Democracy has been one of the most studied themes in the field of comparative political science. This is owed not just to questions of value but also to countries’ growing adherence over time to this type of regime. Studies in this area have sought to answer two main questions: what are the conditions under which democracy emerges and what are the conditions under which democracy survives. The debates have been permeated by different interpretations and currents, most of which use only domestic variables for their explanations.

One of the main theories about the process of democratization attributes the emergence and consolidation of democratic regimes to the existence of favourable socioeconomic conditions (Lipset 1967). Modernization theory, whose precursor is the study by Lipset, takes the view that economic development brings about a gradual build-up of social changes that prepare society for democratization, so that it is more likely that democracies emerge in economically developed countries. However, empirical studies (Przeworski and Limongi 1997; Przeworski et al. 2003) have revealed that there is no relation between the level of economic development and the emergence of democracies. The emergence of democracy is a random event, deus ex machina, but, once established for reasons other than structural factors, it is more likely to survive in a developed country.

Since the 1990s, we have witnessed the development of a literature that has questioned the consensus according to which democracy results only from domestic factors, without any influence from international forces. By analysing four recent empirical studies — Brinks and Coppedge (2006), “Diffusion Is No Illusion: Neighbor Emulation in the Third Wave of Democracy”; Gleditsch and Ward (2006), “Diffusion and the International Context of Democratization”; Mainwaring and Pérez-Liñan (2009), “International Factors and Regime Change in Latin America, 1945-2005” and Wejner (2005), “Diffusion, Development, and Democracy, 1800-1999” — I intend to discuss how this literature has contributed to the insertion of international factors into the theoretical debate on democratization. All of these authors espouse the idea that there exist mechanisms of diffusion or contagion through which the democratic regime gets propagated, for democracies’ grouping in time and space suggests the occurrence of diffusion or of cross-border dependencies that influence the development and persistence of political institutions. In a book on the state of the art in democratization theory, Munck (2007) states that this field of analysis has provided considerable evidence of the influence of international factors on processes of democratic transition and consolidation.

The empirical analyses of diffusion literature are conducted by means of models that take into account not just diffusion variables, such as the proportion of democratic countries in the region and the occurrence of transitions in neighbouring countries, but also control variables representing the main competitor-theories, such as Gross Domestic Product (GDP) per capita, GDP growth, literacy rate, colonial heritage, system of government and internal and external conflicts. As well as corroborating the thesis that socioeconomic factors are not related to the process of democratic transition, the results also reveal that international factors do influence the emergence of democracies. Hence, diffusion theory helps fill in an explanatory gap, by demonstrating that democratic transitions are not completely random and that, yes, there is a pattern to the emergence of democracies on a regional ambit.

With respect to conditions for democracy's survival, Przeworski et al. state in What makes democracies endure?, published in 1996, that economic factors are not the only ones that contribute to the durability of democracy, and that international factors — the proportion of other democracies in the region and in the world — are probably more important to the regime's survival than the country's level of development. In the empirical study conducted later (Przeworski et al. 2003), the variable “international political climate”, constructed as the proportion of democracies in the world, revealed an impact on the stability of democracies, but not on democratic transition. Unfortunately, these authors did not add a regional political climate variable, for diffusion literature demonstrates that international processes that favour democratization operate more strongly at the regional rather than world level. The model proposed by Wejner (2005), for example, points out that structural indicators lose their forecasting power when analysed in conjunction with regional diffusion factors.¹

Gleditsch and Ward (2006) analyse the influence of the diffusion mechanism from 1951 to 1998, by means of the proportion of neighbouring democratic states within a 500 km radius, the proportion of democracies in the world and the occurrence of transitions in neighbouring countries.² The possibility of transition from an autocratic to a democratic regime remains low, 0.015, when a small number of neighbouring countries are democratic and there are no transitions in the vicinity. When the proportion of neighbouring democratic states rises above half, the likelihood of transition increases, and sometimes even exceeds 0.10, when 75% of the neighbouring states are democratic. The percentage of observations correctly classified by the model is around 98.1%.

Upon testing the role of diffusion as a determinant of the magnitude and direction of changes in regime during the period from 1972 to 1996, Brinks and Coppedge (2006) found evidence that countries change their regime so as to approach the average level of democracy or non-democracy found among their contiguous neighbours. Furthermore, they confirmed that countries tend to follow the direction along which most the world's countries are going. Countries belonging to the US sphere of influence tended to become more democratic in the period examined, but it must be stressed that this indicator did not reach conventional significance levels.

