^{1}

^{2}

^{*}

^{3}

The objective of this research is to examine impacts of exchange rate volatility and FDI on efficiency of the Vietnamese agricultural sector at the provincial level for the period 1998-2011. Due to the characteristic of high uncertainty in agricultural production, the chance-constrained programming model would be used to estimate efficiency of the agricultural production sector. In order to study impacts of exchange rate volatility and FDI, we employ the two-stage model. In the first stage, we use the chance-constrained programming model to measure technical efficiency and ARIMA model to quantify exchange rate volatility. In the second stage, we use the fixed effect model to evaluate impacts of exchange rate volatility and FDI on efficiency of agricultural production in poor and rich provinces. The estimated results show that fluctuation in exchange rate volatility would reduce efficiency in agricultural production but FDI has an insignificant impact on the efficient production in Vietnam agricultural sector.

After economic transformation, Vietnam, which was the rice importer before 1980, has become the third largest rice exporter in the world in 1989. The volume of rice exports increased from 2 million tons in 1990 to over 3.6 million tons in 2000. The major rice importers of Vietnam consist of China, Malaysia, Singapore, Indonesia, Hong Kong (China), and East Timor. In 2010, India and Thailand were the two largest rice exporters in the world while Vietnam was at the third position. In 2011, Vietnam exceeded Thailand in terms of volume of rice exports to become the second largest after India. However, in terms of value of rice exports, Vietnam is still behind Thailand because the quality of exported rice from Vietnam is lower than one of Thailand, and thereby the price is also lower.

Red river and Mekong river deltas are two major areas of rice exportation in Vietnam. Even though these two deltas account for only 15% of total area square of the whole country, the volume of rice production is two third of the one of the whole country. The Red river delta contributes to 16.7% of total volume of rice exportation of Vietnam.

The volume of rice exportation increases over time; however, the value of exportation is not quite high. One of the main reasons is the low quality of Vietnamese rice which could not satisfy the requirement of customers in the world. There have been several researches of efficiency in food production in Vietnam. For instance, Minh and Long (2009) in [

1) The first issue is that the output of agricultural production is uncertain. The agricultural production can be influenced by several stochastic factors such as weather condition and they can cause a serious damage to agricultural production. This research would not bring the factors of climate change into the model but it still considers the output of agricultural production being uncertain and models them under the chance-constraints.

2) The second issue is that the efficiency of agricultural production could be significantly affected by exchange rate volatility. This argument implies that surprising changes in exchange rate would have impact on decisions of rice exportation because of direct impacts of exchange rate on inputs such as fertilizer, pesticide, seed… and on supply of rice for exportation; hence they would affect production efficiency. This argument is quite close to ones of Ethier (1973) in [

3) The third issue is to examine differences in terms of effects of FDI flows on efficiency in agricultural production between rich and poor provinces.

As known so far, many countries try to attract FDI by undertaking different packages of policies because they expect that FDI inflow could bring new technology, know-how and contribute to improvement in productivity and competitiveness of domestic industries. They grant foreign enterprises favorable treatments such as subsidy or favorable tax treatment because policymakers believe that FDI can bring positive externalities to the economy (Caves (1974, 1996) in [

All issues above would be brought in different ways into the model.

All issues mentioned above would be presented in this section. Firstly, we would model the uncertainty of out- put in agricultural sector by chance-constrained data envelopment model (CCDEA) to estimate efficiency, produc- tivity growth in agriculture of sixty provinces in Vietnam. Then, models to evaluate impacts of exchange rate vola- tility and foreign direct investment on efficiency of agricultural production sector in Vietnam would be given.

We consider a set of N provinces, each consuming amounts of P inputs to produce M outputs. Assuming that each provinces has at least one positive input and one positive output and we can construct the production frontier

H satisfies the free disposability of inputs and outputs and includes all provinces.

The CRS input-oriented measured of technical efficiency for the r_{0} th province is calculated as solution to the following mathematical programming problem:

The scalar value l represents a proportional reduction in all inputs such that 0 £ l £ 1, and _{0}th provinces.

Maximum technical efficiency is achieved when

There are several papers about efficiency which employ chance-constrained programming model. Land et al. (1993) in [

Assumption 1: Stochastic output.

