FDI & College Major Choice

Methodology

Jose Rojas-Fallas

2025

Data

Administrative University Application Data

• Preferred Major Choice
• Age and Sex of Applicant
• Application Entry Score

• High School Attended Type (Public/Private)
• District-Canton of Residence

Firm-level Data for Free Trade Zone Regime

• ISIC Rev. 4 Industry Code (2-digit)
• 289 Unique Firms

• Year-of-Entry into Nation
• District-Canton of Operation

Multinational Corporations Industry Presence Index

Using a Gravity Model approach, I construct a district-year MNC Presence Index as follows:

\[\begin{align*} \Gamma_{djt} = \sum_{d'} \dfrac{tenure_{j \in J, d't}}{exp(\tau^{dd'})} \end{align*}\]

Subscripts

\(d =\) student’s district, \(d' =\) firm’s district, \(j \in J\) firm \(j\) in industry \(J\), \(t =\) year

• \(tenure_{j \in J, d't}\): Tenure of firm \(j\) in industry \(J\) in year \(t\)
• \(exp(\tau^{dd'})\): Distance (in km) between district \(d\) (student residence) to district \(d'\) (firm residence)

Utility Model

Individuals choose a field-of-study subject to the utility function:

\[\begin{align*} U_{im} = \alpha_{c} + \alpha_{t} + \beta_{mj} \Gamma_{djt} + \gamma X'_{i} + \varepsilon_{im} \end{align*}\]

Subscripts

\(i = \;\) individual, \(\; m = \;\) field-of-study, \(\; c = \;\) canton, \(\; t = \;\) year

• \(\alpha_{c}\) : Labor Market Area Effect
• \(\alpha_{t}\) : Year Effect
• \(\Gamma_{djt}\) : Vector of MNC Industries Presence Index
• \(X'_{i}\) : Vector of Individual Characteristics

Estimation Method

Multinomial Logit (MNL) model of the following equation:

\[\begin{align*} P(Y_{i} = m) = \dfrac{ \overbrace{exp(\alpha_{c} + \alpha_{t} + \beta_{mj} \Gamma_{djt} + X'_{i}\gamma + \varepsilon_{im})}^{\text{Utility of chosen field-of-study}} }{ \underbrace{\sum_{m'} exp(\alpha_{c} + \alpha_{t} + \beta_{m'j} \Gamma_{djt} + X'_{i}\gamma + \varepsilon_{im'}))}_{\text{Utilities across all possible field-of-study choices}} } \end{align*}\]