Bank Of Greece Sa · ATHEX:TELL
Bank Of Greece Sa — Investor Presentation (September 2022)
45 slides from Bank Of Greece Sa's investor presentation, September 2022
Slide 1: Patents that Match your Standards: Firm-level Evidence on Competition and Innovation
Patents that Match your Standards: Firm-level Evidence on Competition and Innovation A. Bergeaud1 , J. Schmidt2 & R. Zago2 1 HEC 2 Banque de France ESCB cluster 2 - Athens September 2022 [The views expressed herein are ours and should not be attributed to BdF]
Slide 2: Introduction
Introduction - Standard Organizations (SSOs) are independent institutions empowered to select the set of technologies to be adopted for specific production/industry Bergeaud, Schmidt & Zago Introduction September 2022 1
Slide 3: Introduction
Introduction - Standard Organizations (SSOs) are independent institutions empowered to select the set of technologies to be adopted for specific production/industry - SSOs choose these technologies since I they are the best among competing ones or they surpass old ones, i.e. they are frontier techno
Slide 4: Introduction
Introduction - Standard Organizations (SSOs) are independent institutions empowered to select the set of technologies to be adopted for specific production/industry - SSOs choose these technologies since I they are the best among competing ones or they surpass old ones, i.e. they are frontier techno
Slide 5: Introduction
Introduction - Standard Organizations (SSOs) are independent institutions empowered to select the set of technologies to be adopted for specific production/industry - SSOs choose these technologies since I they are the best among competing ones or they surpass old ones, i.e. they are frontier techno
Slide 6: Introduction
Introduction - Standard Organizations (SSOs) are independent institutions empowered to select the set of technologies to be adopted for specific production/industry - SSOs choose these technologies since I they are the best among competing ones or they surpass old ones, i.e. they are frontier techno
Slide 7: Introduction
Introduction - Standard Organizations (SSOs) are independent institutions empowered to select the set of technologies to be adopted for specific production/industry - SSOs choose these technologies since I they are the best among competing ones or they surpass old ones, i.e. they are frontier techno
Slide 8: Introduction
Introduction - Standard Organizations (SSOs) are independent institutions empowered to select the set of technologies to be adopted for specific production/industry - SSOs choose these technologies since I they are the best among competing ones or they surpass old ones, i.e. they are frontier techno
Slide 9: Introduction
Introduction - Standard Organizations (SSOs) are independent institutions empowered to select the set of technologies to be adopted for specific production/industry - SSOs choose these technologies since I they are the best among competing ones or they surpass old ones, i.e. they are frontier techno
Slide 10: Example: ITU Standard 2015, IMT-2020 5G protocol
Example: ITU Standard 2015, IMT-2020 5G protocol - Date of release: 9 June 2015 - Key minimum specifications: I bandwidth to be at least 100 MHz I bandwidths up to 1 GHz are required for higher frequencies (above 6 GHz) I connection density is 1 million devices per square kilometer I downlink peak d
Slide 11: This Paper
This Paper - This paper studies the firm-level implications of large-scale technology adoption for competition, innovation and growth Bergeaud, Schmidt & Zago Introduction September 2022 3
Slide 12: This Paper
This Paper - This paper studies the firm-level implications of large-scale technology adoption for competition, innovation and growth - This research question implies three measurement problems: 1. how to measure the new (adopted) technology frontier 2. how to measure firms’ past innovation choices
Slide 13: This Paper
This Paper - This paper studies the firm-level implications of large-scale technology adoption for competition, innovation and growth - This research question implies three measurement problems: 1. how to measure the new (adopted) technology frontier 2. how to measure firms’ past innovation choices
Slide 14: Findings
Findings 1. Patent-level results - Our measure of proximity to the technology frontier is economically meaningful as patents semantically closer to the new standards have higher economic value, are more cited and renewed Bergeaud, Schmidt & Zago Findings September 2022 4
Slide 15: Findings
Findings 1. Patent-level results - Our measure of proximity to the technology frontier is economically meaningful as patents semantically closer to the new standards have higher economic value, are more cited and renewed 2. Firm-level results - We exploit the exogeneity of the timing and content of
Slide 16: Findings
Findings 1. Patent-level results - Our measure of proximity to the technology frontier is economically meaningful as patents semantically closer to the new standards have higher economic value, are more cited and renewed 2. Firm-level results - We exploit the exogeneity of the timing and content of
Slide 17: Literature
