Fashion Buying

Pre-season commitment · Quick response

How many units of a short-lifecycle fashion item should a retailer commit to before the season starts? Seasons are 8–12 weeks; in-season replenishment is either impossible or expensive; unsold inventory is heavily marked down. The quick-response option lets the retailer place a smaller pre-season buy and top up after observing early sell-through. Foundational papers: Fisher (1997) on functional vs. innovative products; Fisher & Raman (1996) on accurate response; Cachon & Swinney (2011) on fast-fashion and strategic consumers; Iyer & Bergen (1997).

Why it matters

Scale of the buy decision · documented quick-response lift

~30%
Share of US fashion-apparel revenue sold at marked-down prices — the pre-season buy quantity is a structural lever for this entire downstream loss.
Pashigian (1988), American Economic Review.
~50%
Reduction in pre-season buy commitment achievable with quick-response: replenish mid-season once early sell-through is observed.
Fisher & Raman (1996), Operations Research 44(1).
~40%
Share of merchandise designed and produced during the season at Zara — vs. ~20% industry standard. Quick-response in industrial form.
Caro & Martínez-de-Albéniz (2012); industry analyses.
Q†
Strategic-consumer interaction (Cachon & Swinney 2011): customers anticipating markdowns wait, eroding initial-price revenue and reshaping the optimal pre-season buy.
Cachon & Swinney (2011), Management Science 57(4).

Where the decision sits

Pre-season commitment · mid-season top-up · end-of-season clearance

Fashion buying applies whenever a retailer must commit to a quantity well before observing demand, with limited or expensive recourse during the selling window. The classic example: an apparel retailer placing orders 6–9 months before the season starts with offshore manufacturing. The quick-response variant adds a second decision point mid-season: after observing a fraction of total demand, place a smaller top-up order at a price premium for speed. The buy decision feeds directly into the markdown problem — whatever pre-season quantity is committed becomes the inventory the markdown DP must clear by season end.

Forecastseason demand \(D\)
Set pre-season buyquantity \(Q_{\text{pre}}\)
Observe sell-throughearly signal \(S_{\text{early}}\)
Quick-response ordertop-up \(Q_{\text{qr}}\)
Markdown unsoldclear leftover

Problem & formulation

Newsvendor base case · two-stage stochastic with Bayesian update

OR family
Two-Stage Stochastic Programming
Base case
Newsvendor (closed form)
Solver realism
★★ Closed form + Monte Carlo
Reference
Fisher & Raman (1996)

Sets and uncertainty

SymbolMeaningDomain
\(D \sim F\)Total season demand — uncertain at pre-season, refined after early salesrandom, \(\geq 0\)
\(\tau \in (0, 1)\)Fraction of season elapsed before quick-response decisione.g., 0.3
\(S_{\text{early}}\)Observed sales in the first \(\tau\) of the seasonrandom, observed

Cost and revenue parameters

SymbolMeaningUnit
\(p\)Selling price per unit at full price$ / unit
\(c_{\text{pre}}\)Pre-season unit cost (offshore, lead time \(\sim\) months)$ / unit
\(c_{\text{qr}}\)Quick-response unit cost — premium for speed, \(c_{\text{qr}} > c_{\text{pre}}\)$ / unit
\(v\)Salvage value per leftover unit (clearance / outlet), \(v < c_{\text{pre}}\)$ / unit

Decision variables

SymbolMeaningStage
\(Q_{\text{pre}}\)Pre-season order quantity (committed before observing \(D\))first
\(Q_{\text{qr}}\)Quick-response top-up quantity (chosen after \(S_{\text{early}}\) revealed)second, recourse

Base case — pure newsvendor (no QR)

When quick response is unavailable, the entire season inventory must be committed pre-season. Underage cost \(c_u = p - c_{\text{pre}}\) (lost margin on a sale not made); overage cost \(c_o = c_{\text{pre}} - v\) (sunk cost on a leftover unit). The standard newsvendor result:

$$Q_{\text{pre}}^{\ast} \;=\; F^{-1}\!\left(\frac{c_u}{c_u + c_o}\right) \;=\; F^{-1}\!\left(\frac{p - c_{\text{pre}}}{p - v}\right)$$

The critical fractile balances stocking deeper (more sales risk-free upside) against stocking shallower (less leftover risk).

