Complete Intelligence

Categories
News Articles

How AI-based ”nowcasts“ try to parse economic uncertainty

This post was published originally at https://www.emergingtechbrew.com/stories/2022/06/17/how-ai-based-nowcasts-try-to-parse-economic-uncertainty?mid=13749b266cb1046ac6120382996750aa

This month, the S&P 500 officially hit bear-market territory—meaning a fall of 20+ percent from recent highs—and investors everywhere are looking for some way to predict how long the pain could last.

Machine learning startups specializing in “nowcasting” attempt to do just that, by analyzing up-to-the-minute data on everything from shipping costs to the prices of different cuts of beef. In times of economic volatility, investors and executives have often turned to market forecasts, and ML models can offer a way to absorb more information than ever into these analyses.

One example: Complete Intelligence is a ML startup based outside Houston, Texas, that specializes in nowcasting for clients in finance, healthcare, natural resources, and more. We spoke with its founder and CEO, Tony Nash, to get a read on how its ML works and how the startup had to adjust its algorithms due to market uncertainty.

This interview has been edited for length and clarity.

Can you put the idea of nowcasting in your own words—how it’s different from forecasting and the nature of what you do at Complete Intelligence?

So Complete Intelligence is a globally integrated machine learning platform for market finance and planning automation. In short, we’re a machine learning platform for time series data. And nowcasting is using data up to the immediate time period to get a quick snapshot on what the near-term future holds. You can do a nowcast weekly, daily, hourly, or minutely, and the purpose is really just to understand what’s happening in markets or in a company or whatever your outlook is right now

And what sort of data do you use to fuel these predictions?

We use largely publicly available datasets. And we’re using billions of data items in our platform to understand how the world works…Macroeconomic data is probably the least reliable data that we use, so we use it for maybe a directional look, at best, at what’s happening. Currencies data is probably the most accurate data that we use, because currencies trade in such narrow bands. We use commodities data, from widely traded ones like oil and gold, to more obscure ones like molybdenum and some industrial metals. We’re also looking at individual equities and equity industries, and we track things like shipping times for goods—shipping times…are usually pretty good indicators of price rises.

Who are your clients, and how are the nowcasts used in practice?

Our clients range from investors and portfolio managers, to healthcare firms and manufacturing firms, to mining and natural resources firms. So they want to understand what the environment looks like for their, say, investment or even procurement—for example, how the current inflation environment affects the procurement of some part of their supply chain.

In fact, we’re talking to a healthcare company right now, and they want to nowcast over the weekend for some of their key materials. In an investment environment, of course, people would want to understand how, say, expectations and other variables impact the outlook for the near-term future, like, days or a week. People are also using us for continuous budgeting—so revenue, budgeting, expenses, CFOs, and heads of financial planning are using us…to understand the 12- to 18-month outlook of their business, [so they don’t have to have an annual budgeting cycle].

Tell me about how the AI works—which kinds of models you’re using, whether you’re using deep learning, etc.

There are basically three phases to our AI. During the pre-process phase, we collect data and look for anomalies, understand data gaps and how data behaves, classify data, and those sorts of things.

Then we go into a forecasting phase, where we use what’s called an ensemble approach: multiple algorithmic approaches to understand the future scenarios for whatever we’re forecasting. Some of those algorithms are longer-term and fundamentals-based, some of them are shorter-term and technical-based, and some of them are medium-term. And we’re testing every forecast item on every algorithm individually and in a common combinatorial sense. For example, we may forecast an asset like gold using three or four different forecast approaches this month, and then using two forecast approaches next month, depending on how the environment changes

And then we have a post-process that really looks at what we’ve forecasted: Does it look weird? Are there obvious errors in it—for example, negative numbers or that sort of thing? We then circle back if there are issues…We’re retesting and re-weighting the methodologies and algorithms with every forecast that we do.

We’ve had very unique market conditions over the past two years. Since AI is trained on data from the past, how have these conditions affected the technology?

