The concept of political instability in economic research

When it comes to the economic implications of political instability, the topic that attracted the most attention from researchers was its impact on economic growth. The theoretical models of economic growth that incorporate political instability show that it has a detrimental effect on investment and thus on economic growth. Carmignani [2003] presents the following basic theoretical framework, which incorporates political instability in a growth model similar to the one proposed by Romer [1986]. A generic firm i has the following production function: Yi=A(Ki)α(KLi)1−α {Y_i} = A{\left( {{K_i}} \right)^\alpha }{\left( {K{L_i}} \right)^{1 - \alpha }} where Y_i denotes output, K_i and L_i are firm-specific inputs of capital and labor, K is the aggregate stock of capital in the economy, and A and α are parameters (A > 0 and 1 > α > 0). Importantly, the production function exhibits constant returns to scale in K_i and K, which yields a positive growth rate in a steady state.

Political instability enters the model through the expected profits function. The logic behind it is that political instability generates uncertainty over critical elements such as future economic policies (e.g., taxation), effective degree of enforcement of property rights and contracts, and the broader socio-economic environment (e.g., shutdowns, and strikes, among others). As a result, the generic company i is uncertain about the share of the output it will be able to appropriate. The expected profits function takes the following form: Eπi=(1−σ)Yi−wLi−(r+δ)Ki+σλYi−wLi−(r+δ)Ki=(1−σ+σλ)Yi−wLi−(r+δ)Ki E{\pi _i} = \left( {1 - \sigma } \right){Y_i} - w{L_i} - \left( {r + \delta } \right){K_i} + \sigma \lambda {Y_i} - w{L_i} - \left( {r + \delta } \right){K_i} = \left( {1 - \sigma + \sigma \lambda } \right){Y_i} - w{L_i} - \left( {r + \delta } \right){K_i} where E is the expectations operator, π_i denotes profit, σ (0 ≤ σ ≤ 1) is the probability firm i will be able to retain a fraction λ (0 ≤ λ ≤ 1) of output Y_i, w and r are unitary prices of labor and capital, and δ is the depreciation rate of capital. A representative household faces a standard intertemporal utility maximization problem (constant intertemporal elasticity of substitution). The solution of the profit maximization problem yields the following Euler equation: c˙c=1θ(r−ρ)=1θ[pαAL1−α−δ−ρ] {{\dot c} \over c} = {1 \over \theta }\left( {r - \rho } \right) = {1 \over \theta }\left[ {p\alpha A{L^{1 - \alpha }} - \delta - \rho } \right] where c denotes per capita consumption, ρ denotes the rate of time preference, and p ≣ (1 − σ + σλ) is an inverse index of political (policy) instability. The Euler equation, which yields the steady-state growth rate of all per capita variables (including output), shows that the larger is the level of political instability, the lower is the steady-state growth rate of capital. Lower investment, in turn, results in lower steady-state growth rates of output and consumption.

There are a number of other theoretical models, many of them being variations of the model summarized above, that investigate the relationship between political instability and economic growth. Svensson [1998] constructed a model that investigated the relationship between political instability, understood as propensity to government change, security of property rights, and investment. The process of reforming the legal system (which is necessary to increase the security property rights) takes time and resources. The incumbent government fully internalizes this cost, but not the future benefits. The model shows that in an environment of high political polarization and instability, the incumbent government does not have sufficient incentives to undertake a reform of the legal system, which leads to a lower equilibrium level of security of property rights, investment, and economic growth. Stevens [2000] focused on foreign direct investment and created a model with a representative firm maximizing its expected present value under uncertainty. The factors causing policy uncertainty are expropriation and shutdowns. Darby et al. [2004] constructed an endogenous growth model in which the aggregate output depends on the level of public investment which is financed though taxation. A high probability of government change reduces public investment and therefore economic growth. Annett [2000] constructed a model in which the society is fractionalized into different ethnic and/or religious groups and the government appropriates rents that are distributed to only some of these groups. This incentivizes the excluded groups to overthrow the government. A higher degree of fractionalization leads to a higher level of political instability, which in turn lowers investment and the output growth rate. Rodrik [1991] investigated the relationship between policy uncertainty and private investment. He argues that even otherwise desirable reforms may have a negative impact on investment if companies have doubts regarding their permanence, as policy uncertainty has similar effects to a tax on investment.

