1. Introduction
Agricultural prices inherently exhibit volatility—the degree of price movement, or the possibility of substantial, unexpected changes in the commodity price—and it is inextricably linked to and affects the welfare of both producers and consumers [
1]. Due to this, there is a marked impact on the supply chain which simultaneously affects other factors like investment, market performance, societal development, etc. The concern is spiraling food price volatility, especially in staple foods like wheat and rice, which has an adverse effect and can affect the stakeholders’ interest along the supply chain. The price of an essential product determined by the extent of demand-supply is, for the most part, fixed disposed to the level of production, stock, Government procurement, support price, and other policies and various subsidies that are eventually reflected in the marketplace [
2].
Extreme prices have the potential to exacerbate and influence broader societal concerns in terms of access to food, human development, and political as well as economic stability. Extensive research has been carried out already [
3] on the past prevalence of food price crises, and it has been attributed to political and economic circumstances, such as the transmission of crises from agriculture practiced in rural to urban regions, among other reasons. The occurrence of regional crises altered, as markets became more integrated. Analyses of the 1970s’ worldwide food price crises focused on production and trade shocks [
4]. Commodity markets that are efficient and functioning well are successful in communicating price signals geographically (across spatially separated regions) and temporally (through varying periods), thus facilitating market resource distribution and encouraging investment [
5,
6,
7,
8].
Research on market integration in agricultural commodities draws potential inferences for economic wellbeing, efficiency, and performance [
9,
10]. It provides information on whether such integration has improved over time due to policy initiatives [
9] and the degree of required intervention to correct the inefficiency [
11]. Disintegrated markets send inaccurate information on prices resulting in wrong policy decisions and resource allocation [
10]. Alternatively, the degree (and transmission direction) of price cointegration determines market performance since the price stabilization measures in one significant market produce the desired outcome in others via the arbitrage process [
12].
Agricultural commodity production is affected by regional topography, soil conditions, etc., and the consumers’ preferences add to the commodity prices to become noisy, non-stationary, and possibly leptokurtic, making it challenging to capture the dynamics [
13]. Extreme prices are a significant concern because they have determined the ‘fate and fortunes’ of several emerging economies due to their linkage to the flow of evidence-based information, especially in a free-market situation. In the realm of foodgrains, wheat and rice are the staple food for several nations, including the agrarian Indian economy. Hence, it becomes vital to understand the extent of price behavior and cointegration as it affects not only production but also consumption decisions. Wheat and rice prices as well as price cointegration across markets serve as essential aspects to examine from the stakeholders’ point of view since they have central control on production and procurement through the major role they play in influencing the market efficiency and performance.
Researchers have focused on the asymmetry of price transmission [
14] and adjustments, the volume of production, and the time period during which volatility was observed along the food chains [
15]. Several research studies on price analysis and spatial market integration have been undertaken in India in the recent past [
2,
9,
12,
13,
14,
16,
17,
18,
19,
20,
21,
22,
23,
24,
25], but little is known about the recent—especially during the COVID-19 incidence period—price dynamics, degree of cointegration, and direction of price transmission in wholesale and retail markets of wheat and rice, a Government-subsidized staple commodity in India. Futher, no studies were carried out on spatial and vertical price integration amidst structural break(s), like the effect of the pandemic, on commodity prices. Monitoring the representative markets for agricultural commodities will help in managing the price shocks prevailing in lagged markets owing to any uncertain situation [
17]. The COVID-19 crisis had a huge impact on food production, processing, distribution, and demand [
26,
27]. Although various disruptions [
28,
29] in the supply chain have been observed, many agricultural markets have performed efficiently [
30,
31] because of timely intervention as agricultural commodities were essential for life sustenance. Thus, in this study, we have been motivated to analyze these aspects, and, in the recent past, questions have arisen about to what extent the agricultural value chain of India’s staple foods has deviated or adjusted under the pandemic situation [
32,
33] and, also, if any such situation may arise in the future. Given the complexities of price behavior in the wholesale and retail markets, a thorough evaluation of the price dynamics and integration will aid in prioritizing investments, removing distortions, and adopting policies to improve overall performance.
