The new economic model making sense of unique circumstances created by COVID-19

The new economic model making sense of unique circumstances created by COVID-19

With the pandemic continuing to wreak economic havoc in many ways including sudden shutdowns, multiple central banks are starting to use a unique statistical model co-developed by our Department of Econometrics and Business Statistics senior research fellow Dr Benjamin Wong.

In early 2020 as COVID-19 morphed into a global pandemic, medical and science journals went into overdrive to speed up publishing the latest research on the virus.

But it wasn’t just medical journals. With the US economy suffering the biggest slowdown since the Great Depression , economics journals also looked to fast-track papers that could help us understand the implications of a virus shutting down huge parts of the world economy.

One such fast-tracked paper was a new statistical model of the US economy, which had already caught the attention of the US Federal Reserve.

The three researchers behind the model had first posted it online in April 2020, shortly after President Trump had declared a national emergency.

“The model can already incorporate real-time financial data like stock markets, but every crisis always draws out new economic tools, and this is a framework that should be able to incorporate them.”

When they started updating it they suddenly had a unique and timely window on how the US economy might actually be tracking amid the sometimes wild gyrations in official data that economists were struggling to make sense of – US Gross Domestic Product numbers during 2020 tumbled and then rose by more than 30 per cent.

“It was just coincidence really that we put the model online and started updating it just as COVID-19 was hitting the US economy,” says Dr Benjamin Wong at Monash Business School, who with co-authors Professor Tino Berger from the University of Göttingen and Professor James Morley from The University of Sydney, had been developing the model’s various iterations over a number of years.

“We were probably lucky timing-wise as I’ll probably never have a paper accepted as quickly as this one was, but I think this was a case where people recognised that our model had potentially something useful to say about the US economy at what was, and still is, a critical time.”

It can often take years or more for economics journals to finally accept a research paper, but the Journal of Econometrics accepted Nowcasting the output gap in late August 2020, less than three months after the paper was originally submitted.

What is the output gap?

The output gap is economic parlance for the amount of slack in an economy, and is a key input into central bank decisions on interest rate levels.

A negative output gap suggests that demand isn’t strong enough to sustain efficient economic production, leading to business shutdowns and job losses that a central bank may seek to head off by lowering interest rates to stimulate activity.

A positive output gap indicates an economy that is overheating, with production struggling to keep up with demand and potentially stoking prices – inflation – which a central bank could try to combat by raising interest rates to dampen demand.

At times of crisis like during COVID-19 when there are sudden downturns in an economy, being able to infer the output gap as accurately and as quickly as possible is crucial to understanding how deep or sustained a downturn or recession may be.

This sort of information can then inform central bank decisions on whether initial rate cuts or other actions to stimulate demand are adequate or even an over-reaction.

Modelling and estimating this output gap isn’t new – central banks and economists around the world are forever seeking to understand where the output gap is at a particular time. This is called “nowcasting”, or using data and forecasts to infer what is happening right now.

The challenge of fast-changing events and data

But it is a fraught exercise. Data by definition is out of date, and important data like quarterly GDP figures can be several months out of date by the time they become available – and even then they will likely be revised. Other data may be more frequent, like monthly unemployment data, which may give earlier signals on where GDP might be trending.

When events are moving fast like the sudden shutdowns in the wake of COVID, this inevitable lag in the data becomes even more problematic.

“When both Sydney and Melbourne went into lockdown earlier this year, the latest economic data, such as the latest unemployment figures, isn’t helpful in gauging what is now happening,” says Dr Wong.

To try and ‘nowcast’ the output gap, Dr Wong says economists have tended to separately model different data sets, like GDP, and then combine the different results in various discussions aimed at settling on a view of where the gap may be sitting.

But Dr Wong, who once worked at the Reserve Bank of New Zealand, says this process is “inevitably open-ended and can be somewhat circular since it is, at heart, still an educated opinion given the uncertainties economists face.”

To an extent this will always be the case, but what Dr Wong and colleagues have done is create a model that can take in all the various data – whether quarterly, monthly or even daily – into a single framework that can provide greater transparency and discipline around these critical discussions.

Dr Benjamin Wong

The benefits of a single framework

“Our idea was that if we could put all the relevant information into a single model, then it would discipline, and make transparent, this process of debate on where an economy may be at any given point,” says Professor Wong.

“On a simple level, this is what our model does. Instead of wasting a lot of time debating different people’s sense of where they suspect real employment or GDP may be, the model provides a framework that can be constantly adjusted to more rigorously test what the overall implication of these different data points and opinions mean for inferring what the output gap is.”

“This makes for a much more rigorous and transparent basis for debate than someone just saying they ‘feel’ that a data estimate is too high or too low.”

“Of course, at the end of the day, it is a model which is only as good as the data and assumptions put into it, so it isn’t something to be blindly followed.

“But a model like ours that combines different data coming out at different times, can hopefully provide a more fruitful starting point for discussions.”

A unique model that’s all about real-time data sources

Indeed, one of the advantages of the model, Dr Wong says, is that it can be adapted to incorporate any future – more “real time” – data sources that economists may identify. He says the speed with which COVID-19 has disrupted economies has economists exploring alternative data sources that could give a more immediate reading on the state of an economy, like for example tracking restaurant bookings or using mobile phone signals as gauges of economic activity. There is also a live debate on how economists could better measure levels of unemployment given concerns that current measures are inadequate.

“The model can already incorporate real-time financial data like stock markets, but every crisis always draws out new economic tools, and this is a framework that should be able to incorporate them.”

By taking in more frequent data releases and combining them with quarterly information, the model has been shown to be able to give timely estimates of the size of the output gap that are broadly consistent with later eventual estimates based on the release of quarterly GDP figures. “That the model has been able to do this so well has been a bit of a surprise, and is a sign that what we’ve done really is useful.”

The sort of multi-frequency data the model has incorporated include US monthly data on various unemployment measures, consumer sentiment surveys and the interest premiums on corporate debt that are a measure of risk.

US Federal Reserve, European Central Bank among those taking notice

Certainly, their work is attracting attention. Even before it had been published Dr Wong had been sharing the details of their model and how it worked with economists from the US Federal Reserve. The researchers have also consulted with the Reserve Bank of Australia on possibly creating a local version, and the model is being adapted for the Euro Area as part of their ongoing work with the European Central Bank.

With support by Monash Business School’s Impact Acceleration Grants Scheme, the public online version of the model, outputgapnow, is being maintained and updated, and versions created for online training.

Through the Malaysia-based Southeast Asian Central Bank Research and Training Centre and Euro Area Business Cycle network, the researchers have also been teaching central bank staff in Asia on this modelling approach.

“Creating a model like this isn’t a trivial undertaking – it has taken us years to get to this point – and the model itself will always be evolving, so in actually using the model there is a skills gap that needs to be addressed in terms of having economists who understand this work,” says Dr Wong.

“We are talking about coding, mathematics, statistics and having the economic intuition that comes from really understanding the discipline.”