Market Update: Diverging Holiday Consumption Trends and Sustained Momentum in the AI Value Chain

With the Mid-Autumn Festival and National Day falling close together this year, combined with extended leave arrangements, markets initially anticipated that the prolonged holiday period could mark a turning point for domestic demand in the fourth quarter. However, actual data suggest that the longer holiday period changed travel patterns without materially expanding overall demand.

According to the Ministry of Transport, total cross-regional passenger movements reached 2.142 billion from October 1–7, averaging 306 million per day. This represented a modest 0.6% YoY increase compared with the daily average during the eight-day Mid-Autumn Festival and National Day holiday in 2025, which serves as the comparison basis throughout this section. Rail passenger traffic increased 12.7% YoY, while road passenger movements, which accounted for more than 90% of the total, declined 0.3%. Non-commercial passenger vehicle movements on expressways and ordinary national and provincial highways fell 1.4%, while civil aviation passenger traffic remained broadly unchanged. Consumers have become less sensitive to travel duration and more focused on value for money. This represents a shift in consumption patterns rather than an expansion in overall demand.

On the consumption front, the prevailing trend was "willing to travel, but cautious about spending." According to the Ministry of Culture and Tourism, domestic tourist trips reached 826 million during the seven-day National Day holiday, with total expenditure of RMB738.375 billion. On a comparable daily-average basis, these figures represented YoY increases of 6.3% and 4.3%, respectively. As tourist arrivals grew faster than expenditure, average spending per tourist declined by approximately 1.9% YoY.

Among the 78 pedestrian streets and commercial districts monitored by the Ministry of Commerce, revenue increased 4.7% YoY and foot traffic rose 2.5% from October 1–6. However, both indicators slowed compared with the first three days of the holiday, when revenue and foot traffic grew 5.3% and 3.4%, respectively. Revenue growth outpacing foot traffic suggests that commercial districts have become somewhat more effective at converting tourist activity into spending.

Entertainment consumption, however, presented a contrasting picture. According to the China Film Administration, nationwide box office revenue during the National Day holiday (October 1–7) totalled RMB1.165 billion, down approximately 36.6% from RMB1.835 billion during the corresponding 2025 holiday period. Cinema admissions totalled 31.9959 million, while the average ticket price was RMB36.41, just RMB0.28 lower than a year earlier. This suggests that weaker demand, rather than ticket pricing, was the primary driver of the decline.

Duty-free consumption also pointed to weaker spending per shopper. According to Haikou Customs, spending at Hainan's offshore duty-free stores increased 8.85% YoY during the first three days of the holiday, while the number of shoppers rose 12.90%. Based on these figures, average spending per shopper declined approximately 3.6% YoY to around RMB6,497. For the full October 1–7 holiday period, duty-free spending reached RMB855 million, with 123,000 shoppers, representing YoY growth of 10.41% in shopper numbers. 

The recovery in durable goods consumption remained relatively slow. According to the Ministry of Commerce, from October 1–6, consumer goods trade-in programmes benefited 6.392 million participants and generated RMB33.48 billion in sales. Within this total, automobile trade-ins involved 68,000 vehicles and generated RMB10.92 billion in new vehicle sales, while household appliance trade-ins covered 2.848 million units and generated RMB12.34 billion in sales. Together, these two categories accounted for approximately 70% of programme-driven sales, demonstrating the continued stabilising role of government support. However, data from the China Passenger Car Association showed that nationwide passenger vehicle retail sales reached 1.618 million units in September, declining 28.0% YoY despite a 5.0% month-on-month (MoM) increase. This indicates a relatively modest recovery during what is traditionally a peak sales season.

Inbound tourism was one of the few segments showing a clear improvement. According to the Ministry of Culture and Tourism, China received 2.983 million inbound tourists during the holiday, representing 6.5% YoY growth on a comparable daily-average basis.

