Empirical Results of Home Equity Bias Puzzle Hypothesis

國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

2.6 Empirical Results

The empirical data scrutinized in this study are panel data. We use Eviews software to analyze regression panel data that has the properties of both cross-sectional and time series. The increase in degrees of freedom can reduce a linear superposition degree, thereby increasing the efficiency of estimates, while at the same to enabling researchers to grasp the differences between individual companies and time series dynamics. It clarifies the relationship between variables and the implied meaning of the behavior, and reduces the possibility of estimation bias by the observation of Individual-Specific Error Term, Time-Specific Error Term, and Random Error Term.

Fixed effects panel data model allows us to control the characteristics of each company and to calculate the differences between individual companies to observe the dynamics of each company and the scenario when it is subject to external shocks (Nucci and Pozzolo, 2001). Asset management and hedge effectiveness may be influenced by unobservable characteristics, such as the effectiveness of the company's managers, thus affecting the value of the company. Therefore, this study references the study of Allayannis and Weston (2001) and uses the fixed effects panel data model in hopes of better analyzing the relationship between making substantial foreign investments and the return on investment while considering the characteristics of individual companies.

2.6.1 Empirical Results of Home Equity Bias Puzzle Hypothesis

Empirical model Equation (1) is used to review the allocation of capital assets of domestic companies and to see whether or not domestic companies display a greater preference to invest at home than do foreign companies. Table 2.10 shows the adjusted R² coefficient is 0.153563. The

β

₁coefficient is significantly greater than 0 at the 1% level. In other words, domestic-owned companies show a greater preference to invest abroad than do foreign-owned companies. Empirical results differ from what the literature suggests concerning the barriers to capital flows created by higher costs associated with transacting in foreign securities, withholding taxes, political risk, information asymmetries, and regulation. Our empirical results reject the Hypothesis H1, that is, the domestic-owned firms’ asset allocation preferences

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

give rise to an Equity Home Bias, which in turn leads to a more aggressive stance regarding international asset allocation than is seen among foreign-owned companies.

Table 2.10 Test Home Equity Bias Puzzle Hypothesis (pooled model)

Ratio

C

0.1932***

(16.9421)

DuO

0.0766***

(4.8473)

Adjusted R² 0.1536

Obs 125

***significance level 1%, ** significance level 5%, * significance level 10%, ( ) t-value

2.6.2 Empirical Results of the Relationship between Foreign Investment and Asset Allocation Performance

2.6.2.1 Test What Factors Affect Foreign Investment Decisions

The empirical equation (1) reveals that domestically owned companies show a preference to invest abroad. This conclusion differs from those of French & Poterba (1991), Cooper and Kaplanis (1994) and Tesar & Werner (1995), all of whom suggest that the Equity Home Bias is a factor. Therefore, this section will examine what factors affect the foreign investment decisions made by Taiwan life insurance industry.

Table 2.11 presents the Equation (2) regression results.

The pooled panel data model adjusted R2 is 0.338662 and the fixed effects panel data model adjusted R2 is 0.763656, indicating that the fixed effects panel data model has twice as much capability as the pooled panel data model to explain the phenomena. The fixed effects panel data model has a good degree of fitness and can be used to reduce the amount of information of individual companies in the

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

information aggregation process and thus decreasing estimation errors by a significant degree. On the other words, fixed effects panel data model can be used to retain each company’s ability to manage the asset and hedge efficiency, which corresponds with the research results of Nucci and Pozzolo (2001), who state that fixed effects panel data model can be used to effectively observe the characteristics of individual companies. In addition, the fixed effects panel data model is used to verify the impact on asset and risk management10 when the competent authorities supervise the ratio of investment in foreign assets and individual companies make their investments in foreign assets. The regression results are summarized as follows:

For all models

γ

¹(Asset) were significantly positive, this indicates that the larger the asset size is the higher the proportion of foreign investment will be. It is verified that the larger the asset size of Taiwan companies, the stronger the tendency to invest in foreign assets in order to overcome the problem of insufficient long-term capital instruments and reduce negative interest-rate problem. This empirical reflecting for reduce interest risk and improve lower interest rate higher premium affect new policy sales problem, the new policy sale strategy from traditional life insurance change to Bancassurance channel, aggressive promotion Investment-Linked Insurance and Interest Rate Sensitivity Policies let high growth premium revenue let asset allocate myth in foreign investment.

For all models,

γ

₂(DomR) returns on domestic investment instruments were negative, and

γ

₃(ForR), the greater the international investment allocation: the higher the return on foreign investment, the greater the international investment allocation. This finding supports Brennan and Cao (1997), which demonstrated that when domestic investors possess a cumulative information advantage over foreign

10 The FSC amended the Article 15 of “The Foreign Investment Management for the Insurance Industry” in 2008, which stipulates that an insurer that has set up measures to process and monitor the foreign investment-related transactions can invest 10% of its insurance funds in foreign investment if the decision is approved by the board of directors. However, if the foreign investment amount reaches the 25-45% limits, the case should be reviewed and approved by the competent authority, depending on the implementation of internal controls, risk management capabilities, whether a risk management department and its risk-control chief are set up, experience in foreign venture capital investment, and the proportion of risk capital.

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

investors regarding domestic market conditions, investors tend to purchase foreign assets in periods when the return on foreign assets is high and to sell when the return is low. This result—showing that the return on domestic of investments is lower than foreign instruments—helps explain why Taiwan life insurance firms have continuously transferred life enterprise funds abroad since the year 2000. Thus, foreign investment ratios have increased from 4.62% in 2000 to 32.17% in 2009.

