Featured faculty: Wu Xian
Written by Putro Harnowo

Against all odds, China managed to rack up record-breaking exports last year with a US$1.2 trillion surplus. It was quite unprecedented, given the US mounting tariffs for months. The tension eventually calmed following the October truce, but a bigger problem is brewing closer to home. The world factory has experienced sustained deflation since 2023.

In contrast to inflation, deflation is a broad decline in the prices of goods and services. While it is not always harmful, deflation may spiral into weak consumer spending, lower business profits, and increased unemployment. The government has committed to stimulating domestic consumption by lowering benchmark interest rates to encourage spending.

“When deflation happens, the government normally wants to boost the economy by lowering the interest rate to raise the inflation rate mildly. However, at some point, the interest rate will hit the floor and cannot be lowered further,” says Wu Xian, Assistant Professor of Finance at the Chinese University of Hong Kong (CUHK) Business School.

To shift away from heavy reliance on traditional interest rate adjustments, the central bank can utilise unconventional monetary policy tools. For instance, the People’s Bank of China, between 2013 and 2014, injected liquidity into the banking system by allowing certain bonds to serve as collateral for commercial banks to borrow. This tool is called collateral-based monetary policy or collateral framework.

“Collateral framework is not unique to China. It has also been adopted widely across the world, for example, the Federal Reserve’s term asset-backed securities loan facility and the European Central Bank’s long-term refinancing operations. However, its impacts were hard to measure,” Professor Wu adds. “There’s a lot of endogeneity in economics, so we couldn’t tell whether the change in markets is due to this policy or other factors, like bond characteristics or the unobservable economic shocks.”

A study by Professor Wu titled Collateral-based monetary policy: evidence from China, in collaboration with Fang Hanming of the University of Pennsylvania and Wang Yongqin of Fudan University, is the first ever to measure the causal positive effects of the collateral framework on the real economy.

Professor Wu and her collaborators find that the collateral framework significantly decreases the bond spread, or the yield gap between a corporate bond and a safer government bond, and provides companies with more capital to invest and pursue other business activities. This tool also has broader and long-term effects. For instance, follow-up studies show that a collateral framework targeting green bonds can stimulate more investments in green initiatives.

Lessons learned from Chinese bonds

China’s central bank provides lending programmes for commercial banks based on loan terms, including the standing lending facility or short-term liquidity operation for immediate cash, the medium-term lending facility with three to 12-month terms, and the pledged supplementary lending with three to five-year term loans to support specific sectors.

To obtain these loans, banks need to put down securities as collateral. Since 2013, the central bank has accepted treasury and local government bonds, as well as AAA-rated corporate bonds, as collateral. This framework was extended in June 2018 to include AA and AA+ rated corporate bonds. While AAA bonds are more secure, an AA rating offers a slightly higher yield for a marginally higher risk profile.

Interestingly, many Chinese bonds are traded on two venues, despite having the same fundamentals. Bonds listed in the interbank market for qualified institutional investors are also available on major exchanges, such as the Shanghai and Shenzhen exchanges, for non-bank financial institutions and retail investors.

Only bonds in the interbank market are eligible as collateral, since the interbank market is regulated by the central bank. Professor Wu and the team use these bonds as the treatment group and the same bonds in the exchange market as the control group. “This allows us to establish a causal relationship between the collateral framework and bond prices, which is not possible in other financial markets.”

The team examines daily bond trading from January to September 2018 and finds that the collateral framework increased the value of eligible bonds. Before the collateral expansion, banks could use the AA and AA+ rated corporate bonds to borrow from other financial institutions in the repo market. The repo market participants don’t lend money based on the full value of the bonds but reduce them by a few per cent.

“Companies can borrow at a lower cost as long as they can issue at least AA or AA+ rated bonds. This allows the central bank to support and ease funding to the real economy.”

Professor Wu Xian

Being included in the collateral framework trims such reduction by three per cent, boosting the collateral value and the prices of these bonds. It reduced the spreads of the newly eligible bonds by 37–53 basis points, equivalently, 10–15 per cent of the average spread in the secondary interbank market, where previously issued bonds are traded among qualified institutional investors.

This effect can further pass through to newly issued bonds in the primary market and allows bond issuers to secure a lower borrowing cost. It reduces the spread of the eligible bonds in the primary market by about 35-56 basis points. This means the bond issuers can borrow at a cheaper rate than before.

“Companies can borrow at a lower cost as long as they can issue at least AA or AA+ rated bonds. This allows the central bank to support and ease funding to the real economy,” Professor Wu adds.

Wider effect in the market

After 1 June 2018, issuing new eligible bonds is cheaper in the interbank market than in the exchange market. Firms can take advantage of this policy and choose the best market to issue bonds. Eligible bonds are more likely to be issued in the interbank market than the exchange market.

By the end of 2018, the People’s Bank of China’s lending facilities totalled more than eight trillion Chinese yuan, about 25 per cent of the monetary base.

“Collateral framework is crucial when interest rates are low,” Professor Wu adds. “This way, the central bank can conduct policy expansion by allowing commercial banks to borrow more and lend to other institutions and businesses.”

The 2018 collateral framework also made the People’s Bank of China one of the first central banks to specifically target green bonds. While her study doesn’t explore the impact on green finance, Professor Wu sees that follow-up studies have shown that green firms received more loans because of this framework.

Furthermore, as the interbank and exchange markets coexist with a shared bond structure, Professor Wu plans to understand more deeply whether this arrangement is efficient. Her next study looks into which type of assets should be traded in the decentralised interbank market, which holds 86 to 90 per cent of all Chinese bonds.

“There’s some benefit to keeping both markets, as each has its own strengths,” she adds. “A centralised market is more transparent, but you can find dealers to trade in large quantities in the decentralised exchange markets. How to design these markets is an interesting direction to explore in the future.”

The post How China fine-tunes bond values to spark economy first appeared on China Business Knowledge.

Featured faculty: Jiang Wenxi and Yao Chen
Written by Joanne Madrid

Across major global stock markets, a new habit seems to be taking shape. The trading floor stays calm for hours until the last stretch before closing at 4pm. In a sudden rush, stocks change hands at a remarkable pace, as if all the day’s energy has been saved for 10 final minutes. All signs point back to index funds.

“The rise of index funds has fundamentally altered trading dynamics in the equity markets. Stocks that are heavily owned by index funds show more buying and selling right before the market close,” says Yao Chen, an Associate Professor in the Department of Finance at the Chinese University of Hong Kong (CUHK) Business School.

An index fund is an investment that mirrors the performance of a specific collection of securities or an index. For instance, if the S&P 500, a stock index of the 500 largest US companies, rises by one per cent, the S&P 500 index fund aims to rise by the same amount. To achieve this, the fund actually buys and sells the stocks that make up the index in the same proportions.

By simply mirroring an index rather than outperforming it, index funds generally deliver strong performance, making them popular among investors seeking low-cost portfolios. In the US, index funds accounted for 61 per cent of equity investments last year, mostly managed by two asset giants, BlackRock and Vanguard.

Algorithms continuously monitor an index fund’s portfolio relative to the corresponding index. Meanwhile, fund managers oversee the fund and execute trades once per day, right after the market closes. This timing, as it turns out, has reshaped the trading pattern.

