• Tidak ada hasil yang ditemukan

Hyperbolic Discounting and Debt: The Case of Japan

5.3 Hyperbolic Discounting and Debt Behavior

5.3.4 Hyperbolic Discounting and Debt: The Case of Japan

The phenomenon in which hyperbolic discounting aggravates debt tendencies among people is observed quite clearly in Japan as well. Let us take a look at the results of analysis based on micro-level data from our 2010 online survey (the Japan Internet Survey on Preferences Relating to Time and Risk 2010), which was also discussed in previous chapters.

In addition to asking about various socioeconomic attributes, the survey asks about short- and long-term discount rates to see whether a respondent has a tendency toward hyperbolic discounting. Since it also asks whether the respondent was able to complete homework assignments as planned when he or she was a child, we can use that response to identify whether a hyperbolic respondent is a naı¨ve or sophisticated decision-maker. Here, a respondent whose responses to the home- work questions indicate that he or she tended to finish doing homework later than he or she had planned is classified as a naı¨ve hyperbolic discounter, while the rest is classified as sophisticated hyperbolic discounters. As a result, 2277 respondents are grouped as shown in Table5.2.

The number of respondents who exhibited a tendency toward hyperbolic discounting is 960 (42.2 %), of whom 576 (25.3 % of total) are classified as

“naı¨ve,” while the remaining 384 (16.9 % of total) are classified as “sophisticated.”

In the following, I will examine how debt tendencies differ among the three groups:

“naı¨ve hyperbolic discounters,” “sophisticated hyperbolic discounters,” and “non- hyperbolic discounters.”

As can be seen from the discussion thus far, we can theoretically expect “naı¨ve hyperbolic discounters” to have a stronger debt tendency than the individuals in the other two groups. Although it is hard to compare “sophisticated hyperbolic dis- counters” and “non-hyperbolic discounters,” it would not be overly unreasonable to predict that the individuals with a tendency toward hyperbolic discounting, even if they are sophisticated, will show a stronger debt tendency than those who do not,

Table 5.2 Distribution of hyperbolic/non-hyperbolic discounters and naı¨ve/sophisticated Hyperbolic discounters

Non-hyperbolic discounters Naı¨ve Sophisticated

# of respondents 576 384 1,317

(%) (25.3) (16.9) (57.8)

Note: Estimates based on theJapan Internet Survey on Preferences Relating to Time and Risk 2010

other than in extreme circumstances (please seeAppendixof this chapter, in which I explicitly discussed borrowing behaviors of naı¨ve and sophisticated hyperbolic consumers by using a simple analytical model).

Figure5.3first compares how the percentage of debt holders and their average amount of debts differ across the three groups, after excluding mortgages from debts.

We can see two things from this figure. First, as expected, the group of naı¨ve hyperbolic discounters exhibits the strongest preference for debt in terms of both the percentage of debt holders and the amount of debt. For example, naı¨ve hyper- bolic discounters, on average, tend to hold JPY 100,000 more debt than those in the other two groups. Such a positive correlation observed between hyperbolic discounting and debt behavior is also large enough to be statistically significant, indicating that hyperbolic discounting has a nonnegligible aggravating effect on debt behavior.

Second, the difference between the groups of sophisticated hyperbolic dis- counters and non-hyperbolic discounters is not very clear. While the group of sophisticated hyperbolic discounters includes a high percentage of debt holders, the group of non-hyperbolic discounters holds a higher amount of debt. However, these differences are not overly large; they are within the range of statistical error. It indicates that even with a tendency toward hyperbolic discounting, those who recognize its effect and make sophisticated choices borrow money in a manner similar to that seen among non-hyperbolic discounters, who control themselves

JPY 584

JPY 450 JPY 476

23.8%

20.3% 18.1%

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

JPY 0 JPY 100 JPY 200 JPY 300 JPY 400 JPY 500 JPY 600 JPY 700

Naïve hyperbolic discounters Sophisticated hyperbolic discounters

Non-hyperbolic discounters

Amount of debt (JPY in thousands) % of debt holders (right axis) (JPY in thousands)

(%)

Fig. 5.3 Hyperbolic discounting and debt holding. Note: Estimated based on the data from the Japan Internet Survey on Preferences Relating to Time and Risk 2010

well. This type of self-controlling tendency among sophisticated individuals is also confirmed in a similar fashion in the analysis that we cover later.

