Passive Investing in Recessions

---

During the 2000 recession, the S&P500 index dropped by over 50% while the Nasdaq index dropped by over 75%. During this recession, the decline took place over the time-frame of over two years. Following the drops of the indices from their all-time highs, it took S&P500 index over 7 years to reach back to its previous high. The case of Nasdaq was much worse, where it took it over 15 years to climb back to its previous all-time-high, in midst of which a second recession occurred.

The characteristics of the price decline during the the Great Financial Crisis (GFC) were quiet different than those of the 2000 recession. While both indices dropped by over 50% over the course of 2 years, this time it took S&P500 and Nasdaq 5.5 and 3.5 years, respectively, to climb back to their previous highs.

Market conditions coupled with the uncertainties associated with the monetary policies adopted the Federal Reserve make it particularly difficult to gauge the depth and duration of each recession. Consider the 2020 recession as an example. During this recession markets recovered to their previous all-time-highs in under 6 months despite the world economies being partly shutdown.

The variability in the depth and duration of recession poses a great challenge to those who wish to invest in recessionary environments, particularly for retail investors. Consider for example a retail investor who wishes to invest a total amount of $100k in the stock market. If the investor waits for 50% decline (2008 recession lows) in the markets to begin investing, it's possible that that the 50% drop never occurs during the course of the recession cycle (e.g. 2020 recession). Alternatively, consider an investor who enters the stock market after 50% drop. Then it is possible that the markets do not climb back for another 5-7 years (e.g. the recession following the dotcom bubble).

The risks associated with investing in individual stocks can be significantly higher than investing in market indices as a recessionary environment can shake out malinvestments and cause stock valuations to come crumbling down (see for example Citibank C after the 2008 recession).

This article therefore focuses on investing in market indices, and considers two simple strategies for investing during previous recessions with a fixed total investment allocation:

1. Case A: Time-based averaging strategy with an percentage-based entrypoint. In this scenario, the investment allocation is blindly invested every month over the course of 2 years following a variable 20-30% drop of the market, depending on the market index. For example, after S&P500 drops by 20%, each month 1/24-th of the total investment amount is invested in an ETF tracking major market indices.
2. Case B: Percentage-based strategy where the allocation is a function of the percentage drop of the market indices. Specifically, percentage invested is assumed to be given by the scaled sigmoid function 0.01/(1+exp(-k*(p-m))), where p is the percentage drop of the index from its all-time-high value, in the range of 0-100 (see figure below), while k and m are hyperparameters that are fixed to be k=0.2 m = 30. Under the strategy and parameters, after a 30% drop from all-time-high, 50% of the allocated investment is invested. This will be also referred to as the "30/50" strategy or "percentage-based" strategy, and is used at 5%-drop increment.

Let's look at how each strategy would have performed in the past three recessions.

S&P500

In the case of S&P500, it is assumed that the investment is done by buying the SPY electronically traded fund (ETF). There are other alternatives with lower index funds such as VOO, but they are not considered here since they generally did not exist during the 2000 recession. Additionally, it is assumed that the paid dividends are reinvested

The considered cases are therefore

• Case A: uniformly-spaced and uniformly-weighted investment contributions done every month after S&P500 dropped by 20% over the coarse of two years.
• Case B: contribution based on percent drop from all-time-high, according to the equation 0.01/(1+exp(-k*(p-30))), where p is the percent drop from all-time high in the range [0, 100], and k=0.2.

Contribution history along with the profits as percent of the total allocated resource.

Figure below shows the monthly history of the profit normalized against the duration invested. The results are also normalized against the total allocated investment amount (not the target invested amount). Case A (uniformly distributed DCA) is comparable to Case B (sigmoid) during the 2000 and 2008 recessions, but there is a significant difference during the 2020 recession.

Nasdaq

In the case of Nasdaq, it is assumed that the investment is done by buying the QQQ ETF (there may be other alternatives that have lower expense ratios). Same as the case of S&P500, it is assumed that the dividends are reinvested.

The cases below are

• Case A: equally-spaced investment done every month after Nasdaq dropped by 28% over the coarse of two years.
• Case B: contribution based on percent drop from all-time-high, according to the equation 0.01/(1+exp(-k*(p-30))), where p is the percent drop from all-time high in the range [28, 100].

Contribution history along with the profits as percent of the total allocated resource.

Figure below shows the monthly history of the profit normalized against the duration invested. The results are once again normalized against the total allocated investment amount (not the total invested amount). Due to the prolonged duration of the 2000 recession, there was negative profit of roughly -10%/year for over 7 years before briefly breaking even, and subsequently losing value due to the start of the 2008 GFC.

During the 2000 recession, Case A (uniformly distributed DCA) considerably outperfoms Case B (percentage-based contribution according to a sigmoid function). In 2008, the performance of the two cases are comparable, whereas in 2020 there was once again a significant improvement if invested using Case B.

Compared to S&P500 where the maximum rate of profit peaked at 40%/year, after 14 years, the Nasdaq investment peaked at a whopping 80%/year after 14 years.

Dow Jones Industrial Average

In the case of Dow Jones Industria Average, it is assumed that the investment is done by buying the DIA ETF. As before, it is assumed that the dividends are reinvested.

The cases below are again

• Case A: equally-spaced investment done every month after S&P500 dropped by 20% over the coarse of two years.
• Case B: contribution based on percent drop from all-time-high, according to the equation 0.01/(1+exp(-k*(p-30))), where p is the percent drop from all-time high in the range [20, 75].

Contribution history along with the profits as percent of the total allocated resource.

Figure below shows the monthly history of the profit normalized against the duration invested. The results are once again normalized against the total allocated investment amount (not the total invested amount). Unlike the case of Nasdaq, in any of the previous recessions, a negative rate of profit did no sustain for more than 2 years. Once again, Case B outperformed case A during the 2008 and 2020.

The maximum achieved profit rate is, relative to the Nasdaq case, are much more modest, peaking at under 40%/year in the case of 2008 recession.

Russel2000

IWM ETF tracks the Russel-2000 index, but it did not exist at the start of 2000. For this reason, the investment is done directly against Russel2000 index, and no dividends are considered.

The cases below are again

• Case A: equally-spaced investment done every month after S&P500 dropped by 25% over the coarse of two years.
• Case B: contribution based on percent drop from all-time-high, according to the equation 0.01/(1+exp(-k*(p-30))), where p is the percent drop from all-time high in the range [25, 75].

Contribution history along with the profits as percent of the total allocated resource.

Figure below shows the monthly history of the profit normalized against the duration invested. The results are once again normalized against the total allocated investment amount (not the total invested amount). In this scenario, negative profit rates were incurred for less than 4 years before breaking even. In this case, Case B (sigmoid) outperformed Case A (uniform DCA) in 2000, 2008, and 2020 recessions.

The maximum achieved profit rate is much more modest, peaking at under 20%/year in the case of 2008 recession, though keep in mind that dividends are not considered.