Printer Friendly

Determinants of market integration and price transmission in Indonesia.

I. Introduction

The recent wave of high international commodity prices has increased interest in understanding the spatial market integration of domestic markets with both international and other domestic ones. The sharp increase in prices, at the same time, constitutes an important opportunity as well as a challenge for natural resource abundant economies like Indonesia. The degree to which Indonesian producers can profit from this opportunity depends firstly on how integrated domestic markets are with world markets, i.e., on how closely domestic prices move with world prices, and secondly on how integrated the different provincial markets are with each other.

There are several reasons why market integration matters for development. Non-integrated markets are to some extent "blind". Producers are not able to "see" what is highly appreciated in world markets and what is not--and therefore are unable to make the best possible decision --lead to inefficient outcomes. (1) Weak integration implies weak domestic supply responses to higher commodity prices. Market integration allows for spreading fixed costs; productivity improvements through specialization; and reaping the benefits of scale economies. Furthermore, better integrated markets contribute to development by helping to stabilize prices. (2) In integrated markets, the effects of local disturbances on prices are mitigated through induced trade between surplus and deficit areas. (3) Related to this, and as argued by Timmer (1986), if markets are integrated, price intervention as a tool of public policy can be powerful. Governments or parastatal agencies could, by operating on small amounts of easily controlled trade flows, affect the entire price structure for the commodity produced and consumed within the country through arbitrage as low-priced commodities find their way to high-priced markets. If markets are isolated instead, intervening to stabilize prices is more costly as these arbitrage mechanisms do not operate and governments may need to intervene in each of the isolated markets. In Indonesia, a country consisting of over 17,000 islands and where the government has an explicit commitment to food price stability, the benefits of market integration cannot be overstated.

As important as understanding the degree of integration within Indonesia, given its peculiar geographical characteristics, is to also understand the factors that explain why some provinces are strongly spatially integrated while others are weakly or not integrated at all. Surprisingly, the analysis of determinants of market integration has generally been neglected in the literature and not much is known about it. In addition, and to the best of our knowledge, there has been no systematic analysis of the determinants of food commodity price differentials.

This paper contributes to the literature by providing new evidence on two separate but related issues: the determinants of market integration across provinces in Indonesia, and the determinants of price differentials across provinces by looking at five commodity markets: rice, soybeans, maize, sugar, and cooking oil.

To analyse market integration determinants, this paper starts by assessing the degree of integration of these markets among provinces in a geographically heterogeneous country like Indonesia. To do so, a definition for market integration needs to be introduced. Conceptually, spatially separated markets of a homogeneous commodity are considered to be integrated when they are connected by trade in such a way that the forces of arbitrage make the long-run price differences that may remain among them reflect transportation costs only. This implies that shocks arising in one region are transmitted to another region. (4)

To make this conceptual definition operational, the paper tests for a common stochastic long-run trend between pairs of provinces using Johansen cointegration techniques. Then, it explores what the determinants of price differences are across provinces and of market integration.

In particular, it attempts to answer the following questions:

Spatial Integration: Are Indonesian provincial commodity markets spatially integrated? Is there a significant degree of co-movement between provincial commodity prices?

Determinants of Price Differences and Spatial Integration: What are the factors that explain price differences across provinces? What are the factors that explain spatial integration and in particular, what is the role of logistic costs? Does "self-sufficiency", or more output produced imply less integration across provinces?

This document is organized as follows: section II summarizes the literature on determinants of spatial integration in commodity markets. Section III describes the dataset. Section IV presents the methodology for testing market integration among Indonesian provinces and the main findings for the five commodities considered. Section V presents the methodology and main findings of the analysis of determinants of price differences, and of market integration across Indonesian provinces. Section VI concludes and draws policy implications.

II. Literature Review

This paper builds upon the question of market integration and geographical price differences by examining the factors that influence these issues. Though inferences related to the drivers of revealed patterns of integration are very informative, these are generally lacking in the literature, which have typically focused on measuring spatial market integration instead. Reviewing the literature on spatial integration measurement is out of the scope of this paper, but Table A1 (Appendix) summarizes the key contributions. (5) The focus of this section is about the literature on determinants of market integration and geographical price disparities.

To provide some structure to this section, two questions guide this review: (a) does the study analyse the determinants of market integration in a systematic fashion, i.e., using regression techniques? (b) Does the study refer to the Indonesian context? (6)

II.1 Non-systematic--Non-Indonesian

Distance between markets has been acknowledged as an important factor affecting market integration. It is common to find in the literature measures of market integration tabulated against markets' distances. However, in most of the cases, no formal empirical analysis of the links is carried out (for these types of informal analyses, see Ravallion 1986; Goodwin and Piggott 2001; Rashid 2004; and Van Campenhout 2007).

Baquedano, Liefert, and Shapouri (2011) look at world market integration for export cash crops (cotton and coffee) and imported food crops (rice) for Mali and Nicaragua and relate the relatively lower degree of market integration and slower price transmission in the African country to institutional differences related to policy decisions. In the case of Mali, where a parastatal enterprise controls cotton production and trade, the government of Nicaragua does not intervene substantially--directly or indirectly--in agriculture.

In a recent contribution, Sekhar (2012) examines spatial integration for selected commodities markets in India using cointegration techniques. The author reveals an interesting pattern in the results associated with existing variations across-commodity in trading institutions. Commodity markets that do not face restrictions to trade, either inter-state or inter-region (gram and to a lower extent edible oils), appear well-integrated both regionally and nationally while those in which inter-state movement restrictions prevail (such as rice) do not show integration at a national level.

II.2 Non-systematic--Indonesian

Marks (2010) provides a historical account of rice market integration in Indonesia over the period 1920-2006 among different cities across the archipelago by using cointegration techniques. The author considers different sub-periods and carefully puts the statistical result in the institutional context of the period, thus hinting at possible causal links. The author argues that while price fluctuations were limited and markets worked "relatively well" during the colonial administration era, the period during the Second World War and the subsequent independence struggles increased uncertainty, devastated infrastructure, and experienced price instability and lower degrees of market integration. High price stability and increased market integration occurred after Suharto came to power, large investments in infrastructure were made, and Bulog was set up (in 1969) with a mandate to stabilize food prices (among others). More recently, the author argues, rice markets remained integrated, although the speed with which price shocks are transmitted spatially has lowered.

II.3 Systematic--Non-Indonesian

The works of Goodwin and Schroeder (1991), Goletti, Raisuddin, and Farid (1995), and Escobar and Cordano (2008) formally address the question of the determinants of integration in a non-Indonesian context. Common to the first two papers is their two-step approach. Firstly, they measure spatial market integration in their relevant geographical setting, and secondly, they regress the measure of market integration on a number of explanatory variables. The last one proceeds in a one-step approach.

Goodwin and Schroeder (1991) use five different cointegration tests to measure integration in livestock markets in the United States over four different periods (from 1980 to 1987). They obtain one test statistic for each pair of markets analysed, and for each period considered. These test statistics are then used as a dependent variable in the second stage. They consider four factors affecting integration: the costs and risks associated with trade between markets (distance between markets); the amount of market information reflected in prices at a particular market (whether the market is a "terminal" market or not); the market volume; and the degree of concentration in the packing market. Their results reveal that distance, as expected, is a significant deterrent of market integration. In addition, they found that concentration in the meat packing market increased the degree of market integration. This result is interesting as the exercise of market power (through, say, the ability of carry out price discrimination) is sometimes argued to decrease the degree of integration of markets. Instead, the authors claim that when firms operate plants in spatially separated markets, transaction costs and uncertainty about market outlets for cattle shipped from one region to another is reduced and it could also facilitate price behaviour coordination among meat-packers across regions.

Goletti, Raisuddin, and Farid (1995) examine rice market integration and its determinants in Bangladesh for the period 1989-92 in sixty-four districts. To measure market integration, the authors combine correlation coefficients on the price series with cointegration coefficients, dynamic multipliers (which measure how much of a shock in market i is transmitted to market j in k periods), and measures of the speed of adjustment (how many periods it takes for a shock in market i to be fully transmitted to market j). Then, they regress these measures of market integration on the hypothesized determinants. Three broad structural determinants of market integration are considered: marketing infrastructure (road distance between markets, density of paved roads, railway infrastructure, number of strikes in the area, telephones per capita, and density of bank branches); volatility of policy (variation coefficient of the stocks that the government agency keeps in each district at the end of the month); and dissimilarity of production (absolute value of the percentage difference in production per capita). The authors' findings revealed that the distance between markets, telephone density, and labour strikes affected integration negatively. Instead, integration was positively affected by more dissimilarity in production and road density (as both factors encourage trade).

Finally, Escobal and Cordano (2008) attempt to gauge the effect of investment in infrastructure on market integration for potatoes using data from ten cities during the period from January 1995 through May 2001 via a one-step approach. The authors test for market integration and the effects of transaction costs in the context of a threshold cointegration approach and find that distance matters for integration, as well as other factors that are susceptible of policy interventions such as information availability and transport and communication infrastructure.

II.4 Systematic--Indonesian

Ismet, Barkley, and Llewelyn (1998) focus on the effects of government intervention on rice market integration in different regions of Indonesia during the period 1982-93. In the first stage, they measure the degree of spatial integration using the multivariate Johansen approach to test for cointegration of the regional price series and explore the dynamics of the price transmission process. Then, they extract the trace statistic (the test statistic for the null of no cointegration) obtained from that first stage procedure and use it as a measure of market integration. Essentially, the larger the value of that statistic, the stronger the evidence is for market integration. In the second stage, the authors regress the trace on measures of: government intervention (purchases and sales of rice carried out by Bulog in each market); infrastructure (road density); market development (income per capita of the region); and a dummy that controls for the periods of self-sufficiency in rice production. The results for the whole period suggest that only the purchases of rice by Bulog had a significant effect on market integration. The rest of the variables do not significantly explain it. For the self-sufficiency period, sales of rice by Bulog also have a significantly positive effect, as well as per capita income.

The present paper is linked to this scarce literature and contributes to it by tackling the question of determinants of market integration and geographical price disparities in a systematic manner for the rice, soybeans, maize, sugar, and cooking oil markets in Indonesia.

III. Description of the Dataset and Descriptive Statistics

We use consumer price time series for the period January 1993 to December 2007 for rice, sugar, and cooking oil and producer price time series for the period January 1992 to December 2006 for soybeans and maize. Consumer price series correspond to 25 capital city averages with data available for the period under consideration, while producer price series summarize fourteen provincial averages. (7) All price series were obtained from the National Bureau of Statistics of Indonesia (BPS). Data were also obtained from CEIC (CEIC Data Company Ltd) and BPS.

Table 1 presents mean, standard deviation, maximum and minimum, and the ratio of the 80th percentile to the 20th percentile of each of the variables used in this analysis, across provinces. Distance is the minimum distance in kilometres to one of the five main cities in the country (Jakarta, Surabaya, Medan, Makasar, or Batam). This measure needs to be complemented. Take Banda Aceh as an example: it is relatively close to one of the largest cities in the country, which is Medan. Still, Banda Aceh cannot be considered as a "central" city. For a given distance to one of the five main cities, centrality depends on the size of the city you are close to. Thus, in this paper, the notion of centrality is captured by weighting the distance in kilometres by the inverse of the population of the closest city. This weighted variable is called Remoteness. (8) Infrastructure measures the quality of roads as the proportion of asphalted roads in total roads. Population is the number of inhabitants by province, while PCI is real per capita income expressed in Indonesian rupiah at constant prices of 1993. Turning to commodity-specific variables, Output PC is the annual average output of the commodity (in kilograms) divided by the population of the province, while Productivity is the average yield per hectare (in tons) over the period. (9) Trace Stat is the trace statistic, which is a measure of the degree of market integration (this measure will be described in more detail in section III). The larger the trace statistic between province i and j, the "stronger" the market integration is between them. The average for province i over all possible j is reported. Price Diff is the average price difference over the period, of one province averaged against all the others, and Price is the average price of the commodity over the period of analysis. Both are expressed in rupiah per kilogram.

