lt;br><br>Resource partner commitment $R(q)lt;br> | Demandâsupply asymmetry driven by quality rather than quantity<br><br>$P_c(q)=q$ , $P_r(q)=1-q$ | | | **2** | quality $q$ | Customer commitment $C(q)lt;br><br>Resource partner commitment $R(q)$ | Statistically capture stakeholders' responsiveness to quality<br><br>$P_c(q)=\dfrac1{1+e^{-\beta_c * q}}$ , $P_r(q)=\dfrac1{1+e^{\ \beta_r * q}}$ | | ## 2.0 Classic Newsvendor Step 0 (classic newsvendor) treats supply quantity as a **decision variable** that the entrepreneur directly controls before observing demand. The entrepreneur minimizes expected cost by choosing quantity $q$ to balance opportunity cost $(D-q)(p-c)$ when demand exceeds supply against overage cost $(q-D)c$ when supply exceeds demand. This yields the familiar critical fractile solution $F(q^*) = (p-c)/p$. Detail in Appendix A.0. ## 2.1 StepâŻ1: Linear Quality Model Step 1 fundamentally transforms Step 0's framework by making supply a **stochastic outcome** dependent on quality decisions. The entrepreneur now chooses product quality $q \in [0,1]$, which influences both customer willingness to buy (probability $P_c(q) = q$) and resource partner willingness to supply (probability $P_r(q) = 1-q$). This creates a 2Ă2 probability grid with four possible outcomes: successful matches generate revenue $V$, mismatches incur either opportunity cost $C_u$ (customer wants but partner can't deliver) or overage cost $C_o$ (partner can deliver but customer doesn't want), and the no-transaction case incurs no cost. Unlike Step 0's single mismatch channel (quantity vs demand), Step 1 introduces dual mismatch risks through quality-driven stakeholder responses, fundamentally changing the optimization from a critical fractile to a dual-cost balancing rule that incorporates the match bonus $V$. ### 2.1.1 Mathematical setup Quality choice $q\in[0,1]$ influences two stakeholders' commitments, $C(q)$ for customer and $R(q)$ for resource partner. This gives the mismatchâloss function in Appendix A.1. This reframes newsvendor's underage cost: opportunity cost from lost sales now stem from _lack of partner supply_ rather than underâordering. ### Proposition 1 _Optimal quality under linear commitment probabilities_ $ q^{*}=\frac{V+2C_o}{2\,(C_u+C_o+V)}, \qquad \frac{\partial q^{*}}{\partial V}=\frac{C_u-C_o}{2\,(C_u+C_o+V)^{2}} . $ ### 2.1.2 Business intuition (4A, 4B) > **When is $q^{*}$ high?** > (i) $C_o\!\gg\!C_u$âexpensive surplus; managers raise quality to pull demand. > (ii) $V$ large *and* $C_u>C_o$âopportunity seekers court customers even at surplus risk. > (iii) $V$ small *and* $C_o>C_u$ârisk avoiders still favour higher quality to dodge leftovers. Throughout this paper, we call proposition 1's result "Quality adjusts to avoid the more expensive type of mismatch", **cost-priority principle**. *Why interesting?* Unlike the classic critical-fractile, which moves with a single cost ratio, here the *direction* of $V$âs impact flips when the cheaper mismatch changes, highlighting a dual-cost amplification mechanism absent in quantity-only settings. 1. **Dualâcost amplification** â raising the cheaper mismatch cost tilts $q^{*}$ toward avoiding the more expensive side. 2. **Match bonus effect** â a larger $V$ pulls $q^{*}$ in the direction of the *scarcer* party, explaining why StepâŻ1 departs from the pure cost ratio of StepâŻ0. todo: need to add asymmetric linear for duality --- ## 2.2 StepâŻ2: Sigmoid Quality Model ### 2.2.1 Mathematical setup Commitment functions become logistic: $P_c,P_r$ in AppendixâŻA.2. This preserves the Bernoulliâstate logic while reflecting empirically observed *Sâshaped* responsiveness to quality signals. ### Proposition 2 _Optimal quality under sigmoid commitment probabilities_ With logistic response curves $P_c(q) = \frac{1}{1+e^{-q}}$ and $P_r(q) = \frac{1}{1+e^{q}}$, the expected loss function $L(q) = C_u P_c(1-P_r) + C_o P_r(1-P_c) - V P_c P_r$ is strictly quasi-convex. The unique optimal quality is: $q^* = \ln\left(\frac{2C_o + V}{2C_u + V}\right)$ The comparative static with respect to match bonus $V$ is: $\frac{\partial q^*}{\partial V} = \frac{2(C_u - C_o)}{(2C_u + V)(2C_o + V)}$ ### 2.2.2 Business intuition (5A, 5B) - Behavioural steepness $(\beta_c,\beta_r)$ moderates the costâpriority principle: when one side reacts sharply to quality, its preferences weigh more heavily in $q^{\dagger}$, potentially overriding small cost asymmetries. The sigmoid model preserves Step 1's cost-priority principle while capturing realistic S-shaped stakeholder responses. The closed-form solution $q^* = \ln\left(\frac{2C_o + V}{2C_u + V}\right)$ reveals several key insights: **Cost-priority mechanism**: When $C_o > C_u$, we have $q^* > 0$ (positive log), pushing quality higher to avoid expensive overage. Conversely, when $C_u > C_o$, we get $q^* < 0$, lowering quality to avoid costly shortages. The match bonus $V$ moderates this effectâlarger $V$ compresses the ratio toward 1, pulling $q^*$ toward 0 (the symmetric point where $P_c = P_r = 0.5$). **Stakeholder's symmetric responsiveness to quality**: With $\beta_c = 1$ and $\beta_r = -1$, the model captures mirror-image behaviorsâcustomers become more willing as quality rises while partners become less willing at the same rate. This symmetry enables the elegant logarithmic solution and ensures that the direction of $V