Excerpt
Contents
List of Figures
List of Tables
List of Abbreviations
List of Symbols
List of Appendices
1. Problem and Course of the Investigation
2. Clustering of OECD countries
3. The Task-Based Approach
4. Determinants of Labor Demand
4.1 Analytical Framework
4.2 Technological Progress
4.3 Trade
4.4 Merge of Technological progress and Trade
5. Conclusion
Appendix
References
List of Figures
Figure 1: Clustering of OECD countries into Northern and Southern ones
Figure 2: O*NET Terminology
Figure 3: Three Conditions of Substitution
Figure 4: Examples of Tasks to the Corresponding Categories
Figure 5: Predictions for the Impact of Computerization Based on Four Categories
Figure 6: Predictions for the Impact of Computerization Based on Three Categories and Assignment to Skill Level
Figure 7: Blinder's Estimation of Affected Workers, Split in the Four Category Model
Figure 8: Development of Transportation Costs Since 1990
Figure 9: Average Annual W age Level
Figure 10: CAGR of Annual Wage Level
Figure 11: Merging of Technological Progress and Trade
List of Tables
Table 1: Blinder’s Four Main Occupational Categories of Offshorability
List of Abbreviations
illustration not visible in this excerpt
List of Symbols
illustration not visible in this excerpt
List of Appendices
Appendix 1: Calculation pattem for Employment
Appendix 2: Clustering of OECD countries into Northern and Southern ones with further clustering of the Northern ones into American and European types
Appendix 3: Calculation Pattern for Appendix 2
Appendix 4: Approaches and Estimations on Offshorability
Appendix 5: Calculation Pattern Transportation Costs
Appendix 6: Calculation Pattern for Wages
Appendix 7: Implications of and Interrelation between Technological progress and Trade
1. Problem and Course of the Investigation
The central concept of microeconomic theory is straightforward: Firms set price and quantity due to profit maximizing purposes. As profit is defined as revenue less cost, this optimization process is based on a production function and a cost function.[1] For relevant inputs to the production function, companies face the decision to substitute domestic labor with either capital or foreign la- bor[2] - provided a given output - whereas costs are deduced from the inputs[3]. This substitution process represents the impacts triggered by technological progress and trade.
In the last few decades, computerization (as the dominant part of technical progress) as well as offshoring (of trade) have caused intensive debates in Northern countries. The public is concerned[4] that these two phenomena might be a threat[5] that functions at the expense of domestic labor[6]. It is therefore to be analyzed if these worries are appropriate; that is, which jobs are at stake and which have future perspectives in terms of existence.[7] This is going to be scrutinized by applying an analytical framework. Within this framework the questions of legality, feasibility and rationality are examined in systematical ordering. Furthermore, as recent literature suggests neither a sectoral- nor an industrial- or occupational-based analysis, but rather one on a task-level basis[8], this task-based approach is used in the framework and necessitates a re-assignment from tasks to jobs.
After this introductory part and a clustering of the OECD countries into Northern and Southern ones in the second part - with the paper being written from the perspective of a Northern country - the third part deals with a description of the task-based approach. The fourth part firstly illustrates the analytical framework, secondly applies it to computerization (Chapter 4.1) and offshoring (Chapter 4.2), and thirdly merges the arguments produced by the two threats (Chapter 4.3). The fifth part concludes.
2. Clustering of OECD countries
It is common wisdom that there are crucial differences concerning the level of the factors determining the economic environment, even between the OECD countries themselves.[9] Hence a cluster analysis that separates the OECD countries into Northern and Southern ones is necessary for setting the stage for a subsequent analysis based on the perspective of Northern countries. The first indicator makes sure, that the employment between the economic sectors is approximately uniformly distributed. The second indicator, GERD per capita, accounts for the differences in the technical equipment. As a result, the 30 OECD countries are clustered into 8 “developing Southern countries” and 22 “developed Northern ones”. Figure 1 shows the classifications at a glance.[10]
illustration not visible in this excerpt
Figure 1: Clustering of OECD countries into Northern and Southern ones
Source: Own representation based on calculations on OECD (2008b), p. 1; CIA (2008)[11]
3. The Task-Based Approach
In recent years, labor market related economy has been facing three crucial phenomena: deindustrialization[12], fragmentation[13] and computerization[14]. Enabled by fragmentation and flanked by computerization and deindustrialization, it is tasks rather than jobs that are threatened today. As there is ambiguous usage of the terms jobs, tasks, and related terms in literature, this chapter presents the O*NET[15] conceptualization, which will be referred to throughout this paper because it is not only mutually exclusive and collectively exhaustive but also commonly used throughout economic literature[16]. Figure 2 depicts the understanding and relation of these terms. Accordingly, every occupation consists of a finite set of job titles, where each job is specified on the basis of nine standardized domains. These domains are scaled by assigning one or more or- dinal[17] rankings to each domain.
