Probabilistic Seismic Risk Assessment in Manizales, Colombia: Quantifying Losses for Insurance Purposes

Int J Disaster Risk Sci (2017) 8:296–307 www.ijdrs.com DOI 10.1007/s13753-017-0137-6 www.springer.com/13753

ARTICLE

Mario A. Salgado-Gálvez1 · Gabriel A. Bernal2 · Daniela Zuloaga2 · Mabel C. Marulanda1 · Omar-Darío Cardona3 · Sebastián Henao4

Published online: 6 September 2017 © The Author(s) 2017. This article is an open access publication

Abstract A fully probabilistic seismic risk assessment was In the case of the buildings, the results were used in the developed in Manizales, Colombia, considering assets of update of the technical premium values of the existing different types. The first type includes elements that are collective insurance scheme. part of the water and sewage network, and the second type includes public and private buildings. This assessment Keywords CAPRA · Earthquake insurance · Manizales required the development of a probabilistic seismic hazard (Colombia) · Probabilistic seismic risk assessment · Risk analysis that accounts for the dynamic soil response, retention and transfer assembling high resolution exposure databases, and the development of damage models for different types of elements. The economic appraisal of the exposed assets was 1 Introduction developed together with specialists of the water utilities company of Manizales and the city administration. The risk Manizales is a city of about 400,000 inhabitants located in assessment was performed using several Comprehensive the Colombian Andes. Similarly to other cities in the Approach to Probabilistic Risk Assessment modules as region, Manizales has a building portfolio that is mainly well as the R-System, obtaining results in terms of tradi- comprised of low and middle-rise masonry and reinforced tional metrics such as loss exceedance curve, average concrete residential, commercial, and administrative units. annual loss, and probable maximum loss. For the case of As a consequence of being located in a region prone to pipelines, repair rates were also estimated. The results for hazards with different origins (for example, landslides, the water and sewage network were used in activities floods, and volcanic ash fall are common in the city, and related to the expansion and maintenance strategies, as well the Colombian Andes are prone to earthquakes), a com- as for the exploration of financial retention and transfer prehensive set of disaster risk management and reduction alternatives using insurance schemes based on technical, activities have been continuously developed during the past probabilistic, and prospective damage and loss estimations. 10 years. These actions have had remarkable achievements in terms of corrective and prospective interventions, which range from the structural retrofitting of hospitals to the & Mario A. Salgado-Ga´lvez design and implementation of a voluntary and collective [email protected] risk transfer scheme using innovative earthquake insurance 1 Centre Internacional de Metodes Numerics en Enginyeria approaches (Marulanda 2013; Marulanda et al. (CIMNE), 08034 Barcelona, Spain 2014, 2016). 2 INGENIAR Ltda., Bogota´, D.C. 110221, Colombia As part of these disaster risk management and reduction activities that have taken place in the city, a multidisci- 3 Instituto de Estudios Ambientales (IDEA), Universidad Nacional de Colombia Sede Manizales, Manizales 170004, plinary and multi-sectorial disaster risk management and Colombia reduction project was developed for the city.1 Among its 4 Universidad Nacional de Colombia Sede Manizales, Manizales 170004, Colombia 1 http://www.gestiondelriesgomanizales.com. 123 Int J Disaster Risk Sci 297 different activities, a fully probabilistic seismic risk results, instead of using solely the opinion of experts. The assessment was developed for all elements of the water and results constitute a starting point not only for future updates sewage network owned and operated by Waters of of the assessment (accounting for changes in the exposure Manizales S.A. E.S.P. (Aguas de Manizales S.A. E.S.P.), characteristics, for example, larger number of buildings, and also for the portfolio of public and private buildings water and sewage network elements with different char- within the city limits. The aim of the assessment was to acteristics, and better understanding of the seismic hazard estimate the order of magnitude of potential damages and in the area under analysis), but for a later integration of losses due to earthquakes, and use these results in the damages and losses associated to other types of hazards design of an earthquake insurance policy for the water and that can be aggregated using approaches like the one pro- sewage network, as well as in the update of the existing posed by Ordaz (2014). voluntary and collective insurance scheme of the buildings This case study provides an example of how scientific, in the city. This is the first time a fully probabilistic seismic engineering, and financial activities can be combined in risk assessment has been performed for water and sewage order to develop comprehensive disaster risk management networks in Colombia, although similar assessments have activities converting the results into a technical basis for been previously performed in other parts of the world the design and implementation of earthquake insurance (Taleb-Agha 1977; Eguchi 1991; Isoyama et al. 1998; schemes in the city of Manizales. Ballantyne et al. 1999; Pitilakis et al. 2006). With respect to the building portfolio, this assessment constitutes an update, bearing in mind that similar studies have been 2 Methodology and Input Data developed using the same methodological framework in other urban centers such as Bogota´ D.C., Medellıń, Bar- This assessment made use of the recently updated local celona, and Lorca (Marulanda et al. 2013; Salgado-Ga´lvez seismic hazard model at bedrock level considering the local et al. 2013, 2014a, b, 2016c). site effects (Bernal et al. 2015, 2017; Salgado-Ga´lvez et al. Because a number of uncertainties related to seismic 2016b). Additionally, detailed and high-resolution expo- hazard and physical vulnerability exist, it has been widely sure databases were developed for the water and sewage acknowledged that probabilistic approaches and method- and building portfolios—in the case of buildings, individ- ological frameworks are required when assessing, in a ual structural characteristics (that is, age, number of stories, prospective way, feasible future losses (McGuire 2004; and construction material) were recorded for each entry, Grossi and Kunreuther 2005; UNISDR 2013; GFDRR whereas in the case of the water and sewage network, 2014). These probabilistic approaches are intended to not characteristics for each segment of the pipelines (that is, only identify the uncertainties but, when possible, quantify material, diameter, and joint type, among others) were and rigorously propagate them throughout the whole assigned. In addition to buried pipelines, other elements analysis process. Recently, different open-source tools and such as water storage tanks and administrative buildings initiatives to perform these evaluations have been devel- were also considered; for these, parameters such as age and oped, with the primary objective of providing access to construction material were also taken into account. After these tools to developing countries. A widely used and identifying the structural characteristics of each element recognized tool is