THRESHOLD OF FINANCIAL DEVELOPMENT AND CURRENCY DEVALUATION THAT STIMULATES STOCK MARKET RETURNS

How to cite this paper: Umoru, D., Ugbaka, M. A

In this study, we endeavored to establish a threshold of financial sector development (FSD) and exchange rate devaluation (EXD) that stimulates stock market returns (SR) based on an analysis of 25 stock exchanges in the Middle East and North Africa (MENA) region.Threshold generalized autoregressive conditional heteroskedasticity (GARCH) regressions were estimated.Only the Istanbul Gold Exchange stock return was found fit for GARCH volatility analysis.However, results from threshold regression revealed that EXD does have significant threshold effects on SR and when EXD falls lower than its threshold of 19.69 percent, FSD had an increased influence on SR by 11.8 percent.The effects of EXD and FSD on SR are greater when the FSD level is beyond the threshold value of 23.45 percent.FSD below thresholds of 23.5 percent, and 51.1 percent would be insignificant in predicting SR.Lagged SR within an economy below the FSD threshold of 50.59 percent will negatively affect SR.By and large, our results reveal that FSD cannot influence returns of stock on their exchange floors given the devaluation of local currencies beyond the threshold value of 19.69 percent.Future studies could extend our threshold regression framework to allow for endogenous threshold variables.

INTRODUCTION
Stock markets are the avenue for investments in different regions.Investors put capital into the markets hoping for returns on their investments (Iyoha et al., 2022).Returns are an offshoot of the overall performance of the stock market.Stock market returns (SR) are theoretical compensations of risks for investors.However, macroeconomic indices could influence returns because the stock market is a part of the financial sector of an economy.Amid these merits, Yartey (2008) states that the liquidity available in the stock market causes high information asymmetry and distortion of reports in a bid to discourage investors from short-term commitments that would have a downturn on the economy.As much as the stock market has been raised as a catalyst for economic growth and development, Okeya and Dare (2020) raise specific indices in the macro economy to assert direct or indirect influence on the market's operations.They opine that strategies relating to financial depth are fundamental to recording tremendous growth in stock markets both in developing and emerging economies.The availability of credit and other attributes of financial development would mean growth in economic activities as surplus units move resources to deficit units for investment.Financial development would also mean increased efficiency in the financial market comprising the stock market itself and other players.The study holds apriori expectation that financial development would improve market efficiency and cause stock prices to reflect information.In addition, Patro et al. (2014) raised that the incessant movements of capital and growth in international trade have positioned the exchange rate as a major determinant of profitability and stock prices.The exchange rate can influence the stock market because future net cash flows are subjected to exchange rate variations.In import-dependent economies like many economies in Africa, exchange rate devaluation (EXD) will mean more local currency units are needed to purchase imports than were needed before the devaluation.Zubair (2013) explains that importing firms in the period of devaluation will have lower stock prices from falling firm value.EXD may also affect stock prices in that devaluation will have an immediate change in firm value and profitability.Devaluation would affect a firm's day-to-day operations, which may, in turn, influence its share price.
Furthermore, African economies are generally characterised by the depreciating value of local currencies to the international standard currency, the United States (US) dollar.Differences, however, exist in the extent of decline and volatility recorded in currency value in respective countries.The constant fall in the value of local currencies implies that investments outside a country require more local currency units.Furthermore, investors may find investments in less volatile currencies more attractive than local investments or foreign investments in countries adversely affected by EXD.The preferential interest of investors in raised is attributable to perceived currency risks.This study seeks to provide evidence on how financial development and EXD will influence SR.More importantly, the study provides results on stock returns in low and high economic conditions.
The objective of this study was to ascertain the threshold of financial sector development (FSD) and EXD that can positively influence SR in the Middle East and North Africa (MENA) region.The relevant question becomes, how significant is FSD in predicting stock returns in stock markets of the MENA region given a designated EXD threshold?To what extent does EXD impact stock returns in MENA stock marked by a determined FSD threshold?The sample of countries whose stock exchange markets covered by the study includes: Lebanon, Bahrain, Malta, Iran, Israel, Jordan, Morocco, Kuwait, Qatar, Oman, Palestine, Saudi Arabia, Syria, Cyprus, Tunisia, United Arab Emirates, Türkiye, Iraq, Egypt, and Algeria.These are countries that belong to the same geographical region, the MENA.All countries in the MENA region are middle-income countries with emerging stock exchange markets.
Previous research focused attention on the relationships between the stock market and economic growth.Some of these researchers include Bello (2022) Firstly, the study provides empirical findings regarding the effects of FSD and EXD on SR, and also, the threshold of FSD and currency devaluation that can positively influence SR in the MENA region.By obtaining non-linear threshold impact as regards the contribution of devaluation and FSD on stock market return, the study is important to policymakers.Additionally, by thresholding the exogenous covariates we contributed empirics of the threshold regression framework upholding that SR in the MENA region perhaps behave differently when the values of FSD and devaluation of currencies exceed a certain threshold.In particular, the study establishes the significance of the threshold of FSD in predicting stock returns in stock markets of the MENA region given an empirically founded EXD threshold.In what follows, the study determines the threshold of EXD that favorably impacts stock returns in MENA stock given a determined FSD threshold.
Another significance of this paper is that we do not use the traditional quantity aggregate of money supply/gross domestic product (GDP) ratio as a proxy variable for FSD, rather, we utilized the ratio of securities market debts calculated as the overall total amount owed in short-term debt securities markets, bond and equity to broad money in circulation.With this ratio, we provided a direct structural measure of a country's level of financial development; we added to the empirical literature of monetization indicators as well as the balance between all financial intermediaries (including banks and financial houses) and securities markets in the financial system; and also, we provided a measure of savings in an economy which rises in response to enhanced price signaling that is signified through positive real interest rates.Accordingly, this paper becomes the first of empirical studies to have applied the ratio of securities market outstanding (SED) to broad money in circulation (BMC) as a measure of FSD.
The structure of this paper is as follows.Section 2 reviews the conceptual, theoretical, and empirical summaries of past works on financial development, EXD, and SR.These works are synchronized to fit study themes.Section 3 raises a theoretical framework backing up likely findings of the study's findings and data sources, study models, and analytical tools used.The next Section 4 contains descriptive statistics, unit root tests, co-integration tests and other inferential statistics specified in the methodology section.The section rounded off with policy implications of analytical outputs for a better connection with the real phenomenon.The last Section 5 contains clearly stated study results, recommendations and conclusions.

