日治時期台灣社會對近代國家制度的調適─貸款的風險貼水：日治時期民事訴訟的調查研究

行政院國家科學委員會專題研究計畫 成果報告

貸款的風險貼水：日治時期民事訴訟的調查研究

中 華 民 國 93 年 12 月 27 日

Final Report for 日治時期台灣社會對近代國家制度的調適-貸款的風險貼水：日

治時期民事訴訟的調查研究

Introduction

During the first half of the 20th century, the risk premiums on loans between friends, neighbors and relatives were much higher in the south than in the north. This project has tried to answer this question by comparing court records from Taichung in the north and Chiayi in the south. We have found that the proximate cause of the larger risk premium was that lenders in the south were more reluctant to take debtors to court. Formal court costs were the same in the north and south, but the informal costs of taking a debtor to court were higher in the south. Lenders had to be almost certain of winning a judgment against the debtor before they would sue, so fewer bad debts were taken to court. This was particularly true in rural areas. We still have not identified the specific cost that kept southern lenders from going to court. In general, southerners were not reluctant to go to court concerning other matters. The evidence also suggests that financial transactions in the south were at least as

transparent as in the north.

One hypothesis which seems to fit the data is that in traditional southern Taiwanese practice, loans were accompanied with an unwritten insurance policy. During the Qing dynasty, loan collection was pursued not through law suit but through the community and southern communities may not have allowed loans to be collected when this imposed too great an economic burden on the lender. Since this

“insurance contract” was unwritten, it was not enforced by the Japanese courts which were set up after 1895. Thus, Japanese courts were much more useful to northerners whose loan contracts contained little insurance. The courts helped much less in the north since the gap between community enforcement practices and Japanese court enforcement led to uncertainty in the lending market and drove lenders to borrow from formal institutions where enforcement policy was clearer.

If the above hypothesis is true (and it needs much further testing) then the question remains as to why loan insurance was higher in the south than in the north. Two possible answers are: (1) Mortality and morbidity rates were much higher in the south than in the north thus there was a greater demand for insurance during the late-Qing and early-Japanese eras. Court records show that southerners were more reluctant than northerners to go to court to force descendents of debtors to repay loans. (2) Communities in the south tended to be older and contained fewer new immigrants than in the north so they may have had greater solidarity and the dynamics of loan collection were different. If (1) were true, it would mean that, at least initially, the

high risk premium was an efficient response to the environment. If (2) were true than it is more likely that debtors were hobbled by an inability to commit which made markets less efficient.

Below, I briefly describe our methodology and results. Then I discuss these results and conclude by explaining how I will proceed with this research.

Methodology and Results

Originally, my proposal was for a two-year project. I have had sufficient time to examine the northern Taichung data which was in my hands last years, but I did not gain possession of the Chiayi data to which I was to compare the Taichung data until this autumn. It took me until the first of December to get the data translated and put into a data base. Thus these results are still preliminary.

The court records used were from arbitration courts that handled civil suits. They were the first court of redress and their decisions could be appealed. We translated Taichung data for several periods beginning in 1896. The Taichung court district included present-day Taichung City, Taichung County, Nantou County, Changhua County and parts of Miaoli County. Since we are interested in north-south

differences, and previous research indicated that Changhua and Nantou County were border cases which had some southern characteristics, we only looked at court cases in which the defendant resided in present-day Taichung County or Taichung City. We have only had time to process three years of Chiayi data—1930, 1931 and 1932. Furthermore, we only looked at the court records in which the defendants resided in modern-day Chiayi City or Chiayi County. All comparisons, therefore, are still from this period and these places only.

During the three years examined, there were 263 court cases relating to debt between individuals in Taichung (city and county) and 303 cases in Chiayi (city and county). Of these cases, 177 cases in Taichung were useable and 212 cases in Chiayi. The other cases could not be used either because they were short and cryptic containing little information or (more often) the court clerk’s handwriting was illegible. I found that there were some significant differences between urban residents of Taichung City and Chiayi City (using the old 1930s boundaries) so I looked at urban and rural data separately. There were 114 legible cases in rural Taichung (plus 63 illegible cases), 154 legible cases in rural Chiayi (plus 66 illegible cases), 63 legible cases in urban Taichung (plus 23 illegible cases) and 58 legible cases in urban Chiayi (plus 25 illegible cases).

The court records consisted of brief reports, usually three to five handwritten pages, stating who was involved in the case, what was being disputed and what the judge

decided. Sometimes it is further noted whether there were witnesses present,

whether written evidence was presented and whether the defendant admitted to owing money. To evaluate the data, I hired assistants whose Japanese was superior to my own. We read a sampling of the cases together and I made a list of information I wanted and could be found in most of the records. The assistants would read the cases and collect all the information they could. I spot checked the results and made enquiries when I thought information might have been overlooked or misread. Some of the records we used contained sentences we could not read due to the handwriting, but I am confident that the data base compiled does not contain too many mistakes. The table shows descriptive statistics for Taichung and Chiayi. The average interest rate statistic does not include interest-free loans in the calculation. Short-term loans are defined as loans due in two months or less. Long-term loans are loans not due for at least 13 months. Cross-generational disputes are those in which one party to the dispute is a descendant of the borrower, lender or guarantor.