CAPRA—Comprehensive Approach to included in both exposure databases, a set of vulnerability Probabilistic Risk Assessment (Cardona et al. 2012). functions was developed in terms of different seismic For the building stock, the probabilistic risk assessment hazard intensity measures (for example, spectral accelera- was performed with the R System (ERN 2014), a propri- tion, and peak ground velocity), to be assigned later to each etary tool that implements the fully probabilistic and event- element. Then the convolution between hazard and vul- based loss assessment methodology proposed by Ordaz nerability was performed using the CAPRA-GIS program (2000). The results from this assessment were used for the (ERN-AL 2011) to obtain the loss exceedance curve update of the technical premium values used for the (LEC), from which other probabilistic risk metrics such as insurance of the complete portfolio of public buildings of average annual loss (AAL) and probable maximum loss Manizales (using one collective policy), as well as for the (PML) can be obtained. For the buried pipelines, average premium values of the voluntary and collective insurance annual repair rates (by km) were also estimated. scheme for the private buildings of the city. In the case of The development of a fully probabilistic seismic risk the water and sewage network, the results were used to analysis usually requires the completion of a set of activ- identify appropriate materials to be used in the expansion ities that provide input data. Among them, the following and maintenance activities of the water and sewage com- can be identified: (1) a probabilistic seismic hazard analysis pany, as well as to serve as a base for the negotiation of an in which events with different characteristics such as earthquake insurance policy based in sound technical location, magnitude, and occurrence frequency are 123 298 Salgado-Ga´lvez et al. Seismic Risk in Manizales: Quantifying Losses for Insurance Purposes modelled, together with the geographical distribution of the spectral ordinates ranging between 0.0 and 4.0 s, and for hazard intensities and the dynamic soil response, which is ground accelerations, velocities, and displacements. Since of high interest usually where there is presence of soft soil a Poissonian process is assumed to occur for the loss layers; (2) the assembly of exposure databases making use generation process, the stochastic event set needs to be of several data sources, with the objective of identifying mutually exclusive and collectively exhaustive; individual and characterizing all relevant elements on which feasible events need to have an associated occurrence frequency damages and losses are of interest; (3) the development of and data of at least the first two statistical moments for vulnerability models that relate different seismic hazard each hazard intensity measure. intensities to the expected losses, where a unique vulner- Due to the characteristics of the soils in Manizales, and ability function is assigned to each element included in the also because of the local nature of the analysis, as well as exposure database, associated to the hazard intensity that its high level of resolution, it is relevant to take into better correlates with damage (that is, spectral acceleration account the dynamic soil response. This was done by using for flexible buildings, peak ground velocity for buried spectral transfer functions based on the results of the pipelines); and (4) the assessment of damages and losses by recently updated and harmonized seismic microzonation convoluting hazard and vulnerability, thus obtaining developed for the city (also developed within the frame- exceedance rates for different loss values (in monetary work of the disaster risk management and reduction pro- terms), from where other relevant risk metrics such as the ject, see Bernal et al. (2015, 2017)) that provides results at AAL and the PML can be obtained (Ordaz 2000; Cardona 377 different locations. et al. 2008a, b). This section summarizes the most relevant aspects of each of these four steps for the water and sewage 2.2 Assembly of the Exposure Databases network of Manizales, as well as for its portfolio of public and private buildings. The development of exposure databases has always con- stituted a great challenge in modeling disaster risk because, 2.1 Probabilistic Seismic Hazard Analysis in most cases, in order to complete the identification and characterization process, data from several sources need to The objective of a probabilistic seismic hazard analysis be combined, leading to a process where assumptions and (PSHA) is to provide a long-term overview of the rela- simplifications are usually required. However, the exposure tionships between strong ground motion levels and their databases developed for these case studies can be consid- occurrence frequencies. For a fully probabilistic seismic ered an exception. Given the previous continuous efforts of risk assessment, a hazard representation by means of the water and sewage company, as well as of the city stochastic events (event-based approach) is required. (cadastral office), in gathering, arranging, classifying, and A PSHA allows the user to obtain feasible earthquakes updating the information related to the exposed assets, all with different locations, magnitudes, and occurrence fre- relevant parameters for a seismic risk assessment (ranging quencies, associated to all the seismic sources of interest. from age to structural system, construction material, loca- For this case, the PSHA was developed using the latest tion, and geometry) could be obtained directly from the version of the program CRISIS, a well-known and accepted official sources (only specific field visits to validate and tool that implements state-of-the-art methodologies to review very specific issues were needed). The economic perform PSHA (Ordaz et al. 2013). This program is also appraisal of each element included in the exposure data- the seismic hazard module of the CAPRA platform and an bases was done using official data in a task where close input data provider for the R-System, used for the case of collaboration was received from specialists working in the buildings. water utilities company of Manizales and the cadastral The PSHA was developed at local level considering office of the city. more than 35 seismic sources for which, in all cases, a The exposure databases for the water and sewage net- geometric area model was used for their representation work, as well as for the public and private buildings, accounting for variations in dip angles and depths (Sal- required the development of specific tasks such as the gado-Ga´lvez et al. 2013, 2016b). Regionally developed and identification of the relevant structural parameters that are calibrated ground motion prediction equations (GMPE) related to the expected behavior of each asset when sub- were assigned to each seismic source (Bernal 2014), thus jected to earthquake hazard intensities. Because of the allowing the consideration of different hazard intensities evident differences between the type of components, and such as spectral accelerations, peak ground velocity (PGV), since the use of the results was not directly linked or and peak ground displacement (PGD). As a result of the combined, these activities were performed in a separate PSHA a total of 1275 stochastic events were obtained, each manner. of them with a set of hazard intensities associated to several 123 Int J Disaster Risk Sci 299

