Most governments in the world have announced, and are implementing, programmes to respond both to the health and economic consequences of Covid-19. However, the magnitude of responses to Covid-19 (including health responses and economic stimulus packages) by governments has varied considerably, from nothing to about half of GDP.[1]
A simple linear regression between the level of support and some possible explanatory factors – the forecast level of impact of Covid-19 on the economy, GDP per capita, and the level of government debt before the crisis – shows that not the severity of the anticipated economic impact Covid-19 but GDP per capita is the most important of these three factors (for more details see below).
In other words, the higher the GDP per person, the higher is the level of Covid-19 measures as a percentage of GDP (Figure 1). The implication of this is that economies and businesses in richer countries receive more support not only in absolute terms but also in relative terms, which in turn means that businesses in developing countries face a high risk of being left behind.
Figure 1: Value of Covid response (in % of GDP) vs. GDP per capita in 2019
 |
Source: Own calculations based on IMF data.
To estimate which factors determine the magnitude of policy responses to Covid-19, a linear regression has been undertaken:
Where
covidmeas = the magnitude of the total Covid-19 response in a country, measured as a percentage of GDP. These values have been obtained from the IMF Policy tracker at https://www.imf.org/en/Topics/imf-and-covid19/Policy-Responses-to-COVID-19 [accessed on 01 May 2020];
covidimp = a simplified measure of the magnitude of Covid-19’s economic impact, as measured by the difference between the forecast GDP growth for 2020 in the IMF World Economic Outlook in April 2020 and the same forecast in the World Economic Outlook in October 2019;
govdebt19 = the level of government debt in 2019 as a share of GDP, from the October 2019 IMF World Economic Outlook database;
loggdppc = per capita GDP (in current USD) in 2019, from the October 2019 IMF World Economic Outlook database.
It goes without saying that the dependent variable covidmeas and also the explanatory variable covidimp are only rough estimates, and therefore, not much effort was spent on the specification of the research design; as such, the results should be interpreted as only a rough indication.
Results are shown in Table 1. The estimation covers a total of 133 observations (i.e. 133 countries); the explanatory value of the equation is modest (R-squared of 0.37). Per capita GDP is the best predictor (significant at the 0.1% level), followed by government debt (significant at the 1% level). Conversely, the magnitude of the Covid-19 impact is not statistically significant. Signs of the coefficients are as expected for GDP per capita (positive, i.e. the higher GDP per person, the bigger the magnitude of the response) and for Covid-19 impact (the higher the expected contraction resulting from the pandemic, the higher the response), and also positive for government debt (i.e. the higher government debt before the crisis, the higher the magnitude of the policy response to Covid-19). This latter observation could be explained by the fact that governments with a high propensity for public spending are also framing their policy response to the pandemic along these lines.
Table 1: Regression results
Source | SS df MS Number of obs = 133
-------------+------------------------------ F( 3, 129) = 24.92
Model | 1990.48172 3 663.493908 Prob > F = 0.0000
Residual | 3435.00861 129 26.6279737 R-squared = 0.3669
-------------+------------------------------ Adj R-squared = 0.3522
Total | 5425.49033 132 41.1021995 Root MSE = 5.1602
---------------------------------------------------------------------------------
covidmeas | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+-------------------------------------------------------------------
covidimp | -.1417267 .1653275 -0.86 0.393 -.4688312 .1853777
govdebt19 | .0402399** .0131416 3.06 0.003 .0142388 .0662409
loggdppc | 2.174681*** .3373617 6.45 0.000 1.507203 2.84216
_cons | -17.06884*** 2.715477 -6.29 0.000 -22.44148 -11.6962
---------------------------------------------------------------------------------
legend: * p<.05; ** p<.01; *** p<.001
[1] It is difficult to assign values to policy measures, and accordingly estimates vary considerably. For example, for Germany, the value of all measures taken together is estimated at more than 60% of GDP by Bruegel (https://www.bruegel.org/publications/datasets/covid-national-dataset/), and 33% by the IMF (https://www.imf.org/en/Topics/imf-and-covid19/Policy-Responses-to-COVID-19) [both accessed on 28 April 2020].