You know, there’s a lag. I would say that in 2020, we lagged the market changes by about six weeks. It took that amount of time for our platform to catch up with the magnitude of change that had happened in the markets. Now, back then, we were not iterating our forecasts more than twice a month. Since then, we’ve started to reiterate our forecasting much more frequently, so that the learning aspect of machine learning can really take place. But we’ve also added daily interval forecasts, so it’s a much higher frequency of forecasting and in smaller intervals, because we can’t rely on, say, monthly intervals as a good input in an environment this volatile.

Categories
Editorials

What nowcasts and unique datasets can tell tech about the coming economic shockwave

This article about nowcast is originally published in Protocol.com at this link https://www.protocol.com/nowcasts-forecast-economic-downturn-coronavirus

 

We are living through an economic event with few historical parallels. There is no playbook for shutting down many of the world’s largest economies, nor starting them back up again. But data-mining tech startups are searching out insights in unlikely places, trying to make sense of the global pandemic.

 

These companies are mining specialized datasets, from the prices of beef rounds and chuck, to traffic levels, to the volume of crude oil stored in tanks. Using a mix of machine-learning techniques, they’re spinning this data into “nowcasts”: small, nearly real-time insights that can help analyze the present or very near future. They’re far faster, more granular and more esoteric than the monthly or quarterly data drops provided by the U.S. government. Nowcasts originated in meteorology but are now being applied in economics, and the unpredictability of weather has never been more relevant to the economic outlook.

 

To glean key tech industry takeaways from the coming shifts, Protocol chatted with three data tech startups about the niche datasets they use to analyze economic events and consumer behavior.

 

One of them, Complete Intelligence, has attempted to build a proxy for the global economy that includes market data from over 700 commodities, equity indexes and currencies. Orbital Insight uses global satellite imagery to gather data on large-scale changes in traffic patterns, the business of marine ports, the movement of airplanes, and pings from cell phones and connected cars. And Gro Intelligence specializes in data related to global agriculture: crops and commodities, foreign exchange rates, and the supply and demand of food products.

 

Since these firms tend to shy away from spinning their nowcasts into takeaways (leaving that to their clients), Protocol also enlisted economists to help analyze the data and compare findings with traditional models.

 

Here’s what may be in store for tech over the coming months.

 

Top-level takeaways

 

The U.S. economy was relatively strong going into the outbreak of COVID-19. And that’s a key differentiator between this pandemic and past downturns: This is, first and foremost, a health crisis that’s spilling over into the economy — meaning that how well the economy recovers will depend heavily on what we learn about and how we handle the virus.

 

The wide range of responses to the pandemic — differing by country and, especially in the U.S., by region — mean that economic recovery will likely be protracted and uneven.

 

The U.S. is currently seeing this play out first hand in the way various states have implemented social-distancing measures. Gro Intelligence’s data showed that prices of beef rounds and chuck — which are more prevalent in home cooking — were at all-time highs in March as restaurants shut down across the country. But by using cell phone ping data, Orbital Insight found that things weren’t quite so uniform. It zeroed in on three cities representing three different stages of the pandemic — San Francisco, New York and New Orleans — then measured the percentage of time people stayed within 100 meters of their home each day. During the second half of March, the average resident of New York stayed home close to 85% of the time; in New Orleans, it was around 75%.

 

“When there is uneven distancing, there will be uneven recovery from the health crisis and therefore the economic crisis,” Krishna Kumar, senior economist and director of international research at RAND, told Protocol over email. “This might wreak havoc with cross-state goods, people movement and domestic travel.”

A heat map of San Francisco

San Francisco’s downtown is normally crowded with people, as the yellow areas on this map indicate. But after a shelter-in-place was ordered in mid-March, business districts emptied out.Image: Courtesy of Orbital Insight

 

Combine that with the far-reaching policy rollouts in the U.S. — such as individual stimulus checks, SBA loans and Federal Reserve actions — and there are a host of variables that could make the next few months difficult to predict. The stimulus may help spark a quicker recovery, but that trajectory depends on how long the downturn lasts. Experts agree that too much help could launch another crisis.

