Market integration of wheat in Pakistan

Sahito, Jam Ghulam Murtaza Working Paper Market integration of wheat in Pakistan Discussion Paper, No. 72 Provided in Cooperation with: Justus Liebig University Giessen, Center for international Development and Environmental Research (ZEU) Suggested Citation: Sahito, Jam Ghulam Murtaza (2015) : Market integration of wheat in Pakistan, Discussion Paper, No. 72, Justus-Liebig-Universität Gießen, Zentrum für Internationale Entwicklungsund Umweltforschung (ZEU), Giessen This Version is available at: https://hdl.handle.net/10419/119870 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by-nc-nd/3.0/de/ Zentrum für internationale Entwicklungsund Umweltforschung der Justus-Liebig-Universität Gießen Market Integration of Wheat in Pakistan von Jam Ghulam Murtaza Sahitoa,b Nr. 72 Gießen, June 2015 a Center for International Development and Environmental Research (ZEU), Senckenberg Strasse 3, 35390, Giessen, Justus Liebig University Giessen, Germany b Department of Agricultural Economics, Sindh Agriculture University Tando Jam, Sindh, Pakistan. Email: [email protected]du.pk ; [email protected]iessen.de ; [email protected] Dieses Werk ist im Internet unter folgender Creative Commons Lizenz publiziert: http://creativecommons.org/licenses/by-nc-nd/3.0/de/ Sie dürfen das Werk vervielfältigen, verbreiten und öffentlich zugänglich machen, wenn das Dokument unverändert bleibt und Sie den Namen des Autors sowie den Titel nennen. Das Werk darf nicht für kommerzielle Zwecke verwendet werden. Acknowledgment: I am thankful to Higher Education Commission (HEC) Pakistan and German Academic Exchange Service (DAAD) Germany, for the financial support to pursue my doctorate in Justus Liebig University Germany. This is part of the Ph.D research project. Special thanks are due to my supervisor Prof. Dr. Peter Winker and co-supervisor Prof. Dr. Roland Herrmann for their valuable comments and suggestions on the earlier draft of this paper. Market Integration of Wheat in Pakistan ABSTRACT: Understanding market integration in developing countries is an important issue in current research. This study is an attempt to analyze wheat market integration in Pakistan. Previous research on the subject has attempted at analyzing market integration in Pakistan’s south and north Punjab regions, mainly relying on co-integration only and not considering advanced dynamic models and transaction costs to analyze the degree of integration. Therefore, this study is a first attempt to analyze the extent of market integration in the whole country using a dynamic model. Monthly wholesale price data of five regional markets from January 1988 to April 2011 are used for this study. Price series were tested for stationarity with the Augmented Dickey Fuller (ADF) test and it was found that all prices are integrated of order one, commonly written as I(1). Cointegration was also identified in all price series pairs using Johansen’s co-integration test. The Vector Error Correction Model (VECM) was then applied to the data to analyze the extent of market integration. As a result, it was found that the adjustment to shocks or disequilibrium was higher for the Lahore and Rawalpindi markets as compared to the Hyderabad and Peshawar markets. It might be because of the high consumption, low production and developed infrastructure in these regions. Adjustment coefficients were significant for most of the market pairs. The Threshold Vector Error Correction Model (TVECM) with a band of non-adjustment was applied to incorporate transaction costs, without relying on observations for these costs, which were not available for the study. It was found that linear ECMs or VECMs provide misleading results as compared to TVECMs. Short-run adjustments in the TVECM model provide mixed results depending on regimes as well as markets. Strong adjustments were found in the upper regime, which shows that when price differences are above the second threshold markets tend to adjust significantly. Keywords: Market integration, co-integration, wheat, commodity prices, error correction, thresholds. JEL-Classification: C32; F15 1 1 Introduction Market integration describes the degree of price transmission within vertically or geographically separated markets. Spatial or vertical market integration of homogenous commodities especially in developing countries has been the center of interest for economists in the last few decades. Special attention has typically been given to basic food crops such as wheat and rice, because food insecurity is a major issue for developing countries. Market integration studies in agriculture, especially for developing countries are the tools to examine, evaluate, regulate and reform price polices for food security and price stability. In the context of Pakistan, a developing country, wheat is the major food crop, providing the largest source of calorie intake, thus it is important from food security