Modes of Contractual Governance in an On-Demand Service Economy


What has been dubbed the ‘Uber-for-X’ approach in the startup world, and the transformation to a on-demand service economy by scholars such as Harvard’s Yochai Benkler, turns out to be an interesting subject of study for economic organization theory. It is interesting to ask what modes of contractual governance occur in the relationships between the platform provider (“Uber”) and its don’t-call-them-employees (“partners”) as well as its customers.

Starting with Coase in 1937, a formal or semi-formal definition of the nature of the firm became the subject of scientific inquiry. Coase’ timeless contribution consists of two fundamental insights: the firm can be understood as the boundary where the price-mechanism of public markets is superseded by non-market-based hierarchical mechanisms of contract governance. To understand the nature of the firm therefore becomes the study of when it is preferable to vertically integrate some aspects of the firm’s production function with the means of a more complex contract regime than is characteristic of the spot contracts of classic exchange markets. The second insight is that the two alternate means of governance entail different costs under certain circumstances: e.g. a resource of which a constant supply is needed may be acquired more cheaply through a long-term supply contract rather than through regular spot exchanges on public markets.

The conceptual framework laid down by Coase is powerful because immediately there come many different reasons to mind when a vertical integration of the firm’s external dependencies may prove cheaper in the long run. Technically, it is of secondary importance, if those reasons are modeled as transaction costs, as Coase did, or some other quantifiable or even non-quantifiable metric[1].

In case of the rising on-demand economy, several interesting remarks can be made when mapped onto the Coasian model, starting with the concept of vertical integration.

Vertical Integration
The firm in a sharing-economy regime makes use of a new form of vertical integration. Firms such as Uber and AirBnB integrate their dependency on physical and human infrastructure through a technology-enabled, digital network. Such a digital network exhibits characteristics of both market-based and hierarchical governance mechanisms. The workings of the firm’s algorithm in distributing available resources based on signals and abstract rules may be compared to a decentralized market’s price-based coordination, at least in effect. Towards the network’s agents, on the other hand, the firm exerts control that is akin to hierarchical non-market regimes. This yields astonishing results: while achieving the same positive gains as classical hierarchical modes of governance, especially with regards to human labor, the network-based approach removes most of the costs typically associated with a large workforce integrated through long-term employment contracts .[2] Moreover, the physical infrastructure comes for free together with the human labor.[3]

Boundaries
Firms of the given kind gain an additional boundary in the Coasian sense because of their new way to integrate external dependencies.[4] The firm exists as a market-facing network and a network-facing platform. Those two boundaries differ in that the rules of governance within the firm’s network are determined by a different set of factors than those outside of it. More specifically, the firm-as-a-platform determines the rules of the network in the middle of which it positions itself, whereas the the firm-as-a-network competes on public markets against other firms, such as given in the case of AirBnB vs. hotels. Moreover, the firm-as-a-platform also employs people, e.g. server administrators and software programmers, in relation to which it uses classical hierarchical models of governance.

Even more Boundaries
The on-demand economy is being called sharing economy for a reason: the network-approach integrates not only the firm’s dependency on labour and physical infrastructure but the customers of its services as well. The platform integrates both parties to a contract — provider and consumer of services — into the same network-based governance mechanism. As a consequence, the two main kinds of economic relationships the firm has are governed by a mechanism that is neither completely like a market nor completely hierarchical in sense of the Coasian theory of the firm.

Algorithmic Markets
Within the network itself, the firm-as-a-platform mediates the economic relationships between buyers and sellers. It serves the same purpose as a public market but doesn’t use price as its sole coordination mechanism. It may or may not be understood as a new kind of market — an algorithmic market. An algorithmic market in this sense is foremost characterized by the fact that it is governed by a centrally planned (and planning) entity: the algorithm. Since markets are classically defined as being decentralized and only coordinated through price, it may be strongly argued that an algorithmic market is no market at all. However, if one defines a market as an autonomous mechanism, that uses abstract signals as a means to coordinate the formation of economic relationships between actors with complementary wants and needs then an algorithm can be understood as or at least reasonably compared to a market. The comparison carries weight: the information a firm’s distribution algorithm processes is still decentralized. The key difference is that this decentrally created information is observed, collected and retransmitted to a central hub of information storage and processing, i.e. the firm-as-a-platform (see also ‘Medium of Governance’ further down below).

