Economic forecasting is particularly challenging during global crises, with predictions often hampered by uncertainty and external factors.
Hwee Kwan Chow and Keen Meng Choy from Singapore Management University examine Singapore’s GDP and inflation forecasts during the Global Financial Crisis and COVID-19 pandemic, uncovering the influence of government projections and herding behaviour that affect prediction accuracy.
Read the original research: doi.org/10.1007/s00181-022-02311-8
Image credit: Adobe Stock / SergeyBitos
Transcript:
Hello and welcome to Research Pod. Thank you for listening and joining us today. In this episode, we look at the work of Hwee Kwan CHOW and Keen Meng CHOY from Singapore Management University’s School of Economics, who examine professional economic forecasts of Singapore’s GDP growth and inflation through major world events such as COVID-19 and the Global Financial Crisis, or GFC.
Studying a broad spectrum of economists, the researchers suggest that while growth predictions did not seem to be directly linked to Government projections during the GFC, both direction from national leadership and peer influence may have had a greater impact on forecasters during COVID-19. Here we’ll take an in-depth look at the accuracy of the predictions and the unique challenges that economic forecasters faced.
By its very nature, economic forecasting is never easy. Events such as the GFC or global pandemic are rare and, therefore, it is even less feasible to make predictions based on empirical evidence. Take, for example, the scale and intensity of the COVID-19 pandemic. Such an event was unprecedented and both the impact of the disease itself and the attempts to mitigate it, such as the implementation of multiple lockdowns, made it extremely difficult to predict the wider economic effect.
The researchers’ paper on Economic forecasting in Singapore: The Covid-19 experience reveals that the blueprint for economic prediction in the case of COVID-19 had flaws from the outset. The last major epidemic had been 2003’s SARS outbreak which was quickly controlled and did not spread beyond Asia hence lessening the global economic impact. Its economic effects within Singapore were therefore a poor comparison for measuring the impact of COVID-19. Put simply, the economic turbulence caused by 2020’s pandemic was extremely complex and impossible to compare with the past. Consequently, forecasters were unable to employ the usual methodology for their predictions.
We’ll now take a deeper look at the predictions of a group of Singapore’s professional forecasters, collated by the country’s central bank, and how their analysis differed during the pandemic. This will be directly compared to the same forecasters’ predictions during the GFC.
The forecasts used for the purpose of this research were taken from the Monetary Authority of Singapore’s Survey of Professional Forecasters and employ a number of key factors that the private sector uses for their analysis. The latest export rates, employment levels, inflation and real GDP are all meticulously considered to influence long and short-term forecasts.
So, what is the story behind the experts? The vast majority of participants in the survey are economists working across a variety of companies within Singapore’s financial sector. Chow and Choy’s research paper focuses primarily on two main types of forecast. Point forecasting focuses on singular future events. Probability distribution forecasting, however, examines the range of possible occurrences and the probable likelihood of each one. The findings focus on the forecasts of two main variables for the two global crises: the GDP growth rate and Consumer Price Index, CPI, yearly inflation rate.
The research suggests that forecasting inaccuracies for growth were worse in COVID-19 than in typical economic times. However, this is not so when compared to the GFC.
Interestingly, the inaccuracies in the CPI forecasts for COVID-19 are less straightforward. Over a one-year period, the magnitude of forecasting errors were clearly higher than both during the GFC and more typical economic times. However, the opposite is true for the longer-term two-year forecasts. This variability further highlights the complexity in making accurate economic forecasts during both COVID-19 and the GFC.
To explain why these inaccuracies might have existed, the research first examines the likelihood of bias. There is evidence to suggest that the surveyed economists may have produced biased growth forecasts in the GFC leading to lower growth forecasts than expected. However, no evidence of such bias was found during the pandemic. During both COVID-19 and the GFC, there was evidence of bias regarding inflation meaning, both times, forecasters underestimated the elevated inflation rate.
Chow and Choy examine why this bias may have occurred. First, they suggest forecasters were influenced by government reports and statistics in an atypical manner to normal predictions. Typically, in Singapore, official government predictions are given as a range. The authors suggest that the rate at which professional forecasters’ predictions fell outside of the government range for growth forecasting during the GFC demonstrates some evidence of independent analysis.
During COVID-19, however, a far greater proportion of growth and inflation predictions were within the government’s estimated range. Indeed, predictions outside of these ranges were close to zero indicating a potential unwillingness to stray too far from leadership’s forecasts.
The authors secondly look at the effect of what they term herding behaviour. The economic forecasting world in a relatively small country such as Singapore means increased familiarity and peer-led influence within the group. In unprecedented times of uncertainty, making popular predictions ensure a forecaster is not singled out or questioned should the prediction prove to be false. Economists may also be reticent to develop a reputation for extreme forecasts in such a small group.
There is also the fact that such a well-connected group within the economic landscape will be sharing key information with each other from which their predictions are derived.
Conversely, it should be noted that other forecasters may make alternative predictions for precisely these reasons. If they go against the grain, so to speak, and their predictions if proved correct, they stand to gain positive publicity and an elevated reputation.
Either way, these possible behaviours demonstrate prediction bias on top of the information available to them,
The authors hypothesise that it would be reasonable to say herding behaviour is more prevalent in times of economic uncertainty where they have less access to reliable tangible data to work from. To examine this, the research looks at the differences between individual forecasts and general consensus at the beginning of each year for the GFC and COVID-19, in order to see if they remain close to zero indicating evidence of herding. During both events and for both the variables of growth and CPI inflation, the differences in predictions are not significant indicating that herding behaviour has occurred.
In summary, professional forecasters can predict growth better during Covid than GFC. This is likely because they think the authorities have more information and hence follow official forecasts closely during Covid but not during GFC.
The research also looks at probability distribution forecasts to examine disagreement among forecasters in their predictions.
Over a 19-year period, the diverging opinions on future growth between forecasters on a year-on-year basis were relatively stable – other than during the GFC and COVID-19.
Notably, it was the year after the GFC that caused more disagreement amongst forecast predictions than during the GFC itself. The authors suggest this may have been due to the European debt crisis in addition to uncertainty and differing opinions over the long-term recovery from the crisis itself. Comparatively, there was a significant fall in forecaster disagreements between 2018 and 2019 despite trade challenges between China and the USA – the authors hypothesise that this was due to lower forecasted growth, smaller forecast ranges and hence reduced scope for alternative opinions in the survey responses.
For inflation forecasts, disagreement levels in both current and next year’s predictions were relatively stable even when Covid struck. The authors surmise that this was due to a lack of evidence to forecasters then that COVID-19 would have such a significant impact on inflation rates.
In conclusion, considering the unprecedented nature of COVID-19, the challenges of Singapore’s economic forecasters are not surprising. Objective uncertainty significantly increased after the pandemic leading to an increased likelihood of ‘herding behaviour’ and an increased unwillingness to deviate from official government forecasts.
Forecasts for inflation were significantly lower during the pandemic. Respondents showed evidence of both herding behaviour and leader influence and demonstrated little increase in disagreement at the start of the pandemic. This suggests that, at least in the short term, predictions for inflation were largely similar in the studied period. While forecasting failures may have occurred, the behavioural patterns amongst professional forecasters further highlight the unpredictable nature of this unprecedented event.
That’s all for this episode, thanks for listening. Links to the original research can be found in the shownotes for this episode, and don’t forget to stay subscribed to ResearchPod for more of the latest science!
See you again soon.
Leave a Reply