perspective. The World Trade Organization (WTO) considers Pakistan as the most food insecure among net wheat importing developing countries (GoP 2011-2012). Pakistan has not yet achieved self-sufficiency, especially in wheat production, and has remained largely a net importer of wheat. In fact, Pakistan has only exported a small amount of wheat as a result of bumper crop between the years from 2000 to 2006. Overall, the production of wheat has been volatile in Pakistan during the last two decades (GoP 2011-2012). Price transmission among domestic markets will enable us to understand the vulnerability of the population to food market shocks in Pakistan. Market integration studies provide valuable information about the efficiency of market functioning and about the dynamics of price adjustment in the markets. Information of spatial market integration infer the efficiency of pricing, effectiveness of arbitrage and competitiveness of markets, which implies the efficient market functioning (Sexton et al. 1991). There are many hindrances to the efficient functioning of the agricultural commodities market in Pakistan. Some major issues include insufficient transportation infrastructure, restrictions on the movement of wheat within provinces and districts, no or sparse access to market information, market structure and changes in the costs of production (Tahir and Riaz 1997). For example, intraprovince movement restriction of wheat in the months of harvesting and support price policy of 2 wheat are direct interventions of the government. Transportation infrastructure, information and communication are other factors affecting market integration. The government of Pakistan has been involved in interventions within the wheat sector via support prices, procurement, storage, transportation and distribution of wheat to flour millers since independence. Two major objectives of this intervention are, first, to protect consumers from higher import prices, and second, to protect producers via procurement and support prices in an effort to reduce price volatility (Ahmed et al. 2006). The government of Pakistan procures about 25 to 30 percent of total wheat production every year (GoP 2011-2012). These government interventions are considered as the fiscal burden on the economy in case of higher degree of market integration (Mushtaque et al. 2007 and Dorosh and Salam 2008). Higher degree of market integration and quicker adjustment of prices to form a new equilibrium as a result of shocks to the market prices also explains the efficient functioning of markets. Hence, it is worthwhile to assess the degree of market integration of wheat markets in Pakistan. There are only few market integration studies regarding the food markets of Pakistan. Unfortunately, most of them have focused only on one or two regions of Punjab province and relied on co-integration coefficients or error correction mechanism only. There has been a lot of development in the last two decades regarding the methods to investigate market efficiency and integration, which has not been applied to food markets of Pakistan. Many models and methods have been developed to analyze integration of markets. Every method has its own strengths and weaknesses. However, due to intuitive interpretation, error correction models have gained the attention of the majority of studies. Most of these studies rely only on time series data of prices and do not take into account transaction costs or trade flows. A brief review about these studies is provided in section two. Although, Barrett (1996) and Barrett and Li (2002) are of the opinion that one cannot describe spatial market relationships only by prices but by their combination with transaction costs. However, transaction costs are neither easily available nor can any other proxy be used to incorporate these costs. Threshold models estimate a neutral band linked with unobservable 9 different applications of market integration (Greb et al. 2012; Hassouneh et al. 2012; Meyer and von Cramon-Taubadel 2004). Parameters of price transmission between two spatially separated markets having variable transportation costs cannot be fixed over time. In this case, the first type of linearity is a very hard restriction. Barrett and Li (2002) describe the difficulties in observing all possible transaction costs, like: trade flows, risk assessment, discount rates and other possible costs. They also implied the possibility of trade and adjustment of short-run prices due to arbitrage, if the difference between two market prices is higher than the transaction cost, because of the unobservable costs, policy interventions and different strategies. Hence, if the price difference is less than a certain threshold, there is no arbitrage benefit for traders. Balke and Fomby (1997) introduced the concept of threshold co-integration, based on discontinuous long-run equilibrium adjustments. This concept allows addressing the abovementioned