Estimating the size of GDP loss in Southeast Asia due to the COVID-19 pandemic

ABSTRACT

This Research Note explores how the major Southeast Asian countries were hit with negative growth rates due to the pandemic in 2020. National income was markedly reduced to a lower level and, even with present growth rates back at pre-pandemic levels, there will be permanent effects. The level trajectory for GDP will necessarily be lower compared to a non-COVID scenario with positive economic growth. Exponential smoothing is applied to GDP data for 2000–2019 with the purpose of estimating incomes for a non-COVID case and the deviation from the actual GDP in order to measure the loss of incomes. For the pandemic years, the negative impacts are most severe for countries with relatively large service sectors (Malaysia, Thailand) and less so for countries with a diversified economy or a large agricultural sector (Vietnam, Laos).

KEYWORDS:

• COVID-19
• GDP
• forecasting
• economic modeling
• economic recovery

Introduction

The COVID-19 pandemic hit the Southeast Asian countries with severe consequences for incomes and in many cases also worsening of health and general living conditions. This Research Note explores how although the virus caused similar problems throughout the region, the impact on economic growth was not of the same magnitude among these countries and likewise the timing of the economic slowdown differed. Even though all of the countries in the region imposed restrictions in order to limit the detrimental effects of the pandemic some countries, like for example Cambodia and the Philippines, suffered larger GDP losses compared to neighboring countries like Indonesia and Vietnam.

When it comes to quantifying GDP loss, this topic can be dealt with in terms of lower economic growth rates in a certain time period or as the permanent – or cumulative – loss of GDP level. During the years 2020 and 2021, all of the Southeast Asian countries experienced significantly lower growth rates compared to the preceding years and, hence, an estimate of the loss can be calculated from such data.

Now, GDP growth rates are nearly back to their former levels, but due to the negative impact of low growth rates during the pandemic GDP is now on a lower trajectory or path compared to a scenario with no COVID-19. During 2020–21 there was a severe drop in the GDP level for some of the countries, and they are not catching up to the income levels that would have taken place in case of a non-pandemic scenario. The latter approach is applied in the present analysis where GDP data from the decade up to 2020 is used to forecast potential GDP levels for the 2020–2022 period. The forecasted values are compared to actual GDP during the pandemic years in order to assess the loss of national income.

The slowdown in economic growth due to COVID-19 was a global issue, and in relation to the OECD the early study by Martinho (2021) – including also some non-OECD members – shows that former signs of convergence with respect to GDP were eliminated, and also that more fragile economies suffered more due to the pandemic. Negative impacts in relation to a range of social and sustainability issues for Southeast Asia are part of the pandemic as reported by Suriyankietkaew and Nimsai (2021). An assessment from April 2021 by the Asian Development Bank (ADB - Sawada and Sumulong 2021) estimates the contraction of GDP to be 0.4% in 2020 with a partial recovery for 2021 with higher growth rates for developing Asian countries.

This early forecast turned out to be too optimistic which will also be revealed in this analysis. The study by Gagnon et al. (2023) deals with a global analysis including quarterly GDP data for 90 countries for the period 2020Q1 to 2021Q4. They find the stringency of lockdown actions to negatively influence GDP, but also that the restrictions were most injurious to emerging and developing countries. Additionally, global trade was a significant channel for the spill-over of negative economic effects across countries. Therefore, countries like Malaysia, Thailand, and Vietnam with high shares of export of goods and services are most likely to be affected by the global slowdown in demand (Suvannaphakdy 2021).

The restrictions imposed with the purpose of curtailing the negative impacts of the pandemic have been quantified by the Oxford COVID-19 Government Response Tracker project (published June 2023). A part of the project is to calculate a Stringency Index (SI) (https://ourworldindata.org/COVID-stringency-index) based on nine different metrics or variables including workplace closures, stay-at-home restrictions, and travel controls. Additionally, an index measuring containment and health (CHI) is calculated from workplace closures and travel bans as with the SI index, but also including testing policy, contact tracking, and face coverings.

Both indices range from 0–100 and cover 2020–2022. For the respective countries, the indices have similar values at specific points in time. Restrictions were imposed in the spring of 2020 and at the end of the year the indices show values in the range of 50–75 for all countries except Laos, where more restrictions were imposed during the first part of 2021. Thus, governments seemed to react by implementing more or less identical restrictions to fight the pandemic, but the detrimental effects to the economies have been very heterogenous, and that is most likely due to differences in economic and demographic structures.

The next section presents the method and data sources and afterward the results from the analysis are presented. The Conclusion sums up the main findings from the analysis.

Method and data

The data included in the analysis are from the most often used international statistical agencies such as the World Bank (The WDI database) and the Asian Development Bank. Both data sources report GDP for Southeast Asia, but the WDI database is the most recently updated with values for the time span of 2000 to 2022 for all of the eight countries included in the analysis. Hence, the data from the WDI is applied, and GDP is measured in local currencies and constant prices in all cases. The focus of the analysis is not a cross-country comparison of GDP levels – in which case measurement units might be US$ or PPP rates – but is instead about the loss of real domestic income due to the growth-slowdown during the pandemic. The analysis will focus on the total GDP instead of per capita values as the time span is very short, but the results and conclusions would be similar in case of investigating the topic with GDP per capita.

