Essays in Network Econometrics and Data Science: Theory and Empirics Restricted; Files Only
Rojas Baez, Diego Israel (Spring 2022)
Published
Abstract
The increasing availability of social and economic interactions has sparked the academic effort to use models taking into account the effect that peers have on the decision of economic agents. One challenge to tackle on the way is how to approach the massive amount of data that firms and governments collect on economic interactions. Another challenge is how to use this large amount of information to model and better understand economic interactions with peer effects, Understanding that interactions between can have different natures. The first essays of this dissertation provide insights on how to use data science to approach using big data in administrative records. The third essay offers a way to model multilayer networks, networks with different types of connections, using the characteristics of the economic agents and a dyadic regression.
The first chapter describes a unique large administrative set of records from the Bank of Canada’s Currency Inventory Management Strategy. The data from a single note inspection procedure generates a sample of 900 million banknotes. After defining the duration of the banknote circulation, several data science techniques are employed to describe the duration patterns across different banknotes: K–prototype clustering, Network graphs, and statistics, among others. A hazard model estimates the survival curve of banknote circulation cycles based on their clusters and characteristics.
The second chapter explores an intervention of The Bank of Canada in the year 2017. It introduced a change in the imagery of forty million CAD 10 banknotes. The chapter sets to explore whether slight differences in the appearance of a banknote can elicit a change in the behavior of economic agents enough to measure it. Specifically, we construct a measure of the duration of the circulation cycle of a note using proprietary data of the Bank of Canada.
The third chapter extends the primary tool to model network formation: the dyadic regression. This chapter proposes that dyadic regression provides a sound framework to model multilayer network formation under standard regularity conditions. This paper introduces an extended version of a gravity model that can accommodate a larger network class that allows for multiple layers of connections.
Table of Contents
1 Survival Analysis of Bank Note Circulation: Fitness, Network Structure, and Machine Learning 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Single Note Inspection Data . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.1 Institutional Background . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Structure of the IMS dataset . . . . . . . . . . . . . . . . . . . . . . . 5
1.2.3 Sample description . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3 The Network and Spatial Patterns of Bank Notes . . . . . . . . . . . . . . . 15
1.3.1 The Cycle Duration and Bank Note Fitness . . . . . . . . . . . . . . 15
1.3.2 Money Circulation Network . . . . . . . . . . . . . . . . . . . . . . . 19
1.4 Banknote Clusters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.5 Hazard Model for Bank Notes . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2 On the Effect of Changing the Appearance of Money: Evidence from a Country-Wide Event 43
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.2.1 The intervention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.2.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.3 Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.3.1 Decision frames and Money . . . . . . . . . . . . . . . . . . . . . . . 51
2.3.2 Identification strategy . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.3.3 Estimating equation . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.4.1 Duration patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.4.2 Estimates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3 Multilayer Dyadic Gravity Models 73
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
3.2 Mulilayer Dyadic Data Model . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.2.1 Multi-layer Gravity Model . . . . . . . . . . . . . . . . . . . . . . . . 80
3.3 Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.4 Numerical Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.5 Empirical Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.5.1 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
3.5.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
About this Dissertation
| School | |
|---|---|
| Department | |
| Degree | |
| Submission | |
| Language |
|
| Research Field | |
| Palavra-chave | |
| Committee Chair / Thesis Advisor | |
| Committee Members |
Primary PDF
| Thumbnail | Title | Date Uploaded | Actions |
|---|---|---|---|
|
|
File download under embargo until 26 May 2024 | 2022-03-25 16:59:35 -0400 | File download under embargo until 26 May 2024 |
Supplemental Files
| Thumbnail | Title | Date Uploaded | Actions |
|---|