In a study titled How index funds reshape intraday market dynamics, Professor Yao and Professor of Finance Jiang Wenxi, along with their PhD student Wu Siyuan, examine how index funds drive the stock rush near the market close. The study also finds that short sellers are increasingly timing their trades during the rush to camouflage themselves.

Save the best for last?

The stock indices fluctuate throughout the day. However, index funds are valued based on how closely they track their corresponding indices, which are calculated using closing prices of the underlying stocks. Trading closer to the closing price reduces the tracking error or gap between the fund’s return and the index’s return.

“The closer the index fund managers execute their trades to the end of the trading day, the lower their tracking error will be,” says Professor Yao.

Having analysed US equity data from 1993 to 2019, the researchers find that in the last five minutes, not only are more stocks available for trading, but the gaps between the highest bid price and the lowest ask price are also smaller. These indicate a highly liquid market, where trading stocks becomes easy, fast, and cost-effective.

“The closer the index fund managers execute their trades to the end of the trading day, the lower their tracking error will be.”

Professor Yao Chen

More liquidity but less information, just noise

More liquidity does not necessarily make stock prices more informative. During the day, stock prices move in response to economic news, earnings reports, global affairs, or other industry developments.

As the bourse winds down, index funds trade to rebalance their portfolios to match their corresponding index, driven by investor inflows or outflows rather than by real performance. Stocks with higher index fund ownership tend to move with overall market trends, creating noise that can overwhelm information about the actual picture of underlying companies.

“Therefore, if the closing price becomes the main target for most trading, causing a high portion of buying and selling to happen right at the end of the trading day, liquidity may dry up during other trading hours of the day,” Professor Yao adds.

Short sellers go with the flow

The liquidity rush benefits short sellers and informed traders. These traders typically have more genuine information about the stocks and the underlying companies, and try to maximise returns by keeping their transactions discreet to avoid tipping others off. Trading during high liquidity allows them to blend in with the crowd.

“Short sellers bet against the stock and strategically wait until the end of the trading day to trade to coincide with the price reveal from the indices. With this, they can hide their intentions and avoid drastically moving the stock price, allowing them to make more profits,” says Professor Yao.

Despite their smart disguise, collective actions by many short sellers still send a signal. The study finds that a notable increase in short sales right before the market closes is a reliable indicator that the stock price will likely fall further in the subsequent trading days. Eventually, the true information about the company’s performance becomes apparent to the market.

Can regulators entangle the rush hour?

To address the last-minute rush, stock exchanges worldwide have offered closing auctions that allow traders to determine a single closing price for each security. Stock exchanges normally set a cutoff time for placing auction orders, e.g., 3.45pm for the NYSE and 3.50pm for NASDAQ, and the closing prices are announced at 4pm.

Not all trading activity occurs through closing auctions. While the study finds that the volume of securities traded through the auction has grown steadily since 2004, it remains insufficient to slow down the last-minute trading rush. Index fund managers participate largely because closing auctions enable them to trade in large quantities at a precise price, but they rely more on regular trading.

The reasons are not only the high auction fee but also the uncertainty in trading execution. If investment flows change after the cutoff time but before the market actually closes, index fund managers can no longer change the orders and bear the risks of tracking errors.

Finally, although the study focuses on US equities, Professor Yao notes that its implications extend beyond borders, as index funds in other major markets operate in similar ways. “For example, Hong Kong indices are also calculated using closing prices of their underlying stocks,” she adds.

The post What drives last-minute rushes in the stock market? first appeared on China Business Knowledge.

Featured faculty: Tony Ke
Written by Sally Ho

Amid the dominance of streaming platforms, digital piracy, or the unauthorised and illegal downloading or streaming of movies, music, and other copyrighted content, has been a persistent headache for the entire creative industry. A data company, MUSO, reported that visits to digital piracy websites have surged from 130 billion in 2020 to 216 billion in 2024.

Piracy websites offer free illegal content to watch and download, while monetising the visitor traffic through sponsored ads to generate revenue. Most jurisdictions treat such commercial piracy as a criminal offence and punishable by fines, imprisonment, or both. However, shutting down piracy websites has been an uphill battle for law enforcement.

Authorities have been blocking many piracy websites to protect copyrighted content, but piracy firms frequently change their web domain since they can host their servers anywhere to evade detection. Therefore, regulators are now trying to curb piracy from the demand side by introducing penalties aimed at users.

“To understand anti-piracy policy, you have to model a strategic ‘cat-and-mouse’ game, where penalties change piracy firms’ technological advancements and consumer behaviour.”

Professor Tony Ke

A notable example is the US Prioritising Resources and Organisation for Intellectual Property (PRO-IP) Act, which imposes US$1,000 to US$200,000 civil penalties, and France’s Regulatory Authority for Audiovisual and Digital Communication (ARCOM) Law, with a €1,500 or around US$1,750 fine for downloading, copying and streaming illegal content.

“These demand-side regulations are unique to digital content, as digital consumption can be traced by regulators more easily than physical transactions, making it feasible for law enforcement to identify users who download or stream pirated content,” says Tony Ke, Professor of Marketing at the Chinese University of Hong Kong (CUHK) Business School.

However, piracy firms constantly assist their users in hiding their digital footprint through encryption, automated random IP addresses, virtual private network providers, and many more. Most users are aware of this support, as the piracy firms make anti-tracking methods apparent on their websites, providing a sense of security from law enforcement.

This dynamic can explain why, despite high penalties, users continue to flock to piracy websites in large numbers, as measured by Google Trends below. An important question then emerges: Do penalties for digital piracy consumption actually work?

In his new study, Regulating digital piracy consumption, Professor Ke, in collaboration with his PhD student Chen Jieteng and Gao Yuetao of Xiamen University, suggests that penalties may, in fact, increase piracy rather than curb it. In some cases, heavier penalties can even encourage more piracy.

A cat-and-mouse game in anti-piracy policy

“To understand anti-piracy policy, you have to model a strategic ‘cat-and-mouse’ game, where penalties change piracy firms’ technological advancements and consumer behaviour,” Professor Ke says. “In this model, consumers understand the penalties, but pirate platforms also respond by investing in anti-tracking tools that make enforcement less effective.”

Specifically, the researchers design a model consisting of copyright holders that offer high-quality content and set a price, piracy firms that host a free and low-quality pirated version, the regulator that sets a penalty on consumers who get caught accessing pirated content, and consumers whose preferences depend on product quality, price, and the expected penalty.

“Demand for piracy often arises when copyrighted content is unavailable elsewhere,” Professor Ke adds. “Consumers with medium incomes are prone to piracy and willing to risk penalties due to budget constraints, while still seeking access to diverse content. The piracy firm’s anti-tracking technologies significantly influence their behaviour.”

Through various analyses, the study suggests that regulators should carefully balance the level of penalties with the cost and availability of anti-tracking technology. “When investing in anti-tracking technology is relatively cheap and easy for piracy firms, the regulator should avoid harsh penalties since doing so can trigger even stronger anti-tracking investment,” says Professor Ke.

