# RECORD 91b7dd63d779b9bfc5961da2cf5599ed5160c16d
# TITLE: Stratification and Public Utility Services in Colombia: Subsidies to Households or Distortion of Housing Prices?
# YEAR: 2006.0  VENUE: nan  DOI: 10.32468/be.422
# AUTHORS: C. Medina; L. Morales
# PAGES: 46  CANDIDATE TABLE PAGES (1-based): [19, 29, 32]
# ABSTRACT: Domiciliary public utility services in Colombia have a cross subsidy system which charges subsidized rates to the households who live in houses located in strata associated to low wealth levels, and taxed rates to the better off. We assesses the hypothesis that the flow of subsidies that potentially come from a particular house, are discounted by housing market agents so that most of them are transferred to the prices of the houses that generate the subsidies. By estimating a hedonic prices model applying a regression discontinuity approach, we find that the increment in house value estimated because of subsidies is similar in magnitude to the present value of the flow of subsidies. Likely effects are found on the rent amount. We conclude that subsidies to the poor population through public spending in domiciliary public utility services in Colombia is being achieved, if anything, in a very limited way. Most of the financial effort on this subject ends up distorting housing relative prices according to socioeconomic strata, with an annual cost of up to 0.7% of GDP in supposed gross subsidies to domiciliary public utility services.


===== PAGE 1 (score 1, numeric-paren 0, table-words 0, var-words 1) =====

Stratification and Public Utility Services 
in Colombia:Subsidies to Households 
or Distortion of Housing Prices?
T
here is ample consensus about the desirability of subsidizing the con-
sumption of public utility services because of the positive externalities
that result when they are supplied and the high public costs generated by
their absence. On that basis, most Latin American countries subsidize the
supply of services and constantly try to improve their targeting systems while
minimizing social losses associated with the subsidy schemes.
States in the region that allocate subsidies for residential public utility ser-
vices (servicios públicos domiciliarios, or SPD) have always found the target-
ing and coverage of subsidies a source of controversy, debate, and criticism.1
Among the ways used in the region to reach households with SPD subsidies
are cross subsidy schemes, subsidies to supplying utilities, cash transfers, and
so forth. Because of the variety of alternatives and the socioeconomic and cul-
tural diversity of the countries of the region, no regional consensus exists on
how subsidies should be targeted.
To the problem of lack of technical consensus can be added the difficulty
of reforming the targeting systems derived from the complex political econ-
omy of subsidy policies in the region. That difficulty has become even greater
41
C A R L O S  M E D I N A
L E O N A R D O  M O R A L E S
kOpinions expressed in this paper are those of the authors and do not necessarily reflect the
views of Banco de la República (Colombian Central Bank) or of the members of its board of
directors. We thank Raquel Bernal, Carmen Pagés, and Maximo Torero for detailed comments
on successive drafts, Carlos Esteban Posada and participants in the Economía and Banco de la
República seminars for helpful comments, and Laura Angarita and Lina Marcela Cardona for
assistance.
1. The Colombian case is a clear example. Among recent studies that have assessed Colom-
bia’s current system and formulated proposals to improve it are Fernández (2004), Meléndez
(2004), and INECON (2006). The Colombian government has also taken some steps toward sys-
tem reform with proposals such as one by the Departamento Nacional de Planeación (DNP 2005).