Figure5.4shows a similar comparison conducted by focusing on excessive debt.

Although it is difficult to define “excessive debt,” I will look at the excessive debt tendency of people from the following five viewpoints.

1. The total amount of debt the individual holds is 30 % or more of his or her annual household income.

2. The individual holds credit card debt.

3. The individual has previously borrowed through unsecured consumer loans.

4. The individual has had a loan application rejected for some reason.

5. The individual has experienced debt consolidation or personal bankruptcy.

The first criterion is somewhat arbitrary; however, considering that the Revised Money Lending Act, which became effective in 2010, essentially prohibits a consumer from adding a new loan that will bring the total debt balance to an amount exceeding one-third of his or her annual income, an amount of debt excluding mortgages that reaches 30 % of household income implies a considerable debt load, at least in terms of social conventions. As for the credit card debt mentioned in the second criterion, it refers to interest-bearing debt due to agree- ments such as revolving credit; it does not include charges paid in full or by installment without incurring interest. Although it is difficult to say whether credit card debt should be considered as part of excessive debt, given the fact that fewer than one in ten people in the sample population in this survey held credit card debt, I

10.4%

12.5%

10.9%

16.3%

3.3%

8.9%

7.3%

8.1%

10.4%

1.0%

6.8% 7.0% 7.2%

10.1%

1.9%

1. Debt ≥ 30% of household income

2. Credit card debt 3. Experience of consumer financing loan

4. Experience of rejected loan application

5. Experience of debt consolidation/personal

bankruptcy Naïve hyperbolic discounters Sophisticated hyperbolic discounters Non-hyperbolic discounters

Fig. 5.4 Hyperbolic discounting and excessive debt. Note: Estimated based on the data from the Japan Internet Survey on Preferences Relating to Time and Risk 2010. For details, see Ikeda and Kang (2015)

have determined that credit card debt holders are probably highly likely to have a debt tendency. While criteria 1 and 2 question whether or not the individual currently holds debt, the remaining three relate to his or her past experiences.

This is why I specifically speak of “the excessive debt tendency of survey respondents.”

As shown in the figure, the results indicate that the group of naı¨ve hyperbolic discounters includes the highest percentage of individuals who exhibit excessive debt tendency according to each of the five criteria. For example, while more than 12 % of naı¨ve hyperbolic discounters are credit card debt holders, only 7 % or so of individuals in the other two groups are credit card debt holders. The percentages of individuals whose loan application has been rejected and individuals who have experienced debt consolidation or personal bankruptcy within the naı¨ve hyperbolic discounter group are also 1.6–1.7 times higher than those in the other groups.

The group with the next highest percentage of individuals who exhibit an excessive debt tendency is that comprising sophisticated hyperbolic discounters;

however, the difference with the group of non-hyperbolic discounters is not as considerable as that with the group of naı¨ve hyperbolic discounters. This is the same as in the case of debt holders, as seen in Fig. 5.3. However, the distribution of individuals with/without excessive debt differs across the three groups under all five criteria. Those differences are statistically significant.

The above was an examination of the effect of hyperbolic discounting on each of the five criteria that seem to indicate an excessive debt tendency. By using a statistical method called principal component analysis and examining these data together, we can extract the common factors included there. The score obtained there can be considered to represent the hidden inclination to make choices that lead individuals to exhibit the specific excessive debt behavior seen above. Let us refer to this as the “excessive debt tendency index.” This index is set up in such a way that the mean equals zero and a higher index value implies a stronger excessive debt tendency. Figure5.5, which compares how the index value varies among the three groups, shows more definitively than Fig.5.4that the group of naı¨ve hyperbolic discounters bears the strongest excessive debt tendency.