One of the striking patterns in Table 1 is provincial heterogeneity. This is clear when one looks at the difference between the maximum and the minimum. Take infrastructure for example: in one province almost all roads are asphalted while in others, only 15 per cent are asphalted. When looking at the commodity-specific variables, besides the provincial heterogeneity for each particular commodity, there is also important heterogeneity across commodities. For instance, it can be observed that there are important price differences from province to province. (10) While the price differences for soybeans and maize across provinces can be higher than 30 per cent of the average prices, the differences in the rice and sugar markets are of 10 per cent and 6 per cent, respectively. This is consistent with the higher trace statistic values in the latter two markets relative to the former two. As expected, in general, cointegrated markets, i.e. markets whose prices exhibit a common long-run trend, exhibit lower price differences.

To unveil this provincial heterogeneity in a simple way, Table 2 presents some summary statistics for key variables considered.

Table 2 shows that Irian Jaya is the most remote region (both in terms of distance and remoteness) and unsurprisingly, it exhibits the highest price differences with respect to all other provinces. In terms of transport infrastructure, quality is low in Irian Jaya and the Kalimantan provinces (with the exception of South Kalimantan). PCI is largest in East Kalimantan and Jakarta, and lowest in East and West Nusatenggara.

IV. Measuring the Degree of Spatial Integration

In this section, we briefly present the strategy used to measure spatial integration among Indonesian provinces in the markets for rice, soybeans, maize, sugar, and cooking oil using monthly price time series, and the main findings.

Following Fackler and Goodwin (200I), two markets are defined as being integrated when shocks arising in one region are transmitted to the other. More specifically, the market for good x in region i is said to be spatially integrated with that of region j if a shock that shifts, say, demand in i but not in j affects the price in both i and j. This implies that the price series for commodity x in region i shares a long-run stochastic trend with that of region j. If there is perfect integration, the effect of the shock on both prices would be the same.

To obtain a measure of the degree of integration in each market, a testable concept associated with a pair of provincial prices sharing a long-run trend needs to be introduced. For that, the concept of cointegration, first introduced by Granger (1981) and further elaborated further by Engle and Granger (1987), is of help.

Two price series are "cointegrated" if they are both integrated of the same order, say 1 (1), and there exists a linear combination of them, [[beta].sub.1] [p.sub.1t], + [[beta].sub.2][p.sub.2t], which is stationary. (11) The tests for cointegration check if stationary linear combination exists. In this paper, we use Johansen's cointegration test (Johansen, 1988). (12) The test suggests cointegration when the trace statistic (Johansen's cointegration test statistic) is higher than a critical value. The two series are then said to share a common stochastic long-run trend. The higher the trace statistic for a pair of provincial prices, the more strongly cointegrated the series are. Therefore, we can conclude that the higher the degree of integration of the two provinces is.

There are caveats to this test for market integration. Cointegration is neither necessary nor sufficient for market integration. Cointegration is not necessary because, for instance, with non-stationary transaction costs, commodity markets in two regions may be fully integrated but a common stochastic trend between the two will not be found. Cointegration is insufficient and needs to be complemented. For example, assume away non-stationary transaction costs. If a common stochastic trend is found between price series for two regional commodity markets but the nature of cointegration implies a negative relationship between the two, the result is at least discomfiting. (13) The choice of an appropriate indicator to measure the extent of market integration is a moot point (see Barrett 1996 for a discussion on the drawbacks of cointegration and other methods to assess market integration). In this paper, acknowledging the caveats of cointegration-related statistics, we use the trace, conditional on the nature of the cointegrating relationship being positive (that is, that the co-movement is positive in sign).

Johansen cointegration tests were performed on all pairs of provincial prices for the period of analysis and for the commodities under consideration (January 1993-December 2007 for rice, sugar, and cooking oil; January 1992-December 2006 for soybeans and maize). In all cases, the resulting cointegrating vector implied positive relationships between prices in the pair of provinces considered (although these were not always statistically significant). (14)

Table A2 (Appendix) shows the trace statistic obtained for the rice market. Take for example the cell in the first column and second row: the trace statistic obtained when testing cointegration between the rice price series of Central Java and Bali is 30.4. This is higher than the critical value of 15.41 (with 5 per cent significance), and thus strongly suggests a high degree of cointegration, which in turn implies that the two markets are spatially integrated (the higher the trace statistic, the higher the strength of the cointegration relationship). Looking at the first column, thirteenth row, the evidence suggests that for South Sulawesi's and Bali's markets, we cannot reject the null of the markets not being spatially integrated as the value of trace statistic is lower than the critical value. In total, 300 cointegration tests (all possible combination of provincial prices) were performed, of which 229 suggest spatial integration. Thus, for the case of rice, we found evidence of spatial market integration in 76 per cent of the cases.

Tables A3 to A6 (Appendix) show the same estimation for the markets of soybeans, maize, sugar, and cooking oil, respectively. In the soybean market, only 26 per cent of the pair of provinces are spatially integrated; 28 per cent in the case of maize; 83 per cent in the case of sugar; and 29 per cent in the case of cooking oil.

The values of the trace statistic for every pair of provinces in each commodity market will be a key input for the analysis of determinants of integration in section V.

V. Determinants of Price Differences and Market Integration

In this section, the determinants of price differences across provinces and of market integration across provinces are examined.

Price differences between province i and province j, and their trace statistic tend to be negatively correlated. Provincial prices that are highly cointegrated exhibit lower price differences. Yet, the two are not equivalent. The notion of market integration between two provinces is compatible with significant price differentials, as long as these differentials are stable over time. (15) In the presence of logistics costs (transport and distribution costs), a pair of provinces can exhibit a high price differential and still form a market with price signals flowing smoothly.

Still, examining price differences across provinces is enlightening. Understanding whether price differences are driven by distance, poor infrastructure, market power, etc., gives essential information to the policy-maker at the time of deciding where to allocate scarce resources to increase availability of key staples.

This paper makes use of both measures. We proceed first with a preliminary examination of price differences and try to identify regularities associated with them. This will increase the understanding of the effect of potential government policies on reducing them. Then, we proceed to examine the determinants of market integration by looking at the factors that explain the trace statistic.

Before turning into the econometric analysis of determinants of price differences and of market integration, a correlation matrix is constructed to understand how these variables co-move, and identify possible sources of collinearities in the subsequent analysis. Table 3 shows bivariate correlation coefficients.

The price differences in the rice, maize, and sugar markets are significantly correlated with Distance and Remoteness, as expected. For all markets, price differences show a negative correlation with (transport) infrastructure. Better transport infrastructure would reduce transport costs and therefore, allow for price convergence. PCI is positively correlated with price differences in rice, soybean, and maize markets. This may be due to PCI capturing patterns of product quality differences in consumption. The correlation is not significant for sugar and cooking oil. (16)

One interesting feature is that for the rice, soybeans, and maize markets, the degree of market integration (the trace statistic) is significantly and negatively correlated with Distance, and the absolute value of the correlation increases when considering Remoteness instead. Remoteness attempts to capture transportation costs as well as "being part of a hub". (17) Therefore, this variable is capturing two interacting forces. On one hand, it captures the physical cost of moving goods, which should negatively affect integration. On the other, it also captures the "market potential" (18) effect or the effect of being closer to a "hub", which could be associated with higher information flows and a better functioning market, which should positively affect integration. The fact that Remoteness is more strongly correlated to the market integration than distance suggests that it is important to factor in the "market potential" effect and we should not only approximate the transport costs with the plain distance. (19)

V.1 Determinants of Price Differentials

Important price differences in the markets considered were documented in section III, as well as provincial heterogeneity in several dimensions (production conditions, geography, infrastructure, income per capita, etc.). The next step is to examine the extent to which this heterogeneity can explain price differences across provinces. The average price difference between province i and province j, over the period January 1993 to December 2007 is estimated to see the effect of a number of covariates on these differences. (20)

This exercise is an attempt to explain divergences from the law of one price. With trade being costly, one could restate the law of one price as the following condition:

[absolute value of ([p.sub.i] - [p.sub.j])] [less than or equal to] t (1)

The absolute difference between the price in i and the price in j is expected to be lower or equal to the transport and distribution costs, t. In other words, if the price of rice that results from the interaction of domestic supply and demand forces in Jakarta ([p.sub.j]) is well above the price of rice in West Java ([p.sub.WJ]) plus the cost associated with transporting the rice from West Java to Jakarta ([t,.sub.WJ, J]), then West Javanese producers would send their rice to Jakarta and the price in Jakarta would go down to [p.sub.WJ] + [t,.sub.WJ, J].

If instead, the initial difference is lower than the transportation cost (either because transport costs are high or because initial price differences are very small), then prices in different locations will reflect supply and demand conditions in the province, ff price differences are to be examined, these may then lie in the differences in supply conditions, which will be determined by how efficient the process of production is (the level of efficiency will depend on how productive labour, capital, and land are, and by weather conditions), as well as by differences in demand conditions, which will depend on consumers' purchasing power and population size.

Finally, another source of price differences is related to the unobserved quality heterogeneity. Take again the example of rice: different types are consumed in different provinces. One could argue that this could be solved by collecting data on a particular type of rice, say, IR-II, and then comparing prices of IR-II across provinces. However, a long enough time series of prices for IR-II across provinces is not readily available. (21) The problem of quality differentials is difficult to avoid. However, if richer households consume better quality, and therefore more expensive rice, then PCI should control for quality differentials.

The model specified is the following:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2)

where [P.sub.i]-[P.sub.j] is the price difference between province i and j measured in rupiah; [Remote.sub.i] is the distance of province i to the closest main city weighted by the inverse of the population of that main city, and controls for transportation costs. Then, Remote*[Infra.sub.i] is an interaction term between the weighted distance and a measure of transport infrastructure (quality of roads), and [Infra.sub.i] captures the level effect of infrastructure on price differences. It is expected that distance will affect the price more the worse the quality of the transport infrastructure is. The dummy variable contiguity, which takes the value of 1 if the two provinces share a border and 0 otherwise, is trying to capture the fact that road transport is relatively cheaper than other types. Supply conditions are captured with [Productivity.sub.i], which is a measure of yield per hectare, and Output [PC.sub.i], which is the level of output of the commodity normalized by the population in the province. [PCI.sub.i] is per capita income and captures demand-push effects, quality differences across provinces, and government intervention, as the latter is greater in poorer provinces, [e.sub.ij] is an error term capturing all other factors affecting price differences, orthogonal to the included regressors. (22)

As a robustness check, we used some ballpark estimates of transportation costs for rice, obtained from the Indonesian Organization of Transport Companies (ORGANDA). ORGANDA provided rupiah cost estimates of transporting a ton of rice from any given pair of provincial capital cities in 2008. We use these estimates in the model for rice, dropping all the variables that attempt to capture transport costs (remoteness, infrastructure, and contiguity). To check the validity of these ballpark estimates, we regress them on remoteness, contiguity, infrastructure and the interaction of remoteness and infrastructure, and found that these variables explained about 55 per cent of the total variance of transportation costs. Of particular relevance, transporting rice between contiguous provinces is 5 per cent cheaper for a given score of remoteness and infrastructure.

