**3. Methodology**

## **3.1 Model specification and estimation techniques**

The main purpose of this chapter is to examine the drivers of inflation the Malawian economy, with a special focus on its recent continuous decline since 2013, and also to assess whether this decline is a reflection of the economic fundamentals performance and whether this decline will persist in the short- to medium-term.

It makes use of the Phillips curve (aggregate supply) or a price-setting model that evaluates the effect of past and expected inflation, fiscal deficits, import prices, output gap (measured as output relative to its potential output<sup>2</sup> ) and exchange rate to capture the external effects of an open economy. Furthermore, the proposed model encompasses both the monetarists' and the structuralists' approach in determining factors that affect inflation, for instance the conventional output gap term is included to capture the rigidity of the labor market [17]. An increase in output will lead to a rise in labor demand as firms will need to hire more workers to further expand production. This increase in labor demand will lead to an increase in inflation as real wages and marginal costs rise. However, if labor supply is highly elastic, then the rise in real wages will be small, marginal cost will not rise significantly, and inflation will not move a lot in response to changes in output gap [18]. The model also takes into consideration full and immediate pass-through of imported prices (and hence exchange rate changes) into consumption prices. The exchange rate reflects the price effects of exchange rate changes on imported goods in the consumption basket which is common in small open economies like that of Malawi.

With reference to the effects of changes in money supply, literature reveals that one of the basic tenets of the quantity theory of money is that a change in the growth rate of money induces an equal change in the rate of price inflation [19]. Since nominal interest rates are based on real return and the expected rate of inflation, it suggests that the level of nominal interest rates should be positively correlated with average rates of inflation in the long-run. Furthermore, nominal

<sup>2</sup> Potential output was estimated using the Hodrick-Prescott (HP) filter.

## *Will Malawi's Inflation Continue Declining? DOI: http://dx.doi.org/10.5772/intechopen.91764*

they also observe that the Keynesian interest rate view of the monetary policy transmission mechanism does not apply to Malawi. Interest rates are found to affect inflation through the cost of production effect rather than through money supply effects. Also as is the case with Mangani [8] and Jombo et al. [9], these studies show that exchange rates have a much stronger effect on price levels in Malawi, reflecting the country's high level of openness and import dependence, making it highly vulnerable to foreign reserve situation due to the country's reliance on a narrow range of sources, most notably foreign aid and tobacco exports [13]. Matchaya [14] looks at the possible sources of inflation in Malawi and finds out that changes in money supply, exchange rates, past values of inflation, recessions and booms were the main determinants of inflation. These results are further supported by the findings from Simwaka et al. [15] where the results indicate that monetary and supply side factors drove inflation in Malawi over the period January 1995–March 2011. The study finds that money supply growth, exchange rate adjustments and decreases in output growth had a significant positive impact on inflation over the period. This result is also supported by Lungu et al. [16] which found that output gap has a negative impact on inflation over the period to some extent signifying the

*Linear and Non-Linear Financial Econometrics - Theory and Practice*

dominance of food prices in the consumer price index in Malawi.

output gap (measured as output relative to its potential output<sup>2</sup>

The main purpose of this chapter is to examine the drivers of inflation the Malawian economy, with a special focus on its recent continuous decline since 2013, and also to assess whether this decline is a reflection of the economic fundamentals performance and whether this decline will persist in the short- to medium-term. It makes use of the Phillips curve (aggregate supply) or a price-setting model that evaluates the effect of past and expected inflation, fiscal deficits, import prices,

to capture the external effects of an open economy. Furthermore, the proposed model encompasses both the monetarists' and the structuralists' approach in determining factors that affect inflation, for instance the conventional output gap term is included to capture the rigidity of the labor market [17]. An increase in output will lead to a rise in labor demand as firms will need to hire more workers to further expand production. This increase in labor demand will lead to an increase in inflation as real wages and marginal costs rise. However, if labor supply is highly elastic, then the rise in real wages will be small, marginal cost will not rise significantly, and inflation will not move a lot in response to changes in output gap [18]. The model also takes into consideration full and immediate pass-through of imported prices (and hence exchange rate changes) into consumption prices. The exchange rate reflects the price effects of exchange rate changes on imported goods in the consumption basket which is common in small open economies like that of Malawi. With reference to the effects of changes in money supply, literature reveals that

one of the basic tenets of the quantity theory of money is that a change in the growth rate of money induces an equal change in the rate of price inflation [19]. Since nominal interest rates are based on real return and the expected rate of inflation, it suggests that the level of nominal interest rates should be positively correlated with average rates of inflation in the long-run. Furthermore, nominal

<sup>2</sup> Potential output was estimated using the Hodrick-Prescott (HP) filter.

