La dimensión social del uso de la bicicleta en la ciudad

The dependent variables of the quantitative chapter (chapter 5) are the degree of state building and the number of governance institutions in a de facto state in a given year. The dependent variables are summarised in the table below (table 7). All variable descriptions are sourced from Florea’s 2016 and 2018 code books apart from the newly generated variables.33

Table 7 Dependent Variables

dfsbuild Ordinal variable measuring degree of state building from 1 (low) to 4 (very high)

dfsbuildmod Binary variable comparing low degrees of state building (0) with moderate, high and very high degrees of state building (1)

dfsbuildstrong Binary variable comparing low and moderate degrees of state building (0) with high and very high degrees of state building (1)

dfsinst Number of governance institutions in a de facto state in a given year The first dependent variable (dfsbuild) is an ordinal variable that represents the degree of state building in a given year in a de facto state. Florea (2016: 14) codes the degree of state building by applying 4 categories:

1 low degree of state building: de facto state separatists control the means of violence in the territory and provide minimal public goods (such as physical security)

2 moderate degree of state building; in addition to 1, separatists allocate resources for civilian governance (such as minimal public administration, social security, education, health)

3 high degree of state building: in addition to 2, separatists develop coherent governmental structures (institutions for extraction and redistribution; internal security and border management; courts; ministries) 4 very high degree of state building; the polity has most characteristics of a state (including external relations, representative offices abroad, commercial relations with international partners).

33 _{The variable definitions in tables 7, 8 and 9 are taken from Adrian Florea’s De Facto States}

Dataset Codebook (Versions 1.1 February 2016 and 2.0 September 2018). I do not place the variable descriptions in quotation marks, but with the exception of some newly generated variables in tables 7, 8 and 10, all descriptions are Florea’s work.

103 As the variable dfsbuild is an ordered categorical variable, I decided to generate two binary variables from the original state building variable: dfsbuildmod and dfsbuildstrong. The former focuses on the step from category 1 (low degrees of state building) to category 2 (moderate degrees of state building) and the latter focuses on the step from moderate degrees of state building to high degrees of state building. Generating these two new dependent variables has two advantages in both a practical and an analytical sense. First, dividing the ordinal variable into two stages allows me to run binary logistic regressions across the time series cross-sectional data set, which would not be possible for an ordered logistic regression. The time series cross-sectional panel data literature offers more sophisticated statistical functions on binary variables than for ordered categorical variables (see e.g. Beck et al. 1998; Tucker 1999; Beck 2001; 2008). Second, as this thesis is interested in individual state and institution building developments, recoding the dependent variable allows me to analyse the steps from separatist control to basic forms of civilian governance and from basic forms of civilian governance to coherent governmental structures more thoroughly and precisely. This enables the regression models to measure the effect of independent variables on the change from low to moderate degrees of state building and from moderate to high degrees of state building.

The second dependent variable (dfsinst) is a discrete variable that counts the number of governance institutions in a de facto state in a given year. The variable encompasses ten categories (Florea 2016: 14-15) and does not distinguish between the origin of these institutions, which means that institutions can be inherited from previous autonomy rights or built by local separatist authorities. Florea weighs each governance institution equally, which presents a possible analytical limitation as foreign affairs institutions and militarily supported executives are coded as equivalent despite the latter’s relatively higher relevance for state building. Yet, for this research, not the equal weighting presents a caveat, but that the coded institutional presence does not represent institutional capacity. In more practical terms, the presence of an independent central bank in a given de facto state does not signify the entity’s capacity to perform central banking responsibilities. Abkhazia’s central bank, for example, is physically present and coded as existent in

104 Florea’s data set, but it does not perform all monetary policies usually associated with central banks. Every governance institution that is captured by the variable dfsinst is listed below:

1 an executive supported by a military (coded as present if there is a clear executive authority that makes decisions in the de facto state)

2 a legislature and/or regional councils (coded as present if there is a legislative body in the de facto state capital and/or regional councils)

