Determinants of Inflation in Somalia: An ARDL Approach

Inflation, recognized as a crucial monetary phenomenon, has attracted significant global attention due to its persistent nature. Conceptually, inflation is characterized by a continuous and sizable increase in price levels, which results in a significant diminution of purchasing power [1]. This phenomenon triggers consequential economic behaviors, prompting individuals to allocate substantial financial resources towards consumption activities. This consumption orientation places downward pressure on savings and investments, amplifying unemployment rates and hindering overall economic growth.

In addition to its national implications, inflation casts a wide influence over the international economic landscape. The phenomenon introduces a unique disruption to a country's trade dynamics, principally by raising the valuation of domestic commodities above the standard benchmarks of regional and global market prices. As a result, a country's competitive position in international trade experiences a setback [2]. This heightened pricing structure cultivates an environment where the ability to compete effectively in the global market is jeopardized.

In essence, the significance of inflation as a monetary phenomenon is underscored by its pervasive impact, prompting careful consideration among global economists. The relentless rise in prices, coupled with subsequent impacts on purchasing power, not only shapes the dynamics of domestic economies but also reverberates across the complex web of international trade relations.

In academic discourse, inflation is often portrayed as a unique form of taxation that operates without legislative intervention [3]. It is generally regarded either as a monetary phenomenon [4], or as a discernible pattern of rising prices, as outlined by references [5-7].

The escalation of conspicuous inflation rates across Africa, including Somalia, has stirred a comprehensive investigation of the underlying causes. The persistent presence of price inflation emerges as a significant concern that deeply affects various economic entities. Ensuring the stability of general price levels, a critical macroeconomic goal across nations, is essential for sustainable economic development.

Focusing on Somalia, recent years have seen a marked challenge posed by inflation. The country's susceptibility to heightened inflation is rooted in a combination of factors such as political instability, conflict, and lack of economic investment. The depreciation of the Somali shilling exacerbates this phenomenon, propelling the upward trajectory of prices. The effects of inflation negatively impact living standards in Somalia, making essential commodities increasingly inaccessible for a substantial portion of the population.

In response, the Somali government has initiated a series of measures to alleviate the adverse effects of inflation. These strategies include efforts to stabilize the currency and implement monetary policies to control the money supply. However, the effectiveness of these measures continues to be hindered by the enduring economic and political challenges that the nation faces.

Inflation in Somalia's economic history followed a relatively stable trend until 1992, as illustrated in Figure 1. However, between 1992 and 2000, inflation experienced a significant surge due to the introduction of counterfeit currency by certain merchants, warlords, and regional state administrations. In 2001, inflation markedly decreased, possibly attributed to the citizenry's optimism about the new administration's commitment to curbing illicit activities related to money.

1.png

Figure 1. Inflation in Somalia

Source: Author's calculation

From 2002 to 2007, Somalia experienced a substantial surge in inflation, likely influenced by increased spending by warlords, Islamic court authorities, faction leaders, and the introduction of counterfeiting machines. In contrast, inflation declined from 2008 to 2010, due to steady remittances, streamlined government processes, and the effect of dollarization.

Addressing the escalating inflation in Somalia necessitates coherent policy strategies, yet the challenge is compounded by the need for extensive literature. Particularly, there is a notable demand for empirical studies investigating the determinants of inflation within Somalia. Most existing research on inflation determinants in developing economies excludes Somalia. Thus, it is crucial to include Somalia in nation-specific and cross-country empirical studies. This study aims to bridge this knowledge gap and contribute to the broader global conversation on the factors influencing inflation.

The paper is structured as follows: Section 2 reviews the relevant literature, Section 3 outlines the research methodology and data, Section 4 presents the study's findings, and Section 5 concludes with policy implications.

3.1 Data

Data from 1970 to 2010 were used to construct a time series for analysis. The set of readily available data establishes sampling observations. The data was obtained from the Organization of Islamic Cooperation (OIC - SESRIC) and the World Bank. The variables used in this study are GDP deflator (a measure of inflation), money supply, exchange rate, and gross domestic product (GDP).

Among the pantheon of variables, the selection we present bears the mantle of being the most fitting contender to portray the underlying determinants of inflation within the context of Somalia. The rationale behind this discerning choice is firmly rooted in the pragmatic reality that several other pertinent variables, most notably the interest rate, remain conspicuously absent from the realm of available data within Somalia's data ecosystem. Consequently, the meticulously curated variables stand as the lodestones that steer our analysis, illuminating the intricate web of inflation dynamics within this distinctive economic milieu (Table 1).

Table 1. Definition of variables

3.2 Model specification

Titled "Determinants of Inflation in Somalia", this study delves into various factors. In this exploration, we will use inflation as the key focus, the variable we are trying to understand. We will also consider the Exchange rate, Money supply, and Gross Domestic Product as our explanatory factors.

Our primary goal here is to provide a more precise measurement of how Somalia's money supply (MS), GDP, and exchange rate (EXR) influence inflation. By doing this, we aim to grasp better how inflation works in the short and long term for the country. To help us with this, we will use an approximate theoretical model as our framework.

$\mathrm{INF}=(\mathrm{MS}, \mathrm{GDP}, \mathrm{EXC})$                                           (1)

The above regression model was translated into a regression equation as stated below:

$\mathrm{INF}=\beta_0+\beta_1 \mathrm{MS}+\beta_2 \mathrm{GDP}+\beta_3 \mathrm{EXC}+\mu$                                          (2)

where,

INF= Inflation

MS = Broad Money Supply

GDP = Gross Domestic Product

EXG = Exchange Rate

$\beta_0$ = Intercept

$\beta_1, \beta_2$, and $\beta_3$  = are coefficients of the explanatory variables, and each, as expected ≠ 0

μ = is Stochastic error term.

