The World’s Largest Open Access Agricultural & Applied Economics Digital Library

This document is discoverable and free to researchers across the

globe due to the work of AgEcon Search.

Help ensure our sustainability.

Give to AgEcon Search

AgEcon Search
http://ageconsearch.umn.edu
[email protected]

Papers downloaded from AgEcon Search may be used for non-commercial purposes and personal study only.
No other use, including posting to another Internet site, is permitted without permission from the copyright
owner (not AgEcon Search), or as allowed under the provisions of Fair Use, U.S. Copyright Act, Title 17 U.S.C.
 1 Rice Price Transmission between Wholesalers and Retailers in the

2 Philippines: Are Prices Integrated in Local Markets?

3
4 Bijay Chaudhary1, L. Emilio Morales1 and Renato Villano1

1
5 UNE Business School, University of New England, Australia.

6
7 ABSTRACT

8 Increasing attention has been given to raising commodity prices due to its negative effects

9 on poverty and undernutrition. An example of this problem are the growing rice prices in

10 Philippines, which are causing high living expenses to the population across the country.

11 To assess the competitiveness of agro-food chains, price transmission has been used as an

12 indicator of market integration. Using monthly data for the period 2000 to 2016, this study

13 tests vertical price transmission between wholesale and retail prices and dynamic

14 relationship between them in five local markets in Philippines. Results demonstrate that

15 retail prices are granger caused by wholesale prices in all local markets. An autoregressive

16 distributed lag (ARDL) model confirms that asymmetry in rice price transmission between

17 wholesale and retail levels in Metro Manila and Davao. In addition, the ARDL model also

18 confirms retail rice prices in all markets studied in Philippines depend on previous retail

19 prices, contemporaneous wholesale prices and wholesale prices lagged one and two

20 periods, depending on the location. Impulse Response Functions (IRFs) show the retail

21 price response initiates almost immediately or at most one month later after shock, i.e.

22 negative and positive change, on wholesale price, and the duration of full price adjustments

1
23 tend to be considerably longer in all five local markets in Philippines. [EconLit citations:

24 C32, L11, Q13].

25 1. INTRODUCTION

26 Rice price in Philippines is higher in comparison with other major rice producing Asian

27 countries such as Vietnam, Thailand and China. The largest sources of higher rice price in

28 the Philippines are the costs for transportation, milling, packaging, working capital and

29 import restriction (The Philippine Rice Research Institute (PhiRice), 2016). The PhiRice

30 (2016) also said that the gross marketing margin (GMM) is higher at the different stages

31 of rice supply chain in Philippines, and it is due to the high costs of marketing and the

32 enormous returns to trade management. The rice prices in Philippines have fluctuated

33 dramatically in the last decade, with consumers facing increasingly high prices that reached

34 exceptional levels before falling during the financial crisis over the second half of 2007

35 and first half of the 2008 (FAO, 2016). According to FAO (2011), Zorya, Townsend and

36 Delgado (2012) and Morales (2018), imperfections in price transmission are factors that

37 have contributed to exacerbate price fluctuations of food commodities due to the lack of

38 incentives transmitted to chain actors for markets adjust to shocks in supply and demand.

39 The degree of market integration in agro-food chains is affected by variations in magnitude,

40 delays and asymmetries in price transmission between positive and negative price shocks

41 (Bunte, 2006; Aramyan and Kuiper, 2009; Swinnen and Vandeplas, 2014). In this context,

42 market prices could be imperfect signals sent to actors, which could allocate suboptimal

43 resources to production. Under this scenario, the quantity and quality of products offered

44 in the market could be affected, with negative consequences for consumers and actors

45 across the chain (Norwood and Lusk, 2008).

2
46 According to Rapsomanikis and Mugera (2011), imperfections in price transmission are

47 considered as evidence of market failure and require policy interventions to control the

48 level of market power of some actors in agro-food chains. Producers/wholesalers when

49 increase prices, the retailers instantly and completely increase their prices to maintain their

50 normal profit margins, but when producers/wholesalers decrease prices, the retailers keep

51 constant their prices or takes time to reduce prices to capture higher profit margins

52 (Schroeder, 1988; Vavra & Goodwin, 2005). Swinnen and Vandeplas (2014) argued that

53 consumers in developing countries are hurt by increasing food prices, while producers are

54 not benefiting from high prices for their products, increasing poverty and hunger. Meyer

55 and von Cramon-Taubadel (2004) also claimed that the asymmetric price transmission

56 (APT) possibly results on consumers not benefitting from price reductions at the producers’

57 level, and producers might not benefit from price increases at the retail level. The

58 asymmetric price transmission, in terms of magnitude and time delay in price adjustment

59 mechanism, raised serious concerns in Philippines about market integration between

60 wholesale and retail markets. Very few studies have been conducted on price transmission

61 in the Philippines rice markets, and most of them were done before the global economic

62 crisis in 2007-2008. Therefore, to our knowledge, this is the first study that investigates the

63 dynamics of price adjustment and vertical price transmission between wholesale and retail

64 prices of milled rice in local markets in Philippines. In this paper, we examine the causal

65 relationships and empirically observe asymmetries in price transmission between

66 wholesale and retail prices, and the dynamics of price adjustment in milled rice prices in

67 rice markets in Philippines.

3
68 Market imperfections in agro-food markets are more prevalent in developing countries

69 compared to developed countries (Morales, 2018). Imperfections in price transmission are

70 due to several factors such as market power, processing and marketing costs, costs of

71 transportation, government intervention, and product homogeneity and differentiation, in

72 addition to market failure (Meyer and von Cramon-Taubadel, 2004). Frey and Manera

73 (2005) stated that the main cause of imperfect transmission from wholesale to retail is that

74 retailers allegedly try to maintain their “normal” profit margin by increasing retail prices

75 when wholesale prices rise, but they try to capture the larger margins keeping constant the

76 retail prices when wholesale prices fall, which results at least temporarily in APT. In the

77 case of Philippines, in the context of a developing country, we expect market imperfections

78 affecting rice markets.

79 Rice is the most consumed food across the Philippines, with a share of the total food

80 consumption per person very high and increasing from 68.56% in 1999-2000 to 78.99% in

81 2008-2009 (Philrice, 2016). Growing rice prices in the Philippines represent high living

82 expenses to the population across the country and more adverse effects on poverty, because

83 the share of rice in total food consumption is high for poor peoples in Philippines which

84 increases the expenditure for food consumption (Philrice, 2016). The historical data on rice

85 consumption rate shows that it tends to increase over time, though the rice price rise,

86 causing the rice consumption rate is inelastic to its price in the Philippines (Philrice, 2016

87 & FAO, 2016). The degree to which price shocks at one level of the rice chain are

88 transmitted to other levels in local markets is often taken to be an important indicator of

89 market power in supply chain. The high food prices to consumers and large marketing

90 margins to traders at certain stages in supply chain, therefore the unbalanced marketing

4
 91 margins among traders are most important issues facing policy makers. Thus, deeper

92 understanding about magnitude, speed and asymmetry to which wholesale prices are being

93 transmitted to retailer prices is a key factor in designing appropriate policy measures to

94 reduce the level of living expenses to individuals. Thus, the imperfections in rice markets

95 could have serious economic impacts to households in the Philippines. Policy initiatives in

96 this country indicate that market reform in rice market can lead to a reduction in the number

97 of poor people in the country as it helps to reduce the food expenses for individuals

98 (Cororaton, 2004), and to achieve such kind of benefits perfect price relationships between

99 various market levels in rice supply chain in markets is an essential condition.

