For the past century, researchers and public officials responsible for allocating public expenditures have sought a method to quantify the “value” of public access venues such as libraries and museums (Rea, 1916; Dana, 1920; Van House et al. 1987; Weech, 1998; and Smith, 1994). One method that has emerged in the past decade is benefit cost analysis (Holt & Elliot, 1998; Holt & Elliot, 2003; Aabo, 2005; Chung, 2008; Aabo, 2009). Benefit cost analysis is a general method of putting disparate measures into like terms, usually monetary terms (Krutilla, 1967; Zerbe, 2007; among many). Benefit cost analysis is one lens used to inform policy makers, and does not replace alternative methods for evaluating proposed projects. Accordingly, this benefit cost analysis is intended to inform policy makers, not to substitute as a decision making tool.
This research is novel in three ways. First, it applies benefit cost methods in new research areas, including developing countries. Second, it contrasts findings across three types of public access venues. Third, this research is the first attempt to estimate both an upper and lower range of the benefits of public access for multiple venues in multiple countries, using mixed benefit costs methods.
This chapter addresses two questions:
- What are the costs and benefits of public access to information and communication technologies?
- How do those cost and benefits differ across countries, demographic characteristics, location of the public access venue, and public access venue type?
The geographic scope of this work includes Bangladesh, Brazil, Chile, Ghana, and the Philippines, examining three types of public access venues: cybercafés, libraries, and telecenters. Within each of these contexts the study analyzes how the independent variables of gender, location, and usage correlate with perceived benefits. The study found broad and statistically significant evidence of willingness to pay for public access to ICT, among both users and non-users.
Measures and methods
Benefit cost analysis assesses a specific good or service in terms of the expenditure it requires, the willingness to pay for it, and the benefit to the individual. The benefits of the good or service are estimated using two types of measures: revealed preference and stated preference.
Broadly, revealed preference measures are responses to the question, “what have you paid,” while stated preference measures are responses to the question, “what would you pay.” Researchers collect revealed preference measures either by observing choices or asking respondents what choices they have made in the past. A revealed preference measure might be an individual’s direct expenditures on a good or service, or the costs an individual incurs to access a good or service. Stated preference measures, on the other hand, ask respondents to indicate what they would pay for a good or service. This study uses both stated and revealed preference measures, in four benefit cost analysis methods which are outlined here and described in more detail below.
- Analysis of benefits
- User: revealed preference
- Non-user: stated preference
- User and non-user: stated preference (Chile case study)
- Analysis of costs
- Venue operator: Stated expenditures
For the analysis of the benefits accrued to users of public access, the study employed the travel cost method, a method of analysis for revealed preferences. The travel cost method is an aggregation of estimated non-market and market costs an individual incurs to access a good or service. Specifically, the study estimates the time and money an individual uses to travel to a venue that provides ICT. The time estimate is monetized as a function of income, and the travel costs are aggregated to estimate the total monetary costs. These two monetized estimates are summed, and then multiplied by the total number of trips to visit the venue per year. This method results in a minimum value that an individual is willing to spend annually to use public access venues. The estimate is a minimum, as users might have been willing to travel farther or to incur higher costs than they were observed to have incurred. Note that the travel cost measure does not include analysis of other, non-travel-related costs of using public access services.
The data for the travel cost method are drawn from the user survey. The user survey asked for travel distance, travel time, transportation costs, travel methods, and frequency of trips to public access venues. The benefit cost analysis framed three hypotheses for users: (1) that males will incur greater travel costs than females, (2) that rural users will incur greater travel costs than urban users, and (3) that users will incur greater travel costs to access libraries than cybercafés or telecenters. The survey also asked about demographic characteristics, such as age, gender, employment status, wage, and home access to the internet. Finally, country level information on minimum hourly wages was used to establish a reasonable baseline “cost of time,” for individuals with missing or no income data.
The simplified model below yields total costs of travel, where T is the travel time, H is hourly wage, C is cost of transportation, and F is frequency of usage. As an estimate of the cost of time, 50% of the individual’s hourly income is used. For individuals with no employment, including students, the travel cost of time as 50% of the minimum wage is set. Finally, all wages are converted to purchasing power parity (PPP) dollars to enable international comparisons.
