A production manager at a Tier-2 parts supplier in Japan faces a practical choice every planning cycle: keep inventory close to the line and trust the next kanban replenishment signal, or add weeks of stock in case a disruption arrives before the next delivery. The first option protects inventory turns and storage capacity. The second protects takt time when a supplier misses a shipment, a model change moves faster than expected, or a component becomes difficult to replace.
After the 2021 semiconductor shortage, many automotive suppliers increased their buffers. That decision was understandable. It also created a new problem when demand and supply conditions stabilized. Stock that once looked like insurance became working capital, warehouse space, and the risk of holding parts for a configuration the plant no longer needed. The tension between just-in-time and just-in-case is therefore not a choice between discipline and carelessness. It is a question of when uncertainty justifies a buffer, how large that buffer should be, and what evidence should release it.
JIT and JIC solve different operational risks
Just-in-time inventory, developed through the Toyota Production System, is designed to bring parts to the production line when they are needed. Its benefits depend on dependable replenishment, clear kanban signals, stable processes, and close coordination between plants and suppliers. When those conditions hold, cycle stock can remain low without putting production at immediate risk.
Just-in-case inventory addresses a different exposure. Safety stock absorbs supply interruptions, demand spikes, transportation delays, and the time required to qualify an alternative source. It is not consumed according to the normal production rhythm. It exists for the abnormal case.
The distinction between cycle stock and safety stock matters because the two types of inventory should not be managed by the same logic. Cycle stock follows expected usage and replenishment frequency. Safety stock follows uncertainty. A supplier that simply adds more weeks to every SKU may improve protection for a few vulnerable parts while tying up cash in parts with short lead times and predictable demand.
Pure JIT still works well for high-frequency, short-lead-time parts in stable supply relationships. The problem comes when one policy is applied uniformly across parts with very different lead times, supplier dependencies, demand patterns, and substitution options. A high-volume fastener with two qualified sources does not carry the same risk as a specialized molded component with a long tooling lead time.
The cost of holding a buffer can hide in plain sight
Consider an illustrative Tier-2 wiring harness supplier in Nagoya. After the 2021 shortage, the supplier adds eight weeks of buffer to several harness components because the cost of a line stoppage appears much higher than the cost of inventory. The policy protects production while uncertainty is high.
Later, assembly schedules settle. The supplier continues to produce against the same buffer because no one has defined a clear release condition. The warehouse now contains parts purchased for an earlier risk profile. Some stock still supports current orders. Some is tied to a model mix that has changed. Meanwhile, purchasing continues to replenish against standing orders, so the buffer does not decline as quickly as expected.
This is where inventory turns and service protection can become confused. A buffer may prevent an expensive disruption, but it also has a carrying cost. That cost includes storage, financing, insurance, handling, obsolescence, and the planning attention required to distinguish usable stock from aging stock. For automotive suppliers, the physical part may be small while the exposure is distributed across many tiers and variants.
The answer is not to remove the eight weeks automatically. The right question is whether the conditions that justified eight weeks still exist. Has supplier lead time returned to its normal range? Are orders deviating from the established pattern? Has the plant announced a shutdown, a model change, or a production shift? Is the part constrained by a single source? Buffer policy should respond to those questions rather than remain fixed because it was once prudent.
Dynamic safety stock starts with the trigger
Dynamic safety stock is often described as a formula, but the operational challenge starts earlier: identifying the signal that should change the formula. A buffer needs an activation condition and a release condition.
An activation condition might be a sharp deviation from standing order patterns, a sustained increase in forecast demand, a longer supplier lead time, or a known event that changes the assembly schedule. A release condition might be the return of orders to their normal range, the completion of a temporary production run, or improved confidence in replenishment timing.
These conditions should be evaluated at the appropriate tier and part level. A vehicle plant may see a stable finished-vehicle plan while a Tier-2 supplier sees a sudden change in demand for one connector family. Conversely, a short-term order spike may be noise rather than a reason to build a large buffer. The planning system needs enough history and context to distinguish a meaningful deviation from normal variation.
That is why the timing of a signal matters as much as its size. A forecast increase arriving after a supplier has already placed a long-lead purchase order may not change near-term availability. A smaller but earlier signal may give the supplier time to adjust production without creating excess stock. The value lies in connecting the forecast to the decision window, not in producing a number that looks precise.
From fixed rules to signal-driven decisions
At Lean Path, we generate tier-specific demand forecasts weekly. The purpose is not to replace kanban or dictate one inventory policy for every supplier. It is to provide a view of expected demand across the supply chain and identify where current orders depart from their established pattern.
Our exception detection flags those deviations from standing order patterns. That flag is the signal a supplier can use to review whether a JIC buffer should be activated. The review can include current inventory, open orders, supplier lead time, part criticality, and the time required to recover after a disruption. A flagged exception does not automatically mean "buy more." It means the normal replenishment assumption deserves attention.
The forecast also incorporates known model changes, plant shutdowns, and seasonal signals specific to the assembly schedule. These details are important in automotive planning because a planned event can look like abnormal demand if it is not represented in the forecast. Treating every change as a surprise encourages unnecessary stock. Treating no change as meaningful leaves suppliers exposed.
The operating principle is simple: match the order to the signal, not the anxiety. In practice, that means a supplier may activate a buffer for a constrained part while keeping ordinary cycle stock unchanged for stable, replenished items. It may also reduce a buffer after the exception passes, rather than allowing a temporary response to become a permanent inventory target.
Heijunka needs flexibility around the edges
Heijunka, or production leveling, reduces unevenness by smoothing the schedule. It remains valuable because a level plan makes supplier replenishment more predictable. Yet leveling does not eliminate every source of variation. Model introductions, shutdowns, recovery production, regional demand changes, and upstream shortages can still disrupt takt time.
The practical implication is that lean systems need controlled flexibility around their stable core. Kanban can continue to manage frequent replenishment while an exception process governs unusual risk. The buffer should be visible as a temporary response with an owner, a reason, and a review date. Without those fields, safety stock tends to become part of the baseline and loses its connection to uncertainty.
This approach also helps separate rational buffer activation from anxiety-driven accumulation. A JIC buffer is rational when the expected cost of a disruption, adjusted for its likelihood and recovery time, exceeds the cost of holding the stock. It becomes anxiety-driven when the buffer grows because no one trusts the forecast, no one owns the release decision, or a past disruption remains vivid after its conditions have changed.
That boundary is especially important for smaller suppliers. They may not have excess warehouse capacity or the cash to protect every part equally. A tier-specific view helps them direct attention toward components where a disruption would affect the line and where replenishment cannot respond quickly enough.
What better buffer decisions require
There is no universal number of weeks that defines a responsible JIC policy. The appropriate level depends on demand variability, lead time, supplier concentration, part criticality, substitution options, and the cost of a takt time disruption. Those variables also change over time, so a buffer target should be reviewed as conditions change.
A useful operating review can ask:
- What risk caused this buffer to be activated?
- Is the risk visible in current order and forecast signals?
- How long would replenishment take if the risk occurred today?
- Which portion is cycle stock, and which portion is safety stock?
- What evidence would support reducing or releasing the buffer?
These questions turn the JIT versus JIC debate into a planning process. They also preserve the strengths of lean manufacturing. The objective is not maximum inventory or minimum inventory. It is inventory that reflects the current operating conditions.
For a small team like ours, the useful test is whether the forecast and exception signal help a supplier make a more specific decision: hold this part, for this reason, at this level, until this condition changes. Buffer activation becomes more accurate when it is tied to tier-specific demand, known production events, and deviations from normal ordering patterns, rather than to a general memory that the supply chain once became dangerous.