One of the most interesting features of the study by Brinks and Coppedge (2006) is that countries’ political regime is not treated in dichotomic fashion — democracies or non-democracies — as in most other studies. By using the notion of degrees of democracy based on the Freedom House classification, the model manages to capture even small alterations towards an increase in democratization. The hypothesis, confirmed by the authors, is that the greater the difference in degree of democracy between contiguous states, the greater the diffusion effect. The average impact of the neighbourhood on the degree of democracy of a certain country is substantial: when the average difference between the target state and its neighbours is of 1 point, one expects an annual change of 0.25 in the Freedom House scale, or a 1 point difference every 4 years. Contrary to what one's intuition might suggest, the neighbouring country's capacity to influence is not conditioned by its size in terms of GDP, territory or population.

Unlike the studies by Przeworski and Limongi (1997) and Przeworski et al. (2003), the analysis by Brinks and Coppedge (2006) indicated that greater wealth is associated with a move towards democratization, but the discrepancy may be related to the different treatment given to the dependent variable — dichotomic in Przeworski et al. and continuous in Brinks and Coppedge.

Diffusion theory also sheds some light on the process of democratization in Latin America. Przeworski et al. (2003) state that several Latin American countries experienced democratic periods despite unfavourable economic conditions, under which other countries tended towards dictatorships. Mainwaring and Pérez-Liñán (2009) show that the likelihood of an authoritarian regime transitioning to democracy grew from 1.5% in the 1945-1977 period to 4.9% in the 1978-2005 period, while the probability of transition to semi-democracy rose from 2.9% to 6.1%. The risk of semi-democracies becoming authoritarian fell from 17.5% to 2.3%, and the likelihood of a democracy becoming an authoritarian regime went from 3.9% to zero. According to the structuralist thesis, democracy is stable in countries with a per capita income above US$ 6,000 and vice-versa. How, then, to explain the survival of democracy in countries like Paraguay, with a per capita income of US$ 2,180, or Peru (US$ 3,990), Bolivia (US$ 1,460) and Colombia (US$ 4,660)?³

Mainwaring and Pérez-Liñán (2005) take the view that certain regions, such as Latin America, possess specific political dynamics and processes, with distinct patterns of causality in comparison with systemic patterns. One of the specificities of this region is related to the causal impact of the level of development on democracy: the level of development has a weak impact in this region, and the relation between development and democracy is an N-shaped curve, whereas the global pattern is curvilinear. Inferences based on world samples would lead to a false understanding of the factors responsible for democratization in this region. Regional factors possess a considerable impact on the political regimes of Latin America. For each additional point in the region's democracy level on the Polity scale, a typical Latin American country increases its level of democratization by 0.71.

In a later study, Mainwaring and Pérez-Liñán (2009) argue that the process of democratization in Latin America during the 1945-2005 period can be better explained by the regional context (proportion of democracies and occurrence of transition), the commitment of the political elite to the regime and by radicalization in the political process (in the form of violent rioting or guerrilla campaigns), rather than by theories that emphasise structural conditions like economic modernization, class and dependence on natural resources. Competitive regimes (democracies and semi-democracies)⁴ are less likely to break down when other countries in the region are democratic, in the absence of radicalism and when the elites are committed to free elections. The influence of the United States is measured by the general orientation of each US administration with respect to democracy;⁵ as in the study by Brinks and Coppedge (2006), this variable did not attain statistical significance.

According to Mainwaring and Pérez-Liñán (2009), the regional variable plays a key role in explaining the major transformation that took place from 1977: zero likelihood of breakdown of competitive regime. The regional context changed from an average of 29% of democracies in the 1945-1977 period to 62% in the 1978-2005 period, contributing both to an increase in the probability of transition and to a reduction in the risk of breakdown of competitive regimes. Against the odds, Latin American competitive regimes have survived despite weak economic and social results. The period since 1978 has shown that elected governments can last under adverse economic and social conditions if the international political climate is favourable.

In spite of somewhat enhancing one's knowledge of conditions favourable to the establishment and survival of democracies, diffusion theory is unable to forecast the probable order of the advance or the contour of the regional grouping's borders. More importantly, the theory does not explain where and why the first democracy in the region emerges. As pointed out by Gleditsch and Ward (2006), the likelihood of an autocracy becoming a democracy is very slight, under 0.015. Would the emergence at least of the first democracy not be random, a deus ex machina, as Przeworski and Limongi (1997) state? The article by Brinks and Coppedge (2006) is the only one to stress that international forces alone are not enough to bring about regime changes, for the country must be in some way ready to respond to them. Furthermore, the authors agree with transition literature, which underlines the need for a trigger — state breakdown (Skocpol 1973), economic crisis (Przeworski et al. 1996), rapid economic growth (Huntington, 1968), divisions within the authoritarian elite (O'Donnell and Schmitter 1986), removal or death of the head of the executive (Londregan and Poole 1996)⁶ — that breaks the inertia of the existing regime and subsequently brings about the adoption of a new structure.