criticism on linear co-integration and justifies the use of threshold models for price adjustment. In particular, this model allows for a no-arbitrage band. Adjustments only occur, when the deviations in the long-run equilibrium are greater than transaction costs or a particular threshold, where the error-correction term determines the threshold parameter. As the TVECM is a special form of asymmetric VECMs, price adjustment can be different depending on the regimes. This model is extendable, by incorporating constants or intercepts and lags in each regime. Regimeswitching models have attracted several researchers of price transmission analysis, and have been extended and applied by many researchers such as, Lo and Zivot 2001; Goodwin and Piggott 2001; Hansen and Seo 2002; Meyer 2004; and Seo 2006. A bivariate TVECM Model with two thresholds (three regimes) can be defined as: �∆𝑃𝑃1𝑡𝑡 ∆𝑃𝑃2𝑡𝑡�=�𝛼𝛼1 𝛼𝛼2�+∑�𝛽𝛽𝑖𝑖𝑃𝑃1,𝑃𝑃1𝛽𝛽𝑖𝑖𝑃𝑃1,𝑃𝑃2 𝛽𝛽𝑖𝑖𝑃𝑃2,𝑃𝑃1𝛽𝛽𝑖𝑖𝑃𝑃2,𝑃𝑃2� 𝑘𝑘 𝑖𝑖=1 ×�∆𝑃𝑃1𝑡𝑡−1 ∆𝑃𝑃2𝑡𝑡−1�+�𝜑𝜑1 𝜑𝜑2�[𝐸𝐸𝐸𝐸𝐸𝐸𝑡𝑡−1]+�𝜀𝜀1𝑡𝑡 𝜀𝜀2𝑡𝑡�,𝑖𝑖𝑖𝑖 𝐸𝐸𝐸𝐸𝐸𝐸𝑡𝑡−1 ≤𝛾𝛾1 �∆𝑃𝑃1𝑡𝑡 ∆𝑃𝑃2𝑡𝑡�=�𝛼𝛼1 𝛼𝛼2�+∑�𝛽𝛽𝑖𝑖𝑃𝑃1,𝑃𝑃1𝛽𝛽𝑖𝑖𝑃𝑃1,𝑃𝑃2 𝛽𝛽𝑖𝑖𝑃𝑃2,𝑃𝑃1𝛽𝛽𝑖𝑖𝑃𝑃2,𝑃𝑃2� 𝑘𝑘 𝑖𝑖=1 ×�∆𝑃𝑃1𝑡𝑡−1 ∆𝑃𝑃2𝑡𝑡−1�+�𝜑𝜑1 𝜑𝜑2�[𝐸𝐸𝐸𝐸𝐸𝐸𝑡𝑡−1]+�𝜀𝜀1𝑡𝑡 𝜀𝜀2𝑡𝑡�,𝑖𝑖𝑖𝑖 𝛾𝛾1≤𝐸𝐸𝐸𝐸𝐸𝐸𝑡𝑡−1 ≤ 𝛾𝛾2 �∆𝑃𝑃1𝑡𝑡 ∆𝑃𝑃2𝑡𝑡�=�𝛼𝛼1 𝛼𝛼2�+∑�𝛽𝛽𝑖𝑖𝑃𝑃1,𝑃𝑃1𝛽𝛽𝑖𝑖𝑃𝑃1,𝑃𝑃2 𝛽𝛽𝑖𝑖𝑃𝑃2,𝑃𝑃1𝛽𝛽𝑖𝑖𝑃𝑃2,𝑃𝑃2� 𝑘𝑘 𝑖𝑖=1 ×�∆𝑃𝑃1𝑡𝑡−1 ∆𝑃𝑃2𝑡𝑡−1�+�𝜑𝜑1 𝜑𝜑2�[𝐸𝐸𝐸𝐸𝐸𝐸𝑡𝑡−1]+�𝜀𝜀1𝑡𝑡 𝜀𝜀2𝑡𝑡�,𝑖𝑖𝑖𝑖 𝐸𝐸𝐸𝐸𝐸𝐸𝑡𝑡−1 ≥𝛾𝛾2 (2) 10 Here, 𝛾𝛾1 and 𝛾𝛾2 are the threshold parameters. 𝑃𝑃1 and 𝑃𝑃2 represent the prices in two markets respectively. The autoregressive parameters differ, based on regimes, whether the variables are below, between two regimes or above the higher threshold. These models have three regimes, namely, lower, middle and higher. Each regime should contain at least 5 to 15 percent of all observations for the empirical application following Goodwin and Piggott (2001), Hansen and Seo (2002) and Meyer (2004). Estimation of this model takes place with a two-dimensional grid search over the thresholds and co-integrating values based on maximum likelihood estimator using “tsDyn” package in R developed by Stigler (2010). To test for threshold effects, the SupLM (Supremum Lagrange Multiplier) test developed by Hansen and Seo (2002) has been used, setting the null hypothesis of linear co-integration against the alternative hypothesis of threshold co-integration. This test uses the co-integration coefficient parameter from the linear VECM representation and applies a grid search over the threshold parameter. Critical values and the p-values are generated by a fixed regressor bootstrap method. The advantage of this method is that LM-like statistics allow for heteroskedasticity of unknown form in the same way as White’s consistent heteroskedastic standard errors, hence it achieves the correct first-order asymptotic distribution. The Sup LM test statistic can be denoted as: 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆=𝑠𝑠𝑆𝑆𝑆𝑆 𝛾𝛾𝛾𝛾 ≤𝛾𝛾≤𝛾𝛾𝛾𝛾𝑆𝑆𝑆𝑆(𝛽𝛽, �𝛾𝛾) (3) Where 𝛽𝛽, � co-integration value is 𝛽𝛽 estimated and 𝛾𝛾 is the threshold parameter. 𝛾𝛾𝑆𝑆 is the trimming parameter (𝜋𝜋0) of the constraint set for the number of observations below the threshold parameter and 𝛾𝛾𝛾𝛾 is (1−𝜋𝜋0) number of observations above the threshold. The restriction for the number of observations in the regimes (trimming parameter) must satisfy the following expression. 𝜋𝜋0≤𝑃𝑃(𝐸𝐸𝐸𝐸𝐸𝐸𝑡𝑡−1 ≤𝛾𝛾)≤1−𝜋𝜋0 (4) In this analysis, 𝜋𝜋0 is equal to 0.10, as Andrew (1993) recommends that the value of 𝜋𝜋0 should range from 0.05 to 0.15. Further, 5000 bootstrap replications are used in the analysis to calculate asymptotic critical values and the p-values for the test. 11 4 Results This section reveals the estimated results of co-integration, VECM and TVECM models. Before presenting the results, it is important to show the contribution of provinces in the wheat production of the country. This is meant to provide an idea regarding the trade flow of wheat within the different provinces of Pakistan. Table (1) presents area and production of wheat crop in Pakistan (province wise). The statistics depicts that in the cropping year 1987-88, 7308.4 thousand hectares were sown, producing 12675 thousand tonnes of wheat. Both the area under wheat crop as well as production increased in the last twenty-five years, however, area increased only by over one thousand hectares, while production almost doubled until the year 2011-12 as compared to 198788. Table 1: Area and Production of Wheat Crop in Pakistan and Provinces in the Years 1987-88 and 2011-12. Area in 1000 Hectares and Production in 1000 Tonnes. Particulars 1987-88 2011-12 Area Production Area Production Pakistan 7308.40 12674.40 8649.80 23473.40 Punjab 5343.80 (73 %) 9203.80 (73 %) 6482.90 (75 % ) 17738.90 (75 %) Sindh 1024.80 (14 %) 2180.40 (17 %) 1049.20 (12 %) 3761.50 (16 %) KPK 756.50 (10 %) 899.20 (7 %) 729.30 (8 %) 1130.30 (5 %) Baluchistan 183.30 (3 %) 391.00 (3 %) 388.40 (5 %) 842.70 (4 %) Source: Agricultural Statistics of Pakistan, 