This paper examines market efficiency in the foreign exchange market in Tanzania. A nonlinear Markov switching model was used to analyse the Tanzanian shilling against the US dollar spanning from 2 January 2009 to 30 April 2020. The estimation period was 2 January 2009–28 May 2019, and the post-sample forecast period was 28 May 2019–30 April 2020. From the results of the two-state Markov switching model, we define State 1, which is more volatile, as the inefficient state and State 2, which is less volatile, as the efficient one. In the unconditional probabilities from the Markov switching model, the exchange rate has a 94% probability in the inefficient state, which indicates that 94% of all speculators fully extract information when predicting future exchange rates, while 30% of all investors do not; suggesting that the foreign exchange market undergoes conditions of efficiency and inefficiency. In addition, findings showed that the exchange rate is expected to spend approximately 18 days in regime 1 and 8 days in regime 2. The policy implication of these findings is that foreign exchange market participants should pay attention to regimes and underlying market conditions when designing and applying their investment strategies.

中文翻译：

---

坦桑尼亚外汇市场的效率

本文考察了坦桑尼亚外汇市场的市场效率。非线性马尔可夫转换模型用于分析2009年1月2日至2020年4月30日期间坦桑尼亚先令兑美元汇率。估计期为2009年1月2日-2019年5月28日，样本后预测期为2019年5月28日- 2020 年 4 月 30 日。根据二态马尔可夫切换模型的结果，我们将波动较大的状态 1 定义为低效状态，将波动较小的状态 2 定义为有效状态。在马尔可夫转换模型的无条件概率中，汇率有94%的概率处于低效状态，这说明94%的投机者在预测未来汇率时充分提取信息，而30%的投资者没有；表明外汇市场经历有效率和无效率的条件。此外，调查结果显示，汇率预计将在制度 1 中花费大约 18 天，在制度 2 中花费 8 天。这些调查结果的政策含义是外汇市场参与者在设计和设计时应注意制度和潜在市场条件。应用他们的投资策略。