With quick response — two-stage stochastic

After observing early sales \(S_{\text{early}}\) from \(Q_{\text{pre}}\) units in fraction \(\tau\) of the season, update the posterior on remaining demand \(D_{\text{rest}}\). Then place a quick-response order:

$$Q_{\text{qr}}^{\ast}(S_{\text{early}}) \;=\; \max\!\left\{0,\; F_{\text{rest}}^{-1}\!\left(\frac{p - c_{\text{qr}}}{p - v}\right) - \big(Q_{\text{pre}} - S_{\text{early}}\big) \right\}$$

QR is itself a newsvendor on the residual demand, but at the higher unit cost \(c_{\text{qr}}\) — so the QR critical fractile is lower and the QR top-up is conservative.

The first-stage problem is then to choose \(Q_{\text{pre}}\) anticipating optimal QR recourse:

$$\max_{Q_{\text{pre}} \,\geq\, 0}\; \mathbb{E}_{D}\!\left[ \,p \cdot \min(D,\, Q_{\text{pre}} + Q_{\text{qr}}^{\ast}) \;-\; c_{\text{pre}}\, Q_{\text{pre}} \;-\; c_{\text{qr}}\, Q_{\text{qr}}^{\ast} \;+\; v \cdot \big(Q_{\text{pre}} + Q_{\text{qr}}^{\ast} - D\big)^{+} \,\right]$$

Solved numerically; the optimal pre-season buy is reduced relative to the no-QR newsvendor because QR partially insures against under-stocking.

Strategic consumers (Cachon & Swinney 2011)

When customers anticipate markdowns, full-price demand falls. Counter-intuitive insight: a smaller pre-season buy helps the retailer when strategic consumers are prevalent, because committed inventory is the source of markdown risk. Quick response and fast fashion are partially substitutes for the same problem — both reduce pre-season commitment.

Backup-agreement model (Eppen & Iyer 1997)

A specific contractual form of QR: the supplier commits to a fixed backup quantity \(B\) at a contracted price \(c_B \in (c_{\text{pre}}, p)\); the retailer pays a per-unit reservation fee \(r\) up front. The optimal split between firm pre-season order and backup capacity has a closed-form structure when demand is normal.

Interactive solver

Newsvendor closed form + Monte-Carlo two-stage stochastic with QR

Fashion buying solver
Normal demand · closed-form base + simulated QR recourse
★★ Closed form + MC
Total season units
Demand uncertainty
Premium for speed
Outlet recovery
e.g., 0.3 of season
—
\(Q_{\text{pre}}\) without QR
—
\(Q_{\text{pre}}\) with QR
—
Expected profit ($)
—
Expected lost sales
—
Expected leftover
—
Avg QR order
Demand PDF \(Q_{\text{pre}}\) without QR \(Q_{\text{pre}}\) with QR Profit histogram (active policy)

Under the hood

The base case uses the closed-form newsvendor: \(Q_{\text{pre}}^{\ast} = \mu + \sigma \cdot \Phi^{-1}\big((p - c_{\text{pre}})/(p - v)\big)\) under normal demand, with \(\Phi^{-1}\) computed via the Beasley-Springer-Moro algorithm. The QR case is solved by Monte-Carlo: for each candidate \(Q_{\text{pre}}\) on a coarse grid, simulate \(N\) demand draws, compute early sell-through assuming demand is split proportionally over the season, update the posterior on residual demand by scaling the prior by \((1 - \tau)\) and applying a Bayesian variance reduction proportional to \(\tau\), and choose \(Q_{\text{qr}}\) by the residual newsvendor formula. The grid maximum of expected profit gives the optimal \(Q_{\text{pre}}\) under QR. Scales linearly in \(N\); sub-second in the browser.

Reading the solution

What a buyer actually does with the optimal pre-season quantity

Three patterns to watch for

  • QR shrinks the pre-season buy. When QR is available, the optimal \(Q_{\text{pre}}\) drops — often by 30–50%. The retailer trades higher per-unit cost on QR units for the option of not buying them at all if demand turns out weak.
  • QR cost premium matters. If \(c_{\text{qr}}\) is close to \(c_{\text{pre}}\) (small premium for speed), QR dominates; the retailer commits very little pre-season. As \(c_{\text{qr}} \to p\), QR becomes uneconomic and the policy reverts to pure newsvendor.
  • Earlier observation, deeper cut. Smaller \(\tau\) means QR happens too early to learn much — less benefit. Larger \(\tau\) means more learning but less time to use the QR units — also less benefit. There is an interior optimum, typically \(\tau \in [0.2, 0.4]\) for an 8–12 week season.