Theoretical literature review
The International Monetary Fund (IMF, 2018) defines FSD as a state in which financial instruments, financial intermediaries, and financial markets within an economy promote key financial sector functions by ensuring an easier interplay of information, transaction costs, and enforcement of defining policies and regulations.It is stated that financial development is pivotal to economic development through the pooling and allocation of capital, savings, and inflows of foreign investments.A country that scores high in financial development would have less poverty and inequality rates and record faster economic growth.The recorded growth is spurred by small and medium-scale enterprises' access to finance, which reduces unemployment rates and increases economic activities.Beyond efficient financial intermediation is the presence of superior policies that support growth in the financial market, whether money or capital markets (Yusuf et al., 2020).FSD has been measured using different measures as spelled out in the World Bank's global financial development database.The database measures financial development using stock-to-GDP ratios, financial depth, domestic credit to the private sector as a percentage of GDP, banking efficiency, and stability.The database measures financial development by the state financial markets and financial institutions in countries and regions worldwide.
The theoretical discussion centers on the McKinnon and Shaw (M-S) theory (McKinnon, 1973;Shaw, 1973) which highlights the role of FSD in the economic growth of all emerging economies.These hinge on the narrowing of diffusion in the social rate of return to existing and new investment (McKinnon, 1973) and the reduction of fragmentation in financial markets.According to market fragmentation produces negative returns on investment.According to M-S theory, positive real interest rates stimulate larger financial savings.Hence, with liberalization, investments with positive real interest rates would be undertaken, financial intermediation would rise, and there would be an escalation of monetization of transactions in the economy.Consequently, as the financial system develops with advanced stock or securities markets guidelines and regulations, the range of price diffusion drops to eliminate information asymmetry; and facilitate additional growth of the market over time into markets for financial instruments, such as derivatives (Beck et al., 2001).Nevertheless, whenever excessive transaction costs; and information asymmetry persist, real returns on investments are diffused.According to Lynch (1996), financial sector reforms/deregulation spawn demand for effective risk management, eradication of consolidated controls over prices, liberalization of exchange rates and nominal interest rates, as well as resource delivery that reduces price diffusion.This is achieved through a link of markets across space and time rather than some kind of equilibrating resource flows (Beck & Levine, 2004 Beck andLevine (2002, 2004), and Christopoulos and Tsionas (2004) among other researchers.All contributions disfavored financial sector liberalization.The traditional theory of exchange rates supports that exchange rates significantly affect stock prices because they affect the value of firms on the exchange floor, especially when these values in local currencies are converted to foreign currency bases.In turn, returns expected by investors in such firms get eroded as the value of market capitalization drops from devalued currencies.In another case, an exporting firm might increase firm value with devaluation, especially when inputs are locally sourced and outputs exported (Zubair, 2013).For Patro et al (2014), devaluation will significantly impact asset prices so much that the cash flows of corporate market participants and the value of future equity returns.Overall, this effect becomes aggregate in the stock market influencing overall stock market performance and returns.Economic theory also suggests exchange rates and stock prices share a causal relationship.

