Fuad Aleskerov: 'Decision-Making under Deep Uncertainty'
In an interview for HSE News Service, Fuad Aleskerov, Distinguished Professor of HSE University and Member of Academia Europaea, shares his insights into making decisions under deep uncertainty, highlights the crucial role of cooperation in this context, and unveils plans for the upcoming launch of a specialised course on decision-making and the establishment of a mirror laboratory at HSE University.
— Professor Aleskerov, you have recently returned from a business trip to deliver lectures as a member of Academia Europaea. What kind of organisation is this, and what prompted them to invite you?
— In 2018, upon receiving an invitation to join the academy, I began searching for information about this organisation. It turns out that its membership includes some 3,800 scientists, among them 52 Nobel Prize winners. I have been a member since 2018. They had not hosted any events for an extended period due to the COVID-19 pandemic. This year, I received an invitation to give a presentation, in which I shared my recent work. Specifically, I gave a talk on food security—it was essentially a scenario analysis designed for decision-making under deep uncertainty.
— Could you kindly elaborate for those unfamiliar with the concept of decision-making under uncertainty?
— The theory of decision-making has existed for many years, and its mathematical foundations were laid more than 250 years ago. In the process of making decisions, we typically operate under the assumption that the consequences of specific actions are known. However, in the 20th century, models for planning under uncertainty were proposed, incorporating a probabilistic assessment of potential outcomes based on the likelihood of specific events occurring. Presently, a fascinating frontier within this field has emerged: deep uncertainty, when we do not only lack an assessment of potential consequences but are unclear about such consequences altogether.
— How long have you studied the area of planning under deep uncertainty?
— I first embarked on studying planning under deep uncertainty in 1999, following an earthquake in Turkey that resulted in numerous casualties. By official data, about 17,000 people were killed by the disaster. After the earthquake, we travelled to Istanbul, where I spent some time working at a university involved in surveying city mayors. Since there was evidently a potential for another earthquake to occur, the city mayors were asked, 'How do you plan to prepare for a new earthquake if it were to occur?' One of the mayors responded by stating that he had procured 4,000 body bags for the victims. It really shocked me then. As we approached the end of the 20th century, could we do more than stockpiling body bags? I came up with mathematical models and invited the Department of Computer Science and the Department of Business and Management of Boğaziçi (also known as Bosphorus) University in Istanbul to join in. We lacked any probabilities regarding the nature and likelihood of potential outcomes. It was also unclear how severe the impact might be—whether buildings would collapse, lives would be lost, or no such outcomes would occur. None of this was known. However, we came up with the idea of constructing a scenario analysis to understand the potential outcomes resulting from specific events. With this understanding, we could assert that in the event of an earthquake following a specific scenario, a certain number of buildings would collapse, a specific count of people would be injured, a certain quantity of medicines would be required, and so forth. The paper was published in a leading journal on emergencies in either 2004 or 2005 and continues to be referenced to this day.
— As for your presentation on food security at Academia Europaea's conference, what was it about?
— We at HSE University have recently released a series of papers on food security. These papers analyse food exports between countries. We have been able to apply our model to a situation where the future is uncertain, yet should a certain event occur, we can project its impact on the existing food flows between countries. When we already had this model developed, Pakistan, a self-sufficient country, was hit by a flood. There is a certain normal rate of grain consumption, and in Pakistan, the level of consumption used to be nearly twice this normal rate. Presently, ten million children in Pakistan are experiencing subnormal consumption levels, and the entire country is operating at half the typical consumption level. We simulated specific scenarios, such as the potential outcome if Russia were to decrease its food exports to other countries by 5% and redirect the surplus to Pakistan. It was revealed that the situation in Pakistan would improve, but not significantly. But if Russia were to reduce its exports to other countries by 10% and redirect the surplus to Pakistan, the improvements would be significantly more noticeable. This was the kind of scenario analysis we conducted, and various other scenario analyses pertaining to grain consumption could be performed.
At the conference, we reported the first part of this work. Further, we applied similar models to analyse the exports of oil and rare earth elements.
There is currently significant tension between China and the United States regarding rare earth elements. In the context of our work, the following scenario can be constructed: let's assume that substantial deposits of rare earth metals are discovered in Sweden, prompting European countries to procure them from Sweden and subsequently decrease their imports from China. These have been our more recent projects. I presented the findings from our analysis on rare earth elements, as well as on food security and oil, at Academia Europea's conference.