100 Previous studies on rice markets in the Philippines such as Reyes et al. (2009) analyzed

101 the impact of changes in rice prices on poverty; Pede et al. (2013) investigated dynamics

102 on rice prices, i.e. monthly rice prices changes over the period of January 1990 to December

103 2012 in 16 regions in Philippines at three market levels: farmgate, wholesale and retail;

104 Jolejole-Foreman and Mallory (2011) analyzed the movement of Philippine rice price

105 margins between farmgate and retail affected by government intervention measures; and

106 Ramos, E. V. empirically tested the presence of seasonality in palay and rice price series

107 from 1972 to 2008 and the speed of price transmission between farm, wholesale and retail

108 levels on local markets in Philippines: Nueva Ecija, Illoilo and North Cotabato. But these

109 above-mentioned studies did not conduct empirical test on asymmetries in price

110 transmission between chain levels in local rice markets in the Philippines. Consequently,

111 this study aims to explore whether there are price transmission imperfections in the

112 Philippines rice markets and report its results and welfare implications to policy makers.

113 Hence, this paper i) tests the causality directions of rice prices between wholesale and retail

5
114 levels; ii) examines asymmetries in price transmission between wholesale and retail prices

115 in different rice markets; and 3) assesses the dynamic relationships between wholesale and

116 retail rice prices.

117 The remaining of this paper is organized as follows: Section 2 briefly review relevant

118 literature about vertical price transmission analysis, Section 3 introduces the data which is

119 used for the analysis, Section 4 describes the econometric methods for the vertical price

120 transmission analysis and dynamics of price series, Section 5 presents the main findings

121 and its discussions, and Section 6 provides the conclusions.

122 2. VERTICAL PRICE TRANSMISSION IN AGRO–FOOD CHAINS

123 Vertical price transmission has been studied to better understand the nature of price

124 movements from one level to other in agro-food chains. Several methods have been used

125 in previous studies, including von Cramon-Taubadel (1997), Conforti (2004), Varga

126 (2007) Acosta and Valdes (2014), and Ahn and Lee (2015), to analyze the direction,

127 magnitude and speed with which price changes are transmitted along the various stages of

128 the agro-food chain. The price variations may reveal different kinds of asymmetries in

129 intensity and nature depending upon the direction of price transmission in supply chain.

130 Research and Consulting in Economics (Areté) (2012) argued that in agro-food supply

131 chains, the increase in input prices are more rapidly (and often fully) transmitted to

132 downstream along supply chain, but the reduction in input prices do not transmit or may

133 take more time to be transmitted to the final market levels. The assessment of magnitude

134 and speed of price movement through supply chain is often used as an indicator of the

6
135 effectiveness and efficiency of the chain as well as the degree of competitiveness in food

136 processing and distribution.

137 Vavra and Goodwin (2005), Commission of the European Communities (CEC) (2009)

138 and Areté, (2012) stated that the assessment of vertical price transmission along the supply

139 chain typically aims to address the issues: the magnitude, speed, and the asymmetry of

140 price adjustment through the chain. In recent years, extensive studies have been done to

141 examine market linkages among market levels such as: farm, wholesale and retail levels;

142 and most of the literature on vertical price transmission refers to noncompetitive markets

143 due to market imperfections, i.e. incomplete and time delay in price transmission (von

144 Cramon-Taubadel & Loy, 1996; von Cramon-Taubadel, 1998; Conforti, 2004; Vavra &

145 Goodwin, 2005; Capps & Sherwell, 2005; Acosta & Valdes, 2013; Ahn & Lee, 2015).

146 Developing appropriate models for analyzing price transmission and testing

147 asymmetries is key to study market integration in agro-food chains. In the literature, there

148 are econometric methods for testing APT in agricultural commodities markets which are

149 still being used. In the very previous period, researchers have developed pre-cointegration

150 approaches for testing APT. Tweeten and Quance (1969) introduced a dummy variable in

151 the symmetric and linear price transmission model for estimating APT, and the dummy

152 variables are split the prices into two parts: increasing and decreasing input prices.

153 Wolffram (1971) proposed another empirical model that explicitly includes first

154 differences of explanatory price series in the equation. Houck (1977) developed another

155 model for testing APT, which is like Wolffram’s model, and this model does not consider

156 initial observations of price series data into account, because according to him the level of

157 the first observation do not have power to cause dependent variable while considering

7
158 differential effects. Ward (1982) modified the Houck’s specifications by considering time

159 lags on the explanatory variables. Meyer and von Cramon-Taubadel (2004), Frey and

160 Manera (2005), and Hassouneh et al. (2012) have reviewed the existing empirical models

161 for testing APT.

162 Granger and Newbold (1974) discovered that there could be spuriously significant

163 results between non-stationary and highly autocorrelated stationary time series. To avoid a

164 potential spurious regression, tests have been developed to identify non-stationarity and

165 models to account for co-integration between time series i.e. the time series variables share

166 similar stochastic trends and they never diverge too far from each other. Granger and Lee

167 (1989) proposed a modeling for estimating asymmetric price transmission between co-

168 integrated variables using an error correction model (ECM). Von Cramon-Taubadel and

169 Loy (1996) suggested the empirical specification by splitting the explanatory variable into

170 positive and negative components to allow for more complex dynamic effects. According

171 to Frey and Manera (2005) some researchers also assume that the dependent variable

172 depends on its own lags and on vector of explanatory variables, both contemporaneous and

173 lagged. Thus, they applied an Autoregressive Distributed Lag (ARDL) model to incorporate

174 asymmetries in price transmission by assuming that the explanatory variables have a

175 different impact on dependent variable, according to whether it is increasing or decreasing.

176 In addition, vectors can be used instead of single equational specifications, i.e.

177 multivariate extension of the uni-equational specification for estimating asymmetries in

178 price transmission. The vector models such as Vector Auto Regressive (VAR) and Vector

179 Error Correction Model (VECM) models are generalized from the standard single equation

180 analysis of price asymmetries to system of equations to take account the potential

8
181 interdependencies among time series data and other exogenous variables. Some studies

182 such as Conforti (2004), Acosta and Valdés (2014), and Ahn and Lee (2015) among others,

183 also tested the causality direction of price influences and lag distribution for adjustment of

184 price transmission in agricultural commodity prices in different market levels.

185 Evidence of asymmetries in price transmission has been detected in several previous

186 studies including producer and wholesaler pork prices in Northern Germany (von Cramon-

187 Taubadel, 1998); producer, wholesaler and retailer for several agricultural product prices

188 across Africa, Latin Ameraica and Asia (Conforti, 2004); beef, chicken and eggs in US

189 farm (Vavra & Goodwin, 2005); farm and retail milk prices in US (Capps & Sherwell,

190 2005); pork and dairy products in EU (CEC, 2009); producer and wholesale milk prices in

191 Panama (Acosta & Valdés, 2014); and shipping and terminal prices of fresh apples, table

192 grapes and fresh peaches within Washington and California (Ahn & Lee, 2015).