The model shows that when the anti-tracking technologies are affordable, piracy initially declines as penalties increase. However, sufficiently high penalties can induce piracy firms to invest in anti-tracking technologies that help users evade detection, ultimately leading to higher levels of piracy consumption. In this context, regulators should target a moderate penalty that is just good enough to deter piracy, and tolerating a small amount of piracy can actually lead to better overall outcomes.

“However, in a scenario where investing in anti-tracking technologies is costly for piracy firms, higher penalties can be effective and may even deter demand and supply altogether,” says Professor Ke. In this case, the risk of being caught is higher and steep penalties can reduce piracy significantly.

Implications for content platforms and regulators

For content platforms and the creative industry, counterproductive regulation can be discouraging, as they might face uncertain returns on investment that affect their decisions in producing new content. Therefore, policymakers need to design not just the level of penalties, but also the strategic responses of piracy firms carefully.

“A more effective approach is to balance demand and supply-side actions,” Professor Ke adds. “On the demand side, piracy can be reduced by making legitimate options more attractive, affordable, convenient, and widely accessible, rather than relying on penalising users. On the supply side, enforcement should focus on deterring piracy firms and the infrastructure that enables large-scale piracy.”

Technology plays a pivotal role in enabling piracy firms and consumers alike. In a digital world where anti-tracking tools are increasingly sophisticated, effective regulation may require moderation, creativity, and a deep understanding of the ecosystem, rather than pushing for heavy fines.

“Additionally, digital tracking is constrained by data privacy regulations, and tech companies offer opt-out choices for such tracking features, so policymakers also need to take into account these limitations when designing effective enforcement,” Professor Ke adds.

He also notes that some platforms may have realised the complexity of the cat-and-mouse game in digital piracy and now offer “freemium” structures that combine free access with advertising. This new strategy opens an avenue for further research to understand how these models affect consumers’ willingness to pirate.

The post How to keep digital pirates at bay? first appeared on China Business Knowledge.

Featured faculty: Lisa Wan
Written by Pan Jingyi

It was supposed to be a fun summer trip to Puerto Rico last year, as a Spanish couple had done everything ChatGPT planned, until they were refused to board the plane for not obtaining proper paperwork. In another case, two tourists were lost in a rural Peruvian town trying to find an imaginary destination suggested by AI.

AI has been hailed as the new technological evolution, but these stories remind us not to take technology at face value. On the other hand, these cases also highlight how trip planning has moved from a search bar of internet browsers to ChatGPT, DeepSeek, Grok, and the like. Scrolling through a sea of blue links is gradually replaced with a single prompt.

“Opting for an unfamiliar and novel search method like GenAI can be seen as a risky choice for making concrete plans.”

Professor Lisa Wan

“When ChatGPT was first introduced, we immediately sensed its strong potential for tourism information search, which largely depends on context and user preferences,” says Lisa Wan, Associate Professor of the School of Hotel and Tourism Management and the Department of Marketing at the Chinese University of Hong Kong (CUHK) Business School.

“Unlike traditional search engines that primarily provide fragmented information through hyperlinks, generative AI, or GenAI, can synthesise information, generate narratives, and adapt responses to users’ preferences.”

With much positive and negative news surrounding GenAI, Professor Wan seeks to understand what travellers actually perceive of the new technology. Working with Li Yuan of Zhejiang University, along with Luo Xiaoyan and Ding Xu of Sun Yat-Sen University, she conducted the research Advancing information search through GenAI: the roles of search type, travel motive and GenAI customisation level.

Across a series of studies involving more than 800 participants from different countries, the team examined when people lean towards GenAI or retreat to traditional search engines. They find that travellers’ willingness to use GenAI depends on their search purpose, travel motives, and whether the AI agent is tailored for trip planning.

When GenAI is less trustworthy

Trip planning often starts from curiosity and then turns into concrete actions. Individuals who come across a destination on social media or over casual conversation may want to find more about the must-sees, the overall vibe, and, as their interest deepens, may seek further information on specific prices and booking options.

Based on the above process, the researchers grouped these behaviours into two search types: non-decision-based, where individuals browse for general information about a destination, and decision-based, when more detailed information is sought for final decision-making.

“These differences can influence which search tools people choose,” Professor Wan says. “In the decision-based search, a small bad decision based on inaccurate information can turn into bigger problems, causing consumers to be more cautious.”

When participants are in decision-making mode, they prefer to gather information using traditional search engines. “Opting for an unfamiliar and novel search method like GenAI can be seen as a risky choice for making concrete plans,” she adds.

Professor Wan notes that new technologies often face a natural trust gap, especially when mistakes have significant consequences. Moreover, scepticism towards GenAI also reflects a rational assessment of its limitations in providing real-time and verified data, as reported in recent news.

In non-decision-based situations, however, the pattern shifts. When people are causally exploring or thinking about a destination, GenAI’s conversational style and ability to synthesise broad information become more appealing.

Traveller’s mindset and customisation make a difference

Looking further into what factors might encourage people to use AI in decision-making scenarios, Professor Wan and her collaborators found that the travel motive is the crucial piece. Specifically, participants motivated by a utilitarian goal that focuses on efficiency and convenience reported a higher preference for GenAI, whereas those with a hedonic motive of prioritising fun and pleasure are more likely to stick with traditional search engines like Google.

For utilitarian travellers, GenAI is preferred for its ability to filter information and organise search results, reducing the effort to compare options manually. Meanwhile, hedonic travellers enjoy the traditional browsing experience, mostly because search engines feature a richer mix of photos, videos, maps, reviews, and unexpected discoveries.

“Those prioritising fun and pleasure may find the variety and richness of multimedia content more appealing, providing a more immersive and enjoyable searching experience compared to the textual responses generated by GenAI,” says Professor Wan.

Customisation levels also affect user preference for AI. As booking platforms increasingly embed AI plugins for specific tasks, such as suggesting available hotels based on user preferences and providing customer service via AI chatbots, the study finds that such customisations can boost trust in GenAI.

How the tourism industry should adopt and develop GenAI

Given that GenAI is often more preferred in the non-decision stage, Professor Wan suggests platforms make an AI assistant visible in the main search bar to inform travellers general information about destinations, such as major attractions and cultural highlights. Another application is to display GenAI responses alongside traditional search results, allowing travellers to cross-check information easily.

While AI transformation continues to gain momentum, Professor Wan observes that fundamental challenges remain. “Many firms invest heavily in AI solutions but see limited results in daily operations. Two common obstacles are the lack of in-house talent to integrate AI into workflows and the tendency to adopt generic tools that don’t match real user demand.”

Furthermore, she observes that rapid AI advancements and shifting customer demand require firms to continually adapt this technology. “Rather than treating GenAI adoption as a one-off technological upgrade, firms need to view it as an organisation transformation process that involves gradual development, cross-functional collaboration and iterative experimentation.”

The future of travel planning

Professor Wan believes that AI will not completely replace search engines just yet, at least in the near future. Instead, travel information would be more distributed, with different tools serving different purposes. “GenAI is more likely to complement travel planning rather than substitute the traditional way,” she adds.

RELATED ARTICLE

Will lack of internet skills prevent seniors from travelling?

