ODA FDI Asean Tu Thuy_Vu Thui

8/9/2019 ODA FDI Asean Tu Thuy_Vu Thui

1/45

1

On the Impacts of ODA on FDI:

Does Composition of FDI Matter?

Evidence from Asean Countries1

Tu Thuy Anh2 and Vu Thi Phuong Mai

3

December 2012

Superviser: Dr. Anirudh Shingal

1. Introduction

Majority of studies show that Foreign Direct Investment (FDI) can playa fundamental role by reinforcing the capacity of the recipient country and

using opportunities given by a global economic integration. It is well-known

that a good preparation in infrastructures, in human resources and in

technology is necessary to attract FDI a given country. However, most of

developing countries appear to have limited social and economic conditions,

limited technology, and low quality of the human resources in comparison with

the developed countries. To improve this situation, strong government

expenditure is needed. In case that the government’s budget is not sufficient,

outside financial resource allocation could be a good compensation. Obviously,

this financing method brings an optimum resolution for the developing

1 We would like to acknowledge the SECO-WTI Academic Cooperation Project for funding the

visiting fellowship and twinning agreement under which this research has been undertaken. We wouldalso like to thank Anirudh Shingal for his supervision and Pierre Sauvé for his comments and

suggestions.2 Vietnam Foreign Trade University ([email protected])

3 Vietnam Foreign Trade University ([email protected])

SECO / WTI Academic Cooperation Project Working Paper Series 9 / 2012


2/45

2

countries: use the Official Development Assistance (ODA) as a tool to promote

FDI and economic growth, especially of developing countries.

Association of Southeast Asian Nations (ASEAN) represents one of the

economic regions being known as experiencing very high economic growth

starting from 1980s. Moreover, this economic region is considered to be “all

the developing countries are the best success in attracting FDI and inserting the

foreign firms into national strategies of development” (OECD, 2004, p.81).

Besides, the ODA flows to ASEAN countries play also a significant role. In

fact, the ODA help these countries to improve our levels of infrastructure

development and the provision of essential social services such as health and

education which are important to attract more the FDI. Our research sheds light

on the interaction between ODA and FDI in the ASEAN countries.

In addition to the humanitarian needs, the objective of ODA is to

promote economic and social development of recipient countries. ODA flows

are therefore dependent upon the extent of the needs of the recipient in terms of

development assistance and its ability to use the assistance of effective ways,

rather than its locational advantages economically compared to other countries.

In contrast, the objective of FDI is seeking benefits for companies. Therefore,

the ability of countries to attract FDI depends on its locational advantages

(market size, abundant resources ...) compared to other countries. Benefits such

as market size and cost competitiveness tend to improve economic

development and growth.

In 2005, over 30% of ODA from DAC has been allocated to education,

health, production and other social infrastructure. The economic infrastructure

and welcomed about 11% of production (agriculture and manufacturing) about

5%. The rest went to multisector activities, assistance programs, humanitarian

assistance and activities not specified. At the same time, the inward stock of

FDI in developing countries is distributed among services (58%), including

services to producers but also consumers (the most important sectors such as

business operations, financial services, the Trade, transportation, storage and



3/45

3

communications), the manufacturing sector (31%) and the primary sector (9%)

(UNCTAD, 2007, p.11).

ODA therefore tends to encourage more FDI as investment in education,

health, population and social infrastructure that play an essential role in the

formation of human capital and human development. In contrast, FDI is

targeting production in mining operations, manufacturing and, increasingly,

producer services and infrastructure services that contribute significantly to

other sectors including telecommunications, trade, financial and business

services.

So, we found that the coexistence of FDI and ODA in the utilities

(electricity, gas and water), transport and storage. However, ODA provides

base supports the attractiveness of FDI. Indeed, the efficient allocation of ODA

to the development of human resources, infrastructure development and

capacity building of enterprises in recipient countries. On the other hand, a

good environment of the host country based on the quality of human resources,

infrastructure and capabilities of indigenous firms is the sine-qua-non for

attracting FDI.

We know that with the help of ODA, social and economic

infrastructures are improved. For the direct channel, a good quality of

infrastructure directly contributes to improving the investment environment.

For the indirect channel, the ODA contributes to reform the quality of human

resources. In turn, a high quality workforce encourages promotion of income,

and an increase in consumption and market size. In sum, all direct and indirect

impacts of ODA positively affect the investment environment to better attract

FDI and vice versa. This idea leads to an optimal solution to use ODA as an

instrument to promote FDI in developing countries. In reality, this idea is

already developed not only by the recipient countries but also by development

assistance agencies. Recognizing that the private sector is the primary source of

employment and advance the incomes of the poor, and thus generate income

necessary to allow governments to expand access to health care, education and

infrastructure and contribute to improving productivity and creating growth



4/45

4

including poor and good for them, aid agencies thus guide the development of

increasing ODA to the efforts in encouraging countries development to address

market failures and overcome structural barriers hampering private investment.

The challenge is to ensure that ODA and FDI are complementary. To this end,

first, it should increase aid significantly in meeting the needs of attractiveness

of FDI, on the other hand, it would maximize the benefits of FDI to "ensure

that investments made in these countries translate into lasting gains for

development "(UNCTAD, 2007, p.12).

In conclusion, recall the idea of Ragnar Nurske who believes that "a

country is poor because he is poor" (1953). This is a really a vicious circle for

most LDCs. According to Elsa Assidon (2000, p.13), to break this circle, "the

support of external assistance was needed (...)". To argue, she said that "Europe

itself had she not resorted to the Marshall Plan to reboot its growth after the

war?

There are a few studies that examine the relation between foreign aid

and FDI by using cross-country panel data, most notably Harms and Lutz

(2006) and Karakaplan, Neyapti, and Sayek (2005). Karakaplan et al. (2005)

find that aid has a negative direct effect on FDI and that both good governance

and financial market development significantly improve the impact of aid on

subsequent flows of FDI. Harms and Lutz (2006), on the other hand, find that

once they control for the regulatory burden in the host country, aid works as a

complement to FDI and, surprisingly, that the catalyzing effect of foreign aid is

stronger in countries that are characterized by an unfavorable institutional

environment.

There are also some important papers analyzing specifically the relation

between Japanese FDI and aid flows. Among these, Blaise (2005) finds

positive effects from aid to infrastructure projects, while Kimura and Todo

(2010) find no positive infrastructure effect, no negative rent-seeking effect but

a positive vanguard effect (arising when foreign aid from a particular donor

country promotes FDI from the same country but not from other countries).



5/45

5

The relationship between aid and FDI appears to be controversial and

literature on this underlying relation remains inconclusive. This type of mixed

empirical results can be explained to a large extent by the high level of

aggregation used for the aid variable. Karakaplan et al. (2005) includes only

overall ODA. Harms and Lutz (2006) distinguish between grants, technical

cooperation grants, as well as bilateral and multilateral aid. Kimura and Todo

(2010) apply the idea of different types of aid but do not implement an

effective disaggregation.

There exist two major shortcomings in this literature. Firstly, none of the

existing papers consider FDI at the disaggregate level. Secondly, there is the

lack of a supporting theoretical model. To our knowledge, there are only two

papers analyzing theoretically the relationship between aid and FDI. The first

one is Beladi and Oladi (2007) that set up a general equilibrium model where

all foreign aid is used to finance public goods, but where they unfortunately do

not consider any further disaggregation for the aid flows nor make an empirical

analysis. The second and most remarkable one is Selaya and Sunesen (2012)

that developed a theoretical model explaining the ambiguous relation between

aid and FDI and test this relation empirically.

The originality of our paper is to fill in exactly these two shortcomings.

First, we modify Selaya’s theoretical model by introducing disaggregated FDI

and evaluate theoretically impacts of not only disaggregated ODA but also FDI

invested in complementary factors on FDI invested in physical capital – the

major type of FDI we usually observe. Second, we fit the theoretical

framework to ASEAN’s data to shed lights on this underlying relation at the

South East region’s level.

The paper is organized as follows. The theoretical is derived in section

2. A overall analysis of FDI and ODA in ASEAN countries is described in

section 3. The empirical model is constructed in section 4. Data are overviewed

in section 5. Empirical results are discussed in section 6. Conclusion remards

finalize the paper.



6/45

6

2. Theoretical model

As suggested by Selaya and Sunesen’s model (2012), assume a Solow

model for a small open economy. Assume a Cobb-Douglas production function

where GDP per capita, y, is given by

y Ak α = (1)

where k is the stock of physical capital per capita (K/L), A denotes total factor

productivity and α is a constant.

Assume that the total flow of foreign aid, AID, can be split into aid

invested in complementary factors, A

AID , and aid invested in physical capital,

K AID , so that A K AID AID AID= +

. The part invested in complementaryfactors,

A AID , raises the marginal productivity of all production factors that are

complementary to physical capital. Aid to complementary factors helps for

example to finance infrastructure investments that lead to the interconnection

of markets, or investments in human capital improve technology adoption. On

the other hand, aid invested in physical capital,K

AID , enters the production

function only through its effect on physical capital accumulation and has no

(augmenting) effect on total factor productivity.

To model the augmenting effect of complementary aid on all production

factors that are complementary to physical capital, we allow the flow of A

AID

and AFDI to increase the existing stock of A ( 0 AID ) in the economy:

0 A A A A AID FDI = + + (2)

Allowing complementary aid as well as investment to have a direct

impact on A is for the idea that A AID and AFDI has an augmenting effect on

any production factor other than k (e.g. human capital, public investments, new

technology, etc.) and, thus, it is ultimately able to increase the MPK (marginal

product of capital).