illustration not visible in this excerpt
Figure 2: O*NET Terminology
Source: Own Representation
Tasks and activities are considered to be the most important domains.[18] Activities summarize standardized descriptors among occupations and therefore allow for comparison in the fields of importance and level. Instead, tasks are assigned uniquely for each occupation. Therefore, tasks are more specific but at the same time do not allow for strong comparison. The following example clarifies the usage of these terms. The occupation “Janitors and Cleaners, Except Maids and Housekeeping Cleaners” has linked samples of job titles like “Janitor”, “Cleaner” or “Custodial Worker”. The corresponding activities are, e.g., “handling and moving objects” or “performing general physical activities” whereas the tasks are, e.g., “service, clean, and supply restrooms” or “gather and empty trash”. As activities are standardized descriptors they also apply to the occupation “maintenance and repair workers” whereas exemplary tasks do not. Here, “repair or replace defective equipment parts using hand tools [...]” or “inspect drives [.] and check fluid levels” are the relevant tasks.[19]
For deducing the vulnerable jobs by using a task-based approach, it is required to assign jobs with tasks. This appears to be difficult, as every occupation consists of numerous task inputs from different categories[20]. Therefore, the task inputs into every job are rated in terms of level and importance[21], while only the highest rated ones will be considered as relevant.[22]
4. Determinants of Labor Demand
4.1 Analytical Framework
Generally, three conditions have to be fulfilled for a task, which is actually performed by domestic labor, to be performed by either capital or foreign labor, i.e. substitution of domestic labor to occur. This conception is depicted as a flowchart in Figure 3. The conditions are scrutinized in logical ordering and if one of them is not fulfilled, substitution is not going to occur.
illustration not visible in this excerpt
Figure 3: Three Conditions of Substitution
Source: Own Representation
The first condition, legality, scrutinizes whether governmental restrictions prohibit substitution. The second condition deals with the question whether substitution is generally feasible, whereas especially constitutional reasons might prevent substitution here. If substitution is feasible, i.e. domestic labor is substitutable, substitution is not necessarily implied, since for answering this in affirmative, marginal profit has to increase due to substitution. This question arises in the third condition and is economically founded[23] in terms of a decision calculus. If revenues are assumed to stay constant, the marginal costs occurring in the substitution-case have to fall below the costs in the nonsubstitution case for substitution to occur. Although the framework suggests all conditions to have binary character (due to applicability reasons), they are rather gradual.
As the first condition can only be neglected due to distributional reasons (concerning technical progress) or is rather politically than economically motivated (concerning trade[24] ), it is not going to be addressed further on.
4.2 Technological Progress
The feasibility of substituting domestic labor by capital finds its way into economic growth as the concept of technological progress.[25] This phenomenon suggests a shift in the production technology and is generally not factor- neutral, but rather favors some production factors more than others.[26] The SBTC-hypothesis refers to a shift that favors skilled work more than unskilled work, which implies an increase in its relative productivity and, by conclusion, its relative demand.[27] Hence, the SBTC-hypothesis suggests that upskilled jobs have to be seen as the ones with future perspectives. As since the end of World War II up to the millennium, computerization and upskilling coincided in the U.S.[28], this has been (partly) seen as a corroboration of the SBTC-hypothesis.[29]
The critique of the SBTC-hypothesis comprises that the relation between computerization and the skill level is drawn indirectly - by using economic indicators like education or wages[30] - and creates a fallacious causal relation[31]. This immanent theoretical flaw gained practical relevance as the income inequality stabilized in the 1990s, but computerization continued.[32] This critique has been taken by ALM (2003) for developing a general equilibrium model, with the aim of analyzing the impact computerization makes on the contestability and composition of domestic work. In contrast to the SBTC-hypothesis, their model acts on the assumption that skill requirement can be measured directly[33]. The authors use two variables to categorize all workplace tasks into four different groups: The first variable characterizes the relation between a task and the use of computers for performing it. If a task is well-defined, e.g., it can be broken down into explicit rules; it can be compiled into a computer code and therefore accomplished by computers. In this case it is labeled routine, otherwise it is labeled nonroutine.[34] The second variable separates manual and information-processing tasks, whereas the latter category contains cognitive (for routine) and analytical / interactive (for nonroutine) tasks[35]. A manual task requires physical action for performing it[36], whereas a cognitive task requires acquiring and processing information; an analytical task calls a worker to be able to understand and solve problems, whereas an interactive task mainly calls for communication skills[37]. As computers are able to process a well-defined task in a more efficient way than human workers do; computer capital substitutes domestic labor in routine tasks, by conclusion. As nonroutine tasks are not well defined, computers do not substitute nonroutine workers, but rather complement them by increasing their efficiency by implication[38]. This relationship is just assumed to exist for the interactive and analytic tasks. As for nonroutine manual tasks, they are almost unrelated to computerization[39]. Figure 4 and Figure 5 depict the arguments.
