Sourcing Arbitrage in Fashion PE: A Research Suite

Three documents. One argument. Built from the ground up.

Most PE deal teams model freight and tariff risk as a sensitivity line item — a cost that goes up or down depending on macro conditions. That framing misses the point. Trade-cost volatility is not symmetric. It falls disproportionately on mass-market apparel companies, and within mass-market, it falls disproportionately on companies sourcing from the wrong geographies. Identifying which geographies, quantifying the difference, and translating it into LBO underwriting language is what this body of work does.

The empirical foundation came first.

Using USITC DataWeb import data for HS chapters 61–62, I ran a two-way fixed-effects panel regression on mass-market versus luxury apparel import growth across freight, energy, FX, and tariff shocks. The headline result: a one-standard-deviation increase in the composite trade-cost index reduces mass-market apparel import growth by approximately 8 percentage points. The luxury interaction term offsets nearly the entire effect — luxury import growth is essentially insensitive to the same shock. The mechanism is not brand perception. It is a structural difference in demand elasticity that produces measurable, consistent divergence in financial outcomes across every cost channel tested, every sample period, and every way of defining the luxury segment.

That paper — Trade-Cost Volatility and Apparel Imports (McGlade, 2024) — establishes the factor. What it leaves open is the question that matters most in deal underwriting: does sourcing geography systematically determine how much of that 8-point hit a specific company takes?

The PE memo answers that question.

Extending the panel to the country level, I isolated four structural drivers of geography-level trade-cost beta: preferential trade access (EU DCFTA, Customs Union membership), freight-route volatility (Asia-Pacific ocean lanes versus overland and short-sea alternatives), vertical integration depth and the associated COGS premium, and currency peg structure. The result is a dual beta screen — separate coefficients for EU-revenue-concentrated and US-revenue-concentrated brands — that treats sourcing geography as a quantifiable underwriting input, not a qualitative risk factor.

The LBO model runs that screen through a real hold. A 30% sourcing migration from Bangladesh to Georgia, ramped over 24–36 months, reduces effective trade-cost beta by 25–35%. At a 7.0x entry on a $50M revenue target, the transition absorbs a COGS premium throughout the hold but generates a 0.5x exit multiple premium from demonstrated EBITDA stability — producing a +0.6% IRR lift versus the baseline. The COGS premium is the gating variable: the thesis holds at 8–10% on margin protection alone; above 12%, the exit multiple assumption becomes load-bearing and requires independent buyer validation before IC. The memo says so directly. Georgia political risk is modeled explicitly — 15–20% disruption probability over a five-year hold, sized at −$0.19M to −$0.25M in expected value — not footnoted away.

The investor brief compresses all of it into a first-meeting format. The slide deck reduces it further to the two questions a deal team actually needs to answer at the screen stage: which geographies carry the lowest EBITDA volatility exposure, and what is the realized value of shifting 30–50% of production to them.

The full suite is available on my Projects section on LinkedIn named “Sourcing Arbitrage in Fashion PE — A Research Suite”:

• Trade-Cost Volatility and Apparel Imports (McGlade, 2024) — empirical paper
• Sourcing Arbitrage in Fashion PE — PE investment memo (full LBO model, country screens, Georgia risk scenario, diligence path)
• Investor Brief — compressed IC-ready summary
• Slide Deck — visual leave-behind

Feedback from practitioners in fashion PE, consumer/retail PE, or anyone working on China+1 sourcing theses is welcome.

Data: USITC DataWeb (HS 61–62), Freightos Baltic Index, IMF Brent series, Federal Reserve DXY index, USITC MFN tariff rates. All documents are for discussion purposes only. LBO return figures are illustrative. Country-level beta screens are directional priors; formal panel estimation is ongoing.