http://hdl.handle.net/123456789/1634 Sat, 04 Apr 2026 22:51:54 GMT 2026-04-04T22:51:54Z http://hdl.handle.net/123456789/1635 PREDICTING REGIME CHANGES IN NIGERIAN STOCK MARKET RETURN SERIES ADESIYAN, AMOS OLUFEMI Regime change is the tendency of the Stock Market Returns (SMRs) for global market to change their behaviour abruptly due to changes in financial regulations and policies. This behaviour has no exemption to emerging markets like Nigeria. In literature, Markov Chain Models (MCMs) have been used to capture the stylised behaviour in 2-state regime; namely low and high return states, which limit the forecasting ability of the stock returns. The aim of this work was to extend the 2-state MCMs to 3- and 4-states for an improved forecast performance. The MCM was employed to specify the state transition probability, P , limiting distribution,  , the expected returns,  and the occupancy times, M n( ) . The behaviour of the SMRs was classified into five scenarios comprising 2-state regime defined as low and high return regimes (scenario 1), 3-state regime based on Mean  1SD (Standard Deviation) classification (scenario 2), 3-state regime based on Quartiles (Q) classification (scenario 3), 4-state regime based on Mean  1SD classification (scenario 4) and 4-state regime based on Quartiles (Q) classification (scenario 5). Following the classification, scenario 2 and 3 were defined as low, medium and high returns while scenario 4 and 5 similarly were defined as strong-low, low, high and strong high returns. Index and Price data from All Share Index Return (ASIR), Dangote Cement Return (DANGCEMR) and Guaranty Trust Bank Return (GTBR) covering the period, 3 January 2006 to 29 June 2018, were used. The limiting distribution, lim n n P    , the expected return time, 1    and the occupancy time 0 ( ) (for n 0) n r r M n P     were obtained. The limiting distribution in days obtained for ASIR, DANGCEMR and GTBR, for each scenario were 4, 4, 4 for scenario 1; 15, 7, 8 for scenario 2; 8, 6, 7 for scenario 3; 15, 6, 9 for scenario 4 and 15, 6, 9 for scenario 5, respectively. The identified expected return time for the transition in days were also obtained for ASIR, DANGCEMR and GTBR, for each scenario as: 2, 2; 3, 1; 2, 2 for scenario 1; 716, 1, 716; 10, 1, 10; 11, 1, 9 for scenario 2; 4, 2, 4; 4, 2, 4; 4,2,4 for scenario 3; 778, 2, 2, 778; 10, 2, 5, 10; 11, 2. 3, 9 for scenario 4 and 4, 4, 4, 4; 4, 2, 2, 4; 4, 5, 3, 4 for scenario 5. The limiting distribution of the MCM iv obtained for scenario 1 was lower to that of scenarios 2 to 5 as the returns will transit into steady state at days above 6 as against 4 for scenario 1. Occupancy times obtained for scenarios 3 to 5 gave a lower time period, an indication of short occupancy time. The transition probabilities obtained for scenarios 2 to 5 identified the persistence in state returns. The 2-state regime was successfully extended to 3- and 4- state regimes respectively. The increase in the limiting and expected return times in days for scenario 3 and scenario 4 is good for an investor as it allows more room for investment before return to equilibrium. Fri, 01 Jan 2021 00:00:00 GMT http://hdl.handle.net/123456789/1635 2021-01-01T00:00:00Z