Still, there are three caveats associated with these estimates that make us cautious about using them. First, there are pairs of capital cities between which there is no trade. This implies that the actual transport cost between them is unknown. ORGANDA computed these costs by triangulating through cities that were actually connected through trade flows but these computations may actually overstate or understate the true costs. Second, the costs were estimated for 2008, while the data on price differences considered here are averages for about fifteen years. This introduces an additional problem when matching with the rest of the series. Third, the data provided by ORGANDA corresponded to rice transport costs. For this reason, we restrict the use of this series to the rice models estimated here.

Table 4 presents the results for the five commodities analysed, and reports elasticities which are calculated at the average of the variables, (23) Since data for output and productivity data by province are only available for soybeans, maize, and rice, these variables are omitted in the regression for sugar and cooking oil. For rice, sugar, and cooking oil, data are available for 25 provinces, yielding 300 possible price differences (25 x 24/2). For soybeans and maize, data are available for 14 provinces, yielding 91 possible price differences (14 x 13/2). For ease of interpretation of the results, the model is run on the 300 pairs for which the price difference is positive. This means that the price in province i is always higher than in j. A variable that increases the price in i, ceteris paribus, will increase the price difference, while one that increases the price in j will decrease it. (24)

In the case of rice, we found the expected signs in remoteness, the interaction of remoteness and infrastructure. In particular, we found price difference to be relatively inelastic to remoteness. A 1 per cent increase in remoteness increases the price difference by about 0.236 per cent on average. The effect of remoteness is attenuated by good transport infrastructure, although the level effect of infrastructure is statistically insignificant. Output per capita of the commodity significantly affects the price differences. Provinces that produce more rice relative to their population face a lower price for the product. Productivity differences do not seem to affect the price differential, nor does the contiguity condition. The effect of differences in qualities consumed associated with income per capita seems to be dominating for the case of rice, as the coefficient for income per capita in province 1 is positive and significant.

When we use actual transport costs estimates instead of proxies for transport costs, we found that a 1 per cent increase in transport costs increases price differences by 0.2 per cent on average, while the coefficients on the rest of the variables carry the expected signs: higher income per capita raises price differences, while higher productivity and output of the commodity contribute to lower price differences.

For soybeans, the results are similar, though the size of the coefficients tends to be larger. (25) The only difference is that the "development" effect of income per capita seems to dominate over the quality differentials for this commodity. Infrastructure affects price differences both directly and indirectly through the effect on the remoteness cost. In fact, while a 1 per cent increase in remoteness induces an increase in price differences by a 1.85 per cent on average, the effect falls to 1.42 per cent when considering provinces with infrastructure quality in the top 25 per cent of the distribution.

The results for maize price differences are largely analogous. For this commodity, the coefficients on per capita income and land productivity are poorly determined. This is discomfiting since market integration for maize is much lower than for rice and land productivity is expected to play a more important role in price setting if provinces are less integrated. Contiguity does not play a role in explaining price differences between province-pairs for rice, soybeans, or maize.

For sugar and cooking oil, a reduced set of explanatory variables is incorporated due to data availability constraints. Signs of remoteness and infrastructure interaction with remoteness are as expected in both cases, although for the case of cooking oil they are poorly determined. The effect of different qualities consumed of income per capita seems to be dominating for the case of sugar. Contiguity seems to play a role in reducing price differences for sugar, although not for cooking oil.

These results suggest interesting patterns: remote provinces pay higher prices than central ones with everything else equal. However, it is not a geographic determinism for remote provinces to pay higher prices. Remoteness is less costly the better the transport infrastructure is. Furthermore, domestic production conditions also seem to affect price differences, not only in the less integrated markets of soybean and maize, but also for the case of rice. When ballpark estimates of rice transport costs are introduced instead of remoteness proxies, the results are largely comparable.

V.2 Determinants of Spatial Integration

Attention is now turned to examining determinants of spatial market integration. The dependent variable is the test statistic calculated for the period January 1993 to December 2007 of the cointegration test between a pair of markets (Johansen's trace statistic). A high value of that test statistic provides evidence of strong co-movement of prices and, therefore, of spatial integration; a low value points to the opposite.

The potential determinants of spatial integration are a subset of those that explained price differentials:

remoteness: a higher weighted distance increases transport costs and therefore reduces the degree of spatial integration;

contiguity: a positive sign is expected since this variable attempts to better capture transportation costs. Given Indonesian geography, a measure of remoteness could prove insufficient to capture transportation costs. If for example, transportation by land is cheaper than by sea, one would expect that contiguous provinces are more integrated than those that are not because trade between them is less costly for a given degree of remoteness;

infrastructure (in levels and interacted with remoteness): it is expected that better infrastructure will decrease transportation costs, and thus increase the degree of spatial integration;

PCI: income per capita would control for the fact that richer provinces will consume better quality rice. If that is the case, then when rice price series for different provinces are compared, the comparison may involve prices of different products and a rejection of cointegration would not be indicative of no spatial integration for one specific type of rice. If this effect is predominant, a negative coefficient will be observed for this control variable. On the other hand, PCI may also capture a development effect of the market. Markets with higher income per capita are more developed, exhibit better infrastructure, and so trading tends to be cheaper, ff this effect is predominant, a positive coefficient for this variable will be observed; and

Output PC: output of the relevant commodity normalized by the population of the province. Goodwin and Schroeder (1991) argue that low volume markets have "a bigger potential for exhibiting unwarranted price behavior". On the other hand, it could be argued that provinces that are "self-sufficient" in a certain commodity (they produce enough to cover demand in the province) could be, to some extent, isolated from provincial price movements.

For these reasons, the effect of output on spatial integration is uncertain a priori. Worth mentioning is that this latter "self-sufficiency" effect would mean that the higher the level of output, the lower the degree of integration will be. However, beyond a certain threshold of output, one would expect that the province becomes an exporter of the commodity, which would lead to an increase in its linkages with neighbouring markets. To test if a self-sufficient province is less integrated than one that is not, allowance needs to be made for a non-linear relationship between output and market integration. Thus, the squared value of output PC is added, Sq Output PC. The estimable equation is:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (3)

As in the case of the price differential model, we use transport costs in the rice version of equation (3) as a robustness check, and drop remoteness, infrastructure, contiguity, and the interaction terms.

Table 5 reports the results of estimating equation (3) using OLS, and reports elasticities. The first thing to observe is that the model for the soybean market is not well-determined. None of the covariates are significant, nor is the model as a whole. A quite robust result, for the rest of the commodities considered, is that more remote provinces seem to be less integrated than central provinces. This seems reasonable and is in line with what has been found in the literature (Goodwin and Schroeder 1991; Goletti, Raisuddin, and Farid 1995). The effect of remoteness on market integration is attenuated by the quality of infrastructure, although the coefficients on the interaction term are poorly determined, with the exception of the sugar market case. As it was the case in the price differentials equations reported above, Contiguity seems to affect market integration positively only in the sugar market --given that we are controlling for remoteness, this seems to suggest that road transport is cheaper than other forms of transport. So, for a given degree of remoteness, two contiguous provinces are more likely to be integrated given that land transportation is about 5 per cent cheaper. The quality effect of PCI seems to dominate in most markets, as the coefficient on PCI is negative (statistically insignificant for soybeans and maize). One interesting finding is related to the self-sufficiency hypothesis. The results for the market for rice suggest that market integration is related to output in a non-linear way. More output produced leads to less market integration up to a certain level, after which the relationship changes sign. (26) Self-sufficiency seems to affect market integration only for rice.

VI. Conclusions

In this paper, we have studied the determinants of market integration and price transmission for five major commodities in Indonesia. In a context where commodity prices have been changing dramatically, it is particularly relevant for natural resource abundant countries like Indonesia to understand what drives the transmission of price signals. On the one hand, this will allow the government to take appropriate measures to facilitate price transmission across regions so that producers can take optimal production decisions. On the other hand, it will allow the government to better target its policies geographically to mitigate the impact of a particular price shock on the poor population, and thus reduce the costs of intervention. Using very detailed price data covering twenty-five Indonesian provinces for more than a decade, we shed some light on the drivers of the transmission of price signals and our findings can be summarized as follows.

First, we found that the degree of market integration varies across different commodities. Rice and sugar markets are highly spatially integrated while cooking oil, soybean, and maize markets are much less so. This is consistent with the smaller price differences across provinces in the former markets than in the latter ones.

Second, when focusing on the determinants of these price differences among provinces, we obtained some consistent findings irrespective of the commodity analysed. Remoteness and the interaction between remoteness and quality of infrastructure clearly influence price differentials. Remote provinces pay a higher price but the effect of remoteness is attenuated by good transport infrastructure. Thus, there is no geographical determinism for remote provinces to pay higher prices. Furthermore, price differences are also significantly explained by output per capita and land productivity. Even if provinces are integrated, domestic market conditions matter.

Third, unlike in the case of the previous variables, the effect of income per capita on price differentials varies across commodities. We argue that income per capita is at the same time capturing unobserved quality differences across provinces, as well as "development and local production capacities". In fact, we expect that richer provinces tend to consume higher quality commodities and hence more expensive products, and this is particularly important for those products where there are large quality differences like rice. At the same time, we reckon that for those commodities where quality differentials are not very important, i.e., sugar, soybeans, and maize, the predominant effect is that of "development and local production capacities", and higher local production capacities help to maintain lower prices.

Fourth, when focusing explicitly on market integration, we found that this is clearly explained by remoteness and infrastructure. We also found some limited evidence, though just for the rice market, supporting the argument of "self-sufficiency" being associated with a lower degree of integration.

Our analysis points towards two important policy implications. First of all, it confirms the importance of investing in infrastructure, showing that the constraints generated by geography and remoteness can be alleviated by upgrading the infrastructure quality--which can be achieved through improvements in the investment climate to promote private investment and through investments in public works. Apart from the well-known direct effects that better infrastructure has on growth and development, by contributing to better integrated markets, better infrastructure: increases efficiency; reduces price instability; and decreases the costs of government intervention. Secondly, our findings point towards the importance of strengthening the capacities of farmers and their productivity as an important means not only to improve their livelihoods, but also as an instrument to foster more efficient markets with faster supply responses to changes in prices.

The analysis also has implications for two programmes that the Government of Indonesia is working on to tackle future food crises. The first is the cash transfer programme for the poor, which has been used in the past food crisis but is being fine-tuned to enhance its targeting. The second is the development of an early warning system that will raise the alarm when prices appear to spike. This paper's findings emphasize the importance of taking into account the spatial dimension and infrastructure conditions when forecasting how a price shock will be spread across the country and how it will hurt the population. The analysis also highlights that for some commodities shocks can be expected to be more homogeneously spread across provinces than for others.