The economic appraisal of the whole water and sewage 2.3 Vulnerability Functions network was developed in terms of the replacement cost, that is the amount of money required to repair or rebuild The vulnerability function approach accounts for a con- the damaged element and take it back to the exact same tinuous, quantitative, and probabilistic representation of the conditions it was in before it was damaged; in this case, the physical damage of the assets for different levels of hazard value for each element does not simply correspond to its intensities. For each asset included in the exposure data- monetary cost (for example, 5 m of concrete pipe), but also bases, a unique vulnerability function is assigned in order to other costs associated to activities of installation, such as to estimate the losses associated with different hazard breaking and repairing streets at different depths. All val- intensity levels, for each of the considered stochastic ues were obtained directly from the water utilities company events. Due to the great variety of types of elements, vul- and those data can be said to be complete, reliable, and nerability functions in terms of several hazard intensities updated. Overall, pipelines for primary and secondary were developed, given that different intensities show good water and sewage networks account for most of the correlation with the damage of different types of elements. exposed value with around 80%, followed by the water For example, vulnerability functions were developed in storage tanks that account for approximately 7% of the terms of peak ground acceleration (PGA) for the case of total exposed value of the water and sewage network in buried water storage tanks, in terms of spectral accelera- Manizales. tions for administrative buildings and elevated water stor- For the case of the public and private building portfolio, age tanks, and in terms of PGV for buried pipelines. For the the base information to assemble the exposure database latter, the recommendations provided by the American was delivered by the Secretary of Finance and the Risk Lifeline Alliance—ALA (2001) were followed in order to Management Unit (Unidad de Gestión del Riesgo) of the obtain the repair rates, RR (the number of damages to be city administration, and was complemented with additional repaired by km). parameters from aerial images, existing maps, and field For other types of components such as administrative visits. The first exposure database for private buildings in buildings and water storage tanks, the development of the Manizales, developed for purposes related to seismic risk vulnerability models included the consideration and com- assessment in 2004, included representative variables such bination of different approaches, such as analytical and as age, number of stories, structural system, location, and empirical analyses combined with expert criteria. For all socioeconomic level. The latter is specified in terms of types of components, loss is considered to be a random strata, from 1 to 6, in which strata 1 and 2 group inhabitants variable for which the variation of the first two statistical with the lowest economic capacity and are therefore moments for different seismic demand levels are described granted a property tax exemption, whereas strata 3–6 group by means of vulnerability functions (Miranda 1999; Ordaz property tax payers. At the time the project was conducted, 2000). Figure 2 shows the vulnerability functions (expected more than 85,816 dwellings were included and, for the values of the loss) used for the different water storage tank purpose of the analysis, they were split into three different classifications (left), and some of the vulnerability func- categories as explained in detail in Marulanda et al. (2014, tions used for the building portfolio (right). 2016). One of the tasks within the disaster risk management 2.4 Loss Assessment project developed in Manizales was related to the update and completion of the exposure database to take into Once the input data for hazard, exposure, and vulnerability account not only improvements related to the content and was ready, the loss estimation was made following the fully attributes of the exposure database, but also the consider- probabilistic and event-based procedure proposed by Ordaz ation of new buildings that have been built since 2004. For (2000). This procedure has been implemented in the the purpose of this analysis more than 113,064 dwellings CAPRA-GIS program, which corresponds to the risk were considered. Figure 1 shows the geographical distri- module of the CAPRA platform, in the case of the water bution of the identified structural systems in the city. Non- and sewage network, and the R System (ERN 2014), in the engineered materials and structural systems such as adobe case of the portfolio of buildings. The main outcome of a and other earthen structures (bahareque) can be easily state-of-the-art probabilistic risk assessment corresponds to identified in the oldest and historical areas of the city the LEC, which contains several loss values together with (Cumanday, San Jose´, and La Macarena). Further details their associated exceedance frequencies. The same about the assembly of the buildings exposure database of methodological approach and procedure was followed to Manizales can be found in Gonzalez (2014). estimate seismic losses in the two different exposure databases and, therefore, results in terms of the same risk metrics have been obtained. Under this approach, the LEC 123 300 Salgado-Ga´lvez et al. Seismic Risk in Manizales: Quantifying Losses for Insurance Purposes

Fig. 1 Geographical distribution of structural systems for the public and private buildings in Manizales, Colombia

100% 100%

90% 90%

80% 80%

70% 70%

60% 60%

50% 50%

40% 40%

Mean Damage Ratio 30% Mean Damage Mean Damage Ratio 30%

20% 20%

10% 10%

0% 0% 0 200 400 600 800 1,000 0.00% 0.50% 1.00% 1.50% 2.00% Spectral acceleration (gal) Maximum Interstory Drift Elevated At surface (small) At surface (big) Confined masonry RC Frames + masonry walls RC Frames RC dual system Semi-buried (small) Semi-buried (big) Semi-buried (open) RC bearing walls

