investigating the presence of regional economic growth ... · at an average of 1.73 percent per...

Philippine Journal of DevelopmentNumber 69, Second Semester 2010

Volume XXXVII, No. 2

Investigating the Presence of Regional Economic Growth Convergencein the Philippines using Kalman Filter1

Dennis s. Mapa, Monica FleriDa B. sanDoval,anD DaviD Joseph eMManuel B. Yap ii2

ABSTRACTThis paper investigates the presence of stochastic and dynamic convergence in the 14 regional economies in the Philippines in terms of per capita gross regional domestic product (GRDP) using panel data from 1988 to 2007. Stochastic convergence, which indicates convergence of the regional per capita incomes in the long run, is tested using the Levin, Lin, and Chu (LLC) and Im, Pesaran, and Shin (IPS) panel unit root tests. The presence of convergence, on one hand, indicates that the economically laggard regions are gaining on the economically better-performing regions with respect to per capita GRDP. On the other hand, the lack of convergence indicates a need to reevaluate existing regional and national economic policies on development. Dynamic convergence or convergence in the growth rates of per capita income is determined by the time-varying parameter (TVP) model estimated using the Kalman filter. The paper proceeds to examine the dynamic convergence behavior of

1 Paper presented during the 11th International Convention of the East Asian Economic Association (EAEA) held on November 15–16, 2008 in Manila, Philippines. The authors are grateful to Shandre M. Thangavelu, Michael Alba, and participants in the Institutions and Human Capital session of the said conference for their comments and suggestions. The authors are also grateful to the comments and suggestions of two anonymous referees. The authors also acknowledge the financial assistance provided by the Philippine Economic Society (PES) to attend the EAEA conference. Errors and omissions are sole responsibilities of the authors. 2 Associate Professor and Director for Research, School of Statistics and Affiliate Associate Professor, School of Economics, University of the Philippines, Diliman, [email protected], [email protected]; Foreign Exchange Officer III, Bangko Sentral ng Pilipinas, [email protected]; and Lecturer, School of Statistics, University of the Philippines, Diliman

2 PhiliPPine Journal of DeveloPment 2010

each region based on the value of the estimates of the parameters of the TVP. The results show that out of the 14 regions studied, seven regions are found to converge dynamically toward the average per capita national gross domestic product (GDP) growth rate while six regions lag behind. None of the growth rates in the per capita GRDP of the 13 regional economies converges toward the per capita growth rate of the National Capital Region (NCR), the lead region used in the study.

INTRODUCTIONThe study of regional economics is concerned with the spatial distribution of economic activity across geographic areas within a country and is now considered a forefront of development issues. The development of a country is contingent on the growth of its economy, and the strength of a national economy is derived from the stability of its regional economies. An understanding of the nuances of the regional economies is thus crucial to any effort geared toward national development. As noted by the economist Paul Krugman, “…one of the best ways to understand how the international economy works is to start by looking at what happens inside nations. If we want to understand differences in national growth rates, a good place to start is by examining differences in regional growth” (Krugman 1991: 3).

In the case of the Philippines, the level of regional economic activity has been predominantly unequal through the years where majority of the economic output is concentrated in the NCR and the two adjacent regions of Central Luzon and Southern Tagalog, with these three regions (out of the 17 regions as of 2007) producing about 55 percent of the total national output (Hill et al. 2007). This rather high level of economic disparity across the regions gave rise to the national economic agenda unveiled by President Gloria Macapagal-Arroyo in her 2006 State of the Nation Address (SONA) where she committed government resources, mainly through various infrastructure projects, to enhance the competitive advantage of the natural geographic composition of the regions. The creation of the five “Super Regions3” aims to fast track development initiatives in North Luzon as Agribusiness Quadrangle, Metro Luzon as Urban Beltway, Central Philippines as excellent tourist destination, Mindanao as a center of agribusiness investment, and the Cyber Corridor that will link all the 17 regions through infrastructure and information and communications technology (ICT).

3 Critics argue, however, that the size of the funding commitment for all the infrastructure projects mentioned was simply too large for the government’s financial position.