The countries included in the analysis are Cambodia, Indonesia, Laos, Malaysia, Myanmar, Philippines, Thailand, and Vietnam. Thus, smaller countries like Brunei, East Timor, and Singapore are not part of the analysis. The WDI data has been compared with the GDP data from the statistical agencies of the aforementioned eight countries – and the ADB data – and it seems there are no larger discrepancies between the alternative data sources.

The approach in the present analysis is to estimate the loss in levels of GDP as mentioned in the introduction. With negative growth rates during the pandemic, the GDP level, measured in constant prices, will not be back on the former trajectory unless growth rates are increasing compared to the years before the pandemic. There must be a catch-up process to get GDP back to the levels expected in forecasts made before the pandemic.

The other approach, comparing GDP development in growth rates, reveals that the Southeast Asian countries by now are nearly back to similar growth rates as experienced before 2020. An ERIA, ASEAN (2022) report exhibits a graph with growth rates in the span of 2–7% annually for the Southeast Asian countries before the pandemic, and the growth rate forecasts for 2023–2025 are back in the same range. From the recent WDI data for real GDP, the actual growth rates are calculated and exhibited in Table 1 and with an ADB forecast for 2023.

Table 1. Annual growth rates of real GDP 2015–2019 and 2020–2022%).

	2015–2019	2020	2021	2022	2023 (ADB)
Laos	6.9	0.5	2.5	2.7	4.0
Indonesia	5.1	−2.1	3.7	5.3	4.8
Vietnam	6.8	2.9	2.6	8.0	6.5
Cambodia	7.1	−3.1	3.0	5.2	5.5
Myanmar	7.0	3.2	−17.9	3.0	2.8
Thailand	3.0	−6.1	1.5	2.6	3.3
Malaysia	5.0	−5.5	3.0	8.7	4.7
Philippines	6.6	−9.5	5.7	7.6	6.0

Annual growth rates for the total GDP calculated in local currencies and constant prices. Data from WDI; for 2023 forecast values from the ADB.

Sources: World Development Indicators, The World Bank (data accessed 30 June 2023). GDP Growth, Asian Development Outlook – April 2023, Asian Development Bank.

Thus, the economic growth rates seem to be close to the level from 2015–19 for six of the eight countries, where Laos and Myanmar are still lagging behind in this respect. Vietnam is obviously the country with the best performance in relation to declining growth during the pandemic.

The WDI data for the development in GDP levels since 2010 (indices, constant prices) is exhibited in Figures 1 and 2 and reveals a more permanent loss of GDP level as found when only considering the development in growth rates.

Figure 1. Real GDP 2010–2022 for Cambodia, Indonesia, Vietnam and Laos (index 2010 = 100, constant prices in local currencies).

Source: World Development Indicators, The World Bank (data accessed 30 June 2023).

Figure 2. Real GDP 2010–2022 for Myanmar, Thailand, Malaysia and the Philippines (index 2010 = 100, constant prices in local currencies).

Source: World Development Indicators, The World Bank (data accessed June 30, 2023).

From Figures 1 and 2 it is obvious that the pandemic did not impact the economies in a similar manner. The countries in Figure 1 are the least affected by the negative GDP impacts, and in Figure 2, Myanmar is doing well in 2020, but has a huge drop in GDP level in 2021. The latter is most likely caused by the military coup on 1 February 2021 and is only to a smaller degree caused by the pandemic.

The methodology used in the estimation of the GDP (level) loss is based on the exponential smoothing technique. This means estimating a model for the time-series GDP data where exponentially decreasing weights are assigned to past values of the variables. Let y_t be the smoothed value of a series x_t, i.e. being calculated from lagged values of x_t with exponentially declining weights, where the final recursive formula for calculating y_t will be:

(1) ${\mathrm{y}}_{\mathrm{t}}=\mathit{\alpha }{x}_{\mathrm{t}}+\left(1-\mathit{\alpha }\right){\mathrm{y}}_{\mathrm{t}-1}$(1)

A trend can be included in the model as:

(2) ${\mathrm{r}}_{\mathrm{t}}=\mathit{\gamma }\left({\mathrm{y}}_{\mathrm{t}}-{\mathrm{y}}_{\mathrm{t}-1}\right)+\left(1-\mathit{\gamma }\right){\mathrm{r}}_{\mathrm{t}-1}$(2)

The smoothed series will be y_t + r_t and an n period forecast computed as y_t+n = y_t + nr_t. The software used for the estimation procedure is RATS (Estima.com), and the smoothing parameters (α, γ) are selected in order to minimize the sum of squared errors. This method allows for a close fit in relation to the historical data, and the model is subsequently used for forecasting the immediate future which will be rather close to the recent past values just before 2020.

The WDI data from 2000 to 2019 is applied to the exponential smoothing, and afterward forecast values for 2000 to 2022 are calculated. These will be non-COVID forecasts and are assumed to represent an estimate of the GDP in the case of no pandemic. The deviations from the smoothed forecast values and the actual GDP during the pandemic 2020 to 2022 can easily be calculated and represent an estimate of the GDP loss in level values.

