Missed opportunities and missing markets: Spatiotemporal arbitrage of rice in Madagascar

1
Missed opportunities and missing
marketsSpatio-temporal arbitrage of rice in
Madagascar     

• Christine Moser, Chris Barrett and Bart
  Minten May 18, 2005
• World Bank seminar

2
Motivation

• Markets are crucial for
• -transmitting macro/sectoral policy signals to
  micro-level agents
• - maintaining incentives in the face of
  technological change
• - managing locally covariate risk
• Yet we know surprisingly little about
• - how markets function
• - the correlates of markets which are
  competitive, non-competitive, or segmented
• - scales at which policy interventions ought to
  occur
• This paper addresses these gaps using data from
  Madagascar.

3
Competitive marketing equilibrium conditions with
trade
Where r is rice price, p is paddy price, t is
spatial marketing margin, p is storage and
interest costs, and a is milling costs. Thus,
4
But competitive equilibrium doesnt always hold
Baulch (1997 AJAE) and Barrett and Li (2002 AJAE)
develop appropriate mixture distribution-based ML
methods parity bounds model (PBM) for
estimating the probability that markets are in
one or another regime.
5
PBM (Parity Bounds Model)
Price Difference (absolute)
 
Quarter 1 Quarter 2 Quarter 3
Quarter 4
6
Policy Implications Vary By Regime

• Regime 1 laissez faire, perhaps work to reduce
  marketing margins
• Policy interventions most appropriate under
• Regime 2 marketing margins a binding constraint
  on market-based transmission of policy signals,
  diffusion of local risk, incentives to adopt
  improved technologies.
• Regime 3 market competition a constraint on
  market-based transmission of policy signals,
  diffusion of local risk, incentives to adopt
  improved technologies.

7
Market Integration

• Methods
• Model of trader entry. Basic model of trader
  entry into a market given prevailing transport
  costs, access to finance, and other commune
  characteristics.
• Estimation of Parity Bounds Model (PBM) by MLE.
  This produces the probability that the market is
  integrated. The PBM is used to test spatial and
  intertemporal integration as well as integration
  across space, time and form.
• Explaining non-integration. Using the PBM
  results, estimate a multinomial logit to examine
  the factors affecting whether a market is
  integrated (Regime 1), non-trading (Regime 2), or
  non-competitive (Regime 3).

8

• Data Source
• 2001 commune census

• Rice paddy prices for 4 quarters
• Rainy dry season transport costs for products
  to different destinations.
• Also imported rice prices, whether commune
  exports rice to other areas, access to public
  goods and services, types of economic activity,
  and climate variables.

 
9
Three scales of spatial market integration
National
10
Quarter 2 harvest
Quarter 1 pre harvest
Rice price/kg in FMG
11
The Commune Census
Dry season cost of transporting 50kg to major city
12
Results Spatial Market Integration
Table 4 Spatial PBM Estimation Results
13
Explaining Spatial Market Integration
14
Intertemporal Market Integration (between the
second and third quarters)
Table 7 Intertemporal PBM estimation results
15
Explaining Intertemporal Market Integration
16
Interseasonal flow reversals
Harvest Season
 
17
Price Difference (local city)
18
(No Transcript)
19
Market integration across space, time and form
20
Summary of Results and Implications

• Markets are fairly well integrated spatially and
  across time, space and form at the
  local/sub-regional level.
• At regional level, markets are largely
  non-competitive or in disequilibrium need
  competition policy and foster market entry by
  traders (info, finance, crime, etc.)
• At the national level, transportation costs are
  generally so high as to preclude spatial
  arbitrage transport key.
• Interseasonal storage is generally unprofitable
  due, it appears, to high unobserved
  storage/interest costs.
• Crime and isolation tend to reduce spatial
  integration.

21
Thank you for your comments and interest
22
(No Transcript)
23
(No Transcript)
24
N4716 c2528.1 LL value -3576.85 Pseudo
R20.069
25
Table 9 Difference between sub-region price and
commune price

26
(No Transcript)
27
(No Transcript)
28
Communes for which walking or traveling by ox
cart is necessary