Malaysia Covid-19 Forecast using Generalized Logistic Function (Updated 29-May-2020)

Since the last update on 28 Apr 2020 (Read more here), Malaysia has moved into the Conditional Movement Control Order (CMCO) phase.  Additional easings were introduced, and more economy sectors are allowed to reopen with conditions.

Subsequently, the daily new Covid-19 cases in Malaysia is on the uptrend again.  The original Generalized Logistic Function (GLF) is no longer adequate to predict the outcome as the underlying conditions have changed.

As such, two sets of GLF are now needed to monitor the situation.  The first set of GLF will be spanning from 24 Jan 2020 to 28 Apr 2020 (GLF pre-CMCO), while second set of GLF will be starting from 29 Apr 2020 to the latest date (GLF CMCO).

The GLF pre-CMCO is using the original version GLF but the new data feeding stopped at 28 Apr 2020.  Thus, the subsequent prediction was based on the actual data upto 28 Apr 2020.  The GLF CMCO is forecasted based on the derived CMCO daily data.  The derived CMCO daily data is calculated by subtracting the actual daily data with GLF pre-CMCO predicted data.

For example, on 29 Apr 2020, the actual daily new cases were 94, the GLF pre-CMCO predicted 44 cases on that day.  Hence, the derived CMCO daily data would be 94 – 44 = 50.    Figure 1 shows the derived CMCO daily cases and GLF CMCO daily cases.  Figure 2 shows the 7-days rate of change for the above-mentioned data.

Figure 1: Derived CMCO Daily New Cases

  

Figure 2: 7-Days Rate of Change for derived CMCO Daily New Cases

Figure 3: Actual Daily, GLF pre-CMCO daily & GLF CMCO daily data

 Figure 4: Actual Cumulative Cases vs GLF (pre-CMCO + CMCO)

  

Figure 3 shows the actual daily data, together with GLF pre-CMCO and GLF CMCO prediction.  As the 7-days rate of change has no signs of slowing down, the GLF is predicting the trend will keep moving up for now.  If the conditions do no improve in the next few weeks, the chances of the total cases hitting 10k are high as depicted in Figure 4.

Stay Safe!

Malaysia Covid-19 Forecast using Generalized Logistic Function (Updated 03-May-2020)

After some easing measurements were announced at the beginning of Phase 4 MCO, it seems like the daily new cases have been going up again.  Some argued that the higher cases may be due to imported cases.  Table 1 shows the breakdown of local and imported cases since 29 Apr 2020.

Table 1: Local and Import New Covid-19 Cases

Date	Local	Import
29-Apr-2020	22	72
30-Apr-2020	32	25
01-May-2020	57	12
02-May-2020	94	11
03-May-2020	70	52

The local new cases are trending upward since 29-Apr-2020 and slightly retreated on 03-May-2020.  To study the situation, let’s look at the rate of change curve.  Figure 1 shows the daily new cases plot, together with 7-days, 14-days and 21-days rate of change, exclude the imported cases.

Figure 1: Rate Of Change

While it is obvious that daily numbers have gone up, 7-days rate of change has also crept up.  Both the 14-days and 21-days rate of change is at the juncture of turning up again.  To get a clearer picture, let’s look at the 2^nd derivative of these plots.  Figure 2 is the 2^nd derivative of the rate of change.

Figure 2: 2^nd Derivative of Rate Of Change

 

7-days curve has crossed above the zero line.  While both 14-days and 21-days curve are still in the negative territory, they are approaching zero.  Once both cross above the zero line, it could mean that 2^nd wave is coming.

The data imply that something has fundamentally changed.  It could be due to more sectors have resumed work too early or people are lowering down their social distancing guard.

Take care.  Please continue to Stay@Home if possible!

Aviation Related Stock, Revisit (Part III)

In Part II of this series (Read more here), the basic assumptions for post-Covid19 Monte Carlo model were illustrated.  Table 1 shows the summary of the inputs.

Table 1: Summary of Inputs

2020 EPS Estimation
EPS (RM)	0.28	0.18	0.08	-0.03	-0.15
Probability	15%	50%	30%	3%	2%
Recovery Pattern
Shape	V	U	L
Probability	38%	54%	8%
PE Range (2015 – 2020)
PE	10 – 23	7	30	Other
Probability	~ 50%	0.08%	0.08%	~ 50%
Floor Price
RM2.00 *

 *             This is the revised floor price.  In Part II, the floor price was set at RM3.29 (0.8 of tangible book value as at 31 Dec 2019 RM4.11).  The reason of change is that the lowest price to tangible book value of SAM was 0.619 in the past five years period.  By adding 20% additional discount to the lowest price to tangible book value, the revised floor price would be around RM2.00.