Whilst diminishing the importance of socioeconomic factors to explain democratic transition and consolidation, diffusion theory at the same time reaffirms, though not very clearly, the relevance of domestic political actors, since they are the ones that absorb and process the alterations in the regional environment. If on the one hand one might consider this reaffirmation of the relevance of domestic political forces an important point in this literature, for one of the criticisms made of the structuralist thesis is the absence of actors, on the other, it is also a weak point, since the channels of transmission through which democratic principles are communicated/spread across the countries of a given region are not well specified.

For Gleditsch and Ward (2006), external actors and events can influence the relative power of relevant groups in the struggle for political institutions, as well as their perceptions and strategies. States and transnational actors can promote democratization by means of actions that strengthen domestic actors that want democratic reform and weaken the power of authoritarian regimes, but the authors’ hypothesis that democratic states tend to support opposition movements and government reforms that result in regimes similar to their own does not square, for instance, with the US policy of supporting autocratic regimes during the Cold War. The difficulty involved in the transition can also result from fears regarding the functioning of democracy. Reluctant leaders may become more willing to initiate difficult reforms if other countries’ experience suggests that the costs and consequences of reforms are not that substantial.

Brinks and Coppedge (2006) believe that countries seek to imitate their neighbours’ regime because they are rewarded when their regimes are similar. The reward may be of different natures: peace, mutual security, trade, investments, easy communication etc. Whatever the motive or justification for emulation, what matters is that there exist influential actors, domestic and/or international, that advocate regime convergence.

For their part, Mainwaring and Pérez-Liñán (2005; 2009), state that the dissemination of norms and ideas affects the way actors perceive their political interests and, consequently, their political behaviour and preferences. Changes to convergent regimes may also be stimulated by international actors (countries and regional organizations), by external incentives (sanctions or rewards), by diplomatic support or foreign assistance to certain coalitions.

The point is that the channels of transmission are not directly investigated, and all the hypotheses presented above are justified by the same item of evidence: an increase in the likelihood of transition or survival of democracy, in line with an increase in the proportion of democracies in the region or the occurrence of transitions.

Despite the abovementioned aspects, the literature on diffusion seems to be developing at a promising pace and overcoming earlier criticisms. In 2002, Pevehouse (2002) stated that this new literature had not yet been able to develop systematic cross-national empirical studies. Whitehead (2002) criticized diffusion literature for its excessive parsimony: not considering actors or intentions; not investigating channels of transmission; difficulty in attributing primacy to domestic or international factors responsible for the democratic process; and absence of any distinction between types of or stages in democracies. Furthermore, Whitehead considered that it was not possible to explain the influence of international factors purely through neutral transmission mechanisms that would induce contiguous countries to replicate the political institutions of their neighbours, affecting attitudes, expectations and interpretations of the public, regardless of external agents’ intentions. Interpretations that excluded the role of external actors, their motivations and instruments of action would tend to produce a distorted image of the international dimensions of democratization: the policy of a third power, mainly of the great powers, would be better at explaining not just the propagation of democracy, but also its speed, direction, limits and transmission mechanisms.

The articles analysed in this review not only conducted systematic studies, but also employed sophisticated statistical models, with the adoption of a series of control variables representing competing theories. The influence of internal and external actors has been the object of theoretical attention, and the study by Mainwaring and Pérez-Liñán (2009) has put forward contributions to the empirical aspect, by measuring, albeit indirectly, the commitment of the political elite to the regime. Even the question of the distinction between types of or stages in democracies was dealt with by Brinks and Coppedge (2006), who used a continuous scale of degrees of democracy based on the Freedom House's classification. Moreover, both Mainwaring and Pérez-Liñán's (2009) and Brinks and Coppedge's (2006) model tested the influence of the world's great powers, the United States and the Soviet Union, over the democratization process. Contrary to what Whitehead states, these variables did not display statistical significance.

There still exist important questions that have not been satisfactorily answered by diffusion theory. However, the overcoming of some of the initial criticisms demonstrates an evolution in the debate and a certain maturing of this literature. The impact that the establishment of a democratic regime in Iraq will have on its neighbours must be closely watched, and will serve as a test for diffusion theory. Beyond this, evidence that favourable external environments really do influence the process of emergence and survival of democracies ought to serve as inspiration for future theoretical developments.