Prêmio CBMM
Imprimir Republish


Calculating in the dark

Creativity helps scientists predict the economic impacts of an unprecedented crisis

Léo Ramos Chaves

When the economy suffers a shock, a standard set of models and historical precedents are usually available to answer economists’ questions. How intense will the crisis be, and how long will it last? Which sectors are being affected most? What is happening to the labor market and the principal economic variables, such as the exchange rate, Gross Domestic Product (GDP), and inflation?

The economic crisis brought on by the pandemic, however, is different. It’s not being caused by overproduction or a crash in financial markets, or a sudden change in any fundamental cost index, such as the exchange rate, interest rates, or oil. This time, an as-yet poorly understood disease has led governments around the world to suspend most economic activity for periods that could exceed two months. And estimating the reaction of economies to this new event has been an exercise in creativity.

Calculating GDP is usually relatively uncomplicated. The total amount of household consumption of goods and services is summed with corporate investments, government disbursals, and total nationwide exports and imports during a given period. By knowing how the economy has behaved previously and the trajectory of primary economic variables such as interest and foreign exchange rates, the performance of principal trading partners, and current prices of exports, among other factors, it is possible to predict the country’s production and consumption over subsequent months and years.

The same predictability isn’t possible when economic activity is suspended. “Usually, in major crises, the drop in demand goes to 4% or 5%. In this crisis, there are people talking about 40% or 70%. There’s no parallel. The level of uncertainty for any forecast this year is enormous,” says José Ronaldo de Castro Souza Júnior, director of Macroeconomic Studies and Research at the Institute of Applied Economic Research (IPEA).

One difficulty is the pandemic itself. How long it will last? How much funding is being allocated to medical equipment? How much work is still getting done while people are being kept at home? “When making short-term estimates, it’s not possible to work normally. Even the numbers that come from epidemiology are uncertain, because small variations give very different results. Any uncertainty indicator that I choose for my model is dubious,” laments Souza Júnior. “Right now, I feel more comfortable making long-term forecasts, for 10, 20, 30 years,” he says.

IPEA specialists held “a series of online meetings to understand what could be done from a technical point of view,” the economist recalls. One difficulty is that the crisis has two key moments. In the first, “supply shock,” production is interrupted, which is what occurred in January and February in China, causing a lack of Chinese parts and supplies in industries around the world. “This part we already understand. The problem is with the second key moment, the demand shock,” he notes, referring to the interruption of consumption during social isolation. “We’re using models that we don’t usually use, which analyze the economy sector by sector. The difficulty lies in understanding the extent of the supply shock, and when the demand shock begins. The economic recovery will depend a lot on the behavior of demand,” Souza Júnior summarizes.

Another difficulty lies in obtaining the data that such projections are based on. Most of the numbers come from the Brazilian Institute of Geography and Statistics (IBGE). The primary source is the Continuous National Household Sample Survey (Continuous PNAD), which is published monthly with numbers from the previous quarter. During the pandemic, the institute has been forced to make a series of adaptations. The first was to postpone the Census to 2021, followed by canceling PNAD home visits.

IBGE replaced the visits with phone calls, using various databases it already had. In May, the PNAD-COVID was launched in partnership with the Ministry of Health. They surveyed 193,600 households, mainly on questions regarding the disease, but issues related to work and family income were also included. “We decided to end face-to-face data collection on March 17, and since then we’ve begun a discussion with our technicians about continuing the research, as well as eventually developing other products,” stated IBGE research director Eduardo Rios-Neto at the launch of the initiative. According to the institute’s National Accounts coordinator, Rebeca Palis, the response rate to questionnaires dropped in March. “Even so, we’ve determined that the data needed to feed into the national accounts for the first quarter are adequate,” she says.

Bruna Prado/Getty Images Wearing facemasks and observing social-distancing rules, people seeking emergency assistance line up in front of the Caixa Econômica Federal bank branch in Rio de JaneiroBruna Prado/Getty Images

Sociologist Ian Prates, who conducts research at the Brazilian Center for Analysis and Planning (CEBRAP) and at Social Accountability International, adds that the interruption of normal research procedures is being called a “data blackout.” The pace of data production has changed, the basis of comparison is imperfect, and in the end, some numbers aren’t being produced. “Even though there’s no official data we can’t just cross our arms and wait for the pandemic to end,” he observes.

“We’re using data that we would never use under normal circumstances,” says Prates, referring to data on consumption and circulation collected with the help of companies like Google and Cielo, and data regarding the business climate obtained with the assistance of the Brazilian Micro and Small Business Support Service (SEBRAE). “These are the data that are available, but they are also the data that make sense at the moment. If we’ve paid less attention to them before, it may be due to scientific routine,” he speculates.

“There are some types of data we didn’t use to work with, but which we are working with now because our objective is near-term activity, and not just understanding larger social processes,” adds sociologist Rogério Barbosa from the Center for Metropolitan Studies (CEM), one of the Research, Innovation, and Dissemination Centers (RIDC) supported by FAPESP. One example is tracing people’s geolocations through their cell phones, in partnership with the company Inloco.