) and exchange rate

**3.1 Model specification and estimation techniques**

**3. Methodology**

**256**

interest rates and money growth rates are also expected to be positively correlated because of the positive correlation between average inflation rates and average money supply growth rates [18].

Fiscal deficits are one of the main factors that have been exerting inflationary pressures in most African countries. Especially when a country implements a regime of fiscal dominance or active fiscal policy, and passive monetary policy where monetary policy adjusts to deliver the level of seignorage required to balance the government's intertemporal budget [18]. In this case, the monetary authority is forced to generate enough seigniorage to satisfy the intertemporal budget balance condition. This will have an effect on prices and inflation since the changes in seignorage affect the current and future money supply.

An increase in the bank rate by the monetary authorities induces a rise in short term rates such as interbank rate and treasury bill rates which have an impact on other long-term lending rates. As a result of the rise in lending rates, both households and firms reduce their consumption and investment expenditures respectively, as real cost of borrowing increases. Households will mostly reduce their expenditures on consumption of luxury or durables goods due to the increased costs of borrowing. This leads to a decline in aggregate demand and consequently easing the inflationary pressures in the economy (see [20–22] among others for more details). Furthermore, when domestic interest rates increase relative to foreign interest rates, assuming uncovered interest rate parity, domestic currency depreciates in order to maintain equilibrium in the foreign exchange market of the domestic economy. This expected future depreciation induces an initial appreciation of the domestic currency making domestically produced goods more expensive than foreign-produced goods. Hence leading to a decline in net exports and aggregate demand as well as inflationary pressures. Also, the rise in domestic interest rates above foreign interest rates could also attract capital inflows, leading to an appreciation of the local currency. Thus, the precise impact of changes in exchange rate may be uncertain.

The general form of the model to be estimated can be represented as:

$$\inf\_{t} \, \_t = f\left( (m\mathbf{2}\_t), (\mathbf{i}\_t), \left( \mathbf{j}\delta\_t \right), (\mathbf{e}\_t), (m\_t) \left( \mathbf{y}\_t \right) \right) + \mathbf{e}\_t \tag{1}$$

where *inft* is the inflation rate, *it* is the interest rate, *fdt* is the fiscal deficit, *yt* is the output gap,*et* is the nominal exchange rate, and *εt*is the error term. It is important to note that there could theoretically be interrelationships between the chosen explanatory variables in the model hence having an impact on inflation through different channels. For instance, based on the conventional real interest rate channel, interest rates could affect the output gap *yt* and hence having an impact on inflation. The exchange rate would have a direct impact on domestic prices through its impact on the cost of imported goods and through wages, but also indirectly through its impact on output and net exports, hence affecting the inflation rate.<sup>3</sup> In this regard, this calls for caution in taking care of instances of autocorrelation and heteroscedasticity in the model.

Since the objective of the chapter is to examine the drivers of inflation both in the short- and the long-run, the ARDL framework is deemed appropriate for this analysis because of the advantages it has over other methodologies. Contrary to the traditional error correction methodology, where it is imperative to carry out and establish the stationarity of the variables to be used in the short- and long-run analysis to establish their order of integration, the ARDL methodology, while

<sup>3</sup> See [23] for more details.

utilizing the Bounds Test, does not require this pre-testing of the order of integration. It uses the F- and t-statistics to test the significance of the lagged variables in a univariate error correction system without establishing the order of integration of the data generation process underlying the series. However, it is important to establish the order of integration beforehand to ensure the absence of I(2) series since their presence would violate the properties of an ARDL model which requires variables to be only I(0), I(1) or they should be mutually integrated. The ARDL methodology also has an advantage over other methodologies in that the its parameters can be estimated consistently without invoking exogeneity and residual serial correlation, especially if the order of the ARDL is appropriately augmented by the suitable specification of the lag structure of the variables [25].