1987-88 and 2011-12. Contributions of the provinces in area and production show that Punjab is and was the single largest contributor in terms of production of wheat as well as in the area sown under wheat. Punjab alone contributed 73 percent of area and production in the year 1987-88. This share increased to 75 percent until the year 2011-12. Area and production of Punjab in the year 2011-12 was 6482.90 and 17738.90 respectively. Area sown in Sindh, KPK and Baluchistan in the year 2011-12 was 1049.20, 729.30 and 388.40, respectively, and production in the same year was 3761.50, 1130.30 and 842.70, respectively. The percentage share of Sindh and KPK in area sown under wheat crop, as well as, production has decreased over the last two decades. 12 Wheat provides the single largest source of calories in Pakistan, more than 35 percent of the total energy requirement in the country. However, Pakistan has remained largely a net importer of wheat during most of the last twenty-five years, with a small exporting period between the years 2000 to 2006 (GOP 2010-11). Unfortunately, historical data of wheat consumption are not available as readily as data on production. Province wise consumption requirements data are especially difficult to find. As this paper focuses on regional market price series of wheat from different provinces, it is therefore necessary to have an idea of the demand in different provinces. Due to the aforementioned data availability constraint, shortfall of wheat for the year 2008 is presented here to give an idea of wheat deficient provinces (Figure 2). As it turns out, Punjab is the only province of Pakistan having a surplus in wheat production, producing about 16-17 million MT of wheat every year with a consumption requirement of 12.5 million MT in the province. Figure 2: Wheat Production and Shortfall Province-wise for the Year 2007-08. Source: UN inter-agency assessment report 2008. Furthermore, Sindh, Balochistan and Khyber Pakhtunkhwa (KPK) provinces are deficient in wheat production hence trade takes place more from Punjab to these provinces. In most cases, government transports wheat from the stock of wheat procured during the harvest season or finances the private sector to transport to the wheat-deficit areas of the country, to offset the costs of transportation. Sindh has a wheat production shortfall mainly because its capital Karachi, which comprises of dense urban population, is also the main port where imports arrive. The urban population of Karachi 13 are the primary wheat import consumers. KPK is the largest wheat deficit province requiring the allocation of more than two million MT annually. These provinces buy wheat either from PASSCO or from the Punjab food department. KPK shares the porous border with Afghanistan and a large share of wheat is sent to Afghanistan as informal trade rather than reaching local consumers. (UN inter-agency assessment report 2008) 4.1 Unit Root Test Results of Wheat Prices Results of the Augmented Dickey Fuller (ADF) test for logged price series of five regional markets of wheat in Pakistan at levels and at first differences are presented in Table (2). These results indicate that the null hypothesis of a unit root in all the five markets cannot be rejected for the levels, because the ADF statistics were not smaller than the critical value at the 5 percent significant level provided by Dickey and Fuller (1981). To check the stationarity in the price series at first differences, the ADF test was re-applied to the differenced price series. The ADF statistics indicate the rejection of the null hypothesis of a unit root significantly, implying that all of the price series are stationary at first differences. Table 2: Unit root test results of logged monthly wholesale prices of wheat markets: Markets Levels 1st Difference Hyderabad -0.176 -14.413*** Lahore 0.110 -12.821*** Multan -0.218 -12.415*** Peshawar -0.470 -13.936*** Rawalpindi -0.096 -13.706*** Critical values: 1% level 5% and 10% respectively are -3.454, -2.872, -2.573 Source: Author’s own calculations Since the results indicate that the price series of the wheat markets under study are first-difference stationary, one can infer that all five series are integrated of order one, i.e I(1). Thus, co-integration tests can be applied to see whether there are long run relationship between the markets. 4.2 Co-integration test results Pair-wise co-integration test results for selected wheat markets are presented in Table 3. Results clearly indicate the existence of a long-run equilibrium relationship between all the pairs of regional wheat markets. Both trace statistics and maximum eigenvalue statistics suggest a co-integration relation in all the ten pairs of five markets. It can be concluded that there is a strong long-run relationship between wheat markets of Pakistan. 