chakravorty 99 34 paper - emory...

40
Efficiency and Technical Change in the Philippine Rice Sector: A Malmquist Total Factor Productivity Analysis* Chieko Umetsu 1,2 , Thamana Lekprichakul 3 and Ujjayant Chakravorty 4 1 The Graduate School of Science and Technology, Kobe University, Kobe 657-8501, Japan 2 Program on Environment, East-West Center, Honolulu HI 96848, USA 3 Department of Economics, University of Hawaii at Manoa, Honolulu HI 96822, USA 4 Department of Economics, Emory University, Atlanta GA 30322, USA Abstract We account for regional differences in total factor productivity, efficiency and technological change in the Philippine rice sector during the post-Green Revolution era by using Malmquist productivity indices for the period 1971-90. The Malmquist indices were decomposed into efficiency and technological change. The average annual Malmquist productivity growth was found to be only slightly positive. Productivity growth was negative during the early 70’s, and was followed by a period of positive growth, and negative growth in the late 80’s. The period of positive growth coincided with the introduction of new rice varieties while the declines may have been caused by intensification of rice production in lowland systems. Certain regions such as Central Luzon, Western Visayas, South and North Mindanao exhibited higher rates of technological change than others, which seems to have been contributed by higher investments in irrigation, increased adoption of tractors, higher population growth rates and a better agroclimatic environment. JEL classification: O13; O33; Q16 Key words: Malmquist Productivity Index, Technical Change, Philippine Rice Sector * This research was financially supported by the Foundation for Advanced Studies on International Development, Tokyo, Japan. Please address all correspondence to Chieko Umetsu: phone/fax: +81 (78) 803-5840; e-mail: [email protected].

Upload: buikhanh

Post on 12-Jul-2018

214 views

Category:

Documents


0 download

TRANSCRIPT

Efficiency and Technical Change in the Philippine Rice Sector:A Malmquist Total Factor Productivity Analysis*

Chieko Umetsu 1,2, Thamana Lekprichakul 3 and Ujjayant Chakravorty 4

1 The Graduate School of Science and Technology, Kobe University, Kobe 657-8501, Japan2 Program on Environment, East-West Center, Honolulu HI 96848, USA

3 Department of Economics, University of Hawaii at Manoa, Honolulu HI 96822, USA4 Department of Economics, Emory University, Atlanta GA 30322, USA

Abstract

We account for regional differences in total factor productivity, efficiency and technologicalchange in the Philippine rice sector during the post-Green Revolution era by using Malmquistproductivity indices for the period 1971-90. The Malmquist indices were decomposed intoefficiency and technological change. The average annual Malmquist productivity growth wasfound to be only slightly positive. Productivity growth was negative during the early 70’s, andwas followed by a period of positive growth, and negative growth in the late 80’s. The period ofpositive growth coincided with the introduction of new rice varieties while the declines may havebeen caused by intensification of rice production in lowland systems. Certain regions such asCentral Luzon, Western Visayas, South and North Mindanao exhibited higher rates oftechnological change than others, which seems to have been contributed by higher investments inirrigation, increased adoption of tractors, higher population growth rates and a betteragroclimatic environment.

JEL classification: O13; O33; Q16

Key words: Malmquist Productivity Index, Technical Change, Philippine Rice Sector

* This research was financially supported by the Foundation for Advanced Studies onInternational Development, Tokyo, Japan. Please address all correspondence to Chieko Umetsu:phone/fax: +81 (78) 803-5840; e-mail: [email protected].

1

Efficiency and Technical Change in the Philippine Rice Sector:A Malmquist Total Factor Productivity Analysis

1. Introduction

Agricultural intensification has long been considered the primary means by which

governments and international development agencies induce technological change in developing

countries characterized by high population pressure and low agricultural productivity. However,

some policy advocates now argue that contrary to the Boserup hypothesis which suggests that

population pressure is a sufficient condition for inducing technological change and productivity,

governments need to play a pro-active role in agricultural development by investing in rural

infrastructure and promoting higher input use through incentives such as input subsidies (Lele

and Stone, 1989).

However, the process of agricultural intensification itself may lead to productivity

declines. Recent observations of agricultural productivity in the post-Green Revolution era have

concluded that intensive monoculture production systems have contributed to declining or

stagnant productivity and farm incomes, as well as related environmental impacts such as

increased pest pressure, depletion of soil micronutrients, and changes in soil chemistry caused by

intensive cropping (Bouis, 1993; Pingali et al., 1995).

The purpose of this paper is to analyze trends in productivity growth in the Philippines

rice sector and examine the factors affecting productivity performance by region. We seek to

verify whether agricultural intensification in the Philippines was accompanied by technological

change and productivity growth during the post-Green Revolution era. This is done by applying

an input-oriented Malmquist Productivity Index. Traditionally, technical change has been

modeled in neoclassical growth models with the implicit assumption that a common production

function is available to all countries regardless of human capital, resource or institutional

2

endowment (Ruttan, 1995). This tradition is followed by productivity studies which incorporate

and measure the contribution of technology, human capital and other non-conventional factors of

production using specific functional forms for the production function. These studies usually

analyze national aggregate data to compare productivity among countries (Hayami and Ruttan,

1970, 1985; Kawagoe, Hayami and Ruttan, 1985; Lau and Yotopoulos, 1989). However, when

regional differences in factor endowment and technology are significant, cross-country

comparisons do not provide policy insights for regional development since they do not

incorporate location specific factors of production and technical change.

Other methods of productivity measurement such as growth accounting and index

number approaches also have some limitations. First, they assume that production is efficient by

theoretical construct. Inefficient production activity is largely ignored. Second, price data

requirements make it difficult to apply this method in developing countries where reliable price

information is often not available. Recent developments in nonparametric frontier approaches

provides more flexibility to productivity and technical change analysis. The Malmquist total

factor productivity (TFP) index was suggested by Färe, Grosskopf, Lindgren and Ross (1989)

following the work of Caves, Christensen, and Diewert (1982a). This index is based on the

Farrell measure of technical efficiency (Farrell, 1957) and Shephard's (1953) distance function.

One of the advantages of the Malmquist index is that it requires only quantity data, and it is not

constrained by a specific functional form of the production or cost function.1 The Malmquist

productivity index has recently been applied to cross-country comparisons of total factor

productivity (Thirtle, Hadley and Townsend, 1994; Färe et al., 1994b; Fulginiti and Perrin,

1997).

1 Diamond et al. (1978) assert that the traditional parametric approach to analyzing technology and technical changemay be sensitive to the particular parametric specification utilized.

3

Earlier studies on productivity and technical change in the Philippine agricultural sector

have relied on index number approaches for estimating total factor productivity (David, Barker

and Palacpac, 1985). While their analyses was based on aggregate data for the agricultural sector

of the Philippines, Antonio, Evenson and Sardido (1977) and Evenson and Sardido (1986)

looked at total factor productivity at the regional level. These studies assume that there is no

inefficiency in production. Färe, Grabowski and Grosskopf (1985) estimated the Farrell measure

of technical efficiency of the Philippine agricultural sector to account for production

inefficiency. However, their study is based on aggregate time series data used in David and

Barker (1979) and hence is restricted to national level data.

We estimate the input-oriented Malmquist total factor productivity index of the rice

sector based on panel data for 1971-90 for 12 regions of the Philippines. The method does not

impose specific functional forms on production technology. Results show that the rice sector in

some regions of the Philippines experienced negative long-run growth in total factor productivity

between 1971 and 1990, primarily due to input-biased intensification. In particular, productivity

growth was negative during the early 70’s, the second half of the Green Revolution era, followed

by positive growth in the late 70’s and early 80’s, leading to another negative trend in the late

80’s. The positive growth coincided with the introduction of new rice varieties, and the negative

growth suggests that rice production may have entered a phase where growth was primarily

achieved through input intensification. These findings confirm the results of Pingali (1992) and

others who suggest that intensive production systems may be causing productivity declines and

degradation of lowland rice environments in the Philippines and other agriculture-based

economies in Asia today. At the regional level, our analysis suggests that certain regions such as

Central Luzon, West Visayas and South and North Mindanao have performed markedly better

4

than others. Several factors contributed to this phenomena, including irrigation investments and

tractor use, as well as higher rates of population growth and a favorable agroclimatic

environment, although no single factor was found to be singularly important across regions.

The organization of the study is as follows. Section 2 briefly reviews trends in

agricultural development and factor/resource endowment for each region. Section 3 discusses the

theoretical construct and the method for estimating the Malmquist productivity index and its

decomposition into efficiency and technical change. Section 4 describes the panel data set used

in this study. Section 5 presents the results of the regional Malmquist index and the second stage

analysis. Section 6 concludes the paper.

2. Overview of the Rice Sector and Factor/Resource Endowment in the Philippines

The total rice production in the Philippines more than doubled during the last 30 years

(IRRI, 1995; see Appendix Table 1). The introduction of high yielding varieties during the late

1960’s started the era of the "Green Revolution" in the Philippine rice sector. Otsuka, Gascon

and Asano (1994) divide the period for adopting modern high-yielding varieties into a first and

second generation. The first-generation was initiated in 1966 by the release of IR8 followed by

IR5 through IR34 and C4. These varieties were not resistant to pests and diseases. The second

generation started in 1976 when IR36 was introduced. The land planted by modern varieties

increased from 58% to 86% between 1971 and 1990 (Table 1). By 1979, more than 90 percent of

farmers in Central Luzon, the "rice bowl" of the Philippines, adopted second-generation varieties

(IR36-IR76) which almost doubled the rice yield compared to the traditional varieties. This trend

is indicated in Table 1.

Between 1971 and 1990, the average Philippine palay (rice) yield for high yielding

5

varieties increased from 1.74 to 2.69 metric tons per hectare. The average yield surpassed 3

metric tons per hectare in Cagayan Valley, Central Luzon, and Mindanao regions although not in

Western Mindanao. Southern Mindanao reached the highest yield per hectare in the late 80's

because of favorable soil conditions and a high adoption rate of modern varieties in the early

70's. High yielding varieties contributed to a significant increase in yield, with modern varieties

producing a third more on average than traditional varieties.

