Contrary to the theorists of the new classical economy, the new Keynesian economy, with contributions from [94] and [10], considers that there are imperfections in the markets associated with the information available to economic agents. This reality raises the problem of information asymmetry, which generates market failures.

In the neoclassical approach, the labor market is subject to the law of supply and demand for labor, with wages being the adjustment variable. However, in the functioning of this market, employers are victims of information asymmetry, particularly moral hazard regarding the productivity of their employees ([89]). Employees reduce their efforts and, consequently, their productivity if employers do not constantly monitor them ([89]). However, such monitoring generates costs for employers, and it is difficult to dismiss the least productive workers. These problems of hidden behavior can affect worker productivity and, consequently, TFP. In the credit market, too, information asymmetry exists and is a source of financial instability. Financial markets are imperfect, characterized by information asymmetry between lenders and borrowers ([94]). In these markets, the moral qualities of the actors can affect financial development. Information asymmetry in the credit market generates risks of non-repayment of credit, thus limiting bank financing ([96]) and, consequently, TFP.

Although the problem of information asymmetry arises in various markets, measuring it remains problematic. To date, there is no comprehensive, global measure of information asymmetry in the economy. However, we can approximate information asymmetry in the credit market using payment defaults. Thus, some studies on the subject have used the credit default rate for this purpose ([96]; [27]). In this research, we have focused our attention on this market.

There is research on the relationship between information asymmetry and TFP ([60]; [38]; [19]; [100]; [28]; [36]). According to this literature, information asymmetry is likely to affect TFP through several mechanisms. It can harm investment and, consequently, affect economic growth ([36]). It can also harm financial intermediation and economic growth ([19]; [100]; [28]). Indeed, information asymmetry between lenders and borrowers affects financial intermediation. However, this intermediation can promote economic growth by reducing management costs and channeling funds to the most efficient users ([28]; [100]). Similarly, information issues influence productivity ([38]). More recently, [60] found a direct negative effect and an indirect negative effect of bank credit risk on economic growth in the WAEMU. Based on all this research, information asymmetry is likely to have direct or indirect negative long-term effects on TFP in the WAEMU.

According to [74], institutions are formal and informal constraints and rules that make it possible to monitor and limit the actions of individuals. Our approach to governance fits into this framework and therefore includes several dimensions. In this research, we focus on political governance. We assess this dimension through the political risk factors described in section 3.2. Indeed, the quality of political governance in a country can affect the accumulation of physical and human capital and, consequently, TFP. For example, better control of corruption is likely to promote financial development and improve investment. Corruption can negatively affect the allocation of public funds and thus economic growth ([66]; [97]). Corruption directly affects sources of productivity improvement, technological progress, and investment ([95]; [56]). Furthermore, political stability is essential for financial development, improved investment, and consequently improved TFP. Indeed, in the presence of political instability, it becomes less attractive for entrepreneurs to invest in long-term projects ([87]; [39]; [84]). Furthermore, high-quality regulation enables the conclusion and enforcement of fair contracts between banks on the one hand, and depositors and insurance institutions on the other ([47]). This can therefore contribute to improving TFP.

Studies have analyzed the importance of institutions for TFP in certain contexts. In some of these studies, the quality of institutions impacts physical capital and human capital and, consequently, TFP ([54]; [32]; [64]). In addition to capital accumulation, institutions can affect TFP and business output through other channels ([83]; [30]; [40]; [4]; [52]). Better institutions create a favorable business environment and a structured legal framework, enabling actors to channel investments toward more productive activities ([3]; [21]; [67]). Economic institutions such as property rights over TFP influence TFP in SSA ([33]). More specifically, some research has examined the link between TFP and governance variables. In this regard, studies have highlighted the negative effect of corruption on TFP ([26]; [75]; [66]; [97]). Similarly, political instability hurts TFP ([8]).

Overall, this review of the literature highlights the abundance of work on the determinants of TFP. However, studies have paid less attention to the effects of information asymmetry and political governance factors on TFP, particularly in the WAEMU context.

