Surveilled 92 — Apple Kremlinology and industrial policy
Can Apple Kremlinology help us design successful industrial policy?
At the start of the year, the “state of the iPad” was a hot topic among Apple enthusiasts. It’s a topic I feel personally invested in: I remember sitting in a hotel room in Frankfurt in 2012, with my second generation iPad propped up on a keyboard cover with an Apple wireless keyboard inside. I was trying to make some updates to a Word document, an anodyne task in computing terms, but boy it was an ordeal. Download the document from Sharepoint, convert it to Pages, make the changes, export it as a Word document—losing all the formatting in the process, and finally upload it to Sharepoint again.
It would have been infinitely easier to do this on a real laptop instead. Blame my particular weakness for shiny new gadgets of course. But even back then, the iPad held the promise of enabling real work to be done on it, in a more travel-friendly format than a laptop and with the promise of more versatility for leisure. Apple increasingly positioned the device this way, introducing accessories like the Magic Keyboard and especially the same chips as Macs.
Unfortunately, more than ten years later after my initial experiment and despite the numerous improvements, using the iPad for garden variety office tasks still proves to be an exercise in frustration. Its decade old promise has never really been fulfilled. I recently gave up on my most recent attempt to go “iPad-only”, got an actual laptop, and upon booting it, Steve Jobs’ famous quip “like handing a glass of ice water to someone in hell” came to mind.
The aforementioned enthusiasts have been pointing out the flaws with the iPad for a long time, and so Apple’s seeming inability or unwillingness to address them is puzzling. They seem to have tried to address most complaints on a few occasions, but most attempts seemed half-baked and half-hearted. It raises the question about the decision-making process behind those attempts, and indeed Apple “Kremlinology” is the enthusiasts’ favourite topic.
Exceptional though it is, Apple is still a corporation, and so it has to manage organisational behaviour like any other. Let’s consider there are three possible main drivers behind product decisions.
First, the company genuinely believes that changes to their product or service will not benefit the majority of its customers. The product managers responsible for the iPad, for example, may not believe that the features requested by the enthusiasts to let the iPad replace Macs would add value to the product, or at least not for a big enough group of users that it would justify the investment. This is an entirely legitimate motivation.
A second and more frustrating reason could be corporate politics. This is where it’s important to really understand the internal organisation and incentive structure of the corporation. If divisions have separate profit and loss statements, in Apple’s example one for the iPad and one for the Mac for instance, the manager of the Mac division may be lobbying against improving the iPad, for fear of internal competition.
Lastly, the organisation may simply not have the resources to make the necessary investments. This is highly unlikely in the case of Apple, but it applies to many other organisations. In this scenario, a well-functioning banking system and capital market should be able to provide funding if the return on investment (ROI) is compelling. As it happens, improving that ROI of a new project is an important lever of a government’s industrial policy. Governments can provide funding to the project directly through subsidies, or reduce its costs through tax rebates.
When looking for investment for a new project, corporations will present a business case that forecasts the return investors can expect. Rather than take these forecasts at face value, potential investors will attempt to double-check their accuracy, using their own approach and financial models. However, the issue that outside investors are faced with is that they have less information about the company and its prospects than its management (“insiders”), and this makes it more difficult to correctly evaluate a possible investment. It is essentially impossible for an outsider to assess the chance of corporate political shenanigans that might kill the project, for example, and they will certainly not volunteer this information in the business case either. This situation is axiomatic, to the extent that it has name: the principal-agent problem.
Private investors’ goals for an investment are straightforward: they are mostly focused on the financial return, and will pick projects with the highest ROI. Governments have a more complex set of goals, in which financial return is likely not even the main one. Increasing the economic activity in the country is usually more important, which benefits workers directly and also has positive second-order effects like increases in tax receipts. But while their goals may differ from private investors, governments also want their investments to be effective, and so they also have to manage the principal-agent problem.
