Taiwanese residential mortgage-backed securitisation

Raj Shourie and Stephen Au
Deutsche Bank

In June 2002 the Financial Asset Securitisation Law (‘FASL’) was passed in the Legislative Yuan, providing the legal foundation for financial institutions to issue securities backed by mortgages and other financial assets. In March 2004, First Commercial Bank (‘FCB’) successfully closed the first residential mortgage-backed securitisation (‘RMBS’) in Taiwan.

Since the passage of the FASL, there have been four transactions backed by residential mortgages to date. As in most other markets, residential mortgages comprise the largest class of assets that are eligible for securitisation in Taiwan. For the Taiwanese banks and other financial institutions, the potential in the residential mortgage sector for securitisation is vast. And mortgage-backed securities may potentially overtake the corporate debt market to become the largest non-governmental sector in Taiwan’s fixed income market, as is the case in the US.

Against this backdrop, First Commercial Bank brought the first RMBS deal to the market in March 2004, which was arranged by Deutsche Bank. The (First Commercial Bank 2003 special-purpose trust (‘SPT’)) transaction was backed by residential mortgage loans in excess of NT$4.57 billion (Taiwanese dollar). As the first residential mortgage-backed securitisation in Taiwan, this transaction was viewed as a milestone that epitomised the successful effort made over the past couple of years to establish a mature securitisation market. This transaction set the benchmark for similar deals in the future.

Taiwanese mortgage market

Unlike in other Asian economies such as Hong Kong, South Korea and Singapore, Taiwan’s government has taken a largely laissez-faire approach towards the housing industry. There are only a small number of housing units provided by the government. By one estimate, the private sector provides 95 per cent of all housing in Taiwan (William D H Li, Housing in Taiwan: Agency and Structure? (1998)).

However, Taiwan’s home ownership rate is among the highest in the world. At 82.5 per cent, it is second only to Singapore’s 92.3 per cent among all major Asia-Pacific economies. Even in large metropolitan areas, where typically more habitants may choose to rent, home ownership rates are still very high. For example, the home ownership rate in the municipality of Taipei is 75.7 per cent. Figure 1 compares the home ownership rate in Taiwan to that of six other major Asia-Pacific economies.

Figure 1

To a large extent, Taiwan’s mortgage origination market is built on a retail structure. Banks originate mortgage loans through their branches. The wholesale and correspondent types of origination, which are quite common in other countries such as the US and Australia, are almost non-existent. (In general, wholesale originations refer to loans originated under a third party’s name (and often underwriting guidelines as well) other than the destined holder, while correspondent originations refer to loans originated by a third party under the eventual holder’s name (and guidelines)). One may argue that one implication of retail origination is that it gives the bank more control over the credit quality of a given loan.

Taiwanese banks control the credit quality of mortgages chiefly through three factors. First and foremost, great emphasis is put on the valuation of the property. Banks usually have a dedicated department that verifies the accuracy of the property survey and the validity of the title, and appraises the property. A 20 per cent-to-30 per cent equity level is then usually required for the loan. This is meant to provide a secure cushion against property price decreases and expenses (if the loan defaults).

Second, background checks are performed on the borrower’s credit history. There are two major agencies, both established under the auspices of the ministry of finance. The Joint Credit Information Center acts as the centralised credit bureau, and the Bill Exchange Center is the centralised check-clearing agency. The latter can determine whether the borrower has any history of returned checks. In addition, a background check on the borrower’s identity is also performed through the government’s Domicile Information Services Center to uncover any criminal (especially fraud-related) history.

Third, the borrower’s ability to repay the loan will also be evaluated, using income reported by the borrower. Accepted proofs of income include pay stubs and tax returns. However, most banks do not rely heavily on the income document in loan origination. Only pay stubs from government agencies, public educational institutions and large corporations are viewed as credible proof of income. Therefore, the debt-to-income ratio is used only as a reference in making the decision to grant a loan. Banks are more often willing to accept a third party to provide a guarantee on the loan, if the guarantor has enough assets to qualify.

Servicing of the loan is usually handled by the same bank that has originated the loan. The whole loan market has only recently developed, and is largely limited to non-performing loans. There are no independent mortgage servicing companies.

Most banks have direct debit systems for loan collection. Borrowers often have, or are required to have, savings or checking accounts at the bank from which the periodic debit is made to pay the loan. If a loan fails to be paid as scheduled, collection efforts will be made. Loan delinquency management may be centralised or handled by branches. There is usually a centralised, dedicated group in each bank to handle defaulted loans.

The Central Bank of China (Taiwan’s central bank) defines a loan as delinquent if it is more than 90 days past due, and such loan becomes non-performing when it is 180 days past due. Foreclosures are generally sought after loans become more than 270 days past due.

