How monetary policy affects industrial activity in Malawi: Evidence from ARDL and VAR models

Figures & data

Figure 1. Industrial sector productivity and contribution to GDP.

Table 1. Data description and sources

Variable	Series Used and Order of Integration	Source
industrial output, $ip$	industrial production index (base = 2010) - I(0)	IFS
consumer price, $cp$	consumer price index (base = 2010) - I(1)	IFS
inflation, $inf$	percentage change in CPI - I(0)	IFS
exchange rate, $ex$	Malawi Kwacha per US dollar exchange rate - I(1)	RMB
monetary base, $mb$	monetary base - I(1)	RMB
policy rate, $br$	RBM bank rate - I(1)	RMB
lending rate, $lr$	commercial banks’ average lending rate - I(1)	RMB
bank loans, $bl$	banking systems’ gross claims on private sector - I(1)	RMB
money supply, $m2$	broad money (m2) - I(1)	RMB

Figure 2. Industrial production and monetary policy trends.

Table 2. Descriptive statistics for monetary policy and industrial production

	bank rate	base money growth	industrial output growth
Mean	19.74	26.22	1.13
Median	18.54	24.63	12.79
Maximum	25.33	51.51	2016
Minimum	13.00	7.45	−29.62
Std.Dev	5.02	14.76	11.89

Table 3. ARDL and LSBP parameter estimates

	ARDL	LSBP
Dep. var : $i{p}_t$		2004/04–2013/12	2014/01–2016/5
$b{r}_t$	0.36 (0.99)	−0.30 (0.96)
$b{r}_{t-1}$	−3.49 (1.25)**	−3.63 (1.20)
$b{r}_{t-2}$	2.42 (0.89)**	3.00 (0.87)**
$m{b}_t$	0.16 (0.05)**	0.18 (0.06)**
$e{x}_t$	0.17 (0.08)*	−0.11 (0.10)
$in{f}_t$	0.00 (0.01)	0.01 (0.01)
$i{p}_{t-1}$	0.79 (0.08)**	0.67 (0.07)**
$i{p}_{t-2}$	−0.12 (0.10)	−0.12 (0.10)
$i{p}_{t-3}$	0.01 (0.10)	0.01 (0.10)
$i{p}_{t-4}$	−0.23 (0.07)**	−0.30 (0.07)**
$cons.$	1.73 (0.32)**	2.05 (0.54)**

Notes: Values in the parentheses are standard errors. ** and * mean coefficient is significant at 1% and 5%, respectively.

Table 4. ARDL conditional error correction model

Dep. Var: $\mathit{i}{p}_t$
		Coeff.	Std. Err
Long run
	$ECT$	−0.55**	0.07
	$b{r}_{t-1}$	−1.26*	0.63
	$m{b}_{t-1}$	0.29**	0.09
	$e{x}_{t-1}$	−0.31*	0.14
	$in{f}_{t-1}$	−0.00	0.02
	cons.	3.16**	0.39
Short run
	$\mathrm{\Delta }i{p}_{t-1}$	0.34**	0.07
	$\mathrm{\Delta }i{p}_{t-4}$	0.22**	0.07
	$\mathrm{\Delta }i{p}_{t-3}$	0.23**	0.07
	$\mathrm{\Delta }b{r}_{t-1}$	0.37	0.85
	$\mathrm{\Delta }b{r}_{t-2}$	−2.43**	0.85

Note: ** and * imply significance at 1% and 5%, respectively.

Table 5. H. M. Pesaran et al. (2001) F-Bounds test

${\mathit{H}}_o$: no level relationship							$\mathit{F}=9.827431$
	Kripfganz and Schneider (2018) critical values
	$\mathbf{10}\mathbf{\%}$		$\mathbf{5}\mathbf{\%}$		$\mathbf{2.5}\mathbf{\%}$		$\mathbf{1}\mathbf{\%}$
	I(0)	I(1)	I(0)	I(1)	I(0)	I(1)	I(0)	I(1)
$F$	2.2	3.09	2.56	3.49	2.88	3.87	3.29	4.37

Figure 3. Impulse response of industrial production to monetary policy shocks.