We assume an open economy, therefore capital equipment is financed

by (i) domestic savings (S=sy, where s is a given savings rate); (ii) foreign

direct investments in physical capital ( K fdi ) and (iii) the part of aid invested in



7/45

7

physical capital ( K aid ). Then, capital accumulation in per capita terms is given

by

.

( )K K k sy fdi aid n k δ = + + − + (3)

where n is the population growth rate and δ is a fixed depreciation rate.

With perfect capital mobility, the world real rate of return, wr , pins

down at any point in time the net return to capital (MPK-δ), and thus

1wr MPK A k α δ α δ −= − = − (4)

According to (4), the steady state level of k at any point in time is given

by

11

* Ak

r

α α − =

(5)

where r is defined as a gross world real rate of return, wr δ + .

Rewriting (3) taking (5) as given, the flow of FDI per capita is

determined as the residual

* ( ) *K K

fdi aid sy n k δ = − − + + (6)

where * * y Ak α

= .

Hence, the empirical relationship between aid and FDI are not

monotonic:

K K K K K A A

K A A

fdi fdi fdidfdi daid daid dfdi

aid aid fdi

∂ ∂ ∂= + +

∂ ∂ ∂.

The above equation holds several implications. First and the most

obviously, K aid and K fdi are substitutes ( 1K

K

fdi

aid

∂= −

∂

). Second, since

Aaid and A fdi are perfectly substitutes, its impacts on K fdi are in the same

direction. Since*

A

ys

aid

∂−

∂(or

*

A

ys

fdi

∂−

∂) and

*( )

A

k n

aid δ

∂+

∂(or

*( )

A

k n

fdiδ

∂+

∂)

work in opposite directions, the sign of the second and third effects will be

ambiguous. The total impacts of three effects onK

fdi are therefore

indeterminate and there is room for empirical works to withdraw the final

conclusion:



8/45

8

0K K K K K A A

K A A

fdi fdi fdidfdi daid daid dfdi

aid aid fdi

∂ ∂ ∂ >= + +

∂ ∂ ∂ < (7)

If mobility of capital is imperfect, MPK should be allowed to deviate

from the gross world interest rate by a risk-premium, ρ , that reflectsidiosyncratic country characteristics. In this case, the first-order condition in

(4) should read

r MPK ρ + = , (8)

and the capital stock in (5) should be redefined accordingly:

1

1

* A

k r

α α

ρ

− = +

. (9)

While this renders the effect of aid invested in physical capital

unchanged, the effect of complementary aid becomes somewhat more

complicated. The risk premium impacts FDI directly through (9) but, given

that the marginal effect of Aaid and A fdi will also depend on the risk

premium and thus on country-specific characteristics.

3. FDI and ODA in ASEAN countries: an overview

Figure 1 shows the evolution of FDI inflows and ODA flows to

developing countries over the period 1970-2010. In this figure, FDI takes the

primary axis, while ODA commitment and ODA disbursement take the

secondary axis. In fact, total ODA increased steadily for twenty years to over

60 billion dollars in 1991, almost twice the 1980 figure (33 billion). During the

same period, FDI has increased fivefold (from 7.5 billion in 1980 to 40 billion

in 1991).

Figure 1: Inflows of FDI and ODA flows to developing countries, total

(1970-2010) (USD millions)



9/45

9

Source: database on ODA: http://stats.oecd.org; on FDI: http://stats.unctad.org

Since 1992, while FDI flows have continued to progress, ODA has

declined. In real terms, total aid to LDCs in 2002, 17 billion, was 16% lower

than the figures achieved in the early 90s. The figures for development

assistance in 2007 are far from the targets (the quantitative goal for 2015 set

under the Millennium Development is 0.7%). The official development aid

distributed by all DAC members totaled USD 103.7 billion, marking a decline

of 8.4% in real terms over 2006, according to provisional data reported by

members. At a cost equivalent to 0.31% of gross national income (GNI)

accumulated in these countries in 2006, the share of ODA in GNI in 2007 was

only 0.28%. In other words, for developing countries, it is extremely difficult to

predict the flow of aid they receive. Furthermore, studies have shown that aid

tends to increase during periods of expansion and become scarce during

periods of slowdown or recession. This means that the volume of aid has

tended to contract for specific times when developing countries most in need.

In this regard, the need to increase aid effectiveness and improve the coherence

and harmonization, with a view to increasing the overall volume of aid is

considered more than ever a key objective of recipient countries and aid

agencies. In comparison, FDI inflowing into developing countries has tripled

compared to 2002. These figures allow us to conclude that flows of ODA and

FDI have evolved in different ways over the last fifteen years.



10/45

10

Turning to FDI flows to South-East Asia or the ASEAN sub region

increased by 18% in 2007, to $61 billions of US dollars – resulting in yet

another year of robust FDI growth there. The evolution of FDI and ODA to the

South-East Asia region is shown in figure 2 that follows similar construction

method as figure 1.

Figure 2: FDI and ODA flows in ASEAN (1970-20010)


One sees clearly that nearly all ASEAN countries received higherinflows. Singapore, Thailand, Malaysia, Indonesia and Viet Nam, in that order,

were the largest FDI recipients, together accounting for more than 90% of

flows to the sub region. While FDI growth in 2007 differed considerably

between countries, the newer ASEAN member countries in particular

(Myanmar, Viet Nam, Cambodia and the Lao People’s Democratic Republic,

in that order) recorded the strongest FDI growth, exceeding 70% in each

(World Investment Report, 2008). Favorable regional economic growth, an

improved investment environment, higher intraregional investments, and

strengthened regional integration were key contributory factors. Reinvested

earnings were particularly strong, highlighting the importance of existing

investors as a source of FDI. Increased inflows in Viet Nam were the result of

that country’s accession to the World Trade Organization (WTO) in 2007, as

well as greater liberalization and FDI promotion efforts, particularly with

respect to infrastructure FDI. There were higher FDI inflows in extractive



11/45

11

industries in Myanmar, in telecommunications and textiles and garments

manufacture in Cambodia, and in agriculture, finance and manufacturing in the

Lao People’s Democratic Republic.

Figure 3. FDI inflows and ODA to ASEAN countries, 1970-2010(Millions USD)


In 2010, FDI to ASEAN surged to $79 billion, surpassing 2007’s

previous record of $76 billion. The increase was driven by sharp rises in

inflows to Malaysia (537 per cent), Indonesia (173 per cent) and Singapore

(153 per cent). Proactive policy efforts at the country level contributed to the

good performance of the region, and seem likely to continue to do so: in 2010,

Cambodia, Indonesia improved its FDI-related administrative procedures; and

the Philippines strengthened the supportive services for public-private

partnerships (PPPs).

Figure 4: Percentage of FDI and ODA to ASEAN in 2010 by country


http://stats.unctad.org/http://stats.unctad.org/http://stats.unctad.org/http://stats.unctad.org/


12/45

12


In Singapore, which accounted for half of ASEAN’s FDI, inflows

amounted to a historic level of $39 billion in 2010 (figure 4). As a global

financial centre and a regional hub of TNC headquarters, the island State has

benefited considerably from increasing investment in developing Asia, against

a background of rising capital flows to the emerging economies in general in

the post-crisis era. Due to rising production costs in China, some ASEAN

countries, such as Indonesia and Vietnam, have gained ground as low-cost production locations, especially for low-end manufacturing. ASEAN LDCs

also received increasing inflows, particularly from neighboring countries like

China and Thailand. For instance, the Lao People’s Democratic Republic has

been successful in attracting foreign investment in infrastructure in recent

years; as a result of Chinese investment in an international high-speed rail

network, FDI to the country is likely to boom in the coming years.

Besides, the sum of ODA for Asia (including ASEAN) ranks in 2nd

position, behind Africa. In fact, according to the results of our another

investigation, over period 1960-2007, a big part of worldwide ODA

concentrates in Africa (34,3 %) and in Asia (33,5 %). Latin America is the

third beneficiary (10,1 %). In that time, in developing countries, the two thirds

of the complete fluxes of FDI go in Asia and in America while Africa is

classified in third position. For instance over period 2000-2007, Asia occupies

62,02 % of inflows of FDI towards developing countries, America accepts



13/45

13

28,81 % of the total and, this figure of Africa is 8,96 %. So, Asia records the

only case where the inflows of FDI are so much important as ODA. This

source of outside allocation contributes irrefutably to the improvement of social

and economic infrastructures of the ASEAN countries. So, graphically, the

figure 3 and figure 4 allow us to assume the existence of a positive interaction

between ODA and FDI in ASEAN member countries.