This being the case the objective of the chapter could therefore be investigated by using the ARDL model and Error Correction Model (ECM) frameworks. In this regard Eq. (1) can mathematically be specified as an ARDL model with *p* lags of *inf* and *q* lags of *X* (where *X* is a *kx*1 vector of independent variables, which include in this case money supply growth, lending rate, nominal exchange rate, import prices, fiscal deficits and output gap), ARDL ð Þ *p*, *q* as:

$$\|\mathbf{w}\|\_{t} = \sum\_{i=1}^{p} \theta\_i \left(\|\mathbf{w}\|\_{t-i}\right) + \sum\_{i=0}^{q} \mathbf{a}\_i' \mathbf{X}\_{t-i} + \mathbf{e}\_t \tag{2}$$

equilibrium level relationship is rejected at below 1% error level by the F-test

The Bounds test results in **Table 2** show that F-statistic has the computed value of 8.96 which exceeds the upper bound value of, I(1) which is 3.99 at 1% level of significance, implying that inflation rate and its determinants in the model are cointegrated and approach the long-run equilibrium, calling for the application of the ARDL approach [25]. The implication of this is that the parameters of the model can be estimated consistently without invoking exogeneity and residual serial correlation. The parameter stability test based on the plot of the cumulative sum of recursive residuals squares (CUSUM test) and the plot of the cumulative sum of squares of recursive residuals show that the estimated parameters of the ARDL specification are stable at least over the study period (**Figures 1A(a)** and **(b)** in the

**Augmented Dickey-Fuller test results**

Inflation 2.5932 0.0991 5.1882 0.0000

Lending rate 2.3613 0.1563 6.554 0.0000 Log(Import price) 0.8814 0.7884 3.5046 0.0107 Fiscal deficit 2.467 0.1277 12.2825 0.0000 Nominal exchange rate 1.1499 0.9976 6.9521 0.0000

**Levels (I(0)) First difference (I(1)) t statistics P-value t-statistics P-value**

> 2.5% 2.55 3.61 1% 2.88 3.99

**Variable Test statistics**

log(M2) growth) 3.1711 0.0261

**F-bounds test Null hypothesis: no levels relationship**

**Test statistic Value Signif. I(0) I(1)** F-statistic 8.956687 10% 1.99 2.94 K 6 5% 2.27 3.28

Output\_gap 6.1783 0.0000

*The results of the augmented Dickey-Fuller unit root test.*

*Will Malawi's Inflation Continue Declining? DOI: http://dx.doi.org/10.5772/intechopen.91764*

*Note: Values in parentheses are P-values.*

*Notes: k=no. of explanatory variables.*

*Bounds test results for co-integration relationship.*

**Table 1.**

**Table 2.**

The short-run analysis results in **Table 3** reveal that all the variables, except import prices, have a significant impact on inflation in the short-run. Money supply growth and fiscal deficits are found to have a significant negative impact on inflation, contrary to what the theoretical literature stipulates. However, this could be due to the tight monetary policy being exercised by monetary authorities in Malawi,

statistic (**Table 2**).

Appendix).

**259**

**4.1 Estimation results**

where *t* is the time period, *θ<sup>i</sup>* are *kx*1 coefficient vectors; *α<sup>i</sup>* are scalars and *ε<sup>t</sup>* is a disturbance term with a zero mean and constant variance. Eq. (2) can be reparameterized and expressed in error correction model form as:

$$\|\eta\|\_{t} = \mathfrak{Q}\|\eta\|\_{t-1} + \gamma' \mathbf{X}\_t \sum\_{i=1}^p \theta\_i(\Delta \|\eta\|\_{t-i}) + \sum\_{i=0}^q a'\_i \Delta \mathbf{X}\_{t-i} + \varepsilon\_t,\tag{3}$$

where ∅ is the speed of adjustment and *Δ* is the difference operator.

#### **3.2 Data**

The paper uses quarterly time series data for analysis for the period January 2001-June 2019. The data for all the variables is obtained from Malawi's Central Bank, the Reserve Bank of Malawi, except import price index which is obtained for the Economist Intelligence Unit database [24]. Real GDP were in annual frequency and had to be interpolated to transform them into quarterly data frequency.
