The book adopts a two-pronged empirical strategy to test the theory. First, a deep dive into a country’s political institutions is required, to ascertain whether electoral districts meeting the conditions for group-based clientelism exist. Second, after these electoral districts have been identified, the researcher must gather data on voting behavior, resource allocations, and other confounding variables at the level of the administrative entity within (tournament-possible) electoral districts and devise stringent empirical tests capable of pitting the theory’s expectations against those of rival theories. This chapter implements the first prong of this empirical strategy, using information about Japan. It presents a detailed overview of how electoral district boundaries are drawn, how votes are counted, what the relevant lower-tier entities are (in Japan’s case, they are municipalities) and how these entities are funded. It explains that Japanese municipalities depend for large shares of their revenue on national treasury disbursements (NTD), which are allocated by bureaucrats but vulnerable to influence by politicians. It classifies the universe of electoral districts that have been used in Japan’s Lower House elections, 1980–2014, as tournament-possible (or not), and it presents evidence that election outcomes are qualitatively different across both sets of districts, in the direction expected by the theory.

A vast scholarly literature addresses the question of why voters in certain electoral districts receive larger allocations of discretionary government resources than voters in others. When resource allocations are found to be positively correlated with electoral support for a ruling party across electoral districts, then it is presumed to favor core supporters. When resource allocations are found to be negatively correlated with support for a ruling party across electoral districts, then it is presumed to favor swing voters. The tournament theory offers an alternative explanation for why resource allocations can vary so substantially across electoral districts. When politicians use tournaments between the municipalities in their electoral districts to win elections, then across electoral districts, the size of a district’s resource allocation will be influenced by the relative sizes of municipalities therein. Moreover, this variable – district-level asymmetry – is expected to act on both electoral support and resource allocations: lowering lowers electoral support and increasing resource allocations. This chapter uses a host of regression specifications to test these two hypotheses using data from Japan, 1980–2014. In doing so, it provides an original account for why money is negatively correlated with support across electoral districts in Japan.

Chapters 6–8 use a comprehensive new dataset on the universe of Japanese municipalities in existence between 1980 and 2014 to examine whether tournaments are being conducted within tournament-possible electoral districts. Using regression specifications designed to minimize the influence of confounders, weigh up evidence for rival theories, and take advantage of distinctive features of the two electoral systems Japan has used, I test two hypotheses. First, I show that LDP politicians make the amounts of money municipalities in their electoral districts receive after Lower House elections conditional on where each municipality places in a rank order of municipalities. Second, I show that the difference in amounts received by municipalities at different ranks is larger at higher ranks, which is also evidence of a tournament. In addition to regression analyses, I consider two of the theory’s microfoundations which are that LDP politicians are capable of lobbying bureaucrats and voters are aware money is tied to support, respectively. Case studies, anecdotes, interviews, election manifestos, and voter surveys offer evidence for these microfoundations. Finally, I use the logic of a tournament to explain why prior studies of the relationship between votes and money in Japan reached strikingly different conclusions.

A new literature on advanced democracies questions the capacity of majorities to influence fiscal policies to advance their distributive interests, either because the modern state is undercut by increasingly footloose capital, or because the wealthy subvert the majority will through the power of money. This paper critically assesses the evidence using an amended dataset from the Luxembourg Income Study and new data from the World Inequality Database (WID). We use a three-class setup and axiomatically derive the distributive interests of each class and then assess these predictions against data on transfers, public services, and insurance for eighteen OECD countries since the 1970s. For the middle class, the transfer ratio (transfers and services as a percent of the net income of the rich) is remarkably stable, and with the notable exception of the United States, so is the relative position of the middle class in the overall income distribution. Top-end inequality and measures of globalization play no role, but both the poor and the middle class do better under center-left governments.