Empirical literature review
The empirical literature is subdivided into three namely, the literature between financial depth and the stock market, and the literature regarding exchange rate volatility and currency devaluation policy on stock markets.And finally, a tread of literature that focused on the impact of a pandemic on stock markets.On the first stride of literature, we provide the following laconic review.2013) have all expressed the exchange rate as a highly sensitive variable that predicts the direction and speed of economic activities.The sensitivity of exchange rates and their intoxicating movements or volatilities in international relations has made it a point of study.Furthermore, its stability (rise and fall) is stated to influence growth in reserves holdings, government spending, money demand, investment, and national output.Umoru, Effiong, Okpara et al. ( 2023) executed the Markov-regime switching estimations, on the nexus between oil and exchange rates markets and found significant coefficients of devaluation and high transition probabilities which negatively affected oil returns.Umoru, Effiong, Umar, Okpara, Ugbaka et al. ( 2023) executed the nonlinear autoregressive distributed lag (NARDL) methodology to unravel the reactions of market returns to changes in exchange rates and oil prices in emerging stock markets.The author established that for every 1% devaluation shock, returns contracted considerably by 1.015% and 2.191% for Egypt and Nigeria whereas, and rose in Tunisia, Morocco, and Tanzania by 0.118%, 0.176%, and 1.145%, respectively.Javangwe and Takawira (2022) using the autoregressive distributed lag (ARDL) model to analyze South African quarterly data from 1980Q1 to 2020Q4 stressed that exchange rate policies influence stock market performances and thus, make investments and portfolio managers continuously monitor these exchange rates.The study sought to achieve its objective of examining the relationship between exchange rate behavior and the stock market in the country.A long-term relationship was found between the variables though this relationship was found to be negative.In the short term, the relationship is positive.The research findings by Javangwe and Takawira (2022) and Jameel and Teng (2022) found that current market volatility had varying impacts on market returns in Sri Lanka.Nusair and Olson (2022) found that interconnection runs from stock markets to exchange rate markets in selected G7 countries.Daily data from twenty-seven countries were employed and particular attention was paid to periods of devaluation announcements.The results revealed that devaluation anticipations affect stock markets and cause significant negative abnormal returns even before devaluation occurs, continuing for 30 days.After a year, however, negative abnormal returns return to equilibrium and become positive.Zubair (2013) examined the causal relationship between the stock market index and exchange rates in Nigeria.Johansen's co-integration test and Granger causality were used on the 2001M1 to 2011M12 data.The results revealed the absence of causality from the exchange rate to the stock market index.Yousuf and Nilsson (2013) tested the impact of exchange rates on the performance of the Swedish stock market.The study employed the GARCH (1,1) model and Pearson's correlation to determine the spill-over effect and correlation between exchange rate movements and SR using data from 2003 to 2013.The study found a low correlation between exchange rates and stock returns.