Literature 1. The pros and cons of standardization (IO literature): Katz and Shapiro 1985; Farrell and Saloner 1985; Leland 1979; Rysman and Simcoe 2008; Lerner and Tirole 2015; Schmalensee 2009; Llanes and Poblete 2014; Spulber 2019; Bekkers et al. 2017; Baron and Pohlmann 2018 2. Innovation litera
Slide 18: 1. Data
1. Data Bergeaud, Schmidt & Zago Literature September 2022 6
Slide 19: 1.1 Semantic Matching of Standards to Patents
1.1 Semantic Matching of Standards to Patents - Goal: Calculation of a score between each patent and each standard that measures to what extent their content overlaps Bergeaud, Schmidt & Zago Data September 2022 7
Slide 20: 1.1 Semantic Matching of Standards to Patents
1.1 Semantic Matching of Standards to Patents - Goal: Calculation of a score between each patent and each standard that measures to what extent their content overlaps - Main ingredient: text describing the respective patent and standard Bergeaud, Schmidt & Zago Data September 2022 7
Slide 21: 1.1 Semantic Matching of Standards to Patents
1.1 Semantic Matching of Standards to Patents - Goal: Calculation of a score between each patent and each standard that measures to what extent their content overlaps - Main ingredient: text describing the respective patent and standard - Link between patent and standard via keyword matching (common
Slide 22: 1.1 Semantic Matching of Standards to Patents
1.1 Semantic Matching of Standards to Patents - Goal: Calculation of a score between each patent and each standard that measures to what extent their content overlaps - Main ingredient: text describing the respective patent and standard - Link between patent and standard via keyword matching (common
Slide 23: 1.1 Semantic Matching of Standards to Patents
1.1 Semantic Matching of Standards to Patents - Goal: Calculation of a score between each patent and each standard that measures to what extent their content overlaps - Main ingredient: text describing the respective patent and standard - Link between patent and standard via keyword matching (common
Slide 24: 1.2 Firm-level Data
1.2 Firm-level Data - CRISP for stock market abnormal returns - IBES for market expectation on future EPS - COMPUSTAT for firms fundamentals: sales, market shares (defined at NAICS 3-digit level), R&D, CapX, Tobin’s Q, market capitalization, leverage, age, industry markup - Aggregation of patent-to-
Slide 25: 2. Results
2. Results Bergeaud, Schmidt & Zago Data September 2022 9
Slide 26: 2.1 Patent-level Results
2.1 Patent-level Results Is the score economically meaningful? - For scores based on patents filed 1 year before the publication of each standard, consider this: log (valuei ) = c + α log (1 + scorei ) + β log (1 + citi ) + γZi + εi where valuei is the economic value of the patent from Kogan et al.
Slide 27: 2.2 Firm-level Results
2.2 Firm-level Results Is the release of new standard exogenous to the firm? What are its implications for competition and innovation? - For firms in the COMPUSTAT universe, consider the following: N=16 X 0 Yi,t = αi + φs,t + βn Shocki,t+n + Xi,t−1 η + εi,t n=−12 where I αi is the firm FE I φs,t is
Slide 28: Exogeneity of the Shock
Exogeneity of the Shock - Each standard is voted by an ad-hoc committee. If the committee approves it, the first version of the standard is circulated to sub-committees for comments Bergeaud, Schmidt & Zago Firm-level Results September 2022 12
Slide 29: Exogeneity of the Shock
Exogeneity of the Shock - Each standard is voted by an ad-hoc committee. If the committee approves it, the first version of the standard is circulated to sub-committees for comments - If no adjustment is required from sub-commitees, the standard is published Bergeaud, Schmidt & Zago Firm-level Resul
Slide 30: Exogeneity of the Shock
Exogeneity of the Shock - Each standard is voted by an ad-hoc committee. If the committee approves it, the first version of the standard is circulated to sub-committees for comments - If no adjustment is required from sub-commitees, the standard is published - Otherwise, the standard is revised and
Slide 31: Exogeneity of the Shock
Exogeneity of the Shock - Each standard is voted by an ad-hoc committee. If the committee approves it, the first version of the standard is circulated to sub-committees for comments - If no adjustment is required from sub-commitees, the standard is published - Otherwise, the standard is revised and
Slide 32: Exogeneity of the Shock
Exogeneity of the Shock - Each standard is voted by an ad-hoc committee. If the committee approves it, the first version of the standard is circulated to sub-committees for comments - If no adjustment is required from sub-commitees, the standard is published - Otherwise, the standard is revised and
Slide 33: Exogeneity of the Shock
Exogeneity of the Shock - Consider the following dependent variables: NAICS3 I the abnormal return over a NAICS3-industry portfolio, i.e. ari,t I the change in the 1-year EPS forecast from professional agencies, i.e. ∆E[EPSi,t+4 ] = E[EPSi,t+4 |It ] − E[EPSi,t+4 |It−1 ] Figure: Standardization Shock
Slide 34: Sales & Market Shares
Sales & Market Shares - Consider sales and market-shares (defined at the Naics 3-digit level) as dependent variables Figure: Standardization Shock, Sales and Market Shares (a) Sales (b) Market Share Bergeaud, Schmidt & Zago Firm-level Results September 2022 14
Slide 35: Is it a competition shock?
