trade agreements can affect bi-lateral trade balance (Burfisher et al., 2001). However, those advocating for the implementation of NAFTA argued that the agreement would benefit those involved through increased trade (Ibid). The central claim within this argument was that cheaper imports from Mexico would help consumers and producers alike (Ibid). Furthermore, many in trade liberalization literature have argued that due to the small share of trade conducted with Mexico and low tariffs, the agreement would have a small impact on the United States economy (Lawrence, 1994; Krugman, 1995).

The link that many studies have sought to explain is how change in wages occurs through change in trade policy (Feenstra and Hanson, 2001). Trade liberalization is the most widely cited explanation for the heavy effects on worker’s wages (Feenstra and Hanson, 1999). Thus, trade liberalization policy has been an area of focus for economists, who have become very accurate in estimating gains (Burfisher et al., 2001). In addition, these studies have demonstrated that regional trade liberalization effects trade patterns and production gains (Ibid). Further, Hanson (2003) indicates on top of affecting regional trade patterns, trade liberalization affects the wages of both and high and low-skill laborers. The most frequently cited reason for this phenomenon is increased competition from low-wage countries, which can change the wage structure (Ibid). Furthermore, there is research that indicates competition is not the only factor affecting wages, it is the conditions of the economies of the signatory members of trade agreements that matter (Davis, 1998). This is due to the flexible nature of wages in the American economy, the impact of such agreements is hard to be assessed independently from the economic conditions in trading partners (Ibid).

Of the greatest such factors impacting wages within the study of trade is policy reform, which has proven to be a huge factor (Hanson, 2003). A primary area of focus in policy reform centers around tariff changes, and when examining NAFTA, tariff reductions. Empirical studies suggest that tariff reductions have had one of the greatest impacts on the relative wage of workers (Ibid). Specifically, these studies suggest that

(27)

26 tariff and quota reductions have had an impact on inter-industry wage changes (Ibid). Research has also demonstrated that another factor dictating the changes in wages in the United States and Mexico, can be traced back to the positive correlation demonstrated between tariffs and import-license requirements (Revenga and Montenegro, 1998). Additionally, research indicates tariffs changes should disproportionately affect the wages of low-skill laborers if examined through the logic of the Stolper-Samuelson Theorem (Robertson, 2001).

Similarly, as trade liberalization spurred shifts in the wages structure, many studies have indicated that there has been negative growth in the middle-wage portion of the labor market (Autor et al., 2003; Autor et al., 2006; Autor et al., 2008; Goldin and Katz, 2007; Goos and Manning, 2007; and Goos et al. 2009). This shift came despite positive growth at the high and low ends of the spectrum (Ibid). The movement of employment gains to low and high-wage jobs, with anemic growth in the middle is referred to by many scholars as job polarization (Kurokawa, 2014). Kurokawa (2014) argues that this could be due to increased trade conducted with low-wage countries. The general consensus throughout the literature indicates low-wage countries have benefited from an increase in the relative wage and demand for skilled labor (Feenstra and Hanson, 1999). Research suggests that as trade has increased between the United States and Mexico inequality within the American labor market has heightened. In fact, studies demonstrate trade between United States and Mexican economies has contributed to the increase in wage inequality (Feenstra and Hanson, 1997).

Conceivably, an integral area of examination between trade and wages is whether the relationship causes inequality in compensation level. There has been extensive research exploring the subject of wage inequality and trade throughout the years (Burtless, 1995; Richardson, 1995; Cline, 1997; Williamson, 1997; Wolff, 2000; Hanson, 2003; Kremer and Maskin, 2006; Goldberg and Pavcnik, 2007; Chusseau et al., 2008; Chusseau and Dumont, 2012). This could be due to past studies suggesting a potential relationship between wage inequality and trade (Wood, 1994; Leamer, 1996). Feenstra and Hanson (2001) argue that international trade is indeed a catalyst for increasing the wage gap. Furthermore, Wood (1994) provides empirical backing to argument that there is a significant effect of trade on wage inequality. Within the literature there are some

(28)

27 detractors, as Lawrence (1994) and Krugman (1995) believe the amount of trade conducted is too small to account for the observed wage changes.