METHODOLOGY AND DATA
This study examines trends of related information limited to FSD and exchange rate positions and how these influence stock returns.In particular, the study investigates how effects of financial development in predicting stock returns in stock markets of the MENA region given a threshold EXD, and also, the extent to which devaluation impacts stock returns in MENA stock following an established FSD threshold.Threshold models have wide applicability in different areas of economic analysis.The underlying modeling framework upholds that an economy may perhaps behave differently when the values of a given variable exceed a certain threshold.In other words, different econometric modeling could apply when values are below a threshold than when the same exceeds the threshold.Hence, threshold models follow the fundamental modeling structure of regime-switching models (RSM) where different models apply to different intervals of values of some variables (Qian et al., 2018, Atem et al., 2017).There are alternative methods for estimating threshold regressions.These comprise, instrumental variable estimation techniques such as the two-stage least squares or two-step GMM estimation of linear index threshold regression model with endogeneity, regime-switching regression method, methodology for estimating sample splitting threshold models, wave-length regression method, probit, logit, and normal panel least-squares regressions, conventional ordinary least squares (OLS) method that utilizes an iterative search procedure, in a resolve to minimize the sum of squares, frequency domain regression, Monte Carlo interquartile range estimation method, least square dummy variable method, maximum likelihood estimator, a bootstrap method for threshold interval, etc.The threshold autoregressive (TAR) estimation method model is estimated in this paper for the advantage of unfolding both the conditional variance and mean due to regimes as established by threshold parameters.The TAR model credited to Tong (1990) and Hansen (1996) is given by Eq. ( 1).