Indicator	Reporting Period	YoY Change	Assessment
Total cross-regional passenger movements	Oct 1–7	0.60%	Subdued overall activity
Rail passenger traffic	Oct 1–7	12.70%	Improving travel mix
Road passenger movements	Oct 1–7	-0.30%	Subdued overall activity
Domestic tourist trips	Oct 1–7	6.30%	Relatively strong growth
Total domestic tourism expenditure	Oct 1–7	4.30%	Lagging tourist growth
Revenue across 78 pedestrian streets and commercial districts	Oct 1–6	4.70%	Moderate improvement
National Day holiday box office revenue	Oct 1–7	-36.60%	Significant weakness
Passenger vehicle retail sales	Sept-26	-28.00%	Weak peak-season demand
New home transaction floor area across 30 cities	Oct 1–4	2.50%	Selective improvement
New home transaction floor area in third-tier cities	Oct 1–4	-39.40%	Continued contraction
Sources: Ministry of Transport, Ministry of Culture and Tourism, Ministry of Commerce, China Film Administration, China Passenger Car Association, Haikou Customs, iFinD, and iFAST Research. Cross-regional passenger movements, tourist trips and total tourism expenditure are compared on a daily-average basis against the eight-day Mid-Autumn Festival and National Day holiday in 2025. New home transaction floor area is based on a 30-city sample. Data as of October 12, 2026.

2. Real Estate: Selective Improvements and the Limits of Policy Support

The individual housing mortgage interest subsidy policy introduced in late September took effect on October 1. Under the policy, eligible first homes with a floor area below 120 square metres and a total purchase price below RMB1.5 million receive an annual interest subsidy equivalent to one percentage point of the outstanding mortgage principal. Holiday transaction data indicate that the effectiveness of this policy varies considerably across city tiers.

In first-tier cities, the number of new home transactions declined slightly YoY, although transaction floor area increased. Performance differed significantly between cities. Certain cities recorded substantial improvements following adjustments to their own home purchase restrictions and housing provident fund policies. In contrast, the remaining three cities experienced YoY declines of approximately 20%–40% in transaction volumes. Because the mortgage subsidy targets lower-priced, smaller homes, improvements in first-tier cities were driven primarily by the release of upgrading demand rather than the subsidy itself.

Second-tier cities were the most direct beneficiaries of this round of policy support. The following analysis uses a fixed, comparable sample of 23 cities across three years. The 2026 reporting period covers October 1–6, while the comparison periods for 2024 and 2025 cover October 1–7. Across nine second-tier cities, new home transaction volumes and floor area increased 7.1% and 6.5% YoY, respectively. However, because the current reporting period covers six days compared with seven days in the base period, the unadjusted YoY figures understate the improvement in daily activity. After adjusting for the number of days, average daily transaction volumes increased 25.0% YoY. In some cities, transaction volumes increased while the average floor area per transaction declined, consistent with stronger demand from first-time homebuyers. In the secondary market, four second-tier cities — Foshan, Qingdao, Suzhou and Dalian — recorded comparable YoY increases of 30.5% in transaction volumes and 35.2% in floor area. However, the relatively small sample means these results should be treated as indicative rather than conclusive.

Third-tier cities remained under pressure. Based on the 30-city sample for October 1–4, new home transaction floor area in third-tier cities declined 39.4% YoY. Within the comparable 23-city sample, four third-tier cities recorded a 3.3% YoY increase in transaction volumes but a 5.2% decline in floor area from October 1–6. However, Jiangmen accounted for 84% of transactions within this third-tier sample, making it both small and highly concentrated. The results are therefore insufficient to overturn the broader conclusion that third-tier housing transactions remain in contraction. In first-tier cities, secondary housing transactions returned to more normal levels following the policy-driven surge in the corresponding period last year, with average daily transactions broadly unchanged YoY.