In the United States, more studies have been conducted to try to establish a firm linkage between trade and wage inequality. Borjas and Ramey (1994) demonstrate through research how trade volumes can be linked to wage inequality in the United States. Within the study, the analysis conducted isolated the durable-goods deficit as a percentage of GDP as an explanatory factor for wage-inequality (Borjas and Ramey, 1994). As time progressed opinions of the previous detractors changed, perhaps represented best by Krugman (2008), which argued that it was no longer a safe assumption that the effect of trade on wage inequality was diminutive. Krugman (2008) argues that due to the increase of trade with low-income countries, the circumstances surrounding previous researched changed, thus, it stands to reason the previous theory could no longer be substantiated. Kurokawa (2014) builds on this, arguing that the increase wage inequality is directly correlated with the increase in trade between Mexico and the United States. Finally, Zhu (2004) argues that this trend that can be observed in both developed and some developing countries.

At the center of the rising wage inequality is the shrinking of middle-income jobs in the United States economy. Research has demonstrated that manufacturing wages were notably higher than other industries, particularly services, finance, and trade (Loveman and Tilly, 1988). However, many studies suggest the decline in manufacturing has been a contributing factor (Kakaomerlioglu and Carlsson, 2006). Beginning in the 1970s the manufacturing share of GDP and total labor force began to shrink (Ibid). This phenomenon can be explained by research demonstrating that workers in industries with higher exposure experienced decreased in earnings and employment (Autor et al., 2014; Acemoglu et al., 2015). Further, the losses in employment of the United States manufacturing industry can be observed in concentrations among firms and regions (Fort et al., 2017). Based on these observed trends, a vigorous debate has existed around whether de-industrialization, or the shrinking of the manufacturing industry, is bad for a developed economy’s long-term growth (Cohen and Zysman, 1987). For this reason, the public debate in the United States continues focus

(29)

28 on what the appropriate policy response would be to curb the decline in the manufacturing industry (Fort et al., 2017).

With the increase of trade between low and high-wage countries one could argue a form of convergence is expected. Leamer (1998) claims that prior to the implementation of NAFTA, wage inequality was lower, as estimates suggest the wage of manufacturing labor rose faster than its counterparts. Feenstra and Hanson (2001) contradicts these findings slightly, demonstrating that the wages of non-production labor rising faster than production during the 1980s, increasing wage inequality. In general, Feenstra and Hanson (1999) suggests that the earnings of low-skill laborers in the United States have declined in real terms. Thus, the decline in the wages of low-skill workers in recent decades has become a point of public policy discussion throughout the industrialized world, and particularly in the United States (Feenstra and Hanson, 2001). Meanwhile, wages in Mexico have increased during that same timeframe and are positively correlated with the changes in the United States (Robertson, 2000).

Another consequence of trade liberalization and free trade agreements like NAFTA is the integration of economies. Hanson (2003) suggests that due to the proximity of Mexico to the United States, that trade liberalization was the equivalent of economic integration. Sachs and Warner (1995) argues that open trade can lead a convergence of growth, with higher growth rates in poorer countries than richer countries. When examining the fluctuation of wages, the economic integration experienced after NAFTA and similar trade agreements, indicates these trade policies have benefitted Mexico disproportionately (Hanson, 1996; Hanson, 2003). This is backed further by the empirical evidence presented in Robertson (2000) that suggest the inevitability that over time wages of two economies in such agreements can converge. There is, however, some disagreement in the field as to whether or not this is settled science. Hanson (2003) argues through empirical observations that there is little to no evidence of wage convergence between the United States and Mexico. Whatever the case, increased trade internationally has had a massive impact, which some argue is equivalent to that technological innovation (Feenstra, 1998).

(30)

29 Specific to this study, there is existing literature that suggests that trade policy can play an important role in determining economic conditions of individual regions. (Krugman and Venables, 1990; Krugman and Venables, 1993; Hanson, 2003). Krugman and Venables (1990; 1993) have conducted research specifically on the regional effects of the integration of the European economy. Hanson (2003) suggests that after any substantial trade reforms, regional wages are altered. This is particularly demonstrated in the states along the United States border in comparison to the rest of the country (Hanson, 2003). In addition, Hanson (2003) argues that the regions most exposed to international trade had the largest wage changes. Building upon these findings, Pierce and Schott (2017) claims that regions with higher shares of employment in particular industries demonstrated declines in earnings. These studies can be tied directly back to the rust belt, with effects stretching beyond wage changes to spikes in mortality rates among workers laid off from the steel industry in Pennsylvania during the 1980s (Sullivan and von Wachter, 2009).

