Written by: Brett Shadbolt, CEO Censere

Something I constantly struggle with as a valuer is the effect of survivorship bias and its effects on the valuation of early stage businesses.

As valuers we are tasked with trying to predict what a business might sell for on a given day. In performing this task we reference comparable market transactions and draw on inputs from equity capital markets and debt markets in order to determine the value of a business today. Except, in the case of early stage businesses we don't actually take the vast majority of outcomes into consideration. By referencing such transactions and gathering data from equity and debt markets, we are automatically omitting all those businesses which failed.

For businesses which are established and profitable and have a lengthy track record, it may be appropriate to limit our search to completed M&A transactions and equity market comparables; there is a reasonable expectation that these businesses will continue to operate and will be subject to the same future market forces as those we surveyed. But for early stage companies, is that same expectation necessarily true? According to data from the US Bureau of Labor Statistics, approximately 50% of all new startups fail within the first 5 years and only a third survive beyond 10 years. (https://www.bls.gov/bdm/us_age_naics_00_table7.txt)

We repeatedly hear about the high failure rate of new businesses, but in valuing such businesses I almost never see models which incorporate this fact. If we are using a traditional Discounted Cashflow model (in itself not ideally suited to early stage pre-revenue businesses) it automatically suffers from survivorship bias as key inputs to WACC are solely derived from those businesses which have survived and are currently trading on equity capital markets. We could use a higher discount rate to attempt to address this issue. We often see discount rates of 50% or higher for early stage companies - but if we have that much doubt about the future cashflows, I don't think it is appropriate to try and address them through adoption of an arbitrarily high discount rate.

A better solution may be to use a probability weighted model - even something as simple as a 2 factor model such as: 

  1. an optimistic case based on a DCF model assuming that forecasts can be met, using a market derived WACC; and
  2. a pessimistic case based on a liquidation model assuming the business will fail.

Each case can then be appropriately weighted for probability.

Whichever approach is taken, valuations need to recognise the source of inputs and the inherent biases which may exist within these sources and seek to mitigate the impact on the final analysis.