The result is an estimate of the total costs of travel incurred by an individual to reach a public access venue. This measure gives a minimum value of public access, which may be contrasted with stated preference willingness to pay, as an upper bound of value.
The second method analyzed non-users’ stated preference for public access venues. Non-users were asked to state what they would be willing to pay to prevent the loss of existing public access venues. The value that non-users place on a public service is important for two reasons. First, in the study contexts, the characterization of non-users’ perception of benefits can substitute for broad public opinion. Second, the values that non-users place on a good or service (often called a non-use value) is an important component of the total value to the community (Arrow et al. 1993, among many). These methods generally result in higher total valuations than revealed preferences, for two reasons. First, stated preference surveys are designed to estimate the full willingness to pay for the good, not the amount that an individual is actually charged. Second, because there are no costs incurred in stating a high willingness to pay, respondents may state a willingness to pay that exceeds their budget constraints. For these reasons, an open-ended stated preference question serves as a high estimate of total value of public access. Indeed, aggregating the benefits to non-users will often result in total benefits greater than the aggregation of the users — particularly when the number of non-users greatly exceeds the number of users.
Data for the non-users were drawn from the non-user survey, which asked individuals how much they would be willing to pay annually to prevent the closure of public access venues.
The benefit cost analysis framed two hypotheses for the non-users: (1) that males will state higher willingness to pay than females, and (2) that stated willingness to pay will be greater for libraries than for cybercafés or telecenters. Demographic data was collected to enable tests of hypotheses of independent variables that predict willingness to pay. Finally, all figures are converted to PPP dollars to enable cross-country comparisons.
The in-depth study of the Chilean public used the contingent valuation method, in a referendum format using a follow-up question. Contingent valuation is a stated preference survey method that asks respondents whether they would be willing to pay to prevent the loss of a good or service. In the referendum format, individuals are presented with a bid amount that they can either accept or reject. The follow-up question presents a higher amount if the respondent was willing to pay the first bid and a lower amount if they rejected it, allowing for greater accuracy in estimating stated preferences.
The survey presented a dichotomous choice: Would you be willing to pay [a bid amount] to prevent the reduction of hours of operation of a public access venue? (The percentage of reduction was randomly rotated between 25%, 33%, and 50%.) Following the initial response, the same question was asked with a follow-up bid — roughly half the initial amount if they had declined and double if they had accepted paying the first bid amount. Total willingness to pay was estimated as a linear extrapolation of the percentage of reduction of hours: thus, willingness to pay to prevent loss of the entire venue was represented by doubling the stated willingness to pay to prevent a 50% reduction in the hours of service. Responses were solicited from a representative sample of 1,100 individuals using a random-digit telephone interview. To test hypotheses about predictors of willingness to pay, researchers asked respondents about their usage history of cybercafés, libraries, or telecenters, as well as their demographic characteristics.
This report uses the midpoint of the value range that is bounded by the bids amounts, then treats the valuation responses as continuous variables. This allows a testing of differences in responses for the independent variables of interest: gender, location, and usage history. The framing hypotheses are the same as for the non-user analysis: (1) that males will state higher willingness to pay than females, (2) that willingness to pay will be greater for libraries than for cybercafés or telecenters, and (3) that users will state higher willingness to pay than non-users. In the separate report on benefit cost, the data are modeled to show how additional independent variables predict willingness to pay (Davis, forthcoming).
Finally, the study attempted to compare the costs with the benefits of operating public access, for libraries, telecenters, and cybercafés, in both urban and rural areas. The sole source of data on costs was the venue survey, which asked venue operators two cost-related questions. First, they were asked to estimate line item costs, including rent, electricity bill, internet connection, and “other.” Then they were asked to estimate the total cost of running the establishment. Unfortunately, the response rates were very low on both overall costs and itemized costs, yielding insufficient data to support generalizations. Therefore, in this report, cost data cannot be included to compare with benefits. (The existing results on cost data will be discussed in the full report on benefit cost analysis.)
Summary of findings
The analysis provided evidence that public access users incur annual travel costs ranging between $15 (Ghanaians, to cybercafés) and $83 (Brazilians, to cybercafés). This finding indicates the minimum annual benefit that accrues to users of public access venues, and has relevance for policy and program design (See note to Table 8.1).