To illustrate from the perspective of a government, consider the semiconductor industry. It has a very complex supply chain that became increasingly centred on China over the last twenty years. In today’s geopolitical situation however, the rapidly growing risks to a China-centric supply chain is driving Western corporations to move or duplicate some activities to other countries. Setting up a new factory is a textbook example of a project for which an ROI will be calculated and then used as the primary decision factor by corporate headquarters.
The semiconductor industry is an innovative, fast-growing and increasingly strategic sector, so naturally governments everywhere are trying to attract semiconductor manufacturers with a laundry list of incentives. By the same token, they also want to anchor and develop domestic players in the industry. The incentives on offer usually take the form of subsidies or tax breaks. Every little helps to increase the ROI, so industry players avidly shop for the best deal governments can give them. This is where the principal-agent problem for the government comes in.
While innovative, the semiconductor industry is also segmented, with lots of companies operating in the so-called mature segments of the industry, where margins are low and competition is on costs, rather than features. The economic impact, for instance the potential for growth or the creation of high-paying jobs, of this segment of the industry is lesser than the advanced segment. This makes it less desirable for governments to support it. But market logic dictates that the players in this segment must grow revenues and profits, giving them a compelling motive to overstate their capabilities to attract government support.
A good understanding of the industry and its players is therefore paramount for governments, not only in terms of technology, manufacturing constraints etc., but also internal corporate dynamics. Reducing the information asymmetry between principal (the government) and agent (the corporation asking for support) as much as possible is essential to make successful investments. Among others, this forms a compelling argument for governments to develop their skills internally, instead of relying on outside consultants who come with their own principal-agent problems. Another complementary approach is for governments to co-invest with private investors: several principals see more than one, and can therefore make better-informed decisions.
When dealing with large multinationals, the situation becomes even more complex, because there will be principal-agent problems within the company itself, driven by the incentive structure and corporate politics. Multinationals will have several plants in different countries, for example, that are competing against each other to receive limited resources from the head office. The managers of each plant will try to sway the head office for investment, and including local government incentives in the project proposal can make a significant difference at head office. Each plant will have different levels of sophistication and maturity though, and a different outlook as a result. Supporting an aging plant may not fully realise the government’s objectives, so here too governments need to go beyond the official business case to assess the desirability of investing. This is not a theoretical risk: I’ve personally sat in meetings where companies were pitching for government support, and these dynamics were clearly and even overtly at play.
The role of the government in an economy lies at the heart of one of the main ideological divides in economics. To caricature, neo-conservatives will argue that, among others because of the principal-agent problem, governments cannot assess what industry projects to support, and hence should abstain from doing sxo altogether. In their view, governments should let go of the idea of industrial policy in general, and instead let markets select “winners and losers.” An important part of the reasoning is that markets aggregate the knowledge of all participants, which reduces the information asymmetry between principal and agent, and therefore leads to better decisions. Leaving alone the merits of the theoretical argument, the reality is that every government has at least some form of industrial policy. Governments had then better ensure that their policy yields good results, and taking a cue from the Apple kremlinologists is not a bad way to start.
Noteworthy links
-
Quick AI roundup — (a) AI practically swept the Nobel prizes, with the physics and chemistry awards going to practitioners in the field like Demis Hassabis and Geoffrey Hinton (who both work for Google). (b) Unfortunately, generative AI is helping to make so-called “pig butchering” scams increasingly more sophisticated. (c) Adding a piece of information irrelevant to the question asked of an LLM tends to make it break down and provide wrong answers.
-
“Founder mode” is the controversy of the moment in the tech world. YCombinator’s Paul Graham argues founders are better than professional managers and their approach should be indulged. Detractors flag this as dangerous tech CEO exceptionalism; the stark difference in fates between Aaron Schwartz and Sam Altman was especially painful to read.
-
We may have passed peak obesity, thanks to the new generation of weight loss drugs like Ozempic and Wegovy.