Fig. 2 Examples of some vulnerability functions for different types of water storage tanks (left) and for some of the different types of buildings (right) in Manizales, Colombia is calculated using 50 different loss values ranging between additionally, loss exceedance probabilities for different 0 and 80% of the total exposed value (logarithmically exposure timeframes can be computed. Equation 1 shows spaced) using the following expression: that an event-based approach is required for the hazard XN representation because the associated occurrence frequen- mðlÞ¼ PrðL [ lj EventiÞ Á FAðEventiÞð1Þ cies are required along with the expected hazard intensities. i¼1 More details on the loss assessment methodology can be ´ where ν(l) is the loss exceedance rate (1/year), Pr(L [ l| found in Ordaz (2000) and Salgado-Galvez et al. (2014b). For the case of the water and sewage network, additional Eventi) is the loss probability exceedance conditional to the analyses were considered for particular feasible earth- occurrence of the ith event, and FA(Eventi) is the annual occurrence frequency of the ith event. Once the LEC is quakes in which the approach from the temporal perspec- calculated, other risk metrics such as the AAL and the tive can be considered as deterministic, but which is still PML (for any return period of interest) can be obtained; probabilistic from the loss estimation perspective (since the

123 Int J Disaster Risk Sci 301 uncertainties associated with the strong ground motion relative AAL corresponds to the water storage tanks with intensities and the vulnerability representation are still around 6.6‰ (a considerably high value bearing in mind accounted for). In that case, the expected loss for the par- that the AAL of the whole network is 0.65‰). From these ticular scenario is estimated using Eq. 1 where the occur- results, it is also worth noting that, although pipelines rence frequency of the considered event is assumed to be concentrate the vast majority of the exposed value, their equal to 1.0. AAL is low when compared, in relative terms, to any other component. Furthermore, it is important to see that the administrative buildings have an AAL that by no means 3 Results can be considered as negligible with the aggravating fact that those buildings are to be considered as critical com- The results of the fully probabilistic seismic risk assess- ponents. This is because aside from the physical losses, any ment of the two portfolios for Manizales (the water and important damage on the buildings can gravely affect the sewage network and the public and private buildings) are operation of the network in the aftermath of an earthquake, presented in this section. For clarity, they have been split not only from a purely operational point of view but also by category. Due to confidentiality constraints, the results from the emergency management perspective (Salgado- (in terms of absolute exposed values and losses) for the Ga´lvez 2014). For the case of the buried pipelines it is also water and sewage network are not presented here, although possible to develop risk maps in terms of average annual the relevant share by component is included. repair rates (by km) such as the one shown in Fig. 3 for the sewage network of Manizales. More details about this 3.1 Water and Sewage Portfolio study can be found in Salgado-Ga´lvez et al. (2017). In addition to the fully probabilistic seismic risk Table 1 shows the risk results for the components of the assessment, an event capable of causing losses with a water utilities company in terms of the relative AAL. From return period of 1000 years occurring in the vicinity of the Table 1 it is evident that the component with the highest city was considered (event with magnitude 7.3, with an epicenter located 25 km southwest of Manizales). In this case, instead of a loss exceedance rate, what is directly Table 1 Exposed value and average annual loss (AAL) by component for the Aguas de Manizales network (a), and probable maximum obtained is the mean damage ratio (MDR) of each com- loss (PML) for different return periods of the Aguas de Manizales ponent of the network, which is approximately 3.5% of the network (b) in Manizales, Colombia total exposed value. As in the fully probabilistic assess- Component Relative share (%) AAL ‰ ment, risk is mostly concentrated in the water storage tanks and the administrative buildings. Repair rates for pipelines (a) with different materials and joint characteristics were Administrative buildings 2.8 1.61 compared with published post-earthquake survey data of Adduction network 4.2 0.18 water and sewage systems (Ballantyne et al. 1991, 2002), Water catching 1.2 0.29 finding congruent results in terms of RR. Several of the Inspection chambers 2.9 0.01 tools and programs used for this study have been previ- Conduction pipelines 13.7 0.09 ously tested and validated with particular damage results Distribution pipelines 30.6 0.04 for earthquakes in Lorca, Spain (Salgado-Ga´lvez et al. Sewage pipelines 35.3 0.17 2016a). Tanks 7.3 6.57 Control valves 0.7 0.92 3.2 Building Portfolio System valves 1.2 1.28 Total 100.0 0.64 For the building portfolio, three separate seismic risk Probable maximum loss assessments were performed. The first included only the buildings that are exempt from property taxes (socioeco- Return period (years) Relative loss (%) nomic strata 1 and 2), the second included only the (b) buildings subjected to property tax payments (socioeco- 250 2.1 nomic strata 3–6), and the third included all 113,064 500 2.8 dwellings. 1000 3.5 Figure 4 shows the geographical distribution of the 1500 4.0 relative (to the exposed value) AAL for the public and private buildings of Manizales. The highest values, around 2.25%, correspond to dwellings with non-engineered 123 302 Salgado-Ga´lvez et al. Seismic Risk in Manizales: Quantifying Losses for Insurance Purposes

Fig. 3 Average annual repair rates (RR/km) for the sewage network of Aguas de Manizales in Manizales, Colombia

Fig. 4 Average annual loss (AAL) (relative) distribution for the public and private buildings in Manizales, Colombia

building classes; this value of 2.25% can be understood as deductible levels, in all cases using the methodological losing the total exposed value of each asset on average approach to account for limits and deductibles proposed by every 40 years (Marulanda 2014). Ordaz (2000). The higher the deductible level in a tradi- Table 2 shows the summary of the risk metrics (AAL tional earthquake insurance scheme, the lower the expected and PML) for the analyzed portfolios considering different