The identification of factors explaining TFP has been the subject of modeling in studies. However, due to differences in the contexts of these studies, not all these models can explain TFP in the WAEMU. Our model is based on the following specification of the TFP determinants model by [33]:

where:

tfp it is total factor productivity; gdp it is gross domestic product; pl it is the general price level; govt it is total public expenditure; trade it is trade openness; pop it is the total population; ms it is the M2 money supply; infras it is the physical infrastructure index, hc it is the human capital index, rd it is the level of research and development; inst it is the level of institutions; i and t refer to the individual (country) and temporal (year) dimensions, respectively.

This specification of the TFP model concerns sub-Saharan Africa and is therefore closest to the general context of developing countries, including those of the WAEMU. We adjust this model to the context of our research by considering variables assessing banking development, information asymmetry, and political governance. We therefore specify the basic TFP model in the WAEMU as follows:

where:

GroRat it , is a measure of economic activity; GovExp it , public spending; PopGro it , population growth; BroMon it , money supply; PhyCap it , physical capital index; HumCap it , human capital index; BanDev it , banking development; AsyInf it , information asymmetry; and PolGov it , political governance.

According to the previous literature review, information asymmetry can also affect TFP through the channel of banking development ([60]; [19]; [100]; [28]). We can therefore specify the following TFP model:

Given the diversity of the databases used and the need for the longest possible series to analyze long-term effects, we excluded Benin, Guinea, and Mali from the sample due to the unavailability of long series or missing data on certain variables of interest. For example, in the ICRG database, political governance data are not available for Benin. Similarly, for certain periods, data on the tfp it are not available for Guinea and Mali. As a result, our sample includes five WAEMU countries, namely Burkina Faso, Côte d’Ivoire, Niger, Senegal, and Togo, and covers the period 2003-2023.

Table 1. Description of model variables.

Source: Authors, February 2026.

Estimating the econometric model requires studying the stationarity of the different variables. Since the seminal work of [58], unit root tests have evolved significantly in the late 1990s, notably with the work of [45], and [63]. In our study, we use the LLC unit root tests of [59] and the IPS tests of [46] and [43] because of their robustness and frequent use in the literature. The panel unit root test of [59] assumes a homogeneous autoregressive unit root, while [46] assume a heterogeneous unit root. The latter unit root test takes heterogeneity into account.

We present the results of the various unit root tests in level and first difference form in Table A1 and Table A2. These results of the panel unit root tests are consistent and show that the variables are integrated of order one.

We therefore proceed with panel cointegration tests. Among the tests developed in the panel context, we consider the tests for the absence of cointegration on panel data proposed by [76] and [50]. Pedroni’s tests take heterogeneity into account. [50], meanwhile, proposed tests of the null hypothesis of no cointegration: the Dickey-Fuller test and the augmented Dickey-Fuller test. Unlike Pedroni’s tests, Kao considers the special case where cointegration vectors are assumed to be homogeneous across individuals. Ideally, to test for the existence of cointegration vectors between the variables under consideration, the traditional approach of [79] seems to be a better option. However, this approach is limited when the sample size is small. Using Monte Carlo simulations, [42] showed that the smaller the sample size, the more Kao’s tests dominate those of Pedroni. Therefore, the structure of our panel (N = 5, T = 21) gives an advantage to Kao’s ADF test. As a result, [50] cointegration test approach is more appropriate in the specific case of this research. The various [50] tests performed concluded that there is cointegration between the PTF and the various explanatory variables in the different models and sub-models (see Table A3 and Table A4).