193 3. DATA

194 Monthly wholesale and retail price time series of milled rice for the period January 2000

195 to March 2016 in five local markets in the Philippines were obtained from the “Food and

196 Agriculture Organization of the United Nations – Food Price Monitoring and Analysis

197 (FAO – FPMA) Tool”. The price series in Philippines pesos per kilogram (PHP/kg) were

198 obtained for five selected local rice markets in the Philippines, including Metro Manila,

199 Cebu, Davao, Iloilo and South Cotabato, which are indicated in Figure 1.

200 [Figure 1 about here]

9
201 The series were deflated to the base year 2000 using the consumer price index (CPI) for

202 the Philippines (Index Mundi, 2016). Table 1 provides a summary of statistics of wholesale

203 and retail rice price series for the five selected market locations, where the wholesale and

204 retail prices reached highest levels in ‘Davao’ than other market locations with high

205 standard deviations in both markets, wholesale and retail, implying a high price variation.

206 In contrast, the standard deviations for both wholesale and retail market prices are smaller

207 in ‘Metro Manila’ than other markets.

208 [Table 1 about here]

209 Figure 2 shows that wholesale and retail price series fluctuated during the period under

210 analysis, and they reached a peak in all markets during 2008, which is related to the global

211 financial crisis. Though Philippines is an eight largest rice producer, it is also a rice deficit

212 country that imports around 10 percent of the rice consumption to meet its demand which

213 makes it a single largest rice importer in the world (FAO, 2016; Philippines Ricepedia,

214 2016). Being the largest rice importer, global rise in rice prices transmitted to the

215 Philippines rice market and it caused high rice prices in domestic markets. After the peak

216 value, the price series in all rice markets started to slightly decline. Figure 2 also

217 demonstrates that the margin between wholesale and retail markets are comparatively

218 higher in Metro Manila and Iloilo than the other three markets – Cebu, Davao and South

219 Cotabato. This could be due the concretized relationship between large retailers and

220 manufacturers in Metro Manila and Iloilo where manufacturers could deliver larger amount

221 of product to the retailers’ own centralized warehouse (Dueñas-Caparas, 2005). The setup

222 could help the retailer to internalize the wholesaling and transportation function into its

223 own activities which could provide more market power to the retailers.

10
224 [Figure 2 about here]

225 4. ECONOMETRIC METHODS

226 This research tests asymmetry in vertical price transmissions of milled rice, i.e.

227 transmission of price shocks between wholesale and retail rice prices in different local

228 markets to investigate the extent of impact of shocks at one market level (wholesale or

229 retail) to the other market level (retail or wholesale). Before developing the appropriate

230 empirical modeling for price transmission between price series, the characteristics of price

231 series and the causal direction between them must be confirmed at first. Therefore, in this

232 study the first step was to determine whether the price series have a unit root or not. The

233 Augmented Dickey-Fuller (ADF) (1979) test is usually carried out for testing the

234 stationarity characteristics of price series data (Dickey & Fuller 1979; Frey & Manera

235 2005; Hill et al., 2012; Greb et al., 2012).

236 The reliability of unit root test is highly dependent on the inclusion of the intercept and

237 time trend in the model equation. So, these terms are considered in the equation only if

238 they appear significant in value. Sometimes ADF tests cannot capture the trend in time

239 series data, therefore the Elliott, Rothenberg and Stock (ERS) (1996) and Ng-Perron (2001)

240 tests were also performed to confirm the stationarity of time series price data. Rapach and

241 Weber (2004) stated that ERS (1996) and Ng-Perron (2001) tests are more reliable because

242 of its detrending data and size adjusted properties (Morales et al., 2017).

243 The tests found the price series do not contain unit root, so they are not conintegrated.

244 The bivariate VAR model in matrix form, presented in equation (1), was used to determine

11
245 the optimal lag orders and Granger Causality to assess the possible direction of the price

246 transmission (Brooks 2014, p. 333; Ahn & Lee, 2015):

𝑛
𝑃𝑤,𝑡 𝛼1 𝛽 (𝑘) 𝛽12 (𝑘) 𝑃𝑤,𝑡−𝑘 𝜀 (𝑘)
247 (1) [ ] = [ 𝛼 ] + ∑ [ 11 ][ ]+[ 1 ]
𝑃𝑟,𝑡 2 𝛽21 (𝑘) 𝛽22 (𝑘) 𝑃𝑟,𝑡−𝑘 𝜀2 (𝑘)
𝑘=1

248 where 𝑃𝑤,(𝑡) and 𝑃𝑟,(𝑡) are rice price series at wholesale and retail levels, respectively, 𝛽𝑖𝑗

249 is the coefficient at kth lag and 𝜀𝑖 (𝑘) is a white noise residual with mean zero, 𝑘 = 1,2 ⋯ 𝑛,

250 and ‘n’ is the optimal lags determined from equation (1). The optimal lag order is selected

251 based on the Schwartz Bayesian Information Criterion (SBIC), minimum value criteria.

252 Granger causality tests are performed based on the expressed individual equations from

253 equation (1), i.e. 𝑃𝑤,𝑡 = 𝛼1 + ∑𝑛𝑘=1 𝛽11 (𝑘) 𝑃𝑤,𝑡−𝑘 + ∑𝑛𝑘=1 𝛽12 (𝑘) 𝑃𝑟,𝑡−𝑘 + 𝜀1 (𝑘) and

254 𝑃𝑟,𝑡 = 𝛼2 + ∑𝑛𝑘=1 𝛽21 (𝑘) 𝑃𝑤,𝑡−𝑘 + ∑𝑛𝑘=1 𝛽22 (𝑘) 𝑃𝑟,𝑡−𝑘 + 𝜀2 (𝑘), where the lags are

255 specified using the findings on optimal lags. Therefore, to determine the causal direction

𝛽11 (𝑘) 𝛽12 (𝑘)

256 between variables, all cross-lag coefficients or coefficient matrix, 𝑀 = [ ]
𝛽21 (𝑘) 𝛽22 (𝑘)

257 where 𝑘 = 1,2 ⋯ 𝑛, can be tested by Wald Statistics. From this Granger causality tests, we

258 can get four possible causal results between two price series 𝑃𝑤,𝑡 and 𝑃𝑟,𝑡 : i) 𝑃𝑤,𝑡 causes

259 𝑃𝑟,𝑡 but 𝑃𝑟,𝑡 does not cause 𝑃𝑤,𝑡 ; ii) 𝑃𝑤,𝑡 does not cause 𝑃𝑟,𝑡 but 𝑃𝑟,𝑡 causes 𝑃𝑤,𝑡 ; iii) 𝑃𝑤,𝑡

260 causes 𝑃𝑟,𝑡 and 𝑃𝑟,𝑡 also causes 𝑃𝑤,𝑡 ; and iv) 𝑃𝑤,𝑡 does not cause 𝑃𝑟,𝑡 and 𝑃𝑟,𝑡 also does not

261 cause 𝑃𝑤,𝑡 .

