RDP 2015-10: The Life of Australian Banknotes 8. Conclusion

The traditional and Feige steady-state methods are relatively easy to calculate and simple to interpret. The two methods predict that the median life of Australia's banknotes spans a wide range across different denominations – 3.5 years for the $5 banknotes and up to 10 years for $50 banknotes – though all Australian banknotes are produced with the same technology and have identical security features. Using the same method of calculation, the previous ‘paper’ banknotes are estimated to have had a shorter average life (between 6 months and 1.5 years for the $5 and $50 denominations respectively) compared with the equivalent polymer denominations. However, these steady-state methods are unable to adjust to supply and demand shocks, and are thus subject to considerable volatility over time that makes them difficult to use for policymaking.

The survival models result in more stable and intuitive estimates of banknote life compared to the steady-state equations. Over the period of polymer banknote issuance, the survival models' median estimates of the life of Australian banknotes are similar to the results of the two steady-state methods, at around 2 years for the $5 denomination and almost 11 years for the $50 denomination. Although the survival models are more complex, they can be used to produce a range of interesting metrics other than the simple measures of central tendency, allowing a more in-depth exploration of banknote survival. The results suggest that the survival of banknotes is not only dependent on the ways in which banknotes become unfit, but on aggregate, they are also influenced by fluctuations in demand, the currency issuer's choice of distribution or processing arrangements, and the dual use of banknotes (in transactions and as a store of value). Intuitively, the models showed that $50 banknotes are more likely to be used as a store of value as well as in transactions, whereas lower-value denominations are predominantly used in transactions.

Despite this, some questions over the true shape of the survival and hazard functions remain as it is quite likely that there is still some heterogeneity across banknote usage that cannot be captured by the current modelling techniques. In the future, technology improvements that make it possible to record the date of first issuance and destruction for each banknote would make it feasible to investigate this heterogeneity further using more conventional survival analysis techniques.

The sampling data collected through the RBA's banknote quality programs provide a useful way to test and verify the age distribution of unfit and circulating banknotes predicted by the survival models. The age distribution of those banknotes in circulation and those that have been destroyed, as predicted by the steady-state methods and survival models, are generally similar to those in the two banknote samples. On average, however, the survival models' predictions are closer to the outcomes of the banknote sampling programs.