14 Table 3: Pair-Wise Cointegration Test Results Logged Wheat Market prices: Market Pairs Null Hypothesis Alternate Hypothesis Trace Statistics Maximum Eigenvalue Statistics LogLahoreLogHyd r=0 r≤1 r≥1 r≥2 41.284 (15.494)*** 0.033 (3.841) 41.251 (14.264)*** 0.033 (3.841) LogLahoreLogMultan r=0 r≤1 r≥1 r≥2 32.892 (15.494)*** 0.000 (3.841) 32.891 (14.264)*** 0.000 (3.841) LogLahoreLogPindi r=0 r≤1 r≥1 r≥2 54.744 (15.494)*** 0.000 (3.841) 54.744 (14.264)*** 0.000 (3.841) LogLahoreLogPeshawer r=0 r≤1 r≥1 r≥2 16.853 (15.494)*** 0.014 (3.841) 16.838 (14.264)*** 0.014 (3.841) LogHydLogMultan r=0 r≤1 r≥1 r≥2 40.627 (15.494)*** 0.029 (3.841) 40.598 (14.264)*** 0.029 (3.841) LogHydLogPindi r=0 r≤1 r≥1 r≥2 38.019 (15.49)*** 0.026 (3.841) 37.992 (14.264)*** 0.026 (3.841) LogHydLogPeshawer r=0 r≤1 r≥1 r≥2 22.452 (15.494)*** 0.068 (3.841) 22.383 (14.264)*** 0.068 (3.841) LogMultanLogPeshawer r=0 r≤1 r≥1 r≥2 15.731 (15.494)** 0.020 (3.841) 15.710 (14.264)** 0.020 (3.841) LogMultanLogPindi r=0 r≤1 r≥1 r≥2 43.079 (15.494)*** 0.003 (3.841) 43.076 (14.264)*** 0.003 (3.841) LogPindiLogPeshawer r=0 r≤1 r≥1 r≥2 15.984 (15.494)** 0.016 (3.841) 15.967 (14.264)** 0.016 (3.841) Critical values at 95% confidence interval are in parenthesis. Source: Author’s calculations Table 4 presents the results of joint co-integration tests for all five wheat markets of Pakistan. The trace statistics as well as the maximum eigenvalue statistics suggest that all the five markets are co-integrated and converge to the long-run equilibrium. Table 4: Joint Cointegration Test Results Logged Wheat Market prices: Equation Tested Null Hypothesis Alternate Hypothesis Trace Statistics Maximum Eigenvalue Statistics LogHyd LogLahoreLogMultan LogPindi LogPeshawer r=0 r≤1 r≤2 r≤3 r≤4 r≥1 r≥2 r≥3 r≥4 r≥5 171.08 (69.818)*** 111.80 (47.856)*** 60.962 (29.797)*** 15.973 (15.494)** 0.073 (3.841) 59.278 (33.876)*** 50.844 (27.584)*** 44.989 (21.131)*** 15.900 (14.264)** 0.073 (3.841) Critical values at 95% confidence interval are in parenthesis. Source: Author’s calculations Test results reveal that there are four co-integrating relationships in the joint co-integration analysis of all five wheat markets. As, Greene (2003) proves that there can be at most K-1 co-integration 15 vectors in the joint co-integration test. Where, “K” indicates the number of variables in the system. This implies that there are four linear independent combinations of the variables; each combination is stationary. It also shows that there is at least one common stochastic trend. 4.3 Linear VECM results The error Correction Model (ECM) was applied to estimate a long-term coefficient along with short-term dynamics. A linear VECM model results are presented in table 5. Results show a highly significant adjustment of prices in almost all the pairs of markets except the Hyderabad market. Adjustment to equilibrium from the Hyderabad market is slower as well as insignificant in some cases because this market is far away from the other four markets but still well connected to Lahore, Multan and Rawalpindi by means of transport and communication. Hence, there is no surprise in the quicker response of Lahore, Multan and Rawalpindi to Hyderabad. Table 5: VECM Results of Wheat Markets of Pakistan Logged Wheat Market Pairs Speed of Adjustment Logged Wheat Market Pairs Speed of Adjustment LogLahoreLogHyd -0.183 *** 0.050 LogHydLogPindi -0.085* 0.156*** LogLahoreLogMultan -0.249*** 0.063 LogHydLogPeshawer -0.033 0.095*** LogLahoreLogPindi -0.171*** 0.177** LogMultanLogPeshawer -0.037** 0.068*** LogLahoreLogPeshawer -0.041* 0.060** LogMultanLogPindi -0.047 0.246*** LogHydLogMultan -0.090** 0.151*** LogPindiLogPeshawer -0.042* 0.057** Note: *, ** and *** show the significance at 90%, 95% and 99%. Source: Author’s calculations Due to the favorable infrastructure in Lahore and Rawalpindi and higher demand because of dense urban population in these areas, wheat trade to these markets from other parts of the country pushes them to adjust to the equilibrium quickly. Lahore, Multan and Rawalpindi are also well connected as well as close to each other as compared to the other markets under study. Multan is a bigger region in terms of production of wheat. Therefore, both the Lahore and Rawalpindi markets adjust quickly to Multan. Lahore is also one of the major markets in which multidirectional trade takes place. Apart from that, these results are from a linear VECM model without considering transaction 16 costs. However, these results may differ when incorporating transaction costs into the threshold model. 