Overall, the mortgage interest subsidy policy has been effective within its intended price and floor-area segments. However, these segments are naturally concentrated among first-time homebuyers in second-tier and lower-tier cities. As a result, the policy's contribution to overall housing demand remains limited and does not change the broader reality that real estate investment is still undergoing a deep adjustment. Furthermore, high-frequency holiday transaction data are affected by delays in online contract registrations and differences in sample coverage. These figures should not be extrapolated directly to full-month performance.

3. Macroeconomic Data: Divergence Between Domestic and External Demand, and Between Upstream and Downstream Industries

Macroeconomic data outside the holiday period showed the same pattern of divergence as observed during the holiday. On the demand side, total retail sales of consumer goods increased by 1.1% year-on-year (YoY) in September, exceeding market expectations of 0.6%. This was partly attributable to a low base in the same period last year, but growth remained at a low level. Real estate development investment declined by 20.6% YoY from January to September, indicating that the two major pillars of domestic demand remained weak. On the pricing side, the Consumer Price Index (CPI) increased by 0.9% YoY in September, below market expectations of 1.0%, while the Producer Price Index (PPI) rose by 5.0% YoY, exceeding market expectations of 4.5% and accelerating significantly from 3.8% in August, marking the fastest increase since June 2022. Consumer prices remained weak, while upstream prices accelerated. According to September data from the National Bureau of Statistics, industries recording the largest YoY increases in PPI included both energy and raw materials industries (coal mining and washing increased by 35.3%, oil and natural gas extraction by 24.5%, and non-ferrous metal smelting and rolling by 21.4%) and technology-related industries (electrical machinery and equipment manufacturing increased by 6.2%, and computer, communication and other electronic equipment manufacturing by 5.8%). In contrast, automobile manufacturing declined by 2.0%. These price increases were not driven by a broad-based recovery in end-market demand.

External demand presented a markedly different picture. September exports, measured in USD terms, surged 36.3% YoY, substantially exceeding market expectations of 25.5%. Imports increased 31.0% YoY, also above expectations of 24.5%. The trade surplus reached US$133.5 billion, its highest level since January 2025. Strong external demand alongside weak domestic consumption, and accelerating upstream prices alongside subdued consumer prices, point to the same underlying structure. China's growth momentum is increasingly being driven by external demand and upstream industrial activity rather than domestic household consumption. This pattern is consistent with the AI value chain, where revenue growth is supported by exports and capital expenditure. It provides important context for the analysis in the following section.

Indicator	Reporting Period	Actual	Market Expectations
Total retail sales of consumer goods (YoY)	Sept-26	1.10%	0.60%
Real estate development investment (YoY)	Jan–Sep 2026	-20.60%	—
CPI (YoY)	Sept-26	0.90%	1.00%
PPI (YoY)	Sept-26	5.00%	4.50%
Exports (YoY, USD terms)	Sept-26	36.30%	25.50%
Imports (YoY, USD terms)	Sept-26	31.00%	24.50%
Trade surplus	Sept-26	US$133.5bn	—
Sources: National Bureau of Statistics, General Administration of Customs, iFinD, and iFAST Research. Market expectations represent median estimates from Reuters surveys. CPI refers to the Consumer Price Index; PPI refers to the Producer Price Index. Data as of October 2026.

4. In Contrast: Strong Momentum and Pricing Power Across the AI Value Chain

During the same holiday period, developments across the AI value chain pointed in an entirely different direction. According to a forecast published by TrendForce on September 30, 2026, conventional DRAM contract prices are expected to rise 10%–15% QoQ in Q4 2026, while NAND flash memory contract prices are projected to increase 15%–20%. TrendForce also noted that the pace of DRAM contract price increases is expected to moderate compared with the preceding period. The anticipated price increases are primarily attributable to memory manufacturers continuing to reallocate advanced production capacity towards high-demand, higher-margin applications such as servers. Enterprise solid-state drives (SSDs) are the only memory product category expected to experience faster price growth than in the previous quarter.

