Abstract
Following the failure of conventional economic models after the global financial crisis of 2008, economic researchers increasingly turned to agent-based modeling (ABM). One of the key features of this approach is its capacity to analyze the diverse behaviors and perspectives of economic agents, rather than assuming the behavior of a representative agent to be completely rational. Instead, ABM replaces this assumption with bounded rationality. Drawing on this framework, the present study provides  a more accurate analysis of exchange rate fluctuations in the Iranian economy by incorporating the heterogeneous expectations of agents regarding exchange rate behavior and by employing both the Smooth Transition Regression (STR) model and the Markov-Switching (MS) model. Since economic dynamics are not determined by a single observable variable, endogenous changes are captured through the MS model in the associated probabilities. Unlike the STR model, the economy in the MS framework experiences three regimes. The first and second regimes, similar to the STR model, represent periods of currency repression and currency jumps, respectively, while the third regime corresponds to turmoil. This turmoil regime characterizes situations in which the exchange rate is traded at exceptionally high levels due to heightened uncertainty, speculation, and limited central bank intervention, resulting in sharp fluctuations. Moreover, the presence of currency repression and the substantial gap between the market exchange rate and the fundamental exchange rate—conditions that foster exchange rate instability and turmoil—appears to contribute to the formation of heterogeneous expectations among market participants
Aim and Introduction:
In economies such as Iran, where the exchange rate serves as a nominal anchor, fluctuations in this variable can be a significant source of macroeconomic instability. Following the limitations of conventional economic models revealed by the 2008 global financial crisis, researchers have increasingly focused on ABM. A notable feature of ABM is its ability to analyze diverse behaviors and perspectives of economic agents, replacing the representative agent assumption and complete rationality with bounded rationality.
According to the literature, the present study seeks to provide a more accurate understanding of the dynamics of exchange rate fluctuations in the Iranian economy by incorporating the heterogeneous expectations of agents concerning exchange rate behavior.
Methodology:
This paper aims to investigate the efficiency of the regime-switching model in the Iranian exchange rate market under the framework of ABM with heterogeneous expectations.
In the first stage, we estimate the real exchange rate series for the period from 1388:1 to 1400:12 (March 2009 to March 2022). This series is calculated using the monetary approach and the cointegration method. After applying the most relevant tests to determine the most robust model specification, the results of the cointegration tests indicate that the ordinary least squares (OLS) method provides the most robust results.
Next, we use the gap between the estimated series and the unofficial exchange rate series published by the Central Bank of Iran to investigate exchange rate misalignment in the Iranian foreign exchange market. In the following step, we introduce three different agents with heterogeneous expectations that characterize the Iranian exchange rate market. These agents are divided into three categories: chartists, fundamentalists, and market makers. We illustrate the behavioral differences and strategies of these agents through appropriate equations.
We then apply three mechanisms. The first model assumes that the proportion of each agent type is a function of its relative past performance. The second model allows agents to switch their strategies based on macroeconomic fundamentals. The third model, proposed by Chiarella et al. (2012), assumes that agents’ beliefs depend on an MS process. Based on the monthly bilateral exchange rate between the Iranian rial (IRR) and the U.S. dollar (USD), we compare these three mechanisms empirically. 
This study employs both the STR model and the MS model. Since economic systems are not solely driven by a single observable variable, the endogenous changes in model dynamics are captured through the MS framework. Unlike the STR model, the MS model reveals three distinct regimes. The first and second regimes, similar to the STR framework, represent currency repression and currency jumps, respectively, whereas the third regime reflects turmoil. This regime denotes a situation in which the exchange rate trades at very high levels due to widespread uncertainty, speculation, and limited central bank intervention, leading to sharp fluctuations. The coexistence of currency repression and the large divergence between the market and fundamental exchange rates—factors that foster currency instability and turmoil—appears to play a critical role in shaping heterogeneous expectations.
Results and Findings:
Our results demonstrate that different types of agents exist in the Iranian exchange rate market. These agents hold heterogeneous behavioral expectations and act according to their own beliefs, employing distinct strategies that collectively influence market outcomes. Thus, under an ABM incorporating behavioral heterogeneity, excessive exchange rate volatility can be effectively explained.
Each of the three mechanisms applied to model agents’ decision-making dynamics proves efficient in explaining exchange rate volatility in the Iranian market. We then evaluate the performance of the three switching mechanisms in estimating behavioral heterogeneity. The behavioral heterogeneity model (BHM) switching mechanism highlights the importance of past performance in determining the dynamic weights of heterogeneous trading rules. This mechanism, which estimates market fractions of each agent based on recent realized returns and past strategy performance, performs particularly well in modeling the Iranian exchange market.
The STR model, which incorporates macroeconomic variables into the logistic function (with the real exchange rate ultimately selected as the best explanatory variable), reveals two distinct regimes for our monthly time series. The results show that this model possesses strong in-sample explanatory power.
On the other hand, since economic dynamics are not captured solely by observable variables, the MS model identifies endogenous regime changes within the transition probabilities. Unlike the STR model, the MS framework detects three regimes: currency repression, currency jumps, and turmoil. The turmoil regime corresponds to periods of extreme exchange rate volatility caused by uncertainty, speculation, and insufficient central bank intervention. Additionally, the pronounced gap between the market and fundamental exchange rates—accompanied by episodes of currency repression—appears to significantly contribute to the formation of heterogeneous expectations.