Evaluasi kemampuan prediksi model Box-Jenkins pada karakteristik laba tahunan

JUNAIDI, Dr. Jogiyanto HM., MBA

2004 | Tesis | S2 Akuntansi

Suatu peramalan bisnis, terutama peramalan laba dibutuhkan oleh hampir seluruh entitas ekonomi. Peramalan laba dapat digunakan sebagai dasar pengambilan keputusan, baik internal perusahaan, pasar modal maupun di dalam penelitian-penelitian. Terdapat beberapa model peramalan bisnis, misalnya model naïve, regresi, dan Box-Jenkins. Berdasarkan penelitian ternyata peramalan laba yang dibuat dengan metode Box-Jenkins, hasilnya lebih akurat dibandingkan dengan model naïve maupun regresi. Penelitian ini menguji secara empirik hipotesis bahwa tidak ada perbedaan tingkat keakuratan prediksi antara model Box-Jenkins (ARIMA) laba nondeflasi maupun model Box-Jenkins laba yang dideflasi dalam meramal laba di masa yang akan datang. Data diperoleh dari Indonesian Capital Market Directory tahun 1987-2003. Seri laba nondeflasi adalah laba yang dihasilkan (net income). Sedangkan seri laba deflasi adalah laba dibagi ekuitas (stockholders equity) untuk periode sebelumnya. Untuk menguji tingkat keakuratan kedua model, dengan membandingkan nilai adjusted R2 atau nilai root mean square error-nya. Berdasarkan analisis statistik ternyata tidak ada perbedaan yang signifikan, antara peramalan laba dengan laba non deflasi dan laba dideflasi ekuitas.

Business forecasting, mainly earnings forecast is required by almost all-economic entities. Earnings forecast can be used as a guideline to make decisions, internally in the firm, capital market as well as in researches. There are some business forecasting models, such as naïve, regression, and Box-Jenkins. Based on researches, earnings forecast made in Box-Jenkins method is more accurate then the other two models. This study examines empirically the hypothesis that there is no forecasting accuracy difference between Box-Jenkins (ARIMA) non-deflated earnings and deflated earnings in forecasting earnings in the future. The data are obtained form Indonesian Capital Market Directory 1987-2003. Non-deflated earnings series are the resulted net income, while deflated earnings series are net income by the previous period stockholder equities. To examine the accuracy of both models, the adjusted R2 or RMSE are compared in the two models. The statistical analysis shows that there is no significant accuracy difference between Box-Jenkins non-deflated earnings and deflated earnings in forecast earnings.

Kata Kunci : Laba Perusahaan,Peramalan,Model Box,Jenkins,forecasting, earnings, ARIMA