262 In this research the rice price series 𝑃𝑤,𝑡 and 𝑃𝑟,𝑡 are used for estimating asymmetries in

263 vertical price transmission between wholesale and retail levels in rice chains. The price

264 transmission analyses were conducted separately on five different local markets across the

12
265 Philippines. As the unit root tests identified the 𝑃𝑟,𝑡 and 𝑃𝑤,𝑡 are stationary, i.e. I(0), in all

266 five markets, the Autoregressive Distributed Lag (ARDL) model with an n lag length

267 determined by Lag Oder Choice based on SIC criteria, is applied for testing asymmetries

268 in price transmission between these price series. For model specification, we considered

269 𝑃𝑟,𝑡 depends on its own monthly lagged price and the current and monthly lagged of 𝑃𝑤,𝑡 ,

270 where the price series, 𝑃𝑟,𝑡 and 𝑃𝑤,𝑡 , are I(0), and the ARDL model can be represented as:

𝑛 𝑛 𝑛 𝑛

271 (2) 𝑃𝑟,𝑡 = 𝛼 + ∑ 𝛽𝑖+ 𝑃𝑟,𝑡−𝑖

+
+ ∑ 𝛽𝑖− 𝑃𝑟,𝑡−𝑖
−
+ ∑ 𝛾𝑖+ 𝑃𝑤,𝑡−𝑖
+
+ ∑ 𝛾𝑖− 𝑃𝑤,𝑡−𝑖
−
+ 𝑒𝑡
𝑖=1 𝑖=1 𝑖=0 𝑖=0

+ +
𝑃𝑟,𝑡 𝑖𝑓 ∆𝑃𝑟,𝑡−1 ≥ 0 𝑃𝑤,𝑡 𝑖𝑓 ∆𝑃𝑤,𝑡−1 ≥ 0
272 where 𝑃𝑟,𝑡 = { − , 𝑃𝑤,𝑡 = { − .
𝑃𝑟,𝑡 𝑂𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒 𝑃𝑤,𝑡 𝑂𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

273 The tests of asymmetric price transmission are based on the parameter estimates,

274 𝛽𝑖+ , 𝛽𝑖− , 𝛾𝑖+ , 𝑎𝑛𝑑 𝛾𝑖− in equation (2). For example, the hypothesis 𝐻0 : 𝛾0+ = 𝛾0− provides an

275 immediate test of asymmetry between contemporaneous prices, 𝑃𝑟,𝑡 and 𝑃𝑤,𝑡 . If these

276 coefficients are significantly different from each other, contemporaneous asymmetry

277 exists. Estimating the effects of 𝑃𝑟,𝑡 and 𝑃𝑤,𝑡 is simple at the current period because of only

278 one explanatory variable, 𝑃𝑤,𝑡 exists. However, the period moves into the future, the

279 effects of 𝑃𝑤,𝑡 becomes less clear because the term 𝑃𝑤,𝑡 entered as lagged terms in equation

280 (2) at the future period which can influences the future 𝑃𝑟,𝑡 directly as a lagged wholesale

281 prices as well as indirectly through lagged retail prices. Thus, for the comprehensive

282 analysis of price transmission, the dynamic multiplier approach requires which captures

283 both the direct effects of 𝑃𝑤,𝑡 and indirect effects that are realized through lagged retail

284 prices over the multiple periods (Ahn & Lee, 2015). So, tracing all these effects, if

13
285 ∑𝑛𝑖=0 𝛾𝑖+ 𝑎𝑛𝑑 ∑𝑛𝑖=0 𝛾𝑖− are significantly different, asymmetry exists between two price

286 series in long run.

287 In addition to usual test of asymmetry, the present study extends the test of asymmetry

288 to dynamic multiplier effects by performing Impulse Response Functions (IRFs) to

289 construct the dynamic relationships between wholesale and retail prices over the multiple

290 periods in five local markets. The pattern of dynamic multiplier effects for each successive

291 period gives insight about how the retail price adjusts in response to the initial shock in the

292 wholesale price. Therefore, the comprehensive effect of initial shock can be obtained by

293 summing up the dynamic multiplier effect at each period. These complete effects on retail

294 price under the nth lag order can be expressed algebraically. For instance, the positive

295 shock of wholesale price (𝑃𝑤,𝑡 ) under the nth lag order can be expressed as:

296 (3a) 𝑃̂𝑟,𝑡 = (𝛾0+ 𝑃𝑤,𝑡 ),

297 (3b) 𝑃̂𝑟,𝑡+1 = (𝛾1+ 𝑃𝑤,𝑡 ) + (𝛽1 𝑃̂𝑟,𝑡 ),

298 (3c) 𝑃̂𝑟,𝑡+2 = (𝛾2+ 𝑃𝑤,𝑡 ) + (𝛽2 𝑃̂𝑟,𝑡 + 𝛽1 𝑃̂𝑟,𝑡+1 ),

299 ⋯⋯ ⋯⋯ ⋯⋯ ⋯⋯ ⋯⋯ ,

300 (3d) 𝑃̂𝑟,𝑡+𝑛 = (𝛾𝑛+ 𝑃𝑤,𝑡 ) + (𝛽𝑛 𝑃̂𝑟,𝑡 + 𝛽𝑛−1 𝑃̂𝑟,𝑡+1 + ⋯ + 𝛽1 𝑃̂𝑟,𝑡+𝑛−1 ).

301 The structural vector autoregressive (SVAR) model represented in equation (4) is also

302 applied in this study to test the contemporaneous relationships between these price series

303 in markets, where bi-directional causality found:

304 (4) 𝐴𝑃𝑡 = 𝛾 + 𝐵𝑃𝑡−1 + 𝑒𝑡

14
305 where, 𝑃𝑡 is a vector of prices at time t, 𝑃𝑡−1 is first month lag term of prices, A and B are

306 2 × 2 square matrices, and 𝛾 𝑎𝑛𝑑 𝑒𝑡 are 2 × 1 column vector matrices.

307 The price transmission between the contemporaneous prices is estimated by imposing

308 short-run restriction on the SVAR model equation (4) by creating matrix ‘A’ as lower-case

1 0 𝛽 0
309 matrix and matrix ‘B’ as diagonal matrix, i.e. 𝐴 = ( ), 𝐵 = ( 11 ). If
𝛼21 1 0 𝛽22

310 the coefficients of contemporaneous price series (lower case in matrix A) are found

311 significant, contemporaneous effects are existed in price transmission between price series.

312 If the diagonal coefficients in matrix B are found significant, we can say that its lag term

313 is significant in price transmission.

314 5. RESULTS AND DISCUSSIONS

315 Unit-Root Tests

316 The results of the unit root tests reported in Table 2, indicate that for the wholesale price

317 (𝑃𝑤,𝑡 ) and retail price (𝑃𝑟,𝑡 ) there is sufficient evidence to reject the null hypothesis of unit

318 roots, i.e. non-stationarity. The ADF (1979) tests show the sign of stationary for wholesale

319 prices in markets – Cebu, Davao, Iloilo and South Cotabato, and for retail prices in markets

320 – Cebu, Davao and South Cotabato. In contrast, the ADF (1979) test results indicate that

321 both wholesale and retail prices in Metro Manila and retail price in Iloilo are non-

322 stationary. Furthermore, the stronger unit root tests such as ERS (1996) and Ng-Perron

323 (2001) tests result show the evidence of stationarity for wholesale and retail price series in

324 all local markets. Therefore, the wholesale and retail price series in all local markets are

325 stationary, i.e. integrated order zero I(0). This is a similar outcome to those reported by

15
326 Ahn and Lee (2015). So, this study used the price series data at level for the model

327 specification to estimate price transmission. But these unit root test results contrast with

328 those reported by von Cramon-Taubadel (1998), Conforti (2004), Vavra and Goodwin

329 (2005), Capps and Sherwell (2005), and Acosta and Valdés (2013), who identified unit

330 roots in price series of agro-food products, and their first differences were stationary.