The significance of this pricing signal lies not in the magnitude of a single quarter's increase, but in its divergence from consumer spending trends. While households reduced their average spending during the extended holiday, enterprises are expected to continue paying higher prices for AI computing infrastructure. Contract prices are negotiated between suppliers and major customers. If the forecasts materialise, they would indicate continued expansion in corporate demand for AI-related memory rather than merely reflecting short-term market sentiment. However, forecasts do not represent confirmed orders, and actual pricing will depend on subsequent contract transactions.

Capital market developments also point in the same direction. According to media reports, a leading domestic large language model company is pursuing a new financing round potentially worth up to RMB100 billion, implying a valuation of approximately RMB500 billion. This suggests that capital continues to flow into the AI sector. The funds are expected to be used primarily for model training and inference deployment, which would ultimately translate into additional demand for computing capacity and memory products.

Looking over a longer period, the same trend appears across several independent datasets. On the pricing front, producer prices for computer, communication and other electronic equipment manufacturing increased 5.8% YoY in September. On the earnings front, profits in the electronics industry doubled YoY during January–September, contributing 9.6 percentage points to the overall profit growth of industrial enterprises above designated size, according to data published by the National Bureau of Statistics on October 27, 2026.

In foreign trade, integrated circuit exports surged 99.1% YoY during January–September, including a 136% YoY increase in September alone, measured in USD terms, according to the General Administration of Customs. At the index level, the CSI All Share Information Technology Index recorded 20% YoY revenue growth and 74% YoY earnings growth in Q1 2026. We forecast approximately 70% earnings-per-share (EPS) growth for the index for the full year 2026. However, index-level earnings performance for Q3 still requires confirmation from corporate third-quarter results as they are progressively released.

Indicator	Reporting Period	Data	Basis
Conventional DRAM contract prices (QoQ)	Q4 2026	+10%–15%	TrendForce forecast (Sep 30)
NAND flash memory contract prices (QoQ)	Q4 2026	+15%–20%	TrendForce forecast (Sep 30)
Producer prices for computer, communication and other electronic equipment manufacturing (YoY)	Sept-26	5.80%	Actual
Electronics industry profits (YoY)	Jan–Sep 2026	100%	Actual
Integrated circuit exports (YoY, USD terms)	Jan–Sep 2026	99.10%	Actual
CSI All Share Information Technology Index revenue / earnings (YoY)	Q1 2026	0.27027027	Actual (Q3 results pending)
Sources: TrendForce, National Bureau of Statistics, General Administration of Customs, iFinD, and iFAST Research. The electronics industry follows the classification used in the National Bureau of Statistics' industrial enterprise profit commentary. Producer price data refer specifically to computer, communication and other electronic equipment manufacturing. Data as of October 27, 2026.

5. What These Trends Mean Collectively

When holiday consumption data, macroeconomic indicators and AI-related developments are considered together, they do not present contradictory signals. Instead, they reflect the same K-shaped economic pattern, in which emerging growth drivers strengthen while traditional sectors remain weak, across different reporting frequencies. On the domestic demand side, government policies are having an effect, but their primary role is to stabilise activity rather than generate sustained growth. Trade-in programmes have supported durable goods consumption, mortgage subsidies have stimulated first-time homebuyer demand within specific price segments, and consumption vouchers and tourism initiatives have lowered the barriers to travel. The common feature of these measures is that they have increased transaction activity without meaningfully lifting spending per person. The below-expectations CPI reading reflects this weakness in consumer pricing. On the AI side, demand is driven by global capital expenditure on computing infrastructure. Pricing power stems from structural capacity constraints and is largely independent of Chinese households' willingness to spend. Stronger-than-expected exports and accelerating PPI growth reflect this trend at the macroeconomic level.

From an asset allocation perspective, this means that selective improvements during the holiday period should not be interpreted as the beginning of a broad cyclical recovery. Order backlogs and employment indicators have yet to confirm sustained demand improvement, while the property sector remains in a deep adjustment phase. A meaningful recovery in per-capita spending would be a more convincing indicator of a turning point in domestic demand. Such evidence has not yet emerged. Verifiable earnings momentum remains concentrated in the AI value chain, where market upside is supported by realised earnings growth rather than valuation multiple expansion.