𝜎 = 𝜑 + 𝜃 𝜑 + 𝑏 𝜎 + 𝛾 𝑢 𝐷
where,    Going forward in this study, we evaluated the effect of the exogenous threshold variables, namely, FSD and EXD.FSD was measured by using the ratio of SED/BMC.We calculated SED as the totality of the amount owed in short-term debt securities markets, bonds, and equity.BMC was calculated as the sum of savings or short-term bank deposits, money-market deposits and fund shares, and debt securities with a 2-year maturity date.We calculated percentage devaluation with currencies of all the countries in our sample by subtracting the pre-devaluation exchange rate against the US dollar from the devalued exchange rate.That is, given a pre-devaluation exchange rate of 380 units of local currency, and a post-devaluation rate is 420 units of local currency, this gives a difference of 40 units of the local currency on the US dollar.
Dividing 40 by the pre-devaluation exchange rate, we divided 40 by 380 which is 0.105, indicating a 10.5% devaluation.The study thus holds the following apriori assumption: EXD will cause negative impacts on stock returns from eroded investors' confidence in the value of future returns.Financial development is also expected to hurt stock returns because access to finance will improve living standards and reduce the occurrences of abnormal stock returns as information asymmetry will reduce.Table 1 below provides a list of stock markets covered by the study.These are all emerging stock markets of countries with middle income in the same region.The table above contains the results of five unit-root tests used to determine the integration order of study variables.All tests showed stationarity at a level for EXD and SR at the 95% confidence interval.The Breitung test, however, had FSD stationary at first differencing.The presence of unit roots in study variables necessitated the cointegration test.Panel co-integration results as reported in Table 4  In threshold GARCH analysis, we reported the following ARCH effects in Table 5 as a prediagnostic test for GARCH family models.According to Table 5, all but the stock return of the Istanbul Gold Exchange could have volatility tested with the threshold GARCH.The results are presented in Table 5.Table 6 reports volatility estimated with threshold GARCH.The in the mean equation reveals the average stock return of the Istanbul Gold Exchange at 2.47%.The one-year lagged value of the stock return of the Istanbul Gold Exchange is found to predict its current value significantly in a direct relationship.For the variance equation, both the ARCH term (0.2307) and the GARCH term (0.8173) are significant.While the ARCH was significant at 5%, the GARCH term passed significance at 1% respectively.Asymmetric volatility measured by the threshold term was 0.9197.
The significance of the positive threshold coefficient portrays the presence of asymmetric behavior at the Istanbul stock exchange.In effect, the positive coefficient reveals that stock returns at the Istanbul Gold Exchange do respond differently to bad news.In other words, reaction returns when the good or bad news hits the Istanbul Gold Exchange differs considerably.There is also persistence in the volatility of stock returns at the Istanbul Gold Exchange (persistence > 1).In effect, there is the presence of volatility clustering.Table 7 below contains estimates of the threshold regression with EXD as a threshold variable in different estimated models.It expresses the existence or non-existence of a threshold relationship between EXD and SR as a result of influences of currency devaluation and financial development in the model.The static model was first estimated.Its peculiarity is that it takes out the lagged value of the dependent variable -SR.In this model, EXD had a threshold of 19.69 and was significant with a p-value of 0.0021 which is lesser than 0.05.Accordingly, SR is -134.4 when EXD and FSD are null as depicted by the constant term.Examining the variables with _b suffixes representing values in the first regime where EXD is less or equal to the threshold value.Lower levels of EXD strongly predicted SR as revealed by the coefficient of 3.399 (with a significant z-value) which is lesser than the threshold value of 19.69.This is also found to be significant.Within the EXD threshold, financial development is also a determinant of SR with a significant positive effect (11.779; p-value = 0.04).In the threshold relationship between EXD and SR, a percentage rise in financial development explains an 11.779 unit rise in stock returns.Beyond the EXD threshold limit, however, financial development and exchange rate are insignificant in the model.
In the dynamic model, the EXD threshold value is 13.109.Below the threshold of 13.109, past values of SR are indirect predictors of current SR with a coefficient of 0.620.This implies that a percentage increase in past value causes an increase of 0.620 in SR for the current period.However, this is not significant with a p-value greater than 0.05, i.e., 0.675.FSD shows a direct relationship of a 7.869% increase in SR for a percentage rise in FSD.This is found to be significant in the model (p < 0.05).Higher levels of EXD weakly predicted SR at 2.009 (with insignificant z-value).In the higher regime which is when EXD rises above 13.109, the past value of SR becomes a positive predictor of current SR with a coefficient of 10.14, and it is significant in the model.Coefficients of FSD and EXD are also seen to increase in magnitude to 14.369 and 13.647 respectively, though insignificance is maintained.When EXD however goes above 19.69, and 13.109 the effects of both variables weaken in the prediction of stock returns as we find no significance in higher boundary estimates (FSD_d and EXD_d).