331 Consequently, they used price series in first differences for estimating price transmission.

332 [Table 2 about here]

333 Lag Order Choice and Causality Tests

334 The test results of optimum lag order choice presented in Table 3, were based on the VAR

335 model equation (1) and the optimum lag orders were selected using the Schwartz

336 Information Criteria (SIC) – minimum value criteria.

337 [Table 3 about here]

338 The optimum lag orders were found one lag order for Cebu and Iloilo, and two lag orders

339 for the Metro Manila, Davao and South Cotabato which are used for Granger Causality

340 tests between 𝑃𝑟,𝑡 and 𝑃𝑤,𝑡 in all local markets. The price transmission models include two

341 lags, as it is the length that is recommended in most of locations. The Granger Causality

342 tests results shown in Table 4 confirmed the presence of causality between wholesale and

343 retail prices in all five markets. In market locations – Metro Manila, Cebu and Iloilo, the

344 results showed wholesale rice price granger causes retail rice price at the 1% level, but

345 retail rice price do not granger cause wholesale rice price, i.e. there is uni-directional

346 granger causality in these markets. This observed causality direction is comparable to that

16
347 identified by Ahn and Lee (2015), i.e. the upstream prices Granger-cause downstream

348 prices. The results also indicated that the retail price granger cause wholesale price at the

349 1% and 5% level in Davao and South Cotabato, respectively. Therefore, there is bi-

350 directional causality between wholesale and retail prices in these markets. The Granger-

351 causalities identified in this study are significant which are different from those reported

352 by Conforti (2004), who found inconclusive Granger-causality within domestic markets in

353 several agricultural products such as for pork meat in Costa Rica, wheat and bovine meat

354 in Egypt, maize in Ethiopia, sorghum, palm oil and cassava in Ghana, and rice in Turkey.

355 [Table 4 about here]

356 Estimation Results of Price Transmission

357 We specified an ARDL model equation to assess the asymmetric relationship between the

358 wholesale and retail price series in the five local markets. The results of the Granger

359 causality test indicate that in the setting of ARDL, the current retail price series (𝑃𝑟,𝑡 ) is

360 dependent variable and should be on the left-hand side. The estimation results of ARDL

361 tests presented in Table 5, indicate that the current wholesale price (𝑃𝑤,𝑡 ) and one-month

362 lagged retail price (𝑃𝑟,𝑡−1 ) have positive effects on the 𝑃𝑟,𝑡 , and their impact is significant

363 at the 1% level in all local markets. This implies that changes in 𝑃𝑤,𝑡 and 𝑃𝑟,𝑡−1 caused

364 changes in 𝑃𝑟,𝑡 in same direction. In contrast, the one- month lagged wholesale price

365 (𝑃𝑤,𝑡−1 ) has negative effect on the 𝑃𝑟,𝑡 , and the impact was also significant at the 1% level

366 in locations – Metro Manila, Cebu and Davao, and significant at the 5% level in South

367 Cotabato. This result suggests the 𝑃𝑟,𝑡 changes in opposite direction with 𝑃𝑤,𝑡−1 which

368 implies that when wholesale price increase (decrease) caused the retail price decrease

17
369 (increase) after one month. The ARDL outputs also suggests that the two-month lagged

370 retail price (𝑃𝑟,𝑡−2 ) do not have significant impact on 𝑃𝑟,𝑡 in all local markets implying that

371 when shock comes on current retail price, it does not make any changes on retail price after

372 two-months. But the two-month lagged wholesale price (𝑃𝑤,𝑡−2 ) has significant positive

373 effect on 𝑃𝑟,𝑡 in the markets Cebu and Davao at the 5% level, which means that the retail

374 price increase (decrease) after two-months of wholesale price increase (decrease).

375 [Table 5 about here]

376 The vertical price transmission estimation results demonstrate that there is evidence of

377 asymmetry in price transmission between wholesale and retail prices in the short and long

378 run at 5% significance level in the markets, Metro Manila and Davao. This outcome

379 indicates that rice price shocks at wholesale level do not fully transmit to the retail level in

380 the short and long run. In contrast, the results corroborate that there is symmetry in price

381 transmission between wholesale and retail prices in Iloilo and South Cotabato in the short

382 and long run at 5% significance level. In Cebu, the estimated results demonstrate that there

383 is asymmetric price transmission in the short run between wholesale and retail prices, but

384 it is symmetric in the long run at 5% significance level. The vertical price transmission

385 results in the rice markets of Cebu, Iloilo and South Cotabato in the Philippines are different

386 than the results obtained in previous studies where asymmetry in price transmission was

387 found in number of agro-food products along supply chains, including von Cramon-

388 Taubadel (1998) for pork prices in northern Germany, Vavra and Goodwin (2005) for U.S.

389 beef, chicken and egg markets, Acosta and Valdés (2014) for milk prices in Panama, Ahn

390 and Lee (2015) for fresh fruits in the Western United States.

18
391 In addition, the results of the Granger causality tests indicated that there is bi-directional

392 causality between wholesale and retail prices in Davao and South Cotabato. Hence, we

393 estimated the contemporaneous relationships between these price series in these two

394 markets using the SVAR model equations (4) imposing short-run restrictions.

395 The SVAR estimated results showed in Table 6 indicate the lower coefficients in matrix

396 A-1 are statistically significant at 1% level in both locations, Davao and South Cotabato,

397 implying that there are contemporaneous effects between wholesale and retail prices. This

398 could be due to a reduced concentration of market power, which could be the consequence

399 of more competitive conditions in these markets. The results also show the diagonal

400 coefficients in matrix B are significant in 1% level, which implies both price series depend

401 on its own first month lagged terms in both markets.

402 [Table 6 about here]

403 Dynamic Multiplier Effects

404 Based on the expressions (3a) – (3d) the dynamic multiplier effects and parameter estimates

405 presented in table 5, we derive the responses of the retail prices to positive and negative

406 impulses on the wholesale prices. We use the absolute value of one standard deviation

407 (S.D.) as a magnitude of initial shocks of wholesale prices to represent a typical change in

408 monthly wholesale price. The positive and negative shocks are prescribed simply by taking

409 positive and negative change values of these price series.

410 [Figure 3 about here]

19
411 Figure 3 presents the resulting dynamic multiplier effects of retail prices and the lines

412 corresponds to the retail price responses to the positive and negative shocks, equivalent to

413 one S.D., in wholesale price. IRFs presented in figure 3 shows responses of retail prices in

414 all five markets seems similar in terms of magnitude and duration in price transmission.

415 First, IRFs demonstrate the impacts to retail price in second month due to shocks in

416 wholesale price in all five markets; the dynamic multiplier effect and the duration of the

417 full adjustment are long in all markets. Second, the response and the price transmission

418 effect tend to be most intense after several months and its tend to be tamper with time.

419 Third, the dynamic multiplier effect to retail price becomes strong in second and third

420 months due to negative and positive changes on wholesale price respectively in South

421 Cotabato, and the adjustment process is faster in South Cotabato than other four markets.

422 Fourth, the adjustment process extends over many periods till 21st month for negative

423 change and 36th month for positive change in South Cotabato but it spreads over more than

424 48 months for both negative and positive changes in Metro Manila, Cebu, Davao and Iloilo.

425 6. CONCLUSION

426 This study examines the asymmetry of price transmission between wholesale and retail

427 monthly rice prices in five different markets in Philippines, Metro Manila, Cebu, Davao,

428 Iloilo and South Cotabato. We tested the asymmetry by applying ARDL model and

429 outlined the speed of adjustment of retail price response over multiperiod to a change in

430 wholesale price that is differentiated by the direction of the change. This study also derived

431 the dynamic multiplier effects of the retail price in response to the change in wholesale

432 price based on IRFs.