3Mapa, Sandoval, and Yap

Manasan and Mercado (1999) examined the regional economic growth and convergence of the Philippine regions from 1975 to 1997 using the same 14 regional economies used in this paper. The authors utilized the classical approach to convergence analysis using two models of convergence – the δ-convergence and the β-convergence. The regional economies are said to exhibit δ-convergence if the dispersion (measured in terms of the standard deviation) of the per capita GRDP (level) is decreasing over time. The regional economies exhibit the β-convergence when poorer regions grow faster (in terms of per capita GRDP growth) than richer regions. In the study, the authors found that δ-convergence exists in the 14 regional economies over the 22-year period, but that variations of the δ-convergence were found to be significant within subperiods. Dispersion in the regional per capita incomes fell during the period 1975–1985 but rose sharply in 1986–1989 and decreased again in 1990–1997. Moreover, the authors found strong evidence of β-convergence in the regional economies during the period 1975–1997, but the estimated speed of convergence is only 0.6 percent which is much lower than the 2 percent speed of convergence for developed countries during the same period.

The primary objective of this study is to examine the economic growth behavior of the Philippine regions in terms of their per capita GRDP4 over the time period 1988–2007. The study aims to determine if Philippine regions converge to a particular economic growth path or diverge into different growth paths. This study provides an analysis similar to that of Manasan and Mercado (1999) but looks at a more recent methodology in measuring convergence.

The study investigates two types of convergence: stochastic and dynamic convergence. Both types of convergence are essential in characterizing the behavior of regional economies. The presence of stochastic convergence implies long-run convergence and tells us if the differences in the levels of per capita GRDP of the regional economies are decreasing over time. Dynamic convergence investigates the short-run convergence of each region. Dynamic convergence will show if the growth rates in the per capita GRDP converge toward the national average per capita GDP growth rate or toward the average per capita growth rate of the lead region or the NCR. Dynamic convergence also provides information regarding the specific nature of the convergence of each region.

Stochastic convergence is tested using LLC (2002) and IPS (2003) panel unit root tests.5 Both tests work under the null hypothesis of unit root. Rejection of the null hypothesis implies stationarity and the presence of stochastic convergence. Nonrejection of the null hypothesis means nonstationarity and

4 Per capita GRDP is part of the Regional Accounts that measures level of economic development of a region.5 LLC and IPS tests are DF-type tests proposed by Levin, Lin & Chu (2002) and Im, Pesaran & Shin (2003), respectively.


that the regions are divergent in the long run. As argued by Chou (2006), the presence of convergence suggests that laggard regions are catching up with the economically better-performing regions in terms of per capita GRDP, while the lack of convergence indicates a need for the creation and implementation of alternative regional policies. Divergence also indicates that all regions do not approach a common steady state growth path in the long run. It is, however, important to note that this does not mean total lack of convergence among the regions. Dynamic convergence enables the analysis of possible regional convergence subgroupings. The dynamic properties of convergence are explored via the TVP model and estimated using the Kalman filter. The individual convergence behavior of a particular region is governed by the corresponding value and sign of the estimated time-varying parameter. We explore the possibility that the regional per capita GRDP growth rates may: (1) converge toward the average of the national per capita GDP growth rate to serve as national benchmark; (2) converge toward growth rate of lead region – the NCR; or (3) diverge from the average national per capita GDP growth rate. The determination of the convergence behavior of Philippine regional economies would identify the regions that require immediate attention. A profile of the convergence behavior of regional economies could also lead to the discovery of effective regional economic configurations. These are crucial to the development and implementation of economic policies for sustainable economic growth.

The rest of the paper is organized as follows: Section 2 presents an overview of the Philippine regional settings. We then expound on the process of performing LLC and IPS panel unit root tests and the estimation of time-varying parameters using Kalman filter in Section 3. The results of the tests as applied to Philippine regional data are then presented in Section 4, followed by the conclusions in the last section.

PHILIPPINE REGIONAL SETTINGSThe data set used in the study is a regional panel data consisting of 14 Philippine regions6 with per capita GRDP recorded for the period 1988–2007.7 The list of regions is provided in Table 1. Figure 1 is a map of the Philippines showing the 14 regions.

The mean of the national per capita GDP (measured in constant 1985 prices) over the period 1988–2007, as shown in Table 2, is about PHP 12,396 and growing

6 The data set consists of 14 regions, instead of the current 17 regions, for data consistency.7 Figures are in 1985 prices. Adjustments were made to all Mindanao regions for the years 1994–2007 due to changes in regional classification.


Table 1. Philippine regions included in the study

Region Name Region Name

1234567

Ilocos RegionCagayan ValleyCentral LuzonSouthernTagalog*Bicol RegionWestern VisayasCentral Visayas

891011121314

Eastern VisayasWestern MindanaoNorthern MindanaoSouthern MindanaoCentral Mindanao **National Capital Region (NCR)Cordillera Administrative Region (CAR)

Source of basic data: National Statistical Coordination Board (NSCB).Notes: * The old Region IV (Southern Tagalog) is now composed of two regions: Regions IV-A and IV-B. ** Most of Central Mindanao now belongs to the Autonomous Region in Muslim Mindanao (ARMM).