Signals and Transmission
The signaling used to coordinate relationships is different at the two boundaries of the firm in an on-demand economy. Still, since the boundary between the network and the public market is permeable for information to a high degree, the network’s governing algorithm can use the market as an additional source of information.[5] Besides, a firm in an on-demand economy has no direct ownership of any physical or human resources besides what is needed to keep the servers running (and R&D in resource-planning algorithms). Therefore, understanding signals and transmitting them efficiently becomes the key function of the firm. From this perspective, the firm may be defined as an entity occupying a sweet spot in a coordination equilibrium.[6]

Medium of Governance
The digital medium in which the firm acts and exists in an on-demand economy houses novel structures of control that allow for a new form of contractual governance. This mode of governance can be classified as a form of vertical integration as identified above. What differentiates it from classical forms of hierarchical governance is the granularity of factual control over specific performance metrics it allows. Uber the ‘taxi service’ takes this to an extreme by being able to observe both the pre- and post-contractual phases of the relationships between the economic agents in its network at a level of detail and completeness that was unprecedented before the deep diffusion of smartphones in modern society. This insight is the key novelty on which the very existence of the on-demand economy depends. To a much lesser extent the new economy is enabled by the current exploitation of a welfare system under which a firm’s ‘partners’ are left without the benefits that workers on payroll currently enjoy in many countries.

Competition of Governance Mechanisms
Williamson’s organization theory, an approach he dubbed “the lens of contract”, defines the firm not as a production function the way Coase did, but as a governance structure. Applying this perspective to the firm in an on-demand economy appears fruitful: as history unfolds and countless firms based on the ‘Uber model’ will try to establish themselves on their respective market, what we will witness is a competition between governance structures in Williamson’s frame of reference.

Judging by the status quo, it seems to be an unequal competition. Firms using the network-based governance approach greedily[7] integrate essential market functions into their inner, network-based boundary, thereby excluding competition at its very root. So-called “network-effects” are turning the transition from market-based to network-based governance into a one-way street:

Capital specialization in economic networks
Staying within Williamson’s model of Cosian organization theory, network-effects can be understood as a form of capital specialization. It is this process, which turns an actor’s economic interests from a market-oriented mindset coordinated by relative price differences to one which is primarily concerned with the continuity of the existing relationship. Such a transition classically occurs in case of a technological ‘buy in’ of at least one party into a contractual relationship, e.g. an automotive supplier specializing itself through acquisition of physical infrastructure to exclusively produce the parts of one specific car brand. Within the networked boundaries of a firm of the Uber-type, a buy-in happens in many forms and depends in nature foremost on the function an agent fulfills within the network. Disregarding investments in physical infrastructure on the side of the partners, the most interesting aspect of the way capital specialization happens is that it happens through some (1) quantity, that is (2) non-physical and (3) not money. It carries many names among which social capital and reputation stand as two of the most prominent names for the concept. In terms of existing networks (in a more general sense than Uber-type economic networks) social capital is accrued over time in the form of e.g. an extensive social graph (facebook, twitter, et al) and its accompanying set of activity and possibility for interaction. In as much the social-network-type social capital may be considered quantifiable is debatable. In the specific case of economic networks of the Uber-kind, however, social capital is a much more reductive quantity, because “in the end” it translates back into classical economic quantities for the agents active in the firm’s network.

Coming back to the idea of a competition of governance structures it is interesting to note, that capital specialization in the networked governance approach only happens on the side of the firm’s contractual parties, not the firm itself. The firm retains the part of its capital gained by its earnings (e.g. usage or membership fees) in the most general form. Compared to a firm of the classical kind that employs typical hierarchical governance mechanisms with regards to its human workforce as well as its critical physical infrastructure, firms in an on-demand economy are much more agile with regards to new investments into different potential business sectors. Disregarding the major streams of venture capital open to Uber the ‘taxi service’, it is non surprising that Uber is already investing its disposable capital to become the worlds biggest general logistics firm. For a ‘classical firm’ active in the service sector to achieve such a level of economic agility, it would take considerably more time and effort — if possible at all.