DOI: 10.1355/ae30-1b

APPENDIX

TABLE A1

Summary of the Literature on Measurement of Spatial Integration

Authors              Date     Location        Product

Ravallion, M.        1986     Bangladesh      Rice

Goodwin, B.K.,       1991     USA             Cattle

T.C.Schroeder
Alexander, C.,       1994     Indonesia       Rice
J. Wyeth

Goletti, F.,         1995     Bangladesh      Rice

R.Ahmed,
N.Farid

Baulch, B.           1997a    Philippines     Rice

Ismet, M., Barkley   1998     Indonesia       Rice
Llewelyn

Badiane, O.,         1998     Ghana           Maize
G.E.Shively

Baffes, J.,          2001     World           Cotton
M.I. Ajwad                    (Selected
                              Regions)

Goodwin, B.K.,       2001     North           Soybeans
N. E. Piggot                  Carolina, US

Rapsomanikis, G.,    2003     Ethiopia,       Coffee;
D.Hallam,                     Rwanda,         Wheat
P.Conforti                    Uganda;
                              Egypt

Abdulai, A.          2003     Ghana           Maize

Rashid, S.           2004     Uganda          Maize

Van Campenhout,      2007     Tanzania        Maize
B.

Fossati, S.,         2007     Uruguay         Sorghum,
F.Lorenzo,                                    maize
C.M.Rodriguez                                 wheat,
                                              Beef

                     Method of           Dets of
Authors              Analysis            integration?

Ravallion, M.        ECM, IV             No

Goodwin, B.K.,       Cointegration       Yes -
                     Analysis            Reg.

T.C.Schroeder                            Analysis
Alexander, C.,       Error               No
J. Wyeth             Correction
                     Model, Coint.
                     Causality tests

Goletti, F.,         Correlation         Yes -
                     Coeff,              Regression

R.Ahmed,             Cointegration,      Analysis
N.Farid              dynamic
                     multipliers

Baulch, B.           Parity Bound        No
                     Model, mult.
                     cointegration

Ismet, M., Barkley   Johansen,           Yes -Reg.
Llewelyn             Juselius            Analysis

Badiane, O.,         Cointegration,      Yes -
G.E.Shively          ARCH                Simulation
                     models

Baffes, J.,          Error               No
M.I. Ajwad           Correction
                     Model,
                     Cointegration.

Goodwin, B.K.,       Threshold           No
N. E. Piggot         autoregressive
                     cointegration
                     models,
                     impulse
                     response
                     functions

Rapsomanikis, G.,    Multivariate        No
D.Hallam,            cointegration
P.Conforti           (Johansen,
                     Juselius),
                     Causality Test
                     Asymmetric
                     Adj. Tests

Abdulai, A.          Threshold           No
                     autoregressive
                     and
                     cointegration

Rashid, S.           Multivariate        Not
                     cointegration       formally
                     (Johansen,
                     Juselius)

Van Campenhout,      Threshold           No
B.                   autoregressive

Fossati, S.,         Multivariate        No
F.Lorenzo,           cointegration
C.M.Rodriguez        (Johansen,
                     Juselius)

                     Dets of
Authors              Price volat?    Journal

Ravallion, M.        No              American Journal
                                     A. E.

Goodwin, B.K.,       No              American Journal
                                     of Agricultural

T.C.Schroeder                        Economics

Alexander, C.,       No              Journal of
J. Wyeth                             Development
                                     Studies

Goletti, F.,         No              The Developing
                                     Economies

R.Ahmed,
N.Farid

Baulch, B.           No              American Journal
                                     of Agric. Econom.

Ismet, M., Barkley   No              Agricultural
Llewelyn                             Economics

Badiane, O.,         Yes -           Journal of
G.E.Shively          ARCH            Development
                                     Economics

Baffes, J.,          No              Applied Economics
M.I. Ajwad

Goodwin, B.K.,       No              American Journal
N. E. Piggot                         of Agricultural
                                     Economics

Rapsomanikis, G.,    No              Book chapter, in:
D.Hallam,
PConforti

                                     Commodity Mkt
                                     Review
                                     FAO, 2003-2004

Abdulai, A.          No              Book Chapter.

Rashid, S.           No              Journal of African
                                     Economies

Van Campenhout,      No              Food Policy
B.

Fossati, S.,         No              Journal of Applied
F.Lorenzo,                           Economics
C.M.Rodriguez

TABLE A2
Trace Statistics for Rice

Province          1       2       3       4       5       6

Bali
Central Java    30.4
Central Kalim   23.5    15.3
Central Sulaw   18.7    29.9    16.5
East Java       18.0    18.7    15.7    24.0
East Kalim      17.6    17.6    23.5    15.2    13.7
East Nusat      22.7    21.5    12.9    20.5    23.3    16.9
Irian Jaya      22.2    21.8    26.3    20.6    20.2    18.9
Jakarta         24.7    18.9    14.9    20.6    13.3    18.7
North Sulaw     18.0    26.6    11.2    17.0    16.8    15.2
SE Sulaw        18.8    25.6    14.7    24.6    20.9    12.7
South Kalim     20.2    18.2    18.2    15.4    19.9    18.0
South Sulaw     12.7    26.0    13.2    26.5    23.7    12.4
Sumatra Aceh    14.2    20.4    13.6    21.9    18.1    10.4
Sumatra
Bengk           23.6    22.2    15.5    19.6    17.2    19.3
Sumatra
Jambi           15.3    20.8    14.7    20.1    19.9    13.5
Sumatra
Lamp            21.8    29.7    16.4    22.5    31.4    21.8
Sumatra
Medan           16.0    21.5    16.3    25.3    15.9    12.2
Sumatra
Padang          16.3    16.7    13.4    14.6    13.1    19.0
Sumatra
Palemb          24.1    18.7    13.5    16.6    17.2    13.2
Sumatra Riau    18.0    18.9    11.3    18.0    16.8    17.7
West Java       22.1    23.5    15.2    24.6    20.6    14.6
West Kalim      15.8    13.6    12.7    17.3    15.3    9.6
West Nusat      26.8    26.7    17.5    42.6    30.0    23.8
Yogyakarta      28.3    17.6    15.8    23.6    18.8    18.9

Province          7       8       9      10      11      12

Bali
Central Java
Central Kalim
Central Sulaw
East Java
East Kalim
East Nusat
Irian Jaya      15.6
Jakarta         24.9    18.7
North Sulaw     15.3    17.0    19.3
SE Sulaw        19.0    17.1    23.4    15.4
South Kalim     20.4    40.1    19.8    13.8    14.0
South Sulaw     14.8    13.9    17.8    17.4    25.2    14.1
Sumatra Aceh    15.0    10.5    16.3    15.9    23.5    16.5
Sumatra
Bengk           17.5    12.7    25.9    24.3    23.5    15.3
Sumatra
Jambi           20.4    13.1    20.8    18.0    23.3    19.5
Sumatra
Lamp            20.9    22.1    19.2    21.4    20.6    23.3
Sumatra
Medan           17.0    11.2    22.9    18.5    30.4    15.9
Sumatra
Padang          14.0    20.6    15.3    26.4    11.7    11.2
Sumatra
Palemb          15.5    13.5    21.1    24.0    21.7    17.0
Sumatra Riau    21.1    13.7    23.2    21.7    19.1    13.9
West Java       24.4    17.1    17.4    17.0    24.4    15.5
West Kalim      15.4    11.3    17.6    12.2    13.5    18.3
West Nusat      24.5    19.0    20.6    25.5    42.8    21.6
Yogyakarta      22.2    18.5    17.7    21.6    21.2    18.9

Province         13      14      15      16      17      18

Bali
Central Java
Central Kalim
Central Sulaw
East Java
East Kalim
East Nusat
Irian Jaya
Jakarta
North Sulaw
SE Sulaw
South Kalim
South Sulaw
Sumatra Aceh    31.1
Sumatra
Bengk           18.3    20.1
Sumatra
Jambi           19.9    22.9    21.1
Sumatra
Lamp            17.9    19.5    17.4    22.6
Sumatra
Medan           23.0    17.9    28.5    19.0    18.8
Sumatra
Padang          12.5    13.0    19.7    16.4    17.6    15.7
Sumatra
Palemb          17.8    33.0    19.9    23.3    20.3    20.3
Sumatra Riau    12.4    17.7    23.2    18.6    18.2    18.1
West Java       25.9    23.0    21.5    27.1    22.4    21.4
West Kalim      12.9    11.6    17.7    12.7    16.0    11.5
West Nusat      41.8    30.9    25.1    35.6    26.4    31.2
Yogyakarta      17.9    19.4    22.5    22.4    25.4    17.6

Province         19      20      21      22      23      24

Bali
Central Java
Central Kalim
Central Sulaw
East Java
East Kalim
East Nusat
Irian Jaya
Jakarta
North Sulaw
SE Sulaw
South Kalim
South Sulaw
Sumatra Aceh
Sumatra
Bengk
Sumatra
Jambi
Sumatra
Lamp
Sumatra
Medan
Sumatra
Padang
Sumatra
Palemb          15.3
Sumatra Riau    15.4    18.2
West Java       12.8    27.1    17.0
West Kalim      10.5    13.7    13.7    14.0
West Nusat      18.8    27.0    18.0    34.1    19.5
Yogyakarta      15.3    30.3    19.9    26.2    24.1    23.0

NOES: (1) Significant coefficients in bold. (2) Numbers in
column headings correspond to same provinces as in row headings.
(3) Trace statistic is the market integration proxy.

TABLE A3
Trace Statistic for Soybeans

Soybeans                1       2       3       4       5

1     Bali
2     Central Java      6.7
3     Central Sulaw    13.8     4.2
4     East Java         6.4   16.8#     6.0
5     East Kalim        7.9     5.4     7.8     9.8
6     North Sulaw     19.5#     6.3    13.0    12.1    11.1
7     SE Sulaw         10.7     5.6    10.5    13.5   28.5#
8     South Kalim       5.7    11.3     9.9     8.3     9.6
9     South Sulaw       5.1    10.4     4.9   17.2#     8.2
10    Sum Aceh          4.6    10.7     5.3   15.9#    11.0
11    Sum Bengk        11.4     4.0     9.2     6.4     8.8
12    Sum Jambi        11.3     8.2     8.9     9.0     6.8
13    Sum Lamp          7.4    11.8    11.0     8.5    12.0
14    Sum Medan         5.3     9.3     7.2     7.3     6.8
15    Sum Padang        8.6   19.4#     9.8    11.7    14.5
16    Sum Palemb      18.1#   17.7#   16.5#   17.9#    13.3
17    Sum Riau         25.7    10.0     9.5    11.0     6.8
18    West Java         7.6   18.0#     6.0   18.3#     8.4
19    West Nusat        6.1   18.7#     8.3   17.6#   15.9#
20    Yogyakarta        9.0   18.0#     7.7   17.3#    11.2

Soybeans                6       7       8       9      10

1     Bali
2     Central Java
3     Central Sulaw
4     East Java
5     East Kalim
6     North Sulaw
7     SE Sulaw         15.2
8     South Kalim      12.2    13.0
9     South Sulaw       3.8    15.5    10.0
10    Sum Aceh          5.5    17.3    10.8    19.9
11    Sum Bengk        14.6    22.2     9.8     5.1     7.5
12    Sum Jambi        11.6    17.2    13.2    12.3     9.8
13    Sum Lamp         14.1    13.4    15.4    11.8     7.4
14    Sum Medan         5.5     8.9    12.2     8.8     8.2
15    Sum Padang      19.0#   19.1#    15.1    13.6    11.0
16    Sum Palemb      26.8#    15.1    13.8   16.7#   15.4#
17    Sum Riau         11.5     9.2    11.3     7.4     7.8
18    West Java         8.4    10.6    12.7    10.5    10.0
19    West Nusat       11.5   22.0#     7.9   16.2#   17.2#
20    Yogyakarta       18.9    14.5     7.4   16.6#   18.0#

