Why we should consume forecasts with caution

The uncertainty of not knowing what lies ahead is disarming in many ways because it renders us powerless. One way to navigate that ambiguity is to use historical data to forecast the future. The predictions help create some certainty about where the future is headed and, in some way, make things less obscure. Various modelers, for example, had anticipated future COVID-19 infection cases, death, and recovery rates. BMC public health projected that Kenya would reach 1,000 confirmed cases by 14th April 2020 (33 days after the first confirmed case) and 4,000 cases by 21st April 2020 (40 days after the first confirmed case). The Ministry of Health numbers were similar, projecting “1000, 5,000 and 10,000 cases of Coronavirus by early, mid, and late April 2020, respectively ceteris paribus[1]”[2]. In hindsight and according to WHO data, on 14th April, we were at 218 confirmed cases, and by 21st April, we were at 291 confirmed cases.[3]

Other interesting examples of data forecasts include those made at the beginning of the year. GDP was expected to grow to 6% following consistent growth in the last five years, which placed Kenya among one of the fastest-growing economies in Sub-Saharan Africa. The expected growth was attributed to continued government spending of 25% of GDP on infrastructural ‘development’ projects, high remittances of 2.9% of GDP, a solid consumer base which accounts for about 80% of GDP, and increased foreign direct investments due to ease of doing business and high macro-economic stability. The continued extraction of oil from Turkana County and other exports, trade, and anticipated higher investments in the big four sectors - health, housing, manufacturing, and agriculture were also expected to boost the economy significantly. The investments would result in savings on health care, increased access to housing, and enhanced food supply (also propelled by expected favorable weather), which would then bring down the cost of food and reduce the cost of living hence lower inflation rates of about 5%.

Unfortunately, the forecasts have had the opposite results, primarily due to the unforeseen effects of locusts and COVID-19 (events that were completely out of anyone’s know or control). Between April and July 2020, the economy was projected to decline to 1% as government investments “slowed down significantly, and civil works were delayed by global supply disruptions and the limited supply of labor associated with COVID-19”[4]. Remittances declined by 20% between January to April 2020[5]. FDI was projected to contract by 25 to 40 percent in 2020, and the adjusted inflation was projected to increase to 6.5%. Labour force participation declined from 75% to 57% due to job losses meaning that the consumer base had considerably shrunk. Export revenue was expected to reduce by a minimum of 25% (USD 1.5bn), while trade with China was expected to decline by about 40%. The COVID -19 pandemic coincided with the start of the maize planting season, the primary staple food. Following a sharp decline in maize production in 2019 due to early-season drought, the unresolved locust pandemic as well as floods in 2020, food stocks are likely to decline, and food prices considerably increase thus a high cost of living.

The point this article makes is not that forecasts are bad but how they are presented and in turn consumed is a big problem. Modelers present forecasts as future facts instead of indicative possible future trajectories that could happen. When the future does not pan out as predicted, people mistrust data or downplay its importance and, therefore, do not use it to plan or prepare for various eventualities. Forecasts are also inadequate by themselves because they emanate from the fact that data projections are linear. They map historical and current data to show future patterns – which is not necessarily how life works. Past trends can indeed continue in the future but not always. Data analysts try to incorporate assumptions into forecasts to bring out other variations of the trends - hence terminologies like the base case, low case, or high case scenarios. However, these data patterns usually consist of stand-alone parameters which, when interpreted by themselves to discuss multifaceted societal issues, will obviously give a false picture of the whole. This means that the interconnectedness of stand-alone parameters has to be explored together because that is how life works hence an analysis of multi-sectoral patterns, their relationship to each other, and the ensuing implications of those multiple relationships.

There is also a dire need to unpack the reasons why a trend moves a certain way and uncover other factors that would affect the course of that trend in a different way from the current norm.  Understanding these push and pull factors altering patterns enables people to get a more accurate picture of how the future might unfold, which is seldom incorporated in predictive data. Tracking the evolution of these underlying aspects altering data patterns, some of which are not quantifiable but qualitative in nature, is particularly painstaking but important because it points to where the change is likely to happen in a more precise fashion. This way of processing data removes the blind spots that keep people locked inside some obsolete assumptions of what the future looks like and which blocks them from seeing emerging realities that eventually become the main cause of their strategic failures.

Ultimately, data and trends are only meaningful when they are innovatively processed, and important insights are extracted to enrich people’s understanding of the issue. However, the value of data, trends, and patterns is in the analyzed and leveraged information to enable evidence-based strategic decision-making, organizational agility to manage risks, and positioning for resilience. 