Summary
We are witnessing a transition from a classical market-governance vs. hierarchical-governance dichotomy as the defining boundary of a firm to a model, where essential market functions are integrated into a new networked environment, governed by a centrally planned algorithmic model that uses more sophisticated signaling mechanisms than classical, decentralized exchange markets (which are coordinated solely through price).

As such, the very definition of competition changes. Can there even be competition within a networked sphere that is governed by a central algorithm which in turn depends on the whims of a private-interest firm?

The answer is not obvious. In Art. 101 Par. 3 of the Treaty on the Functioning of the European Union, the heart of European competition law regulation, there are established cases of when a failure of competition to emerge because of the behavior of one or more undertakings is considered valid, namely when there is a palpable economic benefit, such as

“improving the production or distribution of goods or […] promoting technical or economic progress, while allowing consumers a fair share of the resulting benefit”.

The question needs to be asked then: are algorithms employed by firms such as Uber with the goal to optimally plan and meet the demand of its customer-base while maximizing its own profit margins better in distributing available resources than the classical market-vs-non-market dichotomy? Are market forces needed — and, if so, available despite the build-up of network-specific capital — to ensure that firms in an on-demand economy continue to provide palpable consumer welfare? In a networked, algorithmically-governed environment, in which economic agents can only indirectly influence their position in the market based on the known metrics the invisible algorithm observes, can legitimate competitive behavior and desired innovation still be sufficiently communicated and economically exploited?

Addendum: regulatory approaches
When policy makers will start to analyze problematic areas of the looming on-demand service economy, e.g. with regard to the systematic disregard of the false self-employment status of the workers involved, it may be asked how an inter-network competition of firms could be ensured on a specific market. Taking into account what I said about capital specialization, it stands to reason that there are conceptually two ways one-sided path dependencies can be counter-acted through regulatory means. One is to prevent capital specialization, one is to counter-balance it. The latter would involve the artificial creation of a two-sided path dependency, by mirroring the degree of capital specialization with regards to the platform’s partners, e.g. with strict labor laws. However, such an act of counter-balancing wouldn’t lead to a higher degree of inter-network competition but simply an even more rigid control structure with the same non-market algorithmical governance mechanisms employed through the firm’s sole discretion.

Therefore, the only possible way to counter-act one-sided path dependence would be to make the accrued social capital of the network’s agents transferrable between competing firms. Such a prescription may seem irritating at first because what social capital is can be extremely network-specific, such as in the facebook-case. With regards to economic networks of the Uber-kind, however, social capital does seem to be very precisely quantifiable and therefore transferrable, if one thinks e.g. of simple customer ratings. Said differently, the network-effects exerted by Uber-type firms could be significantly diminished if reputation were a transferrable commodity between firms offering similar services and/or addressing similar markets.[8]

Footnotes:

[1] There is an academic dispute if semi-formal definitions in economics are useful or if the poverty of mathematical models is (more) honest. At least from a legal perspective, this discussion may be compared to the famous difference between physics and mathematics.

[2] At this point, the looming Uberification of the service sector becomes a political topic.

[3] This is probably the reasons why firms in the on-demand economy call their non-employees ‘partners’: not only do they offer their labor but also the physical infrastructure to conduct that labor. They become entrepreneurs themselves, although they most-likely do not compete on an open market but on a centrally-planned algorithmic one, which they can only influence indirectly through performance-metrics such as customer ratings.

[4] Williamson calls those boundaries “hybrids”.

[5] I haven’t though about this very much but it seems like an interesting subject.

[6] As mentioned by Vitalik Buterin when elaborating on his political views.

[7] In a purely technical meaning of the word.

[8] The reasons behind this thinking become more clear when reading about the Blockchain model and its idea of a public electronic settlement layer.