Last Friday, Lyft, the ride-sharing transportation network company, went public to much fanfare, and news reports indicate that many more tech companies could follow in 2019 including Uber, Pinterest, Airbnb, Slack, and Palantir. These companies are challenging to analyse because generally they are growing quickly but are also incurring losses, which makes forecasting their business and share price performance incredibly difficult.
A simpler starting point is base rate analysis – how do technology IPO’s perform on average, and what have been their observed characteristics? This forms the reference class and the historical range of outcomes make up the “outside view”. Using base rates as the initial estimate has been popularised in recent years by Philip Tetlock (“Superforecasting”) and Michael Mauboussin (“The Success Equation”).
Let’s build a reference class of all US-listed tech IPOs launched between 2010-2018 with at least 12 months of trading history post-IPO. There are 220 companies that meet this criteria.
Notably these deals have historically done very well out of the gate. Tech IPO’s have risen on day 1 about 80% of the time, with a median pop of +21% (Figure 1). For example, Lyft’s +9% opening day gain was about a 32nd percentile outcome applying this base rate approach, so broadly in-line with prior tech IPOs.
Figure 1: US-listed technology IPO day 1 performance
Source: Bloomberg, Janus Henderson, data from January 2010 to March 2018
To analyse further, for each subsequent month we calculate the median and the range of observed returns for this set of tech IPOs. In order to isolate IPO-specific performance, all of the return figures below are relative to the overall tech sector (calculated as long each IPO / short S&P Technology index for every month). This controls for the strong performance of tech over the last decade and appropriately compares IPOs launched at different times.
Figure 2 shows median day 1, by month, and first-year relative returns for these 220 tech IPOs along with 25th and 75th percentile outcome ranges (the vertical lines). This provides a visual sense of the distribution of returns in addition to the median.
Figure 2: Tech IPOs first year as a public company (median and 25th/75th percentile outcomes)
Source: Bloomberg, Janus Henderson, data from January 2010 to March 2019
Note: Calculated relative to S&P technology sector index, rebalanced monthly, excludes transaction costs
There are two key things to note looking at the reference class above.
The first is the period of underperformance in the middle of the first year (see months 5 and 6). This likely corresponds to anticipated and actual increase in share supply and selling pressure around the lock-up expiration date (180 days post-IPO). Generally employees and VC backers must wait for the IPO lock-up period to end before they can sell their shares in the market and convert paper wealth into liquidity. Given that pre-IPO holders are often quite concentrated and need to diversify, they can be price-insensitive and cause significant impact due to their selling flows.
The second is that the average first year return (excluding the day 1 pop) for a tech IPO is negative with a median -19% underperformance relative to the broader tech sector. To be clear, the range of observed tech IPO outcomes has been extremely wide ranging from a near wipe-out to a quadrupling within a year. The variability of returns for a tech IPO is roughly three times greater than the broader market.
Overall, the average US tech IPO has risen on its first trading day, but its relative return though the rest of its first year has been negative, with the bulk of the underperformance coming 4-7 months after the firm goes public. This is the “outside view” without doing any analysis of the particular company. Buying at the IPO deal price or before has generally paid off, but buying post-debut is taking the “inside view” that your forecast is more accurate than what prior tech IPO experience would initially suggest.