The previous models assumed the complete efficiency of the economy.However, the reality is characterized by inefficiency, thus kinks are introduced into the model to account for such inefficiencies.The third model showed that at the kink or discontinued point in EXD, the threshold value for EXD is 12.242 with a p-value above 0.05.At kink, a unit change in EXD accounts for a 3.033 increase in SR before taking an opposite turn though the effect is insignificant t.On the prior side before the kink point, lagged SR value had a strong positive effect of 0.077 on SR (p < 0.05).FSD in this model significantly and negatively influenced the threshold relationship between EXD and SR.Also, unit rise in FSD will cause SR to weaken by 14.423 units.Exchange rate as an independent variable in the same vein negatively associates with SR with a coefficient of -2.058 but this is not found to be significant.
The last model had the threshold variable exempted from the model to reduce covariates showing that at the kink or discontinued point in EXD, the threshold value for EXD is -1.214 with a p-value above 0.05.At kink, a unit change in EXD accounts for a significant -3.738 in SR before the reversal in slope after kink.On the prior side before the kink point, lagged SR and FSD values had weak negative effects of -0.236 and -9.385, respectively, on SR (p > 0.05).FSD in this model significantly and positively influenced the threshold relationship between EXD and SR such that a percentage rise in FSD causes SR to rise by 14.423 percent.Exchange rate as an independent variable in the same vein positively associates with SR with a coefficient of 2.058 but this is not found to be significant.The significant slope value reveals a rapid change in slope as variations in exchange rate move from larger to smaller values.From the models, EXD does have significant threshold effects on SR and when EXD falls lower than its threshold of 19.69%, FSD had an increased influence on SR by 11.8%.Financial development was also used as a threshold variable to determine its threshold effects as shown in Table 8 below.The static model had FSD with a threshold value of 23.45% found to be significant (p < 0.05).When FSD is in the lower regime which is lower than or equal to 23.45%, FSD had a positive and significant influence of 31.466 on SR while EXD had an insignificant but positive influence of 0.055 on stock returns.In the high regions above the level of FSD threshold, FSD refused to switch to a negative sign and rather, maintained positivity with a significant effect of 39.16% on stock returns.In the same light, the magnitude of the effect of EXD increases to 0.579 and becomes significant.In other words, the effects of EXD and FSD on SR are greater when the FSD level is beyond the threshold value of 23.45%.
In the dynamic model, the threshold value rises to 51.05%.In this model, the lagged value of SR had a negative relationship of -1.509 with the present value of SR, but this is found be insignificant.FSD also has a positive and significant effect of 1.562 on stock returns, and EXD becomes significant with a coefficient of 0.085.Beyond threshold levels of FSD and EXD maintain positive values with larger magnitudes (19.815 and 11.329, respectively) than when FSD was below the threshold.However, lagged value of stock returns reverses and has a positive effect of 4.219 on stock returns as opposed to the negative effect of -1.509 in the low region of the threshold.
In the kink option highlighted in the previous section, the threshold value of FSD shrinks to 50.59% (p < 0.05) as a way of taking into account inefficiencies in the economy that could cause drastic changes in variation patterns.The kink slope value was -28.864 (p < 0.05) revealing that at the point of discontinuity in the relationship curve of FSD and SR, FSD would negatively affect SR by 28.864 units.The lagged value of stock returns in this model is a significant predictor of stock returns continuously (1.259, p < 0.05).FSD is seen to have a positive effect of 15.05 on SR while EXD had a positive effect of 0.109.Both variables were found to be significant predictors in the regression relationship.
When the threshold variable is not used as an independent variable, the threshold of FSD is 54.33% (p < 0.05).The regression kink is -10.307(p < 0.05) revealing that at the point of discontinuity in the smoothness of the relationship curve, SR is affected by a -10.307 percentage change for a percentage change in FSD, given the threshold.Stock returns in the previous lagged period significantly influence its current value positively (coefficient at 0.779).EXD exerts a significant negative predictor of current stock returns at -0.91.The results reveal that FSD had a significant threshold effect on SR.However, FSD below thresholds of 23.5%, and would be insignificant in predicting SR.Lagged SR within an economy below the FSD threshold of 50.59% will negatively affect SR.To ascertain the robustness of our threshold GARCH results, we carried out some structural VAR model analysis beginning with the determination of optimal lag length and found that our results are robust to alternative SVAR model specifications.