illustration not visible in this excerpt
Figure 4: Examples of Tasks to the Corresponding Categories
Source: Own Representation Based on ALM (2003), p. 1286.
illustration not visible in this excerpt
Figure 5: Predictions for the Impact of Computerization Based on Four Categories
Source: Own Representation Based on ALM (2003), p. 1286.
The authors model the relations, presented above, in a Cobb-Douglas production function with constant returns to scale as shown in Equation (1). They assume routine and nonroutine tasks to be imperfect substitutes and the mar- ginal productivity of nonroutine task performance to be increasing as routine input accrues. The relevant input factors are labor input for performing the nonroutine tasks, whereas computer capital and routine labor input represent in sum the input for routine tasks. The output Q is sold at unity.
illustration not visible in this excerpt
The driving exogenous variable in the model is a declining market price per efficiency unit ρ for computer capital which is assumed to be supplied with perfect elasticity at market price and to be a perfect substitute for routine task input. Therefore, its price marks the wage-setting constraint for routine wages.
illustration not visible in this excerpt
The work force I is a set of workers, whereas every worker i has a heterogeneous productivity endowment in routine and nonroutine tasks [illustration not visible in this excerpt] with [illustration not visible in this excerpt]. The ratio [illustration not visible in this excerpt] is labeled as relative efficiency. Every worker supplies that amount of labor [illustration not visible in this excerpt] whereas [illustration not visible in this excerpt] , which maximizes her income. Therefore, [illustration not visible in this excerpt] depends on the wage ratio [illustration not visible in this excerpt]. The marginal worker is indifferent between supplying [illustration not visible in this excerpt] or [illustration not visible in this excerpt] when her efficiency ratio [illustration not visible in this excerpt] equals the inverse wage ratio:
illustration not visible in this excerpt
By conclusion, the individual labor supply function [illustration not visible in this excerpt] looks like:
illustration not visible in this excerpt
The authors model the labor supply of the whole labor force in terms of two functions, g(r) and h(r)for routine and nonroutine tasks, respectively. The ratio of routine to nonroutine task input 0 equals:
illustration not visible in this excerpt
Provided that markets clear by worker self-selection, labor market equilibrium is governed.[40]
[...]
[1] Cp. Mas-Colell/Whinston/Green (1995), pp. 135, 137. Economies of Scale are assumed to be existent throughout the paper.
[2] Cp. Blanchard (2006), pp. 248-249.
[3] Cp. Besanko/Braeutigam (2002), pp. 269.
[4] In fact, offshoring was a decisive issue during the 2004 election campaign in the U.S. Cp. Mankiw/Swagel (2006), pp. 6-18.
[5] To be precisely, these two phenomena function on a different level. Whereas computerization by definition does not necessarily imply a substation of domestic labor, offshoring does. However, they are both considered to be a threat as the difference is just semantically.
[6] Cp. Ochsen (2004), pp. 4-5.
[7] To cope with these concerns, this paper only takes already existing jobs into consideration, i.e. newly created jobs are not considered. A net analysis might produce different results. In 2004, the overall U.S. employment increased by 1.2 per cent whereas the growth in computer occupations reached 5.5 per cent. Cp. WTO (2005), p. 282. As another indicator for this presumption, McKinsey Global Institute (2003) estimated that from every dollar spent for offshoring purposes, a net gain of up to 14 cents is generated for the overall domestic economy of the U.S. Cp. McKinsey Global Institute (2003), pp. 7-9. Although data shows that on average about 70% of domestic workforce is employed in the service-sector (Cp. Appendix 1), an exclusive analysis of this sector would be inappropriate because the service-sector itself is widely diversified in terms of tasks. Cp. Head/Mayer/Ries (2008), pp. 3-10.
[8] Cp. Blinder (2007a), p. 5; Cp. Goos/Manning (2003), p.8; Stack/Downing (2005), p. 516.
[9] Mexico for instance is a considerable offshoring opportunity for U.S. firms and an OECD country at the same time. Cp. Goel/Moussavi/Srivatsan (2008), p. 3; Rubin/Tal (2008), pp. 6-7.