Although the average yield per hectare was increasing during 1971-90, the growth of rice

yields stagnated during the late 80’s. Average annual growth of rice yields was 1.6% in the early

70’s and jumped to 4.2% in the late 70’s when farmers switched to second-generation modern

varieties (PhilRice-BAS, 1994). In the early 80’s the growth rate of yields decreased sharply to

2.9% and then to 1.4%. Recently, many studies reported a declining trend of yield growth in Asia

during the post-Green Revolution era since the late 70’s:

"An assessment of farm-level data over time, ..... , shows that yield levels are beingmaintained with increasingly higher input levels, indicating a long-term decline in total factorproductivity." (Pingali, Zeigler, Hossain and Prot, 1995)

In addition, the macroeconomic conditions during the 80’s were not favorable to rice

farming. Volatile political conditions, high inflation and a series of devaluations adversely

affected rice farming; and the Philippines2, which once reached its self-sufficiency goal in 1978,

again turned into a net importer of rice in 1984. A severe debt crisis during the 80’s discouraged

production-oriented government investment projects such as research, irrigation, and credit

(Evenson and David, 1993).

The high yielding varieties required more labor for weeding, as well as increased

2 For a detailed discussion regarding macroeconomic impacts on environment in the Philippines, see Cruz andRepetto (1992), and Montes and Lim (1996).

6

fertilizer inputs and controlled water supply relative to the traditional varieties. The scheduled

irrigated water supply increased the yield with the adoption of high yielding varieties during the

dry season because traditional varieties are photo-period sensitive and not suitable for dry season

production (Otsuka et al., 1994). In areas where irrigated water and favorable rainfall were

available, farmers responded quickly to adopt high yielding varieties. Also, high yielding

varieties increased crop intensity due to a shorter growth duration compared to traditional

varieties.3 ,4 Otsuka (1991) points out that land reform was successfully implemented in areas

where the potential benefit from adopting high yielding varieties was large. These characteristics

of high yielding varieties induced higher input use in the rice producing sector during 1971-90.

Table 2 shows the ratios of factor use by region. Labor use per hectare was high during

the Green Revolution era (1971-1975) and decreased gradually during the post Green Revolution

era mainly due to the introduction of tractors. Central Visayas and Eastern Visayas had high

labor use per hectare reflecting slow adoption of labor saving technologies5 such as the hand

tractor. Fertilizer use per hectare, on the other hand, increased on average more than 1.5 times in

the Philippines during 1971-90, with the highest being in Central Luzon. The Philippine

government implemented various kinds of subsidies to increase fertilizer use. When the Fertilizer

Industry Authority (FIA) was established in 1973, fertilizer prices for priority crops (rice, corn,

feedgrains, and vegetables) were controlled by the government and set at 50-70% lower than

other crops (Francia, 1993). Direct cash subsidies were given to fertilizer companies between

1973 and 1982. In addition to direct subsidies, indirect subsidies in the form of tax exemptions

were given to fertilizer imports. In 1988, when the government launched the Rice Production

3 A crop intensity of more than one shows the existence of double or triple cropping.4 The growth duration of traditional varieties, first-generation varieties, and second-generation varieties, are 155days, 130 days and 115 days, respectively (Estudillo, 1995).

7

Enhancement Program (RPEP), fertilizers were again subsidized for farmers.

Labor saving technology such as tractors also increased during the study period showing

large regional differences in tractor use per hectare. Tractor use was particularly high in Central

Luzon, and lowest in Central Visayas. On the other hand, the carabao (water buffalo used for

animal draft) population did not decrease significantly. This may reflect the fact that during the

oil crisis, imported machinery became expensive and farmers substituted carabao for tractors

(Hayami et al., 1990).

Investment in irrigation expanded rapidly from 81 million pesos (US$12.6 million) in

1971 to 2,481 million pesos (US$88.6 million) in 1990 (IRRI, 1995). As a result, the percentage

of irrigated area out of planted area to rice increased from 37% to 53% in the Philippines as a

whole, and Cagayan Valley, followed by Central Luzon, had the highest proportion of irrigated

rice (see Appendix Table 2). The rapid growth of irrigation investment occurred only during the

70’s and the investment during the 80’s stagnated (IRRI, 1995). While irrigated land increased,

total rice area declined since 1970, which suggests that rice production intensified in the study

period. The quality of road infrastructure is high in Ilocos, Central Luzon, Southern Tagalog,

Bicol, and poor in all Mindanao regions.

Population growth in the 12 regions (not including the National Capital Region (NCR),

Cordillera Autonomous Region (CAR), and Autonomous Region in Muslim Mindanao (ARMM)

showed a declining trend during 1971-90. Population density was high in Luzon Island with

intense population pressure on agricultural lands. Ilocos, Central Luzon and Southern Tagalog

showed a population density of more than 10 persons per hectare of arable land and permanent

cropland during the study period. The growth of population density was high in Central Luzon

5 Other labor saving technologies include direct-seeding, herbicides, threshers which require less labor for plantestablishment and weeding (Otsuka, Gascon, and Asano, 1994).

8

and Western Visayas during the post Green Revolution period and declined sharply during the

1980’s.

Distribution of production to landlords in the region may be considered a proxy for the

degree of implementation of land reform policies. Roumasset and James (1979) argue that

regional differences in output share of landlords could be explained by land quality and

population. If land quality and population density are high, both will positively affect the

landlord share since a high physiological density will lead to low farm wages. In reality, the

landlord share of output has decreased during the study period in most of the regions, especially

in Cagayan Valley, Central Luzon and Southern Tagalog, although both population density and

population growth were relatively high in these regions6 (Appendix Table 3, 4). This is possibly

due to a combination of the following factors: land reform policies which reduced the share of

production accruing to landlords; land-saving technological change such as fertilizers which

decreased profits accruing to land (Roumasset and James, 1979); and finally, competition

between agriculture and other sectors of the economy for labor that did not depress farm wages

inspite of high population densities. On the other hand, Ilocos is the only region where

production share of the landlord was increasing over time and the area under share contracts was

the highest among all the regions. Total area of palay farms under share contract was 42.1% in

Ilocos, substantially higher than the 23.3% in Cagayan Valley and 17.4% in Central Luzon.

Agroclimatic conditions were favorable in Mindanao which is endowed with fertile soil.

According to Department of Agriculture (1993) estimates, average maximum potential palay

yield in Mindanao was over 6,139 metric tons per hectare (MT/ha), which was far higher than

6 The landlord share data in the Regional Rice Statistics Handbook may underestimate the actual shares whichusually fall between 30 and 50 per cent, according to farm-level studies (Roumasset, 1984).

9

that of Luzon (4,477 MT/ha) and of Visayas (4,679 MT/ha). On Luzon Island7, palay production

has already been quite intensive and there was not much potential for a major increase in yields.

Visayas and Mindanao, the southern part of the Philippines, have relatively less rainfall

compared to Luzon. Coastal regions which face the Pacific Ocean usually have a wet season8 of

more than nine months of the year. Eastern and Central Visayas, and Northern Mindanao are

especially disaster-prone areas. When the 1983 drought hit the Island, total rice area harvested in

Central Visayas declined by 43% compared to the previous year (PhilRice-BAS, 1994).

During 1970-89, soil erosion from the agricultural sector accounted for 22.4% of total

soil erosion in the Philippines (IRG, 1992). Soil erosion from agriculture was highest in Southern

Tagalog, 69 million metric tons (MMT). This was followed by Southern Mindanao (58 MMT)

and Bicol (51 MMT) where agricultural production was intensive. Among alternative land use

patterns, upland agriculture has significant effects on soil erosion compared to irrigated lowland

rice paddies (IRG, 1994).

In spite of the "Green Revolution" which brought high yielding varieties to the rice sector

of the Philippines, income of agricultural households did not increase relative to non-agricultural

households. A regional polarization of farmers became apparent after the Green Revolution. In

1971, regions where agricultural household income was lower than the Philippine average were

distributed throughout the country. In 1991, however, regions with above the national average

income can be found only in Luzon and ARMM. The introduction of high yielding varieties

contributed to an increasing income level of rice producing regions, especially on Luzon Island.

The income distribution within the rice sector has been found to be skewed in favor of large

farmers after the introduction of high yielding varieties (David et al., 1994). David argues,

7 In 1992, Luzon Island alone produced 63% of the total rice production in the Philippines.8 Rainfall less than 100 mm per month is a dry month and more than 200 mm per month is a wet month.

10

however, that the high yielding varieties contributed significantly to increasing income for poor

farmers as well as landless farmers.

3. Malmquist Total Factor Productivity Index

3.1 Farrell measure of technical efficiency and the distance function

In order to account for the regional total factor productivity, efficiency and technology

change in the Philippine rice sector, we apply the nonparametric method for estimating those

indices.

Farrell (1957) suggested the measurement of technical efficiency using piecewise linear

technology. Linear programming constructs a “best practice” frontier technology. Farrell’s

efficiency measure is the inverse of Shephard’s (1953) distance function, which provides the

theoretical base for the Malmquist productivity index.

The production technology is represented as the set of all feasible input and output

vectors for time period t. Let x x x xt t tNt= ( , ,..., )1 2 denote an input vector at period t with i=1,..,N

inputs and y y y yt t tMt= ( , ,..., )1 2 denote an output vector at period t with j=1,..,M outputs where

x t N∈ ℜ + , and y t M∈ ℜ + . The technology is expressed by the input requirement set, as follows:

L y x x y S t ,..., Tt t t t t t( ) { :( , ) },= ∈ = .1 (1)

where S x y x yt t t t t= {( , ): }can produce is the set of technology at period t. The input

requirement set, L yt t( ) provides all the feasible input vectors, x t N∈ ℜ + , that can produce the

output vector, y t M∈ ℜ + .