Based on DEA model mentioned above, we assume that the probability that the best practical agricultural output exceeds the observed output must not be less than

can be converted to the following constraint:

Thus, corresponding chance-constrained efficiency measure is calculated as:

Assumption 2:

Assumption 3: All outputs of provinces are randomly independent and

We use two Banker’s asymptotic DEA efficiency tests to test for inefficiency differences between two different efficiency scores:

1) The two inefficiency scores

The test statistic is: _{CCDEA}, 2N_{DEA}) degrees of freedom.

2) The two inefficiency scores _{CCDEA}, N_{DEA}) degrees of freedom.

To analyze impacts of FDI on technical efficiency in agricultural production, two variables are included: FDI represents the FDI inflow to each province in the period 1998-2011. This variable can reflect impacts of vertical spillover of FDI to agricultural production of provinces. The second variable FDI*G is the product of dummy variable G (taking value of 1 if that is a rich province and 0 if not). The reason for constructing this variable includes: 1) firstly, it can be shown that rich provinces have more FDI enterprises than poor ones have, 2) theoretically, richer provinces have more advantages in capacity of absorbing spillover effects of FDI.

There have been a lot of researches concerned with impacts of exchange rate volatility on trade flows. Many measurements of exchange rate volatility have been used. For instance, Thursby and Thursby in [_{t} is the spot rate and t represents the period} as the measurement of exchange rate volatility. McIvoi [

In this research, exchange rate volatility is measured by residuals from ARIMA model. The exchange rate volatility is estimated from quarterly data, however data used to estimate the model of evaluating impacts of exchange rate on efficiency are only available at the annual frequency. Thereby, we would use the two-step procedure to select proxy variable for exchange rate volatility.

1) Estimation of ARIMA (p, q, d)

in which p is the autoregressive order, q is the moving average order, and d is the degree of difference.

The vector

2) From residuals e_{t} of the estimated model ARIMA (p, q, d), we pick up three values each year

denoting s_{e}_{3}, in which e_{t}_{I}, e_{t}_{II}, e_{t}_{III}, e_{t}_{IV} are residuals e_{t} of the model ARIMA (p, q, d) in year t, quarter I, quarter II, quarter III and quarter IV respectively.

Then, we can establish the series {s_{e}_{1}}, {s_{e}_{2}} and {s_{e}_{3}}. These series correspond to three models, in which the models differ from each other at the way to measure exchange rate volatility. The exchange rate volatility in each model can be estimated from a pair of variables s_{e} and s_{eN}. The variable s_{e} {(s_{e}_{1}), (s_{e}_{2}), and (s_{e}_{3})} shows exchange rate volatility while s_{eN} is the product of two variables e and n, in which nis the dummy variable taking value of 1 in the case of poor province, i.e. having income lower than the average level of the whole country. This variable ({s_{e}_{1n}}, {s_{e}_{2n}} and {s_{e}_{3n}}) would help us examine impacts of exchange rate on poor provinces.

To study impacts of exchange rate volatility and FDI on efficiency in agricultural production, we employ fixed effect model to evaluate impacts of exchange rate volatility (s_{e}), FDI, FDIG and s_{N} on variables TECCDEA. The model can be specified as follow:

with i denoting provinces, and t denoting time periods. The dependent variable TECCDEA_{i}_{,t} is the annual tech-

nical efficiency of ith province at time t, estimated from CCDEA model.

province at time t. FDI_{i}_{,t} is foreign direct investment in ith province at time t. s_{et} (residuals) represents exchange rate volatility in year t, estimated from ARIMA(p, q, d) model. FDIG_{i}_{,t} is foreign direct investment in the rich province at year t.

There are three forms of model (7). Model (7a) ((7b) and (7c)) is model (7) with s_{e} in (6a) ((6b) and (6c)).

The parameter a_{i} may have two different interpretations and two different models may be distinguished according to this interpretations. If the a_{i} is assumed to be fixed parameters, Equation (6) is a fixed effect panel data model. Conversely, if the a_{i} are assumed to be random, Equation (6) is a random effect panel data model. In general, fixed effect model is indicated when the regression analysis is limited to a precise set of individuals. For this reason, since our data set consists of the observations over the 60 provinces, we use a fixed effect panel data model to analyze the impacts of exchange rate volatility and FDI on technical efficiency from CCDEA model in agricultural production.

