1. Introduction 4

  2.1 Occupational and Sectoral Exposure 6

  2.2 Concentration in the Service Economy 9

  2.3 Overlap with Non-Market and Hard-to-Measure Output 10

3. An Accounting Framework for the Measurement Gap 11

  3.1 The Sum-of-Costs Approach to Non-Market Output 11

  3.2 A Stylized Example: AI in Public Administration 11

  3.3 Quality Adjustment in Market Services 12

  3.4 Combined Effect: Systematic Underestimation 12

  3.5 Formal Decomposition of Measurement Bias 13

  3.6 Summary Interpretation 14

4. Can the Inclusion of Capital Services Mitigate the Measurement Gap? 14

  4.1 Conceptual Framework: Expanding the Sum-of-Costs Approach 14

  4.2 Capital Deepening Through AI Investment 14

  4.3 Capital Deepening Versus True Productivity Gains 15

  4.4 Implications for Labor Productivity Measurement 15

  4.5 The Distinction Between Current Prices and Chained Volume Measures 16

  4.6 Extension to Market Services 17

  4.7 Summary Assessment 18

5. Rethinking Measurement: Capturing AI-Driven Productivity Gains 18

  5.1 Moving Beyond the Sum-of-Costs Approach 18

    Numerical Example 1: Outcome-Based Measurement in Education 19

    Activity-Based Indicators 19

    Trade-Offs 19

  5.2 Improving Quality Adjustment in Service Prices 20

    Hedonic Adjustment 20

    Outcome-Based Pricing 20

    Task-Based Pricing Models 20

  5.3 Incorporating Exogenous MFP Adjustments in the Measurement of Non-Market Output 20

  5.4 New Goods, New Sectors, and the Limits of Existing Classifications 23

  5.5 Implications for Measuring AI-Driven Productivity 23

6. Conclusion 24

References 27