The Farrell measure is the radial measure of technical efficiency in which the efficiency

is obtained by radially reducing the level of inputs relative to the frontier technology holding the

11

level of output constant. The Farrell measure requires input and output quantity information and

is independent of input prices as well as behavioral assumptions on producers. Similarly, the

output-oriented Farrell measure can be defined by radially expanding the level of outputs relative

to the frontier technology holding the level of input constant. Figure 1 illustrates the input-

oriented Farrell measure and distance function for a two-input case. The frontier technology is

given by the piecewise linear isoquant, L yt t( ) . Efficient production activity is the extreme point

of the convex hull of this frontier (B and C). Line segments extending from B and C, AB and

CD, indicate strong disposability of inputs i.e., disposal of surplus inputs is free. Production

activity c is inside of the input requirement set, thus inefficient. In terms of distance, the Farrell

measure of technical efficiency at period t is given by 0b/0c and the Shephard’s distance function

is the inverse 0c/0b. When the observation is efficient, both the Farrell measure and the distance

function are equal to one. The Farrell measure varies between zero and one, and the distance

function is equal to or greater than one.

Suppose there are k = 1,...,Kt number of firms which produce M outputs mk,ty , m = 1,...,M,

using N inputs nk,tx , n = 1,...,N, at each time period t = 1,...,T. A piecewise linear input

requirement set at period t is defined as follows:

L y x y z y m ,..., M

x z x n ,..., N

z k ,..., K

t t tmk t k t

mk t

k

K

nk t k t

nk t

k

K

k,t

( ) { : ,

,

,

', , ,

', , ,

= ≤ =

≥ =

≥ =

=

=

,

,

},

1

1

0 1

1

1(2)

where z k t, indicates intensity levels, which makes the activity of each observation expand or

contract to construct piecewise linear technology (Färe et al., 1994a). Let us define F y xit t t( , ) as

the input-oriented Farrell measure, and D y xit t t( , ) as Shephard’s input-oriented distance

12

function at period t with constant returns to scale and strong disposability of inputs and outputs

assumption as:

F y x x L y

D y x x L yit t t t t t

it t t t t t

( , ) min{ ( )},

( , ) max{ : ( / ) ( )}.

= ∈= ∈

λ λλ λ :

(3)

where F y xit t t( , ) estimates the minimum possible expansion of x t while D y xi

t t t( , ) estimates

the maximum possible contraction of x t .

The following shows the linear programming problem for (2) for estimating the Farrell

measure, or the inverse of the distance function.

F y xit k t k t( , )', ', = [ D y xi

t k t k t( , )', ' , ]− 1 = min λ, (4)

subject to

,

,

.

y z y m ,..., M

x z x n ,..., N

z k ,..., K

mk t k t

mk t

k

K

nk t k t

nk t

k

K

k,t

', , ,

', , ,

,

,

,

≤ =

≥ =

≥ =

=

=

1

1

0 1

1

1λ (5)

In equation (5), the left hand side of the input and output inequality shows the analysis

set, the observation to be evaluated, and the right hand side shows the reference set. This linear

program evaluates observations at period t relative to the reference (frontier) technology at

period t. In general, both Farrell measure and distance function can be defined with any type of

returns to scale assumptions such as non-increasing returns to scale as well as variable returns to

scale which includes constant, decreasing and increasing returns to scale. By controlling the

intensity variables with additional constraints, z k t

k

K, =

=∑ 1

1

, and z k t

k

K, ≤

=∑ 1

1

in the linear program,

variable returns to scale and non-increasing returns to scale can be imposed (Afriat, 1972). Also,

the strong disposability assumption can be relaxed (Grosskopf, 1986).

13

In order to estimate the Malmquist productivity index from period t to t+1, additional

distance functions are required as follows:

D y x x L yit t t t t t( , ) max{ : ( / ) ( )},+ + + += ∈1 1 1 1λ λ (6)

D y x x L yit t t t t t+ += ∈1 1( , ) max{ : ( / ) ( )},λ λ (7)

and

D y x x L yit t t t t t+ + + + + += ∈1 1 1 1 1 1( , ) max{ : ( / ) ( )}.λ λ (8)

The cross-period distance function, D y xit t t( , )+ +1 1 , indicates the efficiency measure using the

observation at period t+1 relative to the frontier technology at period t, and D y xit t t+ 1( , ) shows

the efficiency measure using the observation at period t relative to the frontier technology at

period t+1. In Figure 1, the input requirement set for period t+1 is illustrated by L yt t+ +1 1( ) , and

D y xit t t( , )+ +1 1 and D y xi

t t t+ 1( , ) are given by 0e/0f and 0c/0a respectively. Cross period distance

functions take value less than, equal to, or more than one. Similarly, D y xit t t+ + +1 1 1( , ) is given by

0e/0d. A linear programming problems of the equation (6), (7), and (8) are similar to equation (5)

once the respective analysis set (observation) and the reference set (frontier technology) are

defined.

3.2 The Malmquist productivity index to measure total factor productivity

The Malmquist productivity index (Färe, Grosskopf, Kindgren and Roos, 1989) is the

geometric mean of two Malmquist indices which were suggested by Caves, Christensen, and

Diewert (1982a). The input-oriented Malmquist productivity index consists of four input-

oriented distance functions. The change of productivity between period t and t+1 is defined as:

14

M y x y xD y x

D y xD y x

D y xit t t t t i

t t t

it t t

it t t

it t t

+ + ++ + + + +

+=

1 1 1

1 1 1 1 1

1

12

( , , , )( , )

( , )( , )

( , )(9)

where D y xit t t+ 1( , ) and D y xi

t t t( , )+ +1 1 are cross-period distance functions.

The Malmquist productivity index can be decomposed into changes in efficiency and

changes in technology as:

M y x y xD y x

D y xD y x

D y xD y x

D y xit t t t t i

t t t

it t t

it t t

it t t

it t t

it t t

+ + ++ + + + +

+ + + += ⋅

1 1 1

1 1 1 1 1

1 1 1 1

12

( , , , )( , )( , )

( , )( , )

( , )( , )

(10)

where the first term defines changes in efficiency from period t and t+1. The second geometric

mean in the bracket indicates changes in technology, i.e., a shift in the frontier from period t to

period t+1. This decomposition provides useful indices for the study of efficiency and technical

change. In the input-oriented case, all three terms, i.e., the change in productivity, and its

decomposition to the change in efficiency and the change in technology, are interpreted as

progress, no change, and regress, when their values are less than one, equal to one, and greater

than one, respectively.

4. Data Set

The time series data set for 1971-90 for 12 regions of the Philippines rice sector was

constructed for estimating regional input-oriented Malmquist productivity indices. The twelve

regions are: Ilocos (Region 1), Cagayan Valley (Region 2), Central Luzon (Region 3), Southern

Tagalog (Region 4), Bicol (Region 5), Western Visayas (Region 6), Central Visayas (Region 7),

Eastern Visayas (Region 8), Western Mindanao (Region 9), Northeastern Mindanao (Region 10),

Southeastern Mindanao (Region 11), and Central Mindanao (Region 12). The Cordillera

Autonomous Region (CAR) was not considered as an independent region due to a lack of time

15

series data before the establishment of CAR in 1988. Therefore, data for Region 1 and Region 2

after 1988 does not include the provinces which were reorganized under CAR. The National

Capital Region (NCA) and the Autonomous Region in Muslim Mindanao (ARMM) were

ignored simply due to their small agricultural sectors.

Two outputs and six inputs were used for estimating the input-oriented Malmquist index.

Due to a lack of regional input data for the rice sector, some data was generated by separating it

from aggregate agricultural input data using distribution parameters from the actual regional

data. Outputs consist of total annual production of high yielding varieties (HYV) and traditional

varieties (TV) by region (PhilRice-BAS, 1994).

Inputs are classified into traditional and modern inputs. Traditional inputs are land

harvested of HYV, land harvested of TV (thousand hectares per year), labor (total man-days per

year), and work animals (carabao head per year). The labor inputs were estimated from the total

cost data per hectare for rice production, i.e., cash cost (hired labor), non-cash cost (hired labor

in kind), and imputed cost (unpaid family and operation and exchange labor) between 1985-90.

Work carabaos by region were estimated from various unpublished data sets from the Bureau of

Agricultural Statistics.

Modern inputs such as fertilizer and machinery are considered to embody technology.

Fertilizer use data by grades was converted into actual nutrient sums of nitrogen, phosphorus and

potash ([N+P2O3+K2O] kg per hectare per year). The number of hand tractors used per year were

considered as machinery inputs. This underestimates the number of four-wheel tractors used in

rice production. However, as Otsuka et al. (1994) describe, four-wheel tractors were largely

replaced by two-wheel tractors by the end of the 70’s. Therefore, hand tractors may be a good

approximation of tractor use since separate data for four-wheel tractor use in rice production was

16

not available.

For estimating the Philippine average for the Malmquist index and its components for

each period, a simple geometric mean of each region discriminates against productivity indices

for relatively large rice producing regions. Bjurek and Hjalmarsson (1995) adopted input-shares

as a weight for estimating industry wide input-oriented efficiency. Due to multiple inputs, in this

paper we used an average output-share, i.e., a share of total palay production in each region, to

obtain a weighted average index for the Philippines. Regional Farrell measures were aggregated

using output-shares for period t for efficiency scores for single-period Farrell measures and the

average output-share of period t and t+1 for mixed-period Farrell measures to derive a

"structural" distance function for the Philippines.

For the 2nd-stage analysis, the following explanatory variables are considered.

Infrastructure variables include irrigation and transport. Rural infrastructure such as irrigation

and roads have a significant impact on technological change in the region (Binswanger et al.,

1993). Irrigation infrastructure is measured in terms of the ratio of irrigated area planted to rice.

The percentage of paved road per total length of road was used as a proxy for transport, which

indicates road quality.

According to the Boserup hypothesis (1965), population pressure induces technical

change and intensification of agriculture (Pingali et al., 1987; Lele and Stone, 1989; Thirtle et al.,

1994). However, if a Boserupean transformation is not possible, population pressure is expected

to cause detrimental effects on production (James and Roumasset, 1992). Regional population

pressure is expressed by population per arable and permanent crop land.9 The economically

active population in agriculture is a suitable estimate of population pressure; however, reliable

17

regional time series data was not available.

