It has by no means been uniformly the case that integration has led to a narrowing of tax differentials. Tax rates have always been higher in the core than in the periphery, and the gap between them actually widened until the late 1970s. Evidently, the growing integration of Europe in the decades following the Treaty of Rome did not make core nations feel more constrained by tax competition from low-wage nations. Since the late 1970s, the difference between core and periphery tax rates has narrowed, producing a hump-shaped tax gap. Yet this narrowing has gone in the opposite direction to that predicted by the tax competition view: rather than a 'race to the bottom' the EU countries seem to have been engaged in a race to the top. Baldwin repeated this exercise with the same countries for tax rates for mobile capital (i.e. corporation tax rates). This painted a broadly similar picture.

Underpinned by the 'new economic geography', the economic geography view emphasises the self-reinforcing nature of firm location by taking full account of the importance of both backward linkages (i.e. near suppliers) and forward linkages (i.e. near customers). Put simply, spatial concentration encourages spatial concentration. This results in a very uneven distribution of economic activity, with industry and high-end service sectors clustered together. The circularity of the agglomeration force means that capital is lumpy – i.e. little relocation of capital is observed for most of the time, but when a certain threshold is passed capital suddenly moves in large quantities. And the strength of the agglomeration force depends on the degree of economic integration: at very low levels of economic integration (i.e. high trade barriers) agglomeration is not feasible; at very high levels of integration agglomeration is not necessary; while at intermediate levels agglomeration is both necessary and feasible. These factors imply that the strength of the agglomeration force initially rises with the degree of economic integration and then falls after reaching an intermediate level. Figure 2 plots this bell-shaped relationship.

Figure2: Bell-shaped curve

Moving these arguments to the context of Europe, Baldwin aligned the economic geography view with the evolution of tax rates in the EU as seen in Figure 1. Up to the end of the 1970s, lowering trade barriers increased agglomeration forces, and this allowed core nations to raise their tax rates faster than periphery nations. More recently, the advantage of being in the core has eroded. Cheap transportation, communications and liberalization have made it less important to be located in a spatial concentration of industry. In response, core governments moderated the rate at which they raised the tax burden. At the same time, liberalization also raised incomes in the periphery, inducing their citizens to demand better, more expensive public services while at the same time boosting their ability to pay higher tax bills. In response, periphery governments increased their tax rates.

Baldwin considered tax competition in the context of this alternate worldview. As in the traditional case, governments set their tax rates so as to attract a tax base. However, whereas in the conventional view the movement of capital is smooth, in the economic geography view it is lumpy. Consequently, tax competition with lumpy capital is a winner-take-all situation, in that a country with a high degree of agglomeration can always win if it sets its rates low enough. For example, core countries start with lots of industry and sophisticated service sectors. Periphery countries start with little. In principle, the periphery countries could try to lure the core's industrial bases by charging low taxes. But since the core has an agglomeration advantage, even a zero tax rate in the periphery might not be enough to induce firms to move. Moreover, the core can meet almost any tax-cutting challenge by lowering rates, so any challenge is ultimately futile. Periphery countries are therefore likely to abandon attempts to compete head-to-head for the core's industry, choosing instead to set their tax rates on criteria that are unrelated to tax competition.

Hence tax competition in this worldview is very much a one-sided affair. The possibility of tax competition from the periphery continually bothers core governments, but since periphery countries know they are unlikely to win on tax rates alone, periphery tax rates are not constrained by tax competition. Subsequently, the periphery set their tax rates mainly with an eye to domestic concerns. However, the core are constrained in that they set their rates low enough so the periphery do not find it attractive to lower their taxes in order to steal the agglomeration. So would tax harmonization be beneficial in the economic geography case?

Baldwin argued that the most natural way to harmonize taxes in Europe would be to 'split the difference' – i.e. to converge on the common rate that is somewhere between the high core rates and the low periphery rates. This would entail core nations lowering their rates and periphery nations raising theirs and would maintain the core-periphery pattern of industry location. After all, with identical tax rates firms would continue to prefer to concentrate where other firms are already concentrated. Indeed, the one-tax-fits-all harmonization might even worsen the distribution of industry since it would neutralize the periphery's tax advantage for economic activities that are not subject to agglomeration forces.

Given this, higher rates would be unambiguously bad for the periphery. Their initially lower rates were freely chosen, so a scheme that forced them to raise taxes without affecting the location of industry would make no sense. Likewise, the core (which is continually bothered by potential tax competition) is only interested in raising rates. A scheme that forced them to lower taxes, and therefore the quality of public services, would be a move in the wrong direction. A split-the-difference tax harmonization would make all countries worse off. So is the best policy to do nothing? No – policy can improve on the current equilibrium. Baldwin highlighted a scheme that seemed to offer gains to all countries: a tax rate floor placed just under the initial rate of the low-tax region. By construction, this would not affect the low tax region (since it has already chosen a rate above the floor). It would, however, rule out the possibility that the periphery would engage in fiscal competition. Once the core knows that a tax war cannot be started, it can raise its rates somewhat, because the very possibility that the periphery might cut taxes affects the rate charged by the core. Specifically, the core has to set a rate that is low enough so that the periphery would not want to compete for the core. When this is a tax floor that rules out the competition, the core can set its rate closer to the social optimal level.