Interestingly, she suggests that social media will be the close contender for search engines. In Chinese Mainland, for example, RedNote has already become a starting point for many travellers for its first-hand reviews. “The real shift is towards interactive, experience-rich and peer-validated information, something that social media and GenAI offer in different ways.”

Another takeaway is a concern about how GenAI can subtly change how travellers engage with places and experiences. Therefore, she encourages travellers to keep interacting with locals and communities. “The goal is not to reject intelligent tools, but to remain attentive to how they reshape human capabilities and experience.”

The post Can AI beat search engines for trip planning? first appeared on China Business Knowledge.

Featured faculty: Jessica Kwong
Written by Joanne Madrid

Lots of praise has been given to persistence, highlighting the eventual rewards of never giving up. Walking away is often harder than doubling down when eyes are on a target, especially considering the time and effort spent. After all, failure is a success in progress, as the famous saying goes.

Any resources are considered gone when they’re spent in personal and business endeavours, no matter the outcome. Economists call this unrecoverable spending sunk costs. A rational mind should not account for sunk costs when facing setbacks and know when to stop, but many real-world cases have shown otherwise.

For example, Kodak invested heavily in marketing its film business for years and refused to embrace the digital camera it once invented before filing for bankruptcy. The British and French governments kept operating Concorde jet despite the skyrocketing costs, and it took decades for them to finally decommission it.

Most people have, at some point, stuck with a bad decision longer than they should have, either holding a losing investment or continuing with a plan even as results disappoint. Behavioural scientists name such stubbornness as escalation of commitment and attribute it to irrational decision-making tendencies or escalation bias.

However, Jessica Kwong, Professor of Marketing at the Chinese University of Hong Kong (CUHK) Business School, has another perspective. “Rather than reflecting stubbornness on poor judgment or irrational decision making, the escalation of commitment can arise even as people become better at planning and coordinating choices.”

In collaboration with Wong Kin-fai of the Hong Kong University of Science and Technology, Professor Kong’s latest study, The development of escalation bias across the life span: A multi-level adaptive learning approach, finds that escalation of commitment is formed as a strategy to execute a series of plans with some room to tolerate failures.

“Rather than reflecting stubbornness on poor judgment or irrational decision making, the escalation of commitment can arise even as people become better at planning and coordinating choices.”

Professor Jessica Kwong

Is escalation of commitment irrational or tactical?

People normally continue a behaviour after positive reinforcement, but quit after negative feedback. However, the researchers highlight that people also escalate their efforts when encountering obstacles instead of giving up quickly, especially when the expected results are rewarding.

Professor Kwong finds that, from a natural learning experience, people build decision strategies or game plans to achieve their goals by conducting a series of actions with different methods, and then evaluating whether the strategies succeed or fail after a certain number of attempts. The strategies that deliver more positive outcomes are more likely to be deployed again.

Therefore, what appears to be irrational persistence may instead reflect a deliberate choice to tolerate short-term failures in pursuing longer-term goals, or, as Professor Kwong calls it, an escalation strategy.

“Once an individual adopts an escalation strategy, they allow themselves the space or possibility to absorb some short-term setbacks,” she adds. “It’s akin to long-term investing. After carefully choosing a stock, reacting to every short-term price swing can undermine the original rationale for the investment.”

Professor Kwong further discovers that such tolerance grows as our metacognitive capacity, or the ability to plan, monitor and coordinate multiple decisions, evolves across the lifespan.

How tolerance to failures changes as people age

Existing literature shows that escalation bias tends to increase as children grow into teenagers, then decreases from young adulthood into old age. To understand more about it, Professor Kwong and her collaborator use three computer simulations to examine escalations in different settings.

A virtual experiment is chosen because testing the metacognitive ability of multiple individuals over long periods not only requires extensive time but could also be unethical. Each simulation consisted of 15,000 virtual participants or programmes designed to act and learn over time like humans, with each participant going through 1,000 decision-making moments.

The results confirm that increasing metacognitive capacity from infancy to adolescence leads to greater tolerance for short-term failures, but this tolerance weakens as people age with shrinking metacognitive capacity. This pattern creates an inverted U-shaped tendency of escalation commitment, which starts low in youth, peaks during employment years, and then declines as people retire.

Increased metacognitive capacity allows people to engage in more complex decision planning and fuel escalation bias. This study offers new insights into how natural learning experiences can progress into persistence to invest in failing projects.

Drawing the fine line between grit and stubbornness

This finding may explain why simply telling an employee in their productive age to stop pursuing a failing project may not be enough to change their minds. Their desired final goals are laid out in their master plan involving an escalation strategy, rather than a lack of awareness.

At the top strategic level, on the contrary, escalation of commitment bias can be fatal or disastrous when it is not backed by a higher-level decision strategy. Acknowledging it and pivoting would save the company’s life, just as Mark Zuckerberg did when rebranding Facebook to Meta in 2021 to develop metaverse-based social media, only to move on to AI a couple of years later, after realising his idea was a flop.

Companies can proactively set up systems that encourage critical re-evaluation of projects, especially when signs of counterproductive escalation appear. This might involve having independent teams review struggling initiatives or establishing clear stop points based on objective, helping employees overcome their natural tendency to keep pushing a losing hand.

Google’s Moonshot Factory is particularly adept with this. The factory is created to invent new technologies and cultivates a culture that celebrates killing promising ideas early, ensuring that the hardest parts of a project are tested first, rather than delaying potential failure. For companies that rely on research and development, perhaps a system to address escalation bias is no longer an option.

The post Why do we like to cling to failing plans? first appeared on China Business Knowledge.

Featured faculty: Cao Xinyu and Wang Jingbo
Written by Putro Harnowo

Many may call it cliché or cheesy when watching a short movie about a bullied employee who turns out to be the CEO’s son or a poor husband revealing himself as a billionaire, but believe it or not, these microdramas have stolen the show. Their plot twists and cliff hangers have helped them find a way to Hollywood, as Fox Entertainment and Disney recently announced investments in producing bite-sized movies.

Microdramas have less than two-minute duration and a vertical format because they are meant to be watched on a smartphone. Their popularity can be traced back to 2018, when Chinese short-video platforms began featuring them. Fast forward to 2024, the country’s microdrama industry surpassed its box-office revenue with US$6.9 billion, according to the China Netcasting Services Association.

Nowadays, China’s microdrama industry has become highly competitive, powered by more than 100,000 enterprises producing around 3,000 series monthly. Video platforms recognise these miniseries as profitable content and not only host but also actively incentivise them through financial rewards, algorithm optimisation, and the like, to encourage more original and high-quality content.

However, the ever-increasing number of creators and limited financial resources have made selecting creators for such incentives more challenging. After analysing data from a major video platform, Wang Jingbo, Assistant Professor in the Department of Marketing, and Cao Xinyu, Vice-Chancellor Associate Professor of Marketing at the Chinese University of Hong Kong (CUHK) Business School, find that creators with a large number of followers tend to be selected for financial support.

This method may be counterproductive since the supports might not provide a significant boost to their already high performance. “The platform may simply select creators who already have the highest content quantity and quality, rather than those who were likely to experience the greatest improvement due to the incentives,” Professor Cao says.