Technology variables include education and a dummy variable for the second-generation

modern rice variety. Schultz (1964) states that "the acquired capabilities of farm people are of

primary importance in modernizing agriculture." Higher education enrollment per total

population was used as a proxy for the education variable. According to Otsuka's (1994)

classification of modern varieties, the switching from first-generation to second-generation

varieties occurred during 1976-79. Therefore, the dummies for detecting the effects of second-

generation varieties on productivity change are zero for 1971-77 and one for 1978-80. The share

tenancy ratio, as well as land holdings are strong candidates as explanatory variables in

explaining owner operators’ and large farmers' incentive to adopt technology.10 However, since

the land holding data was available for census years only, we used the landlord share of

production from tenants (PhilRice-BAS, 1994). Exogenous variables include rainfall and

disasters. The ratio of monthly average rainfall to the average of the highest three months of

rainfall was used as suggested by Nugent and Sanchez (1995). Rice area damaged by typhoon

and other causes was used as a proxy for disasters.

Reliable regional input price data was most difficult to obtain except for wage and

fertilizer prices. In order to obtain the effect of input price ratios, i.e., fertilizer to land price, and

labor to machinery price, land price was approximated by the value of production distribution to

the landlord and the same machinery price index was used for all regions.

5. Regional Malmquist Productivity Index in the Philippine Rice Sector

9 Binswanger and Pingali (1988) suggest using agroclimatic population density based on production potential ofeach country rather than using population per unit area. This method is useful for adjusting regional differences inland quality. However, due to limited data availability, it was not used in this analysis.

18

5.1 Mamlquist indices during 1971-1990

Table 3 provides a summary of the input-oriented Malmquist total factor productivity

indices of the Philippine rice sector, and their decomposition into efficiency change and

technological change. For these input-oriented measures, an index of less than one represents

progress, and an index of more than one represents regress. However, for an easy interpretation,

the numbers in these tables are the reciprocals of the real indices multiplied by 100 so that an

index of more than 100 indicates progress, an index of less than 100 indicates regress, and an

index equal to 100 shows no change. The numbers in Table 3 show annual averages of five and

twenty-year intervals between 1971 and 1990.

Two indices for the Philippines are the weighted arithmetic mean (WAM) and the

geometric mean (GM) for all 12 regions. During 1971-90, the weighted average Malmquist

productivity growth for the Philippine rice sector was 0.7%, indicating slightly positive growth.

This is similar to earlier results by Evenson and Sardido (1986) who reported a 0.21% average

increase in total factor productivity of the Philippine agricultural sector between 1975-84. On the

other hand, the geometric mean showed negative growth of –0.6%. However, the simple

geometric mean tends to underestimate the productivity change in relatively large rice regions

such as Central Luzon which has been at the production frontier, i.e., the best rice technology,

throughout the study period. We therefore take the weighted mean to be a better representation of

productivity change for the Philippine rice sector.

Productivity growth was found to be negative in the in the early 70’s (–2.0%), the second

half of the Green Revolution era, followed by positive growth during the late 70’s (2.4%), and

the early 80’s (3.6%), and again negative growth in the late 80’s (–1.8%). The positive TFP

10 Ruttan (1977) generalized the adoption of HYV and stated that neither farm size nor tenure constrained theadoption of HYV.

19

growth coincides with a period during the late 70’s when IR36 and other second-generation

modern varieties were introduced and rapidly adopted by farmers. These results are supported by

the findings of Evenson and Sardido (1986) who show that the highest total factor productivity

growth of agriculture occurred between 1975-84. The growth of total factor productivity is

mostly attributable to technological change (0.7%) during the study period.

Figure 2 illustrates the trend of efficiency, technological change, and Malmquist

productivity indices (WAM) between 1971-90, using 1971 as a base year.11 The technological

change component, i.e., a shift of the frontier technology, displays similar movement to the

Malmquist productivity index, indicating that a change in total factor productivity largely

consists of technological change in the Philippines rice sector during this period. Two oil price

shocks in 1973 and 79 contributed to negative total factor productivity by increasing the import

price of fertilizer, and as a result, domestic fertilizer sales decreased drastically (PhilRice-BAS,

1994). Compared to technological change, the change in efficiency was quite small, and was not

a major source of productivity growth over the twenty years studied.

At the regional level, the average annual growth of the Malmquist index was positive

only in five regions during 1971-90 (Table 3). These regions are Central Luzon, Bicol, Western

Visayas, Northern Mindanao, and Southern Mindanao. Except for Bicol, these regions are

characterized by high rice yields per hectare, high adoption rates of HYV, and high fertilizer and

tractor use per hectare. The annual average total factor productivity growth is highest in Central

Luzon (8.2%) and the other four regions exhibited modest growth of less than 2%. On the other

hand, seven regions resulted in negative annual productivity growth, with the lowest growth in

Central Mindanao (–7.3%) followed by Central Visayas (–4.8%).

Regional differences in Malmquist productivity are also largely due to regional

20

differences in technological progress. Figure 3, 4 and 5 show the technological change index in

Luzon, Visayas, and Mindanao Island, respectively, with 1971 as the base year. Within Luzon

Island, Central Luzon showed much higher technological progress compared to the other four

regions, with more than 2.5 times the relative shift of the production frontier (Figure 3). On

Visayan Island, Western Visayas exhibited technological progress while the other two regions,

Eastern and Central Visayas, lagged behind, and are among the poorest agricultural areas in the

Philippines (Figure 4). In 1991, average income for agricultural households in Central Visayas

was 27,634 pesos/year and in Eastern Visayas was 29,349 pesos/year. These income levels are

the lowest of agricultural households in all regions (NSO, 1991). On Mindanao Island, Northern

and Southern Mindanao made good technological progress, although Western and Central

Mindanao showed negative progress in the study period (Figure 5).

5.2 Factors affecting changes in productivity, efficiency, and technology

The second stage regression analysis attempts to identify factors affecting the Malmquist

TFP as well as efficiency and technical change indices. Three regional blocks, the Island of

Luzon, Visayas and Mindanao, as well as the Philippines as a whole are considered. Luzon

Island is characterized by good infrastructure and high population density. Visayan Islands, on

the other hand, have relatively lower quality infrastructure but favorable rain fall in the coastal

areas. Mindanao Island has low population density and high infrastructure with relatively good

soil quality and a high production potential.

Table 4 presents the results of regression analysis of the Malmquist TFP index and its

components of efficiency and technological change. The Malmquist index, efficiency and

technological change components for 12 regions are grouped into 3 blocks and regressed against

11 The index starts with 1972 because the index in 1972 requires data from 1971 and 1972.

21

the explanatory variables mentioned above. In each regional block model, heteroskedasticity and

autocorrelation were adjusted for when detected. For the Philippine model, panel data was used

for estimating groupwise regression models.

Among the infrastructure variables, coefficients for irrigation are significant in enhancing

TFP, efficiency and technological change in Luzon and have a negative effect in Visayas.

However, the overall effect of irrigation on TFP in Mindanao and the Philippines was not

significant. Since the adoption of modern varieties largely depends on the availability of irrigated

water, this result is counter intuitive. However, Bouis (1993) also reported that both irrigation

and fertilizer contributed only a small portion of yield increases due to stagnation in the growth

of the irrigated area during the 80's. For transportation, on the other hand, coefficients were

significant and positive in all three indices in Luzon and the Philippines and negative for

technological change in the Philippines.

The parameter for population per arable land can show whether the Boserup hypothesis,

i.e., population induced technological change, can be supported by this analysis. The results,

however, are mixed. Population pressure affected TFP positively in the Philippines and in

Visayas but negatively in Mindanao. On the other hand, population pressure affects

technological change positively in Mindanao where population growth was most rapid during

this period. One reason may be that the favorable agroclimatic condition coupled with the

investment in irrigation and high adoption rate of modern varieties in Mindanao contributed to a

shifting out of the production function. Furthermore, population pressure negatively affected

efficiency in Mindanao, Luzon and the Philippines.

The level of education positively affected TFP, efficiency and technological change in

the Philippines. However, some parameters turned out negative, results for which may be

22

difficult to explain, and may be due to regional variations in higher education enrollment. The

dummy variable for modern variety II (second-generation varieties) was not significant in

explaining TFP change in the Philippines. Moreover, the efficiency change parameter is negative

in Mindanao, which could be because of input-biased technological change during the study

period. When new technologies are introduced, it is possible that inefficient production occurs

because of unfamiliarity with new technology (Arnade, 1994). Distribution of production share

to landlords mostly had a negative effect showing that higher production shares accruing to

landlords has a negative effect on TFP, efficiency and technological change.

The Hayami-Ruttan hypothesis of induced innovation can be considered a two-stage

hypothesis. In the first stage, relative scarcity of resources induces a relative increase in the

factor price of scarce resources. In the second stage, a change in relative factor prices induces

technology to save relatively costlier factors of production. Recently, Olmstead and Rhode

(1993) challenged the plausibility of the first-stage indicating that in the U. S., a change in

relative factor prices did not follow a change in relative scarcity of factors at a regional level.

The purpose of the regression against factor price ratios is to test whether changes in relative

price ratios of inputs positively affected technological change, i.e. a shift in the "best practice

frontier", which represents the second half of the Hayami-Ruttan hypothesis. Following the

Olmstead and Rhode (1993) critique that the relative scarcity of factors does not represent

relative factor prices at regional levels, a factor price ratio instead of a factor endowment ratio

was considered as an independent variable.

All factor price ratios, i.e., an increase in land prices relative to fertilizer prices, wages

relative to machinery, and land relative to wages, have significant impacts on increasing

efficiency. In Visayas and the Philippines, however, a land price increase relative to fertilizer

23

negatively affected technological change. This technological regress through a relative decrease

in fertilizer price may be partly contributing to stagnant technological progress in the Philippines

through excessive use of fertilizers caused by its lower relative price leading to a worsening

input-output combination and a backward shift of the production frontier. However, it increased

efficiency and the overall effect of an increase in the land/fertilizer price ratio on TFP was

positive. A wage increase relative to the price of machinery as well as land prices relative to

wages positively affected efficiency and technological change. In particular, all coefficients for

the labor/machinery price ratio on TFP, efficiency and technological change were highly

significant in Central Luzon where the introduction of hand tractors, a labor-saving technology,

was the fastest (Table 2).