Soybeans               11      12      13      14      15

1     Bali
2     Central Java
3     Central Sulaw
4     East Java
5     East Kalim
6     North Sulaw
7     SE Sulaw
8     South Kalim
9     South Sulaw
10    Sum Aceh
11    Sum Bengk
12    Sum Jambi        12.2
13    Sum Lamp         12.0    15.0
14    Sum Medan         4.9     8.2   17.4#
15    Sum Padang       11.4    10.0   20.7#   15.5#
16    Sum Palemb       15.0   20.7#   19.1#    10.1   17.2#
17    Sum Riau          7.2    12.7    15.3     9.8    10.0
18    West Java         7.0     9.7    15.4     9.1    15.0
19    West Nusat       10.3    15.3     7.5     7.8     9.8
20    Yogyakarta        8.6    12.9     8.7     7.5    11.0

Soybeans               16      17      18      19

1     Bali
2     Central Java
3     Central Sulaw
4     East Java
5     East Kalim
6     North Sulaw
7     SE Sulaw
8     South Kalim
9     South Sulaw
10    Sum Aceh
11    Sum Bengk
12    Sum Jambi
13    Sum Lamp
14    Sum Medan
15    Sum Padang
16    Sum Palemb
17    Sum Riau        20.6#
18    West Java       23.4#     8.5
19    West Nusat      15.9#   18.8#   26.0#
20    Yogyakarta      18.5#   23.3#   23.5#   17.0#

NOTES: (1) Significant coefficients in bold. (2) Numbers in
column headings correspond to same provinces as in row
headings. (3) Trace statistic is the market integration proxy.

Note: Significant coefficients in bold is indicated with #.

TABLE A4
Trace Statistic for Maize

Maize                   1       2       3       4       5

1     Bali
2     Central Java      9.0
3     Central Sulaw    12.7    11.3
4     East Java       15.8#     7.1     7.4
5     East Kalim      18.1#    11.5     4.5    10.1
6     North Sulaw     17.4#    14.8     4.9   17.8#     3.6
7     SE Sulaw        15.8#     8.9   15.5#     8.2     9.8
8     South Kalim       3.0     4.4     5.0     5.1     9.5
9     South Sulaw     19.0#     9.7     4.6   23.1#   18.4#
10    Sum Aceh        15.5#   17.6#     5.9    11.8    10.1
11    Sum Bengk        10.9    13.2     4.0    10.1    15.3
12    Sum Jambi       17.9#    14.5     8.2    11.5    14.8
13    Sum Lamp         13.1    13.4    10.2     8.6    11.6
14    Sum Medan       16.6#   27.0#    12.4     8.9    15.2
15    Sum Padang      49.2#   39.7#   23.4#   33.5#   30.5#
16    Sum Palemb       12.9     6.7    11.7     9.3     9.2
17    Sum Riau         11.1     9.0     8.6     9.2     6.2
18    West Java        14.5   19.2#    12.0     9.9     9.3
19    West Nusat        9.7    12.8     4.5     9.2     4.9
20    Yogyakarta      19.5#     9.7    14.1    11.2    12.2

Maize                   6       7       8       9      10

1     Bali
2     Central Java
3     Central Sulaw
4     East Java
5     East Kalim
6     North Sulaw
7     SE Sulaw          3.7
8     South Kalim       4.7     5.2
9     South Sulaw     19.1#    14.0     4.2
10    Sum Aceh        22.4#     6.8     3.6    11.7
11    Sum Bengk         7.3     8.7   17.5#    13.9     8.5
12    Sum Jambi         8.2    13.2   18.7#   15.6#    15.4
13    Sum Lamp         14.3    10.5     3.6     8.2   23.2#
14    Sum Medan       15.9#    13.3     3.5     9.2   22.2#
15    Sum Padang      40.2#   25.1#    12.8   51.6#   50.2#
16    Sum Palemb        7.2    14.3     2.8     8.9    12.2
17    Sum Riau          3.8    13.0     6.0     4.8     9.1
18    West Java       15.7#     8.7     3.3     9.4   20.0#
19    West Nusat      19.4#     4.6     4.0    10.1    11.3
20    Yogyakarta      20.2#    12.5     6.4   20.2#    13.4

Maize                  11      12      13      14      15

1     Bali
2     Central Java
3     Central Sulaw
4     East Java
5     East Kalim
6     North Sulaw
7     SE Sulaw
8     South Kalim
9     South Sulaw
10    Sum Aceh
11    Sum Bengk
12    Sum Jambi       15.5#
13    Sum Lamp         10.9   18.9#
14    Sum Medan        12.7   17.4#   22.2#
15    Sum Padang      31.0#   37.5#   32.2#   44.9#
16    Sum Palemb        8.8    13.5    10.4    10.8   18.1#
17    Sum Riau          7.3    13.6    11.0    14.1   27.8#
18    West Java         9.1    13.6    14.6   31.6#   39.0#
19    West Nusat        9.4    14.9    12.4    17.3    39.8
20    Yogyakarta       11.4    13.1     9.1    12.0   34.8#

Maize                  16      17      18      19

1     Bali
2     Central Java
3     Central Sulaw
4     East Java
5     East Kalim
6     North Sulaw
7     SE Sulaw
8     South Kalim
9     South Sulaw
10    Sum Aceh
11    Sum Bengk
12    Sum Jambi
13    Sum Lamp
14    Sum Medan
15    Sum Padang
16    Sum Palemb
17    Sum Riau        15.5#
18    West Java         8.9     6.5
19    West Nusat        7.0     5.7    10.8
20    Yogyakarta       9.3      119    10.8    10.3

NOTES: (1) Significant coefficients in bold. (2) Numbers in column
headings correspond to same provinces as in row headings.
(3) Trace statistic is the market integration proxy.

Note: Significant coefficients in bold is indicated with #.

TABLE A5
Trace Statistic for Sugar

Sugar                   1       2       3       4       5

1    Bali
2    Central Java     43.6#
3    Central Kalim    32.2#   23.2#
4    Central Sulaw    58.5#   65.2#   43.4#
5    East Java        17.8#   30.1#   21.3#   34.8#
6    East Kalim       37.9#   38.4#   23.5#   45.4#   16.8#
7    East Nusat       21.2#   14.5    20.6#   25.6#   12.1
8    Irian Jaya       39.0#   54.8#   40.4#   42.8#   40.3#
9    Jakarta          56.8#   33.4#   22.1#   65.1#   21.8#
10   North Sulaw      34.1#   46.2#   38.6#   43.3#   31.4#
11   SE Sulaw         43.9#   50.2#   50.8#   46.4#   22.0#
12   South Kalim      43.9#   47.1#   39.1#   52.5#   31.3#
13   South Sulaw      37.6#   43.7#   32.5#   41.7#   19.5#
14   Sum Aceh         21.5#   15.4#   15.6#   23.2#   15.0
15   Sum Bengk        28.3#   19.0#   21.2#   31.7#   14.0
16   Sum Jambi        22.2#   15.2    21.0#   31.2#   10.0
17   Sum Lamp         39.8#   39.4#   25.9#   48.3#   20.2#
18   Sum Medan        19.1#   11.0    11.9    19.9#   10.7
19   Sum Padang       23.6#   19.6#   18.3#   26.8#   13.9
20   Sum Palemb       24.4#   16.8#   21.4#   36.2#   10.7
21   Sum Riau         20.6#   14.6    17.5#   26.7#   13.6
22   West Java        49.3#   24.4#   16.7#   49.1#   26.3#
23   West Kalim       12.5    9.4     14.3    13.5    7.6
24   West Nusat       32.9#   28.3#   24.6#   48.2#   13.1
25   Yogyakarta       28.6#   32.9#   21.8#   47.2#   27.5#

Sugar                   6       7       8       9      10

1    Bali
2    Central Java
3    Central Kalim
4    Central Sulaw
5    East Java
6    East Kalim
7    East Nusat       20.2
8    Irian Jaya       37.3#   28.3#
9    Jakarta          28.5#   20.5#   46.4#
10   North Sulaw      28.8#   17.6#   33.9#   35.8#
11   SE Sulaw         51.0#   30.3#   20.2#   51.8#   27.3#
12   South Kalim      41.6#   19.8#   44.8#   38.9#   47.0#
13   South Sulaw      36.8#   18.6#   28.4#   38.1#   25.1#
14   Sum Aceh         15.7#   28.8#   22.2#   18.2#   16.2#
15   Sum Bengk        17.2#   27.3#   29.2#   18.9    20.7#
16   Sum Jambi        16.1#   24.1#   25.2#   17.5#   16.2#
17   Sum Lamp         34.9#   16.7#   40.2#   41.4#   32.1#
18   Sum Medan        11.0    20.3#   22.7#   10.9    12.6
19   Sum Padang       19.1#   28.8#   24.9#   20.9#   19.6#
20   Sum Palemb       18.5#   28.9#   25.5#   17.5#   18.9#
21   Sum Riau         17.5#   21.0#   25.8#   18.3#   15.3
22   West Java        22.9#   16.1#   46.8#   27.5#   42.7#
23   West Kalim       9.1     12.7    18.3#   9.7     10.8
24   West Nusat       29.7#   23.3#   38.7#   51.0#   27.8#
25   Yogyakarta       28.8#   16.3#   49.5#   19.9#   42.4#

Sugar                  11      12      13      14      15

1    Bali
2    Central Java
3    Central Kalim
4    Central Sulaw
5    East Java
6    East Kalim
7    East Nusat
8    Irian Jaya
9    Jakarta
10   North Sulaw
11   SE Sulaw
12   South Kalim      54.4#
13   South Sulaw      30.6#   38.7#
14   Sum Aceh         24.2#   16.6#   17.0#
15   Sum Bengk        33.8#   21.0#   22.8#   23.0#
16   Sum Jambi        31.3#   15.1    17.2#   31.9#   41.8#
17   Sum Lamp         38.6#   37.8#   37.8#   17.3#   18.3#
18   Sum Medan        17.7#   13.6    15.6#   19.0#   16.4#
19   Sum Padang       30.1#   19.0#   20.4#   27.2#   25.4#
20   Sum Palemb       34.4#   15.2    17.2#   24.2#   44.9#
21   Sum Riau         25.9#   13.5    17.7#   30.3#   27.2#
22   West Java        49.1#   35.0#   47.1#   14.4    15.3
23   West Kalim       16.3#   11.5    11.6    22.9#   20.3#
24   West Nusat       36.3#   28.9#   31.0#   24.1#   27.1#
25   Yogyakarta       44.0#   51.7#   34.9#   14.4    15.7#

Sugar                  16      17      18      19      20

1    Bali
2    Central Java
3    Central Kalim
4    Central Sulaw
5    East Java
6    East Kalim
7    East Nusat
8    Irian Jaya
9    Jakarta
10   North Sulaw
11   SE Sulaw
12   South Kalim
13   South Sulaw
14   Sum Aceh
15   Sum Bengk
16   Sum Jambi
17   Sum Lamp         14.5
18   Sum Medan        21.4#   13.0
19   Sum Padang       40.7#   17.6#   26.1#
20   Sum Palemb       27.9#   17.9#   21.9#   33.6#
21   Sum Riau         34.6#   16.3#   26.6#   47.5#   22.0#
22   West Java        15.3    34.2#   10.2    16.8#   16.4#
23   West Kalim       22.4#   10.2    17.1#   31.4#   18.7#
24   West Nusat       22.5#   33.1#   18.9#   25.6#   29.5#
25   Yogyakarta       13.3    30.3#   9.4     16.2#   14.2

Sugar                  21      22      23      24

1    Bali
2    Central Java
3    Central Kalim
4    Central Sulaw
5    East Java
6    East Kalim
7    East Nusat
8    Irian Jaya
9    Jakarta
10   North Sulaw
11   SE Sulaw
12   South Kalim
13   South Sulaw
14   Sum Aceh
15   Sum Bengk
16   Sum Jambi
17   Sum Lamp
18   Sum Medan
19   Sum Padang
20   Sum Palemb
21   Sum Riau
22   West Java        13.4#
23   West Kalim       34.6#   8.0#
24   West Nusat       22.5#   31.3#   13.6#
25   Yogyakarta       14.0    20.2    8.4     17.5

NOTES: (1) Significant coefficients in bold. (2) Numbers in column
headings correspond to same provinces as m row headings.
(3) Trace statistic is the market integration proxy.