===== PAGE 19 (score 20, numeric-paren 0, table-words 2, var-words 13) =====

subsidies (for electricity, gas, and piped water and sewerage) for only 8,277
households. On the other hand, the DACD information allows us to obtain
the real estate appraisal values for 8,879 households; when merged with the
information for the ECV2003 households, these give a total of 5,759 house-
holds with complete information.
It can be inferred from appendix A-4 that the sample with complete infor-
mation is not a random sample of the households in Bogotá. In particular, it
includes a lower proportion of households in strata 1 and 2 and a higher pro-
portion in strata 3, 4, and 5. It also includes houses with higher real estate
appraisals per square meter, houses with smaller lots and built areas, and a
larger proportion of houses than of other housing units. It includes houses
with more bedrooms and bathrooms and a higher probability of having gas
and telephone service, a garage, and a terrace—in general, better amenities.
Table 5 shows housing prices and utility subsidy amounts by stratum.33
These data reveal the need to control in our empirical exercise for character-
istics on which the socioeconomic strata are based to minimize the possibil-
ity of obtaining biased coefficients.
Carlos Medina and Leonardo Morales
59
33. The price per square meter is defined as the house price divided by the average of square
meters of terrain and square meters of built area.
T A B L E  3 . Households, by Electricity Consumption, Bogotá, 2003
Stratum
Kilowatt-hours
1
2
3
4
5
6
Total
Number of houses by electricity consumption in each stratum
Less than or equal to 200
75,108
381,010
456,701
87,809
26,981
12,000
1,039,609
Greater than 200
21,488
154,036
264,366
95,375
50,327
44,964
630,555
Total
96,597
535,047
721,070
183,188
77,312
56,970
1,670,18
Percentage of houses by electricity consumption in each stratum
Less than or equal to 200
77.75
71.21
63.34
47.93
34.90
21.06
62.25
Greater than 200
22.24
28.79
36.66
52.06
65.10
78.93
37.75
Total
100
100
100
100
100
100
100.00
Percentage of houses by electricity consumption per stratum across all strata
Less than or equal to 200
7.22
36.65
43.93
8.45
2.60
1.15
100
Greater than 200
3.41
24.43
41.93
15.13
7.98
7.13
100
Total
5.78
32.04
43.17
10.97
4.63
3.41
100
Source: ECV2003 (see table 2).


===== PAGE 20 (score 11, numeric-paren 0, table-words 3, var-words 1) =====

60
E C O N O M I A ,  Spring 2007
T A B L E  5 . House Prices per Square Meter and Subsidies in Bogotá, 
by Socioeconomic Stratum, 2003a
Subsidies
Number of 
Housing price per 
Water and 
Stratum
observations
square meter
Housing price
Energy
sewerage
Piped gas
1
222
85,194
20,500,000
14,658
33,031
2,441
2
1,531
129,128
28,000,000
12,044
26,700
2,176
3
2,462
199,558
48,400,000
4,859
13,083
0
4
738
396,539
76,800,000
0
8,372
0
5
202
510,514
113,000,000
−12,968
−22,385
−1,911
6
133
723,551
151,000,000
−15,046
−47,480
−1,675
Total
5,288
225,470
51,200,000
5,539
14,368
602
Sources: Housing price per square meter: DACD; Subsidies: ECV2003 (see table 2).
a. Housing prices and subsidies are in Colombian pesos. Negative figures represent contributions.
T A B L E  4 . Households, by Piped Water Consumption, Bogotá, 2003
Stratum
Cubic meters
1
2
3
4
5
6
Total
Number of houses by piped water consumption in each stratum
Less than or equal to 20
52,955
352,725
515,297
123,501
51,145
27,440
1,123,061
20 less than 40
17,636
99,118
126,268
43,393
18,657
11,749
316,822
Greater than 40
559
5,974
18,126
6,221
2,978
7,034
40,893
Total
71,150
457,817
659,691
173,115
72,780
46,223
1,480,776
Percentage of houses by piped water consumption in each stratum
Less than or equal to 20
74.43
77.04
78.11
71.34
70.27
59.36
75.84
20 less than 40
24.79
21.65
19.14
25.07
25.63
25.42
21.40
Greater than 40
0.79
1.30
2.75
3.59
4.09
15.22
2.76
Total
100
100
100
100
100
100
100
Percentage of houses by piped water consumption per stratum across all strata
Less than or equal to 20
4.72
31.41
45.88
11.00
4.55
2.44
100
20 less than 40
5.57
31.29
39.85
13.70
5.89
3.71
100
Greater than 40
1.37
14.61
44.33
15.21
7.28
17.20
100
Total
4.80
30.92
44.55
11.69
4.91
3.12
100
Source: ECV2003 (see table 2).


