Every six months, the IMF publishes a report called the World Economic Outlook,

where it gives a forecast for the global economy.

Forecasts like these are a big undertaking. So how are they made?

We're going to give you a behind-the- scenes look inside the process.

We are once again downgrading growth for 2019 to 3%.

So that's Gita Gopinath, she's the chief economist at the IMF.

We caught Gita presenting the IMF's World Economic Outlook report to journalists from

all around the world at the fund's annual meetings in Washington.

The World Economic Outlook is a nearly 200 page-long report that includes

hundreds of thousands of data points from 189 member countries of the IMF.

It's released twice a year and includes economic predictions for the next two years,

along with thematic chapters that look at big-picture themes like productivity, climate change or technology.

We are really the only institution that covers literally everybody.

So how do economists even get to predictions in reports like the World Economic Outlook?

Let's start at the beginning. First, teams collect data from countries all over world.

Government agencies measure and report stats like GDP, inflation or trade balances

this is called “hard data.”  There's also “soft data,” which includes surveys of businesses and consumers

on how they feel about the economic environment now and in the future.

The IMF's country teams then send this data back to Washington.

I would say for each country, we're talking one or two thousand data points.

We collect up to three or four hundred different series for most countries.

The country teams send us new sets of data approximately 1,500 times every six months.

So we're looking at a lot of numbers. A lot of data.

It's not just the big points they look at. GDP, for example,

helps provide a snapshot of a country's overall economy, but individual gauges are also important.

For example, oil prices matter a lot for predictions about GDP growth in countries like Saudi Arabia or Venezuela.

That's because their economies are largely dependent on oil.

How do you make sure this data is good data, it's not manipulated or wrong?

We have a whole series of tests and different ways of looking at it.

Obviously we make sure the numbers actually make sense.

We do consistency checks to make sure, for example, that the components of GDP actually add up to GDP.

We look at charts to make numbers appear to be trending in the right direction.

So once all this data is collected, economists don't just plug it into an equation to generate a prediction.

They have to consider a lot of other factors, too.

Where we really try to provide value added is pulling common threads,

trying to come up with a common story that is not like

we are just adding up countries and seeing what the global result is.

The teams here at the IMF start working on their predictions in the months

leading up to the publication of the World Economic Outlook.

And they are constantly updating their forecasts as the data evolves.

It would be nice if we had a process whereby we come to final numbers and then we write the report.

But the truth is that the report is written as the numbers evolve because we get data releases all the time.

That still doesn't mean economists get it right.

In fact, IMF research itself has shown that economists often fail to predict recessions.

An analysis of 63 countries from 1992 to 2014 found the IMF consistently overestimated GDP growth

leading up to a recession and only revised growth downwards once the downturn was already happening.

Other papers have uncovered similar findings.

Another challenge for economists is translating predictions from wonky data into plain English.

Normally when you come out of graduate school you start with very technical language but as you

go along your career, you start realizing you really need to simplify your language

and boil it down to what it means for individual firms, families, governments.

So we are walking into a meeting where final preparations are underway

ahead of the press conference where the World Economic Outlook is released.

We got a look inside a meeting happening just days before the report was released.

Economists discussed their findings with select members of the press and

studied up on the material so that they'd be prepared to answer questions about any country in the report.

The trick is figuring out how to provide a sensible answer

to questions on places that people may not know very much about.

So today's the big day the report is released.

These are the predictions that everyone has been waiting for and we are going to check it out.

This year's report found global growth will slow to 3% in 2019,

that's the lowest level since the 2008-2009 financial crisis.

A lot of factors affected the IMF's forecast, but some of the biggest included more trade barriers,

geopolitical uncertainty and aging populations in advanced economies.

The data is all published online, so that many other economists around the world

can use it to make their own predictions.

How do you make sure that the predictions here in the forecast are constantly kind of up to date and relevant?

You have a lot of questions in the press conference about the news of the day.

What are you doing to be prepared?

So, these estimates actually they stay alive.

At some point we freeze it because it has to be published and put in the document, but

as of now we have these numbers constantly being updated.

Even after the findings are presented at the IMF's press conference,

they're highlighted throughout the next days and months by IMF officials and other world leaders.

The world economy is slowing down, we are in a synchronized slowdown.

So you might think that after this is all done, economists can take a breather for a little bit.

But it turns out, they're already working on the economic outlook for next year.

We actually start thinking about the next one even before we publish the forecast.

Hey everyone, it's Elizabeth, thanks so much for watching our video.

What do you think about the making of the World Economic Outlook? Let us know in the comments.

And you can always leave us other ideas there too. See you later.