Tuesday, August 19, 2014

Keynes vs Hayek

The problem when looking at economic change is that predictability is not constant. By predictability, I mean our ability to assign an accurate probability that an event will occur. As activities evolve they transition through different economic states - one of peace, one of war and one of wonder. Those different states have different types of predictability.

For example, in the peace state (the time of products and rental services), we know that evolution towards more of a commodity will occur because evolution is driven by competition which depends simply upon the aggregate actions of all players. There is a high predictability [P] of what will occur. The problem is, that evolution involves stepwise changes which depends upon the actions of individual actors and hence the predictability [P] of when the change will occur is low.

So, let us take an activity (I've drawn this in figure 1 below). A[1] is an act that is evolving through competition and has various stages from A[1] to [A4]. We can often say what will happen (e.g. it will become more of a commodity), just not when. Keynes is right, we can predict macro effects e.g. the P[what] but so is Hayek, those changes depend upon individual actors actions, hence P[when] is low.

Figure 1 - Predictability and the economic cycle.


Now, as the act evolves from A[4] to A[5] and becomes more industrialised (normally represented by either commodity or utility services) it initiates a state of war in that industry. This has known effects including disruption of past vendors stuck behind inertia barriers, co-evolution of practice forcing changes to organisation and a rapid increase in novel activities (e.g. B[1]) leading to a state of wonder. Now this 'war' phase is usually reasonable predictable in terms of what will happen and through weak signals also when this change will occur i.e. P[what] and P[when] are moderate.

The state of wonder however has high P[when] i.e. we can often quite precisely say when it will occur. However what will appear? Well that's another matter. The P[what] is low as the novel activities are uncertain by nature, they have the navigational characteristics of uncharted spaces.

So with electricity, as with computing, as with many industries, the P[what] of the act itself becoming a commodity (e.g. A[5]) are quite well known and we can literally reel off the standard effects. But the P[what] of any new higher order systems (e.g. B[1]) built upon this commodity component is low. 

We can say what the effect of computing becoming a utility is likely to do but not what magical wonders will be built upon it.

Now, here is the fun part. When you start to examine predictability e.g. P[what] & P[when], you can categorise different areas of change. It's not all the same - see figure 2.

Figure 2 - P[what] vs P[when] - draft, incomplete, changing


Furthermore when you start to examine multiple points of change, you discover that various aspects of the economy are not in the same economic state - see figure 3.

Figure 3 - Economic state of various points of change - draft, incomplete, changing


Then, when you combine the different points of change plus what is and what isn't predictable with the value chain of different industries, you start to discover that changes themselves can be simultaneously sustaining and disruptive to different industries - see figure 4.

Figure 4 - Impacts of changes on various industries - draft, incomplete, changing


However, though it's complex there's a whole bunch of things you can anticipate. For example, whilst you might not be able to anticipate disruption in the peace phase (i.e. product vs product discontinuity), you can anticipate disruption from the shift of product to utility (i.e. the war part) well in advance often by a decade by looking at weak signals (e.g. publication types). Furthermore, you can anticipate it effects (rapid development of higher order systems, co-evolution of practice) and by looking at the value chains across organisations then you anticipate its magnitude.

So, this leads me back to the whole issue of Keynes vs Hayek. Our economies aren't in one state but different parts are in multiple states, varying between industry in how they're being impacted by multiple points of change at different stages of evolution. In certain circumstance this can manifest itself as larger business cycles but the application of a single method to a market by treating the market as one thing is fundamentally flawed. 

I'm not pro Keynes or pro Hayek, I'm pro both. I view it is necessary to use multiple methods simultaneously based upon deep situational awareness. It's no different to agile vs six sigma, insource vs outsource or push vs pull. With any large complex system, it's never one but both - see figure 5.

Figure 5 - multiple methods on a single large complex system - an old map of High Speed Rail IT.


I strongly view competition as the key element of ... competition. This requires the use of both free market and Government interference for reasons of market failure which includes market under investment, inertia to change etc. The effective use of Government force (the guiding hand) however depends upon good situational awareness. But where better to gain good situational awareness of changes from the market itself?

Alas when you examine the technology industry, then you find situational awareness (as exhibited by strategic play) is not uniform but varies from the good to the fairly atrocious - see figure 6.

Figure 6 - Market variation in strategic play in high tech industry.


But beyond often poor situational awareness, what you quickly learn is that when companies become large and successful then the one thing they don't appear to like is change. Organisations are not good at managing change, they are generally not designed to do so and they become enamoured of their existing business models and practices. Even the financial markets exhibits rampant inertia to change. Many would act in ways to limit competition, forming oligopolies or monopolies where possible, using everything from lobbied regulatory barriers to intellectual property to establish a steady state. 

This resistance to change, these attempts to limit and undermine competition have their counterpart in our social structures where those who have amassed the greatest wealth and income inequality attempt to maintain that position. Whether social or industrial both effects if left unchecked have the same consequence, the limitation and eventually stagnation of competition. Alas, the market exists beyond one nation and such stagnation can only lead to collapse. All nations are in competition and if you allow yours to stagnate either through social or industrial inaction don't expect another nation to return the favour.

There is a fundamentally important role of Government in maintaining competition and interference where necessary. The libertarian ideal of a rampant free market is as much of a fantasy as the centrally planned system, ignoring the importance of inertia and competition combined with the gravitational effect of wealth (it tends to accumulate) and human nature. Both systems will tend to inhibit competition through the establishment of an oligarchy.

However, I don't believe we've seen yet seen the most effective forms of Government organisation. Certainly China seems to exhibit signs of high situational awareness and gameplay. Others tend towards more dogmatic and singular methods. At some point in the future, I suspect we will all learn that multiple methods are needed to manage an inherently complex and evolving environment. But managed it must be, both in terms of industry and society.

In the meantime, when it comes to Hayek vs Keynes - I agree with both.

P.S. Don't confuse this with agreement with Friedmanism.