3 a court or semi-formalized legal system (coded as present if there is a formal or semiformal juridical authority that adjudicates disputes between individuals or institutions in the de facto state)

4 a civilian tax system (coded as present if there are institutions for regularized extraction of taxes from the local population/businesses and/or from the diaspora) 5 an educational system (coded as present if the authorities in the de facto state establish a system of education that functions in parallel with or in lieu of the one provided by the parent state)

6 a welfare system (coded as present if the authorities in the de facto state establish a system of welfare – healthcare and/or pensions – that functions in parallel with or in lieu of the one provided by the parent state) 7 institutions for foreign affairs (coded as present if the authorities in the de facto state engage in diplomacy – establishing missions abroad; engaging in contacts with IGOs and/or foreign governments)

8 media or propaganda system (coded as present if the authorities in the de facto state establish media or propaganda outlets)

9 police and/or gendarmerie system (coded as present if he authorities in the de facto state establish a system of internal control – police and/or gendarmerie – that operates separately from the army)

10 an independent central banking system (coded as present if the authorities in the de facto state establish an independent central banking system that functions separately from the parent state’s banking

The outcome of interest in this thesis is the state building development of de facto states. The state building and governance institution variables of the data set relate relatively well with the qualitative operationalisation of state building of this thesis,

105 which refers to state building as the development of state capacity in form of institutional and governance structures as well as the provision of public services and goods (see chapter 2 for the detailed operationalisation of state building). The state building variable dfsbuild, for instance, uncovers the development of a given de facto state’s state structures and its ability to provide public services and goods such as security, health and education. The state building models of this thesis distinguish between two state building stages. The step from low to moderate degrees of state building engages with a development to a stage of basic public services and goods, such as education and health, while simultaneously covering state capacity aspects such as minimal public administration. The step from moderate to high degrees of state building meanwhile captures a further extension of both the provision of public services and goods (e.g. internal security and border management) and state capacity in form of courts and ministries.

The governance institution variable dfsinst accounts at least theoretically for a de facto state’s institutional capacity by listing the physical presence of a set of governance institutions such as a legislature, tax authorities and a central banking system. The institution variable simultaneously indicates the level of public service and goods provision in a given entity by considering for example education, welfare and legal service provisions. It is important to reiterate that the physical presence of such institutions does not automatically represent the practical capacity to provide services and goods associated with these institutions. Thus, the dependent variables dfsinst and dfsbuild of the large-N study appropriately capture the state building outcome of interest in this thesis and enable an estimation of how variations of direct and indirect involvement of patron states influence the provision of public services and state capacity in de facto states.

The independent variables in the statistical models of this thesis (table 8) centre primarily around patrons and parent-patron state competition. The variable patronspanke classifies patron states according to the definition of patron states laid out in chapter 3 of this thesis. The relparentstrength and relparentgdppc variables refer to the relative military and economic power of parent and patron states. The operationalisation of the independent variables ties in with the theoretical framework and argument of this thesis in a number of ways. The patron variable

106 patronspanke enables the measurement of the effect of the presence of a patron state on the two dependent variables and thereby captures both the direct and indirect diffusion influences of patrons on state building in de facto states. The independent variables relparentstrength and relparentgdppc meanwhile quantitatively capture the indirect diffusion influences of patrons through competitive diffusion channels.

Table 8 Independent Variables

patronspanke Patron classification of this thesis (see chapter 3) relparentstrength Relative parent military and economic strength

relparentgdppc Relative parent GDP per capita compared to patron state

I also included a set of control variables (table 9) to ensure the robustness of the explanations derived from the independent variables, such as prior independence of the de facto state (dfspriorind), the type of de facto state emergence (typeonset), de jure autonomy from the parent state (dfsaut), the war intensity on the territory of the de facto state (dfswarint), the presence of a large diaspora (dias) and the duration of de facto state survival (duration). In addition, I control for domestic de facto state capabilities including the relative rebel capability (relcap) and the fragmentation of the secessionist movement (frag2). See chapter 5 for a detailed theoretical justification for the inclusion of these control variables.