123 Int J Disaster Risk Sci 303

Table 2 Seismic risk results, average annual loss (AAL), and probable maximum loss (PML), for public and private buildings with 0 and 3% deductible levels in Manizales, Colombia

Type of building Private Public Number of buildings 113,064 333 Deductible 0% 3% 0% 3% AAL (‰) 2.86 1.63 2.91 1.62 Return period Expected loss (%) Expected loss (%) Expected loss (%) Expected loss (%) 200 years 5.99 4.28 7.43 5.67 PML 500 years 8.31 6.24 10.53 8.61 1000 years 11.15 8.61 13.62 11.46 1500 years 13.55 10.72 15.85 13.46

losses in terms of either AAL or PML, for any return the construction, maintenance, expansion, and update of period. pipelines. The value of the pure premium, in 2014, for private Regarding the water and sewage network, the results properties was set equal to 2.86‰ without deductible, and were useful for the design of a structural retrofitting strat- to 1.63‰ after applying a 3% deductible. Despite the egy of critical components based on the overall risk results. significant increase in the number of existing properties in Among the components, the water storage tanks were pri- the city in a period shorter than 10 years (from 85,816 to oritized not only due to the high relevance they have in the 113,064), the decrease of the pure risk premium value in operational capacity of the network, but also due to the the city (with 3% deductible) from 1.98 to 1.63‰ is par- high-risk concentration they present, as shown in Table 1. ticularly remarkable. It is rarely possible to measure the For that reason, a probabilistic benefit–cost analysis was seismic risk of a city in different time periods with the developed to explore different structural retrofitting alter- same metrics, and even more so to reveal a reduction of natives, with the selection of the most effective ones from risk which, in the case of Manizales, could be expected due the behavioral and economic perspectives. Regarding the to the efforts made in risk management. In recent years a pipelines, after the disaggregation of the risk results by significant number of buildings have been retrofitted in material and joint type, the difference was evident, Manizales, particularly public sector buildings. This can allowing us to recommend that for expansion and explain the new pure risk premium value of 1.62‰ (with replacement activities (the latter closely related to the 3% of deductible) for these properties. This value is continuous maintenance actions developed by the com- practically the same if compared to the value estimated for pany), flexible joints and materials are to be used. This is in the private buildings. However, the main reason why the order to minimize the expected damages, and therefore, to seismic risk of the city has been decreasing relates to the guarantee a better operational condition of the system after significant increase in construction of new buildings that the occurrence of an earthquake in the city or its effectively take into account all the earthquake resistance surroundings. requirements established in the Colombian earthquake The overall results were considered as useful input data resistant building code—NSR-10 (Marulanda et al. 2016). for the design of a robust and technically based earthquake insurance scheme, while at the same time providing the company with sound information to be reliably used in the 4 Using the Seismic Risk Results in Earthquake negotiation process. Regarding financial protection Insurance Design and Implementation schemes, obtaining future feasible losses in terms of AAL, PML (for any return period of interest) and loss exceedance The seismic risk results obtained from the assessment probabilities for different exposure timeframes (Fig. 5) outlined above have been used in different activities that allows for the exploration of the design, use, and estab- include: the design of cross (subsidized) collective earth- lishment of policy conditions for alternative risk transfer quake insurance schemes, the development of hybrid loss instruments (such as captive insurance companies and/or exceedance curves (Vela´squez et al. 2014), the feasibility catastrophe bonds), an activity that is expected to be assessment of the use of alternative risk transfer instru- completed in a second phase of the comprehensive disaster ments such as captive or cat-bonds for the insurance of the risk management project in Manizales. water and sewage network components, as well as the The results also led to the proposal of the design of a identification of the best materials and characteristics for collective voluntary insurance to cover, in some way, the

123 304 Salgado-Ga´lvez et al. Seismic Risk in Manizales: Quantifying Losses for Insurance Purposes

Fig. 5 Loss exceedance 100% probabilities for different exposure timeframes for the 90% 50 Years components of the water and sewage network in Manizales, 80% 100 Years Colombia 70% 200 Years

60%

50%

40%

Exceedance probability 30%

20%

10%

0% 0% 2% 4% 6% 8% 10% Relative loss losses of the poorest city residents. After analyzing the socioeconomic layer, whose assessed valuation is less than portfolios, it was possible to reveal that the cadastral value or equal to 39 minimum current legal salaries (USD 8284), of the low-income properties was not signiﬁcant when which correspond to 20% of the number of properties and compared to the rest of the value of the private buildings of 4% of the insurable value of the total properties of the city. the city. This opened the possibility to consider a subsidy Before this agreement was put into action, only properties for the low-income socioeconomic layers, where the tax- belonging to socioeconomic layers 1 and 2 were exempt. payers who decide to subscribe to the insurance program Table 3 shows the results of this new situation. assume the low socioeconomic layers’ pure premiums In order to determine the value of the insurance pre- (Marulanda et al. 2008, 2014, 2016; Cardona 2009;Car- mium of the non-exempt tax properties, including the dona and Marulanda 2010; Marulanda 2013). exempt (or exonerated) properties, with 3% deductible, it is According to Agreement 760 of 2011, from 1 January necessary to add approximately USD 0.14 million to USD 2012, of the City Council of Manizales, all urban properties 4.41 million of the non-exempts (not exonerated), and belonging to the socioeconomic layers 1, 2, and 3 will be divide this by the insurable value thereof. This value of the exempted from paying property taxes; the same applies for premium corresponds to 1.69‰, and this would be the all rural residential properties, irrespective of the blanket premium in case that all non-exempts (not

Table 3 Risk results for private buildings (tax exempt and non-exempt), with 0 and 3% deductible (Agreement 760, 2011)a in Manizales, Colombia