Factor intensity variables can be used to test the Boserup hypothesis which postulates that

agricultural intensification leads to technological change. The overall impact of fertilizer

intensity on TFP was negative in the Philippines although fertilizer intensity contributed to

technological change to some extent. These results are somewhat surprising because it is

believed that the level of fertilizer use in many developing countries, the representative mode of

intensification, is still not enough to accelerate agricultural productivity (Lele and Stone, 1989).

Pingali (1992) asserts that the long-term stagnation in yields potential under intensive irrigated

rice production can be attributed to degradation of the paddy environment due to production

intensification and current yield gains can be sustained only with increasing levels of chemical

fertilizers. On the other hand, in Mindanao where intensification positively affected TFP growth,

there is scope for further intensive use of fertilizer to increase productivity.

Tractor intensity, on the other hand, needs a somewhat different interpretation. Except for

Visayas, none of the parameters were significant to increase TFP. Intensity of tractor use, i.e.,

24

hand tractor use per land harvested, decreased technological change. These results may be

explained by the fact that hand tractor use as a labor-saving technology was not saving labor fast

enough to produce positive TFP growth. This result is in contrast to the wage/machinery price

ratio which significantly increased TFP growth in Luzon.

The impact of weather (rainfall) on TFP was not significant while rainfall negatively

affected efficiency and technological change in the Philippines. Disasters did not significantly

affect TFP, although they negatively affected TFP in Luzon. In disaster-prone Visayas, the

weather variable was a significant negative effect on efficiency and technology.

6. Summary and Conclusions

During 1971-90, the average annual Malmquist productivity growth of the Philippine rice

sector was only slightly positive. We find that productivity growth was negative during the early

70’s, the second half of the Green Revolution era, followed by positive growth in the late 70’s

and early 80’s, and finally another negative trend in the late 80’s. The positive TFP growth

coincided with a period during the late 70’s when IR36 and other second-generation modern

varieties were introduced. Negative growth during the early 70’s, the second half of the Green

Revolution era, may indicate that the input-output combination was not favorable for the first

generation modern varieties to continue to yield positive TFP growth. Also, the negative growth

in the late 80’s suggests that growth in the level of inputs outweighed the output growth of the

second-generation modern varieties.

The pattern of growth is mostly attributable to technological progress (0.7%) which

occurred during this period. The technological progress component displays a strikingly similar

trend as the Malmquist productivity index, which suggests that the total factor productivity

25

largely consists of technological change. Technological advances were particularly significant in

the regions of Central Luzon, Western Visayas, Southern Mindanao and Northern Mindanao.

Other regions experienced technological regress primarily due to input-biased intensification.

Compared to technological change, the changes in efficiency were quite small.

The factors affecting productivity, efficiency and technological change were analyzed by

second-stage regression analysis. Irrigation infrastructure had a positive effect on TFP in Luzon.

The impact of population pressure on technological and efficiency change gave mixed results.

The effect of population pressure was positive on technological change in Mindanao and was

negative on efficiency change in Mindanao and Luzon. Production distribution to landlords had a

negative effect on TFP, efficiency and technological change. Relative input price changes in

favor of traditional factor- (land and labor) saving and modern factor- (fertilizer and machinery)

using technological change positively affected technological and efficiency changes. Fertilizer

intensification negatively affected total factor productivity in Visayas.

The results suggest that even though there was overall technological progress in the

Philippine rice sector, there were periods of negative growth. These findings are consistent with

cross-country studies of technological change in the agriculture sector of developing countries

which have found productivity declines even in countries where green revolution varieties of rice

and wheat have been widely adopted (Fulginiti and Perrin, 1997; Lau and Yotopoulos, 1989). In

this study , we find periods of productivity decline at the beginning and end of the study period

(1971-90). It seems quite likely that technological change was most rapid immediately after the

introduction of the second generation of rice varieties (IR36) in 1976. However, preceding the

introduction of the new rice technology, productivity declines could have been caused by

intensification of input use from a decline in yield growth of the first generation of rice varieties.

26

The rapid increase in productivity following the introduction of the new varieties could not be

sustained in the late 80’s because of several reasons including possible input intensification and

macroeconomic policies that discouraged production oriented investment and led to input pricing

policies that led to overuse of inputs by farmers.

At the regional level, our analysis suggests that certain regions, in particular, Central

Luzon, West Visayas and South and North Mindanao performed markedly better than others.

Although there is no immediately obvious reason why these areas exhibited higher rates of

technological progress, certain key factors emerge from the regression analysis. For instance,

irrigation investments and tractor use were a major contributor to productivity growth in Central

Luzon. High labor use intensity, as in Central and Eastern Visayas could have contributed to a

slower rate of technology adoption relative to regions with low labor use per hectare. High initial

rates of population growth, as in Central Luzon and Western Visayas, may have led to

modernization of the rice sector, as predicted by the Boserup hypothesis – a conclusion

supported by the observation that population growth rates declined sharply in the two regions in

the late 80’s. Good soil quality and a favorable agroclimatic environment may have also

contributed to productivity gains from modern technology, as in the case of Mindanao. These

results are supported by Evenson and Sardido (1986) who point out that the Mindanao regions

have benefited from their frontier status as well as from improved infrastructure. However, their

conclusions on regional gains in productivity are somewhat different because they have

examined a complete basket of crops and not just rice.

A possible extension of this study would involve classifying regions under different

criteria such as adoption rates of modern varieties, degrees of population pressure, and

proportion of irrigated area, among others. This would provide further insights into the precise

27

causes of agricultural intensification or the lack thereof. Regional research and extension efforts

have significant impacts on productivity gain in food grains and could enrich the analysis

(Evenson and David, 1993). Environmental factors such as soil quality may be important for

explaining stagnant productivity growth. If enough panel data is available, these factors would be

important explanatory variables for productivity growth in the Philippines.

The study of regional-level agricultural intensification and technological change needs to

be supplemented by micro-level analysis that could support policy analysis. In particular, this

study does not cover upland rice production since not enough statistical information is available.

Some upland areas in the Philippines are experiencing rapid population growth through

migration and dynamic transformation of production technologies as well as environmental

problems such as deforestation and soil erosion. An in-depth analysis of these phenomena would

be helpful in understanding the impact of population pressure on production systems and the

environment.

Finally, the Malmquist total factor productivity approach does not directly address

welfare changes of farmers, i.e., whether or not producer surplus increased in the period

analyzed. It is quite possible that farmers' welfare increased even when total factor productivity

growth was negative. More careful interpretation should be made of changes in input and output

prices, technological change and farmer's producer surplus along the direction of total factor

productivity growth.

28

References

Afriat, S. N. (1972). "Efficiency Estimation of Production Functions." International Economic Review.13:568-598.

Arnade, C. A. (1994). Using Data Envelopment Analysis to Measure International Agricultural Efficiency and Productivity. Washington, D.C., Economic Research Service, U.S. Department of Agriculture.

Binswanger, H. P., S. R. Khandker, and M. R. Rosenzweig. (1993). "How Infrastructure and Financial Institutions Affect Agricultural Output and Investment in India." Journal of Development Economics 41: 337-366.

Bjurek, H. and L. Hjalmarsson. (1995). "Productivity in Multiple Output Public Service: A Quadratic Frontier Function and Malmquist Index Approach." Journal of Public Economics.56:447-460.

Boserup, E. (1965). The Conditions of Agricultural Growth. London: George Allen & Unwin Ltd.Bouis, H. E. (1993). "Measuring the Source of Growth in Rice Yields: Are Growth RatesDeclining in Asia?" Food Research Institute Studies. 22:305-330.

Caves, D. W., L. R. Christensen, and W. E. Diewert. (1982). "The Economic Theory of Index Numbersand the Measurement of Input, Output, and Productivity." Econometrica 50(6): 1393-1414.

Cruz, W. and R. Repetto (1992). The Environmental Effects of Stabilization and Structural Adjustment Programs: The Philippine Case. Washington, D.C., World Resources Institute

David, C. C., V. G. Cordova, and K. Otsuka. (1994). Technological Change, Land Reform, and IncomeDistribution in the Philippines. Modern Rice Technology and Income Distribution in Asia. C. C.David and K. Otsuka, Eds. Boulder, Colorado, Lynne Rienner Publishers.

Department of Agriculture (1993a). Crop Development and Soil Conservation Framework for LuzonIsland 1993. Bureau of Soils and Water Management.

Department of Agriculture (1993b). Crop Development and Soil Conservation Framework for VisayaIsland 1993. Bureau of Soils and Water Management.

Department of Agriculture (1993c). Crop Development and Soil Conservation Framework for Mindanao Island 1993. Bureau of Soils and Water Management.

Diamond, P., D. McFadden, and M. Rodriguez. (1978). Measurement of the Elasticity of Factor Substitution and Bias of Technical Change. Production Economics: A Dual Approach to Theoryand Application. M. Fuss and D. McFadden, Eds. Amsterdam, North-Holland: ch. IV.2.

Estudillo, J. P. (1995). Income Inequality in the Philippines, 1961-91: Trends and Factors. Ph. D Dissertation. Department of Economics. Honolulu, Hawaii, University of Hawaii at Manoa.

Evenson, R. and C. David (1993). Adjustment and Technology: The Case of Rice. Paris, DevelopmentCentre of the Organization for Economics Co-operation and Development.

29

Evenson, R. E. and M. L. Sardido (1986). "Regional Total Factor Productivity Change in PhilippineAgriculture." Journal of Philippine Development. 13: 40-61.

Färe, R., R. Grabowski, and S. Grosskopf. (1985). "Technical Efficiency of Philippine Agriculture."Applied Economics. 17: 205-214.