 

“A key reason for a more rapid decline in the unemployment rate from the near-term peak is the unprecedented size and speed of the fiscal and monetary response to this adverse shock, which contains measures aimed at maintaining payrolls,” researchers wrote in an April report from Deutsche Bank shared with Protocol, which addresses GDP model implications for the U.S. unemployment rate. The report forecasts the labor market returning to more normal levels of unemployment by the end of 2021 (4.4% by the last quarter of 2021 and 4% a year later), while the protracted scenario suggests the labor market won’t normalize until well into 2023.

 

Corporate debt levels hit an all-time high of $13.5 trillion at the end of 2019, and economists worry that too large a government bailout could spark a default crisis down the road — or even a corporate version of the subprime mortgage crisis.

 

“There’s a danger that we can lend carelessly,” Kumar said. “We just have to be prudent in bailing out the businesses that have future prospects and have returns to show.” He added that after the 2008-’09 financial crisis, banks in China lent heavily and, 12 years later, the time of reckoning might have finally come for those loans. “We can learn from that and make sure that we don’t end up having a state of default.”

 

Complete Intelligence’s algorithms suggest that deflation is likely already happening in China and parts of Europe as a result of COVID-19. But the data also posits that the U.S. may avoid outright deflation. The Federal Reserve has “taken unprecedented steps to inject liquidity — it stands ready to buy even junk bonds,” Kumar said. “These steps are even stronger than the ones implemented during the Great Recession of 2008. At least for now, it doesn’t look like the liquidity pipes are freezing.”

 

Oil storage statistics can also signify broader consumer economic indicators like consumption, and as of April 14, there’d been a 5% increase in crude oil stored in floating-roof tanks around the world over the past 30 days alone. (The startup applies computer vision to satellite imagery to analyze the tanks’ shadows to glean their volume.) While lower prices are good for consumers, they’ll also add to deflationary pressures, according to Kumar — and the U.S. energy sector will take a hit, likely putting a dent in GDP.

 

And a GDP hit likely translates to an impact on the already-growing unemployment rate. Using Okun’s law, a common rule of thumb for the relationship between gross national product and unemployment rate, the Deutsche Bank researchers worked out an updated economic forecast. “Our baseline parameterization,” the researchers wrote, “has the unemployment rate peak at over 17% in April — a new post-World War II high, before falling to around 7% by year end. Under a protracted pandemic scenario, the unemployment rate remains above 10% through all of 2020.”

 

What tech leaders should know

 

For one, expect less pricing power and lower margins. With the businesses shuttering across the country and high unemployment numbers, consumers by and large will have less to spend with. This could lead to supply surpluses, and in the world of tech, electronics manufacturers in particular will need to cut down on production, said Tony Nash, founder and CEO of Complete Intelligence. That will likely hit China, where a considerable amount of tech manufacturing still takes place, hard. As executives calibrate capacity and inventory, production runs will likely shrink alongside pricing power.

 

What happens in the U.S. may not affect a company as much as what happens in the global market. That could be especially true for tech companies with traditionally large sales volumes in Europe and Asia. Complete Intelligence’s machine-learning platform predicts that consumer price indexes in Europe will fall into negative territory later this year, but that deflation won’t hit the U.S. as hard as it will Europe and Asia.

 

“When China shut down, Apple had to shutter many of its stores, and Apple was one of the earliest companies in the country to feel the pain of the virus — because of the global output,” Kumar said.

 

COVID-19’s spread across the globe has come in waves, and that makes it difficult to predict its effect on the global supply chain. But experts say one time-honored strategy remains true: Diversification is key. And individual companies’ rates of recovery may depend largely on how localized their supply chains are.

 

That’s partly due to manufacturing delays that could stem from additional waves of the virus in other countries. But countries’ self-interests also play a role, Kumar said. “After 2008, many countries enacted protectionist measures,” he said. “And if they’re not able to import easily, first it’s going to increase the cost of our imports, and second, we might not even have the local capacity.” For example, there are almost no smartphone and laptop screens manufactured in the U.S.