Note: Significant coefficients in bold is indicated with #.

TABLE A6
Trace Statistic for Cooking Oil

Cooking Oil              1       2       3       4       5

1     Bali
2     Central Java       9.8
3     Central Kalim      6.6     7.8
4     Central Sulaw      7.9    13.8     7.0
5     East Java         15.0     9.8     6.5     8.6
6     East Kalim         8.7     9.5     4.3     6.3    10.6
7     East Nusat         5.1     7.6     4.5    10.6     4.1
8     Irian Jaya       15.7#    10.0    10.9     9.2    11.2
9     Jakarta          17.3#    10.8   20.8#   17.1#    10.1
10    North Sulaw      16.0#     9.6    10.3    11.6     9.4
11    SE Sulaw          10.6   16.3#     8.4    11.6    11.6
12    South Kalim      21.6#    11.8     9.3    10.0   24.3#
13    South Sulaw       12.5   16.0#     8.6    15.3     8.8
14    Sum Aceh         17.3#   19.1#    10.2    13.9   17.6#
15    Sum Bengk        33.6#     9.7    11.3    12.5    15.2
16    Sum Jambi          6.3     6.7     8.0     8.0     4.5
17    Sum Lamp         22.6#    12.4     6.3     9.5   25.5#
18    Sum Medan        30.5#    11.6     8.5    10.7   16.9#
19    Sum Padang        12.1    11.1     5.6     8.5   30.6#
20    Sum Palemb       26.1#     9.3     5.8     8.8   21.9#
21    Sum Riau         19.4#    11.7     8.9    15.3   25.3#
22    West Java        24.0#    11.0    15.3   17.8#    10.5
23    West Kalim         9.0     9.1    11.4    12.5     7.3
24    West Nusat       20.7#    10.3     7.1    12.8    15.1
25    Yogyakarta       24.0#     8.9     7.8    13.2    10.6

Cooking Oil              6       7       8       9      10

1     Bali
2     Central Java
3     Central Kalim
4     Central Sulaw
5     East Java
6     East Kalim
7     East Nusat         5.0
8     Irian Jaya         6.4     6.8
9     Jakarta            8.2   24.0#   23.6#
10    North Sulaw        6.4   15.6#   25.3#    12.9
11    SE Sulaw          15.2     8.6     9.1     7.9     8.4
12    South Kalim       10.5     8.4   19.1#    13.2    12.1
13    South Sulaw        7.5     9.1    15.2    13.9    11.6
14    Sum Aceh          13.4     8.3    15.2     7.4    10.5
15    Sum Bengk          7.4     5.4   50.7#     8.8    12.2
16    Sum Jambi          5.4    11.6     6.5     9.1     5.1
17    Sum Lamp           8.8     6.8    11.2    11.9    12.3
18    Sum Medan          7.1     8.1   28.5#   18.9#   20.2#
19    Sum Padang       17.0#     5.5     9.8     9.6     8.8
20    Sum Palemb         8.6     4.5   16.0#     8.2     8.2
21    Sum Riau          12.6     8.1    15.2    10.4    12.1
22    West Java          7.0   18.8#   49.1#    12.1    14.8
23    West Kalim         5.4    13.9    17.5   39.7#    14.0
24    West Nusat         7.4     7.3   21.4#   19.3#   16.9#
25    Yogyakarta         7.2     9.0   27.7#    15.3    14.4

Cooking Oil             11      12      13      14      15

1     Bali
2     Central Java
3     Central Kalim
4     Central Sulaw
5     East Java
6     East Kalim
7     East Nusat
8     Irian Jaya
9     Jakarta
10    North Sulaw
11    SE Sulaw
12    South Kalim       11.1
13    South Sulaw        8.4    12.7
14    Sum Aceh          13.9   16.8#   16.4#
15    Sum Bengk          9.4   20.1#    12.0    12.7
16    Sum Jambi          6.6     6.1     6.3     5.4     4.3
17    Sum Lamp          11.8   18.0#    13.9   21.7#   44.3#
18    Sum Medan          9.6   21.9#   15.5#   19.9#   74.2#
19    Sum Padang        13.7   21.5#    10.9   17.2#   15.1#
20    Sum Palemb        10.6   19.8#     9.6    13.3   15.1#
21    Sum Riau          11.4   17.5#    11.4   19.9#   16.3#
22    West Java          7.6   16.0#   17.0#     9.0   17.8#
23    West Kalim         6.9     8.2    14.8     8.5     9.2
24    West Nusat         8.7   18.6#    12.4   16.8#   47.5#
25    Yogyakarta         8.8   20.3#    12.2    12.9   37.8#

Cooking Oil             16      17      18      19      20

1     Bali
2     Central Java
3     Central Kalim
4     Central Sulaw
5     East Java
6     East Kalim
7     East Nusat
8     Irian Jaya
9     Jakarta
10    North Sulaw
11    SE Sulaw
12    South Kalim
13    South Sulaw
14    Sum Aceh
15    Sum Bengk
16    Sum Jambi
17    Sum Lamp           4.9
18    Sum Medan          6.2    14.9
19    Sum Padang         4.8    12.7     9.3
20    Sum Palemb         3.7   41.7#   29.2#    11.4
21    Sum Riau           5.9   31.7#   23.1#   22.9#   18.9#
22    West Java          5.4   17.3#   27.2#     7.3     9.3
23    West Kalim         9.8     8.9    12.8     6.8     4.8
24    West Nusat         6.1   24.1#   29.7#     7.8   20.3#
25    Yogyakarta         5.3    14.8   32.4#     8.0    11.9

Cooking Oil             21      22      23      24

1     Bali
2     Central Java
3     Central Kalim
4     Central Sulaw
5     East Java
6     East Kalim
7     East Nusat
8     Irian Jaya
9     Jakarta
10    North Sulaw
11    SE Sulaw
12    South Kalim
13    South Sulaw
14    Sum Aceh
15    Sum Bengk
16    Sum Jambi
17    Sum Lamp
18    Sum Medan
19    Sum Padang
20    Sum Palemb
21    Sum Riau
22    West Java         11.3
23    West Kalim         7.1    25.3
24    West Nusat        15.5    29.0     9.9
25    Yogyakarta        14.8    26.3    10.9    28.0

NOTES: try Significant coefficients in bold. (2) Numbers in column
headings correspond to same provinces as in row headings. (3) Trace
statistic is the market integration proxy.

Note: Significant coefficients in bold is indicated with #.


REFERENCES

Abdulai, A. "Spatial integration and price transmission in agricultural commodity markets in sub-Saharan Africa". In Commodity Market Review 2003-2004, edited by FAO, Ch. 7, pp. 163-83. Rome: FAO, 2004.

Alexander, C., and J. Wyeth. "Cointegration and Market Integration: An Application to the Indonesian Rice Market". Journal of Development Studies 30, no. 2 (1994): 303-28.

Badiane, O. and G. E. Shively. "Spatial integration, transport costs, and the response of local prices to policy changes in Ghana". Journal of Development Economics 56, no. 2 (1998): 411-31.

Baffes, J. and M. I. Ajwad. "Identifying price linkages: A review of the literature and an application to the world market of cotton". Applied Economics 33, no. 1 (2001): 1927-41.

Banerjee, A., J. Dolado, J.W. Galbraith and D.E Hendry. Co-integration, Error-Correction, and the Econometric Analysis of Non-Stationary Data. Oxford: Oxford University Press, 1995.

Baquedano, EG., W. Liefert and S. Shapouri. "World market integration for export and food crops in developing countries: A case study for Mali and Nicaragua". Agricultural Economics 42, no. 5 (2011): 619-30.

Barrett, C.B. "Market Analysis Methods: Are Our Enriched Toolkits Well Suited to Enlivened Markets?". American Journal of Agricultural Economics 78, no. 3 (1996): 825-29.

Baulch, B. "Transfer costs, spatial arbitrage, and testing for food market integration". American Journal of Agricultural Economics 79, no. 2 (1997): 477-87.

Engle, R. and C. Granger. "Co-integration and error correction: Representation, estimation and testing". Econometrica 55, no. 1 (1987): 251-76.

Fackler, R and B.K.Goodwin. "Spatial Price Analysis". In Handbook of Agricultural Economics Vol. 1, edited by B. Gardner and G. Rausser, Ch. 17, pp. 971-1024. Amsterdam: Elsevier Science, 2001.

Fossati, S., E Lorenzo and C. Rodriguez. "Regional and International Market Integration of a Small Open Economy". Journal of Applied Economics 10, no. 1 (2007): 77-98.

Goletti, E, A. Raisuddin and N. Farid. "Structural Determinants of Market Integration: the Case of Rice Markets in Bangladesh". The Developing Economies 33, no. 2 (1995): 196-98.

Goodwin, B.K. and N.E. Piggott. "Spatial Market Integration in the Presence of Threshold Effects". American Journal of Agricultural Economics 83, no. 2 (2001): 302-17.

--and T.C. Schroeder. "Cointegration tests and spatial price linkages in regional cattle markets". American Journal of Agricultural Economics 73, no. 2 (1991): 452-64.

Granger, C. "Some Properties of Time Series Data and Their Use in Econometric Model Specification". Journal of Econometrics 16 (1981): 121-30.

Ismet, M., A.P. Barkley and R.V. Llewelyn. "Government intervention and market integration in Indonesian rice markets". Agricultural Economics 19, no. 3 (1998): 283-95.

Johansen, S. "Statistical analysis of cointegration vectors". Journal of Economic Dynamics and Control 12, no. 3 (1988): 231-54.

Marks, D. "Unity or diversity? On the integration and efficiency of rice markets in Indonesia, c. 1920-2006". Explorations in Economic History 47 (2010): 310-24.

Persson, K.G. Grain Markets in Europe, 1500-1900: Integration and Deregulation. Cambridge: Cambridge University Press, 1999.

