初探 Amihud 流動性指標對景氣循環之預測力

Abstract

本文呈現市場流動性與景氣循環之緊密關係。 當景氣較差時，市場參與者透過調整其投資組合而獲得較高流動性，有較高風險或較不流動的資產將首先被剔除於投資組合之外，此極所謂「爭取流動性」(“flightto liquidity”) 假說。奠基於「爭取流動性」假說，我們以動態機率單位模型 (dynamic probit model)與靜態機率單位模型 (staticprobit model) 進行分析。我們使用 Amihud (2002) 所提出的流動性指標 ILR，藉此捕捉市場的流動性; 我們使用美國景氣週期測定委員會(NBER) 所宣佈的景氣燈號，藉此認定經濟體的衰退期。由動態機率單位模型分析結果顯示，對於預測未來景氣衰退，流動性並無法貢獻顯著的預測力，景氣燈號的落後期才是真正重要的預測因子。然而，景氣燈號通常為事後宣佈，在做預測的當期並無法獲得，使得動態機率單位模型在現實中並不可行。考慮到資料在當期能否取得，我們引進靜態機率單位模型。在靜態模型中，流動性扮演了重要的角色，在樣本內外都改善了模型的預測力。同時我們也發現，若我們將流動性來源分為大公司與小公司，後者才是改善預測景氣衰退的主力。透過實證研究，我們發現 「爭取流動性」 假說可以在美國市場成立，並且，依據此假說，我們可以透過一個簡單的靜態機率單位模型，對未來景氣燈號進行預測。

In this paper, we discuss the relation between liquidity and the business cycle.There are some reasons that we should link liquidity with the business cycle. The“flight to liquidity” or “flight to quality” hypothesis might be the most important one. It is more likely that an investor shifts his portfolio from those stocks with lower liquidity or higher risk when the economy is in a recession.Instead of business cycle measures, we use the NBER recession signal as the proxy of real economy. We apply dynamic and static probit model to monthly liquidity proxy, because the NBER recession is not always available in current period. Besides we use Amihud (2002) ILR as our liquidity proxy.The dynamic probit model with liquidity added does not result in a significance improvement on prediction. No matter in sample or out of sample, the pure dynamic probit model performs well with itself history only. However, as we mentioned earlier, NBER recession signal is usually announced well until the recession has happened for months, leading to an infeasibility to implement dynamic model. On the other hand, liquidity plays an important role in static probit model. We find a robust positive relation between liquidity and recession. Adding liquidity as a predictor in the static model would significantly improve the forecast accuracy in th short horizon. This observation remains robust with comparison to other financial variables ( e.g. term spread, credit spread, market volatility, and excess return). We also find the existence of “size effect” that small firm is more sensitive to future recession. Our finding above is consistent with our claims of “flight to liquidity; that is, it is more likely for market participants to shift his portfolio from those stock with lower liquidity or higher risk in advance when the economy would be in a recession. The primary contribution of this paper is that liquidity proxy, Amihud’s liquidity proxy here, does provide useful information when predicting the future recession. We find that the liquidity proxy performs well especially in a short prediction horizon. Besides, the liquidity of small firms is more informative in prediction.