 

We’ll also likely see tech companies prioritize different geographical supply chain footprints for future generations of products. Alongside this shift, tech giants will also likely take a harder look at which jobs they’re able to automate.

 

“We’re hearing more and more electronics manufacturers moving their manufacturing out of China, and what I’m seeing in data especially — at least for the U.S. — is moving to Mexico,” Nash said. “We don’t expect people to necessarily move their current generation of goods out of China, but as they move to new generations of goods, they’ll look for other places to de-risk those supply chains. So they may have an Asia version of that product that they continue to make there, but they may have regional manufacturing footprints for North America, for Europe and so on, so they don’t have to be as reliant.”

 

The shifts won’t just affect how things are made but also what’s being made in the first place. Necessity is the mother of all invention, as the old adage goes, and there’s a reason why so many side-gig-friendly platforms like Airbnb and Uber sprung from the last financial crisis.

 

And that’s not to mention the overhaul of how we work that many are already experiencing. We may see even traditional companies increase leniency on existing remote work and parental-leave policies, according to Kumar.

 

Conflicting recovery forecasts

 

Predictions of what recovery will look like are akin to trying to predict snowstorms in the summer.

 

Gro Intelligence CEO Sara Menker told Protocol that the U.S. could see a V-shaped recovery, similar to China’s, but that’s more likely the sooner recovery begins. Menker does concede that due to the two countries’ substantially different strategies addressing the pandemic, it’s difficult to know when we’ll be on the up-and-up again. One insight supporting the beginnings of recovery in China: the price of white feather broiler chickens. They’re a breed served almost exclusively in restaurants, and the prices now seem to be entering a V-shaped recovery after a precipitous decline. You can even track it against the reopening of Apple stores: Gro’s data shows white feather broiler prices in China started to rebound around March 6 and a clear price spike around time Apple stores reopened in China on March 13.

 

On the other hand, Orbital Insight CEO James Crawford predicts a more linear recovery, based partly on satellite imagery of roads in China’s urban centers. “In Beijing, for example, we’re not seeing a V-shaped recovery in traffic patterns,” he told Protocol. “It’s been very much a linear return, with less than half the cars on the roads now compared to pre-COVID activity levels. Although the evolution of shelter-in-place was and will be different stateside, businesses should plan for a gradual rebuild in activity as confidence grows among wary consumers.”

 

And, using global economic data like CPIs and predictions surrounding the strength of the U.S. dollar, Nash forecasts a slower recovery. “Whether you’re looking at equity markets or commodity markets, what we’re seeing from our platform is a slow return,” he said. Nash predicts volatility over the next four or five months along with the beginnings of a sustainable uptick in July — though, he said, that won’t necessarily mean a straight upward line, as there are a number of other consumption considerations involved: whether school will start again in the fall, whether football season will be reinstated, whether people can trick or treat in October, whether there are holiday parties in December. “That will define the rate at which we come back,” he said.

 

The true shape of the recovery to come is probably somewhere in the middle, according to Kumar. It’s likely too optimistic to expect a V-shaped recovery, but the more pessimistic prediction — several months of stagnation — “assumes that we can never get a grip on this disease, and given that social distancing seems to be broadly working, I think that’s too pessimistic,” Kumar said. And that’s not to mention the stimulus boost enacted by the federal government. The spark here wasn’t a financial system collapse; it was an economic shutdown. He predicts a more “checkmark-shaped” recovery, with a precipitous drop followed by a less steep, drawn-out upward slope.

 

But rolling back social distancing guidelines too early could sideline recovery as soon as it begins. Some scientists believe the potential impact of colder temperatures on the virus’ spread could lead to a second wave of infections in the fall, and even optimistic projections suggest a vaccine won’t be available until 2021.

 

“The uncertainty that we see in the health care crisis, you’re going to keep seeing in the economy,” Kumar said. “You can get sick very fast, but you’re going to recover much more slowly from your sickness. And that’s what’s going to dictate the economic pattern.”