The foreclosure process is somewhat more complex than the judicial foreclosure process in the US. The borrower is allowed to defend against the foreclosure action in court. If he fails in his defence, the court will set up an auction for the property. Although the lender who holds the mortgage initiates the foreclosure process, he cannot assume title of the property unless he wins the auction on equal footing with other bidders. In such auction, the court sets a price floor for the property. If no one bids above the floor, the court will set another auction in which the price floor is reduced, usually by 20 per cent. The whole judicial process for foreclosure can last 18 months. If one considers that usually the judicial process does not start until the loan is 270 days past due, it could be more than two years from the day a loan becomes past due before the loan is resolved via foreclosure.


The transaction faced a number of challenges and it would be relevant to examine these in detail.


In Taiwan, borrowers can exercise their set-off rights at anytime. By crystallising the set-off amount at closing through a second public announcement which was not required by the FASL, FCB was able to limit their exposure to set-off risk.

Table 1

Before the enforcing rule of the FASL was effective, the ‘maximum-amount mortgage’ caused some delay because there was no practice regarding such mortgages in Taiwan at the time. By law, the maximum-amount mortgage was converted into the fixed-amount mortgage without needing the obligor’s consent. After the application for entrustment of the maximum-amount mortgage was filed with the land office, the land office issued a new mortgage certificate evidencing the fixed-amount mortgage.

Since block registration for mortgages was not possible, FCB needed to re-register each of the mortgages with the respective land office.


The basis risks between assets and liabilities are normally mitigated through a basis swap. Because of cashflow uncertainty and the high cost of such a swap, Deutsche Bank structured the first floating-rate adjustable rate mortgage-based (ARM-based) rated certificates without the need of a swap. Instead, the basis risks were sized into the subordination. The cashflow analysis took into account the risk of a mismatch in interest rates on certain types of mortgage loans in the asset pool and those on the rated certificates. Furthermore, different prepayment assumptions were tested to determine the cashflow impact. The analysis also factored in stressed delinquency for the portfolio as well as a stressed recovery period of 24 months.

One of FCB’s objectives was to minimise the subordination size while maintaining flexibility to adjust interest margin to desired customers to maintain customer relationships. As a result, the deal structure was specifically designed to address these concerns. FCB has the flexibility to reset the interest margin of the underlying loans up to 25 per cent of the portfolio, subject to the minimum interest margin of 1.1 per cent. The establishment of a reserve account and early amortisation triggers provided comfort to investors and reduced the subordination size. The cash reserve was set up to mitigate commingling risk, set-off risk, servicer transaction risk and to cover temporary interest payment shortfalls on the rated certificates. The early amortisation triggers will accelerate the repayment of principal to investors if a trigger event or deterioration in the quality of the mortgage portfolio occurs.


At the beginning of the transaction, the regulators were not very familiar with securitisation. Time and resources were devoted to educate the regulators on the subject and, in particular, on the trustee’s different liabilities in a special-purpose trust and a special-purpose company (‘SPC’) and on a subordinated certificate-holder’s veto rights over senior certificate-holders’ decisions.

Although the FASL allows both the SPT and SPC approaches, many Taiwanese securitisation transactions took the SPT approach. This leads to trustee liability issues; thus where the originator is lowly rated, acceptance by trustees of the SPT approach may ultimately be limited. Issuers should be encouraged to adopt the SPC approach. However, this approach requires various steps to set up a company, which may be problematic as the company tax laws for securitisation are unclear. The enforcing rules of the FASL finally came out in August 2003. These made multi-tranching possible. Deutsche Bank also successfully applied to the Central Bank of China to treat the trust certificates as possible liquidity reserves for banks, which increased the potential investor base.

Under the current rules, a deal must close within one month of receiving the Bureau of Monetary Affairs’ (‘BOMA’) approval without getting an extension. However, there was a time gap between BOMA’s approval and the tax ruling. Rating agencies generally require a tax ruling as a condition precedent for closing, but the timing of receiving the tax ruling was uncertain. In addition, different tax officers/advisers had different views as to the tax implications of the asset-backed securities (‘ABS’). This meant that it took relatively longer to apply for a tax ruling to resolve unclear tax issues.

BOMA’s notification to the public about the approval of securitisation and specific deal terms immediately after approval raised some issues. The notice put pressure on competing issuers and advisers, and released sensitive structural information into the marketplace when details were still being finalised. The investing public’s understanding of ABS/MBS products was still in its infancy at this stage. Although rating agencies applied the same criteria to all transactions, the results were very different due to the differences in structures, underlying assets and flexibilities, etc. Nevertheless, issuers tended to compare credit enhancement and coupon levels with completely different structures.

