COVID-19 Forecasts in the Philippines: Insights for Policy-making

| Posted by UP Media and Public Relations Office

Guido David, Ph.D.
Professor, Institute of Mathematics
University of the Philippines
Fellow, OCTA Research (

Ranjit Singh Rye, MPA
Assistant Professor,Department of Political Science
University of the Philippines
Fellow, OCTA Research (

Ma Patricia Agbulos, MBM
Associate, OCTA Research (



In the Philippines, the first case of Covid-19 was reported on January 30, 2020. Since then, the number of cases has soared to more than 3,000, amidst lockdown and quarantine measures implemented by the government. In the Philippines, there are challenges in responding to the pandemic including the shortage of health facilities, health workers, and testing kits. This is aggravated by the economic status of the citizens, and a culture that does not seem to embrace self-quarantine. For these reasons, the number of Covid-19 cases in the Philippines may have been under- reported, especially during the initial period of the outbreak. Given these factors, forecasting the spread of Covid-19 in the Philippines is a challenging task.

Figure 1 below compares the trajectory of total cases in Iran and the Philippines starting from the day when both countries had five cases of Covid-19. It is important to note that Iran did not impose any quarantine measures on its citizens, instead opting for an approach based on self-quarantine. The average transmission or infection rate of Covid-19 in Iran during this period is about 0.4, compared with a transmission rate of 0.3 in the Philippines. This figure already demonstrates in part the effectiveness of imposing a lockdown in Luzon. Within 30 days, the number of cases of Covid-19 in the Philippines was 3,660, compared with 21,638 in Iran. The lockdown may have reduced the number of cases by 83%. Even with possible under-reporting of cases in the initial stages of the disease, the difference brought mainly by the imposition of the enhanced community quarantine (ECQ) is substantial.


Figure 1. Comparison of actual Covid-19 cases of Iran vs. the Philippines, showing the effect of the ECQ (Philippines) vs. no quarantine (Iran).



Given the challenges identified and the trajectory presented, forecasting is indeed challenging but is worth undertaking for its value of providing a tentative guide on how to proceed with managing the pandemic. Forecasts are made using an epidemiologic-based (susceptible, infected, recovered) SIR model of infection. It must be noted that forecasts are highly dependent on the underlying assumptions. The rate of spread of Covid-19 depends on the transmission rate and the recovery rate. The recovery rate is the time until recovery or the time until death. The recovery rate is estimated based on the incubation period of Covid-19, which is usually 5-6 days, plus recovery time of around 1 to 3 weeks. The ratio of these two numbers gives the basic reproduction number, which is used in epidemiologic forecasts. The World Health Organization (WHO) gives the range of the basic reproduction number between 2.0 and 2.5. Note that a reproduction number greater than 1 indicates that the epidemic will spread, while a reproduction number less than 1 means that the disease will die out due to lack of transmission. The higher the reproduction number, the faster the epidemic will spread. Moreover, as stated by WHO, “…estimates for COVID-19 is very context and time-specific, making direct comparisons more difficult.” For these reasons, it is better to have experiential studies of the spread of Covid-19 in the Philippines in order to make better estimates of the basic reproduction number.

Preliminary studies of the spread of Covid-19 in the Philippines gives a recovery time of 15 to 18 days, giving us a recovery rate of 1/18 to 1/15. The transmission rate is calculated daily in order to match the data. The average reproduction number obtained for the period of the study is 4.3. This value, when averaged over the past 10 days, is 3.10, and over the past 5 days, is 2.41.

Scenario 1 forecasts Covid-19 based on the most recent 5-day average transmission rate of 0.08 and recovery time of 18 days (Figure 2). The graph includes the initial 32 days of actual data (from March 10 to April 8, 2020) with 60 days of forecast. Under this scenario, assuming nothing changes, there will be more than 12,200 total cases by May 8, 2020, 4900 of which are active cases, and 980 deaths. By June 8, 2020, the simulation shows a total of 29,500 total cases with 9,800 active cases and 2,600 deaths. This scenario is a simulation based on the most recent trends, assuming those trends do not improve even further. Although the curve for the active cases has not flattened yet, the rate of increase is not very drastic.


Figure 2. Forecast of Covid-19 in the Philippines using recent 5-day average transmission rate of 0.08 and recovery time of 18 days. The number of active cases has not yet peaked, but the spread of Covid-19


Note that the above already includes some of the effects of the ECQ. However, the basic reproduction number for this forecast is 1.40, which indicates that the pandemic will still spread. The number of active cases is still increasing under this scenario. The ECQ, even up to an additional 60 days, will need to be extended until the transmission rate of Covid-19 has improved or there is some change in the dynamics of the spread of the disease due to improved quarantine measures.

One way to reduce the spread of Covid-19 is by decreasing the transmission rate. The decrease in transmission rate due to the ECQ can in fact be observed already. Figure 3 shows the plot of the transmission rate plotted during the 32 days of the observation period. The moving average, a smoothened form of the transmission rate, is also shown, and used for extrapolation. The plot clearly shows a decreasing pattern in the transmission rate of Covid-19.


Figure 3. Calculated rate of transmission over the observation period, and the smoothened curve depicting a Moving Average interpolation of the transmission rate.


From the moving average of the transmission rate, we can estimate the values of the transmission rate in the next few days based on the trend, assuming a continuation of the ECQ. Using a forecast of the transmission rate and a recovery time of 18 days, we obtain a forecast for this scenario (Figure 4). The forecast is very encouraging, as we observe the “flattening of the curve” within 1-2 weeks. The number of active cases by May 10 will have decreased to 2,500 with a death toll of nearly 800, and by June 10 the number of active cases will be down to 2,000 with 1,300 deaths. Under this scenario, the ECQ will not need to be extended as long as all the active cases are already in quarantine, isolation or admitted to hospitals. Although the flattening of the curve will be observed almost immediately, it would not be advisable to end the ECQ for the following reasons: (1) we will need to observe if the forecasts will hold, and (2) ending the ECQ prematurely may cause a sudden increase in transmission rates. In other words, we will need to make sure that the spread of Covid-19 has been managed well before we make changes to the government’s approach to the ECQ.