The proposed state and institution building models use the variable patronspanke as a key independent variable, whereas Florea’s (2014) patron variable serves as a control variable in the robustness checks of the quantitative chapter. As Florea does not specify the definition of patron states in the data set codebooks, I contacted him by email and he revealed to me in 2018 that he determines the presence of a patron state by identifying “regularised patterns of military assistance from a third (state) party.”

107 Table 9 Control Variables

dfspriorind Previous independence or autonomy after 1812 typeonset Variable categorises type of de facto state emergence:

(1) de facto state emerges as post-conflict outcome

(2) de facto state emerges out of the contentious interaction between the parent state and separatists (non-conflictual emergence) (3) de facto state emerges in the wake of state/federal collapse

(4) de facto state emerges during the decolonization process (colonial divestment)

dfswarint War intensity in de facto state coded as: (0) no battle-related deaths

(1) minor: between 25 and 999 battle-related deaths in a given year (2) war: at least 1,000 battle-related deaths in a given year between the

de facto and parent state

dias Presence of large diaspora originating from de facto state

dfsaut De facto state benefits from de jure autonomy within parent state duration Number of months entity has survived as de facto state

relcap Relative rebel capability coded as (1) weaker than the government (2) at parity with the government (3) stronger than the government (4) much stronger than the government

frag2 Fragmentation on unified–fragmented scale (cf. Bakke, Cunningham & Seymour 2012):

(1) no fragmentation […]

(9) extreme fragmentation

patron De facto state has a patron. Florea (2018) defines patrons as regularised patterns of military assistance from a third (state) party

In addition to the already existing variables in the data set, I generated a set of variables that perform as either independent or control variables and unpack the influence of patrons and state building processes in de facto states further, situate patron-de facto state relations in a geopolitical setting and ensure more time-varying variables (see table 10). Thereby, the large-N study captures both the effect of patron states on state and institution building in de facto states as well as the impact of indirect competitive diffusion channels of patron states in form of patron-parent state competition. The supplementary variables, for instance, measure the relative patron and parent state capabilities, specify an alternative patron definition and account for previous state and institution building experiences in de facto states.

108 This thesis utilises the Composite indicator of National Capability (CINC) from the Correlates of War Project (Singer et al. 1972)34 _{and World Bank GDP data to} classify the military (patronstrength and parentstrength) and economic strength (patrongdppc and parentgdppc) of patron and parent states. Subsequently, the military or economic strength of the patron state was deducted from the parent state strength to capture the relative power of parent and patron states (relparentstrength and relparentgdppc).35 _{The data set also includes variables that measure the state} and institution building experience (dfsbuildexp and dfsinstexp) and the number of months since the last state or institution building experience (tsincedfsbuildchg and tsincedfsinstchg).

Table 10 Newly Generated Independent and Control Variables

patronspanke Patron classification of this thesis (see chapter 3)

patronname Name of patron state

patrongdp GDP of the patron state

patrongdppc GDP per capita of the patron state

parentgdp GDP of the parent state

parentgdppc GDP per capita of the parent state

relparentgdp Relative parent GDP compared to patron state

relparentgdppc Relative parent GDP per capita compared to patron state relpatrongdp Relative patron GDP compared to parent state

relpatrongdppc Relative patron GDP per capita compared to parent state patronstrength Military and economic strength of patron state using CINC parentstrength Military and economic strength of parent state using CINC relparentstrength Relative parent military and economic strength

relpatronstrength Relative patron military and economic strength

dfsbuildexp Number of times de facto state has experienced state building changes

dfsinstexp Number of times de facto state has experienced institution changes

tsincedfsbuildchg Number of months since last state building change tsincesdfsinstchg Number of months since last institution change

34 _{The CINC indicator uses six material capacity indicators of material capacity: total}

population, urban population, military personnel, military expenditure, steel and iron production as well as energy consumption (Singer et al. 1972).

35 _{I referred to Israel as Gaza’s parent state (rather than Palestine as Florea suggests in his}

109