（论文）The Economics of Buyer Uncertainty: AdvanceSelling vs. Probabilistic selling

author: Scott Fay, Jinhong Xie
Marketing Sciece (2010)
原论文地址：https://pubsonline.informs.org/doi/10.1287/mksc.1100.0576

Notation

$v_{\theta H}$: consumer $\theta$'s value for her preferred product
$v_{\theta L}$: consumer $\theta$'s value for her less preferred product
Max_Value-Heterogeneity:
$${\Delta }_v=\underset{\theta }{\mathrm{Max}}[v_{\theta H}]-\underset{\theta }{\mathrm{Min}}[v_{\theta L}]$$
Strength-Heterogeneity:
$${\Delta }_s=\underset{\theta }{\mathrm{Max}}[v_{\theta H}-v_{\theta L}]-\underset{\theta }{\mathrm{Min}}[v_{\theta H}-v_{\theta L}]$$
Total Heterogeneity:
$$\Delta ={\Delta }_v+{\Delta }_s$$
Relative Heterogeneity:
$$\gamma =\frac{{\Delta }_s}{\Delta }$$

$R=\underset{\theta }{\mathrm{Min}}[v_{\theta H}]$, where $\theta \backsim U[0,\Delta ]$. Then considering he following value functions for a given consumer $\theta$
$$v_{\theta H}=R+(1-\gamma )\theta$$
$$v_{\theta L}=R+(1-2\gamma )\theta$$

model

Traditional Selling

$$\begin{array}{rl}{\Pi }^{TS}& =(P^{TS}-c){\int }_{v_{\theta H}{P}^{TS}}^{R+{\Delta }_v}\frac{1}{\Delta }d{v}_{\theta H}\\ & =(P^{TS}-c)(1-\frac{P^{TS}-R}{{\Delta }_v})\end{array}$$
when $c\le R-{\Delta }_v$, the optimal price under TS is $P^{TS\ast }=R$

Advanced Selling

a seller is capable of committing in advance to announced spot prices if the latter are observable at the time when customers are making advance purchases.通常情况下，seller都会同时提供advanced price和gate price。

• 需要产生一个实施成本$F_{AS}$

每个产品的期望价值为
$$\begin{array}{rl}EV& ={\int }_{\theta =0}^{{\Delta }_v+{\Delta }_s}\frac{v_{\theta H}+v_{\theta L}}{2\Delta }\\ & =R+\frac{2{\Delta }_v-{\Delta }_s}{4}\\ & =R+\frac{(2-3\gamma )\Delta }{4}\end{array}$$
所以得到的利润为(因为EV一样，所以comsumer会选择价格最低的商品)
$${\Pi }^{AS}=\{\begin{array}{ll}Min[P_1^{AS},P_2^{AS}]-c-F_{AS}& ,Min[P_1^{AS},P_2^{AS}]\le R+\frac{2{\Delta }_v-{\Delta }_s}{4}\\ -F_{AS}& ,Min[P_1^{AS},P_2^{AS}]>R+\frac{2{\Delta }_v-{\Delta }_s}{4}\end{array}$$
因此最优的价格为 $P_j^{AS}=EV$

与TS比较
当${\Delta }_v>\hat{\Delta }$时，${\Pi }^{AS}>{\Pi }^{TS}$。其中，$\hat{\Delta }=\frac{4F_{AS}+{\Delta }_s}{2}$
或者$\gamma <\frac{2}{3}-\frac{4F_{AS}}{3\Delta }$

要求：${\Delta }_v$ is large and ${\Delta }_s$ is small.

Probabilistic Selling

in the spot time, sellers既给特定商品$P_i^{PS}$, 也给概率商品$P_O^{PS}$

每个consumer $\theta$对模糊商品的期望为
$$\begin{array}{rl}{v}_{\theta o}& =\frac{v_{\theta H}+v_{\theta L}}{2}\\ & =R+\frac{\theta (2{\Delta }_v-{\Delta }_s)}{2\Delta }\\ & =R+\frac{(2-3\gamma )\theta }{2}\end{array}$$

当$\gamma <2/3$, $v_{\theta o}$随$\theta$递增而递增；
当$\gamma >2/3$, $v_{\theta o}$随$\theta$递增而递减

在需求函数对称的情况下$P_1^{PS}=P_2^{PS}=P^{PS}$
$D_S(P^{PS},P_o^{PS})$:购买特定产品的数量
$D_o(P^{PS},P_o^{PS})$:购买模糊商品的数量
所以消费者如果购买特定商品，会购买自己prefer的

利润为：
$${\Pi }^{PS}=(P^{PS}-c)D_S(P^{PS},P_o^{PS})+(P_o^{PS}-c)D_o(P^{PS},P_o^{PS})-F_{PS}$$

$\gamma <2/3$
$v_{\theta H}-v_{\theta o}=\frac{\gamma \theta }{2}$
所以，存在$\hat{\theta },{\theta }_L$使得：
当$\theta \in (\hat{\theta },\Delta )$，consumer会购买自己喜欢的特定商品
当$\theta \in ({\theta }_L,\hat{\theta })$, consumer会购买模糊商品
当$\theta \in (0,{\theta }_L)$, consumer不会购买
$${\Pi }^{PS}=(P^{PS}-c)\frac{\Delta -\hat{\theta }}{\Delta }+(P_o^{PS}-c)\frac{\hat{\theta }-{\theta }_L}{\Delta }-F_{PS}$$

