傳統上對社會整體貧窮狀況的測量皆是以貧窮率為指標。然此一指標的適當性自70 年代起已開始被學界廣泛的質疑。因為貧窮率僅能表達社會中民眾的貧窮風險發生率，但卻無法由此指標得知，貧窮人口到底有多窮、窮人所得分配的不均度等訊息。自Sen 以降許多研究者開始發展適當的貧窮指標，這些指標雖然仍在發展中，亦未有唯一受大家所公認最佳的指標存在。但多年發展之下，他們形成的共識是一個好的貧窮指標應至少包含三種訊息：貧窮的發生率 (H) 、貧窮的強度 (I) 與貧窮的不均度 (G)。這些指標所指涉的意涵，正可提供福利政策制訂所需的資訊。 本文以SST (Sen-Shorrocks-Thon ) 指標來測量台灣的貧窮趨勢。由1990 年開始至2002 年，台灣的貧窮趨勢呈現W 型曲線，1994 年與2000 年是W 型曲線的最低點，貧窮程度較低；而1990 年、1996 年與2001 年是W 型曲線的最高點，貧窮程度較高，尤其以2001 年的貧窮程度最高。SST 多元貧窮指標是由貧窮率、貧窮強度、貧窮不均度三指標所構成，經由線性分解後，貧窮率的變化較大、貧窮強度的變化較小，而貧窮不均度取對數後的數值接近一固定常數，由此發現貧窮率仍是一個較具影響力的指標，而貧窮不均度幾乎不影響SST 數值的變化。

Poverty rates have been used as the indicators in measuring the over all poverty conditions in a society in the past. However, the appropriateness of using poverty rates as the indicators began to be questioned by academic circles in the 70s.This is due in part to the fact that poverty rates can only reflect the extent of the incidences of the population in entering into poverty. It fails to show the severity of poverty and the differential distribution of poverty among others. Beginning with Amartya Sen, scholars and researchers have been hard at work in developing appropriate poverty indices. However, to this day, a universally acceptable measure has yet to bed eveloped. Over the years, however, some consensuses began to emerge. It is generally agreed by researchers that a good measure of poverty indicator should include three basic pieces of information: poverty incidences (H), intensity of poverty (I), and differential distribution of poverty (G). The dynamics and the interrelationships among these three variables can provide very useful information for developing social welfare policy in a society. SST (Sen-Shorrocks-Thon) indicators are used in this paper to measure the poverty trend in Taiwan for the period, from1990 to 2002. The trend shows a W shape with the years 1994 and 2000 respectively located at the lowest points on the curve, the lowest poverty levels, while the year 1990, 1996, and 2001 are on the peaks, the higher poverty levels. The year 2001 was the highest among these peaks. SST multi-indicators consist of poverty rate, intensity of poverty, and differential distribution of poverty. After linear decomposition, we found the “change on rate” to be larger; while the intensity became smaller, and the inequality approximated a stable constant. Therefore, we can say that the “poverty rate” is still the more influential indicator, and the “differential distribution of poverty” hardly affects the SST value.