Barbosa and Prates are responsible for research on income, the labor market, social protection, and emergency policy for the Solidary Research Network, created in March to study the pandemic. The network brings together researchers from different disciplines and institutions of higher education, publishing newsletters with analyses on the evolving pandemic in Brazil and its effects on society and the economy.

Despite the difficulties, projections began coming out in the last week of March. IPEA estimated a drop in Brazilian GDP that ranged from 0.4%, if social isolation lasted until the end of April, to 1.8%, if it lasted through June. The projection was published in the Carta de Conjuntura (Economic Outlook Brief), issue 46. In April, the institute launched other, separate projections, in advance of issue 47. All the texts emphasized the speculative nature of the forecasts, announcing that the initial figure of 1.8% had been exceeded.

In the first week of April, the United Nations Department of Economic and Social Affairs (DESA) estimated that world GDP would fall by almost 1%—the previous forecast had been for an expansion of 2.5%. The International Monetary Fund (IMF), in turn, estimated the drop at 3%. The World Bank went even further, projecting a reduction in GDP of 4.6% for Latin America and 5% for the Brazilian economy.

The Bank for International Settlements (BIS) also published, in April, a compilation of projections for hypothetical pandemics based on previous events such as the Spanish flu, which infected approximately 500 million people from 1918 to 1920. Estimates range from a decline in the global economy of less than 1% this year to drops of more than 4%. The economic impacts of the Spanish flu itself, which killed between 50 million and 100 million people according to calculations by the US Center for Disease Control and Prevention (CDC), are unclear. Nonetheless, in March a group of economists led by Robert Barro from Harvard University calculated that from 1918 to 1920, the decline in world GDP was between 6% and 8%.

Still, past health crises do not provide instruments for predicting this year’s economic performance, laments economist Emerson Marçal, coordinator of the Center for Studies in Applied Macroeconomics at the São Paulo School of Economics of the Getulio Vargas Foundation (EESP-FGV). Recent pandemics, such as SARS, in 2003, and the H1N1 influenza in 2009, had either a more limited reach, or a far less intense global effect than COVID-19. The Spanish flu, on the other hand, did have global impacts, but because the event occurred 100 years ago, “the data are nonexistent or insufficient,” says Marçal. “At that time, macroeconomic series were not calculated very well. We only have estimates, which are well done, but they don’t allow us to model what happened so long ago,” he explains.

Léo Ramos Chaves In São Paulo, Rua José Paulino is the heart of the garment district, attracting retailers from all over Brazil. Nonessential trade was suspended due to the pandemicLéo Ramos Chaves

Marçal and his team based their research on two events in recent economic history to calculate, in late March, the pandemic’s effect on Brazilian GDP. As an international influence, they chose the 2008 financial crisis, which interrupted capital flow and caused a drop in GDP in 2009. As a reference for a domestic economic freeze, the researchers used the May 2018 truckers’ strike—which blocked highways throughout the country for 10 days.

The resulting projection, which was presented at the beginning of April and had broad repercussions in the media and financial markets, was for a contraction that could reach 4.5% this year. “At the time, I had the impression that the number we came out with was too pessimistic. Now I’m revising it with [actual] data that are just now starting to be collected, and it seems to me that I sinned by being excessively optimistic,” Marçal states. The domestic factor ended up having more weight than the international factor, since the Brazilian economic suspension is lasting longer than the truckers’ strike. “If I had made a model with a longer suspension [of economic activity], it would have resulted in a weightier number,” he acknowledges.

Gradually, market and government forecasts are converging on numbers similar to those obtained by Marçal. In early March, the Ministry of Finance was still forecasting modest growth for 2020. The Focus bulletin, a weekly publication by the Central Bank that puts together projections from financial institutions regarding Brazilian economic variables, detected only a small contraction. But by mid-May, Focus was already showing a 4.4% retreat and the Finance Ministry had indicated a drop of 4.7%.

These new projections already incorporated the first official data from IBGE. In early May, the institute released the pandemic’s effect on industry in March, showing a 9.1% drop from February numbers. Results for the service sector released the following week were also devastating, with a decrease of 6.9%. According to the Brazilian Central Bank, the Economic Activity Index (IBC-Br), considered a “preview of GDP,” was down by 1.95% in the first quarter compared to the fourth quarter of last year. In March alone, the drop in the index was 5.9%. First-quarter GDP, according to the IBGE, shrank 1.5%. “And the first quarter only felt the effects of the pandemic for two weeks,” recalls Marçal.

Much of the pandemic’s economic impact, in the long run, will also depend on the effectiveness of government measures aimed at preventing bankruptcies, layoffs, and the loss of purchasing power for basic items. In Brazil, these measures include reducing the compulsory reserves held by financial institutions at the Central Bank, reducing the base interest rate (SELIC), paying emergency aid of R$600.00 to the laid-off employees, informal workers, the self-employed, and micro-entrepreneurs (MEI), and authorizing reduced working hours and wages, among other initiatives.

