Croissant and Haynes (2021) argue Asia is one of the central arenas of contemporary autocratisation.

We can see a variety of political regimes, ranging from the democratic (e.g. India) to the hybrid (e.g. Thailand) and the autocratic (e.g. Cambodia) undergoing autocratisation.

One reason why we should focus on Asia: it is a principal theatre of multipolar geopolitical competition.

IE China and the US are engaging in great power competition (REFERENCE FONG?).

Various middle powers such as Australia, India, Japan and South Korea are navigating the geopolitical dynamics in different ways (REFERENCE)

Puzzle: why are some countries autocratising in Asia and others are not?

Puzzle: are international or domestic factors able to explain variation in level of militarisation? I.E. What accounts for more variation? Domestic factors such as over-sized role of strongmen in political fabric of the state? Or international factors such as the role of China and Russia in diffusing / socialising anti-democratic norms?

International correlates

The internationally driven explanation suggests ‘international drivers’, namely autocratic powers’ exporting influence.

Operationalising international drivers of autocratisation: Foreign aid from Russia, China, US, EU FDI from Russia, China, US, EU Trade volume with Russia, China, US, EU(Data: Gravity) Similarity UN voting from Russia, China, US, EU (UNGA vote records)

NB READ AND ADD TO LIT REVIEW Julia Bader, Jörn Grävingholt and Antje Kästner, ‘Would autocracies promote autocracy? A political economy perspective on regime-type export in regional neighbourhoods’, Contemporary Politics 16: 1, 2010, pp. 81–100,

Marianne Kneuer and Thomas Demmelhuber, ‘Gravity centres of authoritarian rule: a conceptual approach’, Democratization 23: 5, 2016, pp. 775–96

Stephen G. F. Hall and Thomas Ambrosio, ‘Authoritarian learning: a conceptual overview’, East European Politics 33: 2, 2017, pp. 143–61

Antje Kästner, ‘Autocracy promotion’, in Wolfgang Merkel, Raj Kollmorgen and Hans-Jürgen Wagener, eds, The handbook of political, social, and economic transformation (Oxford: Oxford Academic, 2019).

material, political and social militarization

First graph looks at the relationship between average tradeflows with China and average MATERIAL militarization levels.

Next graph looks at the relationship between average tradeflows with US and average material militarization levels.

## Warning: Removed 1 row containing non-finite outside the scale range
## (`stat_smooth()`).

Next graph looks at the relationship between average levels of diplomatic disagreement with China and average material militarization levels.

## Warning: Removed 2 rows containing non-finite outside the scale range
## (`stat_smooth()`).

Next graph looks at the relationship between average levels of diplomatic disagreement with China and average material militarization levels.

## Warning: Removed 1 row containing non-finite outside the scale range
## (`stat_smooth()`).

Domestic correlates

Regression output

First we will look at international variables that are related to the three different types of militarisation

tinytable_hax8cox0okayvbeeclj6
Types of Militarization
Material Political Social
+ p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001
All independent variables lagged by two years
All independent variables exponentiated
China Tradeflow 0.007 -0.002 -0.012
(0.009) (0.005) (0.008)
Russia Tradeflow -0.007 0.009** -0.022***
(0.006) (0.003) (0.005)
US Tradeflow 0.008 -0.019*** 0.009
(0.008) (0.005) (0.008)
Num.Obs. 4437 4437 4437
R2 0.005 0.032 0.029
R2 Adj. -0.031 -0.003 -0.005
AIC 10054.0 4812.5 9355.9
BIC 10079.6 4838.1 9381.5
RMSE 0.75 0.42 0.69
Std.Errors Custom Custom Custom

Interpreting and comparing AIC (Akaike Information Criterion) and RMSE (Root Mean Square Error) can help you evaluate the relative quality of statistical models:

AIC Purpose: The AIC is a measure of the relative quality of a statistical model, for a given set of data. It deals with the trade-off between the goodness of fit of the model and the complexity of the model (its number of parameters). Interpretation: Lower AIC values generally indicate a better-fitting model. A model with fewer parameters may have a lower AIC than a model with more parameters, unless the latter provides a substantially better fit. Comparison: When comparing multiple models, the one with the lowest AIC is typically preferred. The difference in AIC (ΔAIC) can also be informative: a ΔAIC of more than 2 is considered meaningful, while a ΔAIC of more than 10 indicates that the model with the higher AIC has essentially no support. RMSE Purpose: RMSE measures the average magnitude of the errors between the predicted values by the model and the observed values. It represents the square root of the second sample moment of the differences between predicted values and observed values or the quadratic mean of these differences. Interpretation: Lower RMSE values indicate a better fit as they signify smaller residuals between the predicted and observed data points. Comparison: Like AIC, when comparing models, the model with the lower RMSE is typically seen as having a better fit. Your Models Material Model: The AIC is relatively high, and the RMSE is also the highest among the three models. This suggests that, in terms of RMSE, it may not predict the dependent variable as well as the Political and Social models. Political Model: This model has the lowest AIC, indicating that among the three models, it might be the one that balances the fit and complexity best. The RMSE is also the lowest, which suggests that the predictions made by the Political model are, on average, closer to the actual observed values. Social Model: The AIC is lower than the Material model but higher than the Political model. The RMSE is between the two other models. When you are comparing these models:

Lower AIC and RMSE are better. The Political model appears to be the best model in terms of both criteria. However, AIC and RMSE are only part of the story. It’s important to consider other factors like the theoretical underpinnings of your models, variable significance, and the context of your research. AIC is particularly useful when comparing models with the same dependent variable, as it accounts for model complexity. A simple model with a marginally higher RMSE might still be preferable if it has a much lower AIC. RMSE is useful for understanding the actual average deviation of the model’s predictions from the observed values, which can be particularly intuitive in terms of the units of the dependent variable.