Under the pooled panel data model,

γ

₄(FinR) the gains and losses on exchange rates are significantly negative, which means the currency risk will reduce the willingness to invest in foreign assets. Because there is an interest-rate spread between Taiwan and the United States, those adopting the hedging strategies are required to pay between 1.5% and 2% in hedging costs. As hedging costs reduce the return on foreign investment, there is a negative relationship between exchange rate risk and foreign investment. Under the fixed effects panel data model, which takes each company's characteristics into consideration, the

γ

₄coefficient is negative but insignificant. It is inferred that the companies adopt different hedging strategies.

Larger companies tend to have a better capability for managing exchange rate risks than do small- and medium-sized companies, thus leading to a negative

γ

4coefficient, albeit not at a significant level.

In summary, regression Equation (2) confirms that the scale of a company and the returns on foreign investment will have a significantly positive effect on foreign investment, while hedging costs and exchange rate fluctuations have a significant negative impact on foreign investment.

‧

Table 2.11 Test the Factors Influencing the Foreign Investment Ratio

Ratio

Pooled model Fixed effect

C

-0.405463***

Adjusted R² 0.338662 0.763656

Obs 124 124

1. ***significance level 1%, ** significance level 5%, * significance level 10%, ( ) t-value 2. The variable

FinR misses one sample. Also see footnote 8.

2.6.2.2 Empirical Results of the Relationship between Foreign Investment and Investment Performance

Regression Equation (3) examines the relationship between foreign investment and investment performance. The dependent variable is return on investment (Return).

The controlling variables are company size, the gains and losses on exchange rates, and the proxy variables of firms. Table 2.12 shows estimated results of regression (3).

Four models are estimated. The first three are conventional models that only consider the casual link from Ratio to Return while the last further controls the possible effect of two-way interdependence of Ratio and Return. To solve the endogeneity between the two variables, two-stage least square estimation is

‧

Table 2.12 Test the Relationship between Foreign Investment and Investment Performance

Return

(i) (ii) (iii) (iv)

Pooled model Pooled

model Fixed effect Two Stage Estimation

1. Instruments: c, log(Asset) t-1, log(Asset) t-2, FinR t-1 FinR t-2, Duo t-1

2. ***significance level 1%, ** significance level 5%, * significance level 10%, ( ) t-value 3. The variable FinR misses one sample. Also see footnote 8.

For all models,

α

¹(Ratio) the ratio of foreign investment coefficients are positive but insignificant. The empirical results indicate that there is no significant correlation between the foreign investment ratio and return on investment. The U.S.

sub-prime mortgage crisis in 2008 caused huge losses in security bonds, and may affect the empirical results. It is more important for the life insurance companies to

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

enhance their efficiency in the allocation of foreign investment instruments and risk management capabilities than to raise the proportion of foreign investment.

For all models,

α

₂(Asset) coefficients are negative; under the fixed effects panel data model at 1% level significant negative, it indicates that the amount of the assets available to a life insurance company exists in an inverse relationship with the rate of return on asset allocation. In other words, the larger the asset, the lower the utilization efficiency of funds. This suggests that life insurers should focus on enhancing investment performance in a manner that corresponds to increases in market share.

For all models,

α

₃(FinR) the rate of exchange gains and losses are positive, representing that the hedging strategy on exchange rate changes will enhance the rate of return on the overall asset allocation. The adoption of hedging strategies requires a company to pay between 1.5% and 2% hedging costs. As the coefficient is positive, it means that the life insurance industry is not 100% hedged. The exchange rate fluctuations bring positive benefits for those that hedge. Due to difficulties associated with data collection, we cannot distinguish whether the gains or losses on foreign exchange derivatives are for speculative or hedging purposes. The hedging strategies adopted by the life insurance industry are worthy of a follow-up study in the future.

Under the pooled panel data model

α

₄(DuO) the dummy variables of company nature coefficient is negative but not significant, indicating that the asset allocation efficiency of domestically owned companies is worse than that of foreign-owned companies. In order to avoid contamination due to the U.S. sub-prime mortgage crisis loss outlier sample affect empirical result. In the future, we will examine outlier impacts on the return on investment.

‧

New Taiwan dollar appreciated, causing huge losses for marked-to-market currency hedgers. During the second half of the year, the New Taiwan dollar devaluated, greatly reducing the effects of currency hedging. According to the Taiwan Stock Exchange Market Observation Post System and the annual financial reports published by individual companies, seven listed companies reported gains on foreign exchange in their annual reports, with the best-performing gaining NT$13.168 billion.

Additionally, many life insurance companies recognized huge losses caused by structured investment products in their foreign investment portfolios due to the U.S.

sub-prime mortgage crisis in 2008. To capture the effect of financial crisis on domestic-owned firms’ Return, we estimate the following regression:

, 1 , 2 , 3 , domestic-owned firms’ have suffered financial loss on foreign investment during 2008 financial crisis.

The empirical results are shown in Table 2.13. During 2008 financial crisis, domestic-owned firms’ return on foreign investment is significantly negative up to -7.18%, on average. On the other hand, the return of domestic-owned firms’ foreign investment during 2003~2007 is positive but insignificant. This suggests that the performance of foreign investment is poor even the economic situation leaves recession.

11 The exchange rate of NTD was 32.368 against one USD in January 2008 and appreciated to 30.407.

在文檔中 海外投資及避險策略與保險公司價值之探討 - 政大學術集成 (頁 39-46)