These inputs were fed into the Monte Carlo model.  Total 3000 cases were simulated.  The results of the simulation were plotted in Figure 1.  The snapshot (first 30 lines) of the Monte Carlo model could be found in Appendix 1.

Figure 1: Monte Carlo Forecast Heatmap

The average price for 2020 is between RM3.0 – RM3.5, 2021 could be around RM4.0 – RM4.5, 2022 could be around RM5.0 – RM5.5, while 2023 could improve to RM7.0 – RM7.5.

However, there is one significant difference between post-Covid19 and pre-Covid19 Monte Carlo model – the probability of Min value!

In ordinary Monte Carlo model, Min or Max value have very low probability of occurrence, normally less than 1%.  Meanwhile, in the post-Covid19 model, the floor value is fixed at RM2.0.  As such, any results that fall below RM2.0 will be reported as RM2.0.  The heatmap shows that in 2020, there are 50% chance that the price could go down to RM2.0 (or below).  As the model did not limit the Max value, thus there are 1% chance that it could hit RM8.0 – RM8.5 range in 2020.

Conclusion

The post-Covid19 Monte Carlo model shows that the average price for SAM is lower than pre-Covid19 model but could slowly recover back to pre-Covid19 level, depending on the development of Covid19.  The simulation results are based on the assumptions illustrated in this series of article (Part I, II, and III).  The assumptions are not final and must be adjusted frequently in order to reflect the volatile conditions.  Once the situation has become more stable, a final version of the model would be updated.

The probability heatmap of Monte Carlo model is an additional risk assessment tool to investors.  The closing price of SAM on 30-Apr-2020 was RM5.82.  At this level, risk averse investors might stay on the sidelines.  However, on the longer term perspective, the price could be viewed as neutral.

The purpose of this study is to illustrate the concept of constructing a post-Covid19 Monte Carlo model that could incorporate the impact of the pandemic.  It is not meant to provide any buy or sell recommendations.

Stay Safe!

Appendix 1 (First 30 lines of Monte Carlo Model, there are total 3000 lines)

Disclaimer:  The above analysis does not imply any buy or sell recommendation.  The author disclaims all liabilities arising from any use of the information contained in this article.

Disclosure: The author may have interest in the stocks of the companies in this article.

Aviation Related Stock, Revisit (Part II)

In Part I of this series (Read more here), pre-Covid19 Monte Carlo model was compared with actual price range of SAM Engineering (SAM).  The predicted price range correlate well with the actual data.  The author argued that the market might not have incorporated Covid-19 impact thus a post-Covid19 Monte Carlo model is needed to forecast the price.

Several key factors need to be considered in order to build a robust post-Covid19 Monte Carlo model.  They are,

       I.          Earnings impact;

      II.          Volatility (mood and momentum); and

    III.          Economy recovery pattern and duration (V-Shaped, U-Shaped, or L-Shaped).

First, the earnings impact will be assessed using the concept outlined by valuation guru, Prof. Aswath Damodaran (Read more here).  He mentioned three key crisis-specific inputs:

1.      Revenue Change & Operating Margin in 2020;

2.      Expected Revenue Growth in 2021-2025 and Target Operating Margin; and

3.      Failure probability and consequences.

He also posted a very comprehensive spreadsheet for users to do their own valuation (spreadsheet), and a video guide to use the spreadsheet (guide).

Second, the volatility of the price movement will be assessed using the latest five years Price to Earning (PE) ratio with assigned probability.

Third, the shape and duration of the recovery.  Basically, three types of recovery pattern are considered, V-Shaped, U-Shaped, and L-Shaped.  Each type of recovery pattern will be assigned with probability.

Earnings Impact & Recovery Pattern

Now let us examine the earnings impact by first looking at point number 3 in Prof. Damodaran’s key crisis-specific inputs – Failure probability and consequences.  Prof. Damodaran pointed out that smaller, younger and more indebted company are likely to fail in this crisis.  Based on market cap definition, SAM is classified as small cap company in Bursa Malaysia.  It has been in the industry for more than 10 years thus it is not really a young company.  Thus, the key parameters to consider here is indebtedness.

Table 1 shows the liquidity and solvency ratios of SAM Engineering.  The financial position of SAM Engineering is healthy based on the most recent quarter (MRQ) or trailing twelve month (TTM) information as at 31 Dec 2019.  The liquidity and solvency ratios are healthy and their cash in hand is believed to be enough to pay wages and interest in the near term.