Results

The GDP loss calculations are exhibited in Table 2 where the numbers (per cent) represent the deviation between the actual GDP and the forecast (non-COVID) values. The calculation for 2023 is based on two forecast values. The 2000–2019 data are used to forecast the 2023 value, and afterward the actual GDP 2019–2022 is used for forecasting the year 2023. The latter might be a reasonable estimate of the actual 2023 GDP as data from the pandemic years are applied. Then the deviation or loss can also be calculated for 2023.

Table 2. Deviation of real GDP from the non-COVID forecast values (per cent).

	2020	2021	2022	2023	2022 (ADB)
Laos	−4.7	−7.2	−9.5	−11.8	−14.3
Indonesia	−7.1	−8.7	−8.9	−9.2	−9.0
Vietnam	−4.0	−8.0	−7.2	−6.9	−6.4
Cambodia	−9.6	−13.2	−15.0	−16.7	−15.0
Myanmar	−0.6	−21.6	−22.3	−23.0	-
Thailand	−7.4	−8.4	−8.3	−11.3	−10.8
Malaysia	−9.8	−11.3	−8.1	−4.7	−11.8
Philippines	−15.1	−15.7	−14.8	−13.9	−16.0

The calculation in the ADB report is done based on the deviation between the actual GDP and the estimate from a trend growth model based on 2009–2019 GDP data.

Sources: World Development Indicators, The World Bank (data accessed 30 June 2023). Southeast Asia Rising from the Pandemic, Asian Development Bank, March 2022.

The first four columns in Table 2 are based on the WDI data and calculated with the exponential smoothing technique. The last column is from the Asian Development Bank where a trend growth model is applied to the ADB data and the deviation is measured like the other results in Table 2.

If the smoothed forecast values are assumed to be the counterfactual GDP levels, i.e. in case no pandemic took place, then the deviations represent annual losses and the cumulative sum of lost GDP is very high in all cases. This is probably an overestimation of the COVID-19 detrimental effects and no one knows what would be the alternative GDP developments in the case of no pandemic. The numbers in Table 2 reveal large downturns in countries like the Philippines and Cambodia, and they suggest that the negative effects are not confined to the 2020–2021 pandemic years. The results are also in line with other reports on the COVID-19 hitting Asian and developing countries relatively hard compared to other regions (cf. IMF 2022).

There is most likely not a simple explanation of the heterogeneity in output losses across the countries in Table 2 and as earlier mentioned there were no significant differences in policy restrictions during the pandemic. Explanations may be related to the specific structures of the economies, where service sector activities (travel, tourism, and transport) were immediately affected as revealed by the negative 2020 growth rates for Malaysia, Thailand, and the Philippines (Table 1).

More diversified economies – like Vietnam and Indonesia (which, for example, maintain an export of petroleum products) – and economies with a larger agricultural sector (Laos and partly Cambodia) were less effected in 2020. This picture of the immediate impact of the pandemic does not correspond to the later deviations from a non-COVID scenario, cf. Table 2. Thus, the pandemic might have imposed a structural shock to some of the economies, especially including shifts in foreign investment to the region, where it has been difficult to return to former high-growth rate trajectories.

Conclusion

The COVID-19 pandemic had negative impacts on Southeast Asian countries’ growth rates immediately after the outbreak of the virus and the imposition of restrictions. The shock to the economies comes at more or less the same time, and the reaction in the form of government-imposed restrictions also seems similar across the countries according to the Oxford Stringency Index (OxCGRT 2023).

The WDI data for GDP in the region clearly exhibits a drop in income levels from 2020 onwards. Applying exponential smoothing to the GDP data from 2000 to 2019, the analysis makes an estimate of the counterfactual or non-COVID case, i.e. making a forecast for 2020–2022 GDP in a non-pandemic scenario. The deviations between the GDP forecasts and the actual values reveal a huge loss in incomes for all of the eight countries in the region. The deviations measured in percentage of GDP are in the range of roughly 5–20%, which are high values compared to former economic downturns. Some countries are doing better than others, for example Vietnam avoiding negative growth rates, but with recent growth rates still below the years prior to the pandemic and in that sense also exhibiting GDP loss. The negative impact came to Myanmar with a one-year lag, but the economic downturn was dramatic with a 20% loss – but other events like the military coup in 2021 may have also impacted this case.

The heterogenous impacts from COVID-19 in the region are not easy to explain and may be related to a range of causes. The quality of health systems and the capacity to cope with the virus are important as well as the structure of economies. Some of the countries are very tourist-dependent and were thus immediately hit by the closure of travel, and similarly negative impacts hit if a country is relatively dependent on foreign trade and investments. The opposite case might be an economy having a relatively large agricultural sector – with people living outside the restricted cities – and thus being less impacted during the pandemic. The fact is a loss of GDP, and there may be many causes and reasons to include in the explanation which might be issues dealt with in further research on the impact of COVID-19.

Disclosure statement

No potential conflict of interest was reported by the author(s).