In tests of preferences between venue types, the study found that, for Chile and the Philippines, where reliable travel cost information existed for libraries, travel costs were highest for libraries — indicating high user valuation of benefits. There were few statistically significant differences by gender, and minimal differences in travel costs between urban and rural settings. Within Bangladesh (the only country that had travel costs in both urban and rural areas), the study found higher travel costs for telecenters in urban than in rural areas.
The analysis of non-users revealed evidence of widespread willingness to pay for others to have public access to ICTs. The willingness to pay ranged from $2 (for libraries in Bangladesh) to $64 (for cybercafés in the Philippines). The differences in willingness to pay between libraries and cybercafés may reflect differences in degree of familiarity with different venue types. In Brazil and Chile, the highest willingness to pay was for libraries; in Bangladesh, Ghana, and the Philippines, it was for cybercafés. In contrast to users, there were significant gender differences in non-users’ willingness to pay for different types of venue, suggesting that men and women value venues differently.
As the first multi-country study of willingness to pay for public access, these findings should inform policy makers’ assessment of the value of public access venues, to take into account the benefits perceived even by individuals that do not themselves use the venues. The public at large recognizes the value of public access venues and favors their provision.
Similarly, the in-depth study of the benefits of public access venues in Chile found overwhelming support for public access venues. The mean individual willingness to pay to keep all three types of venues open was $60. Among the responses for individual venue types, support for libraries was highest, with a mean of $49 to keep libraries open. The study found moderate support for telecenters, with willingness to pay of $17, and the lowest support for cybercafés, with a willingness to pay of $7. Women were willing to pay more, for all venues. There were no significant differences in willingness to pay between respondents within Santiago and those outside the capital city. The public evidently prefers libraries as public access venues over telecenters and cybercafés.
Table 8.1 represents a summary of findings from the user, non-user, and in-depth research. Estimates of the non-users represent a high estimate (overestimate) of perceived benefits for the general public; estimates for users represent the minimum annual benefit for users; and the in-depth study results include both users and non-users.
Note that these findings are a simplification, representing extrapolations from the data that were collected and analyzed. The estimates in all the tables in this section represent midpoints of ranges, rather than point estimates. The findings are moreover subject to all the caveats discussed in the findings.
|Table 8.1: Summary of user and non-user estimates of public value per person (PPP dollars)|
|Estimate type||Cybercafés||Libraries||Telecenters||All Venues|
|Bangladesh||Non-users: Stated WTP||$7.09||$5.90||$2.31||$1.60||$6.57||$5.83||$11.11||$7.65|
|Users: Travel Cost||$36.88||$47.83||$15.51||$35.19||$39.51||$36.21|
|Brazil||Non-users: Stated WTP||$12.03||$9.53||$13.73||12.02||$12.82||$10.36||$40.20||$26.23|
|Users: Travel Cost||$80.24||$84.91||$13.56||NA||$31.58||$30.94|
|Chile||Non-users: Stated WTP||$10.58||$12.42||$13.21||$14.66||$12.44||$12.42||$12.79||$20.10|
|Users: Travel Cost||$33.31||$32.15||$43.02||$43.80||$29.03||$40.02|
|Ghana||Non-users: Stated WTP||$17.45||$8.89||$12.88||$9.74||$14.51||$6.81||$27.42||$16.86|
|Users: Travel Cost||$15.01||$15.97||$13.33||$1.40||$65.62||$18.23|
|Philippines||Non-users: Stated WTP||$66.58||$61.86||$46.83||$55.08||$42.64||$41.95||$115.90||$86.84|
|Users: Travel Cost||$34.96||$32.94||$42.14||$55.78||$32.29||$12.63|
Table 8.1 can be illustrated with an example. A policy maker in a community in Bangladesh could estimate the benefits of providing a public access venue using the data from this report, as follows:
For a population of 100, including 45 male non-users and 45 female non-users, the non-user public value of a public access cybercafé would be (45 x $7.09) + (45 x $5.90) = $583.20, as an upper bound.
In the same community, for 5 male users of cybercafés and 5 female users of cybercafés, the minimum annual value for a public access cybercafé would be (5 x $36.88) + (5 x $47.83) = $423.55, as an annual minimum value.