14/45

14

Table 1: FDI of ASEAN countries in 2010, ranking by sectors

(rank, percentage of total FDI of the country)

Sector Cambodia Indonesia Lao Malaysia Myanmar Philippines Thailand Vietnam

Agriculture 1 (33.84%) 2 (15.34%) 1(28.87%) 5(10.63%) 1(36.36%) 3(12.31%) 3(12.42%) 1(20.58%)

Manufacturing 2 (14.91%) 1 (24.82%) 3(9.45%) 1(26.11%) 3(19.52%) 1(21.44%) 1(35.63%) 2(19.68%)

Trade 3 (13.84%) 3(13.72%) 2(20.17%) 2(14.50%) 2(19.84%) 2(17.37%) 2(13.10%) 3(14.59%)

Mining 8 (0.62%) 4(11.15%) 4(7.42%) 4(12.55%) 8(0.91%) 9(1.43%) 7(3.42%) 4(10.86%)

Public administration 6 (1.76%) 8(5.51%) 7(4.49%) 6(764%) 6(2.14%) 7(4.14%) 6(4.36%) 5(7.55%)Construction 5 (5.48%) 5(10.29%) 5(5.07%) 8(3.23%) 5(4.54%) 6(6.12%) 9(2.67%) 6(7.03%)

Transport and communications 4 (7.47%) 7(6.50%) 6(5.00%) 7(6.63%) 4(13.77%) 5(6.51%) 4(6.81%) 7(4.31%)

Electricity, gas, and water 9 (0.55%) 9 (0.78%) 8(3.68%) 9(2.51%) 7(1.03%) 8(3.57%) 8(2.94%) 8(3.53%)

Finance 7 (1.46%) 6(7.21%) 9(3.56%) 3(13.15%) 9(0.08%) 4(6.91%) 5(6.32%) 9(1.89%)

Total value (Millions USD) 76.221731 1.375368 33.65298 910296.6 7779574 129.7999 973.3324 819.0824Source: http://stats.unctad.org

Table 2: ODA of ASEAN countries in 2010, ranking by sectors

(rank, percentage of total FDI of the country)

Sector Cambodia Indonesia Lao Malaysia Myanmar Philippines Thailand Vietnam

Infrastructure 1(50.36%) 1(62.24%) 1(41.66%) 1(54.87%) 1(86.26%) 1(71.79%) 2(23.01%) 1(45.02%)

Transport 2(33.44%) 4(3.52%) 3(21.91%) 7(0.72%) 3(1.29%) 3(7.34%) 1(57.69%) 2(25.77%)

Energy 5(1.82%) 2(20.89%) 2(22.50%) 2(24.15%) 6(0.08%) 4(3.97%) 5(2.93%) 3(12.59%)

Production 3(8.67%) 3(12.26%) 4(9.69%) 3(16.18%) 2(11.32%) 2(13.19%) 4(6.32%) 4(11.85%)Banking 4(4.90%) 5(0.48%) 6(1.37%) 4(1.75%) 5(0.42%) 6(1.14%) 7(0.21%) 5(2.84%)

Business 6(0.44%) 6(0.35%) 5(2.34%) 5(1.31%) 7(0.02%) 5(2.06%) 3(8.70%) 6(1.49%)

Communications 7(0.37%) 7(0.26%) 7(0.53%) 6(1.02%) 4(0.60%) 7(0.51%) 6(1.13%) 7(0.44%)

Total value (Millions USD) 929.8821 2144.219 449.858 71.18099 246.3293 518.0623 479.7059 3121.3Source: http://stats.oecd.org



15/45

15

4. The empirical model

To test the above theoretical model, we derive the relationship between

aid and fdi as taking the following reduced form:4

fdik = f(fdia, aidk, aida, X). (10)

To explain fdik - the amount of fdi invested in physical capital in a

country, we expect the impacts of aidk - the amount of aid invested in physical

capital, aida – aount of aid invested to complementary factors, fdia - the

amount of fdi invested in complementary factors ; and other determinants of fdi

all reflected in vector X, like the gross domestic savings; the current account

balance; the inflation index…In a panel setting, the econometric interpretation

of this aid-fdi relationship is :

fdik it = β0+ β1fdiait+ β2aidait+ β3aidk it+ β4gdpit+ β5 bopit+ β6sit+

β7hdiit+ β8 popit+ β9inflait+ uit (11)

where fdik it is the net flow of FDI invested in physical capital to country i at

period t; fdiait is FDI invested in complementary factors; aida it denotes AID

invested in complementary factors; aidk it is AID invested in physical capital;

gdpit is gross domestic product; bop is current account balance, s is grossdomestic savings, hdi is human development index value, pop is population

and infla is inflation index. Noting that AID is divided into aida and aidk to

model the impacts of aid on fdik by the direct channel (aidk) as well as by the

indirect channel (aida).

We expect β1 to be positive in this specification since higher

investments in complementary factors raise the demand for investments in

physical capital. The net effect of aida on fdik is theoretically ambiguous,

hence will be shown only empirically. From the theoretical analysis, aidk is

expected to crowd out foreign investments, therefore the sign of β3 is expected

to be negative. We expect β4 to be positive since a host country’s size, typically

measured by GDP, is important in determining a country’s FDI inflows (Tsai,

4 Aid (also known as international aid, overseas aid, or foreign aid) is a voluntary transfer of resources

from one country to another, given at least partly with the objective of benefiting the recipient country.The most widely measure of aid is Official Development Assistance (ODA). This term is first defined

by the DAC of the OECD in 1961.



16/45

16

1994). It is also widely argued that FDI and openness of the economy will be

positively related (Caves, 1996; Singh and Jun, 1995). Recall that the degree of

openness of the country can be measured by its export and import both divided

by GDP. So, we include the BOP in the regression and expect that its

coefficient β5 is positive. β6 should be negative since a high level of domestic

savings lowers the need for foreign capital. We also expect β7 to be positive

since a higher quality of labors can encourage more investors in FDI. Finally,

β8 is expected to be positive since a big size of the local market, represented by

the population, allows for an increase in FDI.

Besides, we assume that not all countries start out with the same initial

conditions, and therefore we include time and country effects in the model. To

avoid missing relevant variables, we extend the list of regressors to include a

lagged dependent variable, which basically reflects the persistent nature of fdik.

We believe that the omitted variables bias is substantially reduced by including

a full set of time dummies, individual country effects, the initial level of GDP,

and the lagged level of the dependent variable.

Our final empirical model is specified as follows :

fdik it = β0+ β1fdiait+ β2aidait+ β3aidk it+ β4gdpit+ β5 bopit+ β6sit+ β7hdiit+

β8 popit+ β9inflait+ β10 country+ β11 year+ β12lagfdik+ uit (12)

5. Data

The dependent variable in all our regressions, fdik it, is the sum of FDI

invested in 6 sectors: agriculture, mining, manufacturing, construction, trade

and finance which represent physical capital sector. The independent variable,

fdiait, is the sum of FDI flows in complementary inputs (such as electricity,

transport and communications, public administration, taxes minus subsidies on

products, taxes on imports and less imputed bank service charges). Because we

cannot obtain any FDI database classified by sector, we indirectly construct this

data by ourselves. We collect data on FDI as percentage of GDP from

UNCTAD website, and collect data on GDP by sector from ADB database. We

then derive FDI by sector data basing on the following equation:



17/45

17

(13)

where FDIit is FDI in sector i during period t; GDPit is GDP in sector i during

period t; % (FDI/GDP) is percentage of FDI in total GDP of country j at time

t ; r is exchange rate between local currency and USD at time t.

The aid variables are total net flows of official aid commitments

reported in the OECD aid statistic database. Aid is also decomposed into two

broad categories, relying on the sectoral disaggregation from OECD’s Aid

Activity database:

Aid invested in complementary inputs, aida, is all aid used in social

infrastructure (such as communications, energy, social infrastructure and

transport).

Aid invested in physical capital, aidk, is all aid contributing directly to

production sectors (such as banking, business and production).

These two categories capture the main characteristics of aida and aidk:

aid invested in complementary factors intended to generate positive spillover

effects (public goods, inputs complementary to physical capital) whereas aid

invested in physical capital has a more narrow purpose and could more easily

have been undertaken by private investors.



18/45

18

Figure 4. Compositions of ODA of ASEAN countries

Cambodia Indonesia

Laos Malaysia



19/45

19

Myanmar Philippines

Thailand Vietnam



20/45

20

Figure 5. Compositions of FDI of ASEAN countries

Cambodia Indonesia

Laos Malaysia



21/45


22/45

22

With this classification in hand, composition of ODA to each ASEAN

country is depicted in figure 4. We see that in all countries, a large propostion

of aid is for the complementary sector.

Similar rules applied to divide FDI into FDIk and FDIa. We see in

contrary, in all countries, a large proposition of FDI is for the physical capital

(see figure 5).

The main control variables are the gross domestic savings that are taken

from ADB database ; the human development index value taken from UNDP,

the gross domestic product, balance of payment, inflation and population all

taken from World Economic Outlook Database of IMF.

All the data was converted from current price to constant price 2000 by

applying the following equations:

(14)

where is constant price at time t; def base is the implicit GDP deflator at

2000; def non-base is the implicit GDP deflator at other year; is current price at

time t. The implicit GDP deflator is taken from ADB database.

We summarize the expected sign of each variable and their sources in

table 3 as follows :

Table 3: Variable names, definitions and data sources

Definition Unit Source Sign

Fdik

FDI invested in physical capital(agriculture+ mining+ manufacturing+construction+ trade+ finance)

billions USD,constant price2000

ADB,UNCTAD +

Fdia

FDI invested in complementary factors(electricity, gas & water+ transport&

communications+ public administration+others+ taxes less subsidies on products+ taxeson imports- imputed bank service charges)

millions USD,constant price2000

ADB,UNCTAD +

Aida

AID invested in complementary factors(communications+ energy+ socialinfrastructure+ transport)

millions USD,constant price2000

UNCTAD

+ or-

AidkAID invested in physical capital(banking+ business+ production)

billions USD,constant price2000

UNCTAD -

GDP Gross domestic product

billions USD,constant price2000 IMF +

Bop Current account balance billions USD,constant price IMF +



23/45

23

2000

Infla Inflation

averageconsumer pricesIndex IMF -

S Gross domestic savings

billions USD,

constant price2000 ADB -

Hdi Human development index

HDR/U NDP +

Pop Population millions persons IMF+ or-

Notes :

-

ADB : Asian Development Bank (ADB)/ Key Indicators for Asia and

the Pacific 2011/ www.adb.org/statistics

-

UNCTAD : UNCTADstat/ data extracted on 15 Jul 2012 19:27 UTC

(GMT) from OECD.Stat

- IMF : World Economic Outlook Database

- HDR/UNDP: HDRO calculations based on data from UNDESA (2011),

Barro and Lee (2010), UNESCO Institute for Statistics (2011), World

Bank (2011a) and IMF (2011)./ Accessed: 10/31/2011,11:07 AM from:

http://hdr.undp.org

6. Empirical results

i) OLS

We start by estimating the impact of independent variables on fdik using

OLS to have a benchmark.

Table 4: Estimation Results of Panel Method of FDI Inflows for ASEAN

Countries Using OLS

Dependent Variable= fdik

VariablesCoefficients (OLS)

Coefficient P-value

Intercept -880.8093 0.250

Fdia 5.29712 0.000*

Aida .0578955 0.283

Aidk -.3366853 0.013*

Gdp -.0591393 0.943

Bop -.5645343 0.859

S -.6836061 0.557


http://www.adb.org/statisticshttp://hdr.undp.org/http://hdr.undp.org/http://www.adb.org/statistics


24/45

24

Hdi 817.8138 0.132

Pop 2.444029 0.387

Infla .7025877 0.375

Lagfdik -.131042 0.007*

R 2 0.9966

Obser. 127

Note:** *, ** and * indicate 10%, 5% and 1% level of significant respectively

So, by including a set of dummies for year and country, we are also

controlling the impact of time dimension and country dimension. Except fdia

estimator, lagfdik, aida and aidk estimators are not good as our expectation: the

coefficient of aida is positive but not significative while the coefficient of aidkand lagfdik are all negative.

ii) FE/RE models

Because our dataset is panel data (cross-sectional time-series data), we

have to decide whether fixed effect or random effect model fits our data. For

that purpose, we run a Hausman test where the null hypothesis is random

effects (the preferred model) versus the alternative is the fixed effects.

Table 5: Estimation Results of Panel Method of FDI Inflows for ASEANCountries Using FE Model


VariablesCoefficients (FE model)

Coefficient P-value

Intercept -293.4201 0.376

fdia 5.265827 0.000*

aida .029585 0.692

aidk -.1572828 0.477gdp -.1953439 0.789

bop -4.840637 0.144

s -.4659532 0.280

hdi 448.464 0.254

pop -.6414817 0.870

infla .3537186 0.245

lagfdik -.1081347 0.000*

R2 0.9918

Obser. 127 Note: ** *, ** and * indicate 10%, 5% and 1% level of significant respectively



25/45

25

The result of Hausman test recommend the fixed effect model.

However, two important independent variables aida and aidk are both not

significative at 1%, even at 5% or 10%.

iii) 2SLS

To get a better estimate of the effect of aid on fdi, we try to use aida, hdi

and pop as instruments for aidk, we have the result as follows:


Countries Using 2SLS model


VariablesCoefficients (2SLS)

Coefficient P-value

Intercept 56.66896 0.644

Fdia 5.275206 0.000*

Aidk .365944 0.290

Gdp -.03536 0.960

Bop -3.883426 0.247

S -.5759026 0.575

Infla .5449874 0.444

Lagfdik -.1214462 0.004*

R 2 0.9962

Obser. 127

Note: ** *, ** and * indicate 10%, 5% and 1% level of significant respectively

Instrumenting for aidk here has led to a coefficient of 0.365 but this is

insignificantly different from zero. Now, we run the 2SLS regression to get the

following result :

Table 7: Estimation Results of Panel Method of FDI Inflows for ASEANCountries Using 2SLS model (1

st stage)

Dependent Variable= aidk


Coefficient P-value

Intercept

fdia .0000132 0.246

aida .107984 0.004*

gdp 0002371 0.595

bop .00199 0.333



26/45

26

s -.0001081 0.199

hdi .197392 0.435

pop -.0020961 0.251

infla .0001611 0.059**

lagfdik -8.94e-06 0.087***

R 2 0.6958

Obser. 127

Note: ** *, ** and * indicate 10%, 5% and 1% level of significant respectively

Looking at the first stage we see that few of the instruments have

coefficients significantly different from zero- we might be concerned that the

instrument is weak .

iv)

Arellano- Bond GMM estimator

Several econometric problems may arise from estimating equation (10):

(1)

The capital flows variables in fdik it are assumed to be endogenous.

Because causality may run in both directions- from capital inflows to

investment to investment and vice versa-these regressors may be

correlated with the error term.

(2)

Time-invariant country characteristics (fixed effects), such as geography

and demographics, may be correlated with the explanatory variables.

The fixed effects are contained in the error term in equation (1), which

consists of the unobserved country-specific effects and the observation-

specific errors.

(3) The presence of the lagged dependent variable lagfdik it gives rise to

autocorrelation.

(4)

Heteroskedasticity and autocorrelation within individuals, but not acrossthem.5

(5) Independent variables that are not strictly exogenous, meaning

correlated with past and current realizations of the error. 6

To solve problem (1) and (2), one would usually use fixed-effects

instrumental variables estimation (2SLS) which is what we tried first.

5 See Heteroskedasticity test in Appendix 1

6 See Cross-sectional dependence test in Appendix 2



27/45

27

However, the first-stage statistics of the 2SLS regressions showed that our

instruments were weak. With weak instruments the fixed-effects IV estimators

are likely to be biased in the way of the OLS estimators. Therefore, we decide

to use the Arellano-Bond (1991) difference GMM estimator first proposed by

Holtz-Eakin, Newey and Rosen (1988). The first-differenced lagged dependent

variable (problem 3) is also instrumented with its past levels. Finally, the

problem of heteroskedasticity (4) and cross-sectional dependence (5) will also

be removed.

Before using Arellano-Bond GMM, we need to test for the stationarity

of the model by unit root tests. As Maddala and Wu (1999) describe the IPS

test “…the IPS test is a way of combining the evidence on the unit-root

hypothesis from the N unit-root tests performed on the N cross-section units”.

Fisher-type panel unit-root tests make this approach explicit.7 That is the

reason why we use the Fisher-type tests in our model.

We consequently test for a unit root in fdik, fdia, aida, aidk, bop, gdp, s,

hdi, bop and infla. We will use the ADF test. Because the number of panels is

finite, the inverse χ 2 test is applied.8

As before, we apply the panel unit root tests including constant and

trend in level. The results are shown in table 8 as follows:

Table 8: Panel unit root test results in level

Variable

ADF-Fisher chi square

(constant)

ADF-Fisher chi square

(trend)

fdik 27.2473** (0.0388) 18.1028 (0.3179)

fdia 95.6230* (0.0000) 88.0186* (0.0000)

aida 50.0414* (0.0000) 51.6718* (0.0000)aidk 62.4451* (0.0000) 51.4102 (0.0000)

7 xtunitroot fisher combine the p-values from the panel-specific unit-root tests using the four methods

proposed by Choi (2001). Three of the methods differ in wheher they use the inverse χ 2, inversenormal, or inverse logit transformation of p-values, and the fourth is a modification of the inverse χ 2transformation that is suitable for when N tends to infinity. The inverse normal and inverse logittransformation can be used whether N is finite or infinite.The null hypothesis being tested by xtunitroot fisher is that all panels contain a unit root. For a finitenumber of panels, the alternative is that at least one panel is stationary. As N tends to infinity, thenumber of panels that do not have a unit root should grow at the same rate as N under the alternativehypothesis.8

this statistic has a χ 2 distribution with N degrees of freedom, and large values are cause to reject thenull hypothesis. Under the null hypothesis, as T ∞ followed by N ∞, P tends to infinity so that Phas a degenerate limiting distribution.



28/45


29/45

29

Dhdiit denotes the human development index variable in first difference,

Dpopit denotes the population variable in fisrt difference,

Dinflait denotes the inflation index variable in fisrt difference,

Therefore, we have estimation result as follows:


Countries Using GMM model with year and country effects



Coefficient P-value

Intercept -37.18714 0.573

Lagfdik -.0925815 0.000*

Dgdp -.484664 0.438Bop 1.757468 0.534

Dpop -96.69941 0.031**

Dhdi -704.4586 0.224

Ds -.8772231 0.000*

Fdia 4.946434 0.000*

Dinfla 5.173003 0.000*

Aida .0915501 0.074**

Aidk -.2513535 0.103**

Number of Obs. 112 Number of instruments 113

Wald chi2(24) 42443.86 (0.0000)


We see that all important instruments have coefficients significantly

different from zero: the coefficient on lagfdik is -0.092 at 1% significance, the

coefficient on fdia is 4.95 at 1% significance, the coefficient on aida is 0.915 at

10% significance and the coefficient on aidk is -0.251 at 10% significance.

GMM estimator without controlling time and year effects

From equation (10) we get:

fdik it = α1fdiait+ α2aidait+ α3aidk it+ α4Dgdpit+ α5 bopit+ α6Dsit+ α7Dhdiit+

α8Dpopit+ α9Dinflait+α10lagfdik it+ vit (16)

We will use the one-step Arellano-Bond estimator and request their

robust VCE.



30/45

30


Countries Using GMM model without year and country effects



Coefficient P-value

Intercept

Lagfdik -.0978983 0.000*

Dgdp .1905667 0.711

Bop -.9810358 0.384

Dpop -123.2232 0.211

Dhdi 432.4764 0.396

Ds -1.174967 0.000*

Fdia 4.946891 0.000*

Dinfla 4.715908 0.000*Aida .0695371 0.087***

Aidk -.2305563 0.025**

R 2

Number of Obs. 112

Number of instruments 95

Wald chi2(7) 1701.45 (0.0000)


We can consider that the P-value of aida and aidk in this case is better

than theirs in previous regression except the case of Dpop variable. The results

confirm one dollar of fdi invested on physical capital attracts on average 4.95

dollars of additional fdik. On the other hand, the result shows that one dollar of

aid invested on physical capital crowds out on average 0.23 dollars of fdik. The

table also shows that one aid dollar invested in complementary factors attracts

on average 0.70 dollars of additional fdik. The effect of other controls is eitherinsignificant or goes according to the theoretical predictions: population enters

insignificantly, domestic savings negatively (1 additional dollar of domestic

savings is associated with 1,17 dollars less of fdik, and initial GDP enters

positively (1 additional dollar of GDP at the beginning of each period tends to

attract 0.19 dollar of fdik on average.

Postestimation



31/45

31

Now, we will use Sargan/Hansen test for joint validity of the

instruments after GMM estimation. Theses tests allow testing for

autocorrelation in the first-differenced residuals and testing the validity of the

overidentifying restrictions.

For the GMM model with country and year effects, we can’t perform the

post-estimation tests because we have more instruments (113) than

observations (112) meanwhile for the model without these effects, all the tests

of autocorrelation and overidentifying restrictions are significant- that means

the instruments of the later model are valid.9

By conclusion, our target when including the country and year effects in

the model is to increase the number of observations. However, once this target

can’t be attained; we can focus on the model without these effects. They also

show more statistical significance.

7. Concluding remarks

The analysis shows strict relations between FDI and ODA into a country.

We find robust evidence for ASEAN countries that FDI and ODA invested in

physical capital are substitutes as expected, even though they are not perfectly

substitutes. One possible reason is that the ODA we use in the empirical

analysis is the commitment ODA instead of the disbursement ODA due to the

availability of data. It is also shown that both FDI and AID in complementary

factors are complementary to FDI in physical captital. Results can provide

inputs for decision making of foreign investors and donors in general and in

South East region in particular.

9 View postestimation tests in Appendix 6 & 7



32/45

32

REFERENCES

Assidon E. (2000), « Les théories économiques du développement », La

Découverte, 3ème édition, Paris.

Assidon E. (2000), « Les théories économiques du développement », La

Découverte.

Ben Abdallah M., Drine I., Meddeb R. (2001), “Interaction entre IDE, regime

de change, capital humain et croissance dans les pays émergents”,

document de travail, Ouverture économique et développement,

GDR, Economica.

Blaise, S. (2005). On the link between Japanese ODA and FDI in

China: A microeconomic evaluation using Conditional Logit

analysis. Applied Economics, 37, 51-55.

Chaisrisawatsuk, S. and Chaisrisawatsuk, W., (2007), « Imports, exports and

foreign direct investment interactions and their effects”, in ESCAP,

Towards coherent policy frameworks: understanding trade and

investment linkages- A study by the Asia-Pacific Research and

Training Network on Trade, United Nations, New York, pp.97-115.

Charnoz O., Severino J.-Michel (2007), « L’aide publique au développement »,

La Découverte.

Chauvet L.(2005), “Aide au développement”, http://www.universalis.fr.

Christiansen H., Bertrand A. (2003), « Tendances et évolution récente de

l’investissement direct étranger », OCDE.

Cogneau D., Lambert S. (2006), « L’aide au développement et les autres flux

nord-sud : complémentarité ou substitution ? », document de travail

no. 251, Centre de développement de l’OCDE.

Dormont B. (2007), « Introduction à l’économétrie », Montchrestien.

Doucouliagos H. and Paldam M. (2005) “Aid effectiveness on accumulation. A

meta study”, working paper 2005-12, University of Aarhus,

Denmark.



33/45

33

Dunning, J.H. (1993). Multinational enterprises and the global economy.

Wokingham, England: Addison-Wesley.

Duval, Y. (2008), “Trade and Investment Linkages and Policy Coordination:

Lessons from Case Studies in Asian Developing Countries”, working

paper series, No 55, Asia-Pacific Research and Training Network on

Trade (ARTNET).

Gannagé E. (1984), « Théories de l’Investissement direct étranger »,

Economica.

General statistics office of Vietnam (2003, 2004, 2005, 2006, 2007),

“Statistical yearbook”, Statistical publishing house.

Genral statistics office of Vietnam (2008), “Foreign direct investment in

Vietnam 7 years at the beginning of century 21st”, Statistical

publishing house.

Giuseppe N., Golub Stephen S., Hajkova D., Mirza D., Yoo K. (2003),

“L’influence des politiques sur les échanges et l’investissement

direct étranger”, article journal, revue économique de l’OCDE no.

36.

Harms Philipp. and Lutz Matthias (2006), “Aid, governance, and private

foreign investment” Economic Journal 116(513): 773-790.

Huart J.-Marc (2003), « Croissance et développement », Bréal.

Hurlin Chr., « L’économétrie des données de panel- Modèle linéaires simples

», séminaire méthodologique, Ecole doctorale Edocif.

Karakaplan Ugur M., Neyapti, Bilin, Sayek and Selin (2005) “Aid and foreign

direct investment: International evidence.” Bilkent University

Discussion Papers 05-05, 11, Ankara-Turkey: Bilkent University.

Kimura, H. and Todo, Y. (2007) “Is foreign aid a vanguard of FDI? A gravity

equation approach” Research paper, Research Institute of Economy,

Trade and Industry (RIETI), Tokyo, Japan.

Lall, S.(1997). Attracting foreign investment: new trends, sources and policies.

Economic Paper No. 31 (Commonwealth Secretariat).



34/45

34

LEE, H-L. & HOUDE, M-F. (2000).Recent trends and main characteristics of

foreign direct investment in China. Financial market trends, No. 77,

October, OECD.Hansen and Tarp (2000).

Magalhães, M. and Africano, A., (2007), “A panel analysis of the FDI impact

on Internatinal trade”, FEP (Faculdade de economia universidade do

Porto) working papers.

Mallampally P., Sauvant Karl P. (1999), “L’investissement direct étranger dans

les pays en développement”, Finances & Développement, volume

36, numéro 1, pp. 34-37.

Nurske R. (1953), « Problems of capital formation in underdeveloped countries

», Basil Blackwell, Oxford, first edition, pp.4- 11.

OECD (2002), « Tendances et évolution récente de l’investissement direct

étranger », Perspectives de l’investissement international, Edition

2002, OCDE.

OECD (2003a), « Liste de critères pour apprécier les stratégies d’incitations à

l’investissement direct étranger », Perspectives de l’investissement

international, ISBN 92-64-10361-9, pp.112-152.

OECD (2003b), « Tendances de l’investissement direct étranger dans les pays

de l’OCDE », Perspectives économiques de l’OCDE, pp.193-201.

OECD (2003c), « L’entreprenariat et le développement économique local :

quels programmes et quelles politiques?», les éditions de l’OCDE.

OECD (2003d), « Harmoniser l’aide pour renforcer son efficacité », les

éditions de l’OCDE.

OECD (2006), « Promouvoir l’investissement privé au service du

développement- le rôle de l’APD », un document de référence du

CAD.

Olivier G. (2004), « L’aide publique au développement », Charles Léopold

Mayer.

PHAM Thu Hien (2008), “Effect of ODA in infrastructure in attracting FDI

inflows in Vietnam”, work papers, Ministry of Planning and

investment of Vietnam.



35/45

35

Root, F. R. & A. A. Ahmed. (1979). Empirical determinants of manufacturing

direct investment in developing countries. Economic Development

and Cultural Change, 27 (4): 751-767.

Selaya, P. and Sunesen, E. R., (2012), “Does Foreign Aid increase Foreign

Direct Investment?”, working paper.

UNCTAD (2008a), “Investment policy review”, ISBN 978-92-1-112744-7.

UNCTAD (2008b), « Foreign direct investment and financing for development:

trends and selected issues », issues note by the UNCTAD secretariat,

TD/B/COM.2/80.

UNCTAD (2008c), « World investment report 2008: transnational corporations

and the infrastructure challenge », United Nations publication, New

York and Geneva.

UNCTAD (2008d), «Review of the technical cooperation activities of

UNCTAD», report by the Secretary- General of UNCTAD,

TD/B/WP/202.

UNCTAD (2011), “World investment report 2011: Non- equity modes of

international production and development”, United Nations

publication, New York and Geneva.

UNCTAD. (1999). World Investment Report: foreign direct investment and the

challenge of development. New York and Geneva, United Nations.



36/45

36

APPENDIX

1. Heteroskedasticity test

. xtr eg f di k f di a ai da ai dk gdp bop s hdi pop i nf l a l agf di k, f e

Fi xed- ef f ect s ( wi t hi n) r egr essi on Number of obs = 127Gr oup vari abl e: countr y Number of groups = 8

R- sq: wi t hi n = 0. 9941 Obs per gr oup: mi n = 15bet ween = 0. 9908 avg = 15. 9overal l = 0. 9918 max = 16

F( 10, 109) = 1832. 40cor r ( u_i , Xb) = 0. 3384 Pr ob > F = 0. 0000

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -f di k | Coef . St d. Err . t P>| t| [ 95% Conf . I nterval ]

- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -f di a | 5. 265827 . 0748263 70. 37 0. 000 5. 117524 5. 41413ai da | . 029585 . 0745584 0. 40 0. 692 - . 1181873 . 1773572ai dk | - . 1572828 . 2204934 - 0. 71 0. 477 - . 5942935 . 2797279gdp | - . 1953439 . 7275726 - 0. 27 0. 789 - 1. 637369 1. 246681bop | - 4. 840637 3. 285552 - 1. 47 0. 144 - 11. 35249 1. 67122

s | - . 4659532 . 4294811 - 1. 08 0. 280 - 1. 317171 . 3852645hdi | 448. 464 390. 9469 1. 15 0. 254 - 326. 3801 1223. 308pop | - . 6414817 3. 898563 - 0. 16 0. 870 - 8. 368307 7. 085343

i nf l a | . 3537186 . 3023647 1. 17 0. 245 - . 2455585 . 9529956l agf di k | - . 1081347 . 0205067 - 5. 27 0. 000 - . 1487783 - . 0674911

_cons | - 293. 4201 329. 8004 - 0. 89 0. 376 - 947. 0737 360. 2335- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

si gma_u | 199. 56759si gma_e | 180. 59799

r ho | . 54977435 ( f r act i on of var i ance due t o u_i )- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -F t est t hat al l u_i =0: F(7, 109) = 5. 69 Pr ob > F = 0. 0000

. xt test3

Modi f i ed Wal d test f or groupwi se het eroskedasti ci t yi n f i xed ef f ect r egr essi on model

H0: si gma(i ) 2̂ = si gma 2̂ f or al l i

chi 2 ( 8) = 8857. 45Prob>chi 2 = 0. 0000

The null is homoscedasticity (or constant variance). Above we reject the null and

conclude heteroskedasticity.

2. Cross- sectional dependence test (CSD)

We assume that uit is formed by a combination of a fixed component specific to the

state and a random component that captures pure noise. Therefore, below we have the

results of the model using the FE estimator:



R- sq: wi t hi n = 0. 9941 Obs per gr oup: mi n = 15bet ween = 0. 9908 avg = 15. 9

overal l = 0. 9918 max = 16

F( 10, 109) = 1832. 40



37/45

37

cor r ( u_i , Xb) = 0. 3384 Pr ob > F = 0. 0000


- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -f di a | 5. 265827 . 0748263 70. 37 0. 000 5. 117524 5. 41413ai da | . 029585 . 0745584 0. 40 0. 692 - . 1181873 . 1773572ai dk | - . 1572828 . 2204934 - 0. 71 0. 477 - . 5942935 . 2797279

gdp | - . 1953439 . 7275726 - 0. 27 0. 789 - 1. 637369 1. 246681bop | - 4. 840637 3. 285552 - 1. 47 0. 144 - 11. 35249 1. 67122

s | - . 4659532 . 4294811 - 1. 08 0. 280 - 1. 317171 . 3852645hdi | 448. 464 390. 9469 1. 15 0. 254 - 326. 3801 1223. 308pop | - . 6414817 3. 898563 - 0. 16 0. 870 - 8. 368307 7. 085343

i nf l a | . 3537186 . 3023647 1. 17 0. 245 - . 2455585 . 9529956l agf di k | - . 1081347 . 0205067 - 5. 27 0. 000 - . 1487783 - . 0674911

_cons | - 293. 4201 329. 8004 - 0. 89 0. 376 - 947. 0737 360. 2335- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

si gma_u | 199. 56759si gma_e | 180. 59799

r ho | . 54977435 ( f r act i on of var i ance due t o u_i )- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -F t est t hat al l u_i =0: F(7, 109) = 5. 69 Pr ob > F = 0. 0000

According to the results, one we account for state FE, aida and aidk has no effect

upon fdik. An assumption implicit in estimating (1) is that the cross-sectional units are

independent. The xtcsd command allows us to test the following hypothesis:

H0: cross-sectional independence

To test this hypothesis, we use the xtcsd command after fitting the above panel-data

model. We initially use Pesaran’s (2004) CD test:

. xtcsd, pesar an abs

Pesaran' s t est of cross sect i onal i ndependence = 3. 136, Pr = 0. 0017

Aver age absol ute val ue of t he of f - di agonal el ement s = 0. 306

As we can see, the CD test strongly rejects the null hypothesis of no cross-sectional

dependence. Although it is not the case here, a possible drawback of the CD test is

that adding up positive and negative correlation may result in failing to reject the null

hypothesis even if there is plenty of cross-sectional dependence in the errors.

Including the abs option in the xtcsd command, we can get the average absolute

correlation of the residuals. Here the average absolute correlation is 0.306, which is avery high value. Hence, there is enough evidence suggesting the presence of cross-

sectional dependence in (1) under an FE specification.

Next we corroborate these results by using the remaining two tests Frees (1995) and

Friedman (1937):

. xtcsd, f reesFrees' t est of cr oss sect i onal i ndependence = 0. 398

| - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - |Cr i t i cal val ues f romFrees' Q di str i but i on

al pha = 0. 10 : 0. 1719al pha = 0. 05 : 0. 2262

al pha = 0. 01 : 0. 3351. xtcsd, f r i edmanFri edman' s test of cr oss sect i onal i ndependence = 17. 325, Pr = 0. 0154



38/45

38

As we would have expected from the highly significant results of the CD test, both

Frees’ and Friedman’s tests reject the null of cross-sectional independence. Since T ≤

30, Frees’ test provides the critical values for α= 0.10, α= 0.05 and α= 0.01 from the

Q distribution, Frees’ statistic is larger than the critical value with at least α= 0.01.

Using the RE estimator, we have the results shown below:

. xtr eg f di k f di a ai da ai dk gdp bop s hdi pop i nf l a l agf di k, r e

Random- ef f ect s GLS r egr essi on Number of obs = 127Gr oup vari abl e: countr y Number of groups = 8


Randomef f ects u_i ~ Gaussi an Wal d chi 2( 10) = 21572. 91cor r ( u_i , X) = 0 ( assumed) Pr ob > chi 2 = 0. 0000

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -f di k | Coef . St d. Err . z P>| z| [ 95% Conf . I nterval ]

- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -f di a | 5. 237567 . 0837154 62. 56 0. 000 5. 073488 5. 401646ai da | - . 0756591 . 0616294 - 1. 23 0. 220 - . 1964505 . 0451323ai dk | - . 43269 . 2333102 - 1. 85 0. 064 - . 8899697 . 0245896gdp | - 1. 713865 . 4977064 - 3. 44 0. 001 - 2. 689352 - . 7383787bop | - . 0229533 3. 244583 - 0. 01 0. 994 - 6. 38222 6. 336313

s | - . 4778182 . 4431746 - 1. 08 0. 281 - 1. 346425 . 390788hdi | 121. 3234 239. 2368 0. 51 0. 612 - 347. 5721 590. 219pop | 1. 634705 . 4466665 3. 66 0. 000 . 7592549 2. 510155

i nf l a | . 6893927 . 2382594 2. 89 0. 004 . 222413 1. 156372l agf di k | - . 1208659 . 019442 - 6. 22 0. 000 - . 1589715 - . 0827604

_cons | - 97. 96013 149. 77 - 0. 65 0. 513 - 391. 5039 195. 5836- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

si gma_u | 0si gma_e | 180. 59799

r ho | 0 ( f r act i on of var i ance due t o u_i )- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

The results of this second model are in line with those of the previous one, with aida

and aidk having no significant effects upon fdik (at α= 0.05). We now test for cross-

sectional independence by using the new RE specification:

. xtcsd, pesar an absPesaran' s t est of cross sect i onal i ndependence = 3. 584, Pr = 0. 0003

Aver age absol ute val ue of t he of f - di agonal el ement s = 0. 269

. xtcsd, f reesFrees' t est of cr oss sect i onal i ndependence = 0. 295

| - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - |Cr i t i cal val ues f romFrees' Q di str i but i on

al pha = 0. 10 : 0. 1719al pha = 0. 05 : 0. 2262al pha = 0. 01 : 0. 3351

. xtcsd, f r i edmanFri edman' s test of cr oss sect i onal i ndependence = 17. 800, Pr = 0. 0129

The conclusion with respect to the existence or not of cross-sectional dependence in

the errors is not altered. The results show that there is enough evidence to reject the

null hypothesis of cross-sectional independence.



39/45

39

3. OLS

. xi : r egr ess f di k f di a ai da ai dk gdp bop s hdi pop i nf l a l agf di k i . count r yi . year, vce ( r obust)i . count r y _I count r y_1- 8 ( nat ur al l y coded; _I count r y_1 omi t t ed)i . year _I year_1995- 2010 ( natural l y coded; _I year_1995 omi t t ed)

Li near r egr essi on Number of obs = 127F( 32, 94) = 205. 49Prob > F = 0. 0000R- squared = 0. 9966Root MSE = 181. 23

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -| Robust

f di k | Coef . St d. Err . t P>| t| [ 95% Conf . I nterval ]- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

f di a | 5. 29712 . 1582804 33. 47 0. 000 4. 98285 5. 611389ai da | . 0578955 . 0536636 1. 08 0. 283 - . 0486548 . 1644457ai dk | - . 3366853 . 1323415 - 2. 54 0. 013 - . 5994524 - . 0739183gdp | - . 0591393 . 8207194 - 0. 07 0. 943 - 1. 688697 1. 570418bop | - . 5645343 3. 1581 - 0. 18 0. 859 - 6. 835016 5. 705947

s | - . 6836061 1. 158699 - 0. 59 0. 557 - 2. 98423 1. 617018hdi | 817. 8138 538. 4393 1. 52 0. 132 - 251. 2701 1886. 898

pop | 2. 444029 2. 813773 0. 87 0. 387 - 3. 142782 8. 030841i nf l a | . 7025877 . 7881125 0. 89 0. 375 - . 862228 2. 267403

