Of course, the problem with that reasoning is the conceptualization of gov't as a single individual who has 'self-interest.' That is a blatant fiction. Organizations or 'fictitious persons' do not have self-interests, only 'physical persons' do. In which case, the actions of an organization result from some form of combination of the interests of the organization's staff - as the institutionalists maintain. Thus, we need to examine the self-interest of individual actors as well as their power within the organization to explain how that organization ("leviathan") behaves.
In this specific case, while government staffers may find it in their best interest to increase the tax base (wage increases, job tenures), that is not so for the elected officials. The latter's main source of income (in a long run) is not their public sector's salaries, but payments received in various forms (e.g. political contributions, speaking fees, contracts, or after-the-term-in-the-office sinecures) from the constituents. It thus follows, that such individuals' self-interest is to protect the interest of their paying clients i.e. constituents whose best interst is maximizing private wealth my minimizing taxes (the prisoner's dilemma applies here).
Moreover, it is the elected officials (congress-men and -women, the prez) who decide the interest rates. It thus follows that the 'self-interest of Leviathan' - to sue the authors' terminology - will be determined by the elected officials, whose self-interest are aligned with the highest bidder, whose self-interest include minimizing the taxes - even though in th elong run it may cause Leviathan to fail to fulfill its mission of enforcing social contract.
To summarize, the authors' conceptualization of the 'Leviathan problem" is fundamentally flawed in that it fails to account for the collective nature of that entity, and the prisoner's dilemma influence on its behavior. That is an 'internal criticism' i.e. using the conceptual framework of the rat-choice approach assumed by the authors.
wojtek