20
433 The empirical results demonstrated asymmetry in price transmission between wholesale

434 and retail prices in Metro Manila and Davao in long run, but the symmetric price

435 transmission was found in Cebu, Iloilo and South Cotabato in long run. The price

436 adjustment process was faster in South Cotabato than other markets, which took twenty-

437 one months for full adjustment. But the IRFs showed the response for retail prices in Metro

438 Manila, Cebu, Davao and Iloilo gradually tampered with time and it takes more than forty-

439 eight months for full adjustment. Using monthly data enables us to find that the retail price

440 response initiates almost immediately or at most one month later after the shock and that

441 the full price adjustments tend to last a considerable time, more than forty-eight months

442 except South Cotabato.

443 In this regard of price transmission, this study suggests that the different rice markets

444 have distinct competitiveness in Philippines, and the policy makers require to pay close

445 attention in designing mechanisms other than traditional transfer approaches from

446 wholesale to retail level to increase the competitiveness in the rice markets in the supply

447 chain. Therefore, it can reduce the food expenses to the all Filipinos and help to decrease

448 a substantial number of poor peoples across the Philippines.

449

450

451

452

453

21
454 REFERENCES

455 Acosta, A. & Valdés, A. (2014). Vertical price transmission of milk prices: Are small dairy

456 producers efficiently integrated into markets? Agribusiness, 30(1), 56–63.

457 Ahn, Byeong-il, & Lee, H. (2015). Vertical price transmission of perishable products: The

458 case of fresh fruits in the western United States. Journal of Agricultural and Resource

459 Economics, 40(3), 405–424.

460 Aramyan, L., Kuiper, M., 2009. Analyzing price transmission in agri-food chains: An

461 overview. Measuring Business Excellence 13(3), 3–12.

462 Areté – Research & Consulting in Economics. (2012). Study on price transmission in the

463 sugar sector. European Commission, DG Agriculture and Rural Development,

464 TENDER No AGRI – 2011 – EVAL – 03.

465 Brooks, C. (2014). Introductory econometrics for finance (4th ed.). Cambridge University

466 Press.

467 Bunte, F., 2006. Pricing and performance in agri-food supply chains, in C.J.M Ondersteijn,

468 J.H.M. Wijnands, R.B.M. Huirne, O. van Kooten (eds.), Quantifying the agri-food

469 supply chain. Springer, Dordrecht, pp. 37–45.

470 Capps, O., & Sherwell, P. (2005, May). Spatial Asymmetry in Farm-Retail Price

471 Transmission Associated with Fluid Milk Products. Paper presented at the annual

472 meeting of American Agricultural Economics Association, Rhode Island, USA.

473 Conforti, P. (2004). Price transmission in selected agricultural markets. FAO Commodity

474 and Trade Policy Research Working Paper No.7.

22
475 Commission of the European Communities. (2009). Analysis of price transmission along

476 the food supply chain in the EU. Brussels, 28.10.2009, Commission Staff Working

477 Document.

478 Cororaton, Caesar B. (2004). Rice Reforms and Poverty in the Philippines: A CGE

479 Analysis. ADB Institute Research Paper 57.

480 Department of Agriculture, Philippines Rice Research Institute (DA–Philrice) (2016).

481 Accessed 20 October 2017 from http://www.philrice.gov.ph/.

482 Dickey, D. A., & Fuller, W. A. (1979). Distribution of the estimators for autoregressive

483 time series with a unit root. Journal of the American Statistical Association, 74(366),

484 427–431.

485 Dickey, D. A., & Fuller, W. A. (1981). Likelihood ratio statistics for autoregressive time

486 series with a unit root. Econometrica, 49(4), 1057–1072.

487 Dueñas-Caparas, Ma. T. (2005). State of Competition in the Wholesale and Retail Sector.

488 Philippine Institute for Development Studies. Discussion Paper Series No. 2005–05.

489 Elliott, G., Rothenberg, T. J., & Stock, J. H. (1996). Efficient tests for an autoregressive

490 unit root. Econometrica, 64, 813–836.

491 Food and Agriculture Organization of the United Nations (FAO). (2011). The state of food

492 insecurity in the world. How does international price volatility affect domestic

493 economies and food security? FAO Publishing Policy and Support Branch, Rome,

494 Italy. Accessed 18 January 2018 from http://www.fao.org/docrep/014/i2330e/

495 i2330e.pdf.

23
496 Food and Agriculture Organization of the United Nations (FAO), (2016). Food price

497 monitoring analysis tool (FPMA). Accessed 10 January 2017 from

498 www.fao.org/giews/food-prices/en/.

499 Frey, G., & Manera, M. (2007). Econometric models of asymmetric price transmission.

500 Journal of Economic Surveys, 21(2), 349–415.

501 Google (2016). Map of selected rice markets in Philippines. Accessed 10 December 2016

502 from http://www.freeworldmaps.net/asia/philippines/philippines-outline-

503 Granger, C. W. J. & Newbold, P. (1974). Spurious regressions in econometrics. Journal of

504 Econometrics, 2(2), 111–120.

505 Granger, C. W. J., & Lee, T. H. (1989). Investigation of production, sales and inventory

506 relationships using multicointegration and non-symmetric error correction models.

507 Journal of Applied Econometrics, 4, S145–S159.

508 Greb, F., Jamora, N., Mengel, C., von Cramon-Taubadel, S., & Wurriehausen (2012, 30

509 October – 02 November). Cereal price transmission from international to domestic

510 markets in Africa. Paper prepared for the African Economic Conference, Kigali,

511 Rwanda.

512 Hassouneh, I., Cramon–Taubadel, S. v., Serra, T., & Gil, J. M. (2012). Recent

513 Developments in the Econometric Analysis of Price Transmission. Working Paper No.

514 2, Transparency of Food Pricing TRANSFOP.

515 Hill, R. C., Griffths, W. E., & Lim, G. C. (2018). Principles of econometrics (5th ed.).

516 Wiley, Haboken, United States.

517 Houck, P. J. (1977). An approach to specifying and estimating non-reversible functions.

518 American Journal of Agricultural Economics, 59(3), 570–572.

24
519 Index Mundi (2016). Consumer price index, Philippines. Accessed 15 December 2017

520 from http://www.indexmundi.com/facts/philippines/consumer-price-index.

521 Joleiole-Foreman, Maria Christina and Mallory, Mind L. (2011, July). Analyzing price

522 margins, government intervention and weather shocks for rice market in the

523 Philippines. Paper presented at the annual meeting of Agricultural & Applied

524 Economics Associations, Pittsburgh.

525 Meyer, J., & von Cramon-Taubadel, S. (2004). Asymmetric price transmission: A survey.

526 Journal of Agricultural Economics, 55(3), 581–611.

527 Morales, L. E., Hoang, N., & Stuen, E. (2017). Spatial price premium transmission for

528 Meat Standards Australia-graded cattle: The vulnerability of price premiums to outside

529 shocks. Australian Journal of Agricultural and Resource Economics, 61(4), 590–609.

530 Morales, L. E. (2018). The effects of international price volatility on farmer prices and

531 marketing margins in cattle markets. International Food and Agribusiness

532 Management Review. Accessed 18 January 2018 from:

533 http://www.wageningenacademic.com/doi/abs/10.22434/IFAMR2017.0020.

534 Ng, S. & Perron, P. (2001). Lag length selection and the construction of unit root tests with

535 good size and power. Econometrica, 69(6), 1519–1554.

536 Norwood, F. B., Lusk, J. L., 2008. Agricultural marketing and price analysis. Pearson,

537 Prentice Hall, New Jersey.

538 Pede, Valerien O., Valera, Harold Glenn A., Alam, Mohammad J. and Mckenzie, Andrew

539 M. (2013, August). Nonlinearities in regional rice prices in the Philippines: Evidence

540 from a smooth transition autoregressive (STAR) approach. Paper presented at the

541 annual meeting of Agricultural & Applied Economics Associations, Washington DC.

