20 Mar 2020 15:38

Forecast update under extreme uncertainty

Håkan Frisén, Head of Economic Forecasting

The COVID-19 pandemic has rapidly changed forecasting conditions. Due to lockdowns around the world, a deep economic recession is inevitable. Our main scenario is that GDP in the 36 OECD countries will fall by 2 ½ per cent in 2020 as a whole; a downward revision by around 4 percentage points compared to our February forecast. The scenario is based on the assumption that the situation will to some extent begin to normalize in a few months. The subsequent recovery will be gradual with a rebound in GDP growth to 3 ½ per cent in 2021. Unemployment will reach levels well above those of recent years. In the light of the great uncertainty, we also present growth paths for two alternative scenarios.

Extreme change of scene in a short period

The widespread lockdown decisions of recent weeks in the Western world have dramatically changed the outlook for economic forecasters. For a fairly long time after the coronavirus outbreak, the predominant view was that its economic impact would be minor, especially outside China. In mid-February the International Monetary Fund (IMF) estimated in its main scenario that the negative GDP impact at global level would only be a marginal 0.1 percentage points for 2020 as a whole. On March 2, the Organisation for Economic Cooperation and Development (OECD) published a more complete forecast update, with a main scenario that lowered global GDP growth in 2020 by 0.5 points compared to its November 2019 forecast. On March 13, when lockdowns started becoming more widespread in Western Europe, the European Commission estimated that European Union GDP will fall by 1.5 per cent in 2020, compared to its earlier forecast that it would increase by about 1½ per cent.

Limited historical experience to rely on

Of course the main reason that estimates of the virus’s economic impact changed so radically in just a few weeks was that forecasters could not imagine the dramatic lockdown measures that have now been implemented. It is also conceivable that influential organisations tended to avoid publishing alarmist forecasts, in order not to increase concerns among the general public and in financial markets − in a situation already filled with such great uncertainty. But now that the authorities have clearly spelled out the situation and lockdowns are obviously affecting economic activity, economists must start trying to quantify the consequences, despite this great uncertainty. It is already apparent that experience from earlier pandemics is of limited value in assessing what economic effects we now face, since they did not lead at all to the type of lockdowns we are now seeing. Of course the recessions of recent decades (the real estate crisis of the early 1990s, the bursting of the IT/dotcom bubble around the turn of the millennium and the Lehman Brothers crisis of 2008-09) were initially fuelled by completely different forces, but they may still teach us lessons about negative reinforcement mechanisms and help us to compare different types of policy responses.

A focus on the virus’s spread and lockdown decisions

Yet we must admit that we are mainly in uncharted territory. So far, the depth of the downturn is not so much due to lack of optimism, reduced capital spending needs or other traditional forces that trigger recessions. Instead the speed of COVID-19’s spread and especially the preventive steps the authorities are taking in an effort to slow it are the main factors behind the economic downturn. It is thus misleading to lean too heavily on historical correlations, for example between sentiment indicators and actual GDP growth outcomes.

Forecasting second quarter GDP growth

In the immediate future, forecasters will focus on trying to estimate how large the negative GDP contributions will be from sectors that are now almost entirely shut down. In the first quarter (Q1), the last three weeks of March will leave their mark, but most indications are that the really dramatic decline will occur during Q2. How deep a downturn we will see then will be crucial to the overall forecast situation and to how dramatic the downturn will be on a full-year basis. At present, forecasters are discussing possible GDP declines in the range of 5-7 per cent, calculated as seasonally adjusted quarter-on-quarter figures. Measured the American way, using “annualised” quarterly figures, we would thus see GDP declines in the range of 20-30 per cent. This is a significantly steeper decline than we saw during the worst period of the Lehman crisis.

The recovery pattern will depend on several factors

What then happens after this deep economic downturn will depend on a variety of factors. The future spread of the COVID-19 disease will naturally be important, but public sector decision makers and political leaders will continuously face decisions about the scale of their lockdowns. Although medical considerations will remain crucial, it will be difficult not to also weigh in the consequences of economic and social developments, which include increasing risks of breakdown and collapse in some areas. These, in turn, might influence the chances of achieving a rapid expansion in health care capacity. If the rate of COVID-19’s spread slows as the weather becomes warmer, and then rises again in the autumn, for example, policy makers may very well choose a strategy different from today’s. 

