Scholarlink

<p>In this paper, panel VAR models are developed to generate earnings and price forecasts. Earnings play a significant role in asset valuation models, and stock price forecasts are crucial for investment decisions. Long time-series panel data from 1979 to 2019 is utilized to run five-by-five matrix panel VAR models and predict earnings and price forecasts for 2020 and 2021. This paper builds upon traditional time-series models to develop a panel VAR model that simultaneously forecasts earnings and prices. The panel VAR forecasting outperforms ARIMA model in price and earnings forecasts. The main reason for this superiority is that the ARIMA model is considered technical analysis, using past prices to predict future prices. On the other hand, the panel VAR model corresponds to models that are regarded as fundamental analysis, incorporating accounting and price variables to predict future performance of both accounting earnings and market prices. Panel VAR models are an extension of ARIMA model, incorporating more variables into the forecasting model. The results demonstrate that panel VAR models are more accurate and generally unbiased compared to analyst forecasts, which often exhibit significant optimistic bias. The current analysis extends the literature on earnings and price forecasts and provides insights into the bias in analyst forecasts.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Earnings and price forecast with ARIMA and panel VAR time-series models

  • 期刊论文

  • 发表日期：

    2026年2月6日

查看全文
  • Wen-Chi Yeh,
  • Bharat Sarath

摘要

In this paper, panel VAR models are developed to generate earnings and price forecasts. Earnings play a significant role in asset valuation models, and stock price forecasts are crucial for investment decisions. Long time-series panel data from 1979 to 2019 is utilized to run five-by-five matrix panel VAR models and predict earnings and price forecasts for 2020 and 2021. This paper builds upon traditional time-series models to develop a panel VAR model that simultaneously forecasts earnings and prices. The panel VAR forecasting outperforms ARIMA model in price and earnings forecasts. The main reason for this superiority is that the ARIMA model is considered technical analysis, using past prices to predict future prices. On the other hand, the panel VAR model corresponds to models that are regarded as fundamental analysis, incorporating accounting and price variables to predict future performance of both accounting earnings and market prices. Panel VAR models are an extension of ARIMA model, incorporating more variables into the forecasting model. The results demonstrate that panel VAR models are more accurate and generally unbiased compared to analyst forecasts, which often exhibit significant optimistic bias. The current analysis extends the literature on earnings and price forecasts and provides insights into the bias in analyst forecasts.