Therefore, in a study titled A Deep-DiD method to estimate heterogeneous treatment effects: Application to content creator selection, Professors Cao and Wang, as well as Cheng Yan of Shanghai University of Finance and Economics, Shen Zuo-Jun (Max) of the University of Hong Kong, and independent researcher Zhang Yuhui, propose a new method to optimise a digital platform’s selection process.

Practical experiment and application in the digital world

The short video platform launched a signing programme in three countries in 2022. Selected creators were asked to sign a contract and receive monthly payments based on their performance.

Professor Cao and her team then try to measure how effective the signing programme was, using three key indicators: the number of video uploads per day, the time users spent on each video, and user engagement from likes, comments, shares, and follows. To create a fair comparison, they match 2,343 creators who signed the programme with the same number of other creators who did not sign up based on their performance trajectories.

Through a statistical method called difference-in-differences (DiD), which has been widely used in economics and social sciences to estimate the effect of specific interventions, the researchers find that the signed group shows a significant boost in all key indicators. After signing up, their average number of videos per day temporarily increases, while the positive impacts on user time and engagement last longer.

However, the researchers also find that the effects vary widely across different creators. “The DiD method calculates the overall average impact by blending all the individual effects,” says Professor Cao. “If an intervention affects different individuals in several ways, or if the impact changes over time, the analyses could lead to an inaccurate picture of the actual impact.”

“The platform may simply select creators who already have the highest content quantity and quality, rather than those who were likely to experience the greatest improvement due to the incentives.”

Professor Cao Xinyu

Therefore, Professor Cao and the team further examine the specific impact on each creator with an advanced method by leveraging a computer programme called a deep neural network. Dubbed as the Deep-DiD method, it possesses extra layers to process information and find hidden patterns that traditional techniques overlook.

How does Deep DiD work, and how good is it?

First, the researchers develop a Deep DiD model by integrating deep neural networks into a difference-in-differences framework to flexibly estimate individual-level heterogeneous treatment effects as nonparametric functions of high-dimensional pre-treatment features.

By inspecting rich data from the platform, the model can predict which creators would have improved the most if they joined the programme. “With this advanced predictive analysis, we can examine how a specific programme helped some individuals more than others,” Professor Cao adds.

Creators selected by the Deep DiD model as favourable candidates show 57 to 123 per cent higher performance than those selected by the platform. Overall, creators chosen by the model consistently show a 72 to 114 per cent higher performance.

In out-of-sample evaluations, creators selected by the Deep-DiD model exhibit substantially larger performance gains than those selected by the platform. Among signed creators, those also identified by the model experience 70 to 80 per cent higher realised performance jumps relative to the average signed creator, across both user time contributed and user engagement.

When comparing selection rules directly, creators ranked highest by the model have 57 to 123 per cent higher estimated treatment effects than platform-selected creators, indicating significant scope for improvement in targeting. Notably, nearly half of the creators identified by the model were not signed by the platform, reflecting systematic differences in selection criteria.

What makes the Deep DiD method remarkable is its flexibility. While the three key indicators above reflect outcomes that platforms are likely to consider, platforms can tailor any other metrics depending on their goals. This way, they can maximise the impact of their programmes by investing only in those who will gain the most.

“If a platform’s revenue depends on users’ watching time, then this metric can be prioritised. The goal can also be defined as a weighted combination of metrics or other customised outcomes,” Professor Cao says. “The process can be repeated multiple times to reduce randomness and improve the stability of the results. Whenever the platform plans to implement a new intervention, new rounds of estimation should be conducted using the corresponding data and inputs.”

Beyond the digital platform, the Deep DiD method may also be applied in other settings. For instance, a retail shop launching a rewards programme must choose its target wisely to ensure only the right customers contribute to revenue, a company deploying bonus systems needs to determine which employees would achieve the largest productivity boost, and a government introducing a subsidy scheme should predict which individuals would benefit most.

In the real world, figuring out how much an intervention truly impacts a beneficiary is never easy since many variables interact in complex and hidden ways. These intricate and unknown linkages are nearly impossible to figure out using traditional mathematical methods, but the deep neural network acts much more like a sophisticated brain and is exceptionally good at discovering complex and subtle patterns.

The post How should video platforms spot their next big stars first appeared on China Business Knowledge.

Featured faculty: Ko Chiu-yu
Written by Putro Harnowo

The fuss and feathers around bitcoin and cryptocurrency sometimes obscure the basic technology behind them. Indeed, blockchain, as the fundamental technology, has the potential to revolutionise industries beyond finance.

Blockchain is a distributed, encrypted digital ledger or record that everyone can see and agree on through computers. Its unique feature has opened a new mechanism called a smart contract, a self-executing agreement implemented as code on a blockchain that automatically performs actions without a middleman. Once deployed, this contract has a unique and immutable address that prevents unauthorised changes.

Leveraging this technology, a new study by Ko Chiu-yu, Associate Professor at the Department of Decisions, Operations and Technology at the Chinese University of Hong Kong (CUHK) Business School, examines how smart contracts can be utilised to solve disputes in the supply chain.

“Supply chains involve multi-tiered arrangements with numerous bilateral contracts. When disruption arises, figuring out who should bear the loss can be complicated,” says Professor Ko. “Smart contracts can automatically manage, track, and settle these losses fairly, especially when an action of a party initiates sequences of unanticipated damages that affect others.”

For instance, when a supplier fails to ship a component on schedule to an electronic assembler, the assembler would face compensation demands from the distributors for delayed delivery. A question then emerges: When actions of a party affect another party’s agreements with a third party, how should the liability be shared between the initiator and other parties for the damage?

In a study titled Sharing sequentially triggered losses: Automated conflict resolution through smart contracts, Professor Ko and his collaborators propose “the fixed-fraction rules” to fairly share responsibilities among parties in a supply chain. Each liability is split between the party that initiated the problem and the other parties that caused further damages later in the sequence.

Professor Ko compares it to a choose-your-own-adventure game with a dial. One end shows 0, which means the first party is responsible for all the costs, and the other end shows 1, which means everyone only pays for their own mistakes. The easy middle ground is to split the costs fairly between the initiator and the next party affected, using simple fairness principles like “let’s all chip in equally so nobody gets the short end” to keep the peace.

“Smart contracts can automatically manage, track, and settle losses fairly, especially when an action of a party initiates sequences of unanticipated damages that affect others.”

Professor Ko Chiu-yu

How can blockchain help?

Along with Jens Gudmundsson and Jens Leth Hougaard of the University of Copenhagen, Professor Ko takes a step-by-step approach in creating the fixed-fraction rules. They began by defining core allocation principles to balance fairness and incentives, making liabilities shared among each party based on a fixed fraction of the total loss.

A party is only aware of the agreements it participates in, so each party is only responsible for the loss associated with its own purview. Therefore, the initiator shall cover the rest of the loss since they could have done something to avoid the damage. For the loss incurred by several connected parties, the liabilities are shared equally among the initiator and the other parties. With these rules, the initiator is incentivised to avoid starting the chain of loss while also acknowledged for their limited control over further damage.

“The fixed-fraction rules operate on a similar principle to a common term in the supply chain called fixed share rate contracts, where the costs, risks, or liabilities arising from issues like delays, defects, or product recalls are shared among the involved parties based on predetermined fixed fractions,” says Professor Ko. “These terms have been shown to incentivise improved product quality.”

