From "The Unreasonable Fundamental Incertitudes Behind Bitcoin Mining" (page 40):
With current bitcoin software [6, 7], at certain moments in time the reward for mining is divided by two in one single step. This is NOT compensated by the fact the the diffculty of mining increases all the time. It just adds sudden adjustments every 4 years to a diffculty curve which typically goes systematically up due to a steady increase in the production of new hashes. We predict that in the future the difficulty curve will have a discontinuity at the moment of the 4-year halving. Until now it has not happened because apparently only very small percentage of active mining devices were switched off on 29 November 2012.
Inevitably, on one day many devices will stop being profitable and many people may lose their interest in bitcoin. We are talking about the human factor. Investors may decide that they are no longer going to give lots of money to a high-tech industry which has just decreased production of hashes per second and is binning many mining machines at a massive scale due to a strange rule which has no justification and could easily be modified. They might move their money elsewhere and invest in another cryptocurrency. Overall we expect that a sudden slump in profitability of mining is likely to provoke some sort of a much larger ripple in the bitcoin markets, potentially lasting up to 4 full years.
The paper appears to make the argument, that the miner compensation is already decreasing through the rising difficulty, and that the reward halvings will cause sudden drops in mining power which in turn will "provoke some sort of a much larger ripple in the bitcoin markets, potentially lasting up to 4 full years." As "people may lose their interest in bitcoin", when their devices stop being profitable. This is because "We are talking about the human factor.".
The presented argument is weak and thinly supported. While the taking effect of the reward halving is sudden, it is well-known and predictable years in advance within a reasonably small timeframe. Professional miners can easily plan their mining capacity and purchases to factor in the reward halving. There is likely to be a reduction in mining capacity, but it is not obvious, let alone inevitable, why the halving event should turn away people. Also, the mining equipment which will be obsoleted would mostly be comprised of less efficient older models, likely to present a smaller portion of the total mining power.
Further, it seems likely that corporate entities with a sufficient stake in Bitcoin would consider either some in-house mining or subsidy of global mining an acceptable cost to protect their previous investments by supporting the network, should the security of the network be in serious doubt.
Should Bitcoin become sufficiently pervasive, we might actually see Bitcoin mining financed as a public good, similar to other networks such as electricity and roads.
Conclusion: The paper does not comprehensively discuss potential effects of the halving, but rather asserts a specific scenario, which it supports unconvincingly.