The variables are employed in their log form to avoid problems with non-normality, heteroskedasticity, and mis specified functional form. The model is as follows:

$\mathrm{LINF}_{\mathrm{t}}=\beta_1 \mathrm{LMS}_{\mathrm{t}}+\beta_2 \mathrm{LGDP}_{\mathrm{t}}+\beta_3 \mathrm{LEXC}_{\mathrm{t}}+\mu_t$                                 (3)

This investigation utilizes two analytical methodologies: the Autoregressive Distribution Lag (ARDL) and the Error Correction Model (ECM). In order to ensure the suitability and reliability of the ECM before its implementation, a series of rigorous Unit Root and Cointegration tests are planned. This preliminary phase is geared towards establishing the foundation of our time series data generation process.

To elaborate, our primary objective is to ascertain whether the variables outlined in Eq. (1) exhibit a consistent stationary behavior or if they exhibit fluctuations. This pivotal step is critical in shaping the groundwork for subsequent proficient econometric modelling endeavors. To this end, we have employed the Augmented Dickey-Fuller (ADF) test, drawing upon Eq. (2) as our reference point, to scrutinize the presence of non-stationary tendencies meticulously.

$\Delta \mathrm{X}_{\mathrm{t}}=\beta_0+\beta_1 \mathrm{X}_{\mathrm{t}-1}+\sum_{i=1}^n \alpha_i \Delta X_{t-1}+\varepsilon_t$                                    (4)

where, n is the number of lags incorporated into the model, meaning: $\varepsilon_t \sim \operatorname{NID}\left(0, \sigma^2\right)$ and X_t reflects each variable in Eq. (4) for MS, EXR, GDP, and INF.

In pursuit of our defined goal, the ARDL model emerged as a pivotal tool in this study. ARDL bounds testing has exhibited marked advantages over conventional cointegration tests across several dimensions. A distinctive trait of ARDL lies in its versatility, accommodating variables of differing orders of integration – be it I(0), I(1), or a hybrid thereof [7]. Moreover, the ARDL approach adapts adeptly to modeling frameworks, flexibly incorporating an optimal number of lag terms that capture the essence of data generation.

One of ARDL's distinguishing attributes lies in its ability to harmonize short-term adjustments with long-term equilibrium dynamics by formulating an error correction mechanism (ECM). This synthesis is accomplished through a streamlined linear transformation, effectively encapsulating the underlying interaction without obfuscating excessive long-run information. ARDL's efficacy is notably accentuated compared to alternatives like the Johansen and Juselius methodology, particularly in small sample settings.

Within the framework of a multivariate model, the intricate interplay between Inflation, Broad money supply, Gross Domestic Product, and Exchange rate assumes the following delineation:

$\mathrm{LINF}_{\mathrm{t}}=\beta_0+\beta_1 \mathrm{LMS}_{\mathrm{t}}+\beta_2 \mathrm{LGDP}_{\mathrm{t}}+\beta_3 \mathrm{LEXC}_{\mathrm{t}}+\varepsilon_t$                          (5)

where, LINF_t is the Log of inflation, LMS_t is the log of money supply, LGDP_t is the log of gross domestic product, LEXC_t is the log of exchange rate and εt is the error term. We employed the ARDL model to assess the short- and long-run cointegration between the explained and explanators. Based on the empirical findings [6], the mathematical expression for ARDL cointegration can be written as follows:

$\begin{aligned} & \Delta \mathrm{LINF}_{\mathrm{t}} =\alpha_0+\beta_1 \mathrm{LINF}_{\mathrm{t}-1}+\beta_2 \mathrm{LMS}_{\mathrm{t}-1}+\beta_3 \mathrm{LGDP}_{\mathrm{t}-1}  +\beta_4 \mathrm{LEXC}_{\mathrm{t}-1}+\sum_{\mathrm{i}=0}^{\mathrm{q}} \Delta \alpha_1 \mathrm{LINF}_{\mathrm{t}-\mathrm{k}} \sum_{\mathrm{i}=0}^{\mathrm{p}} \Delta \alpha_2 \mathrm{LMS}_{\mathrm{t}-\mathrm{k}}  +\sum_{\mathrm{j}=0}^{\mathrm{p}} \Delta \alpha_3 \mathrm{LGDP}_{\mathrm{t}-\mathrm{k}} \sum_{\mathrm{i}=0}^{\mathrm{p}} \Delta \alpha_4 \mathrm{LEXC}_{\mathrm{t}-\mathrm{k}}+\varepsilon_{\mathrm{t}-\mathrm{k}}\end{aligned}$  (6)

In the equation, α0 represents the constant term, while α1 – α4 are the short-term variables' coefficients. Similarly, β1 to β4 signify the long-term elasticity parameters. The parameters q and p denote the optimal lags for the explained and explanatory variables. The symbol Δ indicates the first difference, reflecting the short-term variables, and εt represents the error term. The ARDL cointegration methodology initiates with bound testing.

The null hypothesis (H0) posits that β1 = β2 = β3 = β4 = 0, suggesting the absence of cointegration among the long-term variables. Conversely, the alternative hypothesis (H1) asserts that β1 ≠ β2 ≠ β3 ≠ β4 ≠ 0, indicating the presence of cointegration among these variables. Evaluating the null hypothesis involves employing Critical values and Wald-F statistics. If the calculated Wald-F statistics surpass the upper bound of critical values, it signifies the rejection of the null hypothesis, signaling a long-term connection among the variables. Conversely, if the Wald-F statistics fall within the critical bounds, it suggests a lack of long-term linkage between the variables.