Research on the economic implications of political instability is not limited to economic growth. Another important strain of the literature investigates the relationship between political instability and seigniorage and inflation. The most influential theoretical study in this field is by Cukierman et al. [1992]. The hypothesis of this paper is that countries with more unstable and polarized political systems are characterized by less efficient tax systems (i.e., the costs of administering and enforcing regular taxes are high), as a higher perceived probability of government-change lowers the government's willingness to invest in reforming the tax system. As a result, more politically unstable countries must rely to a larger extent on seigniorage, which leads to higher inflation. Another important theoretical contribution to this topic is made by Edwards and Tabellini [1991a], who explore two political economy models of inflation – a model with a “myopic” government, in which inflation is a deliberate outcome of the strategic behavior of politicians, and a model with a “weak” government, where inflation is an inevitable result of a struggle between different political fractions. The work of Edwards and Tabellini builds on the previous work by Friedman [1969], Phelps [1973], Alesina and Drazen [1991], Végh [1989], Aizenman [1992], and others, who present a positive theory of inflation tax. Finally, Carmignani et al. [2008] argue that political instability can hamper central bank independence and thus indirectly lead to higher inflation.

There are also theoretical arguments that a high degree of political instability has a detrimental effect on fiscal deficits and public debt accumulation, which is attributed to the “strategic use of public debt.” The most important theoretical contributions are by Persson and Svensson [1989], Alesina and Tabellini [1990a, 1990b], Drazen [2000, Chapter 14], and Carmignani [2003, p. 10]. The concept of the strategic use of public debt can be briefly described as follows. If a government faces a high probability of being voted out of office in the upcoming election, it is not politically optimal to pursue an economically optimal fiscal policy. If the purpose of the incumbent government is to come back to power in yet another term, the optimal policy is to elevate public debt in order to “tie the hands of its successor” (so that it will not be able to pursue desired policies) and therefore decrease its chances of remaining in power. Another strain of research that supports the argument that political instability may increase public deficits and debt focuses on fragmentation of decision-making power. The bottom line of this literature is that governments that are supported by coalitions of parties, as well as minority governments, are more fiscally irresponsible, especially if the level of political polarization is high (parties forming the coalition have different ideologies and/or represent different groups of interest). A higher degree of political polarization results in a higher risk of government collapse and a higher uncertainty regarding future economic policies. Moreover, each party belonging to the coalition internalizes only a fraction of the cost of government spending, which is often referred to as the “tragedy of common goods.” Important theoretical contributions in this strain of literature are made by Alesina and Drazen [1991], Hallerberg and von Hagen [1997], and Velasco [1999].

Brunetti [1997] stated that early empirical research on the impact of political instability on economic growth was far from being conclusive. Out of 11 studies reviewed by him, only five reported an unqualified negative effect of political instability on economic growth. Three studies also showed a negative relationship, but this result was sensitive to the particular specification chosen, and three other studies showed that there is no statistically significant relationship between the two variables. Brunetti also pointed out that research on this topic faced several methodological problems that were not properly addressed in most of these studies. The first major issue is the problem of endogeneity, as the reviewed studies relied on the assumption that the direction of the causal relationship was from political instability to economic growth and not vice versa. This assumption can, however, be questioned, as one can easily imagine economic stagnation or recession being a source of social unrest that could potentially lead to political instability. Moreover, these studies used OLS regressions, which do not reveal the direction of causation and in cases of endogeneity will produce biased estimations. Another cause of concern, according to Brunetti, was the robustness of the results with respect to the specification of the growth equation, as most of these studies did not provide adequate sensitivity analysis. These concerns justified a need for further research in this area.

One of the most influential empirical studies on the relationship between political instability and economic growth is Alesina et al. [1996]. The authors address the problem of endogeneity by simultaneously estimating two equations – one for economic growth and one for political instability. The authors report a strong negative relationship between the two variables under investigation, with major government changes having a somewhat stronger impact than “normal” government changes, as an increase in values of the variables GCHANGE and MJCHANGE result in a decrease in the growth rate by 1.4% and 1.8%, respectively. These results are statistically significant. At the same time, the coefficients for the growth variables in the political instability equation are small and statistically insignificant, which suggests that the observed negative correlation between political instability and economic growth is the result of a high propensity to government change leading to low growth rates rather than vice versa. Gurgul and Lach [2013] used a very similar methodology to examine the nexus between political instability and economic growth in Central and Eastern European countries over the period 1990–2009. They found evidence of negative and statistically significant relationship between major government changes and economic growth. The evidence for regular government changes was only significant at the 10% level. The authors found no evidence of reverse causality.

Feng [2001] investigated the impact of political variables on private investment. The author provides empirical evidence that (1) political freedom facilitates economic growth, particularly through fostering human capital accumulation; (2) political instability, defined as variability of political freedom, has a detrimental effect on private investment; and (3) policy uncertainty negatively affects private investment. The study also shows that the transition toward democracy and openness alleviates the negative consequences of political instability. The author argues that such a transition can help create a better environment for economic development, ultimately boosting both economic growth and political stability.

