Peru is currently in a critical point in its development process. Currently included in the urbanized countries of Latin America (World Development Report 2008), Peru has seen the gradual decrease in importance of agriculture in GDP (agriculture production accounts for 9% of total GDP for 2008). Yet, Peru still faces many of the characteristics found in transforming countries characterized by rising rural-urban income disparities and extreme rural poverty. Development resides on wherever Peru can overcome its challenges both as an urbanized and as a transforming country. A vast number of economists have emphasized the importance of the agricultural sector not only as a means to provide food resources but also as a necessity for development. The link between economic growth and an efficient agricultural sector has been proven explicitly or implicitly by many literatures as a backbone for the direct and indirect achievement of development (Hayami, 2001).

According to previous studies, Peruvian agriculture development faces two major challenges. The first challenge is the problem of low agricultural productivity which is vital for stimulating economic growth and reducing rural-urban income disparities. For the last two decades all efforts by the public sector had focused on increasing agricultural productivity, yet this has been done without paying much attention to the relationship between product and market.

The second challenge is the accessibility to a competitive commercialization system. In the current setting of the Peruvian agriculture, market competition is a crucial factor in assuring that the system of prices assigns resources in an efficient way. Productivity must be accompanied by an adequate infrastructure system, with information and institutions that provide a fair value to the producers as well as to consumers.

Both of these challenges are associated with efficiency, at the production level and at the commercialization level. To analyze one without considering the other could lead to ineffective implications, since both sectors are highly integrated.

The problematic in the Peruvian agriculture development that this study focuses on is twofold. The first deals with the problem of low productivity at the Peruvian agriculture based on the viewpoint of the disproportionate regional resource endowment. The second problem deals with market inefficiency of the Peruvian agri-food system.

The purpose of this study, after examining both arguments, is to reach an integrated understanding of the efficiency problems in the Peruvian agriculture in order to determine opportunities to alleviate the challenge faced.

This paper is divided into two main parts. The first tackles the productivity problem of the Peruvian agriculture based on the viewpoint of the disproportionate regional resource endowment and its relationship to efficiency and technological change. The second part tries to grasp the market efficiency of the Peruvian agri-food sector by applying vertical and spatial price transmission analysis.

Chapter 2 tries to quantitatively examine the evolving regional agricultural process in Peru for the period 1970-2003 applying a labor productivity decomposition analysis (Kuroda, 1995) to the three main regions. This study provides a segregated regional analysis with results that might differ from those of traditional aggregated national examinations.

According to the regional Peruvian agriculture dual cost estimates, labor productivity levels were greatly dependant upon technical change only in the Costa region. The Sierra and Selva regions share very similar characteristics (increasing returns to scale and low TFP levels). This might be explained by the differences in natural resources, crop production technology, and geography in the Costa region as compared to the other two regions. Low growth rates of labor productivity have been the trend for both the aggregated view as well as the disaggregated regional views, with differences between the Costa region and the Sierra and Selva regions. Factor price change effects have contributed positively to increase labor productivity in all the regions except for the Selva region, implying that farm units were partly sensitive to factor price changes. In the Selva region the contribution by factor price changes to increase labor productivity has not been substantial, possibly due to the lower availability of agricultural labor and the higher prices of fertilizer and machinery inputs. In addition, the biased technical change effects were labor-saving for the Costa and Selva regions. For the Sierra region the labor-consuming trend might be due to the high agricultural labor pressure on arable land and relatively low wages. Machinery use has been significantly biased toward saving in both the Sierra and Selva regions. This might be related to their small farm size which precludes the necessity of machinery; the ample availability of labor in the Sierra region as a cheaper alternative; and the fact that the imported technology was probably not suited for the specific geographical requirements of the mountainous Sierra region and the rainforest environment of the Selva region.

Chapter 3 focuses on productivity at the provincial level and tries to examine the unobserved characteristic of the Total Factor Productivity (TFP). Although it can be decomposed it into its many measurable effects (Antle and Capalbo 1988), there still remains an unobserved characteristic that cannot be molded into quantitativeness. This unobserved trait which stems from different sources, e.g. management attributes, propensity to try new technology, and entrepreneurship, could be obtained through panel data estimates (Gardebroek and Oude Lansink, 2003). By applying this methodology to the case of Peruvian agricultural output we can determine which factors could explain the recent trend of productivity divergence between traditional and non-traditional crops. Traditional crops, which have been the main output of Peruvian agriculture, have started to decrease in importance, while non-traditional crops such as asparagus have seen an exponential increase in their share of agricultural GDP in the last decades. The objective of this chapter is to try to econometrically examine the unobserved factor components in agricultural TFP for the different provinces of Peru, and its relationship with traditional and non-traditional agricultural outputs.