6. Risk Considerations

• Demand destruction from rising memory prices: Rapid increases in memory contract prices will raise the cost of finished products, and some device manufacturers have already increased selling prices. If higher prices significantly weaken end-market demand, memory prices could retreat from elevated levels, potentially triggering downward revisions to earnings expectations.
• Pace of global AI capital expenditure: Revenue growth across China's AI value chain is heavily influenced by exports and capital expenditure. If global investment in AI computing infrastructure slows or external trade and economic conditions become more volatile, earnings expectations could face downward revision pressure.
• Sustainability of strong export growth: September's export growth substantially exceeded market expectations. However, this may partly reflect a low comparison base and temporary front-loading of exports. If external demand weakens in subsequent months, earnings momentum across export-oriented industries could deteriorate.
• Slower-than-expected domestic demand recovery: If per-capita spending remains weak and the property market adjustment continues, the K-shaped divergence is likely to persist. Broad market performance may continue to lag the earnings performance of AI-related sectors.
• Distortions in holiday data comparisons: Differences in the timing of the Mid-Autumn Festival and extended leave arrangements have altered the distribution of travel and consumption activity. Property transaction data are also subject to delays in online contract registrations. As a result, high-frequency holiday data should not be extrapolated directly to monthly or quarterly trends.
• Sensitivity of earnings forecasts: The asset allocation conclusions and potential upside estimates in this report depend on continued earnings delivery across the electronics value chain. If actual earnings growth falls short of expectations, or if market valuation multiples for growth assets decline, the potential upside would narrow accordingly.

7. Investment Opportunities and Asset Allocation Framework

The areas of strong earnings momentum identified in this report are concentrated in China's domestic electronics and semiconductor value chain. The GF CSI All Share Information Technology ETF provides direct exposure to this investment thesis.

Focus Vehicle	Ticker	Exchange / Domicile	Benchmark Index	Supporting Investment Thesis
GF CSI All Share Information Technology ETF	159939.SZ	Shenzhen Stock Exchange	CSI All Share Information Technology Index	Rising memory contract prices and AI value chain earnings growth
Sources: Public disclosures from the fund manager and index provider; compiled by iFAST Research.

The ETF tracks the CSI All Share Information Technology Index, with semiconductors and electronic hardware accounting for a significant share of its holdings. These are precisely the segments supported by the memory contract price increases, industrial profit growth and foreign trade data discussed in this report. The evidence supporting the sector's earnings momentum was previously examined in detail in "PMI Returns to Expansionary Territory as AI Momentum Continues."

According to the Group's Q4 Investment Outlook, based on data as of September 23, 2026, the CSI All Share Information Technology Index has a target level of 17,224 points, derived from a fair price-to-earnings (P/E) multiple of 55x and forecast 2028 EPS of RMB1.72, using the Group's index valuation methodology. This implies approximately 63% potential upside from the index level on September 23, 2026, with a target horizon of FY2028. The projected upside is primarily attributable to earnings growth rather than valuation multiple expansion. The upside estimate is measured from the September 23 reference level. Following the recent market correction, the same projected earnings trajectory would imply greater potential upside from the lower prevailing index level. However, if actual earnings growth falls below expectations, the potential upside would narrow accordingly. This allocation framework reflects a medium- to long-term investment horizon. Short-term fluctuations in holiday and monthly high-frequency data do not, by themselves, justify changes to the overall portfolio allocation structure. Investors should determine their portfolio weights based on their individual risk tolerance and investment horizons.

Related Articles

• Market Update: Manufacturing PMI Back in Expansion, and the Continuation of China’s AI Cycle
• Market Update: Pricing Momentum Shifts to Technology
• China's 2Q GDP slowed, but AI and Semiconductors are driving the next growth cycle