Färe, R. and S. Grosskopf (1992). "Malmquist Productivity Indexes and Fisher Ideal Indexes." TheEconomic Journal. 102:158-160.

Färe, R., S. Grosskopf, B. Lindgren, and P. Roos. (1989). Productivity Developments in SwedishHospitals: A Malmquist Output Index Approach. Department of Economics, Southern IllinoisUniversity, Carbondale.

Färe, R., S. Grosskopf, B. Lindgren, and P. Roos. (1992). "Productivity Changes in Sweedish Pharmacies 1980-1989: A Non-Parametric Malmquist Approach." The Journal of Productivity Analysis. 3:85-101.

Färe, R., S. Grosskopf, B. Lindgren, and P. Roos. (1994). Productivity Developments in Swedish Hospitals: A Malmquist Output Index Approach. Data Envelopment Analysis: The Theory,Applications and the Process. Charnes, A., W. W. Cooper, A. Y. Lewin and L. M. Seiford, Eds.Boston, Kluwer Academic.

Färe, R., S. Grosskopf, and C. A. N. Lovell. (1985). The Measurement of Efficiency of Production.Boston: Kluwer-Nijhoff.

Färe, R., S. Grosskopf, and C. A. N. Lovell. (1994a). Production Frontiers. New York, CambridgeUniversity Press.

Färe, R., S. Grosskopf, M. Norris, and Z. Zhang. (1994b). "Productivity Growth, Technical Progress, andEfficiency Change in Industrialized Countries." American Economic Review.84:66-83.

Farrell, M. J. (1957). "The Measurement of Productive Efficiency." Journal of the Royal StatisticalSociety. 120 (Part III):11-290.

Francia, S. O. (1993). The Philippines: Executive Summary. Fertilizer Policy in Asia and the Philippines. S. Ahmed and A. Clark, Eds. Tokyo, Japan, Asian Productivity Organization.

Fulginiti, L. E. and R. K. Perrin (1997) "LDC agriculture: Nonparametric Malmquist productivityindexes." Journal of Development Economics. 53:373-390.

Grosskopf, S. (1986). "The Role of the Reference Technology in Measuring Productive Efficiency." TheEconomic Journal. 96:499-513.

Hayami, Y., M. Kikuchi, et al. (1990). Transformation of a Laguna Village in the Two Decades of GreenRevolution. Manila, Philippines, International Rice Research Institute.

Hayami, Y. and V. W. Ruttan (1970). "Factor Prices and Technical Change in Agricultural Development:The United States and Japan, 1880-1960." Journal of Political Economy 78: 1115-1141.

30

Hayami, Y. and V. W. Ruttan (1985). Agricultural Development: An International Perspective. Baltimore.

International Resources Group. (1992). The Philippine Natural Resources Accounting Project (NRAP-PHASE I) Main Report. Manila, Philippines.

IRRI (1995). World Rice Statistics, 1993-94. Manila, Philippines, International Rice Research Institute.

Lau, L. J. and P. A. Yotopoulos (1989). "The Meta-Production Function Approach to TechnologicalChange in World Agriculture." Journal of Development Economics 31: 241-269.

Lele, U. and S. Stone (1989). Population Pressure, the Environment and Agricultural Intensification:Variations on the Boserup Hypothesis. Washington, D.C., The World Bank.

Montes, M., F. and J. Y. Lim (1996). "Macroeconomic Volatility, Investment Anemia and Environmental Struggles in the Philippines." World Development. 24(2).

National Statistical Office (1991) Unpublished Family Income & Expenditure Surveys.

Nugent, J. B. and N. Sanchez (1995). The Local Variability of Rainfall and Tribal Institutions: The Caseof Sudan. University of Southern California and College of the Holy Cross.

Olmstead, A. L. and P. Rhode (1993). "Induced Innovation in American Agriculture: A Reconsideration." Journal of Political Economy 101(1): 100-118.

Otsuka, K. (1991). "Determinants and Consequences of Land Reform Implementation in the Philippines." Journal of Development Economics 35: 339-355.

Otsuka, K., F. Gascon, and S. Asano. (1994). "Green Revolution and Labour Demand in Rice Farming:The Case of Central Luzon, 1966-90." Journal of Development Studies 31(1): 82-109.

PhilRice-BAS (1994). Regional Rice Statistics Handbook, 1970-1992. The Philippine Rice Research Institute and the Bureau of Agricultural Statistics.

Pingali, P., Y. Bigot, H. P. Binswanger. (1987). Agricultural Mechanization and the Evolution ofFarming Systems in Sub-Saharan Africa. Washington, D.C., Johns Hopkins University Press.

Pingali, P. L. (1992) “Agriculture-Environment Interactions in the Southeast Asian Humid Tropics”,Southeast Asian Journal of Agricultural Economics, 1(2): 107-124.

Pingali, P. L., R. S. Zeigler, M. Hossain, and J.-C. Prot. (1995). "Humid Tropics of Asia: Coping withHuman Pressure and Environmental Stress." Paper prepared for the Planning Meeting of theEcological Initiative for the Humid Tropics at the International Rice Research Institute, LosBaños, Laguna, Philippines, 18-22 September 1995.

Roumasset, J. and W. James (1979). "Explaining Variations in Share Contracts: Land Quality, Population Pressure and Technological Change." Australian Journal of Agricultural Economics 23(2): 116-127.

31

Ruttan, V. W. and Y. Hayami (1995). "Induced Innovation Theory and Agricultural Development: APersonal Account." In B. M. Koppel (ed.), Induced Innovation Theory and InternationalAgricultural Development: A Reassessment. Baltimore: Johns Hopkins University Press.

Shephard, R. W. (1953). Cost and Production Functions. Princeton, N.J., Princeton University Press.

Schultz, T. W. (1964). Transforming Traditional Agriculture. New Haven, Yale University Press.

Thirtle, C., D. Haddey, and R. Townsend. (1995). "Policy Induced Innovation in Sub-Saharan AfricanAgriculture: A Multilateral Malmquist Productivity Index Approach." mimeo. Department ofAgricultural Economics and Management, University of Reading.

32

Figure 1. Input-oriented Distance Functionand the Malmquist Productivity Index

Input

Input

x1

x20

xt+1

Lt (yt )

a

b

c

de

f

Lt+1(yt+1)

xt

AB

C

D

33

Inde

x

Figure 2. Efficiency, Technological Change and Malmquist Indices (CRS) for the Rice Sector, Philippines, 1971-1990 (1971=100, weighted mean).

1972 1974 1976 1978 1980 1982 1984 1986 1988 19900

50

100

150

200

250

300

350

IlocosCagayan C. LuzonS. TagalogBicol

Figure 3. Technological Change Index (CRS) for the Rice Sector,Luzon Island, Philippines (1971=100).

Inde

x

1972 1974 1976 1978 1980 1982 1984 1986 1988 199080

90

100

110

120

130

140

150

EfficiencyTechnologyMalmquist

34

1972 1974 1976 1978 1980 1982 1984 1986 1988 199040

60

80

100

120

140

160

180

W.VisayasC. VisayasE. Visayas

Figure 4. Technological Change Index (CRS) for the Rice Sector,Visayas Island, Philippines (1971=100).

Inde

x

1972 1974 1976 1978 1980 1982 1984 1986 1988 199020

40

60

80

100

120

140

160

180

200

220

240

W. MindanaoN. MindanaoS. MindanaoC. Mindanao

Figure 5. Technological Change Index (CRS) for the Rice Sector,Mindanao Island, Philippines (1971=100).

Inde

x

35

Table 1. Estimated Average Yield of PALAY Production by Varieties and Adoption of Modern Variety by Regions, 1971-1990.

Region Total (MT/Ha) HYV (MT/Ha) TV (MT/Ha) % Area Adopted HYV

1971-75 1976-80 1981-85 1986-90 1971-75 1976-80 1981-85 1986-90 1971-75 1976-80 1981-85 1986-90 1971-75

Philippines 1.54 1.97 2.33 2.56 1.74 2.18 2.50 2.69 1.24 1.34 1.53 1.78 58.46Ilocos 1.53 1.93 2.40 2.64 1.64 2.04 2.49 2.68 1.37 1.70 1.89 2.02 58.84Cagayan Valley 1.77 2.06 2.49 3.08 1.97 2.25 2.64 3.18 1.49 1.54 1.63 1.98 54.48Central Luzon 1.94 2.76 3.18 3.20 1.98 2.81 3.21 3.23 1.83 2.26 2.37 2.37 72.88Southern Tagalog 1.54 1.89 2.24 2.51 1.85 2.29 2.45 2.68 1.20 1.27 1.32 1.64 51.60Bicol 1.69 2.06 2.03 2.18 1.88 2.24 2.19 2.28 1.27 1.21 1.19 1.47 68.14Western Visayas 1.57 2.00 2.38 2.48 1.73 2.14 2.45 2.52 1.24 1.15 1.21 1.47 64.60Central Visayas 1.33 1.62 1.41 1.54 1.40 1.73 1.52 1.66 1.22 1.30 1.08 1.08 55.59Eastern Visayas 1.23 1.47 1.81 1.86 1.41 1.70 1.95 1.98 1.01 1.04 1.24 1.27 54.28Western Mindanao 1.62 2.49 2.30 2.56 2.04 2.95 2.67 2.81 1.21 1.33 1.56 1.82 46.11Northern Mindanao 1.25 1.61 2.56 3.07 1.43 1.76 2.70 3.27 0.90 0.97 1.43 1.97 63.51Southern Mindanao 1.76 2.37 2.92 3.40 2.01 2.62 3.11 3.50 1.17 1.45 2.03 2.80 70.92Central Mindanao 1.43 1.72 2.88 2.93 1.71 2.06 3.24 3.18 1.17 1.32 1.89 2.12 47.87Source: Regional Rice Statistics Handbook (1994).Note: HYV = High Yielding Variety, and TV = Traditional Variety.

Table 2. Factor Input Ratios by Regions, 1971-1990.