Source: Department of Agriculture website (www.da.gov.ph).

Figure 1. Map of the Philippines

CAR REGION II

REGION I

REGION IV

REGION VI

REGION VII

CaragaREGION X

ARMMREGION XII

REGION XI

REGION III

REGION V

REGION VIII


at an average of 1.73 percent per year.8 The NCR, considered as the lead region, had an extremely high average per capita GRDP of PHP 29,669 as compared to the other regions, and average yearly growth rate of about 2.37 percent. The Cordillera Administrative Region (CAR), however, has the highest mean yearly per capita GRDP growth rate of 3.04 percent. The Bicol Region has the lowest mean per capita GRDP, amounting only to roughly PHP 5,620.

The high income disparity in the country is highlighted by the fact that the average per capita income in Metro Manila (or NCR) is around 2.4 times the average per capita income of the whole country and about 5.2 times the average per capita income of the poorest region (Bicol Region).

Using the national average as a benchmark, only three out of the 14 regions have average income higher than the national average: NCR, CAR, and Southern Tagalog (Region IV).

In terms of socioeconomic indicators, the poverty incidence in Table 3 shows that location really matters. In 2006, poverty incidence for the entire country is estimated to be 33 percent.9 A large disparity appears once we examine poverty incidence across the regions. On one hand, the poverty incidence in the NCR is only 10.4 percent, while the two adjacent regions of Central Luzon (Region 3) and Southern Tagalog (Region IV-B) have poverty incidence of 20.7 percent and

8 The average per capita growth rate for the entire country during this period is less than 2 percent. At this growth rate, it will take about 35 years before real per capita income doubles.9 With the rising food prices, experienced since the start of the year 2008, economists have predicted that the poverty incidence will increase further in 2008.

Region Region NameMean per Capita GRDP

(in 1985 prices)Mean per Capita Growth Rate (%)

Mean per Capita RGDP Index (National=100)

1234567891011121314

Ilocos RegionCagayan Valley Central Luzon Southern TagalogBicol RegionWestern Visayas Central Visayas Eastern Visayas Western Mindanao Northern Mindanao Southern Mindanao Central Mindanao NCRCARNational GDP

6,6076,816

10,94313,2605,620

10,42511,3075,8098,169

10,54312,0697,505

29,66914,76612,396

2.092.130.690.752.122.62.311.442.831.51.810.152.373.041.73

53.30 54.99 88.28

106.97 45.34

84.10 91.21 46.86 65.90 85.05 97.36 60.54

239.34 119.12 100.00

Sources: NSCB; authors’ computations

Table 2. Descriptive statistics of per capita GRDP of 14 regions, 1988–2007


Region 2000 2003 2006

Region IRegion IIRegion IIIRegion IV-ARegion IV-BRegion VRegion VIRegion VIIRegion VIIIRegion IXRegion XRegion XIRegion XIICARARMMCaragaNCRPhilippines

35.3030.4021.4019.1045.3052.6044.5036.2045.1044.8043.8033.3046.8037.7060.0051.207.80

33.00

30.2024.5017.5018.4048.1048.5039.2028.3043.0049.2044.0034.7038.4032.2052.8054.006.90

30.00

32.7025.5020.7020.9052.7051.1038.6035.4048.5045.3043.1036.6040.8034.5061.8052.6010.4032.90

Source: NSCB

Table 3. Regional headcount poverty, selected years (in percent)

20.9 percent, respectively. On the other hand, the Autonomous Region in Muslim Mindanao (ARMM)10 has the highest poverty incidence of about 62 percent.

In Figures 2 and 3, the transitions of per capita GRDP of the regions as well as that of the national per capita GDP over 1988–2007 are illustrated. NCR’s (Region 13) extremely well-off status relative to the other regions, even relative to the national benchmark, is clearly showed in Figure 2. NCR continues to have by far the highest per capita income, and its increasing income differential relative to the national average and to those of the other regions persists over the period 1988–2007. The national per capita GDP and the per capita GRDP of the remaining regions do not seem to converge toward the per capita GRDP of NCR over time. In Figure 3, NCR is excluded in the plot to better examine the relative transitions of per capita GRDP of the remaining regions and that of the national GDP. Also, the plot would show the trend if the individual paths tend to fluctuate around the national benchmark or move toward a common limit. In general, some of the regions appear to follow the path of the national per capita GDP, but the overall economic progress of the regions differs substantially as the paths of their per capita GRDP do not follow a common trend nor show any

10 The ARMM is the region of the Philippines that is composed of all the Philippines’ predominantly Muslim provinces (used to be called Central Mindanao), namely: Basilan (except Isabela City), Lanao del Sur, Maguindanao, Sulu, and Tawi-Tawi, and the Philippines’ only predominantly Muslim city, the Islamic City of Marawi. The regional capital is Cotabato City, although this city is outside of its jurisdiction.