===== PAGE 29 (score 10, numeric-paren 0, table-words 3, var-words 2) =====

When the NPV is compared with Δvaluation using a 10 percent annual real
interest rate, we find that both magnitudes are similar, which implies that the
SPD subsidies are transferred almost entirely to housing prices.
Finally, table 10 illustrates not only how the net subsidy becomes in fact a
tax but also how it is distributed by income decile, for both EE and AA. Only
when the EE subsidy is discounted at a 10 percent annual real interest rate is
a positive subsidy for the poorest households found. However, that is the pop-
ulation that is expected to have a higher opportunity cost of money, and so it
also would be expected to be more likely to discount the flow of subsidies at
higher rates. There are important reductions in AA subsidies due to housing
capitalization in the case of AA, in which only the poorest 20 percent of the
population end up receiving a somewhat relevant amount.
In short, the estimates obtained allow us to conclude that SPD subsidies are
almost entirely transferred to the value of the house that receives them, without
generating an apparent net benefit and distorting housing prices.
Carlos Medina and Leonardo Morales
69
T A B L E  9 . Comparison of the Net Present Value of Subsidies with Their Incidence 
on Housing Pricesa
Table of average results for annual discount rate of 10 percent
NPV
Δvaluation
Δvaluation per NPV
Due to change 
Due to change 
Variable
Elasticity
Subsidy
Contribution
in subsidy
in contribution
Subsidy
Contribution
Energy (EE)
0.0297
1,016,810
−1,811,439
1,320,298
4,014,419
1.30
−2.22
Water and 
0.0295
2,147,633
−4,216,391
1,455,403
3,808,256
0.68
−0.90
sewerage (AA)
EE + AA
3,164,444
−6,027,830
2,775,701
7,822,675
0.88
−1.30
Total subsidies
0.0505
3,083,605
−6,139,050
2,497,108
6,520,789
0.81
−1.06
Table of average results for annual discount rate of 15 percent
Energy (EE)
0.0297
692,125
−1,233,015
1,320,298
4,014,419
1.91
−3.26
Water and 
0.0295
1,461,857
−2,870,024
1,455,403
3,808,256
1.00
−1.33
sewerage (AA)
EE + AA
2,153,982
−4,103,039
2,775,701
7,822,675
1.29
−1.91
Total subsidies
0.0505
2,098,956
−4,178,745
2,497,108
6,520,789
1.19
−1.56
Source: Authors’ calculations.
NPV = net present value.
a. Net present values of subsidies and contributions, as well as changes in valuations, are in Colombian pesos of 2003. Results are obtained
with the sample located at an average 250 meters (m) from the border (500m between each house and to the closest one from another stra-
tum on the other side of the boundary), and correcting for measurement error. The change in house valuation, “Δvaluation,” is generated by
a 100 percent change in subsidy.


