As prices soar it is common for investors to dream up ways to justify valuations. In previous stock market booms these included everything from new definitions of earnings (let’s exclude share options from expenses!) to citing low inflation as a reason to permanently cut equity risk premiums (a discount rate of zero equals infinite upside — yippee!).
That said, here are two out-of-the-box arguments why rapid advances in artificial intelligence could raise returns in perpetuity. The point is not to suggest current valuations are justified. Rather now seems like a sensible moment for investors to ask whether some new technologies are so different from previous advancements they could upend the laws of economics and accounting.
First consider how artificial intelligence can alter the concept of depreciation. Everyone understands that assets such as buildings, machines and mainframe computers deteriorate over time. They do so at different rates, however the trajectory is only downward. Ageing appears as depreciation in company financial statements and in 2017 was equivalent to about a third of operating cash flows globally, or a tad less than total capex.
But what happens to a computer with artificial intelligence that actually becomes smarter over time? Machine learning by definition is the opposite of decay. If an asset grows more valuable with use then it should be negatively depreciated. For certain technology firms the earnings kick would be substantial. At Google, say, depreciation in the past year was $6bn. If only a tenth of that was reversed the company’s market capitalisation would rise by about one-fifth, based on current multiples after tax.
Nor must depreciation move from red to black to improve returns. Simply lowering the rate at which assets tire out is enough. Artificial intelligence combined with the internet of things will result in physical things becoming more adaptive and responsive — thereby extending their useful lives. And slowing the ageing process comes at a crucial moment.
For US non-financial companies, for example, the ratio of accumulated depreciation to the gross value of plant and equipment has risen by a quarter to 50 per cent over the past 25 years. That means the average American firm’s capital stock is halfway towards the scrap heap.
So one reason for optimism is a reassessment of what depreciation means in a world of increasingly clever assets. A second argument involves the other main factor of production: labour. The reality is that what it means to be human — and therefore a worker — will change over the coming decades. Breakthrough technologies such as brain-computer interfaces, robotic exoskeletons, and virtual assistants have the potential to dramatically increase productivity.
This is desirable in and of itself as wages can potentially grow with less pressure on inflation or margins. But also remember how employee costs are accounted for. Remuneration is expensed as a profit and loss item — a direct hit to current earnings. Human beings are therefore treated differently to other assets, which sit on balance sheets and are depreciated.
Should this remain the case if a tenth of a construction worker’s marginal productivity of labour is derived from an exoskeleton? What about 70 per cent? Or how about a call centre operator plugged into a future version of Salesforce via a brain-computer interface? At some level of bionics or digital enhancement the world’s accounting bodies may decide that capital and labour have blurred enough so workers can be depreciable assets, too.
The result would be a near-term jump in profitability, although in theory long-run valuations should not change. However, if human productivity improves as cyborg worker-machines learn on the job then the negative depreciation story above also applies here. Again the potential uplift to margins is huge as the labour share of output in the US, for example, is about 60 per cent. Capitalising wages should also benefit workers in absolute money terms if the productivity spoils are shared.
To be sure this is some out-there stuff. But so is the idea of intelligent machines that learn and improve with age. It is therefore essential to explore every implication and equity investors should at least entertain the possibility that some of the building blocks of finance will crumble to their advantage.