探討流動性及資訊成本在市場資金緊縮時對投資人的行為的影響─以美國金融股為例

Abstract

資金緊縮時，投資人趨吉避凶，會產生投資轉移的行為，而資產轉移可分為流動性轉移(flight to liquidity)及安全性轉移(flight to quality)兩種。本研究就2000年2013年共14年的美國上市櫃金融業股票做為研究標的，欲求其在資金緊縮時，投資人是否有投資轉移之現象，而轉移究竟是轉出或轉入。接著，再分析投資人進行資產轉移的原因，究竟是受股票流動性影響較多，還是受股票的品質影響較多。 本研究以Amihud（2002）建立的非流動性指標衡量股票流動性，泰德利差(TED spread)代表資金流動性，也就是資金緊縮程度，用各股每日的非流動性指標對當日的泰德利差逐年進行時迴歸分析，得出19132個迴歸係數，代表資產轉移的程度及方向，其中有52%顯著，而顯著的係數中，有73%的係數為正，印證了本研究「在資金緊縮時，投資人對美國金融股有資產移出的現象」之預期。 接著，本研究以跟隨的分析師人數、分析師預測歧異度及分析師預測準確度三種指標組合成資訊成本指標，衡量股票品質。以資訊成本指標及美國金融股的非流動性指標這兩個自變數及其他控制變數對投資轉移作逐步OLS迴歸分析，並取標準化迴歸係數。發現流動性越小及資訊成本越大的股票，投資人資產轉出的程度越大。再把資訊成本及美國金融股的流動性的標準化迴歸係數相比，發現資金緊縮時，投資轉移的原因，大多是受流動性轉移影響。最後，本研究把資訊成本指標拆解，發現單獨看分析師預測歧異度對投資轉移的影響，即足以解釋資訊成本吸引的安全性轉移現象。

During times of tightness in the funding market, investors will avert risk. According the past researches, they will transfer their investment mainly to seek for the liquidity or quality. We call this behavior of investors “flight to liquidity’ and “flight to quality.”

This thesis took the American stocks in finance industry from year 2000 to year 2013 as researched samples, daily Amihud’s illiquidity indicator (Amihud (2002)) measured the liquidity of stocks as dependent variable, and daily TED spread to measure the tightness in the funding market as independent variable in the OLS regression model year by year. We find when the funding market is tight, investors will rearrange their funding to flow out the American stocks in finance industry.

This thesis also find ‘flight to liquidity’ is more powerful than “flight to quality in the American stocks in finance industry during times of tightness in the funding market by the regression analysis. The information cost indicator and the yearly Amihud’s illiquidity are independent variables and the coefficients in the first OLS regression model are the dependent variable. The information cost indicator measured the quality of the stocks is built from the analyst number, the dispersion and the accuracy of the estimation for stocks; the yearly Amihud’s illiquidity indicator is measured the liquidity of the stocks. Moreover, by disassembling the information cost indicator. We find the dispersion of the estimation for stocks is the most important, and it almost can replace the whole information cost indicator.