===== PAGE 30 (score 5, numeric-paren 0, table-words 1, var-words 2) =====

70
E C O N O M I A ,  Spring 2007
T A B L E  1 0 . Distribution of DPS Subsidies Net of Their Incidence on House Value 
in Bogotá, 2003a
Energy (EE)
Discount rate = 0.10
Discount rate = 0.15
Subsidy as 
Net subsidy 
Net subsidy 
Monthly amount 
percentage 
Amount of 
as percentage 
Amount of 
as percentage 
Decile
of subsidies
of income
net subsidy
of income
net subsidy
of income
1
381,543,074
5.5
7,018,174
0.1
−168,676,493
−2.4
2
488,009,656
2.2
16,445,345
0.1
−204,771,828
−0.9
3
545,908,526
1.7
−30,098,973
−0.1
−300,311,872
−0.9
4
617,593,442
1.3
24,199,277
0.1
−254,169,879
−0.6
5
674,633,707
1.0
−6,097,110
−0.01
−325,437,100
−0.5
6
629,887,029
0.7
−58,899,267
−0.1
−382,018,229
−0.4
7
612,863,369
0.5
−155,701,525
−0.1
−516,245,621
−0.4
8
591,676,242
0.3
−280,949,748
−0.2
−690,310,420
−0.4
9
463,921,284
0.2
−286,651,201
−0.1
−638,754,762
−0.3
10
224,639,582
0.1
−231,103,448
−0.1
−444,898,606
−0.1
Total
5,230,675,911
0.43
−1,001,838,476
−0.08
−3,925,594,811
−0.32
Water and sewerage (AA)
1
959,510,155
13.8
543,167,767
7.8
347,856,128
5.0
2
1,171,353,736
5.4
682,522,562
3.1
453,205,078
2.1
3
1,323,456,238
4.1
716,325,415
2.2
431,512,307
1.3
4
1,449,931,661
3.1
795,147,452
1.7
487,979,104
1.1
5
1,598,213,756
2.4
874,442,736
1.3
534,911,601
0.8
6
1,558,287,689
1.8
786,617,779
0.9
424,616,910
0.5
7
1,530,139,810
1.3
610,165,333
0.5
178,592,787
0.2
8
1,494,818,640
0.8
428,601,977
0.2
−71,574,625
−0.04
9
1,356,714,458
0.5
237,980,898
0.1
−286,832,010
−0.1
10
874,734,539
0.2
−189,031,175
−0.05
−688,057,709
−0.2
Total
13,317,160,684
1.09
5,485,940,744
0.45
1,812,209,571
0.15
Source: Authors’ calculations.
a. Results are obtained with the sample located at an average 250 meters (m) from the border (500m between each house and to the
closest one from another stratum on the other side of the boundary), and correcting for measurement error.
b. Amounts of subsidies are expressed in Colombian pesos.
Results for Rents
The ECV2003 asks households that pay rent to report their monthly payment.
In addition, it asks those who live in their own house for the amount of rent
that they believe the house would generate if it were rented. Using as the
dependent variable the logarithm of the rents reported in both instances, we
repeat the previous exercise. The results show a positive relation between EE
and AA subsidies and the logarithm of the rent paid by households.