In the milieu, an attempt has been made in this study with the specific objective of examining the price behavior in wholesale and retail markets of rice and wheat in addition to analyzing the extent of price cointegration in the major grain markets of the selected staple commodities. Such types of study have to be revisited due to the changing economic scenario post-pandemic and breakthroughs in the analytical methods of researching the causal factors and effects in the market ecosystem.
2. Data and Methods
India was purposively selected, being one of the largest producers and consumers of rice and wheat. The study was marshaled on the wholesale and retail monthly prices of rice and wheat for four selected markets in India viz. Chennai (South), Delhi (North), Mumbai (East), and Patna (West), covering four major regions (one representative market from each zone). The price data in ₹ per Kg (₹ refers to the Indian Rupee, the national currency, also denoted as INR) were compiled from the Food Price Monitoring and Analysis (FPMA) (
https://fpma.fao.org/giews/fpmat4/#/dashboard/tool/domestic, accessed on 2 October 2022) tool of the United Nations Food and Agriculture Organization (FAO) for the agricultural year spanning from July 2000 to June 2022. Along with descriptive statistics, various analytical tools and techniques (analyzed using Excel, SPSS ver. 26, and EVIEWS ver. 13 software) were used for better interpretation of the research findings and to draw some valid conclusions.
2.1. Theoretical Framework
Deciphering the fluctuations in the prices of agricultural commodities across spatially separated markets helps us to understand the dynamic behavior of the time series across regions, which facilitates sketching out the economic implications [
14,
16,
20]. In general, agricultural commodity prices tend to be noisy, non-stationary, and largely leptokurtic, posing challenges in capturing the dynamics. To start with, analyzing the growth in wholesale and retail markets’ price series helps to capture the spatial dynamics. Similarly, estimating the instability in prices gives a clue about the extent of risk involved [
13,
16]. It is expected that price series showing a higher growth has high risk due to the discernible change in the prices from the base period to the recent period. In the next stage, estimating the variation in prices offers additional information on the behavior of the prices across markets, which guides policy makers to devise appropriate decisions [
13,
16]. As agricultural commodities production is seasonal in nature, price deseasonalization has to be undertaken to capture the actual variation [
16,
20]. Comparing the seasonal indices across the different markets would enhance our understanding of the price variation as the commodity is produced at some period of the year, but consumption takes place throughout the year. Intra-year price rises and average seasonal price variations are some indicators in price dynamics analysis that evaluate the degree of seasonal price variation.
To model the price series for analyzing the extent of integration between two or more markets, cointegration analysis attains significance. However, considering the longer period of the data, there exists the possibility of structural breaks in the price series, and these have to be detected to produce reliable results. Literature contains a vast amount of work on the issues related to structural change, the majority of it specifically designed for the case of a single period, but macroeconomic parameters generally have more than one structural break and most researchers have shown interest in knowing the basic procedure to structural breaks. As far as the significant literature is concerned, Quandt [
34], Brown et al. [
35], Ploberger et al. [
36], Andrews [
37], and Hansen [
38] have contributed to testing the problem of structural breaks. Structural breaks can be divided into two types—i.e., known structural breaks like the 2008 financial crisis or the COVID-19 pandemic, and unknown structural breaks. When the structural break is known, the Chow test is more appropriate. However, in the case of unknown structural breaks, no prior understanding of the structural change that manifested itself in its time, nature, and shift is needed.
Bai and Perron [
39] provided a comprehensive analysis of several issues regarding multiple structural change models, and developed some tests. These tests are helpful to capture the present change(s) and also determine, endogenously, the points of break with no prior knowledge. Under this approach, first, the structural breaks are detected using the time-series properties for all series in a system mode. Later, the price series are subjected to the stationarity test by determining whether they are trend stationary or difference stationary using the augmented Dickey–Fuller test [
40,
41]. This approach is effective in managing small samples and, if augmented sufficiently, it avoids uncertainty about variable exogeneity. After testing for stationarity, price series are subject to the cointegration test.
