The transportation sector is now the largest source of greenhouse gas emissions in the United States. By displacing gasoline and diesel fuels, electric vehicles (EV) offer important public health benefits from reduced air pollution and tailpipe emissions.

One fundamental barrier to EV adoption is consumer confidence in the availability of public charging points. Public charging stations are important assets for electric mobility. In fact, the market for electric vehicle supply equipment (EVSE) is expected to grow from 2.5 billion USD in 2019 to 27 billion USD worldwide by 2027.

Some scholars have argued that because consumers will increasingly have access to home charging, centralized public investment in EV infrastructure should not be critical to vehicle electrification or related corporate sustainability strategies over the long run. Others have claimed that public spending on EV infrastructure would be “wasteful” and would go towards infrastructure that would have been built anyway by the private sector. In a classical debate on the public versus private provision of environmental goods and services, the private sector, according to one view, should deliver charging services more efficiently and with higher quality.

That reasoning has turned out to be largely inaccurate—at least from the standpoint of the consumer experience.

In a recently published article, we analyzed streaming consumer data from a national sample of electric vehicle users reviewing charging stations on a popular mobile app. This was mainly unstructured text that updates in near real-time as users check-in to stations. We note that previous studies on EV consumer preferences have mainly relied on surveys or simulations from smaller datasets, so it was not previously possible to evaluate EV consumer data at a large scale.

However, we encountered one practical challenge during the research. We estimated that even at a rate of 100 consumer reviews per hour, it would take about 32 weeks for a human expert to classify the unstructured consumer texts. Meanwhile, the streaming reviews available for research evaluation would continue to grow.

As a solution to this problem, we implemented a popular machine learning algorithm known as a convolutional neural network (CNN) to classify the consumer reviews automatically. The result provided us with the ability to merge consumer sentiment scores with spatial information about precise station locations and amenities. This capability meant that we could now directly evaluate the performance of both public and privately run EV stations using computational methods and importantly, at scale.

To our surprise, after controlling for observable characteristics and time-varying factors, we did not find evidence that privately run charging station locations (e.g. points of interest such as retail and shopping locations, food establishments, healthcare facilities, workplaces, and other non-residential locations) were outperforming publicly run or managed station locations (e.g. government or municipal buildings, public libraries, rest areas, transit or visitor centers, and public parks). In fact, we kept finding evidence of statistical parity in consumer sentiment scores across all major U.S. metropolitan statistical areas.

With 5 years of data from the entire U.S. market, evidence that government-run charging stations were somehow of lower quality or had significantly poorer operation and maintenance was not borne out by the evidence. This is counter to analysts’ beliefs or predictions from theory. For example, one user at a given station writes: “Be careful if you plan on charging here, there are two cars that tend to bogart these chargers. Try the city hall chargers.

However, not all of the performance data was favorable. Conditional on using mobility apps, we also found evidence that about 4 in 10 EV drivers had negative experiences at public charging stations. These quality issues were primarily concentrated in urban locations as opposed to non-urban locations.

In lieu of action by the federal government, state and local governments have begun to accelerate investments in EV charging infrastructure and are inspiring a new generation of would-be entrepreneurs looking to expand the market in response to rising demand. For example, the state of New York recently announced an investment program that would allocate $750 million to build public charging stations through 2025. As these types of investments will continue to accelerate regionally, most EV policies, such as tax credits, rebates, exemptions, and other policies have mainly emphasized increasing the number of connected cars and infrastructure.

Based on our evidence from national consumer reviews, we argue that it is not only the number of connected cars and infrastructure that matters to consumers but also the quality and reliability of the charging experience.

Our 2019 ARCS conference paper was published in the June 2020 issue of Nature Sustainability.

Link to paper: https://www.nature.com/articles/s41893-020-0533-6