Table 6. Pairwise granger causality test

Null Hypothesis	F-statistic	p-value
$br$ does not Granger cause $ip$	9.43587	0.0001
$ip$ does not Granger cause $br$	0.31229	0.7323
$mb$ does not Granger cause $ip$	1.36623	0.2584
$ip$ does not Granger cause $mb$	0.18395	0.8322
$cp$ does not Granger cause $br$	6.67746	0.0017
$ex$ does not Granger cause $br$	5.88303	0.0035

Table 7. Forecast error variance decomposition for industrial production

Month	br	mb	ex	cp	ip
1	0.13	1.73	0.00	0.02	98.10
3	3.74	4.05	0.45	2.54	89.19
6	5.43	11.12	5.81	5.29	72.33
9	6.96	12.17	6.85	5.30	68.70
12	8.38	12.23	6.89	5.32	67.16
18	10.33	12.11	6.94	5.44	65.15
24	11.43	11.95	7.03	5.54	64.03
30	12.05	11.82	7.13	5.61	63.36
36	12.40	11.76	7.25	5.65	62.91

Figure 4. Impulse responses for the interest rate channel.

Figure 5. Impulse responses for the money supply channel.

Figure 6. Impulse responses for the exchange rate channel.

Table A1. VAR Lag Order Selection Criteria for the baseline model

Lag	LogL	LR	FPE	AIC	SC	HQ
0	−22.80307	NA	1.01e-06	0.380864	0.483042	0.422381
1	745.2263	1472.933	3.83e-11	−9.797620	−9.184551*	−9.548516*
2	780.1543	64.59283	3.34e-11	−9.933620	−8.809659	−9.476929
3	807.5899	48.85801	3.24e-11	−9.966985	−8.332132	−9.302707
4	831.9809	41.76536	3.29e-11	−9.958642	−7.812898	−9.086777
5	852.8053	34.23185	3.51e-11	−9.901442	−7.244806	−8.821990
6	873.5167	32.62753	3.78e-11	−9.842694	−6.675166	−8.555655
7	898.6482	37.86940	3.84e-11	−9.844495	−6.166076	−8.349870
8	925.6446	38.83054	3.84e-11	−9.871844	−5.682534	−8.169631
9	959.1425	45.88743	3.53e-11	−9.988253	−5.288051	−8.078453
10	990.4904	40.79527	3.38e-11	−10.07521	−4.864117	−7.957824
11	1027.947	46.17937	3.00e-11	−10.24585	−4.523864	−7.920876
12	1089.629	71.82160*	1.94e-11*	−10.74834*	−4.515467	−8.215783

* indicates lag order selected by the criterion

LR: sequential modified LR test statistic (each test at 5% level)

FPE: Final prediction error

AIC: Akaike information criterion

SC: Schwarz information criterion

HQ: Hannan-Quinn information criterion

Table A2. Roots of Characteristic Polynomial for baseline VAR (2) model

Root	Modulus
0.998642	0.998642
0.951352	0.951352
0.811676	0.811676
0.561987–0.264018i	0.620914
0.561987 + 0.264018i	0.620914
−0.328452	0.328452
0.051395–0.251406i	0.256605
0.051395 + 0.251406i	0.256605
0.233694	0.233694
−0.130116	0.130116

No root lies outside the unit circle.

VAR satisfies the stability condition.

Pesaran, H. M., Shin, Y., & Smith, R. J. (2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics, 16(3), 289–326. https://doi.org/10.1002/jae.616

 Web of Science ®Google Scholar

Kripfganz, S., Schneider, D.C. 2018 ardl: Estimating Autoregressive Distributed Lag and EquilibriumCorrection Models 2018 London Stata Conference London 6-7 September

 Google Scholar