Type of building Tax exempt properties Non-exempt tax properties Number of buildings 22,501 (19.90%) 90,563 (80.10%)

Insurable value (USD Million) 107.52 (3.85%) 2,682 (96.15%)

Deductible 0% 3% 0% 3% Average Annual Loss (USD M) USD M ‰ cadastral USD M ‰ cadastral USD M ‰ value USD M ‰ value Pure Premium (‰) 0.24 2.19 0.14 1.28 7.74 2.88 4.41 1.64 Return periods USD M ‰ cadastral USD M ‰ cadastral USD M ‰ value USD M ‰ value 200 years 5.20 4.84 3.46 3.21 162.15 6.05 115.58 4.33 PML 500 years 7.25 6.75 4.98 4.63 224.69 8.38 168.43 6.31 1000 years 9.69 9.02 6.85 6.37 301.47 11.24 232.34 8.7 1500 years 11.37 10.58 8.26 7.68 366.65 13.67 289.24 10.83 a USD = 2900 Colombian pesos (COP). Exchange rate as per 4 January 2017

123 Int J Disaster Risk Sci 305 exonerated) would participate in the insurance program. mention that the possibility of covering the poorest With these results the city made a call for proposals to the socioeconomic strata of the population and promoting, in insurance companies to implement the new collective general, the insurance culture in the city of Manizales have voluntary insurance program to cover the low-income been targets of special interest of the municipal adminis- house owners in the city from 2016. tration. This initiative has primarily sought a social benefit and has been promoted by the local government. Support from the national government for the replication in other 5 Conclusion parts of the country can also be possible. The benefit–cost ratio is clear from the point of view of sustainability, The practical use of the results of probabilistic seismic prevention, welfare, and socioeconomic and financial pro- hazard and risk models has been presented in this com- tection. It is based on technical studies made with robust prehensive case study developed for the water and sewage risk models, but the most important elements of this ini- network and the portfolio of buildings of Manizales, tiative have been the political will, governance, citizen Colombia. The case study shows how these models provide solidarity, and risk perception of society and government valuable information that can be incorporated in a direct and officials. This innovative instrument represents a successful explicit way in decision-making processes and how these experience and is a good practice promoted between local decisions can cover a large variety of aspects that range government and the private sector that could be replicated from the design of traditional insurance schemes, to the in other developing countries prone to disasters, if appro- design of innovative cross (subsidized) insurance schemes priate risk studies for implementation are made. (Gurenko et al. 2006; GFDRR et al. 2015), and to the use of specific materials for the replacement, update, and expan- Acknowledgements This work was carried out within the framework sion plans of the water and sewage company, together with of the inter-institutional agreement between the Universidad Nacional de Colombia, Manizales headquarters, and the Corporacioń Autoń- the update and harmonization of contingency plans. oma Regional de Caldas (Corpocaldas)—the regional environmental The water and sewage network of Manizales is the first authority—to improve information systems for risk knowledge and in Colombia that has been subjected to a seismic risk understanding; monitoring of hazard events and warning systems; assessment, not only at a high-resolution detail level, but mainstreaming risk into planning; and to improve risk awareness in the city of Manizales, Colombia. The authors also express their also with a fully probabilistic approach. This case also gratitude to the three anonymous reviewers who contributed to serves to show that having updated, complete, and reliable improving the original version of the manuscript. information regarding exposure yields a set of reliable results that are useful for activities that are not directly Open Access This article is distributed under the terms of the Crea- tive Commons Attribution 4.0 International License (http://creative related to the daily operation of the system. The results commons.org/licenses/by/4.0/), which permits unrestricted use, obtained for the expected losses and after the analysis for distribution, and reproduction in any medium, provided you give single and specific scenarios are consistent (and within the appropriate credit to the original author(s) and the source, provide a link order of magnitude) with the findings by other authors in to the Creative Commons license, and indicate if changes were made. the field (Ballantyne et al. 1991, 2002). The regular completion, update, and improvement of References exposure databases for both the water and sewage network and the public and private buildings in any city, is a high ALA (American Lifelines Alliance). 2001. Seismic fragility formu- priority task for which the needed economic resources lations for water systems. American Society of Civil Engineers. should be allocated. The two different exposure databases https://www.americanlifelinesalliance.com/pdf/Part_1_Guide line.pdf. Accessed 30 Dec 2016. used in the case studies presented for Manizales were Ballantyne, D., J. Eidinger, and C. Howard. 2002. Lifelines. mostly developed using official information showing how Earthquake Spectra 18(S1): 343–361. data, first gathered and arranged for other purposes, is Ballantyne, D., A.R. Guerrero, M. O’Rourke. and E.L. Krinitzsky. clearly useful for disaster risk identification purposes. 1991. Lifelines. Earthquake Spectra 7(S2): 93–117. Ballantyne, D., D. Kessler, W. Elliot. and C. Scawthorn. 1999. Multi- Future risk assessments for the water and sewage net- hazard risk assessment for lifelines—Part 2—case study for the work owned and operated by Aguas de Manizales can Portland, Oregon water supply system. In Proceedings of the 5th explore the consideration of hazards of different origins US Conference on Lifeline Earthquake Engineering. Seattle, (for example, flash floods, landslides, and volcanic ash Washington, USA, August 1999. Bernal, G.A. 2014. Methodology for the modelling, calculation and fall), which can also be assessed and integrated using the calibration of seismic hazard parameters to assess risk in a same probabilistic methodological framework for the esti- probabilistic way (Metodología para la modelación, cálculo y mation of feasible losses explained in this study. calibración de parámetros de la amenaza sísmica para la Finally, regarding the use of risk results for the collec- evaluación probabilista del riesgo). Ph.D. thesis. Universidad Politećnica de Catalunã. Barcelona, Spain (in Spanish). tive insurance of private buildings, it is important to 123 306 Salgado-Ga´lvez et al. Seismic Risk in Manizales: Quantifying Losses for Insurance Purposes