Rapid economic policy responses

In the past week, economic policy makers have reacted with speed and determination. Both governments and central banks have realised that the situation is serious and have been keen on wasting no time. The US Federal Reserve quickly slashed its key interest rate to zero, and large-scale quantitative easing (QE) programmes have been launched on both sides of the Atlantic. In this respect, the situation is completely different from the Lehman crisis. At that time, there was a more or less explicit unwillingness to act in a way that would benefit the market actors that had practised various kinds of excesses, especially in the US mortgage loan market. The period between the first serious signs of crisis in the sub-prime market in August 2017 until the Lehman Brothers bankruptcy in September 2008 was dominated by a balancing act between trying to keep the financial system going as much as possible while avoiding major rescue actions. Only in October 2008 did the authorities launch powerful responses as the financial system teetered on the edge of total collapse. Since the current economic crisis is a direct consequence of drastic medical and political decisions, there is no equivalent dilemma, at least not initially. 

Stimulus measures being blunted

Although today’s vigorous actions may be impressive, there are many unanswered questions about the effectiveness and purpose of such stimulus policies. In several of our research publications, we have tried to compile the actions of both central banks and governments. One fundamental difference compared to earlier crises is that both the demand and supply sides of the economy are highly constrained right now. When international borders as well as shops and restaurants are closed, important areas of potential household consumption disappear. On the supply side, many other sectors are suffering, for example because their employees have difficulty working from home. Many stimulus packages have also been designed to make it easier for employees to stay home. These measures thus indirectly contribute to pushing down GDP in the short term. We are now also beginning to see major manufacturing companies halting production, as closed borders and lockdowns prevent the necessary importation of input goods and workers. In such a situation, expansionary fiscal measures can soften the short-term GDP decline only to a limited extent, regardless of how big a dose of stimulus we may formally add to the economy. The purpose of many stimulus measures is instead to decrease long-term detrimental effects by enabling businesses to retain their employees or ease the consequences to those who temporarily or permanently lose working income.

Weak monetary policy transmission

Central bank actions have been highly important in stabilising the financial system, but here too there are major constraints on their ability to influence economic activity in the short term. Such constraints are not always so evident when these measures are being unveiled. Loan facilities of the kind that Sweden’s Riksbank announced on March 16, totalling SEK 500 billion, can help finally stressed businesses to survive. But in order for businesses to be genuinely ready to take out new loans, they need to see some light at the end of the tunnel in terms of demand for their products. And the credit risk still rests with commercial banks, which thus retain their traditional role as credit evaluators.      

Bigger economic policy tensions on the way

Despite quick responses by governments and central banks, so far they have clearly delivered relatively uncontroversial measures. Looking ahead, they will probably be forced to take actions that require more difficult trade-offs and whose drawbacks will also be more obvious. This may include targeted subsidies that favour some economic sectors and companies over others and – for example at the euro area and ECB level − favour some countries over others. We will probably also see more obvious interventions in the market economy, with governments and central banks taking over credit risks in some areas. Current policy frameworks and limitations on mandates will probably need to be temporarily reassessed by both budget and central bank policy makers.

Traditional macroeconomic issues will reassert themselves

Although our short-term analysis focuses entirely on the impact of all the lockdown measures, fairly soon we will also have to consider more traditional secondary effects. Unemployment will also increase a bit further ahead, affecting consumption and the housing market. The stock market decline will also have a dampening effect on consumption for some time. The need for capital spending will decrease in some sectors, and many businesses will have worse financial potential to expand their capacity. Overall, this means that the GDP level over the next couple of years will remain lower than estimates that were made only a few months ago.

Sharp GDP decline in 2020, but rebound in 2021

Our main scenario implies that the level of GDP will fall somewhat during the first quarter of 2020, then plunge more than 6 per cent (about 25 per cent at an American-style annualised rate). This projection assumes that lockdowns and restrictions will successively ease as summer approaches and that the economy will begin to recover in the middle of the second quarter. Then there will be a gradual recovery during the second half of the year and in 2021 (see chart). Given such a scenario, annual averages will be dramatically different from our earlier forecasts. We expect GDP in the 36 mainly affluent OECD countries to fall by 2.6 per cent in 2020 as a whole: a downward adjustment of more than 4 percentage points compared to the February issue of Nordic Outlook. We are instead revising our GDP growth forecast for 2021 2 points higher. This implies that the full-year 2021 GDP level will be about 2 per cent below our previous forecast, due to the negative after-effects discussed above. Despite the efforts now being made to enable businesses to retain employees, the jobless rate will rise significantly in the coming year. In the United States, unemployment will climb from today’s 3½ per cent to around 7-8 per cent a few months into 2021 and then begin to fall. In Europe, the upturn will be somewhat milder, in keeping with historical patterns during recessions. We expect Swedish unemployment to climb to 10 per cent.