4.4 Testing for thresholds The SupLM test for threshold co-integration clearly rejects the null hypothesis of linear cointegration against the alternate hypothesis of threshold co-integration at the 5% significance level. This holds true for seven out of ten pairs of different wheat market price series of Pakistan. While, for three pairs of price series namely Multan-Peshawar, Hyderabad-Peshawar and LahorePeshawar, the null hypothesis is rejected at the 10 % significance level. The SupLM test results provide enough conclusive evidence of threshold co-integration to justify an application of the TVECM to the data. Estimates of SupLM test with 1 lag and 5000 bootstrap replications on price series of wheat markets of Pakistan are provided in Table 6. Table 6: SupLM Test Results for Wheat Markets of Pakistan Market Pairs Cointegration Vector β Threshold Parameter γ SupLM Test Value Critical Value (P-Value) LogLahoreLogHyd -1.006 -0.066 20.161 18.828 0.026 LogLahoreLogMultan -0.982 0.139 20.414 19.334 0.031 LogLahoreLogPindi -0.962 0.230 34.650 16.117 0.000 LogLahoreLogPeshawer -0.997 0.018 16.865 17.464 0.063 LogHydLogMultan -0.976 0.153 19.461 15.714 0.008 LogHydLogPindi -0.954 0.179 26.348 15.554 0.000 LogHydLogPeshawer -0.982 0.074 17.575 18.388 0.080 LogMultanLogPeshawer -0.978 0.077 27.437 18.868 0.000 LogMultanLogPindi -1.011 -0.112 17.415 18.558 0.084 LogPindiLogPeshawer -1.036 -0.272 18.503 15.252 0.012 Source: Author’s calculations 17 4.5 Threshold Vector Error Correction Model Table 7 presents the estimation results for the TVECM model with two thresholds (three regimes). The band between the two thresholds (regime 2 or middle regime) is the band of non-adjustment because deviations from the long-term equilibrium as compared to adjustment costs are so small that they will not cause an adjustment process of related prices within the band. As expected, the threshold error-correction model produced different results from the previous simple model. Cointegration clearly describes the long-run relationship among different wheat markets of Pakistan, and it can be seen from the threshold model that short-run adjustment to disequilibrium is somehow mixed. The results reveal that some market pairs show higher adjustment in both regimes, while others only indicate significant adjustment either in the upper or in the lower regime. Meyer (2004) referred to price adjustment due to disequilibrium in one direction or in one regime to be insignificant, considering the unidirectional trade flows or significant transaction costs. The adjustment parameters are higher and significant in most cases as compared to the results of the linear VECM, which shows that the threshold model describes the short-run adjustment in the prices as quicker and higher in magnitude. Lahore (LHR) and Hyderabad (HYD) markets adjust quickly, when the shock is higher than the second threshold, which implies that prices adjust quickly when they are higher and adjustment is slow when the price difference is below the lower threshold. Lahore being the major production and consumption region in Punjab province of Pakistan forces other markets of Punjab, namely Rawalpindi (PINDI) and Multan (MLTN), to adjust quickly. These two markets are close to Lahore in terms of distance and are well connected through favorable infrastructure supporting transportation. This holds equally true in terms of information and communication. The linear VECM estimated a higher extent of adjustment for the Lahore market, which was somewhat surprising as Lahore market is considered the leader rather than the follower. In most cases, higher and significant adjustments revealed by the estimation occur in the upper regime. When these deviations are above the second threshold and provide the opportunity for traders to take advantage of the arbitrage, then as expected, prices adjust quickly to form a new equilibrium. 18 Table 7: TVECM Results of Wheat Markets of Pakistan. Market Pairs Regimes Speed of Adjustment Constant P 1t-1 P 2t-1 LLHR LHYD Lower Regime -0.098 (0.193) 0.124 (0.087)* 0.010 (0.056)* 0.012 (0.021)** 0.081 (0.318) -0.043 (0.579) -0.039 (0.665) 0.270 (0.001)*** Upper regime -0.686 (0.001)*** -0.343 (0.099)* 0.036 (0.017)** 0.036 (0.013)** 0.029 (0.856) -0.101 (0.513) 0.011 (0.967) 0.349 (0.171) LLHR LMLTN Lower Regime -0.324 (0.227) 0.651 (0.007)*** 0.004 (0.752) 0.024 (0.022)** -0.319 (0.100) -0.355 (0.041)** 0.221 (0.247) 0.346 (0.044)** Upper regime 0.188 (0.321) 0.369 (0.030)** -0.023 (0.163) -0.029 (0.047)** 0.044 (0.735) 0.262 (0.024)** 0.237 (0.121) 0.195 (0.154) LLHR LPINDI Lower Regime 0.022 (0.801) 0.402 (2.9e5)*** 0.014 (0.004)*** 0.023 (3.2e-5)*** 0.032 (0.750) -0.120 (0.268) 0.130 (0.170) 0.402 (0.000)*** Upper regime -0.089 (0.589) 0.156 (0.376) 0.004 (0.705) 0.004 (0.714) 0.222 (0.157) -0.152 (0.366) 0.159 (0.316) 0.283 (0.097)* LLHR LPSHWR Lower Regime -0.029 (0.567) 0.094 (0.126) 0.008 (0.230) 0.007 (0.376) 0.270 (0.011)** 0.110 (0.395) 0.047 (0.522) 0.237 (0.008)*** Upper regime -0.292 (3.7e-5)*** -0.070 (0.404) 0.039 (3.4e-6)*** 0.018 (0.063)* 0.343 (0.000)*** 0.200 (0.086)* 0.009 (0.938) 0.252 (0.072)* LHYD LMLTN Lower Regime -0.190 (0.490) 1.115 (1.0e-5)*** -0.004 (0.863) 0.097 (8.2e6)*** -0.202 (0.375) 0.069 (0.738) 0.333 (0.026)** 0.248 (0.065)* Upper regime -0.361 (0.042)** 0.004 (0.980) 0.032 (0.055)* 0.016 (0.286) 0.283 (0.006)*** 0.170 (0.067)* -0.113 (0.374) 0.039 (0.732) LHYD LPINDI Lower Regime -0.492 (0.001)*** -0.263 (0.146) -0.015 (0.090)* -0.022 (0.033)** 0.284 (0.064)* 0.151 (0.399) 0.023 (0.838) 0.094 (0.474) Upper regime -0.118 (0.298) 0.237 (0.075)* 0.017 (0.212) -0.013 (0.423) -0.021 (0.870) -0.010 (0.949) -0.029 (0.778) 0.044 (0.713) LHYD LPSHWR Lower Regime -0.016 (0.772) 0.250 (0.000)*** 0.007 (0.145) 0.011 (0.046)** 0.181 (0.118) -0.366 (0.008)*** 0.036 (0.630) 0.569 (0.000)*** Upper regime -0.310 (7.4e-5)*** -0.125 (0.173) 0.045 (4.6e-6)*** 0.029 (0.014)** 0.120 (0.183) 0.189(0.077)* -0.043 (0.683) 0.135 (0.287) LMLTN LPINDI Lower Regime -0.283 (0.175) 0.801 (0.000)*** -0.004 (0.768) 0.057 (0.001)** 0.338 (0.032)** 0.292 (0.104) 0.121 (0.382) 0.240 (0.130) Upper regime 0.043 (0.699) 0.369 (0.004)*** 0.006 (0.452) -0.003 (0.779) 0.041 (0.784) -0.178 (0.290) 0.081 (0.554) 0.123 (0.428) LMLTN LPSHWR Lower Regime -0.099 (0.003)*** 0.031 (0.490) -0.002 (0.627) 0.002 (0.725) 0.366 (2.4e-5)*** -0.013 (0.910) -0.078 (0.164) 0.192 (0.009)** Upper regime -0.235 (0.012)** -0.213 (0.082)* 0.030 (0.009)*** 0.045 (0.003)*** 0.366 (0.000)*** 0.222 (0.121) -0.105 (0. 343) -0.049 (0.734) LPINDI LPSHWR Lower Regime -0.126 (0.012)* 0.034 (0.523) -0.011 (0.164) -0.001 (0.992) 0.249 (0.007)*** -0.079 (0.414) 0.020 (0.819) 0.351 (0.000)*** Upper regime -0.207 (0.009)*** -0.041 (0.630) 0.018 (0.002)*** 0.015 (0.017)** 0.142 (0.201) 0.054 (0.644) 0.147 (0.230) 0.104 (0.421) Source: Author’s calculations ii No. 7 RUBIOLO, M. (01.2002) EU and Latin America Biregionalism in a Globalizing World? No. 8 GAST, M. (02.2002) Zollkontingente bei US-amerikanischen Käseimporten. (gekürzte Fassung erschienen in "Agrarwirtschaft", Jg. 51, H. 4, S. 192-202.) No. 9 BISCHOFF, I. (08.2002) Efficiency-enhancing Effects of Private and Collective Enterprises in Transitional China. No. 10 KÖTSCHAU, K. M., PAWLOWSKI, I., SCHMITZ, P. M. (01.2003) Die Policy Analysis Matrix (PAM) als Instrument zur Messung von Wettbewerbsfähigkeit und Politikeinfluss - Zwischen Theorie und Praxis Das Fallbeispiel einer ukrainischen Molkerei. No. 11 HERRMANN, R., MÖSER A. (06.2003) Price Variability or Rigidity in the Foodretailing Sector? Theoretical Analysis and Evidence from German Scanner Data. No. 12 TROUCHINE, A. (07.2003) Trinkwasserversorgung und Armut in Kasachstan Aktueller Zustand und Wechselwirkungen. No. 13 WANG, R.; GIESE, E.; GAO, Q. (08.2003) Seespiegelschwankungen des Bosten-Sees (VR China). No. 14 BECKER, S.; GEMMER, M.; JIANG, T.; KE, CH.. (08.2003) 20th Century Precipitation Trends in the Yangtze River Catchment. No. 15 GEMMER, M.; BECKER, S.; JIANG, T (11. 2003) Detection and Visualisation of Climate Trends in China. No. 16 MÖNNICH, Ch. (12.2003) Tariff Rate Quotas Does Administration Matter? No. 17 GIESE, E.; MOßIG. I. (03.2004) Klimawandel in Zentralasien No. 18 GIESE, E.; SEHRING, J. TROUCHINE, A. (05.2004) Zwischenstaatliche Wassernutzungskonflikte in Zentralasien iii No. 19 DIKICH, A. N. (09.2004) Gletscherwasserressourcen der Issyk-Kul-Region (Kirgistan), ihr gegenwärtiger und zukünftiger Zustand No. 20 CHRISTIANSEN, Th.; SCHÖNER, U. (11.2004) Irrigation Areas and Irrigation Water Consumption in the Upper Ili Catchment, NW-China No. 21 NARIMANIDZE, E. et al. (04.2005) Bergbaubedingte Schwermetallbelastungen von Böden und Nutzpflanzen in einem Bewässerungsgebiet südlich von Tiflis/Georgien - Ausmaß, ökologische Bedeutung, Sanierungsstrategien No. 22 ROMANOVSKIJ, V.V.; KUZ’MIČENOK, V.A. (06.2005) Ursachen und Auswirkungen der Seespiegelschwankungen des Issyk-Kul’ in jüngerer Zeit No. 23 ZITZMANN, K.; TROUCHINE, A. (07.2005) Die Landwirtschaft Zentralasiens im Transformationsprozess (nicht mehr lieferbar!) No. 24 SEHRING, J. (08.2005) Water User Associations (WUAs) in Kyrgyzstan - A Case Study on Institutional Reform in Local Irrigation Management No. 25 GIESE, E., MAMATKANOV, D. M. und WANG, R. (08.2005) Wasserressourcen und Wassernutzung im Flussbecken des Tarim (Autonome Region Xinjiang / VR China) No. 26 MOSSIG, I., RYBSKY, D. (08.2005) Die Erwärmung bodennaher Luftschichten in Zentralasien. Zur Problematik der Bestimmung von Trends und Langzeitkorrelationen No. 27 GAST, M. (09.2005) Determinants of Foreign Direct Investment of OECD Countries 1991-2001 No. 28 GIESE, E., TROUCHINE, A. (01.2006) Aktuelle Probleme der Energiewirtschaft und Energiepolitik in Zentralasien No. 29 SEHRING, J. (06.2006) The Politics of Irrigation Reform in Tajikistan iv No. 30 LANGENOHL, A. / WESTPHAL, K. (11.2006) Comparing and Inter-Relating the European Union and the Russian Federation. Viewpoints from an international and interdisciplinary students' project No. 31 WEBER, S./ ANDERS, S. (3.2007) Price Rigidity and Market Power in German Retailing No. 32 GAVARDASHVILI, G. / SCHAEFER, M. / KING, L. (8.2007) Debris Flows at the River Mletis Khevi (Greater Caucasus Mountains, Georgia) and its Assessment Methods No. 33 TEUBER, R. (5.2007) Geographical Indications of Origin as a Tool of Product Differentiation – The Case