Sensitivity questions the model answers instantly

  • What if the QR premium drops 20%? — \(Q_{\text{pre}}\) falls further; expected profit rises sharply.
  • What if salvage \(v\) goes to zero (forced destruction)? — \(Q_{\text{pre}}\) shrinks aggressively and QR becomes more valuable.
  • Higher demand uncertainty \(\sigma\)? — QR option value rises; the gap between QR and no-QR profits widens.

Model extensions

From single-SKU buy to fast-fashion, multi-product, and strategic-consumer variants

Strategic consumers

Customers anticipate markdowns and wait. Cachon & Swinney 2011 show pre-season buy is reduced because committed inventory is the source of strategic-wait risk.

Supplier backup-agreement

Eppen & Iyer 1997 backup-agreement model: supplier reserves a fixed backup quantity at a contracted higher price; retailer pays a reservation fee. Closed form for normal demand.

Multi-product joint buy

Buy quantities for color/size SKUs of the same style with shared budget or shared lead-time slot. Demand correlation drives portfolio risk reduction.

Pre-season + in-season pricing

Jointly optimise the buy quantity and the markdown schedule. Larger \(Q_{\text{pre}}\) is rational if the retailer expects an aggressive markdown to clear leftovers.

Markdown →
Newsvendor base case

The pure-newsvendor reduction without QR — the foundational single-period inventory model. Identical critical-fractile structure.

Newsvendor →
Assortment + buy

Joint decision: which SKUs to carry (assortment) and how much of each to buy. Demand substitution effects make the buy quantities interdependent.

Assortment →
Fashion-cycle dynamics

Multi-season learning: this season's sell-through informs next season's buy. Caro & Gallien (2010) Zara network model with cross-store reallocation.

Quick-commerce / on-demand

Modern fast-fashion (Zara, Shein) push QR to its limit: design + manufacture cycles in weeks. Effectively shifts \(Q_{\text{pre}} \to 0\) and replaces it with continuous in-season production.

Key references

Foundational fashion buying and quick-response literature

Fisher, M. (1997).
What is the right supply chain for your product?
Harvard Business Review 75(2): 105–116.
Fisher, M. & Raman, A. (1996).
Reducing the cost of demand uncertainty through accurate response to early sales.
Operations Research 44(1): 87–99. doi:10.1287/opre.44.1.87
Cachon, G. P. & Swinney, R. (2011).
The value of fast fashion: Quick response, enhanced design, and strategic consumer behavior.
Management Science 57(4): 778–795. doi:10.1287/mnsc.1100.1303
Iyer, A. V. & Bergen, M. E. (1997).
Quick response in manufacturer-retailer channels.
Management Science 43(4): 559–570. doi:10.1287/mnsc.43.4.559
Eppen, G. D. & Iyer, A. V. (1997).
Backup agreements in fashion buying — the value of upstream flexibility.
Management Science 43(11): 1469–1484. doi:10.1287/mnsc.43.11.1469
Caro, F. & Martínez-de-Albéniz, V. (2012).
Product and price competition with satiation effects.
Management Science 58(7): 1357–1373. doi:10.1287/mnsc.1110.1465
Caro, F. & Gallien, J. (2010).
Inventory management of a fast-fashion retail network.
Operations Research 58(2): 257–273. doi:10.1287/opre.1090.0698
Hammond, J. H. & Raman, A. (1996).
Sport Obermeyer Ltd.
Harvard Business School Case 9-695-022 (foundational teaching example for fashion buying with quick response).

Back to the retail domain

Fashion buying sits in the Place × Strategic intersection — the upstream commitment that determines what the markdown DP and assortment problems must work with.

Open Retail Landing
Educational solver · normal-demand assumption with simplified Bayesian update · validate against your own demand-history covariance and supplier QR contracts before live buys.