In ensuring fairness, the researchers further set out four principles. First, if losses occur in two different cases, the system can simply add up each party’s responsibility from both cases to get the total loss. Second, if the system combines two separate losses into one, the way of sharing the liabilities should stay fair and consistent. Third, if the number of parties involved changes, the responsibility shares also adjust accordingly. Lastly, parties not involved in causing any loss shouldn’t be responsible for anything.

To illustrate, the researchers extend the model to a scenario represented by a loss tree below, where any party can initiate the chain of loss. When B fails to meet its agreement with E, a total loss of US$19 occurs, which is calculated from a direct link with E and indirect links with F and G. According to the fixed-fraction rules using a benchmark middle-ground split, B pays US$14 to cover the full loss with E plus half of the losses with F and G (each taking an equal half as the balanced default for shared accountability). Accordingly, E cover the remaining US$5.

This approach ensures the initiator bears primary responsibility without overburdening downstream parties, maintaining incentives for prevention across the chain. Meanwhile, A, C, and D aren’t affected.

The fixed-fraction rules can be implemented in smart contracts by storing the agreements in a loss tree. Involved parties shall make deposits to cover potential losses, and with a set of functions, the blockchain can automatically compute and distribute liabilities in case of damage. If everything goes well, the deposits will be returned to conclude the deal.

The future of smart contracts

Supply chain is not the only one with an interwoven network. Given its ability to address the chain of loss, Professor Ko believes that fixed-fraction rules can be applied in other industries, such as automating cost-sharing in loan defaults, handling group claims after disasters in insurance, sharing penalties for construction delays in real estate, and many more.

However, there is a limit on how the rules can work, since their success depends on reliable data sources and integration with their supporting ecosystem. In many countries, the legal framework for smart contracts is also still evolving, and there is no uniformity in regulation about blockchain worldwide.

“Complex and high-value disputes will likely be settled through courts, as it takes time and effort to verify evidence and examine the terms stated in the contracts,” says Professor Ko. “Therefore, automatic conflict resolution works best for clear-cut, data-driven disputes with less room to argue, and facts can be verified automatically.”

With that being said, smart contracts will work best to settle smaller disputes, such as in delivery delays verified by sensors or tracking systems, e-commerce issues confirmed by delivery logs, and licensing or intellectual property breaches tracked through secure digital records.

In the era of AI, Professor Ko anticipates technologies to enhance smart contracts to be more adaptive by using predictive analytics to detect patterns and prevent disputes. Natural language processing would also be able to interpret ambiguous terms, and automated verification may validate data and trigger more accurate settlements.

“By combining AI-driven analysis with smart contracts for complex cases, and monitoring regulatory compliance in real time, technologies can improve efficiency, especially in sectors such as public services, e-commerce, and insurance,” he adds.

The post Can blockchain solve supply chain disputes? first appeared on China Business Knowledge.

Featured faculty: Zhou Xiaolu
Written by Putro Harnowo

Corporate social responsibility, or CSR, is an integral part of modern business strategy. Traditional CSR practices have revolved around charity, philanthropy, and volunteering, but the growing demand for sustainable business practices has transformed CSR activities into a broader scope well beyond simply giving back to society.

While regulators worldwide have now imposed mandatory CSR and sustainability reporting, a genuine desire to contribute to society may motivate corporate good deeds. However, the late American economist Milton Friedman famously argued that “there’s no such thing as a free lunch.” For businesses, CSR can improve brand perception and build public trust, but it comes at a cost.

Zhou Xiaolu, Assistant Professor of Accounting at the Chinese University of Hong Kong (CUHK) Business School, notes that CSR activities can be seen from two viewpoints: either firms genuinely care about the social and environmental impacts over profits, or they seek to enhance brand value and ultimately boost profits. An extensive body of literature shows the latter is more prevalent.

“CSR investment mostly depends on stakeholders’ demand and what the firm can offer cost-effectively,” says Professor Zhou. “Firms are constrained by limited resources to keep their reputation and meet regulatory requirements, so managers need to balance the various aspects of CSR.”

As companies face growing pressure to act responsibly on several fronts, navigating all dimensions of social responsibility becomes more challenging. Her latest study, Substitution between corporate social responsibility activities: Evidence from hiring and mistreating unauthorised workers and pollution, sheds light on the practicality of CSR initiatives and reveals that increasing one facet of social responsibility may lead to a reduction of another domain.

“When a regulation mandates one aspect of CSR to improve, managers may prioritise compliance and reduce investment in other CSR activities. Those with the highest marginal cost or the lowest return are most likely to be curtailed, resulting in a substitution among CSR initiatives,” she adds.

Evidence from the fields

In the study, Professor Zhou collaborates with Huang Ying and Li Ningzhong of the University of Texas at Dallas, and looks into a US government policy called E-Verify that allows companies to verify with the federal data whether their employees are eligible to take the job.

Nine states require universal E-Verify, compelling all employers to verify the work authorisation of employees. Meanwhile, 13 states enforce public E-Verify, which only orders public employers and contractors to verify all of their employee data. The researchers find that this policy profoundly influences CSR activities.

The Pew Research Centre’s report shows that the US had around 8.3 million undocumented workers in 2022, or almost five per cent of the workforce, who often took low-paying jobs in manufacturing, metal mining, electric power generation, and waste treatment. Fear of deportation left them vulnerable to workplace abuses, such as underpayment and unsafe working conditions.

“Firms are constrained by limited resources to keep their reputation and meet regulatory requirements, so managers need to balance the various aspects of CSR.”

Professor Zhou Xiaolu

The researchers analyse the Occupational Safety and Health Administration data from 2000 to 2018 and discover that the adoption of a universal E-Verify mandate reduces the likelihood of labour violations due to hazardous chemicals by 1.4 per cent and the likelihood of being fined for labour violations by 5.9 per cent.

Progress in one CSR area can undermine another

The researchers then compare the E-Verify data from 2000 to 2018 with the pollution data of 39,063 plants from the Toxic Release Inventory Programme that details the release of toxic chemicals to the air, water, and ground. Around eight per cent of observations in the sample are in states with universal E-Verify mandates, and 14 per cent are in states mandating public E-Verify.

The analyses suggest that when states adopt the universal E-Verify, pollution increases by 10 to 11 per cent with the effect becoming stronger over time. The public E-Verify mandate leads to a four to five per cent increase in pollution.

In states with a higher fraction of undocumented workers, toxic releases increase by 17 to 22 per cent after the universal E-Verify mandate and 13 to 15 per cent after the public E-Verify mandates.

Further analyses find that E-Verify mandates do not lead to financial constraints, and the increased pollution is mainly attributed to reduced emission efficiency. This means that companies substitute between different types of CSR activities as a trade-off when one area demands more attention.

“CSR substitution is a rational business decision to reallocate resources. Therefore, stakeholders may need to evaluate CSR holistically rather than in isolated aspects,” says Professor Zhou. “A company’s true commitment to sustainability can be understood by looking at the full picture and the trade-offs the company makes.”

