## 1 Introduction

## 2 Background

### 2.1 Crowdfunding and related research

### 2.2 Internalization of external effects

## 3 Analyzing an internalization approach for overfunding on crowdfunding platforms

### 3.1 Methodology

#### 3.1.1 Theoretical background for the decision-making behavior of agents

#### 3.1.2 Methodological framework

### 3.2 System analysis

### 3.3 Model design

#### 3.3.1 Purpose

#### 3.3.2 Entities, state variables, and scales

#### 3.3.3 Process overview and scheduling

#### 3.3.4 Design concepts

#### 3.3.5 Initialization and input

^{1}we use a Poisson distribution with \(\lambda =170\) representing the average number of new projects per day until the end of 2014 (matching the respective year of observations in our data set).

Name | Configuration | Description | Source |
---|---|---|---|

Simulation days | 360 | Duration of the simulation after initialization | Modeling choice |

Projects per day | Poisson(\(\lambda =170\)) | Distribution of the number of new projects per day | Kickstarter statistics |

Funding amount | Poisson(\(\lambda =80\)) | Distribution of the amount of money a funder pledges | Kickstarter statistics |

Individual preference | \({\mathcal {N}}\)(\(\mu =0.5\), \(\sigma =0.25\)) | Individual preference factor distribution | Modeling choice |

Funding period | Frequency distribution | Duration of a campaign’s funding period | Data set |

Funding goal | Frequency distribution | A campaign’s funding goal | Data set |

Pictures | Frequency distribution | Number of pictures provided by a campaign | Data set |

Videos | Frequency distribution | Number of videos provided by a campaign | Data set |

Text length | Frequency distribution | Text length of a campaign’s project description | Data set |

Categories | Frequency distribution | A campaign’s category | Data set |

Taste parameter | \({\mathcal {U}}\)(0,1) | A campaign’s taste parameter | Modeling choice |

Active funder | \({\mathcal {U}}\)(0,1) | Probability of a funder to get active | Modeling choice |

Registered funders | Calibrated | Number of simulated funders | Modeling choice |

Initial funding | Calibrated | Probability whether a funder does an initial funding | modeling choice |

Observed projects | Calibrated | Number of observed projects per funder | Modeling choice |

Category preferences | Calibrated | Number of categories a funder can have preferences for | Modeling choice |

Funding threshold | Calibrated | Funder only funds projects above the funding threshold | Modeling choice |

Supported projects | 10 | Maximum number of supported projects | Modeling choice |

#### 3.3.6 Submodels

(i) | Benefit type (high values preferred): | \(n_i = \frac{a_i-min\{a_1,\ldots ,a_n\}}{max\{a_1,\ldots ,a_n\}-min\{a_1,\ldots ,a_n\}}\), |

(ii) | Cost type (low values preferred): | \(n_i = \frac{max\{a_1,\ldots ,a_n\}-a_i}{max\{a_1,\ldots ,a_n\}-min\{a_1,\ldots ,a_n\}}\), |

(iii) | Fixed type* (values close to a fixed value \(\alpha \) are preferred): | \(n_i = \frac{|a_i-\alpha |}{max\{|a_i-\alpha |\}}.\) |

[* Be careful—there are some issues with the notation used in Xu (2015).] |