Users: Benefits of access
The analysis of benefits sought to accomplish two tasks: first, to estimate the annual expenditures of users on public access venues; and second, to test the hypotheses that the independent variables of venue type, gender, and location influence the annual expenditures of users on public access venues.
Annual expenditures were found ranging from a high of $83 for Brazilian users of cybercafés, to a low of $15 for Ghanaian users of cybercafés. These expenditures establish a minimum annual value for users of each venue. Table 8.2 shows the mean annual travel cost as reported by respondents that use each of the public access venues, for each country, in purchasing power parity dollars.
|Table 8.2: Mean reported annual travel cost in PPP dollars, by country|
|Note: Shaded cells have a sample size of fewer than 29 observations, and therefore the findings are less generalizable|
The most surprising finding shown in Table 8.2 is that travel costs are roughly equal for the three types of venue, in the countries for which consistent data exists — Bangladesh, Chile, and the Philippines (excluding the shaded items, with too few observations). There was a slightly higher cost incurred by users of libraries as compared to cybercafés, in both Chile and the Philippines, though these differences were not statistically significant due to small sample size and high variance in the data. If these patterns are borne out in further research, however, this finding suggests that users in Chile and the Philippines are willing to spend more to reach a library than a cybercafé.
Cybercafé users in Brazil incur the greatest travel costs of any venue for any country, whether because of a strong preference for cybercafés or because of lack of alternatives. In fact, the majority of public access venues in Brazil are cybercafés or other independent, stand-alone facilities (see Chapter 3). Given limited options, Brazilians are incurring the greatest travel costs to access public access venues.
Table 8.3 is a summary table of the travel costs incurred by urban and rural users in Bangladesh — the only country with data on travel costs in rural areas. The costs of accessing cybercafés is the same for rural and urban users, but urban users in Bangladesh pay more than double the rural cost for accessing telecenters. (With sample sizes of 255 and 152, these two costs are statistically different at the p <.001 level.) These findings imply that either urban users in Bangladesh strongly prefer telecenters to cybercafés, or telecenters are more abundant and therefore more frequently used than cybercafés.
|Table 8.3: Mean travel costs by location (Bangladesh)|
|Note: Shaded cells have a sample size of less than 29 observations, and therefore the findings are less generalizable
* P < .0001
In contradiction of the framing hypothesis regarding gender distinctions, the study found no significant differences by gender in the cost of travel to access any venue, in any country. Table 8.4 shows the mean travel costs incurred by men and women for each venue type and for each country. For users of public access venues, no gender difference in the travel costs was found.
|Table 8.4: Mean travel costs by gender|
|Note: Shaded cells have a sample size of less than 29 observations, and therefore the findings are less generalizable|
The travel cost analysis represents a first effort to quantify the travel costs incurred by users in different settings, for three types of venue. The study was able to test hypotheses on differences in travel cost by venue type, gender, and location. These findings establish a minimum public value for users of the public access venues. The summary tables above can serve as tools for policymakers, to understand the tradeoffs involved in providing (or subsidizing) public access venues.
Non-users: Willingness to pay to maintain access
Are non-users willing to pay for public access, and is there an association with either type of venue or gender of respondent? The analysis shows unambiguous evidence of widespread willingness to pay on the part of the non-user population, to keep public access venues open for the general public. Willingness to pay ranged from a high of $101.30 in the Philippines, for all venues, to a low of $2.02 in Bangladesh, for libraries. Responses were internally consistent, showing willingness to pay more to keep all venues open than to keep a single venue open. These findings provide strong evidence of wide perception of the benefit of public access ICTs, and should inform policy makers’ assessment of the importance of providing public access facilities. These findings are particularly important because they come from the population of non-users. In many countries, the non-user population is in the vast majority, and the values they express can serve as a proxy for the nation’s population as a whole. (Note, however, that the sample for the non-user survey was limited to people living near a public access venue, and therefore does not fully capture the characteristics of the national population of non-users.)
Table 8.5 is a summary of non-users’ stated willingness to pay, by venue type and by country, showing the mean amount that respondents were willing to pay, as well as the standard deviation and the number of observations. For Brazil, for example, Table 8.5 shows data on 299 non-users, whose average willingness to pay to prevent the closure of all three types of public access venue was $33.38. Every value is significantly different from zero (p < .05).
