This study tests the long-run and short-run integration of maize markets in Malawi using the co-integration approach within the Vector Autoregressive modeling framework. The analysis is extended to Wald- F Granger Causality tests and innovation accounting to see the direction of causation between maize markets. A total of six maize markets, two from each region, were analyzed. Three are urban markets, while two of the three rural markets are border markets. The study uses monthly maize retail prices for the period January 2000 to May 2008. Study findings show that nine out of the fifteen market pairs are integrated in the long-run, but the degree of short-run market integration is low, implying that the transmission of price information is slow. Transaction costs seem to have a significant impact on the integration of market pairs involving border markets. Furthermore, there is no market that qualifies to be a central maize market in this study. The study concludes with a discussion of policy action to improve maize market integration and food security in Malawi

1. Maize Price Differences and Evidence of
Spatial Integration in Malawi: The Case of
Selected Markets
BY
LOVEMORE NYONGO
ECAMA RESEARCH SYMPOSIUM: LILONGWE.
8-10 OCTOBER 2014

2. PRESENTATION OUTLINE
• Introduction
• Motivation and study Objectives
• Literature review
• Sample description and data sources
• Estimation Techniques
Co-integration Analysis and Error Correction Model
• Results and interpretation
• Conclusion and policy implications

3. INTRODUCTION
• Efficient markets are key to achievement of food security in
countries where many people are net food buyers.
• Spatial market integration (SMI) becomes a useful tool in
allocating food within the economy.
• SMI refers to a measure of the extent to which demand and
supply shocks in one location are transmitted to another
location
• As such, competition among arbitragers ensures a unique
equilibrium where local prices in regional markets differ by no
more than transportation costs.
• In more integrated markets, farmers specialize in their
production, consumers pay less and the society benefits from
economies of scale.

4. MOTIVATION AND STUDY OBJECTIVES
• In their study, Chirwa and Zakeyo (2003) reported that 93.2
percent of farming households cultivated maize.
• The country’s CPI is dominated by maize
• Specifically, the study had the following objectives:
 To investigate the price transmission mechanism across selected maize
markets in the economy.
 To establish if there are central maize markets in the economy.
To assess the impact of transaction costs on maize market integration.

5. LITERATURE REVIEW
• In a competitive market economy, markets transmit
information that is useful in decision-making to economic
agents.
• Pricing signals regulate production, consumption and marketing
decisions over time, form and place (Kohls and Uhl, 1998).
• The price relationships between spatially separated markets are
generally analyzed within the framework of spatial price
equilibrium theory developed by Enke (1951), Samuelson
(1964) and Takayama and Judge (1964).
• The key assumption underpinning the theory is that price
relationships between spatially separated competitive markets
depend on the size of transaction costs.
• When the price difference between markets exceeds
transaction costs, arbitrage opportunities will be created.

6. SAMPLE DESCRIPTION AND DATA SOURCES
• The study analyzed monthly retail maize prices for 6
geographically separated markets from January 2000
• At least one commercial center (Mzuzu, Lilongwe and Limbe)
and one rural area in each region (Chitipa, Ntchisi and Muloza)
were included in the study.

7. ANALYTICAL FRAMEWORK
Unit root test to determine
order of integration (ADF)
Unit root test to determine order
of integration (ADF)
If I(k) If I(0)
Accept
Test null of no cointegration
btwn
prices at different markets
(Johansen or Engle and Granger)
Reject
• Notes: k>0, k is the order of integration
Estimate VAR model in 1st
differences, perform Granger
Causality tests and innovation
accounting
Specify and estimate (V) ECM to assess dynamics
and speed of adjustment, conduct Granger
Causality tests and innovation accounting.
Evaluation

8. Co-integration Analysis and Error Correction
Model
• The long-run equilibrium, according to the theory of law of one
price (LOP), is specified as:
Pi
t = β1 + β2Pj
t + εt (1)
• If εt is stationary and β2 is unity, then the markets are
completely integrated.
• In the study, equation 1 was modified to include
• variables found or assumed to influence market
integration
• natural logarithms within VA framework
• transaction costs (TC).

9. Co-integration Analysis and Error Correction
Model Cont’d….
ln å ln - å ln ln 2
p i
=b + a p + h p i
+b TC +e t t -
l
t t
n
l
l
n
i
j
i t i
= =
1 1
1
(2)
If co-integration is established, the relationship can be expressed
in an ECM which depicts the process of adaptation in the short
run. Johansen Co-integration test was conducted.
To determine the number of co-integrating relations in the
system, the study invoked the Johansen Trace test and Maximum
Eigenvalue. Failure to accept the null hypothesis of no co-integration
ln å ln å ln ln 1
i
t D p = b + a D p + q D p + l c +f TC +n -
i t t
j
t l
q
l
l
i
t l
q
l
l
=
-
=
1 1
1
confirmed the need to re-specify equation 2 as a VECM
as in equation 3 below:
(3)

10. Co-integration Analysis and Error Correction
Model Cont’d….
• To appreciate the impact of transaction costs, equation 2 and 3
were estimated with and without transaction costs.
• The Granger causality test was conducted to determine the
direction of price adjustment.
• Wald F-test was conducted for linear restrictions to find out if
one market’s lagged prices and transaction costs jointly
contribute to predictability of maize prices in another market.

11. RESULTS AND INTEPRETATION Cont’d…
• Johansen Co-integration Test Results
• 9 market pairs had 1 co-integrating relationship
• 6 market pairs had no co-integrating relationships
• Out of 5 market pairs involving Ntchisi, 4 indicate the
absence of a long-run relationship
• Out of 5 pairs involving Chitipa, 3 are not integrated.
• a VEC model (equation 3) with 1 co-integrating relationship
was estimated for the 9 co-integrated market.
• a VAR model (equation 2) was estimated for the market
pairs without cointegrating relationships.

12. RESULTS AND INTEPRETATION Cont’d…
• Impact of Transaction Costs
• have a significant impact on market integration, especially
on equations involving the border markets and those
markets with poor road network.
• short-run speed of adjustment between market pairs ranges
from 10 percent to 72 percent if transaction costs are
considered and 21 percent to 66 percent when transaction
costs are not considered.
• Government policies, licensing procedures, delays in
accessing price information and capacity constraints
pertaining to storage are important factors to consider.

13. RESULTS AND INTEPRETATION Cont’d…
• Granger Causality Tests
• No market is causing all other markets without being caused
by any of them.
• However, Muloza and Limbe seem to granger cause 5 and 4
other markets, respectively and, therefore, can be good
markets for policy intervention.
• Lilongwe seems to be a major supplier of maize to all three
regions because it is Granger caused by Mzuzu, Ntchisi,
Limbe and Muloza.

14. CONCLUSION AND POLICY IMPLICATIONS
• Short run integration is very low implying that it takes a longer
period for maize markets to respond to localized shocks.
• Policy makers should consider market infrastructure
development as a key priority to ensure linkages of maize
markets.
• Maize marketing in Malawi is complex and dynamic hence the
need to continuously study it.

15. END OF PRESENTATION
THANK YOU FOR YOUR ATTENTION!!!