===== PAGE 32 (score 27, numeric-paren 0, table-words 5, var-words 3) =====

T A B L E  1 1 . House Price Model, with Additional Controlsa
1,000m
800m
700m
600m
500m
Variable
Coefficient
t
Coefficient
t
Coefficient
t
Coefficient
t
Coefficient
t
Subsidy of energy
2.41E-06
1.56
3.02E-06
1.77
3.61E-06
1.95
3.78E-06
1.94
4.01E-06
1.86
Subsidy of piped water and sewerage
2.68E-06
2.83
2.92E-06
2.84
2.79E-06
2.55
2.80E-06
1.60
2.89E-06
1.63
Subsidy of piped gas
5.58E-07
0.08
−5.82E-06
−0.77
−4.51E-06
−0.56
−6.73E-06
−0.77
−7.36E-06
−0.75
Subsidy of energy squared
7.12E-11
1.69
7.20E-11
1.56
3.40E-11
0.71
2.14E-11
0.35
1.67E-12
0.02
Subsidy of piped water and 
−1.47E-11
−0.97
−2.95E-11
−1.70
−1.76E-11
−0.98
−1.06E-11
−0.32
−4.77E-12
−0.14
sewerage squared
Subsidy of piped gas squared
4.13E-10
0.45
−1.12E-10
−0.11
−4.37E-10
−0.45
−5.23E-10
−0.40
−1.53E-10
−0.11
Ln(τ)
−0.024
−1.79
−0.038
−2.74
−0.038
−2.68
−0.030
−2.07
−0.036
−2.33
Subsidy of energy
2.33E-06
1.50
2.99E-06
1.75
3.46E-06
1.88
3.59E-06
1.84
3.72E-06
1.72
Subsidy of piped water and sewerage
2.69E-06
2.83
2.90E-06
2.81
2.76E-06
2.51
2.80E-06
1.59
2.89E-06
1.61
Subsidy of piped gas
3.52E-07
0.05
−5.78E-06
−0.76
−4.26E-06
−0.53
−6.45E-06
−0.74
−7.36E-06
−0.76
Subsidy of energy squared
6.96E-11
1.65
7.14E-11
1.55
3.15E-11
0.66
1.86E-11
0.30
−2.72E-12
−0.04
Subsidy of piped water and 
−1.47E-11
−0.97
−2.99E-11
−1.72
−1.85E-11
−1.04
−1.26E-11
−0.38
−7.12E-12
−0.21
sewerage squared
Subsidy of piped gas squared
4.09E-10
0.44
−9.03E-11
−0.09
−3.83E-10
−0.40
−4.65E-10
−0.35
−1.09E-10
−0.07
Ln(τ)
−0.025
−1.80
−0.0381
−2.7
−0.0375
−2.68
−0.030
−2.10
−0.037
−2.38
SR
−0.037
−1.47
−0.032
−1.20
−0.049
−1.68
−0.042
−1.62
−0.051
−1.75
Source: Authors’ calculations.
m = meters; SR = subsidized regime (see text).
a. The dependent variable is the logarithm of the house appraisal. Results are the equivalent to those reported in table 8 once Ln(τ) (the effective property tax rate) and SR are included. T statistics are estimated on
the basis of robust standard errors.


===== PAGE 33 (score 0, numeric-paren 0, table-words 0, var-words 0) =====

In addition, the inclusion of the SR has negligible effects on the relevant 
coefficients.45
Other Potential Biases
Although we think that our results make a good case for the capitalization of
public utility subsidies on house prices, our methodology is not free of caveats.
First, there are several issues common to hedonic regressions that might be
generating bias in our estimates, such as the presence of substantial heterogene-
ity across households, spatial correlation, and so forth. Second, although the
evidence provided supports to a large extent the validity of the RD assumptions
listed earlier (namely subsidies change discontinuously at the boundaries and
observable and unobservable characteristics of houses change continuously at
the boundaries), in the case of the second assumption, whether unobservable as
well as observable characteristics change smoothly around the boundaries
would still need to be proven. In addition, the third and fourth assumptions (the
effect of public utility subsidies on house prices is continuous at the boundaries
and the amount of subsidies is independent of its effect on house prices at the
boundaries) imply more demanding requirements in our case.
The second assumption would not be satisfied if, for example, differences
in preferences across races led individuals from different races to segregate
themselves. Since people would value the network that each neighborhood
offers differently, that could be an example of an unobservable characteris-
tic that we would not be controlling for that could change discontinuously
around the boundaries. In general, since people can estimate in advance the
benefits that they could receive from locating on either side of a boundary, one
could argue that some sort of sorting around boundaries on unobservable char-
acteristics, such as households’ preferences, should be taking place in prac-
tice. As long as that sorting becomes a characteristic that affects house prices
in a discontinuous way around boundaries, the mechanics of the sorting would
imply a violation of the fourth assumption, which in this context would come
along with a violation of the second assumption and, in some cases, the third.
45. Intuitively, however, one could expect the inclusion of the SR to have a positive
effect on house prices in the future, as the recently implemented “New Sisbén” score is con-
structed so that households in lower strata have lower scores and so are more likely to be eli-
gible to benefit from SR, thus increasing the probability that a household will be a beneficiary
if it moves just across the border between strata (in particular when the household moves
from strata 1 or 2 to stratum 3 or higher or vice versa) and affecting house prices much as
public utility subsidies do.
Carlos Medina and Leonardo Morales
73