These values represent a high estimate of the perceived public value of public access venues, since these are not actual costs incurred but a stated willingness to pay. The previous section, on users, provided actual cost figures that tend to understate willingness to pay (since users might be willing to pay more than the current actual costs). Together, the two measures provide a range of the estimated value of public access. These two measures are not a perfect match, however; the estimates for users are an annual measure, while the stated willingness to pay from non-users is a one-time stated preference.
|Table 8.5: Non-user willingness to pay, by country and type of venue|
|Note: * Statistical indicators are: mean, standard deviation (s.d), and number of observations (n)|
Respondents’ willingness to pay differed by venue, indicating their understanding of the differences between the venue types as well as the types of services offered in each venue. The internal value ranking of venue type could be at least partially explained by differences in respondents’ familiarity with the venues. In Bangladesh and the Philippines, with relatively few libraries and plenty of cybercafés, stated willingness to pay for cybercafés was much higher than for libraries.
The study also found statistically significant gender differences in willingness to pay for cybercafés and telecenters in Brazil and Ghana. In each of those cases, men stated a higher willingness to pay than women. And for Bangladesh, Brazil, and Ghana, men were willing to pay more for all venues than women were. However, there was no statistically significant gender difference in any country in willingness to pay for libraries. Table 8.6 summarizes the findings.
|Table 8.6: Non-user willingness to pay, by gender and country|
|Note: * P < .1. ** P < .05.|
For the surveyed countries with greater per capita income, libraries ranked above other venues in non-user willingness to pay. In Brazil and Chile, non-users valued libraries higher than cybercafés and telecenters — the reverse of the pattern in Bangladesh, Ghana, and the Philippines. One reason for this might be that public services, including public libraries, are better in wealthier countries, and therefore individuals value them more highly. In less wealthy countries, on the other hand, libraries are relatively scarce and therefore less familiar to non-users. Here, the private sector provides public access in the form of cybercafés, and greater familiarity with these services results in a higher public valuation.
|Table 8.7: Gross domestic product (GDP) per capita|
|Source: World Bank: http://data.worldbank.org/indicator/NY.GDP.PCAP.CD|
Another characteristic shared by Brazil and Chile, as higher-income countries, is that a high percentage of the total population uses the internet. Chilean internet usage in 2009 was estimated at 41%, and in Brazil at 38%. One hypothesis is that in a society that is broadly familiar with ICT services, individuals value public access highly even if they do not use the venues themselves.
|Table 8.8: Population, internet users, and percentage of the population online (2009)|
|Population||Internet Users||Users (%)|
|Source: Population and internet usage statistics from the CIA World Factbook. Accessed Sept. 2012: https://www.cia.gov/library/publications/the-world-factbook/rankorder/2153rank.html|
Libraries in Bangladesh were the by far the least valued public access venue, with a stated willingness to pay of just $2.02 PPP. The low valuation may reflect a general lack of familiarity with the services that a library provides: Bangladesh had an internet usage rate of only 0.4% in 2009. The low willingness to pay for library services in this case indicates the internal validity of the survey: where publicly provided services are relatively unknown, they ranked lower than privately operated services in willingness to pay.
In countries with low per capita income, non-users assign greater value to cybercafés than to other public access venues. If per capita GDP is used as a predictor of the level of provision of social services, such as public libraries, a working hypothesis would be that lower income countries tend to value cybercafés over public libraries in part because of lack of familiarity with libraries. These data also suggest the need for research on the provision of public (social) services more generally, investigating the relationships among the revenue model, the level of service provision, and the public’s stated willingness to pay for public services. For example, where paying for social services is not a norm, it may be that users assign low valuation because they reject the payment vehicle as opposed to rejecting the provision of the service.
Summary of non-user survey findings
The non-user data provided valuable insights on the creation of public value through public access venues. Non-users understood that public access was valuable, they understood the differences between venue types, and they reported willingness to pay to prevent the loss of the public value. In countries with higher per capita income and more widespread internet usage, libraries were valued more highly than cybercafés and telecenters. In countries with low per capita income and low internet usage, cybercafés were valued more highly than libraries. Finally, with the exception of Chile, male non-users expressed greater willingness to pay for the provision of public access venues.