${\theta }_L$是买模糊商品和不购买的边界：$v_{{\theta }_{L}o}-P_o^{PS}=0\Rightarrow P_o^{PS}=R+\frac{(2-3\gamma ){\theta }_L}{2}$
$\hat{\theta }$是买模糊商品和购买特定商品的边界：$v_{\hat{\theta }o}-P_o^{PS}=v_{\hat{\theta }H}-P^{PS}\Rightarrow P^{PS}=\frac{\gamma \hat{\theta }}{2}+P_o^{PS}$
代入利润方程，得到分别对${\theta }_L,\hat{\theta }$的凹函数，分别求导，得到：${\theta }_L<0forc\le R-\Delta$, $\hat{\theta }=\Delta /2$

因此，${\theta }_L=0,P_o^{PS}=R,\hat{\theta }=\frac{\Delta }{2},P^{PS}=R+\frac{{\Delta }_s}{4}$
$${\Pi }^{PS}=R-c+\frac{{\Delta }_s}{8}-F_{PS}$$

$\gamma >2/3$
$v_{\theta o}$随$\theta$递减
$v_{\theta H}$随$\theta$递增
所以存在${\theta }_H,{\theta }_L$使得
当$\theta \in ({\theta }_H,\Delta )$,consumer会购买特定商品
当$\theta \in ({\theta }_L,{\theta }_H)$,consumer不会购买
当$\theta \in (0,{\theta }_L)$,consumer会购买模糊商品
$${\Pi }^{PS}=(P^{PS}-c)\frac{\Delta -{\theta }_H}{\Delta }+(P_o^{PS}-c)\frac{{\theta }_L}{\Delta }-F_{PS}$$

${\theta }_L$是会购买模糊商品的最大$\theta$:$v_{{\theta }_{L}o}-P_o^{PS}=0\Rightarrow P_o^{PS}=R+\frac{(2-3\gamma ){\theta }_L}{2}$
${\theta }_H$是会购买特定商品的最小$\theta$:$v_{{\theta }_{H}H}-P^{PS}=0\Rightarrow P^{PS}=R+(1-\gamma ){\theta }_H$
代入利润方程，得到分别对${\theta }_L,{\theta }_H$的凹函数，分别求导，得到：${\theta }_H\le 0,forc\le R-c$
所以，${\theta }_H={\theta }_L$
$${\Pi }^{PS}=P^{PS}\frac{\Delta -\hat{\theta }}{\Delta }+P_o^{PS}\frac{\hat{\theta }}{\Delta }-c-F_{PS}$$
同样的方法，得到$\hat{\theta }=\frac{(1-\gamma )\Delta }{\gamma }$
$${\Pi }^{PS}=R+\frac{(1-\gamma {)}^2\Delta }{2\gamma }-c-F_{PS}$$

所以
$${\Pi }^{PS}=\{\begin{array}{ll}R+\frac{(1-\gamma {)}^2\Delta }{2\gamma }-c-F_{PS}& ,\gamma \ge 2/3\\ R-c+\frac{{\Delta }_s}{8}-F_{PS}& ,\gamma <2/3\end{array}$$

与TS比较

$${\Pi }^{PS}-{\Pi }^{TS}=\{\begin{array}{ll}\frac{(1-\gamma {)}^2\Delta }{2\gamma }-F_{PS}& ,\gamma \ge 2/3\\ \frac{{\Delta }_s}{8}-F_{PS}& ,\gamma <2/3\end{array}$$
当$\gamma <2/3$，函数随$\gamma$递增；当$\gamma \ge 2/3$，函数随$\gamma$递减。所以最大值为$\Delta /12-F_{PS}$($F_{PS}<\Delta /12$)
则，当$\frac{8F_{PS}}{\Delta }<\gamma <\frac{\Delta +F_{PS}-\sqrt{F_{PS}(2\Delta +F_{PS})}}{\Delta }$时，${\Pi }^{PS}>{\Pi }^{TS}$
或者$8F_{PS}<{\Delta }_s<\frac{{\Delta }_v^2}{2F_{PS}}$

necessary condition:${\Delta }_v>4F_{PS}$

Advanced Selling & Probabilistic Selling

$${\Pi }^{PS}-{\Pi }^{AS}=\{\begin{array}{ll}\frac{\Delta (2-\gamma (6-5\gamma ))}{4\gamma }+F_{AS}-F_{PA}& ,\gamma \ge 2/3\\ \frac{\Delta (7\gamma -4)}{8}+F_{AS}-F_{PA}& ,\gamma <2/3\end{array}$$
是个严格增函数

当$\gamma >\hat{\gamma }(\Delta )$时，${\Pi }^{PS}-{\Pi }^{AS}>0$
$$\hat{\gamma }(\Delta )=\{\begin{array}{ll}1& ,\Delta <{\Delta }_1\\ \frac{6\Delta -4(F_{AS}-F_{PS})+\sqrt{[6\Delta -4(F_{AS}-F_{PS}){]}^2-40{\Delta }^2}}{10\Delta }& ,{\Delta }_1\le \Delta {\Delta }_2\\ \frac{4(\Delta -2(F_{AS}-F_{PA}))}{7\Delta }& ,\Delta \ge {\Delta }_2\end{array}$$
${\Delta }_1=4(F_{PS}-F_{AS})$
${\Delta }_1=12(F_{PS}-F_{AS})$

将两个条件合并，得到三个销售方式的选择情况。