[10] A more profound analysis could suggest to additionally distinguishing of Northern countries in terms of regulatory patterns - like flexible or fixed wage-setting processes - into “European” and “American” countries. Appendix 2 presents a clustering based on EPL data. As a crucial implication, an increasing wage inequality in the American countries would be reflected as an increasing unemployment rate in European countries, which therefore has impact on the research question. (Cp. Krugman 1995, pp. 329-330.) Due to complexity purposes, this separation is not centrally accounted for.
[11] See Appendix 1 for calculation pattern in detail.
[12] Deindustrialization describes the shift of labor force allocation from the manufacturing to the services sector - accompanied by a shift of value adding as a percentage of GDP into the same direction. Cp. Rowthorn/Ramaswamy (1997), pp. 7-9.
[13] Fragmentation refers to the re-organization theory and the value chain concept. It indicates that a vertically integrated production process is being sliced up into separable components. Cp. Krugman (1995), pp. 333-335. In other words, fragmentation suggests the technical and institutional separability of tasks in time and space. Cp. WTO (2005), p.269.
[14] Computerization signifies the diffusion of computers into daily work. Cp. Anonymous (2005), Article "computerize verb".
[15] O*NET is an occupational information network developed through the sponsorship of the U.S. Labor Department, see http://www.onetcenter.org/. See http://online.onetcenter.org/ for the online database.
[16] E.g., cp. Blinder (2007a), p. 10. It is also predominantly consistent with the one used by ALM (2003) as they use empirical data from the DOT which had been replaced by the O*NET in a continuous process. (Cp. Mariani (1999), p. 3). ALM (2003) define work as a “series of tasks” (Cp. ALM (2003), p. 1282) but does not explicitly distinguish between tasks and activities (Cp. ALM (2003), p. 1282). A the used conceptualization has to be consistent with the empirical data used for testing the underlying model, other authors constitutionally refer to other databases. Spitz-Oener (2006), for West Germany, refers to the German Federal Institute for Vocational Training and the Research Institute of the Federal Employment Service, Goos/Manning (2003), for the U.K., to the New Earnings Survey and the Labor Force Survey.
[17] This can be deduced from the fact, that the scales are being standardized from the collected scale to a standardized scale between 1 and 100. Cp. Anonymous (2008).
[18] Cp. Blinder (2007a), p. 14.
[19] Cp. http://online.onetcenter.org/. The relevant O*NET-SOC codes are 37-2011.00 for Janitors and 49-9042.00 for Maintenance Workers (Also quotes).
[20] This is a reasonable assumption as there is empirical verification, cp. Spitz-Oener (2006), pp. 261-263.
[21] This can either be done by a labor organization or through a survey among employees. Cp. Spitz-Oener (2006), pp. 241-242. This approach has been used by Blinder (2007a) (Cp. Blinder (2007a), p. 27. and in an alleviated mode by ALM (2003) (Cp. ALM (2003), p. 1293).
[22] As an example, imagine a janitor who, besides the tasks mentioned above, has to order new equipment by time. But as this does not represent a dominant task, it will be discarded.
[23] Cp. Mankiw (2004), p. 7.
[24] For an appraisal of this question cp. Stack/Downing (2005).
[25] Cp. o.a. Mankiw (2003), pp. 207-208.
19-26.
[27] Cp. ALM (2003), pp. 1279-1280.
[28] Cp. Card/DiNardo (2002), pp. 6-8. To be proficient, computers and ICT in general do not present the production factor capital themselves but rather significantly have been increasing the productivity of capital since they became part of daily business in the beginning of the 90s. Cp. Mankiw (2003), pp. 218-219; Blanchard (2006), pp. 6-7.
[29] Cp. ALM (2003), p. 1280.
[30] Cp. ALM (2003), p. 1283.
[31] Cp. ALM (2003), p. 1285.
[32] Can be deduced by the given examples cp. ALM (2003), p. 1286 or deduced by a comparison of the explanations “routine cognitive and manual” (Cp. Autor/Katz/Kearney (2006a), p. 191) and “cognitive and physical” (Cp. Autor/Katz/Kearney (2006b), p. A2).
[33] Cp. ALM (2003), p. 1293.
[34] Cp. ALM (2003), pp. 1286-1287.
[35] Cp. ALM (2003), pp. 1287, 1302.
[36] For the descriptive part of the model cp. ALM (2003), pp. 1286-1289.
[37] For recent studies cp. Nordhaus (2007), pp. 151-154. Besides that, Moore’s law (1965) already suggest a similar growth.
[38] Cp. Nordhaus (2007), p. 158.
[39] Cp. Blinder (2007a), pp. 4, 8.
[40] In fact, the increase of computer capital more than compensates the decrease in routine labor input, e.g. the overall share of routine task input increases. Cp. ALM (2003), p. 1289.