Macroeconomic variables and stock indices: an asymmetric evidence from quantile ARDL model

摘要

Purpose - This study examines the short-run and long-run impact of macroeconomic variables such as industrial production index, inflation, exchange rate, interest rate, foreign direct investment and trade balance, on KSE 100 index and sectorial stock indices under bearish, bullish and normal states of the stock market prices. Moreover, we take into account the effect of three crises observed from 2005 to 2009. Design/methodology/approach - This study uses quantile autoregressive distributed lag (QARDL) model for examining the short-run and long-run effect across various quantiles of the dependent variables and compare its' results standard autoregressive distributed lag (ARDL) model. Findings - ARDL estimates indicate that, in the long-run, industrial production index, trade balance and foreign direct investment significantly affect stock prices. These findings remain same when three crises have been taken into consideration. In addition, estimates from QARDL model indicate that, in the short-run, the effect of exchange rate, interest rate, consumer price index and foreign direct investment, varies across bearish, bullish and normal states of the overall stock prices. Moreover, the short-run findings for Auto Assembler, Cement, Commercial Banks sector are consistent with overall stock indices, whereas other sectors, such as, Oil and Gas and Power Generation and distribution are asymmetrically affected by all macroeconomic variables. In the long-run, the effect of all macro-variables varies across different states of the stock markets except industrial production index for Auto Assembler sector, Oil and Gas sector and composite index of KSE 100 index. Originality/value - We take into account the effect of three crises observed from 2005 to 2009 and also examine the macroeconomic effect across bullish, bearish and normal states of the sectorial stock indices and composite index of Pakistan stock exchange. Finally, we use novel approach, called QARDL model, which has several advantages over other techniques.