In the discussion thus far, I have compared debt behavior by merely dividing the survey respondents into three groups, based on whether they are hyperbolic or not and, if hyperbolic, whether they are naı¨ve or sophisticated when making a decision;

I did not exclude the effects of various confounding factors such as respondents’ socioeconomic attributes and preferences other than the subjective discount rates.

However, even when these are taken into consideration, the bottom line remains the same: hyperbolic discounting increases debt tendency, and it becomes particularly evident among naı¨ve hyperbolic discounters who cannot predict their own self- control problems.

Table 5.3shows the estimated extent to which hyperbolic individuals have a stronger debt tendency than non-hyperbolic individuals, when the effect of other factors—the presence of the sign effect, gender, age, education, income, assets, etc.—

is excluded. In other words, it shows the estimated marginal effects on debt tendency that would occur when a non-hyperbolic individual becomes hyperbolic, assuming that other factors are held constant (for more details, see Ikeda and Kang 2015).

Table 5.3 The effect of hyperbolic discounting on debt holding behavior Naı¨ve hyperbolic discounters

Sophisticated hyperbolic discounters Overall debt Probability of having a debt Increases by 4.7

percentage points

Increases by 1.3 percentage points

Amount of debt Increases by JPY

110,000

(Decreases by JPY 10,000)

Excessive debt Probability of debt30 % of income

Increases by 2.5 percentage points

(Increases by 1.3 percentage points) Probability of having a credit

card debt

Increases by 4.8 percentage points

(Increases by 0.4 percentage points) Probability of having used

consumer financing

(Decreases by 0.9 percentage points)

(Increases by 0.6 percentage points) Probability of having had a

loan application rejected

Increases by 4.0 percentage points

(Decreases by 0.3 percentage points) Probability of experiencing

debt consolidation/personal bankruptcy

(Increases by 0.3 percentage points)

(Decreases by 0.6 percentage points) Excessive debt tendency index Increases by 0.27 (Decreases by 0.02) Note: Estimated based on the data from theJapan Internet Survey on Preferences Relating to Time and Risk 2010. The figures in parentheses are not statistically significant. For details, see Ikeda and Kang (2015)

0.24

-0.07

-0.10 -0.15

-0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30

Naïve hyperbolic discounters

Sophisticated hyperbolic discounters

Nonhyperbolic discounters

Fig. 5.5 Excessive debt tendency index. Note: Estimated based on the data from theJapan Internet Survey on Preferences Relating to Time and Risk 2010. For details, see Ikeda and Kang (2015)

We can see from the table that the debt tendency—both in terms of overall debt and excessive debt tendency—becomes stronger in the same way when an indi- vidual is naı¨ve and underestimates his or her own self-control problems. The probability of holding debt increases by 4.7 percentage points and the amount of debt increases by JPY 110,000 when the individual is hyperbolic. It increases the probability of having a credit card debt by 4.8 percentage points. This effect is far from small, considering the fact that the percentage of individuals within this sample population who hold credit card debt was less than 9 %. In contrast, the effect of hyperbolic discounting is not so clear among sophisticated individuals who correctly understand their own self-control problems. In fact, whereas the effects on naı¨ve decision-makers all exceed the margin of error, in the case of sophisticated individuals, they are all statistically insignificant.

Among naı¨ve individuals, hyperbolic discounting also increases the probability of having had a rejected loan application by 4 percentage points. However, caution is required in its interpretation; lenders cannot tell whether a loan applicant is hyperbolic or not and thus whether he or she is a naı¨ve or sophisticated hyperbolic discounter. Of course, assuming that the household financial situations (such as income and assets) of naı¨ve hyperbolic individuals are poor (which is indeed confirmed statistically to some extent), it is certainly possible that their loan applications get rejected on the basis of their poor ability to repay. However, since the results in Table 5.3 are obtained after excluding the effects of other variables like income and assets, such an explanation will not hold true. Instead, it would be more logical to conjecture that even when their household financial situations do not look particularly good in terms of being able to sufficiently cover future repayments, hyperbolic and naı¨ve individuals try to borrow the maximum amount at which the application could actually be rejected and, as a result, their loan applications are rejected at a higher probability.