In-depth study: Chile
The Chilean in-depth study focused solely on capturing the perceived benefits of public access, testing a set of distinctions across venue types and user groups. The factors examined as potentially affecting willingness to pay for public access venues included: venue type; location (rural or urban); and user’s gender and usage history. With a sample of 1,100 respondents, including both users and non-users, the study found statistically significant differences in the willingness to pay by venue type as well as specific differences by gender and usage history. The data showed no significant differences in willingness to pay by location.
On average, respondents were willing to pay $7.14 to prevent the reduction of hours of cybercafés, $16.92 for telecenters, $48.93 for libraries, and $59.99 for all venues (all results significantly different from zero willingness to pay, Table 8.9). Moreover, the responses were internally consistent in assessing the value of all three venues as greater than any single one.
The results indicate that:
- The Chilean public understands the differences between venue types and places different value on different venue types.
- Chileans are willing to pay more to prevent the closure of libraries than telecenters, and more for telecenters than cybercafés.
|Table 8.9: Venue valuations overall, Chile (USD)|
|Note: * Statistical indicators are: mean, standard deviation (s.d.), and number of observations (n). ** P < .001.
The study found no support for the hypothesis that gender affects the perceived benefits of different types of public access venues: there was no statistical difference between the valuation of men and women for cybercafés, libraries, or telecenters. However, women placed a greater value than men on all three venues together (Table 8.10), stating that they would be willing to pay $64.16 to prevent the reduction of all services, whereas men stated that they would be willing to pay $55.79.
|Table 8.10: Venue valuations, by gender, Chile (USD)|
|Note: * Statistical indicators are: mean, standard deviation (s.d.), and number of observations (n). ** P < .05.|
The hypothesis regarding location was that that rural users would place a higher value on public access than urban users. To test this hypothesis, the results for Chile’s capital city of Santiago were compared with results from areas outside the city. There was no significant difference between the two locations in the valuation of cybercafés, libraries, telecenters, or all three venues (Table 8.11). This rather surprising finding may reflect the particular population distribution of Chile. In a country of 17 million people, more than 10 million live in the capital city or the surrounding metropolitan area. Furthermore, 89% of the population is urbanized, making it difficult to test for differences by location. (Population statistics for 2010 are found in https://www.cia.gov/library/publications/the-world-factbook/geos/ci.html).
|Table 8.11: Value by location, Chile (USD)|
|Note: * Statistical indicators are: mean, standard deviation (s.d.), and number of observations (n)|
The study also compared the responses of users and non-users of public access venues in three ways:
- A comparison of the willingness to pay of users vs. non-users of public access venues.
- A comparison of the value users placed on the type of venue they used vs. the value that others (users and non-users) placed on that venue.
- A comparison of the value users placed on the venue they used, for users who only used one type of venue, vs. the value that users of other venues placed on that type of venue. (For example, the value that library-only users placed on libraries was compared to the value placed on libraries by people who only used cybercafés or telecenters.)
The first test showed no statistical difference between the valuations of users and non-users for cybercafés and telecenters (Table 8.12). However, users of any public access venue valued libraries (as well as all venues) more highly than did non-users. In other words, users of public access venues differ from non-users only when valuing libraries (and all venues combined), not when valuing cybercafés or telecenters. Moreover, the values assigned by all respondents to libraries were much higher than those for the two other venues.
|Table 8.12: Users of no venues contrasted with users of any venue, Chile (USD)|
|Note: * Statistical indicators are: mean, standard deviation (s.d.), and number of observations (n). ** P <.01. *** P<.0001.|
Second, the study compared the value a user placed on the venue they used vs. the value that all others (users of public access venues and non-users) placed on that venue (Table 8.13). People who reported using cybercafés valued cybercafés at $8.21, while people who did not use cybercafés valued them at $6.39, suggesting that users of cybercafés are willing to pay more for them than non-users of cybercafés; similar results are shown for libraries. Finally, the results show that people that use any venue are more willing to pay for all venues than people that use no public access at all. Users of any public access venue were willing to pay $65.59, whereas non-users were willing to pay $52.44. This finding shows that both users and non-users are willing to pay for access, though at different levels, and that even non-users perceive differences between venues.