of Coffee D No. 34 DOSTAJ, Ž. D. (in Zusammenarbeit mit E. Giese und W. Hagg) (6.2007) Wasserressourcen und deren Nutzung im Ili-Balchaš Becken No. 35 FLATAU, J./ Hart, V. / KAVALLARI, A./ SCHMITZ, P.M. (7.2007) Supply Chain Analysis of Olive Oil in Germany No. 36 HART, V. / KAVALLARI, A. / SCHMITZ, P.M. / WRONKA, T. (7.2007) Supply Chain Analysis of Fresh Fruit and Vegetables in Germany No. 37 MÖSER, N. (7.2008) Analyse der Präferenzen russischer Fachbesucher für ausgewählte Messeleistungen mit Hilfe der Choice-Based Conjoint-Analyse No. 38 BISCHOFF, I. / EGBERT, H. (8.2008) Bandwagon voting or false-consensus effect in voting experiments? First results and methodological limits No. 39 BREDL, S. / WINKER, P. / KÖTSCHAU, K. (12.2008) A Statistical Approach to Detect Cheating Interviewers No. 40 HERRMANN, R. / MÖSER, A./ WEBER, S. (01.2009) Grocery Retailing in Poland Development and Foreign Direct Investment No. 41 HERRMANN, R. / MÖSER, A./ WEBER, S. (02.2009) Grocery Retailing in Germany Situation, Development and Pricing Strategies v No. 42 GÖCKE, M. (05.2009) Efficiency Wages and Negotiated Profit-Sharing under Uncertainty No. 43 KRAMB, M. / HERRMANN, R. (05/2009) Wie wirken gemeldete SPS-Maßnahmen? Ein Gravitationsmodell des Rindfleischhandels der EU No. 44 BREDL,S. (10/2009) Migration, Remittances and Educational Outcomes the Case of Haiti No. 45 BELKE, A. / GÖCKE, M. / GUENTHER, M. (11/2009) When Does It Hurt? The Exchange Rate “Pain Threshold” for German Exports No. 46 EGBERT, H. / FISCHER, G. / BREDL, S. (12/2009) Advertisements or Friends? Formal and Informal Recruitment Methods in Tanzania No. 47 RAKHIMOV, M. (01/2010) The European Union and Central Asia Challenges and Prospects of Cooperation No. 48 NAJMITDINOV, A (01/2010) Central Asian integration as a way of guaranteeing regional security and economic growth feasibility and prospects No. 49 JETPYSPAEVA, Y (03/2010) Central Asia Changing Politics. Case of Kazakhstan No. 50 JONBOBOEV , S. (03/2010) Humanities in Transition Liberation of Knowledge in Central Asia and possible Impact of European Union No. 51 KULAKHMETOVA, , A. (03/2010) Protection Mechanisms and services for young Workers in Central Asia and the European Union No. 52 MOMOSHEVA, S. (03/2010) The EU strategy for Central Asia and Kyrgyzstan foreign policy vi No. 53 EGBERT, H. / FISCHER, G./ BREDL, S. (06/2010) Different Background – Similar Strategies Recruitment in Tanzanian-African and Tanzanian-Asian No. 54 GÖNSCH, I. (11/2010) Determinants of Primary School Enrolment in Haiti and the Dominican Republic No. 55 GÖNSCH, I. / GRÄF, S. (05/2011) Education for All and for Life? An Introduction into Primary School Education in Senegal No. 56 BREDL, S. / STORFINGER, N. / MENOLD, N. (08/2011) A Literature Review of Methods to Detect Fabricated Survey Data No. 57 STORFINGER, N. / WINKER, P. (08/2011) Robustness of Clustering Methods for Identification of Potential Falsifications in Survey Data No. 58 STORFINGER, N. / OPPER, M. (09/2011) Datenbasierte Indikatoren für potentiell abweichendes Interviewerverhalten No. 59 PAWLOWSKI, I. (04/2012) Climate Risk Management in Central Asian agriculture A situation analysis No. 60 SEHRING, J. (06/2012) Forests in the context of climate change in Kazakhstan No. 61 GÖCKE, M. (11/2012) Play-Hysteresis in Supply as Part of a Market Model No. 62 FEDOSEEVA, S. (04/2013) Do German Exporters PTM? Searching for Right Answers in Sugar Confectionery Exports No. 63 GÖNSCH, I. (04/2013) Does the availability of secondary schools increase primary schooling? Empirical evidence from northern Senegal vii No. 64 AHMED, M.N./ PAWLOWSKI, I. (09/2013) Conference proceedings “Natural resource use in Central Asia Institutional challenges and the contribution of capacity building” No. 65 LÜDERING, J. (08/2014) The Measurement of Internet Availability and Quality in the Context of the Discussion on Digital Divide No. 66 FEDOSEEVA, S. (10/2014) Is there a Difference? Exchange Rate Nonlinearities in European Agri-Food (versus total) Exports to the US No. 67 GIERHAKE, K. (1/2015) Integrierter Umbau der Raumstruktur im Metropolitandistrikt Quito (Ecuador) – ein Ansatz zur Diskussion geographischer Innovationsforschung No. 68 GIERHAKE, K. (2/2015) „Espacio público“ –öffentlicher Raum in Quito (Ecuador). Ein Instrument zur Steuerung räumlicher Entwicklung No. 69 GIERHAKE, K. / Vergara, A. (4/2015) Dos miradas sobre el espacio público en Cartagena (Colombia) y Quito (Ecuador): de componente relevante conceptualmente pero relegado en la práctica a instrumento central de gestión urbana No. 71 GIERHAKE, K. AND C. JARDÓN (5/2015) Lokales Wissen – ein Faktor für soziale Innovationen. Kommunalplanung in Quito (Ecuador) No. 72 SAHITO, J.G.M. (5/2015) Market Integration of Wheat in Pakistan Stand 23. Juni 2015 Die Diskussionsbeiträge können auf der Homepage des ZEU http//www.uni-giessen.de/zeu im Menü „Forschung“, „Publikationen“ kostenlos heruntergeladen werden.