Table 1:  Liquidity and Solvency Info

	MRQ/ TTM
D/E	0.27
Quick Ratio	1.43
Current Ratio	1.92
Operating Cash Flow to Debt Ratio	0.88
Interest Coverage	13.3
Interest Expense	RM2.60 mil
Cash	RM20.61 mil

Let us move to point number two – Expected Revenue Growth in 2021-2025 and Target Operating Margin.  This portion can be analysed together with point (III), recovery pattern.  The revenue growth shall follow V-Shaped, U-Shaped or L-Shaped pattern?

The good thing about Monte Carlo model is one can incorporate all recovery patterns into the model, then assigning the probability of occurrence for each pattern.  However, the probability assignment now becomes the main issue.  A poll by Ernst & Young (EY) showed that 38% of the global executives said the recovery will be V-Shaped, 54% said U-Shaped, while 8% said L-Shaped (Read more here).  This Monte Carlo model will use these numbers as input but with different recovery level.  Table 2 shows the recovery assumptions.

Table 2:  Recovery Pattern (Revenue)

Recovery Shape	Back to pre-Covid19 level by	Recovery Path	Probability
V-Shaped	2023	Straight Line	38%
U-Shaped	2025	S-Curve	54%
L-Shaped	2030	Flat-S	8%

Finally, let us look the most difficult parameter - Revenue Change & Operating Margin in 2020.  This is the most important parameter as it serves as the reference point for the model.  Based on rough estimation, if the revenue dropped by 30%, at least 10% work force reduction is needed in order to maintain positive EPS.  Thus, SAM’s 2020 EPS may go negative if the revenue dropped more than 30%.  As such, in accordance with L-Shaped recovery pattern, the minimum price for SAM shall be determined by using discounted tangible book value.  As at 31 Dec 2019, the tangible book value per share of SAM is RM4.11.  Assuming it may trade at 80% of its tangible book value, it will be around RM3.29.  This would serve as the floor value for 2020 till 2023.  (Floor price changed to RM2.0 on 3 May 2020, see Part III for details).

Table 3 shows the assumptions for revenue forecast and the happening rate for 2020.  The EPS is highly dependent on the cost cutting measure and other government support scheme.  For this model, only work force cut is assumed.  Tax rate and other measures are assumed similar to 2019.

Table 3: Revenue Forecast & Probability

Revenue Drop	10%	20%	30%	40%	50%
Work Force Cut	5%	10%	15%	20%	25%
EPS (RM)	0.28	0.18	0.08	-0.03	-0.15
Probability	15%	50%	30%	3%	2%

Figure 1 shows the EPS forecast for various recovery pattern.  (The U-Shaped recovery pattern looks more like a Nike swoosh).

Volatility (Mood & Momentum)

The volatility impact (mood & momentum) is studied using past five years PE range.  Figure 2 shows the daily PE range from Mar 2015 to Mar 2020.  The PE could go as high as 30 and hit the lowest at around 7 but their probability of occurrence is 0.08%.  They only happen once in five years.  About 50% of the occurrence happened between 10 to 23.  See Table 4 for details.

Table 4: PE Range and Occurrence (2015 – 2020)

PE	Occurrence	Percentage of Occurrence	Cumulative Occurrence
13.7	63	5.2%	5.2%
13.8	57	4.7%	9.9%
13.9	34	2.8%	12.7%
13.6	28	2.3%	15.0%
15.1	23	1.9%	16.9%
13.5	21	1.7%	18.6%
15.8	20	1.6%	20.3%
15.9	20	1.6%	21.9%
13.1	18	1.5%	23.4%
14.2	18	1.5%	24.9%
14.3	18	1.5%	26.4%
15.2	18	1.5%	27.9%
15.6	18	1.5%	29.3%
15.5	17	1.4%	30.8%
14.4	16	1.3%	32.1%
14.6	16	1.3%	33.4%
14.8	16	1.3%	34.7%
15.4	16	1.3%	36.0%
23	16	1.3%	37.3%
13.4	15	1.2%	38.6%
14.7	15	1.2%	39.8%
14.1	14	1.2%	41.0%
10.9	13	1.1%	42.0%
13.2	13	1.1%	43.1%
15.3	13	1.1%	44.2%
21.4	13	1.1%	45.3%
22.9	13	1.1%	46.3%
13	12	1.0%	47.3%
14.5	12	1.0%	48.3%
21.8	12	1.0%	49.3%
22.4	12	1.0%	50.3%

The PE range with their probability of occurrence will be fed into the Monte Carlo model, together with the earning estimates and their happening rate in accordance with different recovery pattern.

The simulation results will be reviewed in Part III of this series.

Stay Safe!

Disclaimer:  The above analysis does not imply any buy or sell recommendation.  The author disclaims all liabilities arising from any use of the information contained in this article.

Disclosure: The author may have interest in the stocks of the companies in this article.