|Table 8.13: Differences in valuation of users and non-users of the venue they are valuing, Chile (USD)|
|Note: * Statistical indicators are: mean, standard deviation (s.d.), and number of observations (n). ** P <.10. *** P<.05. **** P<.01.|
The third test compares the value that individuals who only used one type of venue placed on that venue type vs. the value placed on that venue type by individuals who only used other types of venue. The results showed that cybercafés were valued more highly by the exclusive users of cybercafés, at $9.82, than by the users of libraries and telecenters, at $5.40 (Table 8.14). Interestingly, however, the valuation of libraries by exclusive users of libraries did not differ from the valuation of libraries by exclusive users of cybercafés or telecenters. This finding suggests that public access users who do not use libraries still appreciate the value of public access services in libraries, and are willing to pay to continue the provision of this service.
|Table 8.14: Valuation by users of that venue contrasted with users of other venues, Chile (USD)|
|Only use this venue||mean||$9.82**||$49.75||$22.43|
|Only use other venues||mean||$5.40**||$52.78||$17.58|
|Note: Shaded cells have a sample size of less than 29 observations, and therefore the findings are less generalizable. * Statistical indicators are: mean, standard deviation (s.d.), and number of observations (n). ** P <.01.|
In conclusion, the in-depth study tested differences in stated willingness to pay by gender, usage history, location, and venue type. The larger sample size of 1,100 respondents yields a conclusion that public access has widespread support, across venues as well as by users and non-users. Moreover, the consistently higher willingness to pay for libraries indicates a public preference for libraries over other venue types. This preference holds when controlling for gender, location, and usage history.
In the benefit-cost study, efforts to estimate the costs of providing services were largely unsuccessful. First, the response rate was very low — especially for libraries, rural venues, and telecenters (outside of Bangladesh). Second, the survey showed little internal consistency: the sum of the line item estimates rarely agreed with the estimated total costs. Finally, the reported itemized costs were sparse; a venue might report internet costs but not staff costs. For this reason, the research team chose not to include a table of benefit cost ratios.
The most important finding from the effort to estimate costs is a methodological observation. Large sample survey methods are inappropriate for complex accounting questions such as itemized costs, which are difficult for respondents to answer. A more promising approach would survey smaller samples of representative venues, using in-depth methods that are designed to build trust, and would work closely with respondents to ensure that they understand and can answer accounting questions.
What are the costs and benefits of public access to ICTs, and how do those differ by venue, gender, location, and usage history? By using mixed methods, the study was able to estimate benefit levels, establishing minimum values (actual costs reported by users) and maximum values (willingness to pay, as reported by non-users). Contrasting public benefits with costs was not possible, as data on the cost of operating venues were inadequate.
The study resulted in four broad findings. First, it found broad and statistically significant willingness to pay for public access to ICTs, among both users and non-users. Even non-users recognized the public value created by ensuring public access to information and communication technologies. Second, the study found that both users and non-users perceive differences in the value of the three types of public access venues. The Chile in-depth study found much higher valuations for libraries over telecenters or cybercafés, for both users and non-users. Third, the study found few differences in willingness to pay by gender: men and women evaluated the importance of public access venues equally. Finally, there were limited differences in valuation of public access venues across urban and rural contexts.
As with all benefit cost analysis, this research is best used as an informative tool, not as a substitute for other valuation methods or a decision-making formula. The generalizability of the findings is in all cases limited by sample size, estimation error, and methodological limitations. This research, along with other findings in this report, provides an informative tool for policy makers in weighing the costs and benefits of (for example) providing libraries, or promoting alternative forms of public access to information and communication technologies.
- This chapter is an abridged version of the Global Impact Study’s benefit cost in-depth study report, “Perceptions of Value for Public Access to Information and Communication Technology in Five Countries: A Mixed Method Benefit-Cost Analysis Approach for Informing Policy.” For this reason, it has a slightly different character than other chapters in this report. ↵
- Benefit cost analysis does not differ substantively from cost benefit analysis except in placing the emphasis on the objective of public policy: maximizing public benefit. ↵
- Analysis also showed that respondents were internally consistent and rational; willingness to pay for all venues was always greater than for any one venue. ↵