Rapsomanikis, G., D. Hallam and P. Conforti. "Market Integration and price transmission in selected food and cash crop markets of developing countries: Review and applications". In Commodity Market Review 2003-2004, edited by FAO, Ch. 8, pp. 187-215. Rome: FAO, 2004.

Rashid, S. "Spatial Integration of Maize Markets in Post-liberalised Uganda". Journal of African Economies 13, no. 1 (2004): 102-33.

Ravallion, M. "Testing Market Integration". American Journal of Agricultural Economics 68, no. 1 (1986): 102109.

Sekhar, C.S.C. "Agricultural market integration in India: An analysis of select commodities". Food Policy 37, no. 2 (2012): 309-22.

Timmer, C.P. Getting Prices Right: The Scope and Limits of Agricultural Price Policy. Ithaca: Cornell University Press, 1986.

--. "Food price policy. The rationale for government intervention". Food Policy 14, no. 1 (1989): 17-27.

Timmer, C. P. "Does Bulog stabilize rice prices in Indonesia? Should it try?" Bulletin oflndonesian Economic Studies 32, no. 2 (1996): 45-74.

Ubilava, D. "El Nino, La Nina, and world coffee price dynamics". Agricultural Economics 43, no. 1 (2012): 17-26.

Van Campenhout, B. "Modelling trends in food market integration: Method and an application to Tanzanian maize markets". Food Policy 32, no. 1 (2007): 112-27.

World Bank. "Boom, bust, and up again? Evolution, drivers and impact of commodity prices: Implications for Indonesia". The World Bank Group, 2010.

NOTES

The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.

(1.) The outcome of market integration will generate greater aggregate gains for society than the absence of integration; however, different groups may either gain or lose. This potentially raises a political economy dimension to question about determinants of market integration. That dimension is out of the scope of this paper.

(2.) As argued by Timmer (1996), in Asia, the link between stable food prices and growth is particularly relevant. In Indonesia, rice accounted for one-quarter of GDP in the mid-1960s and one-third of employment. Today, with the rapid industrialization process that the economy experienced, rice production still explains about one out of ten dollars produced in that economy. From the consumption side, as documented by Marks (2010), 14 per cent of an average family's budget was spent on rice in 2001, and it scaled up to 30 per cent if the poor were considered. For a good description of the links between food price stability and development, see Timmer (1989) and Timmer (1996). Marks (2010) provide an interesting, non-exhaustive summary.

(3.) Perssson (1999, p. 9) refer to this as the "spatial cancelling out of harvest disturbances". Integration does not actually "cancel out" the disturbances but spreads the effects, mitigating the impact on the disturbed region and reducing geographical price disparities. Note also that this spatial mitigation not only applies to harvest disturbances but to any affecting supply or demand conditions.

(4.) These two compatible conceptual definitions are in line with those of Ravallion (1986) and Faclder and Goodwin (2001) definitions respectively.

(5.) An excellent review on methodological issues related to the analysis of spatial market integration can be found in Fackler and Goodwin (2001).

(6.) Systematic here is understood as "explored using regression techniques". Of course, that is not the only "systematic" way of exploring a relationship between two variables, in a broad way.

(7.) Consumer price series were considered for Aceh, North, West and South Sumatra, Riau, Jambi, Bengkulu, Lampung, Jakarta, West, Central and East Java, Yogyakarta, West, Central, South, and East Kalimantan, North, Central, South, South East Sulawesi, Bali, West and East Nusatenggara and Irian Jaya. Producer series were considered for Aceh, North and South Sumatra, Riau, Lampung, West, Central and East Java, Yogyakarta, North, Central, South and South East Sulawesi and Bali. Provinces that were created in 1999 were not considered in the analysis, seven of which have been created since 1999 as data were not available for the whole period considered. These are North Maluku, West Papua, Banten, Bangka-Belitung, Gorontalo, Riau Islands and West Sulawesi. Provinces that existed prior to 1992 but were not included here had price series with constant prices for long periods of time (over 1.5 years in some cases), which suggested misreporting.

(8.) As the measure of remoteness was very small, in the table remoteness?1000 is reported, for convenience of presentation.

(9.) These two variables are only available, at a provincial level, for soybeans, rice and maize only.

(10.) These differences should be analysed in conjunction with the average price.

(11.) A series is said to be 'stationary' if its mean and variance do not vary with time.

(12.) A very good presentation of the Johansen cointegration procedure can be found in Banerjee et al. (1995).

(13.) A common stochastic trend between price series in two markets could also be found in the absence of integration if the two markets were systematically subject to the same shocks. Weather shocks are a typical example. In fact, Ubilava (2012) find that El Nino related events have had short-term impacts on coffee prices. This would lead to a co-movement in prices of markets that were affected by these environmental shocks, irrespective of whether these markets are integrated. Long run analysis of co-movement as performed in this paper may help circumvent this problem.

(14.) The price series considered here were found to be integrated of order one (I(1)). Augmented Dickey Fuller tests were performed on the series. The lag structure was chosen following the Akaike Information Criterium. First differences of the series proved to be integrated of order zero or I(0). The results of the unit root tests are not reported here for brevity's sake, and are available from the authors upon request. A more comprehensive description of the cointegration relationships found among these provinces can be found in World Bank (2010).

(15.) Assuming that logistics costs are stable over time. If they are always increasing, then a constantly increasing price differential may still be consistent with market integration.

(16.) It could be argued that the scope for quality differentials in sugar is lower than in the case of rice. That would explain the insignificant correlation of PCI and Price Diff. The same argument wouldn't hold for cooking oil, though.

(17.) In fact, this is a measure of distance that adjusts for the size of the main market that province i is close to.

(18.) This "market potential" effect is related to the population size of the city.

(19.) Interestingly, for sugar and cooking oil, the correlations are positive and do not change significantly when looking at distance or remoteness.

(20.) The reason a panel is not used to analyse the determinants of price differences along time and across provinces is because data for most of the explanatory variables are available only for selected years and, in general, there is only limited overlapping among them.

(21.) In addition to this, based on discussions with Bulog experts, even a very specific type of rice such as IR-II varies by province.

(22.) In principle, differences in income per capita could also affect price differentials due to supply side reasons for a given quality level, for example, if richer provinces used better technologies that allowed production at lower costs. In our model, however, this effect should be captured by the Productivity variable.

(23.) We use Ordinary Least Squares to estimate both the model of determinants of price differences and of market integration. The assumption that the regressors are uncorrelated with the disturbance term is virtually unquestionable for remoteness, contiguity, infrastructure and income per capita. For the cases of land productivity and output per capita, some concerns of reverse causality may arise. For rice, soybeans and maize, we run the models excluding these variables and the results were barely unchanged, suggesting that the main results are not affected by this potential endogeneity source. The models we report for cooking oil and sugar do not include these variables as data are unavailable for these products.

(24.) It should be taken into account that in the estimation of the model on price differences, the degrees of freedom of the regression are given by the number of provinces we have data on (25) and not by the number of pair-wise combination of provinces. This is taken into account when analysing the significance of the coefficients.

(25.) This is probably related to the reduced sample size for soybeans due to lower data availability. The same consideration applies to maize.

(26.) This turning point was estimated at about 1.36 tons per capita of paddy rice. The conversion from paddy to white rice is generally done at 1.5 kilograms of paddy rice for 1 of white rice, which would imply, assuming no waste, that the turning point is when the province produces more than approximately 955 kilograms of rice per capita.

Gonzalo Varela is an Economist at the International Trade Department, World Bank, USA, and the Center for the Analysis of Regional Integration at Sussex, University of Sussex, U.K.

Enrique Aldaz-Carroll is a Senior Country Economist, World Bank, Phnom Penh, Cambodia.

Leonardo Iacovone is a Senior Economist at Financial and Private Sector Development, World Bank, USA.

TABLE 1
Descriptive Statistics by Commodity

           Variable           Mean     Std Dev      Min

           Distance           570.87     587.84     0.00
           Remoteness          0.071      0.092    0.000
           Infrastructure       0.53       0.24     0.15
           Population          6,538      9,462    1,520
           PCI                 1,998      1,762      682

Rice       Output PC          229.06     173.83     1.74
           P'tivity               40          8       25
           Trace Stat          19.51       5.47     9.60
           Price Diff            259        186        6
           Price               2,520        221    2,174

Soybeans   Output PC            2.95       4.08     0.55
           P'tivity            11.96       1.53     8.48
           Trace Stat          12.14       5.04     3.81
           Price Diff            850        725        4
           Price               2,664        770    1,872

Maize      Output PC           36.55      42.35     0.01
           P'tivity            26.14       6.45    16.00
           Trace Stat          13.85       9.13     2.83
           Price Diff            359        284        3
           Price                 973        316      478

Sugar      Trace Stat          26.69      11.99     7.59
           Price Diff            191        173        0
           Price               3,369        179    3,161

C. Oil     Trace Stat          13.74       8.54     3.71
           Price Diff            565        422        2
           Price               4,192        489    2,958

           Variable            Max      80/20 Pct

           Distance           2381.13         2.82
           Remoteness           0.341       13.780
           Infrastructure        0.98         2.46
           Population          35,000         3.72
           PCI                  7,915         2.47

Rice       Output PC          1442.26         3.37
           P'tivity                55         1.49
           Trace Stat           42.82         1.54
           Price Diff             870         4.54
           Price                3,044         1.17

Soybeans   Output PC            22.88         3.23
           P'tivity             14.96         1.25
           Trace Stat           28.51         2.20
           Price Diff           2,555        11.76
           Price                4,427         1.65

Maize      Output PC           170.78        11.98
           P'tivity             45.00         1.42
           Trace Stat           51.65         2.36
           Price Diff           1,298         5.93
           Price                1,776         1.73

Sugar      Trace Stat           65.19         2.33
           Price Diff             720         5.98
           Price                3,880         1.07

C. Oil     Trace Stat           74.19         2.26
           Price Diff           1,929         4.54
           Price                4,887         1.16

NOTES: Population and PCI are expressed in thousands.

SOURCE: Authors' own elaboration, based on data from BPS and Bulog.

TABLE 2
Descriptive Statistics by Province

Province               Distance    Remote    Population

Aceh                        424     0.037         3,990
North Sumatra                 0     0.000        11,600
West Sumatra                460     0.123         4,396
Riau                        291     0.078         3,734
Jambi                       304     0.082         2,498
South Sumatra               424     0.047         6,512
Bengkulu                    566     0.063         1,520
Lampung                     195     0.022         6,836
Jakarta                       0     0.000         9,000
West Java                   121     0.000        34,900
Central Java                258     0.007        31,400
Yogyakarta                  264     0.008         3,040
East Java                     0     0.000        35,000
West Kalimantan             607     0.163         3,817
Central Kalimantan          624     0.018         1,837
South Kalimantan            485     0.014         3,032
East Kalimantan             583     0.084         2,543
North Sulawesi              953     0.136         1,982
Central Sulawesi            484     0.069         2,072
South Sulawesi                0     0.000         6,985
SE Sulawesi                 367     0.053         1,755
Bali                        317     0.009         3,085
West Nusatenggara           402     0.012         3,843
East Nusatenggara           726     0.104         3,828
Irian Jaya                2,381     0.341         1,633

Province                PCI     Infrast

Aceh                   2,714       0.45
North Sumatra          1,878       0.49
West Sumatra           1,617       0.71
Riau                   4,880       0.35
Jambi                  1,210       0.58
South Sumatra          1,714       0.53
Bengkulu               1,069       0.72
Lampung                  933       0.49
Jakarta                6,298       0.98
West Java              1,526       0.70
Central Java           1,216       0.64
Yogyakarta             1,542       0.76
East Java              1,566       0.58
West Kalimantan        1,667       0.31
Central Kalimantan     2,066       0.15
South Kalimantan       1,854       0.56
East Kalimantan        7,915       0.21
North Sulawesi         1,235       0.72
Central Sulawesi       1,046       0.54
South Sulawesi         1,150       0.51
SE Sulawesi              901       0.45
Bali                   2,223       0.97
West Nusatenggara        858       0.76
East Nusatenggara        682       0.40
Irian Jaya             3,132       0.15

NOTES: Population is expressed in thousands. PCI in rupiah at
constant prices of 1993, expressed in thousands. Distance is in
kilometres, to one of the 5 main cities. Infrastructure is the
percentage of asphalted roads in the province. Remoteness weights
distance to the main city by the inverse of the population of the
main city.

