Limitations and future extensions This chapter initiates the application of panel VAR models in the field of educational studies. In educational research, the analysis of temporal patterns in student outcomes (e.g., performance) is often overlooked in favor of cross-sectional or experimental designs. However, studying temporal patterns provides a framework for capturing dynamics over time, offering a different perspective on educational processes. This paper pioneers the use of panel VAR models in the field of education, applying it to examine skill development. Panel VAR models capture the complex nature of the skill development process, although simplifications are necessary in modeling. For instance, we assume linear relationships between the variables, meaning that we do not directly account for nonlinearity (such as suggested by e.g., Cunha et al. (2010)). Future work could build on our findings. Our model includes an individual fixed effect, accounting for student heterogeneity. Future research might concentrate on differences between high and low achievers, as it is known from the literature that reading has a supportive role in math development, especially for students with average and strong reading skills, but not for students with low reading skills (Erbeli et al., 2021). Another interesting topic for future research is differences between investments in the early and later years of primary education. Moreover, whereas we examine the response to a shock over a time period of five years, students move on to secondary education at the end of sixth grade. In all probability, there will be a structural break in this transition, meaning that the impact of this shock differs over the years. And lastly, this chapter explores how reading, spelling, and math skills interact in a Dutch primary school context. Future research could focus on other skills or on how the interplay be103
RkJQdWJsaXNoZXIy MjY0ODMw