25
542 Rapach, D. E. & Weber, C. E. (2004). Are real interest rates really nonstationary? New

543 evidence from tests with good size and power. Journal of Macroeconomics, 26(3),

544 409–430.

545 Rapsomanikis, G., & Mugera, H. (2011). Price transmission and volatility spillovers in

546 food markets of developing countries, in Piot-Lepetit, I. and M’Barek, R. (eds.)

547 Methods to Analyse Agricultural Commodity Price Volatility. Springer, Dordrecht, pp.

548 165–179.

549 Reyes, C. M., Sobrevinas, A. B., Bancolita, J. & Jesus, J. de (2009). Analysis of the Impact

550 of Changes in the Prices of Rice and Fuel on Poverty in the Philippines. Philippine

551 Institute for Development Studies, Discussion Paper Series No. 2009–07.

552 Schroeder, Ted C. (1988). Price linkages between wholesale and retail pork cuts.

553 Agribusiness, 4(4), 359–369.

554 Swinnen, J., & Vandeplas, A. (2014). Price transmission and market power in modern

555 agricultural value chains. LICOS Discussion Paper Series 347/2014. Accessed 12

556 January 2017 from https://feb.kuleuven.be/drc/licos/publications/dp/dp347.

557 The Philippines Rice Research Institute (Philrice) (2016). Why is per capita rice

558 consumption increasing? Rice Science Decision Makers. Vol. 3 No. 1. ISSN 2094–

559 8409.

560 The Philippine Rice Research Institute (PhilRice) and The International Rice Research

561 Institute (IRRI). (2016). Competitiveness of Philippine Rice in Asia.

562 Tweeten, L. G., & Quance, L. (1969). Positivistic measures of aggregate supply elasticities:

563 some new approaches. American Journal of Agricultural Economics, 51(2), 342–352.

26
564 Vavra, P., & Goodwin, B. K. (2005). Analysis of price transmission along the food chain.

565 OECD Food, Agriculture and Fisheries Working Papers, No. 3, OECD Publishing.

566 Accessed 15 January 2017 from http://www.oecd.org/Agriculture/Agricultural-

567 policies/40459642.Pdf.

568 von Cramon-Taubadel, S. (1998). Estimating asymmetric price transmission with the error

569 correction representation: An application to the German pork market. European

570 Review of Agricultural Economics, 25, 1–18.

571 von Cramon-Taubadel, S. & Loy, J. P. (1996). Price asymmetry in the international wheat

572 market: Comment. Canadian Journal of Agricultural Economics, 44(3), 311–317.

573 Ward, R. W. (1982). Asymmetry in retail, wholesale, and shipping point pricing for fresh

574 vegetables. American Journal of Agricultural Economics 62(2), 205–212.

575 Wolffram, R. (1971). Positivistic measures of aggregate supply elasticities: Some new

576 approaches: Some critical notes. American Journal of Agricultural Economics, 53(2),

577 356–359.

578 Zorya, S., R. Townsend and C. Delgado. 2012. Transmission of global food prices to

579 domestic prices in developing countries: why it matters, how it works, and why it

580 should be enhanced. Working Paper 71268. World Bank, Washington, WA, USA.

581 Accessed 18 January 2018 from: http://tinyurl.com/yb2lqtnn.

582

583

584

27
 Metro Manila

Iloilo
Cebu

Davao

South Cotabato
585

586 FIGURE 1 Locations of selected rice markets in Philippines

587 Source: Google

588

589

590

28
 28

26

24

22
PHP/kg
20

18

16

14
00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16

591 M ManilaW P M M anilaRP

28

26

24
PHP/kg

22

20

18

16

14
00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16

592 CebuW P CebuRP

32

28
PHP/kg

24

20

16

12
00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16

593 DavaoW P DavaoRP

32

28

24
PHP/kg

20

16

12
00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16

594 IloiloW P IloiloRP

30

28

26

24
PHP/kg

22

20

18

16

14

12
00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16

595 S c otabatoW P S CotabatoRP

MM/YY
596 FIGURE 2 Wholesale (WP) and Retail (RP) monthly prices of milled rice in local
597 markets in the Philippines, measured ‘months’ in X-axis and ‘price (PHP/kg)’ in Y-axis.
598 Source: FAO – FPMA

29
599 TABLE 1. Descriptive statistics of wholesale/retail prices from January 2000 to March

600 2016 – Philippine pesos per kilogram (PHP/kg) in base year 2000.

601 Wholesale prices (WP) – (PHP/kg)

602 Location Mean Median SD Min. Max.

603 Metro Manila 18.52 18.05 1.80 15.78 25.25

604 Cebu 18.90 18.86 2.06 15.68 26.08

605 Davao 19.49 19.28 2.53 15.11 28.50

606 Iloilo 17.04 16.85 2.28 12.69 22.95

607 South Cotabato 17.30 17.33 2.13 13.58 25.75

608 Retail prices (RP) – (PHP/kg)

609 Location Mean Median SD. Min. Max.

610 Metro Manila 20.70 20.14 1.90 17.65 26.51

611 Cebu 20.27 19.93 2.12 17.24 27.71

612 Davao 20.87 21.02 2.57 17.02 30.29

613 Iloilo 22.38 22.70 2.20 18.02 29.43

614 South Cotabato 19.30 19.01 2.39 15.36 28.37

615 Source: FAO – FMPA

616

617

618

619

620

621

622

30
623 TABLE 2 Unit root test results of wholesale and retail rice prices in the Philippines

624 Variable t-statistics

625 ADF ERS Ng–Perron (MZt)

626 Metro Manila:

627 Wholesale ‒ 1.79 ‒ 3.07*** ‒ 3.07***

628 Retail ‒ 2.34 ‒ 2.21** ‒ 2.22**

629 Cebu:

630 Wholesale ‒ 3.43* ‒ 2.85* ‒ 2.84*

631 Retail ‒ 3.30* ‒ 2.98** ‒ 2.95**

632 Davao:

633 Wholesale ‒ 5.42*** ‒ 4.74*** ‒ 4.25***

634 Retail ‒ 4.60*** ‒ 4.11*** ‒ 4.02***

635 Iloilo:

636 Wholesale ‒ 5.71*** ‒ 5.11*** ‒ 4.94***

637 Retail ‒ 2.96 ‒ 2.93* ‒ 2.81*

638 South Cotabato:

639 Wholesale ‒ 7.14*** ‒ 6.72*** ‒ 6.40***

640 Retail ‒ 6.45*** ‒ 1.75 ‒ 1.48

641 Null Hypothesis H0: Series has unit root → Non-Stationary.

642 ADF = Augmented Dickey-Fuller (1979); ERS = Elliott, Rothenberg, and Stock (1996); and Ng-Perron = Ng

643 and Perron (2001).

644 (***), (**) and (*) indicate statistical significant at the 1%, 5% and 10% level respectively.

645

646

647

31
648 TABLE 3 Lag order choice based on SIC – minimum value criteria

649 Location Lag 1 Lag 2 Lag 3 Lag 4 Lag 5

650 Metro Manila 2.1531 2.1451* 2.2307 2.2848 2.3626

651 Cebu 2.5320* 2.5369 2.6308 2.7305 2.8191

652 Davao 3.7643 3.7326* 3.8265 3.8968 3.9968

653 Iloilo 4.1945* 4.2034 4.2744 4.3262 4.3813

654 South Cotabato 4.1827 4.1636* 4.2684 4.3597 4.3525

655 *Minimum value that determines the optimal Lag Order Choice.