Throughout the body of literature there is a well-established link between trade liberalization and changes in the wage structure. Amongst others, the research in Feenstra and Hanson (1999) suggests that trade liberalization can influences wages. Additionally, Autor et al. (2003) and other studies have demonstrated that after trade liberalization wages became more polarized, with middle-wage jobs having large levels of negative growth and causing wage inequality. The research in Feenstra and Hanson (1997) suggests that the trade between the United States and Mexico has contributed to increased wage inequality. Wage inequality could also be caused by the concept described in Sachs and Warner (1995) called economic convergence. This thesis argues that as wages in Mexico rose, the wages in the United States fell due to, as Hanson (2003) suggest, a more integrated nature between economies of the United States and Mexico. With a lack of trade barriers, like tariffs, wages also became more integrated between the trade partners (Robertson, 2000). The falling wages, theoretically, should be experienced in the most vulnerable sectors of the economy, like the manufacturing industry. Thus, this thesis argues that trade with Mexico should have a measurable association with the average annual wages of workers in the manufacturing industry of the Rust Belt states. Therefore, the third hypothesis in this thesis that will be tested is:

(31)

30 H4a: There is a significant association (α< 0.05) between the trade conducted with Mexico and

variation in the average annual wages (thousands of USD) of workers in the manufacturing industry in a Rust Belt state, adjusting for trade deficit, amount of import trade, the primary trade partner, and size of the economy (GDP).

The corresponding null hypothesis that will be tested is:

H4a: There is no significant association (α> 0.05) between the trade conducted with Mexico and

variation in the average annual wages (thousands of USD) of workers in the manufacturing industry in a Rust Belt state, adjusting for trade deficit, amount of import trade, the primary trade partner, and size of the economy (GDP).

(32)

31

2.5 H

YPOTHESES IN

S

UMMARY

In conclusion, four hypotheses were formulated that will be tested in this study:

1. There is a significant association (α< 0.05) between the trade conducted with Mexico and predicting variation in the number of manufacturing firms (per 100,000 people) in a Rust Belt state, adjusting for trade deficit, amount of import trade, the primary trade partner, and size of the economy (GDP). 2. There is a significant association (α< 0.05) between the trade conducted with Mexico and predicting

variation in the number of manufacturing jobs (per 100,000 people) in a Rust Belt state, adjusting for trade deficit, amount of import trade, the primary trade partner, and size of the economy (GDP). 3. There is a significant association (α< 0.05) between the trade conducted with Mexico and predicting

4. There is a significant association (α< 0.05) between the trade conducted with Mexico and predicting variation in the average annual wages (thousands of USD) of workers in the manufacturing industry in a Rust Belt state, adjusting for trade deficit, amount of import trade, the primary trade partner, and size of the economy (GDP).

(33)

32

CHAPTER 3: RESEARCH DESIGN

This chapter explains how the theoretical framework developed in this thesis can be applied in a scientific manner. It will begin with a section detailing the research type and methods implemented to test the hypotheses formulated. Following, a section will describe the data used for this study and how it was collected. Directly after, the conceptualization and operationalization of the thesis will be elaborated. In this section, dependent, independent, and control variables will be detailed. Then a section will be included to detail the assumptions of the multiple regression models that were tested. Finally, this chapter will conclude with a summary of the limitations of research conducted in this study. From this section, the goal will be to clarify the overall research design to better inform the reader in interpreting the analysis presented in Chapter 4.

3.1 R

ESEARCH

T

YPE

A

ND

M

ETHODS

The relationship between trade conducted with Mexico and the labor market of the manufacturing industry of the Rust Belt has had wide reaching political and economic implications that must be further explored. With such an emphasis being placed on the effect of the trade conducted with Mexico, this thesis argues that research was required to examine its effects. The purpose of this descriptive, quantitative, non-experimental study is to examine whether there is a relationship between multiple predictive variables (trade deficit with Mexico, percentage of import trade conducted with Mexico, the primary trading partner of a state, and GDP) had a significant association (α< 0.05) in predicting variation in the multiple dependent variables (number of manufacturing firms (per 100,000 people), number of manufacturing (per 100,000 people), economic output of the manufacturing (millions of USD), and average annual wages (thousands of USD) within the manufacturing industry of the Rust Belt.