— Did they generate interest?
— I was invited to discuss these findings at the Adam Smith Seminar in Munich and subsequently at Sorbonne University. There was significant interest, with numerous questions asked, as it was the first time such a scenario analysis was presented on matters of critical importance. We will continue this work.
— What are your plans for further exploration of this area?
— We have recently started a collaboration with the HSE Faculty of World Economy and International Affairs to incorporate other types of food in the analysis. Initially, we only analysed the situation with grains, but have since expanded our focus to include other varieties of food. We are planning to undertake substantial work in this field.
For instance, we have recently been calculating the potential outcomes from various government responses in the event of a volcanic eruption in Kamchatka. Assuming a volcanic eruption in Kamchatka, what should be the appropriate response? We are uncertain as to how it might unfold. If the wind blows in a specific direction, everything will be covered in volcanic ash, but if it blows in a different direction, it will carry the ash out to the ocean, which is less severe. Some response is necessary in any scenario. We are now actively engaged in this work.
We are also contemplating the creation of a specialised course on decision-making under deep uncertainty.
— When can we expect such a course to be launched?
— The course is likely to be launched next year, in the winter of 2023/24. All preparations are complete, and we have been in discussions about engaging political scientists in the development of the course.
— What are some of your other plans?
— We would like to open a mirror laboratory at HSE University to focus on decision-making analysis. A colleague of mine who wrote his Candidate of Sciences dissertation under my academic supervision has opened a laboratory in Turkey which analyses different types of decision-making. We have already commenced a collaborative project centred around network analysis of economic sectors and their interactions.
— What are the benefits of conducting network analysis in this context?
— I will illustrate this using banks as an example. Once, I received a phone call from the deputy governor of the Central Bank, who requested that I examine the interactions among banks using network analysis models. There are classical centrality indices that describe interactions between different entities. What specifically is analysed, be it the volume of grain exports, the number of telephone calls, or the number of loans issued between banks, is irrelevant. The centrality index reveals the crucial elements, indicating who plays the central role in the system. What I have observed in the context of banks is that the classical approach fails to consider the characteristics of the banks from which money is borrowed. Indeed, this approach does not differentiate between the situations of non-repayment to a large bank or to a small bank. But if I were to take out a million-dollar loan from a major systemically important bank and fail to repay, the CEO may refuse to greet me, but the impact on the credit institution would not be too severe. In contrast, a smaller bank may face collapse if I were to default on repayment. Now, imagine that both of us have borrowed substantial amounts from a small bank. If one of us repays while the other defaults, the bank may survive; however, if both of us fail to repay, the bank could face dire consequences. Therefore, I have proposed models which take into account the characteristics of the lending bank, as well as the collective impact of groups of borrowers on the bank. Another proposed method takes into consideration what is known as cascade effects.
— Let us return to the subject of establishing a mirror laboratory at HSE University. If you win in the competitive selection, what other activities will the laboratory undertake?
— It will analyse the interactions among sectors of the economy, also referred to as intersectoral interactions. For example, we are currently examining data from Turkey and France. There is available data on each sector's respective contribution to the country's economy. By conceptualising this as a network, we can identify the key elements and understand their significance in various respects. The same approach can be applied to interactions among countries.
— What potential partners does HSE University have in establishing this mirror laboratory?
— First, our Turkish colleagues. We also have plans to engage Chinese colleagues a year later. Regrettably, it seems unlikely that our European colleagues will participate in this initiative. While I have many friends in Europe, when it comes to inter-university agreements, they are instructed by their governments not to collaborate with Russia. Even in the mid-80s, when Ronald Reagan referred to the USSR as the 'Evil Empire,' universities adhered to the principle that 'politics is politics, but we pursue science.'
— What benefits can be gained from involving Chinese colleagues in your work?
— They are conducting highly impactful work on supercomputers, while our models are designed to handle vast amounts of data. It might be possible for us to use Chinese supercomputers. In general, collaboration and the active involvement of colleagues are crucial for addressing such challenges. Contemporary science is inherently interdisciplinary, and models of this nature need to be designed and computed by a team of professionals.