payout period due to the short-time horizon of loans in the study area as well as
findings from a previous study conducted in Vietnam that a 3-month time horizon
was too long for respondents to consider (Anderson et al. 2004). Moreover, the
survey was administered when most respondents had just finished harvesting maize,
which corresponds to when many loans are repaid. The choice of payouts in 1 or
2 months’ time was made to minimize effects of the harvesting cycle on elicited
discount rates. The exercise was repeated with increasing values for x until the
respondents switched to Option B. The amount x was increased by in each step
varied from 2,000 VND to 48,000 VND in order to correspond to different annual
interest rates between 2 % and 200 %. Discount rates associated with each alternative were not shown to respondents and the payouts were hypothetical due to
budgetary and logistical constraints. The point at which a respondent switched to
Option B was then used to calculate an average annual discount rate based on the
upper and lower bounds from this switch point.
12 Respondents who did not choose
Option B at any point were identified as having an average annual discount rate of
200 %, which represented the upper limit based on the questions asked. Had more
questions been asked with larger amounts in Option B, the upper limit would have
been higher. In total, 547 individuals from 289 households completed a series of
questions regarding discount rates between November and December, 2011, with
respondents being the household heads and (if applicable) their spouses.
Results indicated that discount rates in Yen Chau were quite high when compared to the interest rate offered by VBARD and VBSP (16.1 % and 6.7 % per
annum in 2007, respectively; Saint-Macary and Zeller 2011). The mean discount
rate was 75.3 % (see Table 5.5). Figure 5.1 displays the distribution of discount
rates across respondents: 5.7 % of the respondents had discount rates lower than
10 %. Most respondents had discount rates far above interest rates offered by formal
banks, which varied between 6.7 % and 16.1 %.
We analyzed how individual discount rates related to respondent characteristics
in terms of poverty status, relative poverty, gender and education. As Table 5.5
shows, mean discount rates were not significantly different between respondents
living in poor and non-poor households based on the absolute poverty method, nor
for different relative poverty terciles based on the poverty index. The difference in
mean discount rates between males and females was significant (p < 0.01), with
females revealing lower discount rates than the males. The difference in mean
discount rates between female and male household heads however, was not significantly different from zero (not shown). Surprisingly, respondents who had not
completed a single year of formal education had, on average, discount rates lower
than those who had completed at least the first year of primary school – a difference
that is weakly significant (p < 0.10). In a separate analysis of the relationship
between discount rates and risk preferences, we found that Pearson correlations
12 For more detail on the methodology, please see Coller and Williams (1999).
192
C. Saint-Macary et al.
findings from a previous study conducted in Vietnam that a 3-month time horizon
was too long for respondents to consider (Anderson et al. 2004). Moreover, the
survey was administered when most respondents had just finished harvesting maize,
which corresponds to when many loans are repaid. The choice of payouts in 1 or
2 months’ time was made to minimize effects of the harvesting cycle on elicited
discount rates. The exercise was repeated with increasing values for x until the
respondents switched to Option B. The amount x was increased by in each step
varied from 2,000 VND to 48,000 VND in order to correspond to different annual
interest rates between 2 % and 200 %. Discount rates associated with each alternative were not shown to respondents and the payouts were hypothetical due to
budgetary and logistical constraints. The point at which a respondent switched to
Option B was then used to calculate an average annual discount rate based on the
upper and lower bounds from this switch point.
12 Respondents who did not choose
Option B at any point were identified as having an average annual discount rate of
200 %, which represented the upper limit based on the questions asked. Had more
questions been asked with larger amounts in Option B, the upper limit would have
been higher. In total, 547 individuals from 289 households completed a series of
questions regarding discount rates between November and December, 2011, with
respondents being the household heads and (if applicable) their spouses.
Results indicated that discount rates in Yen Chau were quite high when compared to the interest rate offered by VBARD and VBSP (16.1 % and 6.7 % per
annum in 2007, respectively; Saint-Macary and Zeller 2011). The mean discount
rate was 75.3 % (see Table 5.5). Figure 5.1 displays the distribution of discount
rates across respondents: 5.7 % of the respondents had discount rates lower than
10 %. Most respondents had discount rates far above interest rates offered by formal
banks, which varied between 6.7 % and 16.1 %.
We analyzed how individual discount rates related to respondent characteristics
in terms of poverty status, relative poverty, gender and education. As Table 5.5
shows, mean discount rates were not significantly different between respondents
living in poor and non-poor households based on the absolute poverty method, nor
for different relative poverty terciles based on the poverty index. The difference in
mean discount rates between males and females was significant (p < 0.01), with
females revealing lower discount rates than the males. The difference in mean
discount rates between female and male household heads however, was not significantly different from zero (not shown). Surprisingly, respondents who had not
completed a single year of formal education had, on average, discount rates lower
than those who had completed at least the first year of primary school – a difference
that is weakly significant (p < 0.10). In a separate analysis of the relationship
between discount rates and risk preferences, we found that Pearson correlations
12 For more detail on the methodology, please see Coller and Williams (1999).
192
C. Saint-Macary et al.