Discussion of Routine Jobs, Employment and Technological Innovation in Global Value Chains by L. Marcolin, S. Miroudot and M. Squicciarini Philipp Harms (Johannes Gutenberg University Mainz, Germany) Conference on International Trade and Macroeconomic Interdependence in the Age of Global Value Chains, Vilnius, 15-16 September 2016
Overview Offshoring and routine tasks: theory The paper s contributions Data: measuring routine intensity Estimation: global value chains and employment Interpreting the results Concluding remarks
Offshoring and routine tasks: theory Grossman and Rossi-Hansberg (AER 2008): offshoring as trade in tasks a w * Costs of performing task i L β t( i) abroad. a L : Input coefficient w*: Foreign wage β: Aggregate costs of offshoring t(i): Task-specific costs of offshoring 0 1 HR MR LR NR Conjecture: t(i) lower for routine-intensive (HR) tasks i Note: Becker et al. (JIE 2013) would agree, Blinder and Krueger (JLE 2013 wouldn t.)
Offshoring and routine tasks: theory Grossman and Rossi-Hansberg (AER 2008): the effect of globalization (= lowering β) old a w * β a L L w * β new t( i) t( i) w a L Costs of performing task i domestically. 0 HR MR LR NR 1 performed abroad performed domestically i Result: Lowering β results in offshoring. Routine-intensive (HR) tasks are performed abroad.
Offshoring and routine tasks: theory Grossman and Rossi-Hansberg (AER 2008): the effect of globalization (= lowering β) (contd.) Hypothesis: Negative impact of offshoring on domestic employment in routine-intensive occupations. Note: Shift of focus from workers skills to the routine intensity of their occupations. Challenges: Data: How to measure the routine-intensity of tasks/occupations? Estimation: Identify effect of offshoring/global value chains on routine-intensive tasks/occupations.
The paper s contributions Data: measuring routine intensity Companion paper: Marcolin et al. (2016) describing details. PIAAC survey: 105,526 individuals from 22 OECD countries. Routine intensity of individual k employed in sector i and occupation o based on her/his responses to PIAAC survey. RII, = w1 Sequentiability + w2 Flexibility + w3 Plan_own k i, o + k, i, o w 4 Organise_ownk, i, o k, i, o k, i, o Benchmark weights: w i = ¼ for all i. For each occupation: computation of median value of RII M ( RII ). i, o
The paper s contributions Data: measuring routine intensity (contd.) M Occupations ranked according to median routine intensity ( RII i, o ). Note: Ranking possibly differs across countries. Occupations divided into four quartiles: High routine intensive (HR), Medium routine intensive (MR), Low routine intensive (LR), Non-routine intensive (NR). Prevalence of routine intensity in different countries/industries based on the percentage of individuals employed in different occupations (which are characterized by a certain routine intensity).
The paper s contributions Data: measuring routine intensity (contd.) How to interpret this percentage?
The paper s contributions Data: measuring routine intensity (contd.) How to interpret percentage of employment in NR-occupations, e.g. in Poland? High share of employment in occupations that are de facto NR. High share of employment in occupations that are de facto NR in Poland, but possibly LR, MR, HR in other countries. High share of employment in occupations that are perceived as NR in Poland, but possibly perceived as LR, MR, HR in other countries. Question: Are there substantial cross-country differences in the ranking of occupations according to their routine intensity?
The paper s contributions Estimation: global value chains and employment Regression equation: ln ( N q, i, k, t ) = β 0 + β 1 GVC i, k, t + β 2 TECH i, k, t + γ x + ε j j j, i, k, t q, i, k, t (log of) employment in routine intensity q in country i in industry k at time t A measure of the prevalence of offshoring/ global value chains in industry k in country i at time t A measure of technology intensity of industry k in country i at time t Control and dummy variables
The paper s contributions Estimation: global value chains and employment (contd.) Results: Most technology-related variables have the expected effects (e.g. ICT intensity reducing HR employment in manufacturing) Effect of different measures of offshoring: Input offshoring has a positive effect on HR employment in manufacturing. Offshoring of final assembly has a negative effect on NR employment in manufacturing. Domestic outsourcing has a positive effect on MR and HR employment in manufacturing. Services outsourcing has a negative effect on HR employment in manufacturing. Results differ between manufacturing and services.
Interpreting the results What determines the offshorability of a task? M/M/S (and many others): The task s routine intensity Harms/Lorz/Urban (CJE 2012), Baldwin and Venables (JIE 2013): The task s position in the context of an (industry-specific) production process. A simple example with five tasks ( production steps ): 1 2 3 4 5 Assumptions: Task 1, 3, 5: NR, i.e. cheaper to perform in the domestic economy / Tasks 2, 4: HR, i.e. cheaper to perform abroad 1 2 3 4 5
Interpreting the results What determines the offshorability of a task? (contd.) Constellation 1: The sequence of production steps is flexible or tasks can be performed simultaneously. Representative firm exploits all cost differences by offshoring those production steps which are cheaper to perform abroad. Home 1 3 5 Foreign 2 4 This is the scenario described by Grossman/Rossi-Hansberg (2008)
Interpreting the results What determines the offshorability of a task? (contd.) Constellation 2: There is a rigid sequence of production steps (i.e. tasks have to be performed sequentially). Baldwin and Venables (2013) call this a snake. The costs of shifting unfinished goods across borders ( transport costs ) are negligible. Same result as in the G/R-H (2008) model. Home 1 3 5 Foreign 2 4
Interpreting the results What determines the offshorability of a task? (contd.) Constellation 3a: There is a rigid sequence of production steps (i.e. tasks have to be performed sequentially). There are large costs of shifting unfinished goods across borders. Firms may refrain from doing any offshoring. All tasks are performed in the domestic economy. Home 1 2 3 4 5 Foreign
Interpreting the results What determines the offshorability of a task? (contd.) Example: Automotive industry Source: Siemens AG
Interpreting the results What determines the offshorability of a task? (contd.) Constellation 3b: There is a rigid sequence of production steps (i.e. tasks have to be performed sequentially.) There are intermediate costs of shifting unfinished goods across borders. Firms choose an excessive volume of offshoring. (Almost) all tasks are performed abroad. Home 1 5 Foreign 2 3 4
Interpreting the results What do we learn from this discussion? The offshorability of a given task is likely to depend on... skills required to perform it,... its routine intensity,... other characteristics (see Blinder and Krueger, JLE 2013),... the properties of the (industry-specific) production process. Even high-routine tasks may be hard to offshore if they are part of a production process that is... very sequential (rather a snake than a spider )... characterized by high border-crossing costs Consequence: Impact of globalization / GVCs on employment in NR/LR/MR/HR occupations may be highly industry-specific.
Concluding remarks The paper... uses a new data set that allows to assess the routine intensity of occupations. uses this data set to explore whether (and how) the impact of technology and offshoring on employment depends on routine intensity. presents some interesting (and puzzling) empirical results. Suggestions: Present evidence on RII of different occupations for different countries. Allow effect of offshoring/gvcs on employment to differ across (groups of) industries.