Keeping accountability on responsibility

As many countries increasingly enforce certain aspects of CSR, such as Germany’s compulsory seats in supervisory boards for worker representatives, Australia’s mandatory environmental rehabilitation funds, and the like, the study shows that such policies may reduce other CSR activities. The substitution may occur gradually, making the hidden trade-offs harder to detect.

“When crafting policy, legislators should consider safeguards to prevent backsliding in other CSR domains. However, the optimal design will depend on many factors, such as regulatory context, industry characteristics, and resource constraints,” says Professor Zhou.

Furthermore, she notes that although substitution can theoretically apply to all CSR activities, the study does not imply that substitution should exist or should only exist for pollution.

“Cost and benefit trade-offs vary by firm and context, so it is difficult to have a one-size-fits-all prescription for CSR substitution,” she adds. “While our results show substitution between labour standards and pollution control, not all CSR activities are interchangeable, and some trade-offs are just impractical.”

Future studies could explore substitutions among other CSR activities and, if possible, develop a more robust framework for empirically identifying the most substituted activity.

The post Are companies just shifting CSR problems? first appeared on China Business Knowledge.

Featured faculty: Yang Haibin
Written by Pan Jingyi

Global supply chains have been under pressure from trade tensions, geopolitical risks and rising logistics costs. In response, many companies are restructuring their supply chains with nearshoring and friendshoring strategies, or building closer networks with trusted partners.

These multinationals are keen to foster deeper relationships for greater security and collaboration. Strong partnerships with suppliers help ensure reliability, reduce risks and improve overall operational efficiency. Given that a company normally has multiple suppliers, it seems natural to identify the most strategic suppliers and treat them like VIPs by investing more time and resources, while managing others in a standard way. This approach is known as the differentiated relational strategy.

“Using differentiated relational strategy enables buyers to allocate their limited managerial resources effectively to achieve the highest returns,” says Yang Haibin, Professor at the Department of Management of the Chinese University of Hong Kong (CUHK) Business School.

However, there is a catch. In collectivist cultures like China, where equality within a group is valued, such a strategy may be perceived as favouritism and generate resentment. Suppliers serving the same buyer expect similar respect and fair treatment.

“A collectivist country emphasises tightly knit relational networks and the equality principle of reward allocation among network members.”

Professor Yang Haibin

For global companies that rely on Chinese suppliers, whether based in China or operating overseas, effective management of supplier relationships is paramount. Therefore, in collaboration with Yan Zebin of Guangdong University of Finance and Economics, Yang Zhi of Huazhong University of Science and Technology, and Xie En of Tongji University, Professor Yang conducts a study titled Differentiated relational strategies in major supplier networks: A blessing or a curse in collectivist cultures? to examine the impact of favouritism in the supply chain.

Lessons learned from China

The team aims to explore whether the business environment can influence the effects of differentiated relational strategy, and China’s rapidly changing economy provides an ideal environment to study this. Chinese suppliers have seen uncertainties brought by geopolitics and technological turbulence affecting their relationship with buyers.

The researchers conducted a questionnaire survey and follow-up interviews among the management from around 130 manufacturing firms in China. They also obtained the data from the Annual Census Database of Manufacturing Firms of China to measure the company’s actual financial performance.

The research focused on major supplier networks that are highly interdependent, often producing complementary components or outputs for the same buyer. Frequent interactions among these suppliers make each of them aware of its status relative to its peers.

The team’s analysis indicates that companies employing differentiated relational strategies among major suppliers experience poorer financial performance. “A collectivist country emphasises tightly knit relational networks and the equality principle of reward allocation among network members,” Professor Yang says.

Why is equal treatment crucial?

In collectivist societies, group harmony is essential. Playing favourites violates this unspoken rule, and as a result, trust within major supplier networks is likely to be undermined. “Suppliers with relatively weaker relationships with a buyer may lose trust because they feel they are treated unequally,” he adds. “The favoured suppliers may also not fully trust the buyer, worrying that their special status could be at stake at any time.”

Furthermore, the differentiated relational strategy creates invisible divisions, fostering competition rather than collaboration among suppliers. The suppliers become less likely to share sensitive information or work together to solve problems for the buyer.

“Since major suppliers often play complementary roles, this lack of cooperation can lead to production delays, higher costs, and even less innovation, all of which are vital to the buyer,” Professor Yang says.

When favouritism hurts most and least

The researchers delve deeper into the conditions that amplify or moderate the damage caused by favouritism in supplier networks. They find that demand uncertainty and rapid technology change push the impacts of favouritism in opposite directions.

Demand uncertainty refers to fluctuations in how much customers buy. When demand is unpredictable, buyers must frequently revise production plans and place irregular orders. “Demand uncertainty requires buyers to facilitate efficient cooperation within major supplier networks,” Professor Yang says. Under these conditions, favouritism is especially damaging because it limits flexibility in the production lines and makes it harder to mobilise suppliers smoothly.

In contrast, when technology is advancing quickly, favouritism can be less damaging. In fast-moving tech fields, companies often reconfigure their supplier networks and bring in partners with cutting-edge capabilities. Looser ties to existing suppliers can streamline the switch to key partners, helping firms pivot without being held back by prevailing relationships.

Practical takeaways for global supply chains

More and more companies are leveraging algorithms and machine learning to allocate resources. Amid the convenience and speed of the computer brain, Professor Yang underscores the importance of human judgements, particularly since technology may overlook real-world conditions on the ground. “The purchasing managers cannot always rely on technology and may need to step in and incorporate the cultural factor into their decision-making.”

While the differentiated relational strategy may align with individualist cultures, the findings signal that multinational companies should avoid copy-pasting strategies across cultures. What works in individualist markets may backfire in collectivist ones.

RELATED ARTICLE

Beyond cargo, supply chain also transfers AI

“In collectivist contexts, purchasing managers should avoid a differentiated relational strategy and commit to maintaining equal relational strength with their suppliers,” Professor Yang says. “Transparent and fair practices such as consistent pricing, clear performance metrics and standardised communication may help. This builds trust and reduces administrative workload across the supply chain.”

That said, managers also need to weigh the company’s specific challenges, including demand uncertainty and the need for technological upgrades. If the main challenge is volatile demand, a cohesive supplier network with equal treatment is the greatest asset. If the company is in a race for technological innovation, a more flexible or more differentiated approach might be less harmful.

The post Why playing favourites with suppliers can backfire first appeared on China Business Knowledge.

Featured faculty: Choi Sungwoo, Kim Keongtae, Francisco Cisternas, Philip Zhang, Michael Zhang
Written by Pan Jingyi

Technology has been playing a pivotal role in shaping the business landscape in 2025. The world’s first law to regulate artificial intelligence (AI) by the EU is expected to be fully in force this year, while AI companies like Google, DeepSeek, and OpenAI have rolled out their latest models in rapid succession. Down under, Australia has also made history as the first country to ban social media for kids.

Meanwhile, US and Chinese companies have ramped up AI integration in their e-commerce to make “apps for everything,” and Nvidia’s CEO says quantum computing, which can solve problems too complex for classical computers, is approaching a critical turning point from early research to practical, scalable application. In hardware side, robotic companies around the world are vying for dominance.