Results indicated that province-specific effects are correlated with investment and technological progress. Since the province-specific effects account for the unobserved components of TFP, we can conclude that differences in TFP between provinces are related to differences of these quasi-fixed inputs. By looking at the correlation between these fixed inputs and outputs, we can see that the technological trend reduces output for traditional crops and increases output for non-traditional crops. This result may be explained from two possible viewpoints. It seems that the technological trend is clustered only in some provinces and in some crops. Provinces that produce traditional crops seem to lack the ability to positively absorb the technological trend, showing a negative effect. This may be due to the lack of technical assistance and managerial efficiency as well as production and environmental restrictions of traditional crops. Another possibility that is that there seems to be a large divergence between traditional and non-traditional output due to technological trend for some provinces and some crops. Infrastructure investment increases output for both traditional and non-traditional crops, although land and infrastructure investment seem to have a much greater effect on output for traditional crops than for non-traditional crops. This might be explained by the differences in existing resource requirements, crop production technology and investment scale between the two types of crops and the provinces in which they are produced. It can be inferred that effective managerial abilities are a determinant of infrastructure investment or that infrastructure investment help increase TFP levels.

Chapter 4 delves into the market efficiency of the Peruvian agriculture system by analyzing vertical price transmission. Price transmission analysis has been used in countless studies to try to determine the efficiency of vertical or spatial markets. Yet there is one trait common to market players that is of importance, adjustment resistance. This behavior can be interpreted by way of thresholds. For the case of the agri-food supply system of Peru, saturation and a highly competitive distributor sector may provide an understanding for the interpretation of price adjustment resistance. The objective of this chapter is to try to econometrically analyze and assess the presence of nonlinear price transmission behavior between farm-gate, distributor and wholesale agri-food prices in Peru using a Threshold Vector Error Correction Model (TVECM).

Granger-causality results indicated the presence of significant feedback between farm-gate and wholesale prices, yet no significant feedback between farm-gate and distributor prices. Price transmission results indicated that, in general, wholesale prices were significantly more responsive than distributor prices to price variations due to exogenous shocks in the long-run. Results indicated non-linear performance in this system, with distribution prices strongly adjusting to farm-gate price variations only after a certain threshold was surpassed. This behavior might be explained by the high fixed costs, excess capacity, and competition in the distribution market level of the Peruvian agri-food supply system which prohibit perfect price adjustment if the price variation is within a threshold value.

Chapter 5 provides a spatial market integration analysis using a threshold model. Recently the addition of thresholds in VECM has opened the possibility to include an important issue in market arbitrage, transaction costs. For the case of the Peruvian agri-food supply structure, different distribution networks may provide an understanding for the interpretation of arbitrage and transaction costs. The objective of this chapter is to try to econometrically analyze and assess the presence of threshold behavior between separated markets for different Peruvian agri-food commodity crops using a TVECM (Hansen and Seo, 2002), in order to determine possible factors influencing transaction costs and arbitrage.

Results indicated that Lima potato and rice wholesalers are more flexible towards price variation of potato and rice in comparison to other provinces. Observing long-run speed of adjustment estimates between provinces other than Lima we saw that trade flowed in both directions which was expected since most provinces, with the exception of Lima, alter between being net-suppliers and net-consumers of potato and rice during different periods. For the case of brown sugar the results show that trade flows mostly in one direction and could be explained by the sugar agri-food industry which is composed of few companies with high market share, therefore making these provinces net-suppliers at all times. Results also indicated non-linear performance in this system, with market prices strongly adjusting to differential price variations only after a certain threshold was surpassed.

In conclusion, productivity levels were found to differ according to the region and the type of crops produced. Land and infrastructure investment were found to have a much greater effect on output for traditional crops at the Sierra and Selva regions than for non-traditional crops at the Costa region.

Wholesalers were significantly more responsive to price variations due to exogenous shocks than distributors and farmers.

Trade of potato and rice between pair provinces, with the exception of Lima,flow in a bi-directional manner which suggest that most provinces alter between being net-suppliers and net-consumers of potato and rice during different periods.

Based on the results obtained the following general implications can be formulated. Land and infrastructure investment should be mainly directed toward traditional crop producing provinces. The focus should not only be on introduction of new technologies, but also in the appropriate use of such technologies given the managerial abilities and resources of the farmers. Adequate technical assistance and adaptation of technologies according to the productive environment of traditional crop production should be of special concern.