Price cointegration is an ideal situation wherein the prevailing prices of a commodity across locations follow a similar pattern in the long run [
42,
43]. Rapsomanikis et al. [
44] discussed applying price cointegration tools, particularly for developing countries. A series of studies conducted on price cointegration suggested that market functionaries can achieve benefits through integrated markets [
45,
46,
47,
48]. Integrated markets encourage the dissemination of information across time, space, and form. Many studies have used the procedure introduced by Engle and Granger [
49] to examine market integration. Thereafter, Johansen [
50] introduced the alternative technique to examine price cointegration along with multiple cointegrating vectors. In this line, Kumar and Sharma [
51] observed that Johansen’s test, having multiple advantages, is very easy to compute and robust enough with
sans a priori assumptions on variables with testing, simultaneously, the number of cointegration vectors un-imposed earlier. After the cointegration analysis to assess the co-movement, a causality test is widely suggested. So, after post-testing for cointegration using Johansen’s approach, Granger causality analysis [
52] is undertaken to know the cause-and-effect relationship between the selected markets. The purpose of this statistical test is to determine whether a given price series (market A) and its lags explain the prediction of another series (market B). In addition, the current study has calculated impulse response [
53,
54] to obtain the dynamic interrelationships among the prices of different markets. The purpose is to investigate whether the mechanism of shocks, if any, persists in the market ecosystem. The analysis tracks the effect of ‘one’ SD (standard deviation) or ‘one’ unit shock imposed on one price series that is then reflected on the current and future values of all the endogenous variables over a specific time period [
55]. Following the aforementioned theoretical framework, the detailed tools used in the study are given below.
2.2. Estimation of Growth Rate
The growth in wholesale and retail prices of both food commodities was calculated by using the below-mentioned formula [
40,
56,
57]:
The equation was transformed into logarithmic function before estimating the growth rate using the ordinary least squares method [
40].
where
Pt represents the price at time ‘t’,
0 denotes constant, and
b represents the growth rate.
2.3. Price Instability Index
The Cuddy-Della Valle Index (CDVI) approach was used to examine the magnitude of variation and risk involved in the prices of rice and wheat [
58].
where
CDVI is the Cuddy-Della Valle instability Index (%),
CV is the Coefficient of Variation (%), and
is the coefficient of multiple determination.
2.4. Seasonal Price Index
The seasonal index—measured using the 12-months ratio to the moving average method—is a way to measure the variations in prices across the commodity production seasons [
59]. Subsequently, deseasonalization was carried out to eliminate seasonal variations in wholesale and retail prices of rice and wheat for the selected markets in India. The de-seasonalized commodity prices were calculated by adopting the given formula [
59]:
In addition, the Intra-year Price Rise (
IPR in %) and Average Seasonal Price Variation (
ASPV in %) were estimated to evaluate the degree of seasonal price variations in rice and wheat for the selected markets in India.
where
SPIH is the seasonal price index with the maximum value, and
SPIL is the seasonal price index with the minimum value.
2.5. Price Cointegration
Before applying any test, the foremost important step in the time series data is to check the stationarity and detect structural breaks, if any, since the considered time period is too long (July 2000 to June 2022). A stationary time series is one whose statistical properties like mean, variance, and covariance are invariant. The estimated relationship may be counterfeit without any significant implication in the absence of stationarity. The wholesale and retail price series of rice and wheat crops in the selected markets were first checked for stationarity by using the Augmented Dickey–Fuller (ADF) unit root test [
40,
41].
2.5.1. Detection of Multiple Unknown Structural Breaks
Of the two types of structural break, the study considers the detection of multiple unknown breaks following the below linear regression with ‘
m’ breaks (
m+1 regime):
where
𝑦𝑡 is a function of 𝑋𝑡 and 𝑍𝑡; j = 1, …, m + 1; xt (p × 1) and zt (q × 1) are vectors of covariates; β and δj(j = 1, …, m + 1) are the corresponding vectors of coefficients; and ut is the disturbance at time t. The assumption that underlying economic events remain constant across the entire period is relaxed in the structural breakpoint. Before presenting the results, it is important to know the hypotheses of this study.
H0: There is structural stability.
H1: There are structural breaks.
The p-values enable us to accept or reject the null hypothesis. The study of structural breaks is inevitable in macro-econometric modeling due to structural changes or regime shifts such as systemic shocks—e.g., price fluctuations, demand and supply shocks—and economic and institutional changes.
2.5.2. Unit Root Test
The Augmented Dickey–Fuller (ADF) test was carried out to examine the stationarity in the data [
40,
41]. The ADF test is executed by estimating the following equation:
where
P represents price,
is the constant,
t represents time,
q is the number of lag length, and
is the random error-term.
