A empirical study of short-term interest rate around its lower bound

我们根据Clarida等人（1998）提出的一个流行的政策反应函数，实证地研究了短期利率的动态。我们想要模拟名义短期利率的长期序列，其中包括在其下限附近的低利率时期。因此，我们将这一基本模型发展为Tobit模型，以表达短期利率在较高或较低水平上的动态变化。我们还准备了四个可选的一致估计器，三个gmme和一个MLE，来估计这个Tobit模型。使用一个已实现的日元短期利率的长期序列，其中包括其下界周围的足够观察值，我们通过每个估计量估计Tobit模型。我们发现gmme和MLE之间的残差存在相反的偏差。另外一些分析表明，这些偏差可能不是由下界的存在引起的。因此，我们对基本模型进行了怀疑，并检验了其中假设的一个假设，即中央银行将使利率趋于平滑。众所周知，很难从经验上对其进行检验。但现在我们有了一种新的方法来检验它利用利率的下限。事实上，使用日本数据的检验强烈否定了利率平滑假设。

We investigate the dynamics of short-term interest rate empirically based on a popular policy reaction function proposed by Clarida et al. (1998). We want to simulate long-term series of nominal short-term interest rate which includes periods of low interest rate around its lower bound. So we develop this base model to a Tobit model to express the dynamics of short-term rate at either higher or lower level. Also we prepare four alternative consistent estimators, three GMMEs and a MLE, to estimate this Tobit model. Using a realized long-term series of short-term interest rate of Japanese Yen which includes enough observations around its lower bound, we estimate our Tobit model by each estimator. And we find the opposite biases in residuals between GMMEs and MLE. Some additional analyses show that these biases might not be caused by existence of lower bound. So we suspect the base model and test a hypothesis assumed in it that a central bank would make interest rate smoothing. It has been known difficult to test it empirically. But now we have a new way to test it by exploiting the lower bound of interest rate. Indeed the interest rate smoothing hypothesis is rejected strongly by this test using Japanese data.