2016 Fiscal Year Annual Research Report
Simulations of Municipal Mergers
Project/Area Number 
16H06968

Research Institution  Kobe University 
Principal Investigator 
Weese Eric 神戸大学, 経済学研究科, 准教授 (50777844)

Project Period (FY) 
20160826 – 20180331

Keywords  Simulations / Municipal Mergers / Hedonic games / Moment inequalities / Markov chains / Jurisdiction formation / Integer programming / Maximum score 
Outline of Annual Research Achievements 
Over the past six months I completed a working paper ("Inefficiency and SelfDetermination", joint with Masayoshi Hayashi and Masashi Nishikawa), and also developed the empirical framework for a followup paper.
The major improvement to the working paper made this year was the realization that the model we were using to describe Japanese municipal mergers is in fact a type of "fractional hedonic game", which has been studied from a theoretical perspective in computer science. The method we developed to simulate Japanese municipal mergers thus has broader applicability, because in fact it can be used to simulate any fractional hedonic game. For example, this could include how households group together when sending their children to private or public school.
The followup paper currently in progress is based on the idea of modelling political boundary changes (such as municipal mergers) as a Markov process. In particular, I consider a particular model where the disutility of a merger is determined by Euclidean (squared) distance and the benefit is due to efficiencies of scale from a fixed cost that can be shared: in this special case the Markov process actually forms a reversible Markov chain. A reversible Markov chain has special properties that make estimating the parameters of the model based on observed data much simpler than all other models that I have considered thus far. For example, a standard logit model using subsets of potential jurisdiction boundaries is a consistent estimator, as is a standard maximum score estimator.

Current Status of Research Progress 
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
I have all the data necessary to apply a Markov chain model to the Heisei municipal mergers. This includes data on income (at the Ooaza level) as well as travel patterns (somewhat above the Ooaza level, equivalent to the 15000 Meijiera municipalities). It is important to have data for smaller areas than the municipalities that actually merged, because a major question of interest is what would have happened had smaller areas (such as Ooaza) been allowed to break off and join a different municipality without the approval of the "remainder" that they were leaving behind. My hypothesis is that this would have worked out very poorly, because rich areas or areas with a very high tax base would have unilaterally chosen to leave poorer areas that they had been joined with, and the result would have been a pattern of jurisdiction boundaries that would make little sense from the perspective of a rational observer. This question is particularly timely given the Brexit negotiations currently underway: Scotland claims a right to hold another referendum, given that the rest of the UK is leaving the EU. Why does it make sense to afford them this right, given that nobody asserts it for, for example, London?

Strategy for Future Research Activity 
In general, it is extremely unusual to have a model of jurisdiction formation where the parameters of the model can be estimated via standard techniques such as logit. I thus plan to apply the model described above to as many datasets as possible. These include the Heisei (and possibly Showa) municipal mergers in Japan, as well as European political boundaries over the past 1000 years. This latter project includes joint work with Nancy Qian (Northwestern) and David Schonholzer (Berkeley).
One area of particular interest is counterfactual boundaries for places that were colonized and thus did not have a chance to develop their own political jurisdictions "naturally". In particular, what would African national boundaries have looked like in the absence of European colonial interference? Using the model that I have developed, it should be possible to estimate parameters describing how European boundaries are related to the physical characteristics of Europe (mountains, rivers, agriculturally suitable or unsuitable terrain, etc.). We can then ask, using those same parameters, what boundaries we might expect to see in Africa. Are the boundaries that would have arisen in Africa as "stable" as those we see in Europe, given the physical characteristics of Africa? One hypothesis is that because there are fewer "natural borders" in Africa, simulations will reveal that the continent is relatively short of "natural states", which might suggest one reason why strong centralized states failed to develop in the same way as they did in Europe

Research Products
(9 results)