1: \begin{definition}\label{definition: RRC estimator} (RRC estimator)

2:    Let $k\in \{1,\dots,p-1\}$. The RRC estimator for the total effect of $\rvX_i$ $(1\le i\le k)$ on $\rvY$ in a joint intervention on $(\rvX_1,\dots,\rvX_k)$ is defined recursively as follows

3:    (cf. Theorem \ref{theorem: RRC for multiple interventions}):

4:    \begin{align*}

5:       \hat{\effect}_{ip}^{[k]} =   \left\{

6:         \begin{array}{ll}

7:             0 & \text{if $\rvY \in \PaVec_i$} \\

8:             \hat{\effect}_{ip}^{[k]\setminus \{j\}} - \hat{\effect}_{ik}^{[k] \setminus \{j\}}\hat{\effect}_{kp}^{[k]\setminus \{i\}} & \text{otherwise}

9:         \end{array}

10:         \right.,

11:    \end{align*}

12:    where the adjusted regression estimator $\hat{\effect}^{(i)}_{ij} = \hat{\effect}_{ij}$ is defined in \eqref{eq: causal effect estimate} and we fix $j = \max([k] \setminus \{i\})$.

13: \end{definition}

14: