The Evolution of Rice Price Policies over Four Decades: Thailand, Indonesia, and the Philippines. Kei Kajisa* Takamasa Akiyama

Transcription

1 The Evolution of Rice Price Policies over Four Decades: Thailand, Indonesia, and the Philippines by Kei Kajisa* Takamasa Akiyama Foundation for Advanced Studies on International Development, Tokyo Japan July, 2003 *Corresponding author. Address: 2-2 Wakamatsu-cho, Shinjuku-ku Tokyo, Japan; Phone: +81-(0) ; Fax: +81-(0)

2 Abstract Using time series data over the past four decades, 1960s-1990s, this paper examines rice pricing policies in Thailand, Indonesia, and the Philippines. It analyzes the determinants of the paths along which these policies have moved. While this analysis confirms the findings of previous analyses, that price stabilization has been a major policy achievements, it also reveals that stabilization was not necessarily sustained over the entire survey period. It finds that politico-economic factors such as entry into the GATT, increase in per capita GDP and achievement of rice self sufficiency have been among the determinants of rice pricing policy, but the ways in which these factors have affected policy varies among these countries. Such variation, which previous cross-country studies have not analyzed, is a reflection of variations in the roles of rice and in the attitudes of policy makers in these economies. In its conclusion, this study draws policy implications for each country taking into account differences in determinant impact. Keywords: rice; price policy; Thailand; Indonesia; the Philippines JEL classification: F13, O13, Q18 2

3 The Evolution of Rice Price Policies over Four Decades: Thailand, Indonesia, and the Philippines 1. Introduction Rice has a special significance in Asia and to Asians. Because of its economic and political sensitivity, it has been rare for any Asian government to allow the international market to determine domestic rice prices. Instead, government interventions in the domestic market taxes, subsidies, and market operations have been prevalent. Over the four decades examined in this study, 1960s-1990s, many Asian countries experienced dramatic changes in their internal economic condition as well as in the external economic environment. Dissemination of Green Revolution technology since the early 1970s significantly increased rice production. Agricultural technology and developments in non-agricultural sectors contributed to significant and long-term increases in income level, such that some countries succeeded in moving out of poverty into middle income status. Meanwhile the momentum of globalization was exerting pressure in the direction of free trade. Each of these historical factors influenced Thailand, Indonesia and the Philippines, the three countries chosen for this study. Their impact may have been intensified by entry into the GATT/WTO regime. This study asks the question: Given these circumstances, what were the factors that affected rice price policies in each of these three Asian economies and what were the effects? More specifically, in each country what has been the relationship over time between changes in 3

4 macro- and rice-sector factors and rice price policies? And secondarily, what was the impact on each country s rice policy of significant events, substantive or symbolic, such as the achievement of rice self sufficiency or entry into GATT? There are indications that protection of the rice sector in Asian developing countries has been increasing despite clear indications from Asian developed countries, Japan, Taiwan, and South Korea, that protection leads to inefficient rice farming with huge financial burdens for consumers and governments. Thus, answering these questions has practical relevance because without understanding the dynamics of price policy formation, it will be difficult to forumulate and evaluate appropriate rice policies. The existing literature only partially answers these questions. The pattern that has been most commonly observed is a policy shift toward protection as incomes increase (van Bastelaer, 1998; Beghin and Kherallah, 1994; Zietz and Valdes, 1993; Lindert, 1991; Timmer, 1988; Anderson and Hayami, 1986). Recent literature also reveals a practice of using taxes and subsidies to stabilize domestic prices (Dawe, 2001; Timmer, 1993; Schiff and Valdes, 1992). These have been statistically confirmed in Asian rice markets by David and Huang (1993). However, previous analyses have generally relied on cross-country data using both developed and developing countries as samples. Such analyses detect the factors that have influenced the sample countries, but on the average rather than in the particular. For this reason, while the past literature provides useful general ideas on the determinants of protection, it is inefficient for the purpose of evaluating and designing a specific country s rice pricing policies. In order to formulate country specific policies, the generalized ideas provided by cross-country analyses should be examined by country specific time series analyses, as well. This paper explores the forces driving the evolution of rice price policies over the last four decades by conducting specific time series analyses for three major rice producing countries: Thailand, Indonesia, and the Philippines. The period covered is for Thailand and the Philippines and for Indonesia. Akiyama and Kajisa (forthcoming) 4

5 analyze these countries price policies for major agricultural commodities, including rice, over the period Building on that work, this paper focuses specifically on rice with extended newly available time series data. This paper is organized as follows: Section 2 provides details concerning methods of measurement and an overview of rice price policies in the three countries over the past four decades. In order to understand the determinants of change, Section 3 presents analytical and econometric frameworks. Section 4 shows the results of time series analyses. Based on these results, policy implications and conclusions are drawn in Section Methods of Measurement and an Overview of Rice Price Policies in the Three Countries The impact of price policy is measured by the nominal protection rate (NPR) defined as the percentage difference between the real domestic price (RDP) and the real border price (RBP). RDP RBP NPR = RBP RDP = RBP 1 The domestic market price of rice deflated by the consumer price index (CPI) is used as RDP and the world price of rice at border converted at the official exchange rate and deflated by CPI is used as RBP. 1 NPR is negative when rice is taxed and is positive when it is protected. The ratio (RDP/RBP) is called the nominal protection coefficient (NPC), utilized for the econometric analysis in Section 4. Although there are more complete protection measures, such as the producer support estimates of OECD (OECD, 2002) or the effective protection rates, 1 For details of the sources and the computational method of RDP and RBP for each country, see the Appendix. 5

6 NPR is used in this analysis because the ease of its estimation allows time series analysis to be extended long enough for an examination of evolutionary path. Because price stabilization is claimed to be one of the major objectives of price policy, two price stability indices are calculated. One is the ratio of the coefficient of variation (CV) of real border price to that of real domestic price, which is greater than one when the real domestic price is more stable around its mean than is the real border price. Because the CV can be affected if the underlying data have some trend, a second index has been calculated, the ratio of absolute average values of annual percentage changes in RBP to those of RDP over the decades. 2 This ratio is greater than one when the domestic price is more stable than the border price; that is, when the year-to-year RDP percentage changes are smaller on average than those of RBP. Since the first index examines deviations from the average and the second examines annual changes, the former can be regarded as a long-term stability index and the latter as a short-term stability index. These indices for Thailand are presented in Figure 1 and Table 1. Figure 1 shows the trend of RDP and RBP from 1961 to 2000 and Table 1 shows the average NPR and the two stability indices for Thai rice. Thailand exported rice and taxed it throughout the period but the data reveal a changing pattern. Figure 1 and the NPR averages in Table 1 indicate that tax rates were very high in the 1960s and 1970s (on average 30% in the 1960s and 24% in the 1970s even after exclusion of the commodity crisis years) and then were reduced appreciably in the 1980s (10%) and in the 1990s (4%). The stability indices in Table 1 reflect a policy of domestic price stabilization from the early 1970s to the early 1980s; after that the price stabilization indices become closer to one, or even smaller, suggesting that since the early 1980s the government abandoned the stabilization policy. These data, the reduction in taxation and 2 Because these analyses use real prices and because the period considered is relatively short (i.e. 10 years each), the CV is little affected by trend. 6

7 the cessation of rice price stabilization, point to a dramatic shift in Thai government rice policy from the early-1980s toward free trade. In fact, the government abolished the rice reservation deposit in 1982 and several types of rice taxes in It is important to note that Thailand signed the GATT agreement in November The next section will examine the economic and political rationale underlying tax reductions in Thailand prior to the 1980 s and whether a structural break occurred in the policy-making mechanism in the early 1980s. Key statistics on Indonesian rice prices for the period are shown in Figure 2 and Table 2. They indicate that rice was moderately protected at around 15% throughout the survey period excluding the very early years and the commodity crisis and the Asian currency crisis years when RBP surged. The high values of the stability indices show that the domestic price was more stable than the border price, especially in the 1970s and 1980s. The stability is attributed to marketing operations designed and conducted by the parastatal BULOG (Bandan Urusan Logistic, National Logistics Board), established to facilitate Indonesia s top priority rice policy, self sufficiency, announced in BULOG s strategy was to set the rice price high enough to stimulate production without transferring international price fluctuation into the domestic market. The effect of macro or sector variables on BULOG s pricing policies will be examined statistically in Section 4. In the case of Indonesia, entry into GATT occurred before the survey period (1950) and, hence, will not be analyzed. Figure 3 characterizes Philippines rice price policy as starting with protection, switching to taxation in the early 1970s, and then returning to protection in the early 1980s. Broadly speaking, the periods of protection and taxation correspond to the periods of rice importation and exportation, respectively. These results can be traced to market intervention by a parastatal. 3 Unlike in the case of Indonesia, however, the Philippines parastatal was not 3 This parastatal was called the Rice and Corn Administration (RCA) until 1972 when it became the National Grain Authority (NGA). In 1981 it became the National Food Authority (NFA). 7

8 authorized to design policies but only to implement government policies. The Philippines signed the GATT in December 1979 but rice was exempt until 2004 (as of 2002) and the government seems reluctant to open the rice market. Table 3 shows that since the mid-1980s when the country returned to the position of rice importer, the protection rate has increased dramatically (from 16% in the 1980s to 41% in the 1990s) while domestic price stabilization has declined (0.9 in the ratio of CV in the 1990s). Identification of the determinants of these changes in the three countries is attempted in the following section. 3. Analytical Framework 3.1 Structure of the Model We start with a simple price-transmission model in which the relationship between the real domestic price (RDP) and the real border price (RBP) of rice in year t is described as RDP = αrbp (1) t β t where α reflects tax or protection policies determined independently of the level of RBP, as well as marketing and transportation costs if these components are not included in RBP, and β captures the elasticity of price transmission. When β is one, the full percentage change of RBP is transmitted to the domestic price. Although unusual, the value of β can be negative if movements of the two prices are opposing. When β is zero, RDP t is equal to α regardless of the level of RBP t, implying a constant percentage tax/subsidy. The magnitude of β becomes close to zero either when price stabilization policies are effective or when domestic prices fluctuate as much as border prices but in a different direction. Therefore, in order to see the 8

9 effect of price stabilization policies over time, the magnitude of β must be interpreted by the stability indexes presented in Tables 1 to 3. To facilitate analysis of NPC, equation (1) is modified as RDPt ln RBPt = ln NPC t = α ' + ( β 1) RBP (2) t where α is lnα. Both coefficients, α and β, may change over time due to the effects of some variables. Assuming a linear relationship between the coefficients and these variables, each coefficient is replaced with α and β with subscript t and be specified as α = α ln X (3) ' α + t 0 i t + 0 i i i it β = β β ln Z (4) it where X i and Z i are determinants of taxation/protection policies. Combining equations (2), (3), and (4) and adding a statistical error term, the following equation can be estimated: ln t 0 + αi ln Xit + ( β0 1) ln RBPt + βi i i [ ln Zit ln RBPt ] et NPC = α + (5) Estimating this equation allows us to evaluate how αs and βs have changed over time with the effects of determinants X it and Z it on NPC. Furthermore, by computing β t based on equation (4), we can trace the change in the elasticity of transmission over time. Previous cross-country studies suggest candidates for X i and Z i. First, income level is used in order to capture politico-economic factors which encourage taxation on the rice sector in low income countries; these factors include the government s need for revenue from the dominant agricultural sub-sector and the desirability of providing people staple foods at low price (Lindert, 1991; Timmer, 1988; Anderson and Hayami, 1986). These same studies also argue that as incomes grow, factors promoting protection become more powerful; these factors include as nostalgia for rural life which often results in leniency toward an unprofitable 9

10 agricultural sector. Therefore, a positive relationship is expected between income level and protection rate. In the present analysis, per capita GDP (PGDP) is used to measure income level. 4 Second, the self sufficiency ratio of rice (SSR), measured as the ratio of production to consumption, is used to capture the extent to which rice importing countries protect the sub sector in order to achieve food security (David and Huang, 1993; Schiff and Valdes, 1992; Lindert, 1991). Third, the relative wage ratio of the agriculture sector to the industrial sector is used because several previous studies (van Bastelaer, 1998; Beghin and Kherallah, 1994; Zietz and Valdes, 1993; Anderson and Hayami, 1986) argue that governments tend to subsidize the agricultural sector when agricultural wages tend to lag behind industrial wages. This is measured by the ratio of agricultural GDP per total agricultural population to non-agricultural GDP per total non-agricultural population (GDPRTO). Fourth, the competitiveness of the rice sector is measured by rice yield per hectare (YLD) to test the argument of David and Huang (1993) that an increase in productivity may nullify the need for protection and may even make the sector tolerant to taxation. Fifth, in order to examine the influence of powerful interest groups, the relative size of the agricultural sector is used, measured by the share of agriculture in the labor force (AGPOPSHR) or in GDP (AGGDPSHR) as proposed by van Bastelaer (1998); Beghin and Kherallah (1994); Lindert (1991); and Miller (1991). In our time series analysis, we examine the effects of these six variables on the path of nominal protection rate, NPC. 3.2 Econometric Issues As time-series data are used in the estimation of equation (5), what is required is essentially a co-integration analysis of a set of variables. This implies that the choice of specific analytical framework is very important. There are several choices including the 4 The GDP per capita used in our analysis is calculated such that GDP at 1995 constant US$ price is divided by total population. 10

11 two-step Engle-Granger method (TSEG), Johansson s maximum likelihood approach, the fully-modified Hansen-Phillips least squares approach (FM-OLS), and the autoregressive distributed lag (ARDL) approach. The results presented and discussed here are based on the ARDL (dynamic OLS) approach. In addition to its relative computational simplicity, this approach requires fewer assumptions, has attractive large sample properties (Pesaran and Shin, 1997), and permits valid inferences on the long-run parameters. It can be applied regardless of whether the variables are trend- or difference-stationary, making it unnecessary to apply unit-root tests that have low power in small samples. We considered a simple illustration of the approach based on an ARDL(1,1) model for two I(1) variables y and x, i.e y t = φ 1 yt 1 + γ 0xt + γ1xt 1 + εt 1 < 1 φ (6) which has the dynamic representation γ + γ L 0 1 t y t = xt + = θ( L) xt + 1 φ1l 1 φ1l ε u t (7) The long-run relationship implied by this model is γ0 + γ1 y = θ (1) x = x (8) 1 φ where the long-run coefficient is θ (1) 1. The standard errors of long-term coefficients can be obtained by a variety of methods including the delta-method. The model given above can be generalized to allow for additional lags in both y and x and for additional stationary variables such as dummy variables. In practice, information on the order of lags in the model is not available. Model selection criteria can be used to select the appropriate order. In this paper we use the Schwarz criterion, which tends to favor a relatively parsimonious parameterization. We also apply specification tests for serial correlation to confirm that the model is free from misspecification. The occurrence of a structural break in pricing policy is unknown a priori. A popular method 11

12 to detect a break is the cumulative sum of residual (CUSUM) and the cumulative sum of residual square (CUSUM of square) tests. 5 In order not to overlook any structural breaks, we also rely on the method of plotting residual variance over time (Hansen, 2001) Regression Results The regression results are reported in Tables 4 to 6. Our analysis found some variables such as AGPOPSHR, AGGDPSHR, and YLD to be insignificant and they are not shown in the tables. The results indicate that some particular variables may be significant in one country during a certain time period but not necessarily in another country. This suggests that the variables found to be significant in cross-country analyses are not necessarily significant in country- and commodity-specific time series analyses. The following discussion is based on time series analyses specific to each country and uses the results of the long-term response unless otherwise indicated. 4.1 Thailand The path of Thailand s NPC is determined by per capita GDP (PGDP) and real border price (RBP). Before interpreting these variables, however, we test for a structural break because it will change the ways PGDP and RBP affect NPC. Results reported in column (T-1) in Table 4 were obtained under the assumption of no structural break. Using the same 5 The CUSUM test can be applicable to our dynamic model which contains a lagged dependent variable among the regressors (Kramer et al., 1988). 6 A sample is split into two at a possible break year and then OLS regressions are run separately on each segment. Repeating this process, we compute the residual variances for all the years (sum of squared errors divided by the sample size) and plot them against years. The years having global/local minimum are candidates for break years. 12

13 explanatory variables in the model (T-1), CUSUM, CUSUM of squares and the residual variance are plotted (Figures A1-(a) to A1-(c) in the Appendix). These suggest structural breaks in the years 1970 and in Inclusion of a period dummy, which becomes one in and after 1982, and its interaction terms with other factors improves the results substantially. Hence, we can conclude that Thailand s rice price policy experienced a structural break in 1982, which is the year Thailand signed the GATT agreement and also abolished the rice reservation deposit. How is the impact of PGDP different before and after the break? According to the results in the model (T-2) with the period dummy (DUMMY82), the partial derivative of ln(npc) with respect to ln(pgdp) evaluated at the mean ln(rbp) indicates that the impact of PGDP was much greatr on NPC before 1982 than after. For example, a 5% annual increase in per capita income reduces the tax from 20% to 17% before the break; whereas after the break, the impact from the growth of PGDP becomes almost negligible (from 20% to 19.4% with the same income growth rate). For comparison, we run a regression using the relative agricultural wage, GDPRTO; however, it does not explain NPC as well as PGDP does. These results suggest that before the break point, the government did not pay attention to the relative poverty of the agricultural sector but instead reacted to the increase in absolute income level. Presumably, one of the reasons for this is that a tax on rice became less necessary as alternative taxable sources increased with the growth of the economy. Figure 4 shows that the proportion of rice revenue in central government revenue declined dramatically during the 1960s and became very small in the 1970s even though the rice tax rate remained the same (about 30 % on average). The literature suggests another reason, the need to provide low wage goods had declined with the growth of the economy (Lindert, 1991; Anderson and Hayami, 7 Although both CUSUM and CUSUM of squares are within the band of 5% significance, the residual variance suggests possible breaks in 1970 and in 1982 where the variances take lower values. 13

14 1986). Figure 5 shows that during the late 1970s, minimum non-agricultural wages tripled. This suggests that over this time the provision of rice at low prices became politically less necessary due to increases in income among the non-agricultural poor. Given these two reasons and also the finding that GDPRTO, relative agricultural wage or relative poverty in agriculture, did not explain NPC s path well, we can conclude that taxes on rice were not reduced because the government wanted to reduce the burden on rice farmers, or even to protect them; rather because the need for the taxes diminished as a consequence of economic development. Therefore, by the time of the break point the tax had been reduced enough that PGDP had lost its impact as a relevant determinant. How does the other important determinant, RBP, relate to NPC? The evolution of the transmission elasticity in Figure 6, which is computed as the coefficient β t in equation (4), indicates that until the break point was reached, the elasticity had been declining; that is, the domestic market had became more and more insulated over time. This insulation was achieved through adjustment of the tax or premium rates such that abrupt changes of the world prices were neutralized (Dixon, 1999). Domestic prices were stabilized as reflected in the high stability indices in the 1970s in Table 1. However, after the break, the transmission elasticity jumped to unity and since 1982 domestic price has been moving almost in tandem with border price. This implies that using entry into GATT as an opportunity for an abrupt change in rice pricing policy, the government withdrew completely from the domestic market and let the world market determine the price of rice. Nevertheless, this does not mean that both consumers and producers were exposed to the fluctuation of the world rice market. For producers, the government provided a program, the rice pledging program, under which farmers mortgage their rice when market price is low until such time as it returns to normal. Note, however, that under the pledging program farmers have to manage the risk themselves. The government s 14

15 stance toward the provision of price stabilization for rice farmers switched from one of giving direct support to all farmers through intervention in commodity markets to one of providing a price stabilization device which the farmers, themselves, must opt to use. 4.2 Indonesia The CUSUM, CUSUM squares, and the residual variance plots in Figures A2-(a) to A2-(c) in the Appendix are computed using the explanatory variables in the Model (I-1) in Table 5, suggesting possible structural breaks in 1985 and/or in The results shown as Model (I-2), which include DUMMY85 and DUMMY92 interacted with RBP and PGDP, fit the actual NPC path best. This suggests that Indonesia s pricing policies experienced structural breaks firstly as a shift of constant term in 1985 and secondly as changes of magnitudes of the coefficients in The effect of the first structural break is captured by DUMMY85. The negative value for the coefficient for this dummy indicates that the protection rate was reduced from 15% to 3% after 1985 and was kept at that level. It is worth noting here that this break year occurs one year after Indonesia s public announcement of achievement in rice self sufficiency. This suggests that once the government was convinced that rice self sufficiency had been achieved, it reduced the protection that had been used for that purpose. Interestingly, at the same time the results show that the annual change of self sufficiency ratio (SSR) has no explanatory power for the evolution of NPC in the long term. This holds even if DUMMY85 is excluded as shown in Model (I-1). These findings indicate that BULOG did not change its policy of protecting against short term fluctuations of SSR (Figure 6) until it was convinced that rice self sufficiency had been achieved. The second structural break is related to the impact of PGDP which for Indonesia, as for Thailand, was one of the most important factors affecting protection rate. Unlike Thailand, 15

16 however, PGDP became more influential after the structural break in After the break, the coefficient of PGDP increases from 0.58 to 1.138, meaning, for example, that while the nominal protection rate increases from 15% to 18% with a 5% increase of GDP per capita before the break, it increases from 15% to 23% with the same increase of GDP per capita after the break. 8 This is consistent with anecdotal evidence that BULOG s interventions in this period were oriented toward import substitution, partly due to the merger of the Ministry of Trade with the more conservative Ministry of Industry and also partly due to replacement of the management board of BULOG (Amelina, forthcoming). Hence, one explanation for the increase of protection with PGDP at a increased rate after 1992 is an increase in demand for protectionism by policymakers. Regression implications that the drop of SSR in the late 1990s (Figure 6) does not significantly explain the rise of protection lends credibility to this explanation. The transmission elasticity for Indonesia in Figure 6 shows that the effect of the structural break on the transmission elasticity is not discernable. Although the positive and significant coefficient of the interaction term of RBP with GDPRTO indicates the evolution of beta, the aggregated impact is so small that Indonesia s transmission elasticity is low for the entire survey period (around 0.35), a reflection of the immunity of domestic prices to world price fluctuations. The stabilization indices in Table 3 re-enforce this, showing that at least until the late 1980s BULOG stabilized the domestic price against world market prices. In the 1990s, stabilization was achieved to some extent, but not as strongly as in the previous decades. These findings suggest that while it might have been accomplished during the period until the late-1980s, after that time the achievement of price stabilization for the entire county by a single 8 In 1998, the year of the Asian currency crises, NPC sharply dropped with PGDP. There might be some concern that the magnitude of the positive coefficient of PGDP after the break would be exaggerated by that event. However, additional regression analyses show that the exclusion of that year does not change the results appreciably, especially regarding the coefficient of PGDP. 16

17 parastatal became increasingly difficult. In summary, Indonesia s price policies effectively achieved the stabilization of domestic price in the 1980s but became less effective after the early 1990s. While the achievement of SSR had a one time reduction impact on protection in 1984, the protection rate significantly increased with PGDP growth after 1992, presumably due to a rise in demand for protectionism by policymakers. This implies that Indonesia has the potential to take the same path traveled by Japan, South Korea, and Taiwan; a path that has led to high rice prices in insulated domestic markets and a correspondingly heavy burden on consumers. 4.3 The Philippines According to the residual variance plot in Table A3-(c), there are possible structural breaks in the Filipino rice pricing policies in 1979 or/and 1991, also possibly around However, CUSUM and CUSUM squares in Tables A3-(a) and A3-(b) do not suggest any structural breaks. Among the three possibilities, a break is most likely to have occurred either in 1979 when the Philippines entered into the GATT regime or in 1986 when the Marcos regime ended. The inclusion of DUMMY79 or DUMMY86 and their interaction terms do not produce any significant coefficients but the inclusion of DUMMY91 and its interaction terms as shown in the Model (P-2) does. We judge that the structure of Filipino rice pricing policies did not change in 1979 or in 1986, but did change in The structural break in 1991 is clearly reflected in the change of the transmission elasticity in Figure 6. Until 1991 elasticity stayed very close to zero, indicating that Filipino rice price was immune to fluctuations in world price. This insulation of the market was accompanied by price stabilization, as shown in the stability indices in Table 3. However, after the break, transmission elasticity was around -1.4, an unusual value which indicates that domestic prices were changing in opposite direction from border prices. The fluctuation 17

18 depicted in Figure 3 and the CV ratio in Table 3 (less than 1 in the 1990s) reflect the change of transmission elasticity. Many researchers believe that this change was due in large degree to a lack of competence and funding of the rice marketing parastatal, the National Food Authority (NFA) (Balisacan, 1989; Dawe, 1993; Amelina, forthcoming). In the event of a shortage in domestic supply, NFA, which was an international rice trade monopoly, was required to import rice and release it to defend the price set by the government. However, inaccurate projections of the amount of rice to import during times of shortage often caused the domestic price to surge. Budget stringency also limited the NFA s administrative capacity for market intervention. Price fluctuation due to shortages in domestic supply under controlled importation accelerated in the 1990s when the El Niño and La Niña weather phenomena damaged the country s agriculture. Presumably this made the inefficient management of NFA more obvious, a situation which was manifested as a structural break. Rice self sufficiency ratio (SSR) is another important determinant. In the Philippines, unlike Indonesia, indications are that the government adjusted price policies in accordance with annual changes in SSR. As we have seen in Section 2, when rice was exported as a result of the Green Revolution in the late 1970s, it was taxed. However, rice production did not keep pace with population increase so that the country reverted to the position of importer in the mid-1980s. The SSR dropped further in the 1990s (Figure 8). Facing this trend, NFA purchased paddy at subsidized prices with the aim of stimulating domestic production, resulting in positive protection rates (FAO, 2001). NFA s mismanaged import control further increased the domestic price in years of low self sufficiency, strengthening the negative relationship between SSR and NPC. In summary, for the entire period considered, the level of rice self sufficiency and price stabilization have been two influential driving forces of the Philippines rice price polices. As 18

19 we have seen in Section 2, the entry of the Philippines into GATT did not cause a structural break, unlike the case of Thailand. Like Indonesia, parastatal market interventions aimed at price stabilization were effective until the 1980s. The disappearance of the price stabilization effect is more obvious in the case of the Philippines than for the other countries; in fact, the market intervention exacerbated domestic price fluctuation in the 1990s. Unlike Indonesia the recent rise of protection is explained mainly by the decline in rice self sufficiency rather than by income level. 5. Policy Implications and Conclusions For each of these three countries, the analyses confirm what has been claimed by the existing literature, one of the main achievements of pricing policy is price stabilization. The time series analyses in this paper, however, also reveal that stabilization was not necessarily sustained over the entire four decades. The Thai government has been letting the market determine rice price since the early 1980s. In Indonesia and, especially, in the Philippines price stabilization policy was less effective in the 1990s than in earlier periods, reflecting the increasing difficulty of achieving price stability for an entire country through a single marketing parastatal such as BULOG in Indonesia or NFA in the Philippines. Taking account of the differences in the dynamic mechanisms of the rice price policies, some policy implications can be drawn for each country. First, looking at the Philippines: Given the negative and significant coefficient of the self sufficiency ratio (SSR), achievement of food self sufficiency appears to have been one of the major objectives of price policies in the Philippines. However, the social cost of pursuing that goal under the NFA has been very high, as exemplified by recent high domestic prices 19

20 (burden to consumers) and exaggerated price fluctuations (burden to consumers and producers). Opening the rice market to the world may be a more reliable and satisfactory way to achieve the underlying objective, food security. Allowing private importers into the rice market would benefit consumers and challenge producers. Support for farmers, if needed, could be given through access to new technology and by investment in land, capital, and labor. Opening the market would also help smooth price fluctuations, at least to the international level; though in this regard, the dependability of the international rice market must be demonstrated to the Philippine government. Although, the world rice market used to be regarded as thin, recent research by Dawe (2002) shows that rice prices have been low and quite stable since Thailand and Vietnam returned to the market and predicts that they will remain so in the future. Research of this sort can be valuable input to policymakers decisions. Second, the case of Indonesia: The income growth since 1992 indicates that despite having attained self sufficiency, the government continues to subsidize the domestic price relative to the world price. Protection in rice-producing developed countries is a common pattern; however, it generally results in inefficiency and huge burdens on consumers and should serve as an example of what not to do. Information about the negative consequences of price protection is limited and it might be that Indonesia s policymakers are ill-informed. Researchers should disseminate information about the potentially significant adverse effects of large-scale protectionism. Third, better price stabilization mechanisms should be adopted in Indonesia and, especially, in the Philippines. In both countries, price stabilization was effectively achieved in the past but was not sustained. While intervention efforts have not weakened, stabilization has been less successful, indicating that it is increasingly difficulty to achieve price stabilization for an entire country through the procurement and trade controls of a single parastatal. The nature of the parastatal should perhaps be changed from that of agent stabilizing prices to that of agent 20

21 providing instruments through which individuals, themselves, stabilize prices. Thailand s rice pledging program is one example of such an instrument. However, this kind of program would need to be introduced carefully in the Philippines and Indonesia. Poverty is more prevalent in these countries than in Thailand and there is a risk that poor farmers may experience long-term ill effects from a one-time negative price shock. Introduction of news program should be accompanied by safety nets. Fourth, Thai rice protection policy is interesting in that it switched from price intervention to other means in the early 1980s. The results of the present analysis indicate that increases in income induced a reduction in taxes until the structural break in 1982 but did not induce an increase in protection through pricing policies. This is not the same thing as saying that no protection was given to Thai farmers. Since 1981 the major support program for rice farmers in Thailand has been the rice pledging program (the impact of which is difficult to capture by NPC). However, it should be noted that the quantity of pledged rice at subsidized rates has been increasing. According to a review by WTO, this support for rice farmers is designed to address the income gap between agriculture and industry (WTO, 2000; GATT, 1991). This suggests that as relative poverty in the agricultural sector becomes a social problem, protection through non-price policies has increased in Thailand. Based on Japan s experience, Hayami (1988) argues that relative poverty becomes a social problem when the economy enters the mid-income level, a stage which Thailand appears to have entered in the mid-1980s. The future path of protection in Thailand could be more clearly revealed by examining the relationship between non-price policies and poverty in the agricultural sector relative to the non-agricultural sector. These country-specific time series analyses allow us to understand the impact of politico-economic factors such as entry into GATT, increase in per capita GDP or the achievement of rice self sufficiency in ways that earlier cross-country studies have not 21

22 considered: Firstly, that these factors may be relevant determinants in one country during one time period but not necessarily in another country during that same period. Secondly, that the pattern of impact on price policy of these politico-economic factors differs among the countries; in some cases the impact is continuous and uninterrupted and in other cases there is a structural break. These differences reflect the differences in the role and importance of rice in the economy and in the attitudes of policymakers. Earlier cross-country studies have erroneously assumed that the same factors explain each country s evolutionary path in the same way. The present study has found that policy making mechanisms cannot be readily generalized and that country specific time series analyses are needed for each developing country and commodity. The ramifications of policy alternatives are country-specific. This is an important reality that the researcher must accommodate. 22

23 Appendix: Description and Sources of Data Used All the data except prices are from either the World Development Indicators (2002) by the World Bank (WDI hereafter) or FAOSTAT by FAO. Variable Official exchange rate (LCU per US$, period average) Consumer price index (1995 = 100) GDP (constant 1995 US$) GDP share of agriculture % Population, total (000) Agricultural population Rice export Rice import Rice production (MT) paddy Rice yield per hectare Source WDI WDI WDI WDI FAOSTAT FAOSTAT FAOSTAT FAOSTAT FAOSTAT FAOSTAT The rice prices are obtained in the following manner for each of the three countries. Thailand The domestic and international prices from 1960 to 1985 are from Siamwalla and Setboonsarng (1989) (S&S hereafter). From 1986 to 2000, both domestic and border prices are extrapolated using the results from the regression of the prices from S&S on the prices from Rice Statistics by IRRI (2002) (Rice Stat hereafter). The results are: (Domestic Price from S&S) = *(Wholesale Price from Rice Stat), adjusted R 2 =0.999 (Border price from S&S) = *(Border price from RiceStat), adjusted R 2 =0.999 The coefficients are significant at the 1% level. Since Thailand is one of the largest rice exporters, border price is influenced by Thailand s 23

24 taxation. Hence, the adjusted border price, which is the border price under the assumption of no taxation in Thailand, is computed using the formula in S&S on p. 47. Philippines The domestic prices from 1960 to 1979 are the wholesale equivalent domestic prices from Intal and Power (1989) (I&P hereafter). From 1980 to 2000, we use the wholesale price from Rice Stat without adjustment because the prices from Rice Stat and I&P are quite close. The border prices are computed as the sum of Thai fob rice price and transportation costs which are assumed to be US$ 10 from 1961 to 1973, US$ 20 from 1974 to 1978, US$ 25 from 1979 to 85, and US$ 15 from 1976 to 2000 based on oil prices at the world market. Indonesia There are no previous studies along the lines of S&S and I&P for Indonesia. For domestic prices, retail prices issued from BULOG rather than wholesale prices from Rice Stat are used. This is because, according to a specialist from the International Rice Research Institute, recent wholesale prices have still been determined using prices of varieties that are no longer popular in the domestic market. The border price is computed as was done for the Philippines. It should be noted that if the real domestic price had been based on the wholesale price series constructed by the same varieties used in the BULOG series, the nominarl protection rate in Indonesia would have been lower than the figures reported in Table 2. 24

25 References Akiyama, T. & Kajisa, K. (forthcoming) Effects of direct protection for agriculture in three Southeast Asian countries, in: T. Akiyama & D. Larson (Eds) Dynamism of Rural Sector Growth: Policy Lessons from East Asian Countries, (Canberra, Asia Pasific Press). Amelina, M. (forthcoming) Do different regimes distort differently?, in: T. Akiyama & D. Larson (Eds) Dynamism of Rural Sector Growth: Policy Lessons from East Asian Countries, (Canberra, Asia Pasific Press). Anderson, K. & Hayami, Y. (1986) The Political Economy of Agricultural Protection (Sydney, Allen and Unwin). Balisacan, A. M. (1989) Philippine agricultural development in historical perspective, in: M. F. Montes & H. Sakai (Eds) Philippine Macroeconomic Perspective: Developments and Policies (Tokyo, Institute of Developing Economies). Beghin, J. & Kherallah M. (1994) Political institutions and international patterns of agricultural protection, The Review of Economics and Statistics, 76, pp David, C. C. & Huang J. (1993) Political economy of rice protection in Asia, Economic Development and Cultural Change, 44, pp Dawe, D. (1993) Price Stabilization and the Macroeconomy in Low Income Countries, Ph. D. dissertation (Massachusetts, Harvard University). (2001) How far down the path to free trade? the importance of rice price stabilization in developing Asia, Food Policy, 26, pp (2002) The changing structure of the world rice market, , Food Policy, 27, pp Dixon, C. (1999) The Thai Economy: Uneven Development and Internationalization (London, Routledge). 25