Data set used in this research consists of inputs and outputs of agricultural production in sixty provinces in Vietnam in the period 1998-2011. Data of some provinces must be added up due to separation or combination between provinces in this period. For instance, Ha Tay and Hanoi are two separated provinces before 2008, but then they are combined with each other to become new Hanoi. Therefore, statistical data for HaTay would not be available since then. To ensure the consistency of data, we would add up each indicator of HaTay to the corresponding indicator of Hanoi in the period 1998-2008, and now we have new data set for only one province, namely new Hanoi. We do the same transformation for two other pairs of provinces including Dac Lac and Dac Nong, Dien Bien and Lai Chau. The data set used in this research is collected from General Statistical Office and Ministry of Labor-Invalids and Social Affairs.

Most of researches about agricultural productivity in Vietnam use gross value of agricultural value as gross value of agricultural production. The gross value of agricultural value is defined as the gross value of agricultural production in following fields: cultivation, forestry, animal husbandry, fishery, and secondary works. The value of all inputs in agricultural production would also be included in the gross value of Vietnamese agricultural value. Thereby, in this research, the net value or added value of agricultural output (VAO) would be used to measure gross value of Vietnamese agricultural output. The VAO data of provinces in the period 19988-2011 would be adjusted with respect to GDP deflator, in which base year is 1994. The quarterly exchange rate (VND/$) data in the period 1995-2012 and the data of implemented FDI in the period 1998-2011 are collected from General Statistical Office.

We employ the indicator, namely inverse of (2), to measure technology efficiency in agricultural sector for each province in Vietnam in the period 1998-2011. These values of technical efficiency are measured under the assumption of constant returns to scale within sixty provinces in the period 1998-2011.

To have a preliminary picture of technical efficiency in agricultural sector, we would firstly review estimated results of efficiency of provinces in this period. The results show that agricultural production in Ho Chi Minh city has efficiency score lying in the group of ten best provinces within 12 years, and lying on the frontier curve in 9 years. Agricultural production in Tra Vinh also has efficiency score lying in the group of ten best provinces within 12 years, but only lying on the frontier curve in 5 year. Agricultural production in Ba Ria-Vung Tau has efficiency score lying in the group of ten best provinces within 10 years, and lying on the frontier curve in 4 years. Efficiency score of agricultural production of Ben Tre belongs to the group of ten best in 9 years and lies on the frontier curve in 4 years... In contrast, the group of ten least efficient provinces in agricultural production includes provinces such as Quang Ninh and Dien Bien (in 13 years), Thai Nguyen and Quang Binh (in 12 years), Ha Giang and Phu Yen (in 11 years). Especially, Phu Tho has lowest efficiency score of agricultural production within 13 years but in 2011, it steps up to become one of the ten most efficiency in terms of agricultural production. Quang Ngai also lies in the group of ten least efficiency in 1998 but jumps up to the group of ten most efficiency in 2011.

To compare between technical efficiency estimated from CCDEA model and technical efficiency estimated from DEA model, we use two Banker’s asymptotic DEA efficiency tests.

The estimated results of two Banker’s asymptotic DEA efficiency tests can be given in

The test statistics estimated under the assumption of the two inefficiency scores following the exponential distribution (the half-normal distribution) are reported in the second column (the fourth column) in

We conduct a regression analysis to determine the impacts of exchange rate volatility and FDI flow on technical efficiency in Vietnam agricultural sector.

The first column in this table shows the name of variables in the models. The third, fourth, and fifth column give estimation results from three models in which efficiency scores estimated from CCDEA model. All three models have the same independent variables except exchange rate. In the first model, the maximum exchange rate (s_{e}_{1}) is used, and in the second model the minimum exchange rate (s_{e}_{2}) is used and in the third one, the average exchange rate (s_{e}_{3}) is used.