GDP growth in our main scenario

Y-o-y change, %. Difference compared to NO February

 

Forecast 2020


2021

Difference
  2020


2021

Sweden

-2.7

3.4

   -3.8

+1.7

Euro area

-3.6

3.4

   -4.7

+2.2

United States

-1.7

3.7

   -3.5

+1.8

OECD

-2.6

3.6

   -4.1

+2.0

World

 0.5

4.5

   -2.6

+1.2

Source: OECD, SEB

Similar situation in the US and Western Europe

At this early stage it is misleading to project differences between countries in excessive detail, among other things because these figures will greatly depend on the course of the pandemic. At present, however, our assessment is that GDP will show its deepest decline in the euro area, since this region has been hardest hit by COVID-19 and has implemented the most extensive lockdowns. However, because the spread of the virus is lagging in the US, this picture may change. Weaker automatic stabilisers in the American economy – due to such factors as a less far-reaching social insurance system – normally make the economic impact of an epidemic larger there.  However, we are likely to see new types of federal programmes that will change this pattern. This is one reason why we do not believe that US unemployment will climb to levels as high as historical experience indicates (see table below).   

Changes in GDP and unemployment

Per cent and percentage points, US

 

GDP*

Unemployment**

Early 1980s

-2.1

          2.0

Early 1990s

-1.4

          2.3

Financial crisis

-3.9

          4.9

Corona virus scenario

-5.8

          5.5

Source: SEB, Macrobond

* Accumulated top-to-bottom change in level, per cent
** Upturn measured as quarterly averages, percentage points.

Sharp slowdown also in EM economies

At present there are various reasons to believe that emerging market (EM) economies will experience a slowdown roughly similar in magnitude to what they saw after the global financial crisis. The spread of COVID-19 in China appears to have slowed very sharply, and economic activity has started to recover; a hopeful sign that EM growth will remain in positive territory this year. However, we are likely to see a dramatic slowdown in many export dependent economies. A sudden stop of non-resident portfolio investment inflows, a sharp fall in commodity prices and a collapse in tourism will have a severe impact on EM.

Downside risks related to new lockdown decisions

Given the exceptional uncertainty that now prevails, it is natural to work with various alternative scenarios. Different patterns are conceivable when it comes to the spread of the virus, economic policy responses and the actual economic consequences of these factors.  Our negative scenario assumes that the spread of COVID-19 will be so serious that lockdown decisions are extended for significantly longer than our main scenario assumes, remaining in place at least until the end of June. In that case, the economic downturn will be significantly deeper and the decline in GDP compared to our February forecast will be as much as 10 percent at the end of 2020. In this scenario, economic policy makers will need to take dramatic measures that we have not seen so far in peacetime. Even now, for example, the US has already introduced wartime-style legislation that widens the range of potential public sector responses. Yet we still see reasons not to project an alternative that includes an overly catastrophic economic trend, since there will probably still be actionable options to avoid this.

Better growth potential when politics start to work

A more positive scenario can be based partly on a faster-than-expected improvement in the spread of infection, or an earlier easing in lockdown restrictions than in our main scenario. A bit further ahead, it is also conceivable that the recovery will be faster than in our main scenario, partly because economic policy stimulus measures that have had very limited short-term impact have begun to help significantly more once the general social environment has been normalised. In such a scenario, it is difficult to expect governments and central banks to be particularly fast in withdrawing their stimuli out of concern about overheating. We foresee a roughly symmetrical probability for our two alternative scenarios. At this stage, our assessment is that a weaker trend is slightly more probable than our positive scenario. We estimate that the probability of our main scenario is 60 per cent and that our positive and negative scenarios each have a 20 per cent probability.