0

5000

10000

15000

20000

25000

30000

35000

40000

45000

1988 1990 1992 1994 1996 1998 2000 2002 2004 2006

in p

esos

GDPreg1reg2reg3reg4reg5reg6reg7reg8reg9reg10reg11reg12reg13reg14

Figure 2. Per capita national GDP and per capita GRDP of 14 regions, 1988–2007 (in PHP)

Source of basic data: NSCB

Figure 3. Per capita national GDP and per capita GRDP of regions excluding NCR, 1988–2007 (in PHP)

0

5000

10000

15000

20000

25000

1988 1990 1992 1994 1996 1998 2000 2002 2004 2006

in p

esos

GDP

reg1

reg2

reg3

reg4

reg5

reg6

reg7

reg8

reg9

reg10

reg11

reg12

reg14



evidence of convergence through time. More so, some of the paths depict high volatilities in per capita GRDP.

The growth rates of national per capita GDP and the per capita GRDP of the 14 regions are shown in Figure 4, and fluctuations around some common path are apparent. Although per capita GRDP growth rates of a few regions depict high volatilities, in general, the trends relatively suggest the possibility of convergence over time with the national benchmark and among the regions.

STOCHASTIC AND DYNAMIC CONVERGENCE

Stochastic convergenceStochastic convergence is tested using LLC and IPS panel unit root tests. Panel unit root tests are similar, but not identical, to the unit root tests performed on individual series. The panel unit root test considers an Autoregressive model of order one [AR(1)] on the panel data,

(1)

i= 1,2,…,14 regional units and t = 1,….,T (years)

-30

-20

-10

0

10

20

30

40

50

1988

-89

1989

-90

1990

-91

1991

-92

1992

-93

1993

-94

1994

-95

1995

-96

1996

-97

1997

-98

1998

-99

1999

-2000

2000

-01

2001

-02

2002

-03

2003

-04

2004

-05

2005

-06

2006

-07

grow

th ra

te (i

n pe

rcen

t)

reg1reg2reg3reg4reg5reg6reg7reg8reg9reg10reg11reg12reg13reg14GDP

Figure 4. Growth rates of national per capita GDP and per capita GRDP of 14 regions, 1988–2007 (in percent)



where yit is the natural logarithm of the ratio of the per capita GRDP of region i with the average per capita national GDP; the ρi are the autoregressive coefficients; and the errors εit are assumed to be mutually independent idiosyncratic disturbance. If ρi< 1, yi is said to be weakly stationary; on the other hand, if ρi> 1 then yi contains a unit root.

For purposes of testing, there are two natural assumptions that we can make about ρi. First, one can assume that the persistence parameters are common across cross-sections so that ρi = ρ for all i. The LLC test employs this assumption. Alternatively, one can allow ρi to vary freely across cross-sections. The IPS test is of this form. Both tests have a null hypothesis of unit root and rejection of the null hypothesis implies stochastic convergence indicating that disparities in per capita GRDP among regions follow a stationary process, and the differences in the natural logarithm of the per capita GRDP of region i and the average national GDP remain constant. Nonrejection leads to nonstationarity and might lead to permanent deviations from the equilibrium level of per capita GRDP.

Before performing panel unit root tests, the study first examines the individual convergence behavior of regions using Augmented Dickey-Fuller (ADF) Test.11 As provided in the work of Chou (2006), if an individual series is stationary, it is said to achieve stochastic convergence. Hence, the study carries out univariate unit root test for each region using the natural logarithm of the ratio of the per capita GRDP of a region i to the national (average) per capita GDP at a given year. The rejection of unit root indicates convergence of per capita GRDP to the national benchmark. However, as LLC and IPS showed, the panel unit root tests are more powerful than the univariate tests of unit root. This study will make use of the LLC and IPS tests to check if all 14 regions achieve stochastic convergence.

Dynamic convergenceTo measure the dynamic process of convergence in the growth rates of GRDP of the regions over the sample period, the study makes use of the following TVP model based on the formulation of Chou (2006),

( ) ( ) ( )[ ]GR GR t t GR GRb i i i b BASE it− = + − +α β ε (2)

where GR = growth in GRDP; b = national GDP; i = particular region; t = time (year); and BASE = region with the highest average yearly GRDP in the series and is taken to be the NCR.