This thesis implemented several multiple linear regressions to explore the relationship of four predictive variables as they relate to four dependent variables of this quantitative study, within the labor market

(34)

33 of the manufacturing industry in the Rust Belt. This thesis implemented a non-experimental, descriptive, cross-sectional design. The rationale for the selection of Large-N analysis in the social sciences is rooted in the fundamental premises of: 1. There is structure in social and political study that aides the findings in one case to be used to understand another; 2. Causal links in the social and political world are generally weak; and 3. Causal relationships are derived from patterns within variation rather than individual cases or outcomes (Toshkov, 2016).

The study conducted in this thesis is well suited for Large-N design because it seeks to establish causal links over the variation from the data from seven different states. The variation that exists in this study suits theory testing and generating hypothesis (Toshkov, 2016). Additionally, the selection of across case, or cross-sectional method is important for evaluating causal links (Ibid). This is because it can evaluate, through the cases selected, variation in the outcome variable (Ibid). It also can establish causal inferences by examining a section or population of heterogenous cases (Ibid). With a cross-sectional, Large-N design, many variables can be examined within one study, which is well suited to such a complex socio-economic phenomenon like the effects of trade on the labor market of a sector within a regional economy. A design that incorporates a large number of variables, due to the variation in the states being observed, is necessary to draw larger inferences out of the results generated in this study.

If one were to consider another research design, like Small-N, upon careful consideration it could be observed a lack of viability in the application. For one, the Small-N approach relies on comparative approaches by selecting varied cases (Toshkov, 2016). In the case of this study, we are seeking to draw out causal inferences about a larger population, the Rust Belt states as a whole. Secondly, Small-N is often used for theory generation rather than theory testing (Ibid). In this study, the aim is to test a theory, not generate new ones. Furthermore, in this study the aim is to estimate average causal effects prospectively, not retrospectively account for the outcomes in individual cases. Furthermore, the quantitative nature of this study makes it much more suited to Large-N analysis rather than Small-N, which focuses primarily on qualitative analysis (Ibid).

(35)

34

3.2 D

ATA AND

C

OLLECTION

The data that was collected to conduct the analysis for this study consists of crucial economic indicators related to the manufacturing labor market within the Rust Belt. This includes the number of manufacturing firms (per 100,000 people), number of manufacturing jobs (per 100,000 people), economic output of the manufacturing industry (millions of USD) and the averages annual wages (thousands USD) in the manufacturing industry. The data was collected with the aim to provide a holistic picture of the effects being measured on the dependent variable and was done on a state by state basis. The collection of data was performed through a self-administered, web-based search of databases of economic and social institutions in the United States. The sources chosen to be utilized from this search are reputable and consisted of: The United States Census Bureau, Bureau of Labor and Statistics, and National Association of Manufacturers.

The United States Census Bureau was chosen as a source because the institution has the largest database of socio-economic information in the United States, encompassing 320 million people and thousands of companies. The databases in the Foreign Trade and Business & Industry Sections provided much of the data used in this study (see Appendix II). The Census Bureau is also responsible for providing key information about the makeup of the citizens of the United States including race, gender, and other categories. This data is closed-sourced as it is collected by the government and its security is controlled by the Census Bureau, thus it has one of the greatest degrees of certainty of primary sources. Similarly, the Bureau of Labor and Statistics is part of the Department of Labor and measures labor market activity and working conditions in the United States economy, making it a reliable primary source. From the Bureau of Labor and Statistics, the Manufacturing section and its data were used in this study (see Appendix III). It is important to acknowledge that government entities have a certain degree of bias in the information they present to the public. However, both organizations are non-partisan, thus the data they publish is very reliable because they do not have an agenda.

The only secondary data source utilized in this study was state manufacturing data and reports published by the National Association of Manufacturers (NAM) on their website (see Appendix IV). NAM

An Assessment of the Impact of Trade Liberalization: The Relationship Between Mexico and the Manufacturing Industry of the Rust Belt