Year | Exponential type | Critical value | Half-normal type | Critical value |
---|---|---|---|---|

1998 | 0.9602 | 1.5330 | 1.2778 | 1.8360 |

1999 | 0.8646 | 1.5330 | 1.2250 | 1.8360 |

2000 | 1.0395 | 1.5330 | 1.3134 | 1.8360 |

2001 | 1.063 | 1.5330 | 1.3644 | 1.8360 |

2002 | 1.1131 | 1.5330 | 1.3784 | 1.8360 |

2003 | 1.0593 | 1.5330 | 1.3501 | 1.8360 |

2004 | 1.0288 | 1.5330 | 1.3294 | 1.8360 |

2005 | 1.0179 | 1.5330 | 1.3001 | 1.8360 |

2006 | 1.0311 | 1.5330 | 1.3080 | 1.8360 |

2007 | 0.999 | 1.5330 | 1.3005 | 1.8360 |

2008 | 0.9565 | 1.5330 | 1.2659 | 1.8360 |

2009 | 0.8596 | 1.5330 | 1.2187 | 1.8360 |

2010 | 0.8219 | 1.5330 | 1.2043 | 1.8360 |

2011 | 1.0016 | 1.5330 | 1.3144 | 1.8360 |

Source: The authors estimate from the data of GSO.

Chance-constrained model | |||
---|---|---|---|

Model (7a) | Model (7b) | Model (7c) | |

FDI | 0.00026 (0.00002) | 0.000267 (0.00002) | 0.000018 (0.00002) |

K/L | 1.3882^{***} (0.3118) | 1.4055^{***} (0.3139) | 1.4688^{***} (0.3155) |

s_{e } | −2.0277^{***} (0.3128) | −1.7632^{***} (0.3222) | −4.4618^{***} (0.7945) |

FDIG | −0.00004 (0.00003) | −0.00004 (0.00003) | −0.00004 (0.00003) |

s_{en} | 1.1293^{*} (0.6271) | 0.7733 (0.7869) | 5.7488^{***} (1.6723) |

_cons | 0.5077^{***} (0.0055) | 0.5003^{***} (0.0050) | 0.4882^{***} (0.0044) |

/sigma_u | 0.2242 | 0.2251 | 0.2250 |

/sigma_e | 0.1139 | 0.1148 | 0.1147 |

Rho | 0.7946 | 0.7936 | 0.7936 |

R^{2}: within | 0.0775 | 0.0641 | 0.0653 |

Between | 0.0446 | 0.0095 | 0.0133 |

overall | 0.0267 | 0.0169 | 0.0181 |

Number of obs | 840 | 840 | 840 |

Source: The authors estimate from the data of GSO.

The estimation results show that in all three models, the coefficient of exchange rate variable (s_{e}) takes negative value and highly statistical significance. This implies that fluctuation in exchange rate would reduce efficiency in agricultural production. This can be explained by the fact that a large proportion of agricultural production in Vietnam is rice, whose export value is very large. Exchange rate volatility can directly affect exports. This result is compatible with theoretical view that exchange rate volatility would reduce trade flow. This view claims that a sudden change in exchange rate would have effect on decision of risk-adverse traders, therefore volume of trade would decline. This view is also given in Ethier [_{en} is positive and statistically significant (except in the model 2), which means that exchange rate volatility in the case of provinces having no rice exports has positive impacts on efficiency. The possible reason is that exchange rate volatility often makes imported agricultural materials cheaper, so provinces without rice export would benefit due to lower input costs, so the product of these two variables would take positive value and be highly statistically significant.

The estimation results of FDI variable in all three models take positive value but insignificant at the any level. The sign of the variable GFDI is negative and insignificant. It means that FDI flow during the period of 1988- 2011 have not been significant impact on the efficient production in Vietnam agricultural sector.

This research employs chance-constrained DEA approach to estimate technical efficiency in Vietnamese agriculture sector in the period 1998-2011.

This research examines effects of exchange rate volatility and FDI flows on efficiency of production in agriculture sector by using residuals from ARIMA model as proxy variable for exchange rate volatility. By dividing provinces in the country into two groups, namely rich and poor one, we can derive several interesting findings. The results show that in all three models, the coefficient of exchange rate variable is negative and highly statistically significant, that is, fluctuation of exchange rate would reduce efficiency of production in agriculture production. However, exchange rate volatility in provinces without rice exportations has a positive impact on efficiency. The coefficient of FDI variable is positive and insignificant at any level. Meanwhile, the coefficient of FDIG variable is negative and insignificant.

Nguyen KhacMinh,Pham VanKhanh,Nguyen VietHung, (2015) Impacts of Exchange Rate Volatility and FDI on Technical Efficiency—A Case Study of Vietnamese Agricultural Sector. American Journal of Operations Research,05,317-325. doi: 10.4236/ajor.2015.54025