Bernal, G.A., O.D. Cardona, M.A. Salgado-Ga´lvez, and C.P. Villegas. international workshop, anti-seismic measures on water supply, 2015. Update of the seismic microzonation of Manizales International Water Services Association and Japan Water Works (Actualización de la microzonificación de Manizales). In Pro- Association. Tokyo, Japan. ceedings of the VII National Congress of Earthquake Engineer- Marulanda, M.C. 2013. Probabilistic modelling of earthquake ing. Bogota´ D.C., Colombia, May 2015 (in Spanish). economic losses for the estimation of the fiscal vulnerability Bernal, G.A., M.A. Salgado-Ga´lvez, D. Zuloaga, J. Tristancho, D. and financial risk management (Modelación probabilista de Gonza´lez, and O.D. Cardona. 2017. Integration of probabilistic pérdidas económicas por sismo para la estimación de la and multi-hazard risk assessment within urban development vulnerabilidad fiscal del estado y la gestión financiera del planning and emergency preparedness and response: Application riesgo). Ph.D. thesis. Universidad Politećnica de Catalunã. to Manizales, Colombia. International Journal of Disaster Risk Barcelona, Spain (in Spanish). Science 8(3). doi:10.1007/s13753-017-0135-8. Marulanda, M.C. 2014. Risk assessment of Manizales for insurance Cardona, O.D. 2009. Financial disaster risk management: Retention and financial protection of the public and private assets and transfer financial instruments (La Gestión Financiera del (Evaluación del riesgo de Manizales con fines de seguros y Riesgo de Desastres: Instrumentos Financieros de Retención y protección financiera de inmuebles públicos y privados). Pro- Transferencia para la Comunidad Andina, PREDECAN), Comu- grama de Gestioń Integral del Riesgo de Desastres de Manizales, nidad Andina, Lima. Technical report. Gestioń del Riesgo—Manizales. CORPOCAL- Cardona, O.D., and M.C. Marulanda. 2010. Financial mechanisms, DAS (in Spanish). insurance, reinsurance for natural disasters in Latin America and Marulanda, M.C., O.D. Cardona, M.G. Mora, and A.H. Barbat. 2014. the Caribbean. Recent experiences (Mecanismos Financieros, Design and implementation of a voluntary collective earthquake Seguro y Reaseguro contra Desastres Naturales en América insurance policy to cover low-income homeowners in a devel- Latina y el Caribe: Experiencias Recientes). Secretarıá Perma- oping country. Natural Hazards 74(3): 2071–2088. nente del Sistema Econo´mico Latinoamericano y del Caribe, Marulanda, M.C., O.D. Cardona, M.G. Mora, and D.M. Gonzalez. SELA (in Spanish). 2016. Seismic risk assessment for risk transfer: The voluntary Cardona, O.D., M.G. Ordaz, M.C. Marulanda, and A.H. Barbat, collective insurance in Manizales, Colombia. In Proceedings of 2008a. Estimation of probabilistic seismic losses and the public International Conference on Urban Risks, ICUR2016, Lisbon, economic resilience—An approach for a macroeconomic impact June 2016. evaluation. Journal of Earthquake Engineering 12(S2): 60–70. Marulanda, M.C., O.D. Cardona, M.G. Ordaz, and A.H. Barbat. 2008. Cardona, O.D., M. Ordaz, E. Reinoso, L.E. Yamıń, and A.H. Barbat. The financial management of risk from the perspective of 2012. CAPRA—Comprehensive approach to probabilistic risk disasters: Evaluation of the fiscal exposure and alternatives of assessment: International initiative for risk management effec- financial instruments of risk retention and transfer (La gestión tiveness. In Proceedings of the 15th World Conference on financiera del riesgo desde la perspectiva de los desastres: Earthquake Engineering. Lisbon, Portugal, September 2012. Evaluación de la exposición fiscal de los Estados y alternativas Cardona, O.D., M.G. Ordaz, L.E. Yamıń, M.C. Marulanda, and A.H. de instrumentos financieros de retención y transferencia del Barbat. 2008b. Earthquake loss assessment for integrated riesgo). Monografıá CIMNE IS-61, Universidad Politećnica de disaster risk management. Journal of Earthquake Engineering Catalunã, Barcelona (in Spanish). 12(S2): 48–59. Marulanda, M.C., M.L. Carrenõ, O.D. Cardona, M. Ordaz, and A.H. Eguchi, R.T. 1991. Seismic hazard input for lifeline systems. Barbat. 2013. Probabilistic earthquake risk assessment using Structural Safety 10: 193–198. CAPRA: Application to the city of Barcelona, Spain. Natural ERN (Evaluacioń de Riesgos Naturales). 2014. R-System: Program Hazards 69(1): 59–84. for the assessment of earthquake losses in buildings for insurance McGuire, R. 2004. Seismic hazard and risk analysis. Oakland, CA: purposes. Mexico City, Mexico. Earthquake Engineering Research Institute. ERN-AL (Evaluacioń de Riesgos Naturales-Ame´rica Latina). 2011. Miranda, E. 1999. Approximate seismic lateral deformation demands CAPRA-GIS v2.0. Program for probabilistic risk assessment. in multistory buildings. Journal of Structural Engineering 125 Bogota´ D.C., Colombia. (4): 417–425. GFDRR (Global Facility for Disaster Reduction and Recovery). 2014. Ordaz, M. 2000. Methodology for the seismic risk assessment Understanding risk in an evolving world. A policy note. oriented toward earthquake insurance management (Metodología Washington, DC: World Bank. para la evaluación del riesgo sísmico enfocada a la gerencia de GFDRR (Global Facility for Disaster Reduction and Recovery), The seguros por terremoto). Universidad Nacional Autońoma de World Bank Group and JICA (Japan International Cooperation Me´xico. Mexico City, Mexico (in Spanish). Agency). 2015. Pacific catastrophe risk insurance pilot: From Ordaz, M. 2014. A simple probabilistic model to combine losses design to implementation, some lessons learned. Washington, arising from the simultaneous occurrence of several hazards. DC. Natural Hazards 76(1): 389–396. Gonzalez, D. 2014. Buildings exposure database of Manizales (Base Ordaz, M., F. Martinelli, V. D’Amico, and C. Meletti. 2013. de datos de predios de Manizales). Technical report. Gestioń del CRISIS2008: A flexible tool to perform probabilistic seismic Riesgo—Manizales. CORPOCALDAS 2014. http://www. hazard assessment. Seismological Research Letters 84(3): 495– gestiondelriesgomanizales.com/Documentos/Informes/12-BaseDa 504. tos-EdificacionesPublicas-EvRiesgoSismico-DianaGonzalez.pdf. Pitilakis, K., M. Alexoudi, S. Argyroudis, O. Monge, and C. Martin. Accessed 13 Jan 2017 (in Spanish). 2006. Earthquake risk assessment of lifelines. Bulletin of Grossi, P., and H. Kunreuther (eds.). 2005. Catastrophe modeling. Earthquake Engineering 4(4): 365–390. Boston: Springer. Salgado-Ga´lvez, M.A. 2014. Probabilistic seismic risk assessment of Gurenko, E., R. Lester, O. Mahul, and S. Oguz Gonulal. 2006. the Aguas de Manizales System (Evaluación probabilista del Earthquake insurance in Turkey: History of the Turkish catas- riesgo sísmico del sistema de Aguas de Manizales). Technical trophe insurance pool. Washington, DC: World Bank. report. Gestioń del Riesgo—Manizales. CORPOCALDAS (in Isoyama, R., E. Ishida, K. Yune, and T. Shirozu. 1998. Seismic Spanish). damage estimation procedure for water supply pipelines. In Salgado-Ga´lvez, M.A., A.H. Barbat, O.D. Cardona, and M.L. Proceedings of Water and Earthquake ´98 Tokyo, IWSA Carrenõ. 2016a. Comparing observed damages and losses with