Campos and Nugent [2002] argue that the empirical research on the political instability–economic growth nexus suffers from a variety of shortcomings, such as the ad hoc selection of explanatory and control variables, excessively narrow definitions of political instability, insufficient sensitivity analysis, and failure to conclude the direction of causation. The authors use PCA to construct two indices of political instability for 98 countries and non-overlapping 5-year periods from 1960 to 1995. The first index captures more “severe” forms of political instability (such as political assassinations and coups) and the other “moderate” or “lower-bound” forms (government transfers). Using the Granger causality framework and instrumental variables estimations, they conclude that there is no robust evidence of a long-term relationship between political instability and economic growth. They provide two explanations as to why their results conflict with the results of previous research. First, their analysis shows that the relationship is obtained only contemporaneously. Second, the relationship seems to be strongly region-dependent: Sub-Saharan Africa seems to be the driving force behind the conclusion that there is a negative relationship between the two variables, while for other regions there seems to be no evidence to support this hypothesis.

De Haan [2007] also strongly criticizes the latest empirical research on the political instability–economic growth relationship. He argues that it suffers from four fundamental problems. First, arbitrary model specifications – many studies do not include a rigorous sensitivity analysis. Second, possible sample heterogeneity – the author shows that with the frequently used ordinary OLS models, outliers can be a source of strongly biased results. Third, measurement errors – political instability is a latent variable and has various dimensions, which creates significant measurement problems. The author suggests that factor analysis should be used to alleviate this problem. Fourth, treatment of the time dimension – in many studies data coverage is short and divided into arbitrary periods (like in Campos and Nugent [2002], described in the previous paragraph).

Jong-A-Pin [2009] criticized research on the relationship between political instability and economic growth for not taking into consideration the different dimensions of political instability. He pointed out that measures of political instability suffered from measurement errors, which in turn severely affected the reliability of the estimates obtained. Jong-A-Pin used EFA to distinguish four separate dimensions of political instability, which were later jointly used in a dynamic panel econometric model with a GMM estimator and with lagged political instability variables used as instrumental variables (the purpose of which was to overcome the problem of potential endogeneity). The findings of the study are the following. First, instability of the political regime has a negative impact on economic growth. The author suggests that this might reflect the uncertainty of investors regarding the stability of property rights. Second, the study suggests that instability within the political regime is in fact good for economic growth. The author finds this result to be logical, as higher political competition allows the voters to vote incompetent politicians out of office and replace them with ones that introduce more appropriate economic policies. Third, the author finds some evidence for reverse causality, as lack of economic growth triggers politically motivated violence.

Aisen and Veiga [2013] found that the existing literature on the relationship between political instability and economic growth failed to answer some fundamental questions, namely: (1) what are the transmission channels from political instability to economic growth? and (2) how quantitatively important are the effects of political instability on the main drivers of economic growth? In order to answer these questions, the authors constructed linear dynamic panel data models with a system-GMM estimator based on a sample of 169 countries and 5-year periods from 1960 to 2004. The number of “cabinet changes” and six indices of political instability were used as proxies for political instability. The results of the study are the following. Political instability adversely affects economic growth and, in the authors’ opinion, the results are “strikingly conclusive,” as an additional government change per year reduces the average GDP per capita growth rate by 2.39 percentage points. The main channel of transmission is through a lower total factor productivity (TFP) growth rate and, to a lesser extent, through a negative effect on physical and human capital accumulation. The Aisen and Veiga do not find any evidence that would suggest that political instability has a different impact on economic growth in democratic or autocratic regimes.

Section 3 presented the article by Cukierman et al. [1992], which is the most important theoretical contribution in this strain of the literature. In addition to the theoretical model, the article also presented an empirical model, which is described as follows. First, the authors use a probit model to estimate the probability of government change. Next, they estimate five cross-country regressions with the estimates obtained from the probit model as an explanatory variable and seigniorage as the dependent variable. The authors then re-estimate these regressions using the instrumental variables method. In all 10 regressions, the variable for political instability is statistically significant at the 1% level and has the expected sign; therefore, the authors conclude that a higher degree of political instability leads to a higher reliability on seigniorage.