l agf di k | - . 131042 . 0473124 - 2. 77 0. 007 - . 2249819 - . 0371022 _I countr y_2 | 532. 8991 558. 1757 0. 95 0. 342 - 575. 3717 1641. 17 _I countr y_3 | 559. 2849 583. 0009 0. 96 0. 340 - 598. 277 1716. 847 _I countr y_4 | 644. 7411 477. 1533 1. 35 0. 180 - 302. 658 1592. 14 _I countr y_5 | 349. 4011 498. 1253 0. 70 0. 485 - 639. 6383 1338. 441 _I countr y_6 | 114. 7235 342. 0132 0. 34 0. 738 - 564. 3519 793. 7988 _I countr y_7 | - 54. 10324 403. 1581 - 0. 13 0. 894 - 854. 583 746. 3765 _I countr y_8 | 155. 9275 349. 8754 0. 45 0. 657 - 538. 7583 850. 6132 _I year_1996 | - 132. 722 81. 09713 - 1. 64 0. 105 - 293. 7423 28. 29824 _I year _1997 | - 156. 2683 84. 12847 - 1. 86 0. 066 - 323. 3073 10. 77079 _I year _1998 | - 209. 5504 88. 12006 - 2. 38 0. 019 - 384. 5149 - 34. 58598 _I year _1999 | - 205. 5174 96. 02574 - 2. 14 0. 035 - 396. 1788 - 14. 85609 _I year _2000 | - 244. 8052 97. 93137 - 2. 50 0. 014 - 439. 2502 - 50. 36013 _I year _2001 | - 98. 76333 132. 3419 - 0. 75 0. 457 - 361. 5313 164. 0046 _I year _2002 | - 147. 1242 118. 1699 - 1. 25 0. 216 - 381. 7534 87. 50497 _I year _2003 | - 168. 9311 103. 9054 - 1. 63 0. 107 - 375. 2376 37. 37551 _I year_2004 | - 195. 7095 109. 0201 - 1. 80 0. 076 - 412. 1714 20. 75241 _I year_2005 | - 268. 0949 117. 4786 - 2. 28 0. 025 - 501. 3514 - 34. 83836 _I year_2006 | - 367. 2309 95. 27106 - 3. 85 0. 000 - 556. 3938 - 178. 0679 _I year_2007 | - 235. 5673 174. 6601 - 1. 35 0. 181 - 582. 359 111. 2244 _I year_2008 | - 199. 7984 116. 3353 - 1. 72 0. 089 - 430. 785 31. 18808 _I year_2009 | - 301. 3504 141. 7787 - 2. 13 0. 036 - 582. 8554 - 19. 84542 _I year_2010 | - 265. 2107 113. 0739 - 2. 35 0. 021 - 489. 7215 - 40. 69985

_cons | - 880. 8093 760. 3566 - 1. 16 0. 250 - 2390. 515 628. 8965- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

4. FE/RE models




F( 10, 109) = 1832. 40cor r ( u_i , Xb) = 0. 3384 Pr ob > F = 0. 0000


- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -f di a | 5. 265827 . 0748263 70. 37 0. 000 5. 117524 5. 41413

ai da | . 029585 . 0745584 0. 40 0. 692 - . 1181873 . 1773572ai dk | - . 1572828 . 2204934 - 0. 71 0. 477 - . 5942935 . 2797279gdp | - . 1953439 . 7275726 - 0. 27 0. 789 - 1. 637369 1. 246681



40/45

40

bop | - 4. 840637 3. 285552 - 1. 47 0. 144 - 11. 35249 1. 67122s | - . 4659532 . 4294811 - 1. 08 0. 280 - 1. 317171 . 3852645

hdi | 448. 464 390. 9469 1. 15 0. 254 - 326. 3801 1223. 308pop | - . 6414817 3. 898563 - 0. 16 0. 870 - 8. 368307 7. 085343

i nf l a | . 3537186 . 3023647 1. 17 0. 245 - . 2455585 . 9529956l agf di k | - . 1081347 . 0205067 - 5. 27 0. 000 - . 1487783 - . 0674911

_cons | - 293. 4201 329. 8004 - 0. 89 0. 376 - 947. 0737 360. 2335- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

si gma_u | 199. 56759si gma_e | 180. 59799

r ho | . 54977435 ( f r act i on of var i ance due t o u_i )- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -F t est t hat al l u_i =0: F(7, 109) = 5. 69 Prob > F = 0. 0000

. esti mates st ore FI XED

. xtr eg f di k f di a ai da ai dk gdp bop s hdi pop i nf l a l agf di k, r e

Random- ef f ect s GLS r egr essi on Number of obs = 127Gr oup vari abl e: countr y Number of groups = 8

R- sq: wi t hi n = 0. 9932 Obs per group: mi n = 15bet ween = 0. 9980 avg = 15. 9overal l = 0. 9947 max = 16

Randomef f ects u_i ~ Gaussi an Wal d chi 2(10) = 21572. 91cor r ( u_i , X) = 0 ( assumed) Pr ob > chi 2 = 0. 0000

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -f di k | Coef . Std. Err . z P>| z| [ 95% Conf . I nterval ]

- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -f di a | 5. 237567 . 0837154 62. 56 0. 000 5. 073488 5. 401646ai da | - . 0756591 . 0616294 - 1. 23 0. 220 - . 1964505 . 0451323ai dk | - . 43269 . 2333102 - 1. 85 0. 064 - . 8899697 . 0245896gdp | - 1. 713865 . 4977064 - 3. 44 0. 001 - 2. 689352 - . 7383787bop | - . 0229533 3. 244583 - 0. 01 0. 994 - 6. 38222 6. 336313

s | - . 4778182 . 4431746 - 1. 08 0. 281 - 1. 346425 . 390788hdi | 121. 3234 239. 2368 0. 51 0. 612 - 347. 5721 590. 219pop | 1. 634705 . 4466665 3. 66 0. 000 . 7592549 2. 510155

i nf l a | . 6893927 . 2382594 2. 89 0. 004 . 222413 1. 156372l agf di k | - . 1208659 . 019442 - 6. 22 0. 000 - . 1589715 - . 0827604

_cons | - 97. 96013 149. 77 - 0. 65 0. 513 - 391. 5039 195. 5836- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

si gma_u | 0si gma_e | 180. 59799

r ho | 0 ( f r acti on of var i ance due t o u_i )- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

. est i mat es st or e RANDOM . hausman FI XED RANDOM Note: t he rank of t he di f f erenced vari ance matr i x ( 9) does not equal t he numberof coef f i ci ent s bei ng t est ed ( 10) ; be sur e

t hi s i s what you expect , or t her e may be probl ems comput i ng t he t est .Exami ne the out put of your est i mators f or

anythi ng unexpected and possi bl y consi der scal i ng your vari abl es so

t hat t he coef f i ci ent s are on a si mi l ar scal e.

- - - - Coef f i ci ent s - - - -| ( b) ( B) ( b- B) sqr t ( di ag( V_b- V_B) )| FI XED RANDOM Di f f erence S. E.

- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -f di a | 5. 265827 5. 237567 . 0282603 .ai da | . 029585 - . 0756591 . 105244 . 0419615ai dk | - . 1572828 - . 43269 . 2754073 .gdp | - . 1953439 - 1. 713865 1. 518521 . 5307072bop | - 4. 840637 - . 0229533 - 4. 817684 . 5172334

s | - . 4659532 - . 4778182 . 011865 .hdi | 448. 464 121. 3234 327. 1405 309. 201pop | - . 6414817 1. 634705 - 2. 276187 3. 872891

i nf l a | . 3537186 . 6893927 - . 3356741 . 1861636l agf di k | - . 1081347 - . 1208659 . 0127313 . 0065218

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -b = consi st ent under Ho and Ha; obt ai ned f r omxt r eg

B = i nconsi st ent under Ha, ef f i ci ent under Ho; obt ai ned f r om xtr eg



41/45

41

Test : Ho: di f f er ence i n coef f i ci ents not sys t emat i c

chi 2( 9) = ( b- B) ' [ ( V_b- V_B) (̂ - 1) ] ( b- B)= 49. 79

Prob>chi 2 = 0. 0000( V_b- V_B i s not posi t i ve def i ni t e)

5. 2SLS

. xi : i vr egr ess 2sl s f di k fdi a gdp bop s i nf l a l agf di k i . count ry i . year ( ai dk=ai da hdi pop) , vce ( r obust )i . count r y _I count r y_1- 8 ( nat ur al l y coded; _I count r y_1 omi t t ed)i . year _I year_1995- 2010 ( natural l y coded; _I year_1995 omi t t ed)

I nstr ument al var i abl es ( 2SLS) r egr essi on Number of obs = 127Wal d chi 2(29) = 7848. 48Prob > chi 2 = 0. 0000R- squared = 0. 9962Root MSE = 164. 13

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -| Robust

f di k | Coef . St d. Err . z P>| z| [ 95% Conf . I nterval ]

- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -ai dk | . 365944 . 3457708 1. 06 0. 290 - . 3117544 1. 043642f di a | 5. 275206 . 137554 38. 35 0. 000 5. 005605 5. 544806gdp | - . 03536 . 697027 - 0. 05 0. 960 - 1. 401508 1. 330788bop | - 3. 883426 3. 356176 - 1. 16 0. 247 - 10. 46141 2. 694557

s | - . 5759026 1. 026894 - 0. 56 0. 575 - 2. 588578 1. 436773i nf l a | . 5449874 . 7126974 0. 76 0. 444 - . 8518738 1. 941849

l agf di k | - . 1214462 . 041677 - 2. 91 0. 004 - . 2031316 - . 0397607 _I countr y_2 | - 32. 94913 81. 29557 - 0. 41 0. 685 - 192. 2855 126. 3873 _I countr y_3 | - 28. 36329 83. 33979 - 0. 34 0. 734 - 191. 7063 134. 9797 _I countr y_4 | 254. 9483 178. 1031 1. 43 0. 152 - 94. 12738 604. 0239 _I countr y_5 | 67. 0228 68. 38112 0. 98 0. 327 - 67. 00174 201. 0473 _I countr y_6 | - 165. 8551 54. 241 - 3. 06 0. 002 - 272. 1655 - 59. 54472 _I count r y_7 | - 327. 7702 47. 99942 - 6. 83 0. 000 - 421. 8474 - 233. 6931 _I count r y_8 | - 261. 036 99. 06572 - 2. 63 0. 008 - 455. 2013 - 66. 87078 _I year _1996 | - 134. 3955 75. 95101 - 1. 77 0. 077 - 283. 2567 14. 46578 _I year _1997 | - 91. 30764 75. 04342 - 1. 22 0. 224 - 238. 39 55. 77476 _I year _1998 | - 140. 3063 77. 53194 - 1. 81 0. 070 - 292. 2661 11. 6535 _I year _1999 | - 160. 2621 83. 29842 - 1. 92 0. 054 - 323. 524 2. 999849 _I year _2000 | - 120. 8747 94. 73 - 1. 28 0. 202 - 306. 5421 64. 79264 _I year_2001 | - 29. 72452 103. 7509 - 0. 29 0. 774 - 233. 0725 173. 6235 _I year_2002 | - 43. 97234 96. 0431 - 0. 46 0. 647 - 232. 2134 144. 2687 _I year_2003 | - 49. 48728 89. 36801 - 0. 55 0. 580 - 224. 6454 125. 6708 _I year_2004 | - 67. 5958 90. 08821 - 0. 75 0. 453 - 244. 1654 108. 9738 _I year_2005 | - 130. 2547 93. 13816 - 1. 40 0. 162 - 312. 8022 52. 29269 _I year_2006 | - 186. 0564 83. 63181 - 2. 22 0. 026 - 349. 9718 - 22. 14109 _I year_2007 | - 76. 68261 145. 077 - 0. 53 0. 597 - 361. 0282 207. 663 _I year_2008 | - 101. 6831 105. 2958 - 0. 97 0. 334 - 308. 059 104. 6928 _I year_2009 | - 209. 3674 126. 4222 - 1. 66 0. 098 - 457. 1504 38. 41566 _I year_2010 | - 168. 6602 109. 7717 - 1. 54 0. 124 - 383. 8087 46. 48831

_cons | 56. 66896 122. 4495 0. 46 0. 644 - 183. 3276 296. 6655- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -I nst r ument ed: ai dk

I nst r ument s: f di a gdp bop s i nf l a l agf di k _I count r y_2 _I count r y_3 _I countr y_4 _I count r y_5 _I count r y_6 _I countr y_7 _I count r y_8 _I year_1996 _I year_1997 _I year _1998 _I year_1999 _I year_2000 _I year_2001 _I year_2002 _I year _2003 _I year_2004 _I year_2005 _I year_2006 _I year_2007 _I year _2008 _I year_2009 _I year_2010ai da hdi pop

. xi : i vregress 2sl s fdi k fdi a gdp bop s i nf l a l agf di k i . count ry i . year ( ai dk=ai da hdi pop) , vce ( robust) f i r sti . count r y _I count r y_1- 8 ( nat ur al l y coded; _I count r y_1 omi t t ed)i . year _I year_1995- 2010 ( natural l y coded; _I year_1995 omi t t ed)

Fi r st- stage r egressi ons- - - - - - - - - - - - - - - - - - - - - - -

Number of obs = 127F( 31, 95) = 12. 82Prob > F = 0. 0000



42/45

42

R- squared = 0. 6958Adj R- squared = 0. 5965Root MSE = 78. 0576

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -| Robust

ai dk | Coef . St d. Err . t P>| t| [ 95% Conf . I nterval ]- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

f di a | . 0132256 . 0113393 1. 17 0. 246 - . 0092857 . 0357369gdp | . 2370975 . 4439004 0. 53 0. 595 - . 6441562 1. 118351bop | 1. 990006 2. 043794 0. 97 0. 333 - 2. 067439 6. 04745

s | - . 1081177 . 0835845 - 1. 29 0. 199 - . 274054 . 0578185i nf l a | . 1610581 . 084145 1. 91 0. 059 - . 0059909 . 3281071

l agf di k | - . 0089388 . 0051734 - 1. 73 0. 087 - . 0192094 . 0013318 _I count r y_2 | - 397. 2809 335. 0539 - 1. 19 0. 239 - 1062. 447 267. 8853 _I count r y_3 | - 412. 7739 347. 638 - 1. 19 0. 238 - 1102. 923 277. 3747 _I countr y_4 | - 390. 9655 264. 4346 - 1. 48 0. 143 - 915. 9345 134. 0034 _I countr y_5 | - 487. 5525 322. 5936 - 1. 51 0. 134 - 1127. 982 152. 8767 _I countr y_6 | - 263. 9375 210. 5947 - 1. 25 0. 213 - 682. 0208 154. 1457 _I count r y_7 | - 365. 0384 283. 5315 - 1. 29 0. 201 - 927. 9196 197. 8428 _I count r y_8 | - 154. 7559 224. 5134 - 0. 69 0. 492 - 600. 4714 290. 9595 _I year _1996 | 11. 54945 38. 98828 0. 30 0. 768 - 65. 85207 88. 95097 _I year _1997 | - 57. 93269 37. 71147 - 1. 54 0. 128 - 132. 7994 16. 93405 _I year _1998 | - 30. 53556 36. 20092 - 0. 84 0. 401 - 102. 4035 41. 33235

_I year _1999 | 8. 4385 66. 54437 0. 13 0. 899 - 123. 6688 140. 5458 _I year _2000 | - 99. 01842 44. 44722 - 2. 23 0. 028 - 187. 2573 - 10. 77954 _I year _2001 | - 4. 751173 44. 78318 - 0. 11 0. 916 - 93. 65704 84. 15469 _I year _2002 | - 41. 09087 42. 94645 - 0. 96 0. 341 - 126. 3503 44. 16861 _I year_2003 | - 49. 88857 41. 68241 - 1. 20 0. 234 - 132. 6386 32. 86148 _I year_2004 | - 54. 13739 47. 09276 - 1. 15 0. 253 - 147. 6283 39. 35355 _I year_2005 | - 50. 27199 45. 59979 - 1. 10 0. 273 - 140. 799 40. 25503 _I year_2006 | - 97. 99766 50. 59525 - 1. 94 0. 056 - 198. 4419 2. 446611 _I year_2007 | - 56. 23098 50. 61008 - 1. 11 0. 269 - 156. 7047 44. 24273 _I year_2008 | - 42. 46308 42. 12743 - 1. 01 0. 316 - 126. 0966 41. 17044 _I year_2009 | - 56. 51236 42. 77827 - 1. 32 0. 190 - 141. 438 28. 41323 _I year_2010 | - 57. 96341 37. 74526 - 1. 54 0. 128 - 132. 8972 16. 97041

ai da | . 107984 . 0362373 2. 98 0. 004 . 0360439 . 1799241hdi | 197. 392 251. 8047 0. 78 0. 435 - 302. 5037 697. 2876pop | - 2. 096124 1. 81476 - 1. 16 0. 251 - 5. 698878 1. 50663

_cons | 369. 0136 390. 0732 0. 95 0. 347 - 405. 3796 1143. 407- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

I nst r ument al var i abl es ( 2SLS) r egr essi on Number of obs = 127Wal d chi 2( 29) = 7848. 48Prob > chi 2 = 0. 0000R- squared = 0. 9962Root MSE = 164. 13

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -| Robust

f di k | Coef . St d. Err . z P>| z| [ 95% Conf . I nterval ]- - - - - - - - - - - - - +- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

ai dk | . 365944 . 3457708 1. 06 0. 290 - . 3117544 1. 043642f di a | 5. 275206 . 137554 38. 35 0. 000 5. 005605 5. 544806gdp | - . 03536 . 697027 - 0. 05 0. 960 - 1. 401508 1. 330788bop | - 3. 883426 3. 356176 - 1. 16 0. 247 - 10. 46141 2. 694557

s | - . 5759026 1. 026894 - 0. 56 0. 575 - 2. 588578 1. 436773i nf l a | . 5449874 . 7126974 0. 76 0. 444 - . 8518738 1. 941849l agf di k | - . 1214462 . 041677 - 2. 91 0. 004 - . 2031316 - . 0397607

_I countr y_2 | - 32. 94913 81. 29557 - 0. 41 0. 685 - 192. 2855 126. 3873 _I countr y_3 | - 28. 36329 83. 33979 - 0. 34 0. 734 - 191. 7063 134. 9797 _I countr y_4 | 254. 9483 178. 1031 1. 43 0. 152 - 94. 12738 604. 0239 _I countr y_5 | 67. 0228 68. 38112 0. 98 0. 327 - 67. 00174 201. 0473 _I countr y_6 | - 165. 8551 54. 241 - 3. 06 0. 002 - 272. 1655 - 59. 54472 _I count r y_7 | - 327. 7702 47. 99942 - 6. 83 0. 000 - 421. 8474 - 233. 6931 _I count r y_8 | - 261. 036 99. 06572 - 2. 63 0. 008 - 455. 2013 - 66. 87078 _I year _1996 | - 134. 3955 75. 95101 - 1. 77 0. 077 - 283. 2567 14. 46578 _I year _1997 | - 91. 30764 75. 04342 - 1. 22 0. 224 - 238. 39 55. 77476 _I year _1998 | - 140. 3063 77. 53194 - 1. 81 0. 070 - 292. 2661 11. 6535 _I year _1999 | - 160. 2621 83. 29842 - 1. 92 0. 054 - 323. 524 2. 999849 _I year _2000 | - 120. 8747

ODA FDI Asean Tu Thuy_Vu Thui

Documents