These statistics are only for problem loans and are probably not representative of loans in general, but the much higher interest rate in the Chiayi area, particularly in the countryside, confirms that the risk premium in southern areas was higher than in the north. A common feature of both areas is that loan disputes were not brought to court unless the lender was fairly certain he would win the dispute. In the Taichung area, 84% of loan disputes were decided completely in the lenders favor, while 89% of Chiayi disputes were settled thus. Besides the higher rate of interest in Chiayi, the most salient differences between the two areas appear to be: (1) In Chiayi, people are much more likely to borrow from people very close to themselves, either those living at the same “address” (所) or from those with the same surname as themselves. (2) Loans in dispute in Chiayi tended to be of much smaller value than loans in

Taichung. (3) In Taichung, disputants were more likely to bring in witnesses or written documents as evidence. It is questionable, however, whether the lack of witnesses and written evidence in Chiayi was due to poorer record keeping in Chiayi or reflects what actually happened during the court dispute. Our impression is that Chiayi records contained less information concerning the trials than most Taichung records. One further important difference is that Chiayi residents were less likely to use the Japanese courts in a loan dispute. We don’t know for sure how many bad loans there were, but given the amount of borrowing shown by the 33rd volume of the 農業基本調查 and the size of the risk premiums a rough calculation would suggest that bad loan’s chance of being arbitrated in a Japanese court was probably twice as great in Taichung as in Chiayi.

I am not including regression results in this report because I still don’t have any regression results I judge to be reliable. There is a negative correlation between loan size and interest rate, but this explains little of the regional difference. Loans with guarantors do not have lower interest rates, but this is not necessarily surprising since a lender may only insist on a guarantor when the loan is relatively risky. One important proximate cause of the southern risk premium is that southerners made less use of Japanese courts to enforce loans. As explained in the introduction, this is at least partly because the southern enforcement practices differed more from formal Japanese enforcement practices in that there was more insurance packaged with the loan, so cause and effect are difficult to untangle. I am not yet able to determine (1) to what extent this caused southerners to borrow from people close to themselves and (2) to what extent the fact that southerners borrowed from people close to themselves caused them to be more reluctant to go to court.

One way we are trying to solve this problem is to examine the distribution of the loan disputes within Taichung and Chiayi to see if this distribution reflects any other underlying economic features, e.g. irrigation, commercialization of agriculture, etc. If we find any suggestive patterns, we can expand the data set to the Nantou,

Changhua and Yunlin areas. Data from Taoyuan, Hsinchu and Miaoli should also become available within six months.

It is possible that more Showa-era data will allow us to determine between alternate models, but I have decided the most attractive approach now is to look at earlier Taisho-era data. We have begun work on a data base for the years 1914-1916. We earlier examined some of the Taichung data from this period. Interest rates on informal loans in Taichung were much higher in this earlier period and by examining other changes over time in the informal loan market we hope to gain a better

understanding of what determined the regional risk premiums.

We have found one other source of information that may throw light on the risk premium problem. This is the courthouse registers of various sales transactions. Many of these transactions also involved loans. This record of non-problem loans is interesting in itself in that it records the overall change in interest rates in the informal lending markets which most Taiwanese depended on throughout the Japanese-era. For this research, this new source of information will be most important in that it allows us to directly view simple loan contracts in the north and south to search for differences.

Table. Loan Dispute Statistics Characteristics Taichung County Chiayi County Taichung City Chiayi City Taichung Total Chiayi Total Disputes / 10,000 residents 3.5 5.5 15.4 10.1 4.8 6.2

Ave. Value of Loan in Dispute (Yen) 914 258 382 283 724 265 Value of Disputed Loans / 10,000 Residents (Yen) 3180 1463 5870 3121 3475 1640

Ave. Interest Rate (%) 12.4 19.9 16.8 18.4 13.9 19.6

Short-term Loans (%) 16 13 30 28 21 17

Long-term Loans (%) 34 26 22 10 30 22

Loans with Guarantors (%)

30 29 24 29 28 29

Loans with Collateral (%) 6 1 0 2 4 1 Cross-generational Disputes (%) 18 9 11 14 15 10 Loans sold (%) 9 1 2 2 6 1 Disputants Using Japanese Lawyers (%) 17 16 17 15 17 15 Disputants Using Taiwanese Lawyers (%) 17 11 18 11 18 11 Disputes Mentioning Documents (%) 37 20 38 24 37 21 Disputes Mentioning Witnesses (%) 35 18 32 12 34 17

Disputants with Same Surnames (%)

15 21 10 16 13 20

Disputants with Same Address (%) 11 32 6 21 10 29 Debtors Fully Exonerated (%) 13 8 6 2 11 7 Debtors Partially Exonerated (%) 6 4 3 5 5 4