Building on this surge, we bring together insights from the Chinese University of Hong Kong (CUHK) Business School faculty on how technology will redefine the future of business in China and beyond. In the second part, we will also share perspectives on other hot topics relevant to economics and consumer behaviour.

1. Future service robots: Humanlike interactions vs. humanlike bodies

Robots are no longer merely helpful assistants in factories. They have also become tireless servers at restaurants and dazzling performers on stage. A Morgan Stanley report estimates that the number of humanoid robots, or robots that mimic the human body, will likely reach nearly 1 billion by 2050, with China expected to have the highest number by then.

However, for human-centric service industries, such as hospitality and tourism, Choi Sungwoo, Assistant Professor of the School of Hotel and Tourism Management, offers another perspective: “I don’t necessarily think humanoid robots represent the primary direction for the future of service robotics.”

His recent research examined how assistive robots can empower employees with disabilities, but also highlights the importance of personal interaction. Robots can help with daily tasks, but customers still crave genuine human interactions and emotional connection, something that robots can’t truly replicate. Therefore, hiring humans in the hospitality industry remains crucial, even with the rise of robotic technology.

“Although a large body of research has demonstrated the benefits of humanoid robots, the goal of most companies is not to make robots physically resemble humans. Rather, it is to enable machines to interact with customers in more natural and humanlike ways, through tone, language, and behaviour.”

Therefore, Professor Choi expects AI-powered chatbots to lead the way in hospitality and tourism by increasingly interacting with customers, moving beyond simple tasks to take on more engaging roles. “By handling routine call centre-type tasks like answering frequently asked questions, handling bookings and check-ins, and scheduling appointments, chatbots can help free human staff to focus on more value-added and personalised services to create memorable experiences.”

2. Blockchain-verified AI agents bolster stablecoin safety

Blockchain is a decentralised digital ledger that records transactions across many computers, making it secure and transparent. As AI spreads across industries, its combination with blockchain could reshape finance by making systems more autonomous and interoperable, according to Kim Keongtae, Professor of the Department of Decisions, Operations and Technology.

His previous study shed light on the rising cryptocurrency and its influences on other technologies, underscoring the impact and functionality of blockchain on a wide range of applications. As AI takes on human tasks like trading and customer service, its ability to operate across platforms securely becomes critical, but without a universally accepted and verifiable digital identity, the AI programme cannot easily interact with other platforms.

“Blockchain provides the missing infrastructure, a neutral and verifiable identity layer for AI agents,” Professor Kim says. “The synergy extends to fraud detection and trust assurance. As AI models detect anomalies and suspicious behaviour, blockchain provides a tamper-proof source of truth to verify and cross-reference identities, transactions, and behavioural patterns.”

This secure, interoperable foundation also matters for a type of cryptocurrency called stablecoins, which are pegged to a stable asset like a fiat currency or commodities. As Hong Kong implemented a new regulatory framework for stablecoin issuers in August 2025, Professor Kim notes that its smooth launch is crucial for accelerating broader blockchain adoption and establishing the city as a regional leader in digital finance.

A cautious start by limiting issuance to a few trusted and well-funded companies can help establish standards for governance. “Over time, a broader range of HK$‑ or Chinese yuan‑pegged stablecoins could unlock local use cases and reduce reliance on US$‑backed stablecoins,” he adds. “In the near to medium term, investment and trading purposes are likely to remain the dominant use of stablecoins.”

3. More personalised e-commerce experience and smarter algorithms

E-commerce platforms nowadays deliver not just personalised product suggestions, but entire experiences tailored to consumers’ interests, making shopping feel personal, fast, and seamlessly woven into everyday life. Looking ahead, the momentum will only build.

Francisco Cisternas, Senior Lecturer of the Department of Marketing, anticipates that smarter models fuelled by richer data will allow more accurate predictions and better-timed offers. His recent study proposed a new budget-conscious model that can make the online retail recommendations more accurate. “Most routine tasks, such as recommending products, setting prices and designing creatives, are now already handled by algorithms,” he adds.

While AI is moving from a helpful assistant to the main decision-maker in e-commerce, Dr Cisternas points out that this doesn’t mean humans are out of the loop. AI often outperforms humans on repetitive and data-rich tasks, but it can falter when preferences are diverse or the data is scarce. “AI will be the central engine for many day-to-day decisions, while human and AI hybrids are the optimal mode,” he says.

In China, e-commerce giants have built a thorough ecosystem that integrates shopping, payments, logistics and social media under one roof. In the future, the country’s retail sector will continue to push the boundaries of seamless multi-channel integration, Dr Cisternas notes.

However, e-commerce must pivot from the old “right product, right moment” mantra to “right relationship, right lifetime” as shoppers look for more than just a quick purchase. “Personalisation that focuses on building long-term relationships with customers will outperform aggressive one-time price promotions,” Dr Cisternas adds.

4. Low-effort entertainment content dominates on social media

For most of us, social media isn’t just for connecting with friends, but also to see the world. Yet, it also often harbours a darker side that harms mental health and well-being. The Australian ban on social media has highlighted how “predatory algorithms” encourage youth to spend excessive time on screens.

This ban sheds light on the crucial role of humans as social media continues to evolve. “In the age of AI-driven tools flooding social media, humans will not compete with technology, but rather to distinguish themselves through superior judgment by knowing which content truly matters and defining the values and priorities that guide meaningful discourse,” says Philip Zhang, Associate Professor of the Department of Decisions, Operations and Technology.

Professor Zhang’s recent study shows that although AI can generate content and enhance social media consumption by filling missing metadata, it still falls short of matching the boundless creativity of humans. Human role would be even more crucial in the future, as Professor Zhang anticipates that social media will increasingly prioritise effortless consumption.

“If I were to imagine boldly, social media might instead evolve into a brain-computer interface one day,” he says. A brain-computer interface creates a direct communication channel between the brain and an external device, allowing users to control machines with their thoughts.

To complement the preference for low mental effort, simpler and more entertaining content would flourish greatly. This will make it harder for serious material to compete for attention—but not impossible.

He cites a podcast as an example and highlights the needs of seeking uninterrupted and insightful experiences. “While different platforms converge on similar features, the key is to identify niche audiences with specific and underserved needs.”

5. AI won’t fundamentally transform most industries yet

As elaborated above, AI has been used across industries. Our other recent articles have also explored how AI has and will continue to change our lives, from doing our jobs, data analysis, to supply chain management, with the vast majority of AI adoption remaining in early stages, rather than enterprise-wide scaling that reshapes core operations.

While there is lingering concern over the so-called “AI-winter”, where funding, interest, and excitement in AI dwindle, Michael Zhang, Wei Lun Professor of Business AI of the Department of Decisions, Operations and Technology, remains optimistic about its near-term impact. “Over the next few years, AI will undoubtedly boost productivity, especially in language and data-related tasks, but it is unlikely to fundamentally transform most industries, as current large language models still lack robust logical reasoning capabilities,” he says.

“Sectors closely tied to text and language processing, such as content creation, translation, legal documentation, and customer service, will experience the most transformative impact from AI in China,” Professor Zhang adds.

The post China Business Knowledge’s 5 tech waves to watch in 2026 first appeared on China Business Knowledge.