Table 9. VAR optimal lag selection
The above table specifies the optimal lag length as lag 2 as suggested by the FPE, AIC, SC, and HQ.All other tests specified lag 2 as the optimal lag.This indicates that the SVAR (2) specification is the parsimonious model for this study.The impulse response function graphs (Figure 2) for stock returns to innovations within itself reveal that shocks from previous stock returns would cause stock returns to go in a negative direction in the short run (till the third period).After third period, stock returns begin to return to equilibrium and maintain a stable rate.In other words, abnormal returns function only in the short term before the market dynamics adjust for normal stock returns.For changes in financial development, returns respond slightly in the second year and quickly journey back to equilibrium.However, stock returns do not react strongly to innovations as found in the weak rise in the second year revealing stability of stock returns when other things are held constant.
FSD maintains an equilibrium state even when occur in stock returns.In other words, movements in stock returns do not influence FSD.
This implies that increasing values of traded shares on African stock exchanges do not transform into financial development.In the same vein, falling stock returns would not influence the FSD in an economy ceteris paribus.FSD responds significantly to its contemporaneous shocks.The behavior of financial development is such that a return to equilibrium position is not envisaged.Rather, FSD variations have their effect in the long run.This effect is reversed when financial development reactions to innovations in EXD are examined.Shocks that emanate from EXD cause FSD to go on a downward sprawl till the third period and then remain constant without returning to previous values or equilibrium.For EXD, its impulse response to internal innovations is an initial negative response in the short run before stability in subsequent longrun periods.Responses to FSD are swift but only in the short run as a return to equilibrium begins in the short run.The response is at the first positive, before returning to equilibrium.In response to variations in SR, EXD rises in the second period and stabilizes to a place of equilibrium.Response of EXD to Shock3 Source: Authors' elaboration.
In terms of post-estimation SVAR Results, the absence of serial correlation at lag 2 (Table 10) which was the optimal length used for the VAR model confirmed the robustness of the results.Linking the threshold and SVAR results, the research findings are further discussed as follows.The lagged value for SR was found to be a significant predictor of current SR in an inverse direction within the FSD threshold regression kink model.The SVAR analysis supports the autoregressive significance as it revealed that stock returns were largely exogenous with SR revealing a negative response to shocks from variations from its lagged values.The threshold results, however, address specifics on the extent to which SR(-2) can be considered a significant variable in predicting SR, even amid economic inefficiencies.EXD had minimal influence on SR in the SVAR supporting the absence of significance in EXD estimates.The results were not significantly different even when threshold regions were set.FSD had significant threshold values about SR.Overall, the study takes up Model 3 as the most viable model in the threshold model options.Volatility estimations did not show the volatility of study indices or asymmetric effects.From structural VAR estimates, EXD is affected by financial development in the short run.However, EXD has a long-run and negative impact on FSD.In other words, when a currency falls increase, it weakens FSD in that economy, hence, the state of economies of many African states.To improve FSD, the government has to stimulate demand for local currencies to increase and reduce devaluation, which will in turn promote FSD.FSD positively influences stock returns in the short run but reverses to equilibrium after the second year.
Past values of stock returns were found to be the main predictor of stock returns.Therefore, investors can forecast portfolio returns with a level of precision from past occurrences on the exchange floors.It can also be deduced that African stock markets function largely independently of the local economies in which they exist.Nevertheless, in specific thresholds of macroeconomic indicators, past SR may not be viable forecasting variables for future returns.Past SR will be dependable at levels below 50.59% level of financial development as a threshold.Descriptive statistics showed that stock exchanges in Morocco, Namibia, Egypt, Tunisia, and South Africa, which were larger than thresholds, will have weak evidence of the values of stock values influencing current stock markets.stock exchanges with less financial development rates can have investors determine the direction in which stock returns will take from the behavior of past returns.EXD in threshold limits of 12.2%, did not affect the predictability power of past values of stock returns.
When EXD is below its threshold value of 19.96%, FSD hurts SR.However, at rates above 19.96%,FSD would be weak in predicting SR.From the descriptive statistics, the stock markets examined have EXD maximum values of over 20.6% revealing that FSD can only influence returns of stock on their exchange floors given 19.96% or above devaluation of local currencies.Past SR cannot also be used to predict present values in these markets.Overall, investments in African stock markets may be weakened by the low explanatory power of past levels of stock returns to predict expected returns because local currencies are constantly weakening against standard foreign currencies.However, the weakening exchange rates increase the predictability power of FSD indices in influencing stock returns.Therefore, the expansionary effect of devaluation will support the volatility persistence of stock returns through the level of FSD.

