The aggregate inventory count has been the default unit of analysis in floor-plan credit for as long as the product category has existed. A lender approves a credit line of, say, PHP 15 million. The dealer's stock list shows 35 units with a combined stated value of PHP 18 million. The collateral appears to cover the exposure, the loan-to-value ratio looks reasonable, and the credit decision proceeds.
What the aggregate count does not tell the lender is which 35 units are in that portfolio, how quickly each model category turns in the local market, and what each model's realistic recovery value looks like after 60 days on lot. Two dealers can have identical aggregate inventory values while carrying completely different levels of collateral risk.
Why Model Mix Matters
Used-car inventory is not homogeneous collateral. A Toyota Vios and a Kia Seltos carry different liquidation risk profiles. The Vios has a deep secondary market in the Philippines, with strong demand across a wide buyer range and predictable pricing at auction. A higher-end crossover or a slow-moving MPV variant from a smaller brand has a narrower buyer pool and a less predictable recovery trajectory, particularly if market conditions shift mid-cycle.
A dealer whose 35-unit portfolio consists primarily of high-demand, fast-moving models is holding fundamentally different collateral than one whose portfolio is dominated by slow-moving models or units positioned at the upper end of the market price range. Both may show the same aggregate inventory count. Both may have the same stated aggregate value. The credit risk is not the same.
Model-level analysis disaggregates the portfolio into its component risk profiles. Instead of 35 units at PHP 18 million, the lender sees: 12 units in the Toyota and Honda compact segment with average days-on-lot below 25, 8 units in the crossover segment averaging 38 days, and 15 units in the higher-displacement segment averaging 58 days with a narrower recovery band. The aggregate value is unchanged; the risk distribution is visible.
The Turnover Dimension
Days-on-lot data adds the time dimension that stated value alone cannot provide. A unit that has been on lot for 10 days and a unit that has been on lot for 70 days both appear in the stock list at their respective book values. In terms of actual collateral quality, they are in entirely different positions.
Fast-turning models provide a lender with two forms of protection: the unit is likely to sell before significant depreciation occurs, and the demand signal from rapid turnover suggests the dealer has priced the unit in line with market expectations. Slow-turning models in the same portfolio signal either a pricing disconnect, a demand weakness for that model category, or a condition issue. Any of these factors affects the lender's realistic recovery position.
When we look at lot activity data across dealer portfolios in the Philippine market, the model-level variation in turnover speed is substantial. Some models consistently clear in under 20 days across a broad range of dealers. Others routinely sit for 60 days or longer before selling, and their prices typically adjust downward once they reach the 45-day mark. These patterns are not random: they reflect underlying demand dynamics that are consistent and measurable across market cycles.
The Recovery Band Dimension
Auction recovery data adds a third layer to the model-level view: what the market actually pays when units are liquidated under pressure. A lender's exposure on an aged unit is not equal to the book value on the stock list. It is a function of what can be realistically recovered if the unit needs to be sold at regional auction.
Recovery rates vary significantly by model category. High-demand models with strong secondary market depth recover a higher percentage of their market value at auction than slow-moving models where buyer competition is limited. Models in segments experiencing a demand shift downward can lose additional value at auction compared to the same model at peak demand.
A credit model that applies a uniform recovery assumption across the entire portfolio is systematically misrepresenting risk. The assumption may be conservative enough on average to cover losses across the portfolio, but the distribution of outcomes is much wider than the average implies. Lenders who underestimate the tail risk in their model mix are underpricing credit for the dealers whose inventory carries the most liquidation exposure.
Building the Model-Level View in Practice
Constructing a model-level risk view requires more granular data than what most dealer stock lists provide. Standard stock lists typically include make, model, year, and stated value. What they often do not include is the acquisition date that enables days-on-lot calculation, transaction data that establishes turnover velocity, or any reference to auction recovery rates for comparable units.
The data assembly problem is not trivial, but it is solvable. VIN registry data, where accessible, provides acquisition timestamps that enable accurate days-on-lot calculations. Dealer management system exports can supplement this with sales event records that establish actual turnover rates for each model. Regional auction data provides the recovery reference.
When we build a scoring view for a dealer's portfolio at OneLot, the model-level breakdown is the analytical output the lender actually acts on. The aggregate credit line figure is a planning number. The model-level risk distribution is what determines whether that planning number is well-supported by the collateral it is written against.
The Concentration Problem
There is a second risk dimension that model-level analysis surfaces: concentration. A dealer whose entire inventory is concentrated in one model category has made an implicit bet on the continued health of that segment. If that segment weakens due to a competing model launch, a fuel price shift, or a credit tightening that reduces buyers' price range, the entire portfolio is exposed simultaneously.
Aggregate counts obscure this concentration. A dealer with 30 units of essentially the same model variant looks on paper like a healthy inventory holder because the total count is comfortable. A model-level breakdown immediately identifies the concentration and flags the correlated risk it represents.
Lenders who track model-level distribution across their dealer book can identify portfolio-wide concentrations too: if several dealers in the same region all hold significant positions in the same slow-moving segment, the lender has a correlated risk exposure that affects multiple credit lines simultaneously.
What Changes for Lenders Who Use This View
Lenders who shift from aggregate to model-level analysis do not necessarily need to restructure how they approve credit lines. The initial sizing decision can remain aggregate-based. What changes is the monitoring posture: instead of tracking whether the aggregate count stays in range, the lender monitors whether the model mix is shifting toward slower-moving categories and whether days-on-lot trends are extending for the higher-risk segments.
This is not a more complex analysis than what most lenders already do. It is a more specific one. The data requirements are higher, but the insight payoff is substantial: the lender sees where the risk actually lives in the portfolio, not just how large the portfolio is.
A dealer with a 35-unit portfolio concentrated in fast-moving economy segments deserves different credit terms than one with an equivalent portfolio concentrated in slow-moving models. Stock lists, as currently used, do not provide the information to make that distinction. Model-level data does.
This is not an argument that the aggregate count is irrelevant. Total exposure matters for credit limit sizing and portfolio management. The argument is more limited: aggregate count alone is insufficient for assessing the quality of the collateral it represents. A lender who adds model-level turnover and recovery data to their assessment has a fundamentally more accurate picture of what they are holding.