在ACCA考試中,F(xiàn)2階段時(shí)間序列知識點(diǎn)一直都是比較重要且每年都會出現(xiàn)在試卷里,今天會計(jì)網(wǎng)就跟大家詳解這個知識點(diǎn)內(nèi)容。
01、時(shí)間序列的組成部分,及局限性
1.1)Time series can be broken down into 4 categories:
a) Trend.(趨勢)
Key words: underlying long-term movement
√ 根據(jù)當(dāng)下actual sales,通過數(shù)學(xué)計(jì)算得出大致銷售趨勢。
即:默認(rèn)現(xiàn)行趨勢在未來也適用。
b) Seasonal variation.(季節(jié)性變動)
Key words: short-term fluctuations
affect results at different times of the year
√ 在時(shí)間上不一定是按“季節(jié)”發(fā)生的偏差,可以是每天的或每周的有規(guī)律的偏差。
即:短期內(nèi)有規(guī)律的偏差都可歸為季節(jié)性變動。
c) Cyclical variation.(周期性變動)
Key words: longer time period
√ 相對于季節(jié)性變動,它是長期的有規(guī)律的變動,比如:經(jīng)濟(jì)周期。
d) Random variation. (隨機(jī)性變動)
Key words: unforeseen circumstances
√ 通常不可預(yù)見,比如:政變、戰(zhàn)爭。
由于是突發(fā)事件,所以在做預(yù)測時(shí)(forecasting)不考慮該元素。
Ps:在F2中做Time series計(jì)算題時(shí),不考慮Cyclical variation,到了P level才會涉及。
1.2)時(shí)間序列計(jì)算的局限性源于3個前提假設(shè)
a. a straight-line trend exists.
b. seasonal variations are constant.
c. what has happened in the past is a reliable guide to the future
因此,如果事件的發(fā)生帶有不可預(yù)見性,則不適用這種方法。
02、時(shí)間序列的組成部分——Trend計(jì)算
第1步:用moving average方法平滑actual sales units注意:偶數(shù)的時(shí)間跨度比奇數(shù)的時(shí)間跨度多一步計(jì)算!a) 當(dāng)時(shí)間跨度是奇數(shù)時(shí)(以3年時(shí)間跨度為例)
b) 當(dāng)時(shí)間跨度是偶數(shù)時(shí)(以4年時(shí)間跨度為例)
第2步:actual sales units被平滑后,根據(jù)high-low method 求出Trend表達(dá)式,即:Y=a + b XY(因變量)=Trend,X(自變量)=時(shí)間,一個X代表一個時(shí)間跨度注意:時(shí)間是自變量,trend是因變量!所以根據(jù)high-low method的計(jì)算原則,先選出時(shí)間(X)最大值和最小值,再找到對應(yīng)trend(Y)的值,求出表達(dá)值即可
以第1步中3年時(shí)間跨度為例:假設(shè)20X1在X軸上代表1,往后每一年都依次共一個數(shù)字代表,則20X5 在X軸上代表5通過X找到對應(yīng)Y的值,則得到兩個點(diǎn)(1,410);(5,470)求的Y=395+15X
辨析:圖中傾斜向上的直線就是trend表達(dá)式在坐標(biāo)軸上的體現(xiàn)。這條直線沒有具體的X范圍規(guī)定(即:時(shí)間序列計(jì)算的前提假設(shè),暗含現(xiàn)在的趨勢以后仍將繼續(xù))為了便于理解可以將時(shí)間序列分解成兩部分來看
1)當(dāng)下狀況(actual)
圖中藍(lán)色曲線是actual units,根據(jù)前面兩個步驟的計(jì)算得到藍(lán)色傾斜直線。即:現(xiàn)有狀況下的趨勢
2)預(yù)測未來(forecasting)
圖中黑色傾斜直線,是藍(lán)色傾斜直線的延長線,延續(xù)現(xiàn)有趨勢,代表未來趨勢再根據(jù)給出的藍(lán)色粗實(shí)線調(diào)節(jié)對應(yīng)季節(jié)性波動,得到黑色曲線 Forecasting figure
03、時(shí)間序列的組成部分——Trend + Seasonal variation計(jì)算
涉及2元素:Y= Forecasting figure T= TrendS= Seasonal variation上圖中藍(lán)色實(shí)線,是在求出的trend上調(diào)整季節(jié)性波動,最終得到Forecasting figure(Y)a) 加法模型的計(jì)算公式Y(jié) = T + S例子1: Based on the last 15 periods the underlying trend of sales is y = 345.12 – 1.35x. If the 16th period has a seasonal factor of –23.62, assuming an additive forecasting model, what is the forecast for that period, in whole units?解:將x=16代入y = 345.12 – 1.35x,得到y(tǒng)=323.52.這里的y,就是trend。所以在加法模型中對應(yīng)T=323.52又已知S=–23.62代入公式得到Y(jié)=323.52+(–23.62)=299.9
b) 乘法模型的計(jì)算公式Y(jié) = T * S例子2: The trend for monthly sales ($Y) is related to the month (t) by the equation Y = 1,500 – 3t where t = 0 in the first month of 20X8. What are the forecast sales (to the nearest dollar) for the first month of 20X9 if the seasonal component for that month is 0.92 using a multiplicative model?解:這里t=0對應(yīng) 20X8 month1,一年有12個月則20X9 month1,對應(yīng)t=12 (12+0=12) 代入 Y = 1,500 – 3t,求的Y=1464,這里的Y,就是trend。所以在乘法模型中對應(yīng)T=1464又已知S=0.92代入公式得到Y(jié)=1464*0.92=1346.88
c) 兩種模型下季節(jié)性變動求和以一年為一個周期,中間有n個季節(jié)性波動時(shí)間點(diǎn)1) 加法模型∑ Δ S=0即:一年內(nèi),n個季節(jié)性波動相加=0例子3:The multiplicative quarterly seasonal variations for the time series were as follows:
以一年為一個周期,中間有4個季節(jié)性波動時(shí)間點(diǎn),相加=00.82+1.41+?+(-1.09)=0求得?=-1.14 2) 乘法模型∑ Δ S=n即:一年內(nèi),n個季節(jié)性波動相加=n例子4:The multiplicative quarterly seasonal variations for the time series were as follows:
以一年為一個周期,中間有4個季節(jié)性波動時(shí)間點(diǎn),相加=40.82+1.41+?+1.09=4求得?=0.68Ps :在乘法模型中不可能出現(xiàn)某個季節(jié)性波動系數(shù)是負(fù)的的情況
04、Deseasonalization(去季節(jié)化因素)計(jì)算
“seasonally adjusted”是”Deseasonalization”的同義詞辨析:該知識點(diǎn)可以理解為是以上知識點(diǎn)的逆向思維,不涉及forecasting即:已知actual units,和去季節(jié)化因素后,得到當(dāng)下trend結(jié)合圖形理解:給出藍(lán)色曲線(actual units),和季節(jié)性波動系數(shù)(藍(lán)色粗實(shí)線),求藍(lán)色傾斜直線上的點(diǎn)
加法模型:T=Y-S乘法模型:T=Y/SY =the actual sales units S =Seasonal variationT =seasonally adjusted trend = current trend例子5:In January, the unemployment in Ruritania is 567,800. If the seasonal factor using an additive time series model is +90,100, what is the seasonally-adjusted level of unemployment (to the nearest whole number)?加法模型:T=Y-S =567,800-90,100=477,700
來源:ACCA學(xué)習(xí)幫