26 FAO (Food and Agriculture Organization) (2002) FAOSTAT, (2001) Review of Basic Food Policies (Rome, FAO Commodities and Trade Division). GATT (General Agreements on Tariffs and Trade) (1991) Trade Policy Review: Thailand 1991 (Geneva, GATT). Hansen, B. E. (2001) The new econometrics of structural change: dating breaks in U.S. labor productivity, Journal of Economic Perspectives, 15, pp Hayami, Y. (1988) Japanese Agriculture under Siege: The Political Economy of Agricultural Policies (London, Macmillan). Intal, P. S. J. & Power, J.H. (1989) Trade, Exchange Rate, and Agricultural Pricing Policies in the Philippines (Washington D. C., The World Bank). IRRI (International Rice Research Institute) (2002) Rice Statistics (Los Banos, IRRI). Kramer, W., Ploberger, W. & Alt, R. (1988) Testing for structural change in dynamic models, Econometrica, 56, pp Lindert, P. H. (1991) Historical patterns of agricultural policy, in: C. P. Timmer (Ed) Agriculture and the State (Ithaca, Cornell University Press). Miller, T. C. (1991) Agricultural price policies and political interest group competition, Journal of Policy Modeling, 13, pp OECD (2002) Agricultural Policies in OECD Countries: Monitoring and Evaluation 2002 (Paris, OECD). Pesaran, M. & Shin Y. (1997) An autoregressive distributed lag modeling approach to cointegration analysis, in: S. Strom (Ed) Econometrics and Economic Theory in the 20th Century (Cambridge, Cambridge University Press). Prudhisan, J. (1987) Interest and pressure groups, in: S. Xuto (Ed) Government and Policies for Thailand (London, Oxford University Press). Schiff, M. & Valdes, A. (1992) The Political Economy of Agricultural Pricing Policy: A 26

27 Synthesis of the Economics in Developing Countries (Baltimore, Johns Hopkins for the World Bank). Siamwalla, A. (1975) A history of rice policies in Thailand, Food Research Institute Studies, 14. Siamwalla, A. & Setboonsarng S. (1989) Trade, Exchange Rate, and Agricultural Pricing Policies in Thailand (Washington D. C., The World Bank). Timmer, C. P. (1993) Rural bias in the East and South-East Asian rice economy: Indonesia in comparative perspective, Journal of Development Studies, 29, pp (1988) The agricultural transformation, in: H. Chenery & T. N. Srinivasan (Eds) Handbook of Development Economics (Amsterdam, Elsevier). van Bastelaer, T. (1998) The political economy of food pricing: an extended empirical test of the interest group approach, Public Choice, 96, pp World Bank (2002) World Development Indicators (Washington D. C., The World Bank). WTO (World Trade Organization) (2000) Trade Policy Review: Thailand 1999 (Geneva, WTO). Zietz, J. & Valdes, A. (1993) The growth of agricultural protection, in: T. Ito, & A. O. Kruger (Eds) Trade and Protectionism (Chicago, University of Chicago Press). 27

28 Figure 1 and Table 1: Real Rice Prices, NPR, and Stabilization Indexes in Thailand from 1961 to 2000 RDP RBP Baht/ton Year Period Average NPR (-24.4)* Ratio of CV Ratio of Average Absolute Percentage Change * NPR excluding commodity crisis years from 1973 to

29 Figure 2 and Table 2: Real Rice Prices, NPR, and Stabilization Indexes in Indonesia from 1967 to 2000 RDP RBP 2,000 1,800 1,600 1, Rupiah/ton 1,200 1, Year Period Averages Average NPR (14.76)* (18.61)** Ratio of CV Ratio of Average Absolute Percentage Change * NPR excluding commodity crisis years from 1973 to **NPR excluding 1998 when the exchange rate surged due to the Asian currency crisis. Note: Retail price is used for RDP. 29

30 Figure 3 and Table 3: Real Rice Prices, NPR, and Stabilization Indexes in the Philippines from 1961 to 2000 RDP RBP Peso/ton Year Period Averages Average NPR (4.66)* (-16.4)** Ratio of CV Ratio of Average Absolute Percentage Change * NPR excluding commodity crisis years from 1973 to ** From 1976 to 1981 when the domestic rice prices had been set lower than the border prices. 30

31 Figure 4: Proportion of Rice Revenue in Total Central Government Revenue in Thailand from 1961 to % Year Source: Siamwalla, A., and S. Setboonsarng (1989) "Trade, Exchange Rate, and Agricultural Pricing Policies in Thailand," Washington D. C.: The World Bank. 31

32 Figure 5: Minimum Wage Rate Enforced by the Government in Non-agricultural Sector in Thailand from 1973 to 1999 Counry Average Bangkok Baht/day Year Source: Ministry of Labor and Social Welfare (Thailand) 32

33 Figure 6: Evolution of Estimated Transmission Elasticity in Thailand, Indonesia, and the Philippines from 1961 to Thailand Indonesia Philippines Elasticity Year 33