11 ADF Test is one of the many popular univariate tests for presence of unit root.


In estimating the time-varying parameters αi t( ) and βi t( ) , Kalman filter is used. The equation in (2) is the signal equation in the state-space model. The two time-varying parameters, αi(t) and βi(t) can be represented by the state equations, defined as,

State equations:(i) α α να αi i i tt F t( ) ( )= − +1 , with var( )ν σα ναi t i= 2 (3)

(ii) β β νβ βi i i tt F t( ) ( )= − +1 , with var( )ν σβ νβi t i= 2 (4)

Fα and Fβ are the corresponding coefficients of the state-space equations, and ν να βi t i t, are the random errors.

The estimates of the time-varying parameter βi t( ) will determine the convergence or divergence behavior of region i. The study accomplishes this by first estimating the smoothed βi t( ) coefficients for each region over the sample period and computing for the corresponding means and standard deviations of these coefficients per region. If the region’s GRDP growth converges toward the growth rate of the national per capita GDP, the expected mean of βi t( ) for this particular region is zero. If, however, the region’s growth in GRDP converges toward NCR, the lead region, the expected mean of βi t( ) for that particular region is one. A negative value for the mean of βi t( ) signifies that the region is lagging behind the growth rate of the national benchmark. The estimated standard deviation of βi t( ) would indicate volatility of the estimated βi t( ) coefficients for a given region, which, in turn, provides an overview of how a region’s GRDP fluctuates relative to that of the other regions over time. Furthermore, the regions can be subdivided into groups based on the estimated mean βi t( ) coefficient, and the dynamic behaviors of regions within and across groups are analyzed in the same manner.

EMPIRICAL RESULTSResults of ADF regressions as well as LLC and IPS panel unit root tests are summarized in Table 4. The results of the LLC and IPS panel unit root tests indicate lack of convergence of the natural logarithm of per capita income GRDP of the 14 regions with the average national GDP in the long run, as the null hypothesis of nonstationarity is not rejected, indicating divergence in the per capita incomes of the regions.

The individual ADF tests for Bicol Region, Central Visayas, and Southern Mindanao are significant at the 5 percent level, as indicated by the corresponding p-values, implying presence of stochastic convergence and that the per capita GRDP of the respective regions converges toward the national per capita GDP over time. The ADF tests for the remaining regions, on the other hand, show


presence of unit root denoting lack of convergence toward the national benchmark through time. However, as indicated previously, the panel unit root tests based on LLC and IPS are more powerful than the individual ADF tests and the results of the panel unit root tests will be used in the study.

Although there is absence of stochastic convergence, this does not imply that there is total lack of convergence among regions. It may be possible that convergence exists within subgroups of regions, or some regions converge either toward the national per capita GDP growth rate or toward the growth rate of the lead region, NCR, over the sample period. The assessment of the dynamic process of convergence of individual regions is determined via the TVP model, specified in the previous section, using Kalman filter. NCR is disregarded in the subsequent analyses as it is the lead region.

The smoothed βi t( ) coefficients for 13 regions over 1988–2007 are provided in Table 5. The mean and standard deviations of these smoothed coefficients for each region have also been provided. Since the estimated values of the βi t( ) coefficients are close to zero, it would be essential to test if the coefficients are significantly different from zero or not. The t-statistics and p-values of the one-sided test under the null hypothesis that the mean of the βi t( ) coefficients is not significantly different from zero at the 5 percent level are also provided in Table 5.

From the results, we see that seven regions obtained mean βi t( ) coefficients that indicate convergence toward the growth rate of the national per capita GDP over the sample period. The remaining six regions, however, gained negative mean βi t( ) coefficients that are significantly different from zero indicating that

Region Name ADF t-stat P-value

Ilocos RegionCagayan ValleyCentral LuzonSouthern TagalogBicol RegionWestern VisayasCentral VisayasEastern VisayasWestern MindanaoNorthern MindanaoSouthern MindanaoCentral MindanaoNational Capital RegionCordillera Administrative RegionPanel Unit Root TestLLC IPS

-2.0273-1.9768-1.8989-2.4648-6.9728-3.4317-4.9418-2.0131-1.3670-3.0987-3.7684-2.6039-1.9031-1.3553

LLC t-stat/IPS w-stat-0.03600.6030

0.55040.57620.61570.33910.00030.07840.00700.55770.83720.13440.04210.28230.61150.8386

P-value0.48560.7267

Table 4. Unit root tests on stochastic convergence

Source: Authors’ calculations.

13M

apa, Sando

val, and Yap

YearIlocosRegion

Cagayan Valley

CentralLuzon

Southern Tagalog

BicolRegion

Western Visayas

Central Visayas

19881989199019911992199319941995199619971998199920002001200220032004200520062007

MeanS.D.t-stat

(p-value)

0.000000-0.000016-0.0000160.0000010.0009180.0001170.0001220.000449

-0.00000040.0002620.0001720.000170

-0.000050-0.000022-0.000037-0.000258-0.0001220.0000440.000059

-0.000077

0.0000860.0002461.560

(0.064)

0.000000-0.000009-0.0000090.0000180.000874

-0.0004410.0001010.000011

-0.0000010.000408

-0.000139-0.000745-0.000111-0.0000370.000203

-0.0003600.000416

-0.0013130.000198

-0.000017

-0.0000480.000446

-0.477(0.318)

0.000000-0.000012-0.0000120.000008

-0.0009480.000090

-0.000045-0.000031-0.0000003-0.000059-0.0001660.0000520.000074

-0.000072-0.000073-0.000161-0.000564-0.000328-0.000054-0.000062

-0.0001180.000245

-2.158(0.019)

0.000000-0.000003-0.000003-0.000002-0.0004180.0003590.000001

-0.0000500.00000025

-0.000058-0.0000380.0000860.0002690.000050

-0.000049-0.000221-0.0004250.000218

-0.000153-0.000076

-0.000026-0.000190-0.603(0.275)

0.000000-0.0000005-0.0000005-0.000003-0.000075-0.000867-0.000056-0.000250-0.00000020.000079

-0.0000300.000069

-0.0000480.000053

-0.000080-0.000079-0.0000650.000039

-0.0002360.000030

-0.0000760.000205

-1.657(0.053)

0.0000000.0000060.000006

-0.000001-0.000843-0.000485-0.000057-0.0003260.00000007

-0.0003070.000033

-0.000125-0.0001240.000047

-0.000046-0.0000800.0001810.000019

-0.0000390.000032

-0.0001060.000230

-2.051(0.024)

0.000000-0.000002-0.000002-0.000002-0.0000460.000145

-0.0000100.0001000.000000480.0000740.000079

-0.000012-0.0000630.0000260.000034

-0.0000970.0001020.000146

-0.0000540.000081

0.0000250.0000681.631

(0.056)

Table 5. Smoothed βi t( ) coefficients of 13 regions from 1988 to 2007

14P

hiliPPine Journal o

f Develo

Pment 2010

YearEasternVisayas

WesternMindanao

NorthernMindanao

SouthernMindanao

CentralMindanao

CAR

19881989199019911992199319941995199619971998199920002001200220032004200520062007

MeanS.D.t-stat

(p-value)

0.0000000.0000110.000011

-0.0000050.000266

-0.000780-0.000059-0.000128-0.00000012-0.0001100.000050

-0.000005-0.000124-0.0000420.000022

-0.000139-0.000135-0.000246-0.000021-0.000226

-0.0000830.000198

-1.876(0.034)

0.000000-0.000006-0.0000060.000003

-0.000273-0.000556-0.0001240.0011320.000001

-0.0002630.0000360.000137

-0.0000660.0002610.000084

-0.000150-0.0002770.000279

-0.0002910.000008

-0.0000030.000333

-0.047(0.481)

0.0000000.0000140.000014

-0.000001-0.000010-0.0000700.0000060.000064

-0.000001-0.000287-0.0000890.000034

-0.000175-0.000492-0.000642-0.0002260.000004

-0.0000880.0001110.000044

-0.0000890.000192

-2.087(0.022)

0.0000000.0000110.000011

-0.0000020.0004430.000306

-0.0000470.000760

-0.000002-0.0001260.000120

-0.000071-0.0000340.0011180.0012070.0001380.000083

-0.000066-0.000061-0.000005

0.0001890.0003932.152

(0.019)

0.0000000.0000190.0000190.0000050.0001310.000506

-0.000077-0.000995-0.00000018-0.000221-0.000006-0.0000240.000113

-0.000772-0.001048-0.000114-0.000107-0.0003790.000023

-0.000057

-0.0001490.000381

-1.754(0.044)

0.0000000.0000030.000003

-0.0000150.000534

-0.0003260.000187

-0.000366-0.000000240.0007580.000141

-0.000546-0.000110-0.000099-0.000010-0.000292-0.000296-0.000613-0.000168-0.000005

-0.0000610.000323

-0.847(0.201)

Table 5. (cont'd.)



these six lag behind the growth rate of the national benchmark over 1988–2007. None of the growth rates of the regions converge toward the growth rate of the lead region, NCR.

The region with the highest βi t( ) standard deviation is Region 2, Cagayan Valley. This suggests high volatility for smoothed βi t( ) coefficients; hence greater instability in per capita GRDP growth rate of the region relative to the growth rate of national per capita GDP.

We plot the smoothed βi t( ) coefficients of the 13 regions as shown in Figure 5. The plot shows large fluctuations in per capita GRDP over the sample period for a number of regions, specifically those for Regions 2, 11, and 12.

The study then subdivides the 13 regions based on the calculated smoothed βi t( ) coefficients. The first group consists of those regions whose growth rates in the per capita GRDP converge toward the growth rate of national per capita GDP, the national benchmark. The second group consists of those regions with growth rates lagging behind the average national growth rate. The groupings are provided in Table 6.

Figures 6 and 7 illustrate the behaviors of the regions within subgroups with respect to per capita GRDP growth rate over the sample period. In Figure 6, the high volatilities exhibited by Regions 2 and 11 are further emphasized. But, in general, per capita GRDP growth rates of the seven regions tend to fluctuate around a common path and would further suggest convergence with respect to per

-0.001500

-0.001000

-0.000500

0.000000

0.000500

0.001000

0.001500

smoo

thed

bet

as

reg1

reg2

reg3

reg4

reg5

reg6

reg7

reg8

reg9

Figure 5. Smoothed βi t( ) coefficients of 13 regions over 1988–2007



capita GRDP beyond 2005. For the lagging regions in Figure 7, Regions 3, 6, and 8 exhibited extreme declines in per capita GRDP growth rate around the years 1991–1993. Region 12, similarly, may have had experienced significant declines in per capita GRDP in the years 1995, 2001, and 2002. More so, by 2006, the six laggards clearly showed lack of convergence in per capita GRDP growth rate.

CONCLUSIONSThis paper looks at the economic behavior of the Philippine regions in terms of the levels and growth rates in per capita GRDP over the time period 1988–2007. Convergence in the levels of per capita GRDP and the growth rates in the per capita GRDP are evaluated using the concept of stochastic and dynamic convergence, respectively. The panel unit root tests for stochastic convergence

Regions Converging toward Mean GDP Growth Rate Diverging Regions

Ilocos RegionCagayan ValleySouthern TagalogCentral VisayasWestern MindanaoSouthern MindanaoCAR

Central LuzonWestern VisayasEastern VisayasNorthern MindanaoCentral MindanaoBicol Region

Table 6. Regional subgroups based on smoothed βi t( ) coefficients

-0.001500

-0.001000

-0.000500

0.000000

0.000500

0.001000

0.001500

smoo

thed

bet

as

reg1

reg2

reg4

reg7

reg9

reg11

reg14

Figure 6. Smoothed βi t( ) coefficients of seven regions converging toward growth rate of national per capita GDP over 1988–2007



-0.001200

-0.001000

-0.000800

-0.000600

-0.000400

-0.000200

0.000000

0.000200

0.000400

0.000600

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

smoo

thed

bet

as

reg3

reg6

reg8

reg10

reg12

reg5

Figure 7. Smoothed βi t( ) coefficients of six lagging regions over 1988–2007


showed that the per capita GRDP of the 14 regions do not converge to the average national per capita GDP in the long run. The analysis of dynamic convergence, however, revealed that growth rates in the per capita GRDP of seven regions converge toward the growth rate of national per capita GDP; these are Ilocos Region, Cagayan Valley, Southern Tagalog, Central Visayas, Western Mindanao, Southern Mindanao, and CAR. The study also showed that the growth rates in the per capita GRDP of six regions lag behind and do not converge toward the growth rate of the national benchmark; these regions are Central Luzon, Western Visayas, Eastern Visayas, Northern Mindanao, Central Mindanao, and the Bicol Region. The results also showed that none of the growth rates in the per capita GRDP of the 13 regions converge toward the growth rate in the per capita GRDP of the NCR.

The results reiterate the need to improve the socioeconomic infrastructure of regions other than the NCR. The fact that no region is converging toward the growth rate of the NCR is a reminder of the severe systemic imbalance and the staggering amount of untapped economic potential. The creation and implementation of economic policies that seek to generate economic growth outside of NCR should be pursued by national and regional leaders.

