Highlights • Stronger association between breath tests and crashes than earlier estimates. • Breath tests are subject to calculable marginal returns. • The optimal ratio of breath tests per driver estimated to be 1.45–1.54.

Abstract In this paper, we examine the relationship between traffic enforcement (in the form of roadside breath testing for alcohol) and traffic outcomes (alcohol-related traffic crashes) to identify an optimal point of traffic enforcement. In Western Australia (WA), Police are authorised to stop any driver and measure their blood alcohol concentration via a sample of their breath. Using a metric employed by Ferris et al. (2013) and a methodology analogous to that utilised by Cameron (2013), we model the relationship between alcohol related traffic crashes and the saturation of breath testing in WA from January 2008 to April 2018. Our analysis suggests that given the saturation of breath testing in 2018 (1.2 tests per licenced driver), a 10 % increase in breath testing would be associated with a decline of 0.34 alcohol related traffic crashes (ARTC) per 100,000 drivers per month, equating to six fewer crashes per month, state-wide. In addition, using ‘willingness to pay’ and human capital cost metrics to approximate the social costs of ARTC, we employ a cost-benefit analysis to estimate the point at which the social costs of ARTC equal the economic costs of breath tests. Our analysis suggests that an increase in the number of tests to 143 % of all licensed WA drivers (an increase of 450,000 breath tests from the 2017/18 financial year) would be anticipated to save the state AUD$13.3 million annually in the human capital costs of ARTC. Our findings suggest that a further increase in breath tests to 154% of all licensed WA drivers (an increase of 650,000 breath tests from 2017/18) may save the state AUD$31.0 million annually in willingness to pay costs. The analytics below present a novel combination of methods to approximate the relative economic merits of increases in traffic enforcement. Furthermore, the findings outlined here have practical applications for operational policing, while providing an analytical perspective for policy makers faced with making recommendations regarding the volume and saturation of breath tests.