123 Int J Disaster Risk Sci 307

modelled ones using a probabilistic approach: The Lorca 2011 (Comparación de los resultados de riesgo sísmico en dos case. International Journal of Disaster Risk Reduction 19: 355– ciudades con los mismos coeficientes de diseño sismo resistente). 365. Revista de Ingeniería, Universidad de Los Andes 41: 8–14 (in Salgado-Ga´lvez, M.A., D. Zuloaga, S. Henao, G.A. Bernal, and O.D. Spanish). Cardona. 2017. Probabilistic assessment of annual repair rates in Salgado-Ga´lvez, M.A., D. Zuloaga-Romero, G.A. Bernal, M.G. Mora, pipelines and of direct economic losses in water and sewage and O.D. Cardona. 2014b. Fully probabilistic seismic risk networks. Application to Manizales, Colombia. Natural assessment considering local site effects for the portfolio of Hazards. doi: 10.1007/s11069-017-2987-z. buildings in Medellıń, Colombia. Bulletin of Earthquake Engi- Salgado-Ga´lvez, M.A., G.A. Bernal, and O.D. Cardona. 2016b. neering 12(2): 671–695. Probabilistic seismic hazard assessment of Colombia with updat- Salgado-Ga´lvez, M.A., D. Zuloaga-Romero, and O.D. Cardona. 2013. ing purposes of the Earthquake-Resistant Building Code for Probabilistic seismic risk assessment of Bogota and Manizales Bridges CCP-14 (Evaluación probabilista de la amenaza sísmica with and without the influence of Caldas Tear Fault (Evaluación de Colombia con fines de actualización de la Norma Colombia de probabilista del riesgo sísmico de Bogotá y Manizales con y sin Diseño de Puentes CCP-14). Revista Internacional de Métodos la influencia de la Caldas Tear). Revista de Ingeniería, Univer- Numéricos para Cálculo y Diseño en Ingeniería 32(4): 230–239 (in sidad de Los Andes 38: 6–13. Spanish). Taleb-Agha, G. 1977. Seismic risk analysis of lifeline networks. Salgado-Ga´lvez, M.A., M.L. Carrenõ, A.H. Barbat, and O.D. Bulletin of the Seismological Society of America 67(6): 1625– Cardona. 2016c. Probabilistic seismic risk assessment of Lorca 1645. through scenario simulations (Evaluación probabilista del riesgo UNISDR (United Nations International Strategy for Disaster Reduc- sísmico en Lorca mediante simulaciones de escenarios). Revista tion). 2013. Global assessment report on disaster risk reduction Internacional de Métodos Numéricos para Cálculo y Diseño en 2013. Geneva: UNISDR. Ingeniería 32(2): 70–78. Vela´squez, C.A., O.D. Cardona, M.L. Carrenõ, and A.H. Barbat. Salgado-Ga´lvez, M.A., D. Zuloaga-Romero, G.A. Bernal, and O.D. 2014. Retrospective assessment of risk from natural hazards. Cardona. 2014a. Comparison of the seismic risk results in two International Journal of Disaster Risk Reduction 10(B): 477– cities with the same earthquake resistant design coefficients 489.

123