The results also indicate that almost half of Philippine regions are growing slowly relative to the overall growth rate of Philippine regions. Furthermore, these regions are constrained in significantly improving their economic condition because of the generated regional convergence profile. It is thus incumbent


upon regional and national leaders to immediately develop and implement socioeconomic policies that are aimed at generating sustainable economic growth in these six regions.

The Philippine regions, 1988

Region I: IlocosBenguetIlocos NorteIlocos SurLa UnionMt. ProvincePangasinan

Region II: Cagayan ValleyBatanesCagayanIsabelaNueva VizcayaQuirino

Region III: Central LuzonBataanBulacanNueva EcijaPampangaTarlacZambales

Region IV: Southern TagalogAuroraBatangasCaviteLagunaMarinduqueOccidental MindoroOriental MindoroPalawanQuezonRizalRomblon

Region V: BicolAlbayCamarines NorteCamarines SurCatanduanesMasbateSorsogon

Region VI: Western VisayasIloiloCapiz AklanAntiqueNegros Occidental

Region VII: Central VisayasBoholCebuNegros OrientalSiquijor

Region VIII: Eastern VisayasLeyteSouthern LeyteNorthern SamarWestern SamarEastern Samar

Region IX: Western MindanaoBasilanSuluTawi-Tawi Zamboanga del Norte Zamboanga del Sur

Region X: Central MindanaoAgusan del NorteAgusan del SurBukidnonCamiguinMisamis OccidentalMisamis OrientalSurigao del Norte

Region XI: Southern MindanaoDavao del NorteDavao del SurDavao OrientalSouth CotabatoSurigao del Sur

Region XII: Central MindanaoLanao del NorteLanao del SurMaguindanaoNorth CotabatoSultan Kudarat

Region XIII:National Capital Region (NCR)

Cordillera Administrative Region (CAR)AbraIfugaoKalingaApayaoBenguetMt. Province


The Philippine regions, 2007

Region I: IlocosIlocos NorteIlocos SurLa UnionPangasinan

Region II: Cagayan ValleyBatanesCagayanIsabelaNueva VizcayaQuirino

Region III: Central LuzonAuroraBataanBulacanNueva EcijaPampangaTarlacZambales

Region IV-A: CALABARZONCaviteLagunaBatangasRizalQuezon

Region IV-B: MIMAROPAMindoro OrientalMindoro OccidentalMarinduqueRomblonPalawan

Region V: BicolAlbayCamarines NorteCamarines SurCatanduanesMasbateSorsogon

Region VI: Western VisayasIloiloCapizAklanAntique Negros Occidental

Region VII: Central VisayasBoholCebuNegros OrientalSiquijor

Region VIII: Eastern VisayasLeyteSouthern LeyteNorthern SamarWestern SamarEastern Samar

Region IX: Western MindanaoZamboanga del SurZamboanga del NorteZamboanga Sibugay

Region X: Northern MindanaoBukidnonCamiguinMisamis OccidentalMisamis Oriental

Region XI: Southern MindanaoDavao del NorteDavao del SurDavao OrientalCompostela Valley

Region XII: Central MindanaoLanao del NorteNorth CotabatoSultan KudaratSaranganiSouth Cotabato

National Capital Region(NCR)

Cordillera Administrative Region (CAR)AbraApayaoBenguetIfugaoKalingaMt. Province

Autonomous Region in Muslim Mindanao(ARMM)Basilan SuluTawi-TawiLanao del SurMaguindanao

CaragaAgusan del NorteAgusan del SurSurigao del NorteSurigao del Sur

Source: NSCB


REFERENCESHill, H., A. Balisacan, and S. Piza. 2007. The Philippines and regional

development. In A. Balisacan and H. Hill (eds), The dynamics of regional development: the Philippines in East Asia. Quezon City: Ateneo De Manila University Press.

Chou, W.L. 2006. Convergence of service sector employment in China: panel data evidence. Discussion paper presented at the 10th International Convention of the East Asian Economic Association, 18–19 November 2006, Beijing.

Im, K.S., M.H. Pesaran, and Y. Shin. 2003. Testing for unit roots in heterogeneous panels. Journal of Econometrics 115:53–74.

Krugman, P. 1991. Geography and trade. Cambridge, M.A.: MIT Press.Levin, A., C. Lin, and C. Chu. 2002. Unit root tests in panel data: asymptotic and

finite-sample properties. Journal of Econometrics 108:1–24.Manasan, R. and R. Mercado. 1999. Regional economic growth and convergence

in the Philippines: 1975–1997. PIDS Discussion Paper Series No. 99-13. Makati City: Philippine Institute for Development Studies.

investigating the presence of regional economic growth ... · at an average of 1.73 percent per...

Documents