Region Labor/Land (Mandays/Ha) Fertilizer/Landa/ (NPK Kg/Ha) Tractor/Land (Unit/'000 Ha) Carabao/Land (Head/Ha)1971-75 1976-80 1981-85 1986-90 1971-75 1976-80 1981-85 1986-90 1971-75 1976-80 1981-85 1986-90 1971-75

Philippines 95.06 93.64 93.02 93.26 49.46 53.50 51.30 73.85 21.45 42.88 84.29 114.23Ilocos 83.43 82.32 82.11 83.30 66.98 72.20 73.16 95.72 34.50 66.65 115.66 202.03Cagayan Valley 95.02 93.65 93.59 95.20 51.56 57.84 55.46 87.82 13.00 35.78 99.48 166.16Central Luzon 80.80 81.14 76.02 78.65 100.05 101.36 96.67 112.49 88.42 170.33 240.65 294.29Southern Tagalog 71.86 72.21 71.49 72.05 54.17 55.49 51.48 70.65 43.34 61.33 97.47 121.62Bicol 95.67 94.38 91.23 92.09 39.65 45.89 40.43 60.43 18.57 48.40 97.20 132.15Western Visayas 96.13 96.89 96.95 98.79 74.25 69.29 62.04 83.80 26.44 50.36 105.03 162.03Central Visayas 119.40 114.92 112.24 111.23 35.85 35.84 31.50 55.22 8.20 11.27 12.13 11.54Eastern Visayas 112.14 108.57 112.32 109.36 32.48 34.10 33.60 55.98 11.92 32.93 66.48 89.60Western Mindanao 95.60 94.36 93.58 93.30 51.29 59.30 53.64 75.79 7.73 20.99 51.78 73.01Northern Mindanao 106.71 97.84 100.01 98.25 47.18 51.31 53.10 72.63 30.77 54.84 137.75 184.25Southern Mindanao 98.46 96.78 98.23 96.14 35.32 45.80 44.97 69.81 36.37 74.40 143.10 188.09Central Mindanao 95.69 98.73 98.04 98.76 37.73 44.03 48.38 67.77 13.26 20.02 56.73 73.35Source: Census Of Agriculture (1971, 1980, 1991); National Statistical Information Center unpublished data; Regional Rice Statistics Handbook (1994).Note: a Land includes both areas applying and not applying inorganic fertilizers.

36

Table 3. Efficiency Change (E), Technological Progress (T), and Malmquist Total Factor Productivity (M) Indices in PALAY Production with CRS Technology by Regions, 1971-1990.

Region 1971-75 1976-80 1981-85 1986-90E T M E T M E T M E T

Philippines(WAM) 100.2 97.8 98.0 100.1 102.3 102.4 99.9 103.7 103.6 99.6 98.6Philippines(GM) 99.6 97.2 96.9 100.7 101.4 102.1 99.8 101.8 101.6 99.4 97.3Ilocos 98.1 101.3 99.4 96.5 96.7 93.4 103.4 98.8 102.1 98.0 102.6Cagayan Valley 100.0 90.9 90.9 97.0 105.3 102.2 103.1 103.5 106.7 100.0 97.6Central Luzon 100.0 108.2 108.2 100.0 107.7 107.7 100.0 109.9 109.9 100.0 106.9Southern Tagalog 101.1 96.1 97.1 98.7 100.5 99.2 101.3 99.0 100.3 100.0 99.1Bicol 101.7 98.3 99.9 97.1 107.3 104.1 97.7 102.7 100.3 102.6 94.4Western Visayas 103.9 98.5 102.3 98.8 110.2 108.8 100.6 100.9 101.5 95.4 97.7Central Visayas 101.2 92.7 93.7 101.7 95.7 97.4 100.0 93.0 93.0 100.0 96.5Eastern Visayas 98.3 95.5 93.9 103.7 100.6 104.3 96.1 103.2 99.2 97.4 93.8Western Mindanao 103.2 100.2 103.5 97.1 98.1 95.2 99.7 102.6 102.4 103.3 92.0Northern Mindanao 90.8 97.3 88.4 108.6 107.3 116.6 98.2 105.0 103.1 102.6 95.6Southern Mindanao 100.0 100.2 100.2 100.0 101.0 101.0 100.0 106.6 106.6 100.0 95.7Central Mindanao 100.0 88.4 88.4 100.0 92.6 92.6 100.0 96.0 96.0 100.0 93.3Note: Index of 100 means no change, less than 100 means deterioration and greater than 100 means an improvement. Figures are geometric mean of 5 year and 20 year periods. WAM: weighted arithmetic mean using output share of each region as a weight; GM: geometric mean of regional scores.

37

Table 4. Regression Analysis of Efficiency Change,Technological Change, and Malmquist TFP Index by Regions, 1971-1990.

Efficiency Change Technological Change Malmquist TFP IndexVariables PHILS Luzon Visayas Mindanao PHILS Luzon Visayas Mindanao PHILS Luzon Visayas Mindanao

Infrastructure Variables

Irrigation 0.036 0.218 ** -0.656 ** 0.155 0.018 0.115 ** -0.155 ** 0.013 * -0.086 0.803 ** -0.707 ** 0.400(0.84) (2.25) (4.01) (0.76) (1.54) (4.63) (3.63) (1.65) (0.96) (2.71) (4.36) (1.55)

Transportation 0.070 ** 0.269 ** -0.203 0.217 ** -0.019 ** 0.037 ** 0.015 -0.005 * 0.129 ** 0.832 ** -0.046 0.025(3.50) (4.96) (1.22) (2.37) (4.19) (3.06) (0.37) (1.75) (3.66) (5.50) (0.29) (0.22)

Demographic Variable

Population/Land -0.131 ** -0.272 ** 0.808 ** -0.813 ** 0.070 ** -0.006 0.007 0.087 ** 0.086 * -0.235 1.124 ** -0.313 **

(4.82) (3.94) (2.40) (2.94) (11.92) (0.27) (0.08) (19.90) (1.67) (1.00) (3.63) (2.71)Technology Related Variables

Higher Education 0.602 ** 0.086 -2.274 -0.488 0.054 ** 0.349 ** 0.440 0.141 ** 0.661 ** -3.372 ** -3.654 ** 0.954(4.66) (0.16) (1.49) (0.85) (2.01) (3.33) (1.20) (5.46) (2.79) (2.29) (2.79) (1.37)

Modern Variety II 0.092 ** -0.026 0.024 -0.254 ** 0.000 -0.012 -0.023 0.001 0.011 0.097 -0.103 0.035(2.44) (0.42) (0.23) (2.40) (0.02) (1.06) (1.07) (0.15) (0.18) (0.58) (1.12) (0.27)

Institutional Variable

Landlord Share -0.729 ** -0.383 ** -0.836 ** -0.623 ** -0.019 -0.110 ** -0.153 ** -0.002 -0.996 ** -0.908 * -0.367 -0.896 **

(8.14) (2.10) (3.22) (3.13) (1.11) (3.26) (2.14) (0.19) (6.69) (1.82) (1.48) (3.82)Factor Price Variables

Land/Fertilizer Prices 0.482 * 0.635 ** 1.265 * -0.643 -0.135 ** 0.123 ** -0.365 ** -0.012 1.539 ** 0.139 0.807 1.167(1.93) (2.26) (1.81) (0.59) (2.17) (2.19) (2.19) (0.44) (3.14) (0.17) (1.17) (1.24)

Labor/Machinery Prices 14.433 ** 17.114 ** 14.119 -9.526 1.269 * 1.830 ** 5.524 ** 0.197 4.982 38.224 ** 1.454 4.724(5.37) (3.20) (1.46) (1.07) (1.90) (2.17) (2.00) (0.58) (0.86) (2.52) (0.15) (0.40)

Land/Labor Prices 0.590 ** 0.712 * 0.556 * 0.326 0.073 ** 0.125 ** 0.255 ** 0.007 0.467 ** 2.396 ** 0.087 0.548 **

(7.88) (1.85) (1.93) (1.53) (4.83) (1.97) (2.84) (0.88) (3.24) (2.08) (0.30) (2.27)Factor Input Intensity Variables

Fertilizer/Land 0.054 0.310 ** -0.153 * -0.009 0.032 ** 0.061 ** -0.009 0.002 -0.119 ** 0.312 -0.127 0.221 **

(0.84) (2.96) (1.89) (0.08) (5.27) (3.26) (0.37) (0.74) (2.48) (1.11) (1.64) (1.96)Hand Tractor/Land -0.086 * 0.039 0.347 ** 0.007 ** -0.021 ** -0.070 ** 0.039 -0.016 ** 0.023 0.252 0.390 ** -0.178

(1.86) (0.53) (2.88) (6.19) (3.61) (5.54) (1.21) (3.48) (0.45) (1.37) (3.50) (1.33)Purely Exogenous Variables

Weather -0.013 ** -0.012 -0.030 -0.025 -0.006 ** -0.005 0.022 * 0.001 0.014 0.040 0.025 -0.022(1.98) (0.73) (0.73) (0.53) (2.17) (1.58) (1.77) (1.10) (0.51) (0.87) (0.54) (0.57)

Disasters -0.016 ** 0.038 ** -0.060 ** 0.018 0.000 0.000 -0.022 ** -0.001 -0.006 -0.119 ** -0.035 0.011(2.52) (2.52) (2.12) (0.67) (0.05) (0.15) (2.96) (0.81) (0.56) (2.81) (1.23) (0.45)

Statistics

Number of Observations 216 90 54 76 216 90 57 72 216 90 54 72Log-Likelihood Function 118.41 -3.91 -0.70 43.84 461.60 145.76 74.92 237.03 -38.84 -86.75 -3.33 -33.95Homoscedasticity Test1 221.15 ** 80.06 ** 1.42 21.23 ** 230.99 ** 34.08 ** 7.72 22.92 ** 169.80 ** 22.04 ** 18.60 12.29 **

Groupwise Correlation Test2 26.87 ** 16.47 ** 12.07 ** 22.23 ** 145.59 ** 13.22 ** 13.28 ** 28.27 ** 26.86 ** 14.76 ** 15.13 11.16 **

Autocorrelation Test3 755.76 ** 227.57 ** 31.98 ** 114.20 ** 482.40 ** 81.83 ** 13.02 98.60 ** 321.17 ** 32.28 ** 37.06 ** 118.97 **

** denotes significant level of at least 5% and * denotes significant level of at least 10%. Figures in parentheses are z statistic.All independent variables are in log form except factor price variables, factor input intensity variables, disaster and dummy variable for period of second modern variety.1 Log ratio statistic is used and its distribution is χ2 to test the null of homoscedastic variance against heteroscedasticity.2 Log ratio statistic to test the null of no cross group correlation against groupwise correlation.3 Box-Ljung statistic to test the null of no autocorrelation. The statistic distributes as χ2 with 10 degree of freedom.