Unit root testing was carried out by framing the null hypothesis, H
0:
= 0; that is, there is a unit root and the time series is non-stationary, and the alternative hypothesis, H
1:
< 0; that is, the time series is stationary. A rejection of the null hypothesis suggests that the particular price series is stationary [
40].
2.5.3. Cointegration Test
To test the long-run relationship in commodity prices between the wholesale and retail markets, the study used the Johansen’s cointegration test [
50,
60]. The null hypothesis (H
0) of utmost ‘
r’ cointegrating vectors—i.e., rank of error-correction coefficient matrix—against a general alternative hypothesis (H
1) of ‘
r+1′ cointegrating vectors is tested by trace and maximum eigenvalue statistics [
34].
where
r is the number of cointegrated vector,
is the eigenvalue and
is the (r + 1)th largest squared eigenvalue obtained from the parameter matrix of the cointegrated system, and
T is the effective number of observations.
In the present study, we have found that the wholesale and retail price series of rice and wheat crops have structural breaks. In order to see the effect of such breaks on the cointegration of the rice and wheat markets, we have introduced three dummies for each crop in our model. These breaks have been incorporated as the methodology given by Johansen et al. [
61].
2.6. Granger Causality Test
The Granger causality test has been employed to know the path of price transmission across markets [
52]. The Granger causality test is carried out for a market pair, testing whether price series
Pt Granger-causes price series
Qt, and vice versa. All sorts of permutations and combinations are possible within the selected markets viz. univariate Granger causality indicating price transmission from
Pt to
Qt or from
Qt to
Pt, bivariate Granger causality depicting price transmission in both directions, or absence of causality implying no price transmission. The above test is carried out using the following equation:
where
P and Q are the selected market prices series pair,
and
t indicates the time period.
The subscripts, ‘i’ and ‘j’ represent the number of lags pertaining to the respective price series.
The null hypothesis states that ln
Pt does not Granger-cause ln
Qt, and its rejection implies that there is Granger causality between the selected price series pair [
40].
2.7. Impulse Response Function
In the present study, the Generalized Impulse Response Function (GIRF) proposed by Koop et al. [
53] and further advocated by Pesaran and Shin [
54] has been employed. The GIRF of a random current shock ‘
’ and historical shock ‘
’ is given in the below equation:
4. Conclusions
Price analysis on staple food commodities viz. rice and wheat exhibited strong evidence of spatial and temporal dynamics. In addition, a clear-cut seasonality has been witnessed, especially in wheat, linked to its harvest month(s). Further, price divergence between the wholesale and retail markets was witnessed in rice and wheat over time and space. Johansen’s test following structural breaks indicated a strong degree of price cointegration between the wholesale and retail markets. In terms of causation, using the Granger causality test, the price series exhibited unidirectional-, bidirectional- and no causality. Finally, the analysis of impulse response revealed the efficiency of the rice and wheat markets in terms of ‘price discovery’. The analysis also confirms that ‘price discovery’ takes place initially in the wholesale market, and then gets transmitted to the retail market. Overall, the research findings from analyzing the prices of the rice and wheat wholesale and retail markets reveal some vital information to the stakeholders viz. producers, traders, and consumers who have a potential interest in the market ecosystem for taking agri-business decision(s). The derived information from price analysis will facilitate them to take advantage of the price movement in staple commodities being seasonally produced; either in buying, selling, stocking, or distribution. On the policy front, the study advocates for strengthening the existing market intelligence, investing in infrastructure, and reducing the distortion in prices prevailing in geographically separated markets to improve the efficiency and overall performance. Being staple commodities, the Government has substantial control over them through price policy intervention like implementing the minimum support price. Such policies, although cost-intensive, insulate the economy from food inflation. The findings from this investigation will guide the policy makers to suggest the pertinent role of the Government in price stabilization. However, the study has its own limitations with respect to the selection of only two food commodities (i.e., rice and wheat), aggregated price series, and coverage of markets (only four across India). The future research should include a basket of commodities from commodity groups like cereals, pulses, oilseeds, fruits and vegetables, dairy products, etc., and more markets to gain a comprehensive scenario of the Indian food markets.