CONCLUSION
This study uses the SVAR model to determine responses of endogenous variables, financial development, EXD, and stock returns to innovative shocks.This study employed the SVAR, threshold GARCH, and threshold regressions to examine impulse responses, volatility, and threshold effects of specified variables.The study found that financial development has a significant and negative predictor of stock returns in African stock markets.In the specification of a 50.56% threshold for FSD, previous stock returns were found to significantly and negatively determine current returns, but financial development itself and EXD are insignificant.Higher devaluations would make stock returns unpredictable.By and large, our results reveal that financial development cannot influence returns of stock on their exchange floors given the devaluation of local currencies beyond the threshold value of 19.69%.Consequently, investments in the Türkiye Exchange and other exchanges with weaker than 23.5% financial development would cause more predictability of stock returns, which is against the theory of random walk.Within a specified threshold of FSD at 50.56%, past values of stock returns were the major determinant of current stock returns and were found to have a negative effect.We are recommending that structural policies should be formulated to encourage foreign investments to increase local currency demands needed to reduce devaluation rates.Given the fact that our study evaluated the effects of exogenous threshold variables, the estimations in this paper are limited by our assumption of a kink restriction at the threshold point.Future researchers should focus attention on the continuity or discontinuity of the threshold point to estimate the threshold parameter value of asymptotic normality using instrumental variable estimation techniques such as the two-step GMM estimation of linear index defined by endogenous threshold variables based on an inverse Mills ratio bias correction.
Attah-Botchwey et al. (2022) reported that the financial depth of sub-Saharan African countries positively impacted the SR of those countries.Okeya and Dare (2020) examined the relationship between financial deepening and the development of the Nigerian stock market.The study employed stationarity, co-integration, vector autoregression (VAR), and vector error correction model (VEC) models on relevant data.Study data included financial sector contribution to the GDP ratio, banking sector liquidity, and stock market capitalization from 1981 to 2019.Results revealed a long-run positive effect of financial deepening on the Nigerian stock market but significance was absent in the short run.Yusuf et al. (2020) studied the effect of financial deepening (a measure of financial development) on SR in Nigeria from 1985 to 2018.Financial deepening was measured by broad money supply and credit to the private sector.These proxies were analyzed against SR in Granger causality, VAR, and VEC models.Results confirmed that money supply is a positive and significant enhancer of SR in Nigeria, but credit to the private sector was insignificant.Unidirectional causality was also found from financial deepening to SR.Similarly, Alenoghena et al. (2014) uphold that financial depth positively influenced market returns in Nigeria.Asal (2012) explored the nonlinear effects of financial development on stock returns amongst other measures of the economic performance of the euroarea.The study employed the generalized method of moments (GMM) model on panels that consisted of eleven countries from the euro-area and five others from non-euro countries with data from 1989 to 2011.Financial development had volatility, and bank loans to private enterprises were proxies.The study, in addition, modeled the threshold effect of rising public debt.Results indicated a negative association between banking development and volatility of stock returns.The exchange rate refers to the units of a local currency required to purchase currencies of other countries.Studies by Dabor et al. (2023), Umoru, Effiong, Umar, Okpara, Iyaji et al. (2023), and Umoru ( Özbey et al. (2016) studied the Istanbul Stock Exchange and its response to the exchange rate valuation of the local currency, the Turkish lira-US dollar.Monthly frequency data from 2009M1 to 2015M11 were retrieved and subjected to generalized autoregressive conditional heteroskedasticity (GARCH) analysis.It was found that EXD had a positive effect on the risk of investments in the market and a negative one on expected returns.In other words, as the value of the lira falls to the dollar, the riskiness of securities rises, and investors in the stock exchange begins to expect less from their investments.Aside from the study of the positive research findings obtained by Özbey et al. (2016) as regards the effect of devaluation on the stock market, Korsah and Fosu (2016) investigated the influence of the depreciation of Ghana cedis on stock market capitalization.Quarterly data from 1990 to 2013 were used.The results revealed a negative relationship between exchange rates and stock market capitalization in the short and long run.Patro et al. (2014) studied the reactions of different stock markets to currency devaluations.

( 5 )
=  +  +   +  +   +  (6) where,  ,  ,  are (m × m) matrices,  -is a non-singular (m × m) matrix of contemporaneous coefficients.A variable definition is such that EXD stands for exchange rate devaluation, FSD is financial development, and SR is stock market returns in %.The global financial development database of the World Bank was the source of data on financial development.The study employed data from quarterly series from 1975Q1 to 2022Q4.

Figure 1 .
Figure 1.Volatility graph -stock returns at Istanbul Gold Exchange

Table 1 .
Stock markets in the MENA region

Table 4 .
below based on the panel Rho, ADF and PP-Fisher statistics indicate a co-integrating relationship among financial development, stock returns and EXD.Panel co-integration results

Table 6 .
Stock return Istanbul Gold Exchange

Table 7 .
Threshold estimations for exchange rate devaluation (EXD) Note: z-statistics in parentheses followed by p values.** Significance at 5% level.Source: Authors' estimation.

Table 8 .
Threshold estimations for financial sector development (FSD)

Table 10 .
VAR residual serial correlation LM test results