SOURCE: BPS and CEIC Data Company Ltd.

TABLE 3
Correlation Matrix

                          Distance    Remote     Pop      PCI

           Distance
           Remote             0.90
           Pop               -0.38     -0.35
           PCI                0.06      0.10    -0.11
           Infra             -0.46     -0.47     0.21    -0.17

Rice       Output PC         -0.23     -0.22     0.05    -0.16
           P'tivity          -0.47     -0.51     0.58    -0.04
           Trace             -0.14     -0.25     0.13    -0.14
           Diff Price         0.41      0.49    -0.19     0.28

Soybeans   Output PC         -0.06     -0.31     0.02     0.00
           P'tivity          -0.12     -0.15     0.14    -0.05
           Trace             -0.04     -0.09     0.10     0.10
           Diff Price         0.08      0.23    -0.39     0.15

Maize      Output PC         -0.44     -0.41     0.41    -0.44
           P'tivity          -0.60     -0.72     0.82    -0.24
           Trace             -0.16     -0.35     0.18    -0.05
           Diff Price         0.18      0.43    -0.36     0.28

Sugar      Trace              0.19      0.21    -0.21    -0.03
           Diff Price         0.63      0.62    -0.24    -0.06

C. Oil     Trace              0.06      0.05    -0.07     0.03
           Diff Price        -0.01      0.01    -0.01     0.00

                          Infra    Output PC    P'tivity    Trace

           Distance
           Remote
           Pop
           PCI
           Infra

Rice       Output PC      -0.02
           P'tivity        0.83         0.13
           Trace           0.28        -0.07        0.31
           Diff Price     -0.23        -0.40       -0.27    -0.16

Soybeans   Output PC       0.25
           P'tivity        0.12         0.21
           Trace           0.04         0.03       -0.13
           Diff Price     -0.49        -0.45        0.05     0.18

Maize      Output PC       0.15
           P'tivity        0.39         0.32
           Trace           0.26         0.06        0.40
           Diff Price     -0.35        -0.31       -0.64    -0.34

Sugar      Trace          -0.04
           Diff Price     -0.42                              0.01

C. Oil     Trace           0.02
           Diff Price     -0.03                             -0.21

NOTE: Trace is the proxy used for market integration.

TABLE 4
Determinants of Cross-Province Price Differentials

                   Rice (1)      Rice (2)     Soybeans (1)

Transport Cost    0.200 ***
                  (0.06)
Remoteness 1      0.236 ***                   3.392 ***
Remoteness 2      -0.05                       (0.49)
                  -0.558 **                   -0.653 ***
                  (0.18)                      (0.17)
Contiguity        -0.007                      0
                  (0.01)                      (0.02)
Remote*Infra 1    0.01                        -2.815 ***
                  (0.06)                      (0.33)
Remote*Infra 2    0.429 **                    0.551 ***
                  (0.15)                      (0.15)
Infra 1           0.104                       -0.719 ***
                  (0.06)                      (0.19)
Infra 2           0.025                       0.028
                  (0.07)                      (0.13)
PCI1              0.243 ***     0.218 ***     -0.565 ***
                  (0.05)        (0.05)        (0.14)
PCI2              -0.136 *      -0.077        0.11
                  (0.07)        (0.07)        (0.14)
Land Prod 1       -0.092        -0.582 ***    1.499 *
                  (0.17)        (0.16)        (0.58)
Land Prod 2       -0.379        0.22          0.888
                  (0.28)        (0.18)        (0.47)
Output Rel.       -0.280 ***    -0.305 ***    -0.270 ***
Comm.PC 1         (0.03)        (0.04)        (0.05)
Output Rel.       0.340 ***     0.128         0.039
COmm.PC 2         (0.08)        (0.07)        (0.06)

R2                0.42          0.297         0.769
N                 300           300           91
F                 15.459        15.632        57.242

                                             Cooking
                  Maize (1)    Sugar (1)     Oil (1)

Transport Cost

Remoteness 1      1.762 ***    0.617 ***    0.059
Remoteness 2      (0.33)       (0.05)       (0.05)
                  -0.219       0.034        -0.015
                  (0.27)       (0.09)       (0.09)
Contiguity        -0.003       -0.025 **    0.007
                  (0.02)       (0.01)       (0.01)
Remote*Infra 1    -1.546 ***   -0.227 ***   -0.081
                  (0.25)       (0.05)       (0.06)
Remote*Infra 2    0.227        -0.006       0.215 **
                  (0.23)       (0.09)       (0.08)
Infra 1           -0.578 **    -0.062       0.103
                  (0.19)       (0.07)       (0.09)
Infra 2           0.146        0.158 *      0.058
                  (0.15)       (0.08)       (0.07)
PCI1              -0.152       -0.188 ***   -0.028
                  (0.14)       (0.04)       (0.07)
PCI2              -0.33        -0.065       0.137*
                  (0.17)       (0.07)       (0.06)
Land Prod 1       -0.936
                  (0.53)
Land Prod 2       0.165
                  (0.5)
Output Rel.       0.137
Comm.PC 1         (0.11)
Output Rel.       0.207 *
COmm.PC 2         (0.1)

R2                0.659        0.45         0.081
N                 91           300          300
F                 15.796       33.814       4.104

NOTES: * p<0.05, ** p<0.01, *** p<0.001, s.e. in parentheses.

TABLE 5
Determinants of Market Integration in Indonesia

                 Rice (1)      Rice (2)      Rice (3)

Transport Cost                               -0.077 **
                                             (0.03)
Remoteness 1     -0.082 ***    -0.081 **
                 (0.02)        (0.03)
Remoteness 2     -0.007        -0.027
                 (0.02)        (0.02)
Contiguity       0             0.001
                 (0.00)        (0.00)
Remote*Infra 1   -0.056        -0.053
                 (0.03)        (0.03)
Remote*Infra 2   -0.083 **     -0.066 *
                 (0.03)        (0.03)
Infra 1          -0.01         -0.009
                 (0.03)        (0.03)
Infra 2          -0.039        -0.031
                 (0.02)        (0.02)
PCI 1            -0.076 ***    -0.072 *      -0.034
                 (0.02)        (0.03)        (0.03)
PCI 2            -0.065 **     -0.146 ***    -0.142 **
                 (0.02)        (0.04)        (0.04)
Output Rel.      0.015         0.024         0.052
Comm.PC 1        (0.02)        (0.1)         (0.09)
Output Rel.      -0.015        -0.266 **     -0.261 *
Comm.PC 2        (0.02)        (0.1)         (0.1)
Sq Output Rel.                 -0.003        -0.021
Comm.PC 1                      (0.03)        (0.03)
Sq Output Rel.                 0.114 *       0.116*
Comm.PC 2                      (0.04)        (0.05)

R2               0.187         0.214         0.109
N                300           300           300
F                7.106         6.062         4.444

                 Soyb (1)    Soyb (2)    Maize (1)

Transport Cost

Remoteness 1     0.006       -0.049      -0.295 *
                 (0.18)      (0.19)      (0.14)
Remoteness 2     -0.158      -0.138      -0.592 **
                 (0.36)      (0.41)      (0.17)
Contiguity       0.013       0.012       -0.003
                 (0.01)      (0.02)      (0.02)
Remote*Infra 1   -0.137      -0.094      -0.015
                 (0.09)      (0.1)       (0.09)
Remote*Infra 2   -0.297      -0.262      0
                 (0.21)      (0.32)      (0.12)
Infra 1          -0.045      0.004       0.205
                 (0.17)      (0.18)      (0.13)
Infra 2          0.085       0.079       -0.480 ***
                 (0.23)      (0.26)      (0.13)
PCI 1            -0.095      -0.07       0.151
                 (0.11)      (0.11)      (0.11)
PCI 2            0.174       0.152       0.143
                 (0.17)      (0.24)      (0.09)
Output Rel.      0.094       -0.12       -0.01
Comm.PC 1        (0.07)      (0.26)      (0.06)
Output Rel.      -0.028      -0.068      -0.001
Comm.PC 2        (0.05)      (0.23)      (0.06)
Sq Output Rel.               0.134
Comm.PC 1                    (0.16)
Sq Output Rel.               0.022
Comm.PC 2                    (0.12)

R2               0.139       0.146       0.194
N                91          91          91
F                1.545       1.384       2.956

                 Maize (2)    Sugar         Cooking Oil

Transport Cost

Remoteness 1     -0.419 *     -0.252 ***    -0.255 ***
                 (0.18)       (0.04)        (0.07)
Remoteness 2     -0.574 **    0.109 ***     0.133 **
                 (0.18)       (0.03)        (0.05)
Contiguity       -0.006       0.024 ***     0.009
                 (0.02)       (0.01)        (0.01)
Remote*Infra 1   0.035        -0.220 ***    0.062
                 (0.1)        (0.05)        (0.08)
Remote*Infra 2   -0.003       -0.201 ***    -0.093
                 (0.11)       (0.04)        (0.07)
Infra 1          0.32         0.102 **      0.161 *
                 (0.17)       (0.04)        (0.06)
Infra 2          0.462 **     -0.056        -0.175 ***
                 (0.15)       (0.03)        (0.05)
PCI 1            0.115        0.031         0.064
                 (0.12)       (0.04)        (0.04)
PCI 2            0.14         -0.101 ***    -0.117 ***
                 (0.09)       (0.03)        (0.03)
Output Rel.      -0.351
Comm.PC 1        (0.29)
Output Rel.      0.038
Comm.PC 2        (0.2)
Sq Output Rel.   0.161
Comm.PC 1        (0.13)
Sq Output Rel.   -0.023
Comm.PC 2        (0.1)

R2               0.207        0.236         0.075
N                91           300           300
F                2.677        11.568        3.802

NOTE: * p<0.05, ** p<0.01, *** p<0.001, s.e. in parentheses.
COPYRIGHT 2013 Institute of Southeast Asian Studies (ISEAS)
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2013 Gale, Cengage Learning. All rights reserved.

 
Article Details
Printer friendly Cite/link Email Feedback
Author:Varela, Gonzalo; Aldaz-Carroll, Enrique; Iacovone, Leonardo
Publication:Journal of Southeast Asian Economies
Geographic Code:9INDO
Date:Apr 1, 2013
Words:15988
Previous Article:Economic governance: what can the EU and Asean learn from each other after the economic crises?
Next Article:Trends in output and income disparities across the states of Malaysia, 1972-2009.
Topics:

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters