IRA and Local Government Expenditure

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 24

Philippine Political Science Journal

ISSN: 0115-4451 (Print) 2165-025X (Online) Journal homepage: http://www.tandfonline.com/loi/rpsj20

The relationship between IRA and local


government expenditures: evidence from a cross-
section of Philippine cities

Tristan Canare

To cite this article: Tristan Canare (2016) The relationship between IRA and local government
expenditures: evidence from a cross-section of Philippine cities, Philippine Political Science
Journal, 37:3, 167-189, DOI: 10.1080/01154451.2016.1236476

To link to this article: http://dx.doi.org/10.1080/01154451.2016.1236476

Published online: 29 Nov 2016.

Submit your article to this journal

Article views: 6

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=rpsj20

Download by: [The UC San Diego Library] Date: 09 January 2017, At: 05:34
Philippine Political Science Journal, 2016
VOL. 37, NO. 3, 167–189
http://dx.doi.org/10.1080/01154451.2016.1236476

The relationship between IRA and local government


expenditures: evidence from a cross-section of Philippine
cities
Tristan Canare
Department of Economics, Ateneo de Manila University, Quezon City, Philippines

ABSTRACT ARTICLE HISTORY


Intergovernmental fiscal transfer (IFT) is one of the ways in which local Received 11 September 2015
government units (LGUs) source their funds. In the Philippines, the Accepted 29 August 2016
most common IFT is a lump-sum grant called the Internal Revenue
KEYWORDS
Allotment (IRA). Empirical studies show that the unconditional Intergovernmental fiscal
lumpsum grant is the IFT type that stimulates spending the least. transfers; internal revenue
Using data from Philippine cities, this paper finds evidence that total allotment; cities; Philippines;
city government expenditures is positively associated with IRA. A local government
one peso increase in IRA is associated with an almost unitary peso expenditures
increase in city government expenditures. This estimate is much
higher than the result of most similar empirical studies using data
from other countries. Moreover, this relationship varies according to
the capacity of the local government to generate its own funds. IRA
has a smaller marginal effect on local government spending for LGUs
with more locally sourced revenues. Furthermore, not all components
of local government spending are affected by the IRA. Results also
suggest that different income sources have different marginal effects
on spending.

Introduction and objectives


Intergovernmental fiscal transfer (IFT) is one of the ways in which local government units
(LGUs) source their funds. There are two types of IFTs: general purpose (unconditional) grants
and specific purpose (conditional) grants. In the Philippines and in many other developing
countries, the common type of IFT in place is an unconditional grant called revenue sharing
(Shah 2007). Revenue shares are legislated and automatic appropriations from the national
government to LGUs are drawn from collections either by multiple levels of government or
by one level (Rao 2007). In the Philippines, revenue shares come in the form of the Internal
Revenue Allotment (IRA). IRA is divided among different levels of local governments –
provinces, cities, municipalities and barangays – according to population and land area. IRA
is sourced from a predetermined share of national government revenue. Allocation of the
IRA to LGUs is stipulated in the Local Government Code of 1991. It serves the purpose of

CONTACT  Tristan Canare  tristan.canare@gmail.com


© 2016 Philippine Political Science Association (PPSA)
168    T. Canare

decentralizing some fiscal decisions to local governments with the aim of efficiently ­delivering
public services to the local populace.
One of the criticisms against general purpose unconditional grants in general is that they
may have no or little relation with actual expenditures or expenditure needs of the recipient
local government unit. The unconditional grant is the type of IFT that stimulates local gov-
ernment spending the least, usually increasing LGU expenditure by only 50 percent of the
transfer, with the remaining 50 percent going to tax relief for residents so they can spend
more on goods (Shah 2007). The reason for this weak responsiveness of local spending is
there is no condition attached to the grant. Local governments are free to use (or not use)
it the way they want, including parking it as savings or investment, or providing tax relief to
the citizens. In the Philippines, for many municipalities with weak capacity to generate their
own income, IRA accounts for almost all of their revenues. The amount of their expenditures
is therefore largely tied to the amount of IRA that they receive. However, for cities, the pres-
ence of economic and commercial activities gives the local government varied sources of
income such as business fees, local taxes, and real property taxes.
In the case of the Philippines, while IRA is formula based, the amount of local government
expenditures is not part of the criteria for its allocation, whereas land area and population
are. IRA is also unique in such a way that it is not purely unconditional. Over the years, the
Department of Interior and Local Government (DILG) has issued guidelines on how IRA
should be spent. For instance, a 2007 memorandum stipulated that 20 percent of IRA should
be spent on development projects (DILG 2007); while a 2012 order required 1 percent of IRA
funds received by local governments to be used for programs for children (DILG 2012a). This
makes IRA mildly conditional, although what exactly qualifies as a development project or
a children’s program is not explicitly stated in these circulars.
The main objective of this paper is to find out about the relationship between local gov-
ernment expenditures and the amount of IRA received using data from Philippine cities.
Specifically, this paper will look not only at the direction of the relationship between the
two variables, but also the degree of responsiveness of expenditures to IRA. The local gov-
ernment expenditures to be tested involve not only total spending, but also several com-
ponents of LGU expenditures. A secondary objective is to compare the marginal effect of
IRA on local government spending with other revenue sources. This paper is organized as
follows. Following this introduction, the next part will present the hypothesis and conceptual
framework and a discussion on IRA. This will be followed by a review of related literature
and studies. Next, the econometric model used and data sources will be presented. This will
be followed by the discussion and interpretation of results. Finally, the paper concludes with
a summary and some conclusions.

Hypothesis and conceptual framework


Local governments obtain funds from different sources including property tax, other local
taxes, transfers from other local government units, non-tax revenues, and the IRA. For some
small LGUs such as fifth and sixth class municipalities, IRA is almost the only source of funds.
The main hypothesis of this paper is that IRA and local government expenditures have a
positive, but less than one-for-one, relationship because IRA causes expenditures to increase.
There is no compelling reason to believe that the direction of causality is opposite because
IRA is formula-based and expenditure is not part of its determinants. And as noted earlier,
Philippine Political Science Journal   169

some literature finds evidence that unconditional lump sum fiscal transfer is the least
­stimulating type of IFT (Shah 2007), resulting in the hypothesis that the marginal effect is
less than one. As discussed earlier, IRA is some kind of an “impure” unconditional transfer,
because of regulations that a portion of it should be spent on such purposes as development
(20 percent) and children’s (1 percent) programs. However, which programs exactly meet
these purposes is not explicitly stated in the regulations. Nevertheless, this study is designed
such that results will show which components of LGU spending IRA has significant
effect on.
Theoretically, the effect of a lump sum unconditional fiscal transfer is an expansion of the
local government’s budget line (Shah 1994). In contrast to conditional grants, unconditional
IFT does not affect the relative price of public goods because the local government is free
to use it however they want (Shah 2007). And this discretion over how to spend (or not
spend) unconditional fiscal transfers is the reason why it is the least expenditure-stimulating
type of IFT.
Factors affecting local government expenditure can be aggregated into two major groups
– those that affect the need to spend and those that affect the capacity to spend. Even if an
LGU needs to spend, it will not do so if there are no funds available. Conversely, if an LGU
has the capacity to spend, but if there is hardly anything to use those funds for, then it will
rather save them or invest them somewhere else. Factors that affect the “need” and “capacity”
to spend are important in this study because we have to control for them in testing the
hypothesis.

The Internal Revenue Allotment


The IRA is provided to allow LGUs to carry out functions that are decentralized to them by
the Local Government Code (LGC) of 1991 (Republic Act 7160). This law decentralized many
functions previously handled by the national government with the premise that local gov-
ernments are nearer to the people, and thus can better respond to their needs and prefer-
ences. Hutchcroft (2012) describes IRA as the most contentious provision in the LGC, as some
members of Congress opposed the transfer of power to local governments.
The LGC of 1991 mandates that IRA will be allocated as follows: 23 percent goes to prov-
inces and cities each, 34 percent goes to municipalities, and 20 percent goes to barangays.
It is distributed among provinces, cities, municipalities, and barangays according to the
following criteria: 50 percent for population, 25 percent for land area, and 25 percent for
equal sharing. In 2013, the national government allocated PhP71 billion of IRA to provinces,
PhP69 billion to cities, PhP103 billion to municipalities, and PhP59 billion to barangays. This
allocation deviates slightly from the proportion stipulated in RA 7160 because of the
Guaranteed Share component, or the cost of devolved national functions and city-funded
hospitals when the law was implemented (DILG 2012b). It is noticeable that the share of
cities and provinces were almost the same even if the latter has a much wider administrative
scope than the former. Similarly, IRA allocation for municipalities is only 49 percent higher
than cities, even if there are much more municipalities than cities. As of the end of 2015,
there are 1,489 municipalities in the Philippines as opposed to only 145 cities. Both cities
and municipalities are composed of barangays and both cities and municipalities combine
to form a province. While some cities are politically independent from their home province
– the chartered cities – most are component cities, or those governed under provinces.
170    T. Canare

Thus, it can be said that cities and municipalities are at the same level in the local government
structure.
In comparing spending patterns, all provinces spent a total of PhP78.94 billion in 2013
in operating and non-operating expenditures; while cities spent PhP141.85 billion, and
municipalities PhP98.48 billion. Data in the previous paragraph show that overall, provinces
received 2.9 percent more IRA than cities in aggregate, but cities spent almost 80 percent
more than provinces. Similarly, municipalities received 49.3 percent more IRA than cities,
but cities spent 44 percent more than municipalities. When looking at aggregate figures, it
seems that IRA has a weak correlation with LGU spending.
Some literature argues that one of the main problems with IRA is that it promotes depend-
ence among local government units and discourages revenue generation, resulting in IRA
accounting for a disproportionately large share of local government revenues (see Literature
review section). For cities, IRA accounts for 62 percent of total local government revenues
in 2013. But this figure varies widely across cities, particularly among income classes; lower
income cities are more dependent on IRA than high income ones. Among first class cities,
the share of IRA to total revenue is 44 percent, less than 58 percent for second class cities,
67 percent for third class, 77 percent for fourth class, and 87 percent for fifth and sixth class.
This suggests that dependence on IRA could depend on the local government’s other sources
of income.
Cities, compared to municipalities, are less dependent on IRA for their income. Manasan
(2007) reports that from 1992 to 2003, IRA accounts for 47 percent of cities’ income net of
borrowing, much lower than the 74 percent for municipalities. According to the 2013 data
used in this study, the share for cities is 62 percent.

Review of related literature


This section will focus on a review of empirical literature and studies on determinants of
local government expenditures. Aside from looking at previous studies on the relationship
between local spending and revenue shares, this will also give a glimpse of local government
spending behavior and bring forward relevant control variables that should be used. Some
of the studies explored here also looked at the relationship between local government
spending and intergovernmental fiscal transfers using data from other countries. In addition,
this section also reviews the literature on IRA in the Philippines.
Using a simultaneous equation model and data from the state of Florida in the United
States, Loehman and Emerson (1985) estimated that local government expenditure has
an almost unitary positive elasticity with population, 0.05 with income, and 0.04 with IFT.
They also found evidence that income and intergovernmental transfers are positively
correlated with local government spending. Meanwhile, using a wider panel of 48 US
states for 11 years (1984 to 1994), McCarty and Schmidt (1997) concluded that income
and IFT exert a significant but only small influence on local state expenditures. The authors
attributed the small effect to the behavior of state governments of decreasing future taxes
when they experience a surge in income and IFT. They also concluded that past spending
affects current spending, and that an increase in spending on one category does not
necessarily affect spending on other categories. This implies that local governments
finance additional spending in one category through taxes or inward transfers rather than
reduction in spending on another.
Philippine Political Science Journal   171

A related study by Ram (1988) using an even larger body of federal and state data in the
US from 1929 to 1983 employed Granger causality to test causation between government
expenditure and revenue. The author found evidence of bi-directional causality between
revenue and spending in some sectors. However, the general result is that causation runs
from expenditure to revenue in local governments, in contrast to the federal government
where revenue causes expenditure.
On the other hand, Caputo and Cole (1975), using data from American cities with popu-
lations of over 50,000, concluded that proceeds from revenue sharing generally go to specific
expenditures, and are not distributed in overall LGU spending. If this holds in the Philippines,
we may expect to find a positive IRA‒LGU spending relationship only for particular compo-
nents of local government expenditures. A comprehensive study by Schmandt and Stephens
(1963) using 1957 data from the US concluded that state aid is the biggest factor that affects
local government spending per capita. Additionally, LGU spending per capita is positively
related to land area, but it has no significant relationship with population and population
density – a finding that even the authors admitted is controversial.
A broad study by Merrifield (2000) used newer data from 1981 to 1991 on US states’
spending. The author concluded that area, per capita income, and percentage of senior
citizens to the total population are significant determinants of local government spending.
Finally, Farnham (1990) used citizen influence, a rather abstract variable, in explaining local
government spending and found it to be significant.
While there are plenty of empirical studies on fiscal transfers using foreign data, the same
cannot be said for the Philippines, particularly IRA. Most literature about IRA in the Philippines
uses qualitative analysis. Hutchcroft (2012) studied the political dynamics of IRA, and argued
that while it was implemented to enable local governments to perform the functions
devolved to them by the LGC, it is affected by the patronage politics that existed even before
the LGC was enacted. Hutchcroft further concludes that IRA only changed the division of
the pie of patronage politics, transferring some of it from the national to the local government;
and that it promotes dependence among local governments on the national government
for funding.
The view that IRA promotes dependence among local governments is a common
conclusion among several studies (Manasan 2005; Llanto 2009; Capuno 2005). Dependence
on IRA arises from the local government’s lack of incentive to generate its own revenue, as
IRA depends only on population and land area, and leaves out own revenue generation in
computing the IRA share. Capuno (2005) also highlights other flaws in IRA, including
providing disincentives to local governments to increase their tax collection effort, a view
also shared by Manasan (2005), and promoting inefficient spending. One result of not taxing
enough is that it induces citizens to demand less accountability from local officials. To support
this argument, Manasan (2005) used a panel data of local governments, and found empirical
evidence that IRA substitutes for, rather than stimulates, local tax collection.
Some studies point to the dysfunctions of IRA; several studies have called for a revision
of the IRA formula to make it more responsive to development needs and to reward per-
formers. Capuno (2005) proposes that the IRA formula should create incentives to raise
revenue collection among local governments, while Gatmaytan (2001) suggests removing
the bias in favor of cities. Manasan (2005) argues that the IRA formula should be revised such
that it closes the gap in fiscal capacities of local governments. Similarly, Llanto (2009) argues
172    T. Canare

that the IRA distribution formula should be modified in such a way that the spending capacity
of the local governments will match their expenditure needs.

Methodology and data source


Regression model
To test the hypothesized relationship between IRA and local government expenditures, the
following regression equation was estimated in this paper:
expi = 𝛽0 + 𝛽1 irai + 𝛽i Xi + 𝜇 (1)
where expi is the expenditure of city i in Philippine pesos in the year 2013, irai is the IRA
received by city i in the same year, Xi is a vector of control variables that affect spending level
of cities, and μ is the error term.
To control for other factors that affect local government spending, the following control
variables were included (variable names in parenthesis): population (pop), area in hectares
(area), total locally sourced revenue (locsource), loans availed for the year (loan), amount of
inter-local transfers received (intrlocal), budget surplus or deficit (surplus), percentage infla-
tion rate in the region where the city is located (inflatn), and number of barangays in the city
(brgy). This is the base model. Population, land area, and number of barangays are factors
that could affect an LGU’s need to spend; while locally sourced revenue, loans, inter-local
transfers, and budget surplus are the factors that affect an LGU’s capacity to spend.
As will be discussed later, several variations of the base model will be used to test for
stability of results. One of these variations is the removal of area and population among the
control variables. Because these two variables are used in computing for IRA allocation, they
are highly correlated with IRA, which can cause multicollinearity. Multicollinearity causes
the variance of the coefficients of the correlated variables to be larger than they should. The
implication of this is that variables may appear statistically insignificant even if they are
actually significant. In the base model, population and land area were included as control
variables because it is possible that they affect local government spending. To isolate the
relationship between IRA and local government expenditures, other factors that affect
spending should be controlled for as much as possible. If population and land area do indeed
affect local government spending, removing them from the control variables could bias the
estimated relationship between IRA and local spending, and there is a bias‒variance trade-
off. Using different model variations, wherein area and population are included in some and
not in others, allows for testing the stability of results. In particular, if the statistical signifi-
cance and coefficient of IRA does not change much when area and population are included
as controls compared to when they are not included, then it makes the results more
compelling.
To account for possible regional differences and external effects in spending patterns of
LGUs, regional dummies will be introduced. Regional patterns in spending may arise in several
ways. Some regions (e.g. Region 5) are prone to typhoons and this may induce the cities there
to spend more on disaster mitigation and relief operations. Regions with high levels of poverty
may have unusually high spending on redistribution and public services. The use of regional
dummies will be incorporated in a new equation, which will be referred to as Model 2. Model
2 includes all variables in the base model plus the regional dummies.
Philippine Political Science Journal   173

A lagged dependent variable (same variable name prefixed by “l_”) will also be added to
the base model to account for past factors that may cause current variations in local gov-
ernment expenditures that are difficult to account for in other ways. For instance, a natural
calamity that occurred the year before may increase spending in the current year because
rehabilitation activities continue into the present year. The use of a lagged dependent var-
iable will be incorporated in a new equation, which will be referred to as Model 3. Model 3
includes all variables in the base model plus the lagged dependent variable. A fourth model
will be introduced which includes all variables in the base model plus regional dummies
and a lagged dependent variable. The fifth model includes all variables in Model 4 plus an
interaction term between IRA and locally sourced income (ira*locsource). This interaction
term will show if the relationship between IRA and city government spending changes with
locally sourced income, i.e. if IRA has smaller or larger marginal effect on spending for city
governments with more locally sourced revenues. The sixth model includes all variables in
the fifth model less area and population.
All models will be estimated with total expenditures and disaggregated expenditure
components as the dependent variable. This is to test how specific expenses vary with IRA.
Expenditure components used are general public services (exp_genpblc), education, culture,
sports, and manpower development (exp_educ), health, nutrition, and population control
(exp_health), labor and employment (exp_empl), housing and community development
(exp_comdev), social security, social services, and welfare (exp_sswlfr), economic services
(exp_econsrv), interest payments on debt (exp_debtint), and non-operating expenditures
(exp_nonoprex). Non-operating expenditures include capital purchases such as infrastructure
and equipment.
The regression model implemented here is similar to most empirical studies on local
government spending, where spending is expressed as a function of determinants of
demand for public services (Gamkhar and Shah 2007). Also included in the equation are
determinants of ability to spend and availability of funds (local government revenue varia-
bles). Ordinary Least Squares (OLS) was used in estimating Equation 1.

Data sources and data characteristics


Data used is a cross-section of 143 cities for the year 2013. Data on cities’ income and expend-
iture items came from the Bureau of Local Government Finance. Data on population, land
area, and number of barangays came from the Philippine Standard Geographical Classification
of the National Statistical Coordination Board (NSCB). Data on regional inflation also came
from NSCB; while population data came from the results of the 2010 census. Table 1 shows
the summary statistics for the variables included in the equation.

Results and discussion


Estimation results ‒ IRA
Table 2 shows the results of OLS estimation of Models 1 to 6 using total city government
expenditure as the dependent variable. The coefficients of ira in the six models are all positive
and significant. A 1 peso increase in IRA is associated with 0.80 to 1.13 pesos increase in total
city government expenditures, depending on the model. This is much higher than the result
of most similar empirical studies on other countries reviewed by Gamkhar and Shah (2007),
174    T. Canare

Table 1. Summary statistics.


Variable Number of observations Mean Standard deviation Minimum Maximum
Expend 143 993,000,000 1,310,000,000 162,000,000 7,920,000,000
exp_genpblc 143 416,000,000 470,000,000 56,500,000 3,010,000,000
exp_educ 143 75,900,000 181,000,000 999,456 1,550,000,000
exp_health 143 92,900,000 227,000,000 7,445,271 1,930,000,000
exp_empl 143 829,007 2,682,567 0 20,700,000
exp_comdev 143 39,400,000 134,000,000 0 1,170,000,000
exp_sswlfr 143 40,200,000 68,900,000 2,517,621 604,000,000
exp_econsrv 143 141,000,000 200,000,000 13,500,000 1,450,000,000
exp_debtint 143 15,900,000 54,100,000 0 617,000,000
exp_nonoprex 143 170,000,000 316,000,000 4,743,337 3,050,000,000
ira 143 486,000,000 378,000,000 179,000,000 2,810,000,000
pop 143 256,664 329,363 37,219 2,761,720
area 143 26,501 34,665 595 244,361
locsource 143 657,000,000 1,530,000,000 12,100,000 11,000,000,000
loan 143 39,500,000 142,000,000 0 1,370,000,000
intrlocal 143 4,198,378 13,900,000 0 115,000,000
surplus 143 346,000,000 809,000,000 −558,000,000 7,740,000,000
inflatn 143 3 1 2 5
brgy 143 54 80 9 897

where the majority of coefficients are less than 0.60. This provides evidence that IRA received
by Philippine cities stimulate more spending compared to similar transfers to local govern-
ments in other countries based on these studies.1
While IRA is positively correlated to local government spending, the degree of this rela-
tionship varies according to the city government’s ability to generate its own revenues. The
interaction term between IRA and locally sourced revenue (ira*locsource) is negative and
statistically significant, implying that as locally sourced revenues increase, the marginal effect
of IRA on local government spending weakens. Indeed, as discussed earlier, data shows that
as city income class rises (goes from 6th to 1st), the share of IRA in total revenues declines.
This implies that the ability to raise the city’s own revenue lessens the marginal effect of IRA
on the city government’s spending. As locally sourced revenues increase, the ability of IRA
to stimulate city government spending weakens.
Another noticeable result of the regression is the high R-squared, ranging from 0.970 to
0.979, which is high especially for cross-sectional data. This means that the model accounts
for almost all of the variations in local government spending. Moreover, the R-squared and
the coefficients are hardly different between Models 5 and 6. This means that dropping area
and population from the equation to control for multicollinearity did not change the results
materially. This, along with the insignificant coefficients of area and population in Models 1
to 5, further implies that these two variables have insignificant effect on local government
spending. If population and land area were to be considered as determinants of the LGU’s
expenditure needs, then this would have significant implications from a policy point of view.
While the six models show strong evidence that total city government expenditures are
positively associated with IRA, the same is not true for all components of city government
expenditures. The marginal effects of IRA on these components, and the full regression
results, are reported in Appendix 1. Among the nine components of expenditures tested,
IRA shows strong evidence of positive correlation only with two: health, and social security
and services. In the former, IRA has positive and significant coefficient in all six models; while
in the latter, IRA has positive and significant coefficient in four models (all except Models 4
and 5). An additional peso of IRA is associated with 0.06 to 0.51 peso higher spending on
Philippine Political Science Journal   175

Table 2. OLS regression of city government expenditures on IRA and control variables.
Dependent variable: City government expenditures in PhP
Base model Model 2 Model 3 Model 4 Model 5 Model 6

ira 0.796** 0.849* 0.865*** 1.075*** 1.131* 1.084***


(0.345) (0.450) (0.308) (0.407) (0.621) (0.143)
ira*locsource −1.37e‒10* −1.37e‒10*
(7.22e‒11) (7.22e‒11)
pop 126.7 56.38 −108.0 −391.2 −45.93
(445.1) (565.6) (473.4) (680.1) (650.4)
area 2,360 2,229 1,533 757.7 −207.8
(1,748) (2,140) (1,481) (1,714) (2,838)
locsource 1.072*** 1.069*** 1.024*** 1.033*** 1.030*** 1.030***
(0.0918) (0.0919) (0.115) (0.120) (0.115) (0.110)
loan −0.367 −0.396 −0.385 −0.478 −0.706 −0.703
(0.533) (0.572) (0.540) (0.626) (0.551) (0.532)
intrlocal 3.623** 3.671** 2.328* 1.658 2.509** 2.504*
(1.489) (1.658) (1.317) (1.534) (1.218) (1.286)
surplus −0.801*** −0.797*** −0.807*** −0.798*** −0.449* −0.449*
(0.113) (0.133) (0.114) (0.127) (0.249) (0.246)
inflatn −3.912e+07 −7.010e+07 −2.983e+07 −7.094e+07 2.530e+07 −1.412e+07
(2.566e+07) (6.025e+07) (2.268e+07) (6.875e+07) (3.160e+07) (6.371e+07)
brgy −277,933 −266,409 889,294* 1.574e+06* 1.505e+06* 1.497e+06*
(337,797) (401,129) (492,189) (947,580) (836,917) (840,913)
l_expend 0.0921 0.0818 0.107 0.108
(0.0949) (0.0946) (0.101) (0.100)
Region 1 −4.389e+07 −8.067e+07 8.267e+07 1.142e+08
(1.087e+08) (1.366e+08) (8.720e+07) (1.009e+08)
Region 2 −2.879e+07 −1.026e+08 −8.556e+07 −7.625e+06
(9.802e+07) (1.059e+08) (9.657e+07) (7.389e+07)
Region 3 1.229e+07 −1.169e+07 8.573e+07 1.297e+08
(7.032e+07) (8.453e+07) (8.966e+07) (1.025e+08)
Region 4A 2.387e+07 1.996e+07 1.221e+07 5.419e+07
(9.906e+07) (1.236e+08) (1.190e+08) (1.124e+08)
Region 4B 9.929e+07 7.062e+07 2.393e+08 2.775e+08
(2.028e+08) (2.368e+08) (2.195e+08) (2.308e+08)
Region 5 1.314e+07 −1.558e+07 3.046e+07 1.066e+08
(4.957e+07) (6.577e+07) (6.003e+07) (7.802e+07)
Region 6 2.045e+07 1.227e+07 3.705e+07 1.254e+08
(4.110e+07) (4.322e+07) (4.662e+07) (8.287e+07)
Region 7 1.163e+08* 1.233e+08* 1.322e+08
(5.946e+07) (6.904e+07) (1.262e+08)
Region 8 4.682e+07 −5.476e+07 −7.631e+07 3.414e+07
(4.048e+07) (7.309e+07) (7.078e+07) (9.727e+07)
Region 9 −1.069e+08
(9.498e+07)
Region 10 2.353e+07 2.203e+07 −1.820e+06 1.110e+08
(5.196e+07) (4.713e+07) (4.731e+07) (1.061e+08)
Region 11 1.328e+07 −3.154e+06 7.885e+07 1.457e+08
(5.605e+07) (7.020e+07) (6.698e+07) (9.426e+07)
Region 12 2.612e+07 7.757e+07** 9.498e+07* 1.786e+08**
(4.360e+07) (3.408e+07) (5.460e+07) (8.625e+07)
ARMM −2.021e+07 −8.286e+07 −5.650e+06 6.620e+07
(6.555e+07) (8.283e+07) (8.098e+07) (8.869e+07)
CAR 4.735e+07 −3.480e+07 −1.511e+07 5.643e+07
(1.157e+08) (1.083e+08) (7.632e+07) (8.014e+07)
CARAGA 4.221e+07 3.556e+07 5.566e+07 1.472e+08
(7.627e+07) (7.926e+07) (7.792e+07) (1.052e+08)
Constant 2.269e+08* 3.012e+08 1.372e+08 2.310e+08 −2.343e+08 −1.680e+08
(1.300e+08) (2.782e+08) (1.089e+08) (2.736e+08) (1.998e+08) (2.248e+08)

Observations 143 143 132 132 132 132


R-squared 0.972 0.973 0.970 0.971 0.979 0.979
Note: Robust standard errors in parentheses.
***
Significant at 1%; **significant at 5%; *significant at 10%.
176    T. Canare

health; and with 0.04 to 0.17 peso higher expenditures on social security and services. There
is large variation in the coefficients across the six models. But since we controlled for the
possible multicollinearity of area and population in Model 6, the most likely marginal effect
of IRA on these expenditure components are those at the lower part of the range (i.e. 0.06
on health and 0.04 on social security and services).
Aside from health and social security and services, results also show that IRA has a positive
and statistically significant relationship with spending on housing and community devel-
opment, labor and employment, and general public services, but only in Model 6, or when
area and population were removed from the equation to control for multicollinearity.
Respectively, the peso increase in these expenditure components for every peso increase
in IRA are 0.06, 0.0003, and 0.4. The regressions wherein the dependent variables are disag-
gregated spending also show fewer significant control variables and slightly lower R-squared
compared to the equations where the dependent variables are total LGU spending.

Some estimation results – control variables


The coefficients of the control variables also provide interesting results. Locally sourced
revenues are positive and significant determinants of local government spending. In fact,
its coefficient is higher than the coefficient of IRA from Models 1 to 3. However, as more
variables are controlled for, and as multicollinearity is addressed in Model 6, the coefficients
of IRA and locally sourced revenue become almost equal. A 1 peso increase in locally sourced
revenue is associated with a 1.03 peso increase in total expenditures in Models 4 to 6. This
means that IRA and locally sourced revenues have almost equal marginal effects on the level
of spending of city governments.
Inter-local transfers are also positive and significant in five of the six models (all models
except Model 4). Using the results of Models 5 and 6, a 1 peso increase in inter-local transfers
is associated with a 2.5 peso increase in local government expenditures. This is significantly
higher than the marginal effects of IRA and locally sourced income. This means that transfers
from other LGUs stimulate spending more than IRA and locally sourced income. This is pos-
sibly explained by the nature of transfers from other LGUs. While IRA is an (impure) uncon-
ditional grant, which means that recipient local governments can use (or not use) them
however they want, transfers from other LGUs are a mixture of conditional and unconditional
grants. Conditional grants require to be spent on specific goods and services. Moreover,
some transfers from other local governments require counterpart funds from the recipient
LGU.2 Meanwhile, the variable surplus is negative and significant. The lower the city’s budget
surplus – or the higher the deficit – the higher the city government expenditures. This is not
unexpected as local governments that spend more are expected to have higher budget
deficits or lower surpluses.
On the other hand, population and land area are both insignificant determinants of
expenditures. The number of barangays is positive and significant in four of the six models.
Lagged expenditures are insignificant. All six models have high R-squared with all exceeding
0.97 – an exceptionally high value for cross-sectional data. Even if the lagged dependent
variables were insignificant, and even for the models without a lagged dependent variable,
the R-squared was still high. This means that practically all of the determinants of city gov-
ernment spending were included in the model.
Philippine Political Science Journal   177

Discussions and implications


The results in Table 2 confirm our hypothesis that IRA is positively associated with total local
government expenditures. Although causality cannot be inferred just by using OLS, we can
infer that IRA is a cause and expenditure is an effect based on the arguments presented in
the Hypothesis and conceptual framework section. Reverse causality is doubtful because
IRA is formula-based and expenditure is not among the criteria for the computation of IRA
allocation, unless expenditure affects IRA through land area and population. This, however,
is disproved by the results of our regression, indicating that area and population are insig-
nificant determinants of LGU expenditures. Likewise, it seems that a good number of other
factors were controlled for in the equation, as suggested by the high R-squared and the
good number of control variables being highly significant.
While a positive relationship between IRA and local government spending is largely
expected, this study has several relevant results. One is the almost one-to-one response of
local government spending to IRA. This means that a peso increase in IRA leads to about the
same increase in city government spending. This marginal effect of IRA on local government
spending is higher compared to the marginal effect of comparable transfers to local gov-
ernments in other countries, based on the results of similar empirical studies (see Shah 2007
for a summary of these papers). However, although the relationship between IRA and local
government spending is roughly one-to-one on average, results show that this relationship
varies depending on the city’s ability to generate its own revenues. For cities with greater
ability to raise their own revenues, and therefore less dependent on IRA, the marginal effect
of IRA on local government spending is weaker.
Another is the almost similar effect of IRA and locally sourced revenue on local government
spending. The results show that a 1 peso increase in IRA and a 1 peso increase in locally sourced
revenues both lead to about 1 peso increase in local government expenditure. The marginal
effect of IRA is just higher by 5 centavos than the marginal effect of locally sourced revenues
in Model 6, 1.08 versus 1.03. This means that although IRA and locally sourced revenues
account for widely different shares in LGU income, their marginal effect on local government
spending is almost the same. Based on our 2013 dataset, IRA and locally sourced revenue
account for 61.6 percent and 35.9 percent, respectively, of city government revenue.
The revenue source that stimulates greater spending is inter-local transfers. These are
transfers from other local government units, both higher-level LGUs (provinces) and other
cities or municipalities. The marginal effect of inter-local transfers on city government spend-
ing is about 2.5 times the marginal effect of IRA and locally sourced income. As discussed
earlier, this could be due to the fact that much of the transfers from other LGUs are conditional
and require counterpart funding. Inter-local transfers, though, account for a mere 0.5 percent
of city government revenues on average.
Before proceeding, a clarification is in order. The results cited here show the marginal
effects of the different revenue components on local government spending, i.e. how much
an additional peso can stimulate spending. Marginal effect is the amount of increase in LGU
spending for every peso marginal increase in IRA, locally sourced revenue, or inter-local
transfers. Thus, even if inter-local transfers are able to stimulate local spending the most,
having the highest marginal effect, the total effect is not that much because they account
for only 0.5 percent of total city government revenue. It is only that in terms of stimulating
city government spending, inter-local transfers have the highest bang for one’s buck.
178    T. Canare

Another important result is the effect of IRA on the different components of city govern-
ment spending. Using Model 6, it has a positive and significant effect on health, social security
and services, housing and community development, labor and employment, and general
public services. IRA has the strongest effect on general public services – a 1 peso increase
in IRA leads to a 42 centavo increase in this expenditure component. For the other four
components, the marginal effect ranges from 0.0003 to 0.06 peso. It can be noted that among
the nine expenditure components tested in this study, these four components, along with
education, are the ones that can be considered developmental in nature. The regulation
that 20 percent of IRA funds should be spent for development purposes could help explain
the significant and smaller effect of IRA on these expenditure components.
An important incidental finding of this paper is what factors LGUs base their spending
on. Estimation results show that population and land area are insignificant factors in deter-
mining city LGU spending. This is shown by the insignificant coefficients of these two vari-
ables in the models where they are included; and by the results not changing by much when
they are taken out of the equation. Similarly, the number of barangays has a barely significant
effect on spending. On the other hand, IRA, locally sourced revenue, inter-local transfers,
and budget surplus/deficit are all significant determinants of expenditures. If population
and land area are used as measures of the expenditure needs of the local government, these
results imply that city governments base their spending on their capacity to spend rather
than on the need to spend. This is somewhat similar to the findings of Bird and Rodriguez
(1999) that there is not much relationship between fiscal transfers and incidence of poverty
in the Philippines. Indeed, Llanto (2009) argues that the IRA distribution formula should be
modified such that it will be more responsive to the expenditure needs of local governments.
From a normative sense, because the purpose of IRA is to help LGUs with their expenditure
needs on the devolved functions, IRA should be responsive to expenditure needs.

Endogeneity test
Gamkhar and Shah (2007) suggested that one of the critical issues in similar empirical studies
is failing to account for the endogeneity of fiscal transfers on spending. This is especially
worse in conditional transfers, but may also be present in unconditional transfers. The endo-
geneity test suggested by Wooldridge (2008, 527–528) was performed on the variable ira
for Models 1 to 6 with total city spending as the dependent variable. The result is that we
cannot reject the null hypothesis that ira is exogenous in all six models at a 10 percent level
of significance. Thus, there is no endogeneity problem for the six models.
IRA was also tested for endogeneity with health expenditures, social security and services,
housing and community development, labor and employment, and general public services.
For social security and services expenditures, there is no problem of endogeneity in any of
the six models. For health spending, there is a problem of endogeneity for Models 3, 4, 5,
and 6. For the three expenditure components in which IRA turned significant only in Model
6, labor and employment has no endogeneity problem.

Summary, conclusions and recommendations


Intergovernmental fiscal transfer is one way by which local governments source their funds.
Fiscal transfers are especially important to local governments in developing countries, where
Philippine Political Science Journal   179

LGUs have little source of revenue on their own. As in most other developing countries, the
type of IFT in place in the Philippines is a grant called revenue sharing, formally called Internal
Revenue Allotment in the country. Empirical studies show that an unconditional lump-sum
grant is the IFT type that stimulates spending the least.
There are several conclusions that can be derived based on the results of this study. First,
total city government expenditures are positively associated with IRA. A 1 peso increase in
IRA is associated with a 0.80 to 1.08 peso increase in total city government expenditures,
depending on the model. These estimates are much higher than the marginal effect of
comparable transfers to local government on local spending in other countries, using results
from similar empirical studies. Second is that this relationship varies according to the capacity
of the LGU to generate its own funds. IRA has a smaller marginal effect on local government
spending for LGUs that are able to generate more locally sourced revenues.
Next, while IRA has a positive effect on city government spending in aggregate, not all com-
ponents of local government spending are affected by the IRA. Only expenditures on health,
social security and services, housing and community development, labor and employment, and
general public services are positively correlated with IRA. IRA has the greatest marginal effect on
general public services – an additional peso of IRA is associated with 42 centavos greater spending
on this component. IRA has a much smaller marginal effect on the other four components.
Finally, results also suggest that locally sourced income has almost the same marginal
effect on local government spending as IRA. On the other hand, the marginal effect of
inter-local transfer is about 2.5 times greater. It should be emphasized that the results dis-
cussed here are marginal effects, or the peso increase in spending for every 1 peso increase
in IFT. Thus, although inter-local transfers provide the biggest bang for one’s buck in terms
of stimulating local government spending, it does not necessarily have a large total effect
on spending because it makes up only a small share of local government revenues.
It is recommended that future studies extend this paper to include local government
units aside from cities. Some of the results here might be different when applied to munic-
ipalities and provinces. The difference in the ability of cities and municipalities to generate
their own income may alter the results, as well as the cities and municipalities belonging to
different groups in the IRA allocation formula.

Notes
1. 
The local government transfer studies reviewed by Gamkhar and Shah are unconditional
revenue shares. Again, it should be noted here that IRA is not purely unconditional, in the
sense that 20 percent of it should be spent on development projects and 1 percent on children’s
programs. However, as stated earlier, these regulations are not clear on what projects are
considered developmental or for children.
2. 
The author would like to acknowledge this suggestion from an anonymous referee.

Acknowledgments
The author would like to thank two anonymous reviewers for useful comments and suggestions.
However, all remaining errors remain the author’s responsibility.

Disclosure statement
No potential conflict of interest was reported by the author.
180    T. Canare

Notes on contributor
Tristan Canare is a PhD student at the Ateneo de Manila University. Previously, he was with the World
Bank, the Asian Development Bank, and the Asian Institute of Management Policy Center. He holds a
Master’s degree in Economics from the University of the Philippines.

References
Bird, Richard, and Edgard R. Rodriguez. 1999. “Decentralization and Poverty Alleviation. International
Experience and the Case of the Philippines.” Public Administration and Development 19 (3): 299–319.
Capuno, Joseph J. 2005. “The Quality of Local Governance and Development Under Decentralization
in the Philippines.” University of the Philippines School of Economics Discussion Paper No. 0605.
Caputo, David A., and Richard L. Cole. 1975. “General Revenue Sharing Expenditure Decisions in Cities
over 50,000.” Public Administration Review 35 (2): 136–142.
DILG (Department of Interior and Local Government). 2007. Memorandum Circular 2007–5. Quezon
City: DILG.
DILG (Department of Interior and Local Government). 2012a. Memorandum Circular 2012–120. Quezon
City: DILG.
DILG (Department of Interior and Local Government). 2012b. Local Budget Memorandum No. 66: FY
2013 IRA Level and Other Local Budget Preparation Matters. Quezon City: DILG.
Farnham, Paul G. 1990. “The Impact of Citizen Influence on Local Government Expenditure.” Public
Choice 64 (3): 201–212.
Gamkhar, Shama and Anwar Shah. 2007. “The Impact of Intergovernmental Fiscal Transfers: A Synthesis
of the Conceptual and Empirical Literature.” In Intergovernmental Fiscal Transfers: Principles and
Practices, edited by Robin Boadway and Anwar Shah, 225–258. Washington: The World Bank.
Gatmaytan, Dante B. 2001. “Cost and Effect: The Impact and Irony of the Internal Revenue Allotment.”
Philippine Law Journal 75 (4): 629–679.
Hutchcroft, Paul D. 2012. “Re-Slicing the Pie of Patronage: The Politics of the Internal Revenue Allotment
in the Philippines, 1991-2010.” Philippine Review of Economics 49 (1): 109–134.
Llanto, Gilberto M. 2009. “Fiscal Decentralization and Local Finance Reforms in the Philippines.”
Philippine Institute for Development Studies Discussion Paper No. 2009–10.
Loehman, Edna, and Robert Emerson. 1985. “A Simultaneous Equation Model of Local Government
Expenditure Decisions.” Land Economics 61 (4): 419–432.
Manasan, Rosario G. 2005. “Local Public Finance in the Philippines: Lessons in Autonomy and
Accountability.” Philippine Journal of Development 60 (2): 31–102.
Manasan, Rosario G. 2007. “Decentralization and the Financing of Regional Development.” In The
Dynamics of Regional Development: The Philippines in East Asia, edited by Arsenio M. Balisacan and
Hal Hill, 275–315. Quezon City: Ateneo de Manila University Press.
McCarty, Therese A., and Stephen J. Schmidt. 1997. “A Vector-Autoregression Analysis of State-
Government Expenditure.” The American Economic Review 87 (2): 278–282.
Merrifield, John. 2000. “State Government Expenditure Determinants and Tax Revenue Determinants
Revisited.” Public Choice 102(1/2): 25–48.
Ram, Rati. 1988. “Additional Evidence on Causality between Government Revenue and Government
Expenditure.” Southern Economic Journal 54 (3): 763–769.
Rao, M. Govinda. 2007. “Resolving Fiscal Imbalances: Issues in Tax Sharing.” In Intergovernmental Fiscal
Transfers: Principles and Practices, edited by Robin Boadway and Anwar Shah, 319–338. Washington:
The World Bank.
Schmandt, Henry J., and G. Ross Stephens. 1963. “Local Government Expenditure Patterns in the United
States.” Land Economics 39 (4): 397–406.
Shah, Anwar. 1994. The Reform of Intergovernmental Fiscal Relations in Developing and Emerging Market
Economies. Washington: The World Bank.
Shah, Anwar. 2007. “A Practitioner’s Guide to Intergovernmental Fiscal Transfers.” In Intergovernmental
Fiscal Transfers: Principles and Practices, edited by Robin Boadway and Anwar Shah, 1–53. Washington:
The World Bank.
Wooldridge, Jeffrey M. 2008. Introductory Econometrics: A Modern Approach. 4th ed. Mason, Ohio: South-
Western Cengage Learning.
Philippine Political Science Journal   181

Appendix 1. Regression results with expenditure components as dependent variable.

Base model Model 2 Model 3 Model 4 Model 5 Model 6


Dependent variable: City government expenditures on housing and community development in PhP
ira 0.0230 0.0502 −0.108 −0.140 −0.106 0.0608***
(0.166) (0.219) (0.179) (0.215) (0.204) (0.0229)
ira*locsource 3.98e‒11*** 4.00e‒11***
(1.42e‒11) (1.46e‒11)
pop 139.1 113.9 218.1 272.5 152.6
(174.2) (225.9) (188.4) (232.1) (212.9)
area 71.56 −138.6 843.3 858.4 825.1
(821.9) (1,007) (873.1) (1,016) (981.6)
locsource 0.0432 0.0355 0.0308 0.0361 0.0268 0.0257
(0.0291) (0.0272) (0.0249) (0.0240) (0.0252) (0.0251)
loan −0.297** −0.316** −0.176 −0.182 −0.161 −0.169
(0.134) (0.145) (0.111) (0.131) (0.113) (0.121)
intrlocal 0.705 0.930* 0.351 0.515 0.651 0.721
(0.553) (0.488) (0.430) (0.551) (0.538) (0.526)
surplus 0.000742 0.00751 −0.0120 −0.0225 −0.103** −0.103**
(0.0485) (0.0474) (0.0414) (0.0438) (0.0473) (0.0485)
inflatn −4.683e+06 −2.106e+07 1.171e+06 8.120e+06 −1.711e+07 −2.055e+07
(8.290e+06) (2.003e+07) (7.572e+06) (2.301e+07) (1.489e+07) (2.040e+07)
brgy 186,508 177,612 −188,726 −284,067 −231,945 −231,620
(113,053) (124,943) (188,557) (228,013) (241,987) (236,760)
l_expend 0.426* 0.431 0.194 0.180
(0.241) (0.261) (0.240) (0.236)
Region 1 −2.292e+07 3.253e+07 −4.460e+07 −4.678e+07
(3.755e+07) (4.320e+07) (3.695e+07) (4.059e+07)
Region 2 −3.475e+07 2.873e+07 −2.258e+07 −4.440e+07*
(3.079e+07) (3.743e+07) (3.619e+07) (2.416e+07)
Region 3 −2.460e+07 1.442e+07 −4.460e+07 −4.774e+07
(3.012e+07) (3.322e+07) (3.226e+07) (3.460e+07)
Region 4A −5.625e+07* 384,766 −3.733e+07 −3.882e+07
(3.279e+07) (3.737e+07) (3.776e+07) (3.766e+07)
Region 4B 6.593e+07 8.445e+07 1.360e+07 2.519e+07
(8.652e+07) (6.488e+07) (6.263e+07) (6.948e+07)
Region 5 −6.728e+06 1.478e+07 −3.204e+07 −3.022e+07
(2.162e+07) (2.120e+07) (2.215e+07) (2.130e+07)
Region 6 3.998e+06 5.992e+06 −3.262e+07* −2.829e+07
(1.967e+07) (1.908e+07) (1.882e+07) (1.819e+07)
Region 7 3.056e+07 −13,532 6.478e+06
(2.456e+07) (2.211e+07) (1.919e+07)
Region 8 8.181e+06 1.747e+07 −9.700e+06 −1.788e+06
(2.079e+07) (1.770e+07) (1.735e+07) (1.451e+07)
Region 9 −7.207e+06
(1.426e+07)
Region 10 −1.153e+06 −4.421e+06 −3.307e+07* −2.592e+07
(2.366e+07) (1.904e+07) (1.709e+07) (1.782e+07)
Region 11 −7.515e+06 9.006e+06 −4.480e+07 −4.171e+07
(2.401e+07) (2.624e+07) (2.840e+07) (2.880e+07)
Region 12 −7.628e+06 1.398e+07 −2.488e+07 −2.183e+07
(2.592e+07) (1.650e+07) (2.004e+07) (1.843e+07)
ARMM −1.032e+07 1.992e+07 −3.451e+07 −2.856e+07
(2.766e+07) (2.420e+07) (2.614e+07) (2.317e+07)
CAR −7.103e+06 2.884e+06 −1.988e+07 −1.588e+07
(2.362e+07) (2.922e+07) (2.773e+07) (2.506e+07)
CARAGA 1.153e+07 1.497e+07 −2.430e+07 −1.963e+07
(1.734e+07) (1.497e+07) (1.587e+07) (1.311e+07)
Constant −2.384e+07 3.981e+07 −9.865e+06 −3.724e+07 1.089e+08 9.960e+07
(4.429e+07) (9.691e+07) (5.026e+07) (1.086e+08) (8.654e+07) (8.905e+07)

Observations 143 143 132 132 132 132


R-squared 0.735 0.757 0.794 0.802 0.844 0.842

(Continued)
182    T. Canare

Appendix 1. (Continued)
Dependent variable: City government expenditures on interest payments on debt in PhP
ira −0.0674 −0.104* −0.0106 0.00903 0.00912 −0.0244
(0.0554) (0.0543) (0.0488) (0.0866) (0.0862) (0.0211)
ira*locsource −1.39e‒13 −1.87e‒12
(3.03e‒12) (2.70e‒12)
pop 124.7 179.6* −45.77 −62.09 −61.65
(84.55) (92.18) (61.39) (100.6) (105.5)
area 520.7* 485.1** 65.99 −52.72 −53.62
(269.7) (229.0) (297.4) (480.8) (474.9)
locsource 0.0197 0.0248 0.00589 0.00364 0.00367 0.00458
(0.0196) (0.0218) (0.0125) (0.0121) (0.0118) (0.0132)
loan −0.0234 −0.0219 −0.0449 −0.0394 −0.0396 −0.0338
(0.0478) (0.0492) (0.0568) (0.0581) (0.0564) (0.0538)
intrlocal 0.0581 −0.00518 −0.0672 −0.151 −0.150 −0.0932
(0.144) (0.231) (0.140) (0.217) (0.217) (0.246)
surplus −0.0499 −0.0574 0.00588 0.00844 0.00874 0.00648
(0.0422) (0.0449) (0.0128) (0.0125) (0.0177) (0.0194)
inflatn 7.036e+06 1.667e+07 −34,621 −3.106e+06 −3.049e+06 2.021e+06
(8.448e+06) (1.438e+07) (4.403e+06) (4.556e+06) (4.209e+06) (6.059e+06)
brgy −22,615 −33,490 201,834** 226,648** 226,494** 172,233
(73,385) (91,687) (79,543) (109,707) (109,568) (127,088)
l_expend 1.286 1.226 1.225 1.148
(0.852) (0.820) (0.830) (0.784)
Region 1 3.430e+07 −3.203e+06 −3.003e+06 5.807e+06
(2.441e+07) (7.499e+06) (6.601e+06) (1.035e+07)
Region 2 4.768e+07* −5.883e+06 −5.776e+06 4.311e+06
(2.868e+07) (2.295e+07) (2.390e+07) (9.649e+06)
Region 3 2.590e+07* −899,053 −751,190 4.329e+06
(1.430e+07) (8.895e+06) (9.538e+06) (8.811e+06)
Region 4A 1.750e+07 −3.230e+06 −3.197e+06 −168,567
(1.435e+07) (6.139e+06) (6.041e+06) (7.579e+06)
Region 4B 7.274e+07 −7.017e+06 −6.761e+06 1.353e+07
(6.147e+07) (1.139e+07) (1.010e+07) (2.076e+07)
Region 5 6.012e+06 −3.243e+06 −3.120e+06 −415,917
(8.814e+06) (6.369e+06) (6.293e+06) (6.219e+06)
Region 6 3.820e+06 7.806e+06 7.915e+06 8.909e+06
(9.137e+06) (6.223e+06) (6.577e+06) (5.840e+06)
Region 7 −2.052e+07 −8.627e+06 −8.610e+06 −1.393e+07
(2.013e+07) (1.440e+07) (1.430e+07) (1.364e+07)
Region 8 3.597e+06 −2.691e+06 −2.600e+06 1.542e+06
(1.193e+07) (1.154e+07) (1.113e+07) (1.105e+07)
Region 9 ‒ ‒ ‒ ‒

Region 10 6.593e+07 5.636e+07 5.646e+07 5.751e+07


(5.567e+07) (4.591e+07) (4.690e+07) (4.624e+07)
Region 11 8.075e+06 −1.711e+06 −1.563e+06 4.095e+06
(9.617e+06) (8.990e+06) (1.010e+07) (9.604e+06)
Region 12 −1.482e+06 7.911e+06 8.008e+06 7.882e+06
(8.180e+06) (9.542e+06) (9.170e+06) (9.327e+06)
ARMM 6.139e+06 −7.019e+06 −6.869e+06 −2.651e+06
(1.170e+07) (9.963e+06) (9.188e+06) (9.822e+06)
CAR −1.085e+06 −1.186e+07 −1.177e+07 −6.564e+06
(1.473e+07) (9.339e+06) (8.910e+06) (1.039e+07)
CARAGA 1.212e+07 4.276e+06 4.392e+06 7.754e+06
(1.252e+07) (6.830e+06) (6.657e+06) (6.199e+06)
Constant −1.433e+07 −5.453e+07 −2.391e+06 4.017e+06 3.612e+06 −1.185e+07
(2.826e+07) (5.436e+07) (2.299e+07) (2.471e+07) (2.137e+07) (2.109e+07)

Observations 143 143 132 132 132 132


R-squared 0.118 0.250 0.380 0.443 0.443 0.439

(Continued)
Philippine Political Science Journal   183

Appendix 1. (Continued)
Dependent variable: City government expenditures on economic services in PhP
ira 0.295 0.338* 0.0130 −0.107 −0.0784 0.0184
(0.199) (0.189) (0.0956) (0.156) (0.169) (0.0551)
ira*locsource −1.31e‒11 −1.73e‒11
(2.57e‒11) (2.54e‒11)
pop −378.4* −421.0* −116.9 18.58 29.27
(217.7) (245.7) (118.7) (184.0) (171.8)
area 328.1 180.6 531.6 929.6 782.1
(912.1) (904.6) (371.5) (687.8) (691.8)
locsource 0.246*** 0.255*** 0.0681 0.0768 0.0836 0.0803
(0.0697) (0.0835) (0.0587) (0.0662) (0.0711) (0.0704)
loan −0.460* −0.491 −0.108 −0.0937 −0.132 −0.118
(0.271) (0.308) (0.129) (0.137) (0.174) (0.165)
intrlocal 0.543 0.706 0.761** 0.868* 0.941* 1.144**
(0.526) (0.604) (0.355) (0.522) (0.514) (0.538)
surplus −0.282** −0.290** −0.0751 −0.0834 −0.0561 −0.0511
(0.119) (0.139) (0.0728) (0.0855) (0.0948) (0.0964)
inflatn −2.622e+07** −5.127e+06 −3.241e+06 3.306e+07 3.586e+07 4.680e+07
(1.289e+07) (3.574e+07) (6.801e+06) (3.603e+07) (3.676e+07) (3.346e+07)
brgy −131,542 −160,459 463,079** 400,422 389,332 265,885
(268,635) (297,078) (212,811) (331,408) (332,215) (317,828)
l_expend 0.788*** 0.801*** 0.778*** 0.782***
(0.244) (0.232) (0.265) (0.260)
Region 1 588,341 4.363e+07 5.831e+07 7.354e+07
(6.534e+07) (6.869e+07) (7.750e+07) (7.270e+07)
Region 2 −8.141e+07* 3.107e+07 3.395e+07 2.647e+07
(4.861e+07) (4.462e+07) (4.439e+07) (2.999e+07)
Region 3 5.519e+07 6.897e+07 8.042e+07 8.447e+07
(9.268e+07) (6.840e+07) (7.230e+07) (6.907e+07)
Region 4A −1.479e+06 4.660e+07 4.699e+07 4.939e+07
(5.021e+07) (5.053e+07) (4.960e+07) (4.743e+07)
Region 4B 2.058e+07 5.136e+07 6.893e+07 1.204e+08
(1.043e+08) (8.447e+07) (9.275e+07) (9.047e+07)
Region 5 −2.798e+07 1.107e+07 2.069e+07 2.387e+07
(3.397e+07) (2.596e+07) (3.287e+07) (3.131e+07)
Region 6 −4.526e+07 −3.920e+06 5.444e+06 7.202e+06
(3.569e+07) (1.670e+07) (2.675e+07) (2.676e+07)
Region 7 −5.001e+07 −5.625e+07 −5.373e+07 −7.014e+07*
(5.656e+07) (3.994e+07) (3.995e+07) (3.785e+07)
Region 8 −2.361e+07 −2.496e+07 −1.685e+07 −5.709e+06
(4.314e+07) (2.938e+07) (2.631e+07) (2.897e+07)
Region 9 - - - -

Region 10 −3.588e+07 −4.975e+07* −4.068e+07 −4.160e+07


(5.035e+07) (2.908e+07) (3.621e+07) (3.658e+07)
Region 11 −3.646e+07 2.590e+07 3.736e+07 5.043e+07
(4.809e+07) (2.683e+07) (3.876e+07) (3.719e+07)
Region 12 −1.249e+06 7.175e+06 1.581e+07 1.386e+07
(3.880e+07) (2.128e+07) (2.986e+07) (2.984e+07)
ARMM −5.579e+07 −2.254e+07 −1.049e+07 1.213e+06
(4.014e+07) (3.304e+07) (3.926e+07) (3.948e+07)
CAR −1.893e+07 9.088e+06 1.522e+07 2.761e+07
(3.485e+07) (3.170e+07) (3.363e+07) (3.397e+07)
CARAGA −1.109e+07 2.974e+06 1.291e+07 1.903e+07
(6.002e+07) (2.158e+07) (3.059e+07) (2.865e+07)
Constant 1.318e+08** 6.984e+07 5.949e+06 −1.124e+08 −1.408e+08 −1.907e+08
(6.359e+07) (1.493e+08) (3.181e+07) (1.364e+08) (1.511e+08) (1.420e+08)

Observations 143 143 132 132 132 132


R-squared 0.584 0.599 0.835 0.848 0.851 0.848

(Continued)
184    T. Canare

Appendix 1. (Continued)
Dependent variable: City government expenditures on education, culture, sports, and manpower development
in PhP
ira 0.0615 0.0859 −0.0277 0.000501 −0.00160 0.0178
(0.0633) (0.0744) (0.0740) (0.0778) (0.0813) (0.0200)
ira*locsource 3.86e‒12 4.81e‒12
(1.21e‒11) (1.24e‒11)
pop −40.99 −78.94 64.57 42.08 32.28
(71.05) (86.34) (85.92) (106.9) (91.73)
area −433.8 −385.4 21.78 −3.423 22.85
(315.5) (357.5) (366.7) (370.8) (399.6)
locsource 0.137*** 0.131*** 0.129*** 0.128*** 0.128*** 0.128***
(0.0184) (0.0199) (0.0222) (0.0229) (0.0232) (0.0227)
loan 0.246*** 0.255*** 0.268*** 0.281*** 0.288*** 0.285***
(0.0756) (0.0845) (0.0790) (0.0894) (0.0984) (0.0958)
intrlocal −0.323 −0.411 −0.0520 −0.0412 −0.0685 −0.105
(0.221) (0.273) (0.189) (0.219) (0.199) (0.196)
surplus −0.0962*** −0.0889*** −0.0946*** −0.0951*** −0.105* −0.106*
(0.0297) (0.0314) (0.0327) (0.0354) (0.0563) (0.0562)
inflatn −1.193e+06 −1.205e+07 1.568e+06 −3.276e+06 −4.811e+06 −7.607e+06
(4.817e+06) (1.047e+07) (5.271e+06) (1.381e+07) (1.169e+07) (1.091e+07)
brgy 192,378** 206,400** −134,720 −221,060 −219,673 −190,222
(82,688) (84,598) (133,337) (184,346) (182,910) (176,594)
l_expend 0.0471 0.0306 0.0271 0.0240
(0.0455) (0.0480) (0.0466) (0.0457)
Region 1 −3.531e+07* −1.512e+07 −2.057e+07 −2.526e+07
(1.859e+07) (2.544e+07) (1.959e+07) (1.915e+07)
Region 2 −2.095e+07 −6.592e+06 −9.073e+06 −1.395e+07
(1.619e+07) (1.899e+07) (1.542e+07) (1.208e+07)
Region 3 −3.742e+07* −3.078e+07 −3.484e+07* −3.739e+07*
(2.002e+07) (2.368e+07) (2.052e+07) (2.168e+07)
Region 4A −1.047e+07 −997,156 −1.917e+06 −3.430e+06
(1.679e+07) (2.203e+07) (2.090e+07) (2.055e+07)
Region 4B −6.548e+07 −4.753e+07 −5.370e+07 −6.409e+07
(4.350e+07) (5.008e+07) (4.791e+07) (4.798e+07)
Region 5 1.122e+06 9.057e+06 5.616e+06 4.132e+06
(1.070e+07) (1.329e+07) (1.324e+07) (1.368e+07)
Region 6 −4.541e+06 −4.289e+06 −7.533e+06 −8.306e+06
(1.120e+07) (1.176e+07) (1.524e+07) (1.556e+07)
Region 7 1.201e+07 −434,799 −826,113 2.507e+06
(1.550e+07) (1.801e+07) (1.992e+07) (1.988e+07)
Region 8 −1.047e+07 1.047e+07 7.939e+06 5.384e+06
(1.306e+07) (1.434e+07) (1.436e+07) (1.412e+07)
Region 9 ‒ ‒ ‒ ‒

Region 10 −1.921e+07 −1.980e+07 −2.238e+07 −2.219e+07


(1.850e+07) (2.112e+07) (2.448e+07) (2.402e+07)
Region 11 −2.759e+07** −2.118e+07 −2.543e+07 −2.888e+07
(1.185e+07) (1.650e+07) (1.849e+07) (1.866e+07)
Region 12 −6.322e+06 −2.134e+07** −2.430e+07* −2.445e+07*
(1.438e+07) (1.054e+07) (1.438e+07) (1.463e+07)
ARMM −1.634e+07 1.749e+06 −2.475e+06 −4.922e+06
(1.121e+07) (1.518e+07) (1.396e+07) (1.395e+07)
CAR −2.132e+06 2.136e+07 1.890e+07 1.559e+07
(1.399e+07) (1.635e+07) (1.310e+07) (1.298e+07)
CARAGA −1.502e+07 −1.287e+07 −1.608e+07 −1.794e+07
(1.569e+07) (1.725e+07) (1.893e+07) (1.875e+07)
Constant −3.954e+06 4.138e+07 4.732e+06 2.722e+07 3.841e+07 4.675e+07
(2.139e+07) (4.427e+07) (2.510e+07) (5.528e+07) (4.691e+07) (4.230e+07)

Observations 143 143 132 132 132 132


R-squared 0.951 0.955 0.946 0.951 0.951 0.951

(Continued)
Philippine Political Science Journal   185

Appendix 1. (Continued)
Dependent variable: City government expenditures on labor and employment in PhP
ira −0.00124 −3.89e-05 0.000284 0.000193 0.000558 0.000326*
(0.00286) (0.00423) (0.000873) (0.00119) (0.00106) (0.000190)
ira*locsource −2.90e‒13*** −3.05e‒13***
(8.23e‒14) (8.75e‒14)
pop 2.021 0.455 −1.026 −0.960 −0.425
(3.852) (4.957) (1.111) (1.427) (1.165)
area 1.797 −0.524 2.522 2.865 −0.0815
(14.10) (19.63) (4.420) (5.614) (5.035)
locsource −0.000105 −0.000515 1.78e‒05 5.24e‒05 7.00e‒05 6.33e‒05
(0.000780) (0.000743) (0.000196) (0.000197) (0.000169) (0.000161)
loan 0.000332 0.000332 0.000562 0.000583 7.57e‒06 6.42e‒05
(0.00473) (0.00470) (0.000918) (0.00100) (0.000876) (0.000832)
intrlocal 0.00607 0.00704 −0.00343 −0.00443* −0.00242 −0.00188
(0.0114) (0.0104) (0.00212) (0.00227) (0.00174) (0.00189)
surplus 0.00145* 0.00190* −0.000120 −0.000170 0.000626* 0.000643*
(0.000855) (0.000970) (0.000295) (0.000288) (0.000333) (0.000335)
inflatn −452,133** −1.198e+06** −12,591 47,821 106,449 141,609
(212,236) (594,809) (44,991) (124,537) (127,664) (131,992)
brgy −5,619** −5,570* 976.0 1,158 1,157 724.5
(2,722) (2,947) (1,413) (1,690) (1,580) (1,610)
l_expend 1.021*** 1.035*** 1.010*** 1.008***
(0.0677) (0.0710) (0.0462) (0.0450)
Region 1 −2.241e+06** 120,907 422,960* 480,398*
(1.047e+06) (235,252) (235,253) (248,333)
Region 2 −1.264e+06 91,344 191,940 250,048
(831,457) (217,381) (214,597) (167,004)
Region 3 −1.116e+06 198,032 418,352* 445,801*
(980,754) (201,848) (216,590) (230,118)
Region 4A −1.722e+06* 373,099* 361,178 374,606*
(1.034e+06) (214,443) (222,415) (224,562)
Region 4B −1.645e+06 406,400 762,416* 905,434**
(1.179e+06) (370,243) (387,912) (450,095)
Region 5 908,061 −134,029 134,765 155,611
(1.906e+06) (277,824) (206,437) (208,635)
Region 6 −512,972 54,861 279,242** 288,517**
(427,140) (102,784) (139,799) (144,993)
Region 7 608,112 83,721 150,138 104,688
(683,492) (117,559) (175,318) (180,705)
Region 8 280,156 −127,147 60,483 98,492
(497,026) (126,202) (152,429) (153,612)
Region 9 ‒ ‒ ‒ ‒

