Anexo II. Documentación de código
II.3. Resto de ficheros
A.1
Proof of proposition 1.
The optimal rating qualityq 2[0;1]is derived by the …rst order condition:
(1 p)m=c0(q ) (A1.)
E¤ect of p: As p increases, the left hand side of A1. becomes smaller. To keep equality, the right hand side has to become smaller, too. Due to the characteristics of the cost functionc(:), the optimal rating quality q has to decrease.
E¤ect of m: m increases, if d increases and/or b decreases. The left hand side
increases, therefore the optimal rating quality q has to increase, too.
E¤ect ofc0(q ): A more e¢ cient rating technology is expressed with lower costs for a given rating standard. If rating quality increases, the rise of costs is lower for a more e¢ cient technology: c0high(q)< c0low(q). The marginal rating costs for a given rating quality is lower, if cost e¢ ciency becomes higher. The value of left hand side is exogenous and constant. If cost e¢ ciency increasesq , the rating standard has to rise in order to keep the right hand side constant.
A.2
Proof of proposition 2.
The optimal rating standard if rating quality q^ is unobservable and with costly regulation is derived by the …rst order condition:
As the left hand side of (A2.) is the same as the left hand side of (A1.), we have:
c0(q ) =c0(^q) +e0(^q) (A.2’)
Sincec(:) ande(:) are increasing and convex inq, thereforeq has to be larger than
^
q: q >q.^
For the proof of parts (i) to (iii) of proposition 2 see the proof of proposition 1. Part (iv) claims that the optimal rating standard is higher, if optimal penalty^
increases. An example for a rise in^ is an increase in the pro…ts per rating due to market power in the CRA sector. The necessary e¤ort that has to be employed in order to enforce a given rating standard, is given by:
e(^q) =h 1 c(^q)
^ (A3.)
If^increases, the optimal e¤ort corresponding to a given rating standard decreases. From this follows that the marginal e¤orte0(:)decreases either. The right hand side
therefore becomes smaller for a given rating standard. Due to the characteristics of the functions c(:) and e(:), the optimal rating standard q^must increase to ensure equality of equation (A.3).
A.3
Proof of proposition 3.
Why the optimal rating standard in the presence of joint provision of services (q) and~
analogously in the case of a direct bribe, is lower than the optimal rating standard in 3.3 (q), can be shown as follows: The …rst-order condition of the regulator’s decision^
problem is given by
(1 p)m=c0(~q) +e0(~q) (A.4) The left hand side of A4. the same as in equation A.2, therefore it must be true that:
Since the e¤ects are analog for the direct ( ) and the indirect bribe ( ), the proof is presented for > 0 only. Taking into account the potential bribe , the necessary e¤ort to enforce a given rating standard q is larger in the case of joint provision of services:
e(q) =h 1 c(q)
^ < e(q) =h
1 c(q) +
^
The marginal e¤ort with bribe is therefore larger than the marginal e¤ort without bribe for a given rating quality. Therefore the optimal rating standard has to be lower in the case of joint provision of services, to ensure equality of equation (A.4’), given the characteristics ofc(:)and e(:): q <~ q.^
That the optimal rating standard is decreasing as the rent of the joint provision of services increases can be shown by total di¤erentiating equation (A.4) and using e0(~q) = h0 1 c(~q)+ ^ c0(~q) ^ : d(1 p)m | {z } =0 =c00(~q)dq~+e00(~q)dq~+h00 1 c(~q) + ^ c0(~q) ^2 d ! ddq~ = h00 1 c(~q)+ ^ c0(~q) ^2 c00(~q) +e00(~q) <0 (A.5)
Equation is smaller than zero, because of the convexity of e(:), c(:), and h 1().
A.4
Proof of proposition 4.
The social value with joint provision of services is given by V~, whereas the social value with rating only is given by V^. If = 0, then V~ = ^V. Equation (A.5) showed that ddq~ <0. Therefore, given the characteristics ofc(:), it must be true that
@c0(~q)
@ < 0. Above, we have shown that e0(~q) = h0
1 c(~q)+ ^
c0(~q)
^ . Di¤erentiating with respect to yields @e@0(~q) =h00 1 c(~q)+
^
c0(~q)
^2 >0. The partial derivative of
the social value V~ and V^ with respect to is given by: @@V~ = c01(~q)(c0(~q) e0(~q))
and @@V^ = 0. As increases, the di¤erence (c0(~q) e0(~q)) decreases, since c0
the social value V~ decreases further as increases. Therefore, it is optimal to prevent the joint provision of services for any >0, since V <~ V^.
A.5
Proof of proposition 5.
If @@V~ > 0 at = 0, then we must have c0(~q) > e0(~q). From this follows that
~
V increases, if becomes larger, as long as is relatively small. Given the characteristics of @c@0(~q) and @e@0(~q), V~ decreases, for relatively large values of . Therefore, at some critical level = , we have V~ = ^V. For values < , we have V >~ V^, and therefore allowing the joint provision of services is optimal. For values > , we haveV <~ V^, and therefore forbidding the joint provision of services is optimal.