Gas storage in South-East Europe is still widely discussed as a question of how much gas is in the ground. Markets, however, increasingly price storage based on how fast that gas can be mobilised when power systems lose flexibility. The distinction is critical. In this region, storage does not function primarily as a seasonal arbitrage tool; it functions as regional insurance against extreme power outcomes. When that insurance is thin, electricity prices explode regardless of headline inventory levels.
This reframing matters equally for traders and industrial electricity buyers because storage determines whether gas can cap power prices during the narrow windows that dominate annual risk. Seasonal outlooks by ENTSO-E implicitly reflect this by focusing on adequacy during stress periods, but market behaviour shows that storage deliverability, not inventory, sets outcomes.
The structural exposure is clear across the region. Serbia relies primarily on the Banatski Dvor storage facility, with working gas capacity of roughly 0.45–0.50 bcm. On paper, this represents a significant share of annual consumption. In practice, the decisive variable is withdrawal capacity. During winter cold spells, sustainable withdrawals cover only a fraction of peak daily demand, which can exceed 12–14 mcm/day. Once that withdrawal ceiling is reached, storage stops functioning as a buffer and becomes irrelevant to marginal pricing.
A similar pattern holds in Romania, which holds more than 3 bcm of total storage capacity across multiple sites. Despite the larger volume, withdrawal constraints still bind during prolonged cold periods, particularly when domestic production underperforms or export flows toward Hungary persist. Storage inventories remain high on paper while the system behaves as if it is short, because the gas cannot be released quickly enough.
This is why two winters with similar end-of-summer fill levels can produce radically different price outcomes. In one year, storage deliverability is sufficient to bridge short cold spells, and gas caps power prices. In another, the same inventories coincide with prolonged cold, infrastructure maintenance, or coincident regional demand, and storage hits its withdrawal limits. Once that happens, gas becomes marginal at punitive prices, and electricity markets reprice instantly.
The power market impact is severe and concentrated. During recent winter stress events, peak electricity prices in Serbia and Bulgaria exceeded €250–300/MWh, while balancing prices climbed beyond €400–500/MWh, even as storage inventories remained well above 50%. These outcomes were not driven by low inventory; they were driven by exhausted withdrawal capacity. Storage failed as insurance precisely when it was needed most.
For traders, this reframes storage valuation. Traditional seasonal spreads—buy summer, sell winter—capture only a portion of storage value. The dominant value lies in short-duration, high-intensity optionality. A storage asset capable of sustaining 0.05–0.10 bcm/day of withdrawal during cold spells can materially alter local power pricing and intraday volatility, even if its total working volume is modest. Markets reward this capability through scarcity rents during a handful of days, not through smooth seasonal arbitrage.
This explains why balancing and intraday markets often deliver outsized returns to flexible storage positions. Intraday power spreads of €50–100/MWh increasingly coincide with moments when storage withdrawals accelerate toward technical maxima. Traders positioned with deliverable flexibility capture convexity; those positioned only with volume exposure do not.
For industrial electricity buyers, the insurance function of storage explains why average-price-focused procurement fails. Electricity contracts implicitly assume that gas storage will smooth peaks. When withdrawal limits bind, that assumption collapses. Buyers then face imbalance charges, peak surcharges, or supplier pass-throughs that can overwhelm average price savings. In practical terms, winter peak hours—often less than 10% of annual consumption—can drive 25–30% of total electricity spend in tight years.
This shifts the economics of procurement. Paying €3–6/MWh more on average to secure peak caps, load-shifting rights, or transparent imbalance exposure often delivers better outcomes than chasing marginal baseload discounts. The value proposition mirrors insurance: a small premium protects against large, low-probability losses.
Carbon convergence strengthens this dynamic. As coal exits accelerate in Romania and Bulgaria, and residual lignite capacity in Serbia faces rising economic pressure, gas becomes marginal in more hours. That increases the frequency with which storage withdrawal limits matter. Even if total storage volumes expand, failure to expand withdrawal capacity and downstream pipelines will leave the system exposed. Volatility will rise even if average gas prices fall.
The interaction with LNG underscores the point. LNG volumes arriving through terminals such as Krk LNG terminal improve annual balances, but they do not increase withdrawal capacity from underground storage. LNG stabilises seasons; storage stabilises hours. Conflating the two leads to false confidence.
From a regional perspective, storage acts as a public good. When one country’s storage absorbs stress, neighbouring power markets benefit through lower prices and reduced congestion. Yet remuneration remains largely national and volume-based. The result is chronic underinvestment in deliverability, perpetuating volatility that markets then price through higher peak premia.
The unified takeaway is direct. In South-East Europe, gas storage should be analysed and valued as insurance against extreme power outcomes, not as a seasonal inventory play. Traders who focus on withdrawal-driven optionality capture the real value. Industrial buyers who understand storage limits design procurement around peak risk rather than averages.
As dispatchable depth continues to thin and gas marginality increases, the system will test storage not on how full it is, but on how fast it can respond. When withdrawal capacity is exhausted, prices will not drift upward—they will jump. Storage that cannot move quickly enough does not fail quietly; it fails expensively.