Figure 4. Forecast of Covid-19 in the Philippines using forecasted transmission rate (based on the moving average) and recovery time of 18 days, showing an almost immediate “flattening of the curve.” Under this scenario, the ECQ does not need to be extended. However, we will need to observe if the rate of transmission will be maintained at these levels, before making changes to the ECQ.


Another way to reduce the spread of Covid-19 is to alter the recovery time. As mentioned, the recovery time is essentially the average number of days until the patient is no longer infective, which is estimated to be between 15 to 18 days. The spread of the epidemic can essentially be attributed to the infective patients spreading the virus. However, with early detection and isolation of Covid-19 patients in hospitals, health facilities and holding areas away from communities, those carrying the virus will no longer spread the disease. If, for example, these cases (i.e. PUIs) are identified and successfully isolated or quarantined within 7 days of infection, then we obtain the forecast below (see Figure 5), given a transmission rate of 0.08 (same as with the first forecast, see Figure 2). The calculated reproduction number here is 0.95. As can be seen in the graph, the curve has flattened, and the pandemic is dying and the virus is no longer spreading. Just as with the previous scenario, the ECQ will not need to be extended beyond the original target date, but a longer observation period will be needed to determine if the effects of the improved testing and isolation will have a sustained impact on reducing the number of infectious cases.


Figure 5. Forecast of Covid-19 in the Philippines using recent 5-day average transmission rate of 0.08 and recovery time of 12 days (based on active screening, testing and isolation of Covid-19 cases). Given this setting where active screening and isolation of cases has been improved to efficient levels, the ECQ will not need to be extended. However, we will need to observe if the Covid-19 testing and isolation will control the spread of Covid-19.



Forecasts are highly dependent on the model assumptions and parameters used. This study uses experiential analysis of cases to date of Covid-19 in the Philippines in order to estimate the transmission rate, recovery time and reproduction number of the pandemic. Three weeks into the enhanced community quarantine period, we observe that the ECQ has shown an improvement by as much as 83% compared with no enhanced community quarantine (for example, Iran). However, even with this, the current transmission rate of 0.08 based on the recent 5 day average still needs to be improved, since the current trends indicate that the projected number of cases will increase to more than 30,000 over 2 months with nearly 10,000 active cases and 2,600 deaths. This is actually a fair result, compared with the Covid-19 pandemic in other countries.

A second scenario forecasts the decreasing transmission rate using a moving average method. Under this scenario, the forecasts show an immediate “flattening of the curve” and curbing of the pandemic. It is possible that sustained enhanced community quarantine can keep the transmission rates at these levels. However, it is important that we monitor the latest data so we can observe if the effects of the ECQ will be sustained in managing the spread of Covid-19. An extension of the ECQ may be needed in order to observe any changes in the rate of spread of Covid-19.

An alternative solution is more proactive and involves actively testing possible cases, and then isolating confirmed Covid-19 cases in health care facilities and hospitals. Responsible self-quarantine may also have the same effect. Under this scenario, if Covid-19 cases are identified and isolated within 12 days of contracting the virus, then even with the l transmission rate of 0.08, the effect is an immediate control of the spread of Covid-19. Again, the data will need to be monitored closely in order to determine if the effects of testing and isolation will manage the spread of Covid-19. An extension of the ECQ in order to monitor any changes in the spread patterns may be justified.

Given these scenarios, we recommend the following:

1. The government must consider extending the enhance community quarantine beyond April 13, 2020. As shown above there had been gains with the restrictions put in place averting an Iran-like scenario. These gains should be sustained until such time that the country has mass tested and isolated infective cases.

2. Corollary to the need to extend the ECQ, the national government must scale up and proactively do testing for Covid-19, isolate infective patients in health care facilities and hospitals, and improve contact tracing such that cases are identified and isolated sooner. Under this scenario, if Covid-19 cases are identified and isolated within 12 days of contracting the virus, then even with the less optimistic transmission rate of 0.08, the effect is an immediate control of the spread of Covid-19.

3. As a consequence of a push to mass testing, human and material resources need to be increased to support our already burdened national health system. The special powers of the President provided in Republic Act No. 11469 could muster these additional resources. These additional resources include but are not limited to more testing kits, incentives and provision for front liners including proper personal protective equipment, and dissemination of PhilHealth’s coverage for testing and hospitalization to encourage low-income households to seek medical attention.

4. Aside from the medical aspect of the pandemic, the economic needs of those whose incomes are adversely impacted should be provided by the national government especially among the poor. This is to ensure that the extension of the ECQ will not cause social unrest. The provision of cash and relief goods both by the national and local governments should be continued and should reflect the size of the household. Further, monitoring mechanisms should be put in place to ensure that the assistance is delivered to intended beneficiaries.

5. Prospectively, the national government should invest more resources in institutions such as the University of the Philippines, which can do evidence- based research to improve policy and program formulation and evaluation. These researches should underpin national plans to bolster resilience of national institutions, e.g. distance education and blended learning in schools, telecommuting, telehealth, and other similar initiatives.

6. Finally, in the same manner that a national task force is in place to manage emerging infectious diseases, a similar government mechanism must be put in place to enable government agencies to share information and widely institutionalize strategies and programs that enhance national resilience. This crisis and all our collective experience should not go to waste.

Download the report here.