A Markov‐switching analysis of Nigeria's business cycles: Are election cycles important?

Abstract

The study examines Nigeria's business cycles between October 1998 and October 2017 and ascertains the importance of general elections cycles in engendering cyclical fluctuations in different measures of business cycles. A framework based on political business cycles theory was estimated with a dynamic Markov‐switching regression technique. The study finds that election cycles are adequate in predicting cycles in food prices, non‐farm prices, exports, and imports in Nigeria while a significant effect of election cycles on the stock market, general price level, and exchange rate could not be established. The study concludes that cycles in food, non‐farm prices, imports, and exports can be predicted by future general elections while re‐election seeking behaviour of politicians lacks the power to influence stock market performance and exchange rate in Nigeria. Hence, artificial business cycles that result primarily from politicians manipulating certain fiscal tools targeted at stimulating the economy only to increase the re‐election chances could be minimized if monetary and fiscal institutions are strong, effective, and truly independent. This will ensure that policies are not manipulated between elections by politicians but are well targeted at achieving a set of long‐term developmental goals.

Efficiency of Tanzania's foreign exchange market

“…The Markov switching approach is envisaged to take into account nonlinearity issues in time series data such as structural break, regime shifts, and volatility, which are typical in the foreign exchange market. Indeed, the choice of the Markov switching model fits well with random walk behavior because it assumes that movement between regimes is unrelated to the past observation of the process (see, e.g., Ayodeji, 2017; Olakojo, 2020). Studies by Bergman and Hansson (2005), Corporale and Spagnolo (2004), Engel (1994), and Kirikos (2000) have shown that Markov switching models provide better in‐sample and out‐of‐sample forecasts than the random walk specification.…”