Underlying collateral

Before looking at the transaction structure, let us examine the underlying collateral. This consisted mostly of ARMs. FCB introduced its ARM program in late 2002. An ARM has its interest rate tied to an index. In FCB’s case, the index is the average of the one-year fixed deposit rates of seven major banks published by the central bank. Compared with traditional mortgage products offered by Taiwanese banks, which carry discretionary interest rates, this product has the apparent advantage of transparency. Since the introduction of the product, the bank has given its borrowers who carry traditional prime mortgages a one-time option to convert their old loans into ARMs. As expected, many borrowers have taken advantage of this opportunity to do so. The other two types of mortgages in the pool are prime mortgages and fixed-rate mortgages. Prime mortgages are the traditional type of home loans in Taiwan. Interest rates on these mortgages are discretionary, meaning they are subject to change at the lender’s discretion. Fixed-rate mortgages are those with fixed interest rates for one to three years, after which their rates becomes floating (like ARMs).

Transaction structure

First Commercial Bank was the originator and the servicer, while First Commercial Bank 2003 Special Purpose Trust was the issuer. BOMA of the ministry of finance approved the formation of the trust. Deutsche Bank acted as the arranger and the lead manager, as well as the trustee. Taiwan Ratings Corporation assigned ratings to the class A, class B and class C trust certificates. Lee and Li conducted legal due diligence to verify the legality of the underlying mortgage loan agreements. PricewaterhouseCoopers conducted financial due diligence to verify that the originator followed the agreed-upon procedures in its origination process, and provided the comfort letter and the opinion on true sale. KPMG provided the fairness opinion (or the valuation specialist report).

The trust issued four classes of certificates, with three of them rated tw (Taiwan Rating)AAA, twAA, and twA, respectively. There is also a reserve account to cover any liquidity shortfall involving potential servicer transition, the cost of potential servicing transition, and the set-off risk in addition to the amount being covered in the Class D tranche. The amount of the reserve account was funded in cash.

Figure 2

Distribution of the monthly cashflow followed the ‘senior-subordinate’ rule. Generally speaking, in each month, interest collected from the mortgage pool will be distributed (after taxes and various expenses are satisfied) to the bonds as interest payments to class A first, then class B, then class C, according to the rates specified by the bond coupons. Funds will then be used to fund the reserve account (if not fully funded) and pay the servicing fee (if FCB is servicing the loans). Excess interest is then distributed to the class D holder. The principal collection will be distributed strictly in sequential order. Namely, all principal collections are used to retire class A bonds first, then class B, then class C. There are also certain credit performance triggers built in to provide additional protection to bondholders. Currently, the trigger is set (indicatively) at the occurrence of any of the following events:

(a)    the average of the delinquency ratio in respect of the three preceding consecutive collection periods exceeds 2.5 per cent;

(b)    the average of the default ratio in respect of the three preceding consecutive collection periods exceeds 4 per cent;

(c)    the aggregate amount of realised losses (up to and including the relevant payment date) exceeds 50 per cent of the aggregate initial principal amounts of the class D trust certificates; or

(d)    there is failure by the originator or the servicer to perform or comply with any of its material obligations under the transaction documents, which failure is incapable of remedy or continues to be un-remedied for a period of 90 days.

If a trigger event occurs, early amortisation will take place, and interest payment to class D holders can be redirected to retire class A bonds first, then class B and class C. The subordinate nature of FCB’s servicing fee and the retention of class D by FCB will align the servicer’s and investors’ interests, since if the pool does not perform well, it will be the servicer who suffers first.

Table 2

Distribution highlights

Investors were educated by a thorough evaluation of the structure of, and credit and distribution in, the deal as well as by Taiwanese RMBS research (In November 2003, Deutsche Bank published Taiwanese RMBS research by Eugene Xu and Shirley Chu. Some of the information in this chapter was based upon this research.)

This increased their exposure and enhanced their understanding of these more complex investment structures. The transaction was oversubscribed with a diverse investor base, including bill finance companies, banks and insurance companies.

Table 3

Significance of the transaction

Since the FCB transaction was the first RMBS originated in Taiwan, it clearly set the benchmark for subsequent RMBS deals, such as, Taishin’s and Chang Hwa ‘s transactions. Deutsche Bank structured an ARM-based sequential-structure transaction, which naturally matched to the mortgage portfolio. It has become the standard model for Taiwanese mortgage deals. The success of this transaction has prompted other competing RMBS deals to market sooner rather than later.

In addition, this transaction added a new asset class to the Taiwanese capital markets. FCB was able to improve its risk management, trust administration and servicing capabilities. By executing this pioneering residential mortgage-backed securitisation, FCB has contributed to creating a more sophisticated securitisation market (through the education of both investors and regulators).

During the approval process for the issuance of the trust certificates, the regulators addressed and tackled novel issues challenges even as they sought to understand the introduction of a new securitisation structure to the Taiwanese market. This process has contributed invaluably to the development and refinement of laws and regulations pertaining to residential mortgage-backed securities in Taiwan.

In November 2003, Deutsche Bank published the Taiwanese RMBS research by Eugene Xu and Shirley Chu. Some of the information in this article was based upon this research.