656

657 TABLE 4 VAR Granger Causality test results between wholesale and retail rice prices

658 Location Causality Chi2 Test Statistics d.f. p-values

659 Metro Manila H0: Wholesale do not cause Retail 13.88638*** 2 0.0010

660 H0: Retail do not cause Wholesale 0.936402 2 0.6261

661 Cebu H0: Wholesale do not cause Retail 29.59543*** 1 0.0000

662 H0: Retail do not cause Wholesale 1.779629 1 0.1822

663 Davao H0: Wholesale do not cause Retail 24.47100*** 2 0.0000

664 H0: Retail do not cause Wholesale 12.78076*** 2 0.0017

665 Iloilo H0: Wholesale do not cause Retail 21.77164*** 1 0.0000

666 H0: Retail do not cause Wholesale 1.091724 1 0.2961

667 South Cotabato H0: Wholesale do not cause Retail 32.37150*** 2 0.0000

668 H0: Retail do not cause Wholesale 7.612101** 2 0.0222

669 (***), (**) and (*) indicate statistical significant at the 1%, 5% and 10% level respectively.

32
670 TABLE 5 Estimation results for testing vertical price transmission in local markets in the Philippines

671 Metro Manila Cebu Davao Iloilo South Cotabato

672 Regressor Coefficient Coeff. Est. Std. Er. Coeff. Est. Std. Er. Coeff. Est. Std. Er. Coeff. Est. Std. Er. Coeff. Est. Std. Er.

673 𝛼 0.0107 0.2307 0.4084 0.2515 0.2842 0.2393 1.1837*** 0.4338 − 0.7522*** 0.2772
+
674 𝑃𝑟,𝑡−1 𝛽1+ 0.8891*** 0.0781 0.5190*** 0.0790 0.9840*** 0.0761 0.8776*** 0.0828 0.6739*** 0.0757

−
675 𝑃𝑟,𝑡−1 𝛽1− 0.8891*** 0.0790 0.5178*** 0.0803 0.9854*** 0.0769 0.8865*** 0.0852 0.6796*** 0.0770

+
676 𝑃𝑟,𝑡−2 𝛽2+ − 0.0507 0.0729 0.1362* 0.0836 − 0.2131*** 0.0750 − 0.0728 0.0812 − 0.0648 0.0657

−
677 𝑃𝑟,𝑡−2 𝛽2− − 0.0502 0.0732 0.1395* 0.0837 − 0.2121*** 0.0746 − 0.0664 0.0812 − 0.0640 0.0653

+
678 𝑃𝑤,𝑡 𝛾0+ 0.7759*** 0.0444 0.8790*** 0.0548 0.8242*** 0.0381 0.3912*** 0.0604 0.7136*** 0.0404

−
679 𝑃𝑤,𝑡 𝛾0− 0.7833*** 0.0459 0.8860*** 0.0569 0.8386*** 0.0402 0.3926*** 0.0643 0.7204*** 0.0442

+
680 𝑃𝑤,𝑡−1 𝛾1+ − 0.6116*** 0.0943 − 0.2921*** 0.1097 − 0.6456*** 0.0869 − 0.1116 0.0963 − 0.2026** 0.0797

−
681 𝑃𝑤,𝑡−1 𝛾1− − 0.6089*** 0.0956 − 0.2911*** 0.1111 − 0.6482*** 0.0885 − 0.1015 0.0986 − 0.1989** 0.0815

+
682 𝑃𝑤,𝑡−2 𝛾2+ 0.0111 0.0729 − 0.2420** 0.0941 0.0460*** 0.0722 − 0.1041 0.0710 − 0.0402 0.0646

−
683 𝑃𝑤,𝑡−2 𝛾2− 0.0142 0.0731 − 0.2449** 0.0945 0.0475*** 0.0728 − 0.1012 0.0708 − 0.0336 0.0651

684 Null Hypothesis F- Stat. Pr(|F| > c) F- Stat. Pr(|F| > c) F- Stat. Pr(|F| > c) F- Stat. Pr(|F| > c) F- Stat. Pr(|F| > c)
685 (df) (df) (df) (df) (df)

686 𝛾0+ = 𝛾0− 6.5906 0.0111 4.3969 0.0374 16.2504 0.0001 0.0492 0.8246 1.6217 0.2045

687 ∑𝑛𝑖=0 𝛾𝑖+ = ∑𝑛𝑖=0 𝛾𝑖− 12.4591 0.0005 0.9018 0.3436 4.8975 0.0281 2.0945 0.1496 3.1117 0.0794

688 (1, 181) (1, 181) (1, 181) (1, 181) (1, 181)

689 (***), (**) and (*) indicate statistical significant at the 1%, 5% and 10% level respectively.

33
690 Table 6. Coefficients Cholesky decomposition imposing short-run restrictions

691 Coefficients Matrix A-1 Coefficients Matrix B

692 Retail Wholesale Retail Wholesale

693 Davao Retail 1.0000 0.0000 0.7467*** 0.0000

694 Wholesale 0.9989*** 1.0000 0.0000 0.4469***

695 South Retail 1.0000 0.0000 0.8017*** 0.0000

696 Cotabato Wholesale 1.0835*** 1.0000 0.0000 0.5179***

697 (***), (**) and (*) indicate statistical significant at the 1%, 5% and 10% level respectively.

698

699

700

701

702

703

704

705

706

707

708

709

710

711

712

713

714
34
715 Response to Cholesky One S.D. Innovations
Re s p on s e of MMANIL ARP to MMAN ILAWP_N EG R e s p o n s e o f MMAN IL AR P to MMAN IL AWP_ POS
.6 .6

.4 .4

.2 .2

.0 .0

-.2 -.2
716 5 10 15 20 25 30 35 40 45 5 10 15 20 25 30 35 40 45

Res pons e of CEBURP to CEBUWP_NEG Res pons e of CEBURP to CEBUWP_POS

.8 .8

.6 .6

.4 .4

.2 .2

.0 .0

-.2 -.2
717 5 10 15 20 25 30 35 40 45 5 10 15 20 25 30 35 40 45

Res p ons e of DAVAORP to DAVAOWP_NEG Re s p on s e of DAVAORP to DAVAOWP_ POS

.8 .8

.6 .6

.4 .4

.2 .2

.0 .0

-.2 -.2

-.4 -.4
718 5 10 15 20 25 30 35 40 45 5 10 15 20 25 30 35 40 45

Res pons e of ILOILORP to ILOILOWP_NEG Res pons e of ILOILORP to ILOILOWP_POS

.8 .8

.6 .6

.4 .4

.2 .2

.0 .0

-.2 -.2
719 5 10 15 20 25 30 35 40 45 5 10 15 20 25 30 35 40 45

Res pons e of SCOTA BA TORP to SCOTA BA TOWP_NEG Res pons e of SCOTA BA TORP to SCOTA BA TOWP_POS
1.2 1.2

0.8 0.8

0.4 0.4

0.0 0.0

-0.4 -0.4
720 5 10 15 20 25 30 35 40 45 5 10 15 20 25 30 35 40 45

721 Figure 3 Responses of Retail Price to Positive and Negative Shocks in Wholesale Price by One Standard
722 Deviation; measured ‘month’ in X-axis and ‘price (PHP/kg)’ in Y-axis.
723 Source: FAO – FPMA

35