The relationship between political instability and inflation and seigniorage was revisited by Aisen and Veiga [2006, 2008a, 2008b]. In their 2006 study, the authors present evidence that political instability leads to significantly higher inflation, as an additional government crisis per year increases the inflation rate by 16.1% and an additional government change per year increases it by 9.1%. A government crisis is defined as “a number of rapidly developing situations in a year that threaten to bring the downfall of the present regime, excluding situations of revolt aimed at such overthrow” and was taken from the CNTS Database. In their 2008 article, the authors present evidence that a higher degree of political instability and social polarization result in the higher volatility of inflation rates [Aisen and Veiga, 2008a]. Moreover, political instability has a stronger effect on inflation volatility in developing countries, which are characterized by a lower degree of central bank independence and economic freedom. In their second study from 2008 [Aisen and Veiga, 2008b], they investigate the impact of political instability on seigniorage. The authors argue that inflation and seigniorage are two separate phenomena and should be studied separately, even though they are to large extent correlated. They present data showing that the correlation between inflation and seigniorage declines as inflation increases and becomes negative with inflation rates above 400% per year. Moreover, the determinants of inflation might not be the same as seigniorage, as inflation might be a result of a supply side shock such as an increase in oil prices. The authors find that the impact of political instability on seigniorage is higher in an environment of high inflation and in developing countries (as opposed to high-income countries). Other factors that strengthen the observed relationship are high political polarization, history of political instability, high central bank governor turnover (a proxy for central bank independence), a lower level of economic freedom, a higher degree of authoritarianism, high domestic debt levels, limited access to international financing (measured by credit ratings), and low openness to trade.

Empirical evidence on this issue was presented by Grilli et al. [1991], Franzese [1998], Lambertini [1998], De Haan et al. [1999], Petterson [1999], and Carmignani [2000]. This evidence is far from being conclusive, which might be connected to the different methods of measurement and quantification of political instability used in these studies, but only to some extent. Grilli et al. [1991], Franzese [1998], and De Haan et al. [1999] used similar indicators of political instability – government duration in office and/or frequency of government turnover – but reached conflicting conclusions. Grilli et al. found a positive correlation between fiscal deficits and instability, Franzese found no significant effect of instability on accumulation of public debt, and De Haan et al. found a positive correlation between political instability and debt growth. Lambertini [1998], Petterson [1999], and Carmignani [2000] also used similar indicators – probability of defeat of the government – and also reached conflicting results. Lambertini found no significant effect of political instability on public debt accumulation, Petterson found a positive correlation when right-wing governments are in office and a negative correlation when left-wing governments are in office, and Carmignani found some evidence for a positive correlation between political instability and fiscal deficits. These conflicting results suggest that more research is needed in order to discover the true relationship between political instability and public sector deficits and debt accumulation.

Recently, the topic that has been attracting increasing attention among researchers is the impact of political (policy) instability on the functioning of the financial markets. Hartwell [2018a] argues that there are two channels through which political instability affects the functioning of financial markets. The first channel, which has a direct influence, is the execution of monetary policy. The second, indirect channel, is through uncertainty regarding future economic policies and institutions. Hartwell argues that the second channel is related to both the uncertainty regarding the results of regularly scheduled elections and informal instability.

There is an extensive and growing body of empirical research on this topic. Thorbecke [1997] demonstrates that monetary policy affects the level of asset returns. Bernanke and Gertler [2012] provide evidence that uncertainty regarding monetary policy increases the volatility of returns. Carmignani et al. [2008] show that political instability can negatively affect the central bank's independence and Papadamou et al. [2014] present how central bank independence affects stock market volatility, establishing a causal relationship between political instability and financial markets’ volatility. Engle and Ng [1993] present the concept of news impact curve and use it to investigate the behavior of the Japanese stock exchange between 1980 and 1988. The analysis shows that the stock exchange is very sensitive to news, but the sensitivity is asymmetrical, as negative news tends to produce more volatility than positive news. Białkowski et al. [2008] and Goodell and Vähämaa [2013] show that market volatility increases around election dates, which is connected to uncertainty regarding the election results. The effect may be intensified by several factors such as a narrow margin of error, change in the political orientation of the government, or a failure to form a government with the support of a parliamentary majority. Arin et al. [2013] employed a Bayesian Model Averaging method to investigate the impact of political variables (including government changes and political fractionalization) for financial volatility in 17 parliamentary democracies and found evidence that while the impact of political variables on excess returns is weak, political variables have a significant impact on return volatility. Hartwell [2018a] used a novel and comprehensive database containing monthly data on political changes in transition economies to construct a GARCH model investigating the impact of political instability on capital markets. His analysis shows that informal political instability has a strong negative effect on stock returns and formal political institutions increase financial volatility to a larger extent than monetary policy. Hartwell [2018b] investigated the determinants of financial volatility in Central and Eastern Europe and countries which formed part of the erstwhile Soviet Union. He shows that institutional changes (in particular of crucial institutions such as property rights) have been a major driver of financial volatility. Compaoré et al. [2020] provide evidence that political instability leads to an increased probability of banking crises. The main transmission channel is occurrence of fiscal crises. Moreover, political instability in one country may result in spillover effects to the banking systems in neighboring countries.