Souza Júnior at IPEA says it’s not yet possible to calculate the effect of government aid to companies and individuals, except as a counterfactual comparison. “Without these measures, we could be suffering social chaos. If people are restricted to staying at home and at the same time see no prospect of earning money for basic purchases, it would be a bomb that could result in looting and rioting,” he says.

In early May, the team from the Industry and Competitiveness Group of the Institute of Economics at the Federal University of Rio de Janeiro (IE-UFRJ) published a projection for diverse segments of the Brazilian economy, in which activity levels in some sectors are reduced by as much as 10%. This applies to construction, extractive industries (mining, for example), and processing (such as consumer goods factories), in addition to trade.

The researchers worked with three scenarios, using the Input-Output Matrix, a methodology that allows each sector of the economy to be evaluated separately. In the optimistic scenario, which is already obsolete, social isolation would be short, but efficient: coupled with effective government measures to avoid bankruptcies, recovery would be swift. Even so, the decline in GDP would be 3.1%, with the employment rate falling by 4.4%. In the pessimistic scenario, social isolation measures and actions taken to mitigate the pandemic’s economic effects are ineffective. The decline in GDP reaches 11%, with a catastrophic 14% drop in employment.

Léo Ramos Chaves Avenida Paulista, empty at the end of May: the effectiveness of government measures taken to avoid bankruptcies, layoffs, and loss of purchasing power will be decisive in minimizing the effects of the pandemicLéo Ramos Chaves

Between the two is the base scenario, in which social isolation is prolonged and measures to contain the virus take time to take effect, but the situation in Brazil avoids becoming a disaster. In this case, GDP would fall by 6.4% while employment would decrease by an impressive 7.9%.

The manufacturing sector, both in the consumer goods and capital goods industries, could fall by 11.3% (base scenario) or drop as much as 18.8% (pessimistic scenario). “There’s a lot of talk about the decrease in the service sector, but we’re also going to see an important retraction in demand, because income, exports, and investment are also in decline. Industry will be even more affected than services,” says economist Esther Dweck, a professor at IE-UFRJ, and coordinator of the study. “Even if, thanks to agricultural products, exports don’t drop, manufactured goods will lose ground in foreign markets. This will exacerbate the country’s deindustrialization problem,” she predicts.

To better understand the pandemic’s effect on industry, in early April economist Cristina Fróes de Borja Reis of the Federal University of ABC (UFABC), working with a researcher at Technische Universität in Berlin, Germany, interviewed 32 entrepreneurs in the machinery and equipment sectors from different regions of Brazil. They confirmed the heavy impact the pandemic was having on their businesses. According to Reis, 53% of these companies were hit seriously or very seriously, and some spoke of imminent risk of bankruptcy.

With the advancing pandemic, international supply chains have been disrupted and contracts revoked, domestic and foreign demand has fallen, and trade fairs have been canceled. Companies have also been hampered by the inability of their sales teams to physically meet with customers, the economist adds. For industrial sectors that depend on imported raw materials, the recent devaluation of the Brazilian real against the dollar was also bad news.

“This doesn’t mean that industry will simply die, because it can modify its activities, but it’s very bad for the more complex and innovative companies,” says Reis. “What could happen is that a process that was already underway will accelerate: Brazilian industries are becoming ever more dependent on foreign technology, with multinational corporations doing only the assembly stage in Brazil.”

Both the size of the crisis and the shape of recovery are unknowns. On the international stage, the World Economic Forum has compiled various hypotheses. Returning to economic levels expected before the crisis within just a few months, the so-called “V recovery,” is increasingly unlikely. A “Z” recovery, with a brief period of accelerated growth capable of compensating for the months in decline, before returning to the baseline trend of previous years, also seems unlikely to the Forum analysts.

In China, the first country hit by the COVID-19 pandemic, recently published data indicate a slow return to production levels seen prior to the abrupt drop in GDP. This is what’s called a “U-shaped recovery.” The projections of UFRJ, IPEA, the FGV, and the Ministry of Economics are primarily working from the hypothesis of a U-shaped recovery in Brazil. In the alphabet of economic shock, there are two possibilities that are even more worrying. In an L-shaped recovery, economic activity remains at lower levels for much longer than it normally would. In a W-shaped scenario, a brief surge is followed by a second recession.

With the diverse projections and uncertainties involved, the effort to anticipate data and predict how the crisis will unfold over the weeks and months ahead may seem like a fruitless exercise, since gross errors will be unavoidable. Researchers, however, reject the notion of abdicating predictions. Forecasts are indispensable if government and the private sector are to devise strategies for responding to the economic shock of the pandemic.

“Our objective when drawing up projections is to guide public policy actions. If a given projection underestimates or overestimates a bit, it doesn’t matter. What matters is that it’s a good tool for action,” observes Barbosa, of CEM. Souza Júnior, from IPEA, takes a similar view. “We need to continue doing these projections because we have to be prepared, and know what to do, depending on which path the economy takes. What will the impact be on revenues, the primary surplus, on unemployment? On the size of the national debt? What fiscal and monetary policies will be adopted? What can be done in terms of social policies? It all depends on the size of the hole and the pace of recovery,” he says.