Region 10 606,819 122,857 341,307* 341,134*


(619,491) (147,783) (179,701) (175,734)
Region 11 −1.306e+06** 73,621 336,335* 381,353*
(536,101) (153,176) (186,408) (206,192)
Region 12 −548,845 109,748 308,094** 307,164**
(489,946) (114,843) (140,086) (142,387)
ARMM −829,353 43,898 314,797** 346,604**
(587,369) (136,216) (144,649) (155,376)
CAR 144,938 599,976 745,296 790,148*
(818,234) (364,066) (453,941) (474,341)
CARAGA 190,205 −117,390 121,261 147,263
(797,729) (150,097) (166,444) (165,702)
Constant 2.198e+06** 5.158e+06** 23,902 −281,129 −912,375 −1.016e+06*
(1.075e+06) (2.604e+06) (257,809) (581,990) (564,184) (562,056)

Observations 143 143 132 132 132 132


R-squared 0.276 0.340 0.973 0.976 0.983 0.983

(Continued)
186    T. Canare

Appendix 1. (Continued)
Dependent variable: City government expenditures on general public services in PhP
ira 0.00385 −0.0707 0.170 −0.0291 −0.0423 0.416***
(0.437) (0.403) (0.387) (0.454) (0.417) (0.123)
ira*locsource −2.22e‒11 −3.09e‒12
(1.96e‒11) (2.26e‒11)
pop 927.3** 1,024*** 457.3 639.4 680.6
(404.3) (370.7) (369.6) (436.6) (414.9)
area 1,779 1,949 64.42 977.6 807.9
(1,959) (1,810) (1,724) (2,063) (1,836)
locsource 0.327*** 0.334*** 0.276*** 0.275*** 0.264*** 0.277***
(0.0411) (0.0461) (0.0525) (0.0531) (0.0600) (0.0665)
loan 0.349** 0.392*** 0.223* 0.275** 0.234* 0.155
(0.135) (0.132) (0.127) (0.128) (0.132) (0.205)
intrlocal 2.219*** 2.075** 0.454 0.185 0.254 −0.371
(0.710) (0.873) (0.522) (0.613) (0.601) (0.811)
surplus −0.412*** −0.423*** −0.405*** −0.401*** −0.341*** −0.361***
(0.0794) (0.0840) (0.0695) (0.0710) (0.0958) (0.107)
inflatn 1.027e+07 7.420e+06 1.378e+07 1.354e+07 3.683e+07* −2.517e+07
(1.321e+07) (3.412e+07) (1.076e+07) (2.571e+07) (1.940e+07) (3.516e+07)
brgy −1.026e+06*** −1.057e+06*** 497,826 497,953 476,514 1.054e+06*
(344,878) (321,073) (376,693) (486,019) (498,248) (608,797)
l_expend 0.304** 0.306** 0.363** 0.336*
(0.136) (0.130) (0.175) (0.195)
Region 1 6.785e+07 3.270e+07 5.760e+07 −1.976e+07
(5.797e+07) (5.208e+07) (4.959e+07) (6.075e+07)
Region 2 1.760e+08** 1.307e+08* 1.230e+08* 4.164e+07
(8.020e+07) (7.668e+07) (7.235e+07) (3.554e+07)
Region 3 1.802e+07 1.151e+06 1.368e+07 −2.074e+07
(5.591e+07) (4.827e+07) (4.154e+07) (5.127e+07)
Region 4A 9.782e+07* 8.438e+07* 7.822e+07* 6.439e+07
(5.339e+07) (4.383e+07) (4.310e+07) (5.287e+07)
Region 4B −1.165e+07 −6.339e+07 −4.414e+07 −2.166e+08
(1.216e+08) (1.038e+08) (1.001e+08) (1.648e+08)
Region 5 4.123e+07 2.643e+06 −794,818 −7.461e+06
(5.443e+07) (3.485e+07) (2.880e+07) (3.609e+07)
Region 6 7.048e+07 3.863e+07 2.757e+07 3.913e+07
(5.570e+07) (3.713e+07) (2.741e+07) (3.985e+07)
Region 7 6.484e+07 4.277e+07 9.880e+07
(8.307e+07) (4.272e+07) (5.986e+07)
Region 8 1.178e+08 1.348e+07 −5.845e+06 −2.417e+07
(7.373e+07) (4.513e+07) (3.659e+07) (5.147e+07)
Region 9 −2.867e+07
(3.924e+07)
Region 10 3.085e+07 −118,201 −2.091e+07 1.289e+07
(8.200e+07) (5.361e+07) (4.653e+07) (5.316e+07)
Region 11 1.454e+08* 1.050e+08** 1.062e+08** 6.307e+07
(7.433e+07) (4.875e+07) (4.416e+07) (4.862e+07)
Region 12 −3.003e+06 4.311e+07 3.383e+07 5.569e+07
(6.978e+07) (3.731e+07) (3.027e+07) (3.716e+07)
ARMM 1.140e+08** 3.962e+07 4.157e+07 1.867e+07
(5.467e+07) (4.537e+07) (4.162e+07) (4.619e+07)
CAR 1.804e+08 1.224e+08 1.216e+08* 7.647e+07
(1.356e+08) (7.657e+07) (7.031e+07) (9.915e+07)
CARAGA 4.978e+07 3.897e+07 3.095e+07 2.129e+07
(6.675e+07) (3.953e+07) (3.201e+07) (3.985e+07)
Constant 5.513e+07 1.135e+07 −2.971e+07 −4.064e+07 −1.302e+08 2.729e+07
(7.014e+07) (1.277e+08) (7.327e+07) (1.294e+08) (1.052e+08) (1.260e+08)

Observations 143 143 132 132 132 132


R-squared 0.944 0.953 0.965 0.971 0.972 0.965

(Continued)
Philippine Political Science Journal   187

Appendix 1. (Continued)
Dependent variable: City government expenditures on health, nutrition and population control in PhP
ira 0.509** 0.513** 0.241** 0.227** 0.235** 0.0642**
(0.245) (0.213) (0.100) (0.104) (0.106) (0.0319)
ira*locsource −1.13e‒11 −1.28e‒11*
(7.75e‒12) (7.16e‒12)
pop −580.9** −614.9*** −255.2*** −212.8** −185.7*
(227.9) (196.9) (97.18) (102.5) (108.2)
area −2,072* −1,834* −720.8 −602.2 −687.9
(1,077) (945.7) (465.1) (489.4) (480.4)
locsource 0.174*** 0.173*** 0.151*** 0.157*** 0.158*** 0.157***
(0.0202) (0.0239) (0.0218) (0.0250) (0.0235) (0.0233)
loan 0.0749 0.0699 0.156** 0.166** 0.143* 0.154**
(0.0788) (0.0838) (0.0682) (0.0739) (0.0728) (0.0743)
intrlocal −2.090* −2.254* −1.154 −1.025 −0.946 −0.907
(1.119) (1.224) (0.762) (0.672) (0.674) (0.649)
surplus −0.118*** −0.113*** −0.107*** −0.118*** −0.0893 −0.0883*
(0.0374) (0.0415) (0.0372) (0.0441) (0.0544) (0.0532)
inflatn 7.037e+06 8.157e+06 1.160e+07* 3.516e+07* 3.924e+07** 4.513e+07***
(6.232e+06) (1.797e+07) (6.045e+06) (1.841e+07) (1.806e+07) (1.667e+07)
brgy 960,675*** 1.025e+06*** −125,660 −303,657 −307,710 −364,852
(223,521) (216,946) (188,486) (264,389) (262,780) (240,477)
l_expend 0.0606 0.0736 0.0781 0.0883*
(0.0483) (0.0490) (0.0492) (0.0512)
Region 1 −3.120e+07 3.361e+07 4.898e+07 6.108e+07**
(3.030e+07) (3.133e+07) (3.230e+07) (2.938e+07)
Region 2 −5.842e+07 −7.463e+06 −799,766 2.926e+07
(4.563e+07) (3.194e+07) (3.090e+07) (2.049e+07)
Region 3 −1.366e+07 1.438e+07 2.566e+07 3.488e+07
(3.250e+07) (3.373e+07) (3.063e+07) (2.962e+07)
Region 4A −4.606e+06 2.519e+07 2.749e+07 3.341e+07
(3.123e+07) (3.125e+07) (3.149e+07) (3.040e+07)
Region 4B −5.884e+07 −3.040e+06 1.433e+07 2.861e+07
(4.594e+07) (5.009e+07) (4.550e+07) (3.785e+07)
Region 5 −2.628e+07 1.732e+06 1.176e+07 1.702e+07
(2.426e+07) (2.301e+07) (2.031e+07) (1.914e+07)
Region 6 −2.083e+07 −1.676e+07 −7.256e+06 −4.590e+06
(2.785e+07) (2.462e+07) (2.077e+07) (1.989e+07)
Region 7 −1.144e+07 −4.768e+07 −4.614e+07 −5.071e+07*
(3.768e+07) (3.687e+07) (3.093e+07) (3.005e+07)
Region 8 −7.876e+07** −1.184e+07 −4.336e+06 428,519
(3.634e+07) (3.276e+07) (2.569e+07) (2.361e+07)
Region 9 ‒ ‒ ‒ ‒

Region 10 −3.790e+07 −4.674e+07 −3.884e+07 −3.882e+07


(3.789e+07) (3.470e+07) (3.018e+07) (2.951e+07)
Region 11 −6.213e+07* −3.703e+07 −2.476e+07 −1.686e+07
(3.297e+07) (2.956e+07) (2.765e+07) (2.760e+07)
Region 12 2.813e+07 −1.118e+07 −2.621e+06 −515,540
(4.012e+07) (2.669e+07) (2.264e+07) (2.183e+07)
ARMM −5.185e+07 6.305e+06 1.850e+07 2.215e+07
(3.415e+07) (2.379e+07) (2.026e+07) (1.887e+07)
CAR −4.713e+07 2.940e+07 3.642e+07 4.415e+07**
(3.608e+07) (2.702e+07) (2.276e+07) (2.013e+07)
CARAGA −3.167e+07 −2.299e+07 −1.361e+07 −8.750e+06
(2.813e+07) (2.774e+07) (2.265e+07) (2.140e+07)
Constant −9.378e+07*** −7.922e+07 −5.159e+07* −1.231e+08** −1.546e+08** −1.594e+08**
(3.195e+07) (6.478e+07) (2.632e+07) (6.121e+07) (6.582e+07) (6.527e+07)

Observations 143 143 132 132 132 132


R-squared 0.924 0.932 0.918 0.925 0.928 0.927
(Continued)
188    T. Canare

Appendix 1. (Continued)
Dependent variable: City government expenditures on non-operating expenditures in PhP
ira −0.155 −0.132 −0.163 −0.0869 0.0303 0.0413
(0.326) (0.442) (0.228) (0.293) (0.117) (0.0463)
ira*locsource −8.19e‒11*** −8.23e‒11***
(1.46e‒11) (1.40e‒11)
pop 59.23 7.124 −6.003 −124.4 5.476
(435.7) (559.0) (233.9) (306.1) (146.7)
area 2,258 2,234 1,007 861.9 77.64
(1,653) (2,089) (1,063) (1,353) (530.0)
locsource 0.0620 0.0519 0.00708 0.00494 0.0177 0.0175
(0.0897) (0.0887) (0.0268) (0.0259) (0.0229) (0.0219)
loan −0.322 −0.351 0.401*** 0.379*** 0.163 0.164
(0.532) (0.563) (0.122) (0.134) (0.117) (0.110)
intrlocal 2.419 2.510 −0.408 −0.749 −0.102 −0.0854
(1.570) (1.638) (1.004) (1.069) (0.865) (0.868)
surplus 0.218** 0.232* −0.155* −0.146* 0.0870 0.0873
(0.104) (0.125) (0.0814) (0.0773) (0.0618) (0.0616)
inflatn −3.343e+07 −7.396e+07 −2.223e+06 −8.035e+06 1.073e+07 1.159e+07
(2.464e+07) (5.926e+07) (9.990e+06) (3.127e+07) (2.598e+07) (2.233e+07)
brgy −242,917 −188,410 600,605 895,278* 911,510** 901,692**
(322,966) (391,323) (369,756) (479,861) (381,478) (354,755)
l_expend 1.172*** 1.167*** 1.094*** 1.095***
(0.198) (0.206) (0.0889) (0.0883)
Region 1 −6.435e+07 −3.390e+07 6.100e+07 6.215e+07
(1.073e+08) (6.614e+07) (5.348e+07) (4.704e+07)
Region 2 −3.815e+07 −6.547e+07 −3.099e+07 −3.212e+07
(9.320e+07) (6.695e+07) (4.385e+07) (3.397e+07)
Region 3 −6.784e+06 −4.046e+07 3.393e+07 3.416e+07
(6.870e+07) (6.300e+07) (5.457e+07) (5.135e+07)
Region 4A −1.940e+07 −6.505e+06 −499,721 −367,778
(8.981e+07) (5.818e+07) (4.742e+07) (4.538e+07)
Region 4B 7.716e+07 −5.118e+07 6.856e+07 7.274e+07
(1.836e+08) (8.818e+07) (7.210e+07) (6.290e+07)
Region 5 1.281e+07 −3.489e+07 3.449e+07 3.467e+07
(5.048e+07) (4.102e+07) (3.957e+07) (3.796e+07)
Region 6 2.953e+07 −2.482e+07 4.481e+07 4.491e+07
(4.197e+07) (3.668e+07) (3.600e+07) (3.527e+07)
Region 7 1.169e+08* 9.379e+06 3.307e+07 3.171e+07
(5.967e+07) (3.714e+07) (4.296e+07) (4.117e+07)
Region 8 4.083e+07 −7.180e+07 −1.660e+07 −1.573e+07
(4.140e+07) (4.679e+07) (4.634e+07) (4.454e+07)
Region 9 ‒ ‒ ‒ ‒

Region 10 2.429e+07 −4.797e+07 1.533e+07 1.524e+07


(5.141e+07) (3.421e+07) (4.044e+07) (4.028e+07)
Region 11 1.951e+07 −4.560e+07 3.990e+07 4.088e+07
(5.677e+07) (4.763e+07) (4.637e+07) (4.450e+07)
Region 12 3.453e+07 −4.199e+07 2.497e+07 2.476e+07
(4.432e+07) (5.533e+07) (5.824e+07) (5.778e+07)
ARMM −1.736e+07 −7.794e+07* 5.373e+06 6.318e+06
(6.401e+07) (4.207e+07) (3.871e+07) (3.665e+07)
CAR −6.584e+07 −6.068e+07 −2.066e+07 −1.971e+07
(5.848e+07) (3.964e+07) (4.354e+07) (3.989e+07)
CARAGA 4.579e+07 −4.778e+07 2.248e+07 2.292e+07
(7.584e+07) (3.698e+07) (3.445e+07) (3.298e+07)
Constant 1.813e+08 3.021e+08 2.550e+07 5.620e+07 −1.371e+08 −1.414e+08
(1.249e+08) (2.739e+08) (5.344e+07) (1.349e+08) (1.048e+08) (9.877e+07)

Observations 143 143 132 132 132 132


R-squared 0.567 0.576 0.903 0.906 0.945 0.945

(Continued)
Philippine Political Science Journal   189

Appendix 1. (Continued)
Dependent variable: City government expenditures on social security, social services, and welfare in PhP
ira 0.127** 0.168** 0.0839*** 0.0668 0.0716 0.0374***
(0.0527) (0.0713) (0.0261) (0.0495) (0.0437) (0.0113)
ira*locsource −2.76e‒12 −5.08e‒12
(5.48e‒12) (5.18e‒12)
pop −125.4** −154.3** −113.4*** −71.39 −67.21
(58.41) (77.73) (35.05) (52.68) (49.27)
area −93.17 −261.4 7.355 41.36 10.36
(233.6) (302.0) (126.7) (227.3) (194.7)
locsource 0.0626*** 0.0648*** 0.0483*** 0.0536*** 0.0541*** 0.0532***
(0.0126) (0.0130) (0.00907) (0.0101) (0.00994) (0.0101)
loan 0.0651 0.0662 0.117*** 0.130*** 0.124*** 0.133***
(0.0567) (0.0605) (0.0392) (0.0379) (0.0451) (0.0472)
intrlocal 0.0854 0.114 −0.00933 0.0994 0.120 0.213
(0.262) (0.250) (0.239) (0.195) (0.194) (0.194)
surplus −0.0618*** −0.0662*** −0.0580*** −0.0664*** −0.0594** −0.0575**
(0.0194) (0.0210) (0.0122) (0.0152) (0.0247) (0.0235)
inflatn 2.526e+06 1.105e+07 6.305e+06** 2.227e+07*** 2.318e+07*** 2.952e+07***
(5.383e+06) (1.599e+07) (2.893e+06) (7.677e+06) (7.406e+06) (7.643e+06)
brgy −188,579** −230,418*** −53,272 −189,401* −191,327* −266,561**
(73,572) (81,147) (98,832) (106,517) (104,658) (110,362)
l_expend 0.648*** 0.648*** 0.635*** 0.631***
(0.132) (0.108) (0.105) (0.100)
Region 1 9.382e+06 3.968e+07*** 4.305e+07*** 5.328e+07***
(2.690e+07) (1.327e+07) (1.302e+07) (1.355e+07)
Region 2 −1.752e+07 1.260e+07 1.339e+07 2.267e+07**
(2.609e+07) (1.017e+07) (9.764e+06) (9.530e+06)
Region 3 −3.237e+06 2.002e+07 2.244e+07 2.734e+07*
(2.571e+07) (1.434e+07) (1.368e+07) (1.389e+07)
Region 4A 2.484e+06 2.932e+07*** 2.946e+07*** 3.212e+07***
(2.209e+07) (1.014e+07) (1.016e+07) (1.047e+07)
Region 4B 491,523 3.727e+07** 4.090e+07*** 6.596e+07***
(3.080e+07) (1.592e+07) (1.489e+07) (2.168e+07)
Region 5 1.205e+07 2.279e+07** 2.511e+07*** 2.822e+07***
(2.111e+07) (9.169e+06) (9.142e+06) (9.366e+06)
Region 6 −1.624e+07 1.593e+06 3.604e+06 5.097e+06
(1.896e+07) (7.751e+06) (8.059e+06) (8.157e+06)
Region 7 −2.669e+07 −2.165e+07* −2.150e+07* −2.968e+07**
(2.666e+07) (1.186e+07) (1.108e+07) (1.192e+07)
Region 8 −1.100e+07 −3.389e+06 −1.726e+06 4.672e+06
(2.335e+07) (1.197e+07) (1.050e+07) (9.816e+06)
Region 9 ‒ ‒ ‒ ‒

Region 10 −4.006e+06 −3.053e+06 −1.083e+06 −1.369e+06


(2.907e+07) (1.249e+07) (1.252e+07) (1.174e+07)
Region 11 −2.471e+07 3.784e+06 6.301e+06 1.411e+07
(2.242e+07) (8.982e+06) (9.870e+06) (1.124e+07)
Region 12 −1.631e+07 1.729e+06 3.413e+06 3.119e+06
(1.861e+07) (7.826e+06) (8.091e+06) (7.470e+06)
ARMM 1.210e+07 3.286e+07* 3.553e+07* 4.115e+07**
(2.786e+07) (1.876e+07) (1.823e+07) (1.736e+07)
CAR 9.069e+06 2.622e+07*** 2.773e+07*** 3.549e+07***
(1.820e+07) (9.479e+06) (8.370e+06) (8.338e+06)
CARAGA −1.941e+07 −5.377e+06 −3.368e+06 784,707
(2.048e+07) (1.130e+07) (1.130e+07) (1.093e+07)
Constant −7.629e+06 −3.475e+07 −3.059e+07** −9.021e+07*** −9.744e+07*** −1.168e+08***
(2.109e+07) (5.973e+07) (1.203e+07) (2.906e+07) (2.842e+07) (2.955e+07)

Observations 143 143 132 132 132 132


R-squared 0.648 0.665 0.870 0.899 0.900 0.893
Note: Robust standard errors in parentheses.
***Significant at 1%; **significant at 5%; *significant at 10%.

You might also like