---

[1] Latin word meaning all things remaining constant

[2] Ministry of Health website: 

[3] This fact is not meant to belittle the gravity of COVID-19 in any way

[4] World Bank. (2020). Turbulent Times for Growth in Kenya: Policy Options during the COVID-19 Pandemic. Kenya economic update, 21

[5] From 259,392.71 in January to 208,217.73 USD thousand in April 2020

Building a shock-resistant planet in a world of such great difference

I had a rare opportunity to speak with innovative people questioning whether global immune technologies seeking to find systemic solutions to global pandemics can help build a shock-resistant planet. Such technologies include those that "can detect a novel pathogen in the air, water, or soil of the Earth and rapidly sequence its DNA or RNA"[1] to neutralize the pathogen before its damaging effects begins, and in a sense entirely bypass the limitations we have seen with governments in managing diseases like COVID – 19. This conversation can be retrieved from https://bit.ly/2ZNiSqU. 

I live in a continent that shoulders one-quarter of the global disease burden, has less than 2% of the world doctors, invests less than 1% of global health expenditure, and is among the continents with the lowest access to healthcare services in the world. While these impressive innovations towards "precision medicine" seek to prioritize preparedness, prevention, and confinement of disease outbreaks, I often wonder the extent to which such innovations can genuinely be global in a world of such significant difference.

The global health sector is predicated on an exploitative model that prioritizes profits over people. As such, it has built a curative system that is mostly available to those who can pay, and which has proved to be unsustainable with occurrences of global pandemics like SARs and COVID-19, especially among the poor. For any such system to work optimally, it must avail health as a public good and a right while collectively working towards the achievement and maintenance of good health through preventive healthcare that is inherently less costly. This value-based health system would have to centralize the needs of people to produce consistently superior outcomes at the lowest cost. This means universal access to high-quality medical care, patient safety, convenience, cost containment, and ultimate satisfaction in services rendered for the collective wellbeing of all to remove the threat of global infections effectively.

Problem-solving technological innovations have been vital in making enormous strides in the health sector. However, technology has not been without its challenges. A first critical concern is the relationship between humans and machines. Weak collaborations between people and machines resulting from technological misuse - which occurs when people overly rely on automation inappropriately, or technological disuse - which occurs when people do not fully appreciate the benefits of automation, have become increasingly costly and catastrophic in as far as they compromise people's safety. WHO acknowledges that the diffusion of innovations is a major challenge because of: user perceptions - on the benefits of change, on uncertainty levels, on compatibility between values and current needs, as well as on simplicity of use. This means that the process of acceptance and diffusion has its own pace, rules of the game, and complexity level. For the successful and accelerated interface between humans and technology, trust becomes a critical factor. Just as trust aids relationships between people, it also guides and influences reliance and use of the said technology, its mastery, and the extent one feels a sense of security when using the technology, especially in complex and uncertain situations. This means that the way certain technologies are introduced is critical in building or undermining trust.

A predominant characteristic of the health model that has been pursued over time is the top-down prescriptive approach of creating solutions by the 'power holders' because they have the resources and the capacity to do so. The only problem with this approach is that it now has to create solutions for cross-border challenges that manifest differently in different cultures and communities. As such, automated systems have to genuinely understand the users' unique situations in different parts of the globe to design truly inclusive and adaptive systems. This can only be achieved through new processes that are mutually collaborative and participatory – to co-design and co-produce unique gap-filling solutions that value diversity and incorporate indigenous knowledge to build true resilience. Therefore, the system's effectiveness will only be achieved with the dismantling of hierarchical, bureaucratic, and egocentric institutions and constructing transformative ones that are responsive and agile enough to bring about the needed change. The only question here is, are the beneficiaries of the current global order able to shift to the required paradigm?

References 

Berwick, D. M. (2003). Disseminating innovations in health care. Jama, 289 (15), pp. 1969-1975.

Hoc, J. M. (2000). From human–machine interaction to human–machine cooperation. Ergonomics, 43 (7), 833-843.

Lee, J. D., & See, K. A. (2004). Trust in automation: Designing for appropriate reliance. Human factors, 46 (1), 50-80.

Porter, M. E. (2009). A strategy for health care reform—toward a value-based system. New England Journal of Medicine, 361(2), 109-112.

Porter, M. E. (2010). What is value in health care?. New England Journal of Medicine, 363(26), 2477-2481.

World Health Organization. (2006). The world health report 2006: working together for health. World Health Organization.

---

[1] for more information visit the Atlantic Council website.