Adjustment costs at the distribution sector of the agricultural system seem to impede efficient price adjustment; further studies should be made to pinpoint the contributing factors and reduce such asymmetries in price transmission.

Different climate conditions within Peru make certain provinces net-suppliers and net-consumers of potato during different periods. Therefore the government should monitor climate and price changes to create an adequate and efficient harvesting production schedule.

ペルーの農業は、低位にある農業自体の生産性と効率的な市場システムへのアクセスの不足のふたつの課題を抱えている。また、ペルー農業は沿海平野地域(Costa)・高地山岳地域(Sierra)・熱帯雨林地域(Selva)の三つの地域から構成されており、自然条件や生産要素の賦存条件、したがって作物にも顕著な差異がある。本論文は、計量経済学の手法を応用することで・ペルーの農業生産の効率と農産物市場の効率について、統一的な視点から分析を加えたものであり、先行研究のレビューを行った序章、結論と政策的な含意を述べた終章を含む全6章から構成されている。

序章に続く2章では、ペルー農業の歴史的な背景や地理的な特徴を整理したうえで、費用関数の計測結果に基づく総合生産性(TFP)の要因分解を通じて、農業の発展・停滞の定量分析を行っている。計測期間は1970年から2003年までであり、Costa・Sierra・Selvaの3地域の分析に統一的な手法を応用することで、比較可能な結果を導出した点に先行研究にはない特色がある。TFPの要因分解を通じて、3地域のあいだに技術進歩の寄与度と偏向性に著しい違いが確認され、背景として要素の賦存条件の違いに加えて、とくにSierra・Selvaについてインフラの整備水準の遅れや海外からの移転技術の適応不全の問題のあることが示唆された。

3章では、TFPの変化の地域間格差を取りあげ、要因分解で残差とされた部分にさらに立ち入った分析を試みている。計測には主要な投入要素と作物ごとの生産量などに関する州別パネルデータを使用し、対象はデータの制約から1970年から1991年までとされた。Hausman-Taylor推定量などの計測結果から得られたファインディングスを要約するならば、州間格差は管理能力、新技術志向性などの準固定要素の水準と結びついており、とくに馬鈴薯、コーヒー、サトウキビ等の伝統品目を生産する州では、支援策の不足もあって、技術進歩の貢献が低水準にとどまっていた。

4章と5章は、農産物の価格形成と流通システムの効率に着目する。まず4章では、農産物のサプライチェーンを構成する農場・流通業者・卸売のあいだの価格伝達の効率性を定量的に評価している。ペルーの農産物流通の実態を確認したうえで、Hansenらによって開発されたTVECモデルを用いて、価格変動の伝達に関するラグや閾値の検出を行った。使用したデータは農場・運送業者・卸売の主要品目の週単位価格であり、分析期間は2002年4月から08年の3月までとした。各種の検定の結果、農場・卸売間の価格伝達効率が農場・運送業者間のそれを有意に上回ること、運送業者段階の米や馬鈴薯の価格に認められた明瞭な閾値が運送業者の過剰装備・高固定費に起因することなどが明らかにされた。

5章では、農産物の地域間分配の効率性を産地・消費地間の輸送問題として定式化し、4章と同様にTVECモデルを用いた計量分析を試みている。効率性の指標として産地・消費地間の価格裁定と種々の取引費用を念頭に置き、分析には米・馬鈴薯・砂糖の州別の生産量と週単位卸売価格のデータを用いた。対象期間は4章と同様に2002年4月から08年2月までとした。輸送のパターンについて、米と馬鈴薯には生産の季節性に起因する交錯輸送が少なくないこと、首都リマが価格形成上特別なポジションにあることなどを確認するとともに、系統的な検定の結果を踏まえて、いずれの品目についても価格調整の発動に閾値が存在すること、米と馬鈴薯の取引においては、市場間の近接性や迅速な情報伝達によって裁定行動の余地が小さいことなどが明らかにされた。

以上を要するに、本論文はペルーの農業生産と農産物市場の効率性を計量経済学の手法を用いて定量的に評価したものであり、地帯別・州別・品目別のデータを利用することによって・包括的で詳細な分析結果を得るに至っている。申請者の研究は、近年開発された計量経済学的な手法の応用範囲をアグリフードシステムに拡張した点に特色を持つとともに、ペルー農業の今後の発展に向けて、とくに技術進歩の前提条件の観点から少なからぬ示唆を与えている。このように、本論文によって得られた成果は学術上、応用上寄与するところが少なくない。よって、審査委員一同は本論文が博士(農学)の学位論文として価値あるものと認めた。