38

Appendix Table 1. Estimated Average PALAY Production by Varieties by Regions, 1971-1990.

Region Total (Million MT) HYV (Million MT) TV (Million MT)1971-75 1976-80 1981-85 1986-90 1971-75 1976-80 1981-85 1986-90 1971-75 1976-80 1981-85 1986-90

Philippines 5.537 7.272 8.035 9.107 3.712 5.920 7.243 8.346 1.825 1.352 0.791 0.761CAR 0.134 0.164 0.164 0.173 0.044 0.070 0.097 0.120 0.090 0.095 0.066 0.053Ilocos 0.511 0.561 0.725 0.794 0.320 0.394 0.649 0.757 0.191 0.167 0.075 0.037Cagayan Valley 0.616 0.730 0.797 1.077 0.378 0.584 0.720 1.018 0.238 0.147 0.077 0.058Central Luzon 0.912 1.181 1.487 1.603 0.690 1.086 1.443 1.568 0.223 0.095 0.043 0.034Southern Tagalog 0.699 0.837 0.839 0.963 0.438 0.621 0.742 0.856 0.262 0.216 0.097 0.107Bicol 0.558 0.650 0.631 0.670 0.423 0.579 0.586 0.613 0.135 0.071 0.045 0.056Western Visayas 0.651 0.993 1.082 1.082 0.471 0.907 1.047 1.056 0.181 0.087 0.035 0.026Central Visayas 0.119 0.155 0.148 0.174 0.071 0.120 0.120 0.148 0.048 0.034 0.028 0.025Eastern Visayas 0.220 0.269 0.357 0.389 0.138 0.207 0.307 0.341 0.082 0.062 0.049 0.047Western Mindanao 0.224 0.364 0.307 0.357 0.137 0.310 0.236 0.291 0.087 0.054 0.071 0.066Northern Mindanao 0.167 0.257 0.302 0.408 0.124 0.229 0.284 0.366 0.043 0.029 0.018 0.042Southern Mindanao 0.268 0.428 0.534 0.678 0.216 0.368 0.468 0.596 0.051 0.059 0.066 0.082Central Mindanao 0.458 0.682 0.662 0.740 0.264 0.445 0.543 0.613 0.194 0.237 0.119 0.127Source: Regional Rice Statistics Handbook (1994).Note: HYV = High Yielding Variety, and TV = Traditional Variety; Figures are annual averages of five-year period.

Appendix Table 2. Rice Production Area under Irrigation by Regions, 1971-1990.

Region Irrigated Area ('000 Ha) Growth of Irrigated Area (%) % Irrigated Area/Arable Land % Irrigated Area/Planted Area1971-75 1976-80 1981-85 1986-90 1971-75 1976-80 1981-85 1986-90 1971-75 1976-80 1981-85 1986-90 1971-75

Philippines 1362.8 1475.7 1668.8 1892.2 2.23 3.76 3.90 2.41 19.23 19.04 19.80 20.41 36.86Ilocos 122.6 121.7 149.9 161.4 -3.94 0.89 4.26 2.28 48.71 46.10 54.16 55.45 36.65Cagayan Valley 179.4 190.7 204.8 280.0 -4.50 -1.39 9.24 3.28 52.68 51.39 50.00 61.31 49.76Central Luzon 267.6 295.0 328.8 370.1 4.69 1.76 0.70 3.78 57.89 64.82 69.50 71.54 54.11Southern Tagalog 180.0 186.2 185.7 200.3 3.08 -1.98 3.54 1.04 21.87 19.98 17.59 16.85 38.78Bicol 156.8 145.1 160.5 169.7 -0.39 2.53 1.08 0.24 21.37 17.63 18.27 18.87 44.30Western Visayas 90.8 120.4 153.1 158.7 3.02 8.17 2.36 0.28 14.85 19.30 23.35 22.63 21.74Central Visayas 31.1 29.9 36.0 46.7 -2.98 13.89 -1.29 11.25 7.94 6.57 7.25 8.97 31.55Eastern Visayas 44.4 54.8 73.5 77.3 -4.35 11.11 8.98 -4.91 7.63 8.56 10.92 11.31 23.47Western Mindanao 48.5 50.8 54.9 61.2 14.38 2.50 6.48 1.93 10.80 10.10 9.51 9.12 34.69Northern Mindanao 59.7 82.4 79.2 90.7 10.97 1.79 2.14 7.29 10.84 12.37 10.17 10.28 40.17Southern Mindanao 84.5 91.6 117.8 132.8 9.31 0.19 6.22 3.09 12.17 10.92 11.97 11.86 53.98Central Mindanao 97.5 107.0 124.6 143.3 -2.57 5.70 3.11 -0.69 28.77 28.30 27.99 26.53 30.69Source: Regional Rice Statistics Handbook (1994); Census Of Agriculture (1971, 1980, 1991).

39

Appendix Table 3. Landlord Share, Population with Higher Education, and Paved Road by Regions, 1971-1990.

Region % Production Paid to Landlord % Population ≥ 15 Yrs. Educ. % Paved Road1971-75 1976-80 1981-85 1986-90 1971-75 1976-80 1981-85 1986-90 1971-75 1976-80 1981-85 1986-90

Philippines 14.55 12.17 12.42 10.68 6.07 6.51 6.91 7.32 13.35 14.18 11.13 11.69Ilocos 11.78 16.01 19.20 19.54 6.79 7.36 7.78 8.31 18.40 19.78 15.64 20.11Cagayan Valley 18.96 11.66 11.33 9.46 6.03 6.42 6.87 7.27 8.28 8.54 7.13 7.13Central Luzon 16.59 13.31 11.59 8.11 7.02 7.44 7.88 8.30 30.99 32.87 19.38 20.17Southern Tagalog 20.22 16.54 13.33 13.03 7.92 7.53 7.53 7.93 34.66 29.35 18.91 19.03Bicol 15.99 13.82 14.23 12.97 6.27 6.85 7.47 7.83 22.42 23.39 25.79 21.67Western Visayas 17.05 14.02 15.39 11.82 6.23 6.67 7.10 7.51 16.67 14.83 13.48 13.80Central Visayas 15.45 12.19 15.14 10.16 5.78 6.13 6.45 6.81 13.39 13.33 13.36 15.03Eastern Visayas 19.12 17.34 17.15 17.01 5.38 5.81 6.22 6.56 10.91 13.17 12.80 18.56Western Mindanao 11.15 6.89 9.79 8.17 4.92 5.40 5.65 6.10 9.07 9.85 6.92 6.36Northern Mindanao 12.79 9.14 8.40 8.34 6.40 6.91 7.33 7.59 12.56 12.39 8.56 9.13Southern Mindanao 10.83 10.00 9.19 8.07 6.18 6.68 7.06 7.41 3.44 6.48 5.27 5.05Central Mindanao 9.62 9.96 9.34 7.76 4.62 5.43 5.99 6.63 9.14 8.40 4.78 4.86Source: RRSH (1994); IRRI unpublished data.

Appendix Table 4. Population, Population Pressure and Growth Rates by Regions, 1971-1990.

Region Population ('000 Persons) Population Growth (%) Pop./Arable Land (Person/Ha) Growth of Pop. Density (%)1971-75 1976-80 1981-85 1986-90 1971-75 1976-80 1981-85 1986-90 1971-75 1976-80 1981-85 1986-90 1971-75

Philippines 39256 44913 50837 57252 2.53 2.54 2.37 1.99 5.91 6.06 6.20 6.33 0.52Ilocos 2631 2844 3091 3400 1.79 1.40 1.86 2.10 10.45 10.77 11.17 11.68 0.84Cagayan Valley 1592 1831 2081 2331 2.72 2.72 2.50 1.42 4.67 4.93 5.08 5.10 1.03Central Luzon 3985 4581 5199 5881 2.99 2.68 2.45 2.64 8.61 10.06 10.98 11.36 3.28Southern Tagalog 9514 11300 12993 15026 3.75 3.04 2.86 3.38 11.55 12.11 12.30 12.63 1.30Bicol 3112 3372 3746 4105 1.48 1.70 2.34 -0.01 4.24 4.10 4.26 4.56 -0.72Western Visayas 3940 4385 4868 5387 2.75 1.73 2.31 1.21 6.44 7.03 7.42 7.68 2.39Central Visayas 3252 3637 4033 4445 2.25 2.19 2.00 1.88 8.29 7.98 8.10 8.54 -0.73Eastern Visayas 2517 2725 2964 3184 1.77 1.46 1.82 -0.09 4.32 4.26 4.40 4.65 -0.08Western Mindanao 1986 2345 2735 3056 1.89 4.21 2.34 2.04 4.42 4.66 4.74 4.56 -0.32Northern Mindanao 2179 2590 3014 3420 3.40 3.49 2.73 2.06 3.96 3.88 3.87 3.87 -0.40Southern Mindanao 2517 3105 3647 4162 4.21 4.17 2.60 3.10 3.62 3.69 3.70 3.71 0.47Central Mindanao 2030 2197 2468 2854 1.36 1.74 2.62 4.17 5.99 5.81 5.54 5.27 -0.77Source: RRSH (1994); COA (1971, 1980, 1991).Note: Figures for Philippines do not include National Capital Region (NCR), Cordillera Autonomous Region (CAR), and Autonomous Region in Muslim Mindanao (ARMM).