Is it a competition shock? Figure: Aghion et al. 2005, Innovation vs. Competition Bergeaud, Schmidt & Zago Firm-level Results September 2022 15
Slide 36: Is it a competition shock?
Is it a competition shock? - Split Naics 3-digit industries in competitive and non-competitive according to their historical markup (De Loecker et al. 2020) and look at results for R&D Figure: Standardization Shock and R&D (a) R&D (Competitive Ind) (b) R&D (Non-Competitive Ind) Bergeaud, Schmidt & Z
Slide 37: Robustness Checks
Robustness Checks - These results are robust I to different clustering methods: firm-level clustering industry-time double clustering I to sample selection: results are not driven by frontier firms results are not driven by listed firms only I when considering the intensive/extensive margin of the v
Slide 38: 2.3 Industry-level Results
2.3 Industry-level Results What are the sectoral implications of technology adoption in terms of growth? Is growth driven by leaders or followers in the long-run? - Define as leaders (followers) those with Shocki,t > 0 (Shocki,t = 0) - Aggregate variables at industry level and use same econometric m
Slide 39: Followers explain sectoral growth in the long-run
Followers explain sectoral growth in the long-run - Consider the followers’ share of industry aggregate sales and R&D expenditure Figure: Sectoral Sales and R&D of Followers (a) Market Share (Followers) (b) R&D (Followers) Bergeaud, Schmidt & Zago Industry-level Results September 2022 19
Slide 40: Conclusion
Conclusion - This paper studies the implications of large-scale technology adoption at the firm level - To do so, we introduce a new measure of proximity of firms to the technological frontier - This measure is based on text analysis of patents and standard documents and expresses how much the conte
Slide 41: Thank you!
Thank you! Bergeaud, Schmidt & Zago Conclusion September 2022 21
Slide 42: Appendix
Appendix Bergeaud, Schmidt & Zago Conclusion September 2022 22
Slide 43: Linking patents to standards
Linking patents to standards • Basic idea: look at keywords in the abstract of the patent and in the standard document 1 Establish a list of all patent keywords with respective list of patent ids keyword patent id air condit 41 aluminium 41,106 beef 2 electric 7,84,41,91,102 screw 3,9,17,38 2 Establ
Slide 44: Linking patents to standards
Linking patents to standards 4 Find common keywords keyword patent id standard id idf aluminium 41,106 B 0.2 air condit 41,65 B,M 0.5 5 For each keyword: Cartesian product of respective patent and standard ids keyword match idf aluminium 41-B 0.2 aluminium 106-B 0.2 air condit 41-B 0.5 air condit 41
Slide 45: Building up the score
Building up the score Back - Evaluation of importance of each keyword k via calculation of inverse document frequency (IDF) across all patents (N): 1+N IDF (k) = 1 + log 1 + N(k) - For every patent i ∈ P and standard j ∈ J , we extract I the set of k-grams B(i) for patent i I the set of k-grams A(j)
Bank Of Greece — Investor Presentation (September 2022) on 6ix