Codecision procedure biais: the European legislation game

Abstract

The European Union’s codecision procedure is analyzed as a bargaining game between the Council and the European Parliament. The model shows that the policy outcome is biased towards the ideal point of the institution that is closest to the status quo, when the negotiation leads to an agreement. If the Council’s ideal policy is closer to the status quo, as is the case for example concerning the duration of maternity leave, the resulting policy will remain closer to the Council’s than to the European Parliament’s position. Furthermore, the result of the codecision procedure is Kaldor–Hicks inefficient. The developments of common policies concerning asylum, fishery and maternity leave illustrate both the possibility of the failure of negotiations and the influence of the Council on the outcome.

Appendix

Proof of proposition 1

Solving the program (1) is equivalent to solving the following equation:

$$ A {{}^2} {x {{}^2} +(2AC-B {{}^2} )x+C {{}^2} =0} $$

(2)

with

$$ \begin{aligned} A &=2\phi _{EP}\phi _{CM} \\ B &=(\phi _{EP}+\phi _{CM})\frac{3}{2}-2\phi _{EP}\phi _{CM}\sqrt{q}\\ C &=1-\frac{\phi _{EP}+\phi _{CM}}{2}\sqrt{q}\\ \end{aligned} $$

The determinant of the equation is positive for all $q\varepsilon \lbrack 0,x_{CM}^{\ast}]\cup \lbrack x_{EP}^{\ast},1]$ with $x_{CM}^{\ast}=\frac{1}{4\phi _{CM}^{2}}$ and $x_{EP}^{\ast}=\frac{1}{4\phi _{EP}^{2}}.$

$$ \bar{x}_{1}=\frac{-{2AC+B {{}^2} (1}-\sqrt{(B^{2}-4AC)})}{2A {{}^2}} $$

(3)

Therefore, Eq. (3) has two solutions. The relevant one necessarily belongs to the bargaining set $S=[x_{CM}^{\ast},x_{EP}^{\ast}]$. For all $q<x_{CM}^{\ast},$ we find that $\bar{x}_{1}=\frac{-(2AC-B{{}^2})-\sqrt{\Updelta}}{2A{{}^2}}$ is always the relevant solution. For all $q>x_{EP}^{\ast},$ $\bar{x}_{2}=\frac{-(2AC-B{{}^2})+\sqrt{\Updelta}}{2A{{}^2}}$ is always the relevant solution.

We show that $x_{EP}^{\ast}+x_{CM}^{\ast}-2\bar{x}_{1}>0$ when q = 0. In other words, $\bar{x}_{1}$ is lower than the middle of the interval $[x_{CM}^{\ast},x_{EP}^{\ast}].$ Furthermore, we know that $\bar{x}_{1}=x_{CM}^{\ast}$ if $q=x_{CM}^{\ast}.$ Deriving $\bar{x}_{1}$ with respect to q, we show that $\frac{d\bar{x}_{1}}{dq}<0$ on [0,1].

Indeed,

$$ \frac{d\bar{x}_{1}}{dq}=\frac{1}{32c {{}^2} e {{}^2} }[\frac{2ec(E+4ce\sqrt{q}-3c-3e)(E-4ce\sqrt{q}-c-e)}{\sqrt{q}E}] $$

with $E=\sqrt{16c{{}^2}e{{}^2}q-8c{{}^2}e \sqrt{q}+9c{{}^2}-8ce{{}^2}\sqrt{q}-14ce+9e{{}^2}},$ and $E+4ce\sqrt{q}-3c-3e<0$ and $E+4ce\sqrt{q}-c-e>0$ for all $q\in \lbrack 0,1],0<e<c<1$.

Furthermore, we obtain that $x_{EP}^{\ast}+x_{CM}^{\ast}-2\bar{x}_{2}>0$ when q = 1 and

$$ \frac{d\bar{x}_{2}}{dq}=\frac{1}{32c {{}^2} e {{}^2}} [-\frac{2ec(-E+4ce\sqrt{q}-3c-3e)(-E-4ce\sqrt{q}-c-e)}{\sqrt{q}E}] $$

with $-E+4ce\sqrt{q}-3c-3e<0$ and $-E+4ce\sqrt{q}-c-e<0$ for $q\in \lbrack 0,1],0<e<c<1$.

Even at the extremities q = 0 and q = 1, there exists a biais toward q. □