Disclaimers 2

Abstract 3

Acknowledgements 4

Executive summary 7

1 Overview of the literature 9

 1.1. The use of Google Trends for predicting the present 9

 1.2. The bias-variance trade-off in macroeconomic nowcasting 10

 1.3. Structural proximity and global production networks 10

2 Data and pre-processing 12

 2.1. ICT value-added growth and the challenge for nowcasting 12

 2.2. Google Trends topics and pre-processing 14

 2.3. Trade in value added and ICT value-chain linkages 15

3 Methodology and empirical framework 17

 3.1. Global nowcasting model and benchmark performance 17

 3.2. Measuring structural similarity in global ICT value chains 18

 3.3. Structural proximity and the pooling trade-off 20

4 Results 22

 4.1. Performance of the optimal pooling algorithm 22

 4.2. Comparative performance and resilience to global shocks 26

 4.3. Nowcast estimates for 2025 point to continued ICT growth across the OECD, with signs of convergence 30

 4.4. Unpacking the model’s predictions 31

5 Conclusion 33

Annex A. Construction of the ICT structural proximity measure 34

 A.1. Sensitivity analysis 35

Annex B. RMSE dynamics across pool sizes 𝒌 for all Member countries in the sample 36

Annex C. Additional robustness tests 40

Annex D. Country-level evaluation of optimal pooling performance over time 41

Annex E. Country-level ICT value-added growth nowcasts with confidence intervals 45

Annex F. Investigating plausible determinants of structural pooling efficiency 47

References 48

Note 52