In this research, our model uses macroeconomic variables. Some variables in the system may therefore be jointly determined. Endogeneity issues may therefore arise. As a result, many explanatory variables may be correlated with the error term. Consequently, estimating our model using the ordinary least squares (OLS) method may produce inconsistent results due to endogeneity stemming from a possible simultaneity bias. In addition, reverse causality may arise from the explanatory variables. Given these considerations, an appropriate estimation of the PTF model requires the use of an estimator that considers the concerns expressed. We therefore used the Fully Modified Ordinary Least Squares (FMOLS) estimator proposed by [77]. The main strength of this approach is its ability to control the endogeneity of the explanatory variables. In addition, it allows for heterogeneity between cross-sectional units. Better still, for small panels where the individual dimension (N) is smaller than the time horizon (T), such as ours (N = 5 and T = 21), FMOLS estimators are highly effective ([77]).

As part of this study, we performed several robustness checks. These checks help address potential biases arising from the small sample size. When the panel data sample size is small, it can affect the inference of unit root and panel cointegration tests, such as LLC, IPS, and Kao, as well as long-term estimators such as FMOLS. We therefore performed alternative unit root tests (Levin-Lin-Chu, Im-Pesaran-Shin, and Hadri tests) to verify the robustness of the stationarity results. We have summarized the results in Table A1 and Table A2. We also performed cointegration tests (Kao and Pedroni) to verify that the long-term relationship is robust. Pedroni’s cointegration tests confirmed the results of Kao’s tests. In addition, we used alternative long-term estimators (FMOLS and Pooled Mean Group (PMG) estimator) to verify that the coefficients remain stable. The results of our estimates using the PMG estimator produced similar results to those of the FMOLS estimator, thus ensuring the stability and reliability of the results of this study.

We estimated PTF model 1 by substituting the various governance indicators, GovSta, CorCon, DemAcc, SocCon, InvPro, IntCon, ExtCon, MilPol, and nmg, respectively. In this case, we estimated nine sub-models: Sub-model 1 is an estimation of PTF model 1 with GovSta as the political governance variable, sub-model 2 with CorCon, sub-model 3 with DemAcc, sub-model 4 with SocCon, sub-model 5 with InvPro, sub-model 6 with IntCon, sub-model 7 with ExtCon, sub-model 8 with MilPol, and sub-model 9 with nmg. We summarize the results in Table 2.

The estimated long-term coefficients of the variable “AsyInf,” which captures the direct effect of information asymmetry on TFP, appear to be significant and negative in all sub-models. Information asymmetry, therefore, has a direct negative effect on TFP. This result is in line with our expectations. Indeed, as theory indicates, information asymmetry generates malfunctions in the credit market, marked by credit defaults. However, banks’ rationality requires them to distance themselves from these credit risks. In the WAEMU, these dysfunctions therefore prevent the full utilization of productive capacities in this market and consequently lead to productivity losses. This result is consistent with the conclusions of other studies. Indeed, credit risk harms economic growth in the WAEMU ([60]). Similarly, credit defaults are hampering banking development in West Africa ([27]). Furthermore, credit defaults limit bank financing in SSA ([96]).

The coefficients of the “PolGov” variable, which captures the specific effects of political governance factors on TFP, appear to be significant and positive in nine sub-models. It is non-significant in the sub-model with internal conflict (IntCon). Governance, therefore, affects TFP in WAEMU. Indeed, in the WAEMU, an improvement in the quality of each of these components of political governance stimulates TFP. This result is also in line with our expectations. Studies found comparable results. Indeed, productivity growth is higher in better-governed countries ([75]). Similarly, in the context of Italy, better local institutions can help firms to combine inputs more effectively, approach optimal size, and ultimately be more productive ([7]).

Table 2. Summary of results from PTF Model 1 estimates.

The values in parentheses are p-values; *** significant at 1%, ** significant at 5%, * significant at 10%. Source:authors’ estimates, February 2026.

About government stability, improved cohesion in government action, legislative strength, and popular support boost productivity in these countries. Political instability negatively affects growth by lowering productivity growth rates ([8]). Concerning corruption control, reducing corruption reduces resource allocation problems in the WAEMU and, consequently, improves TFP in this region. This result is consistent with economic literature. Indeed, corruption could lead to inefficient allocation of public funds and negatively affect economic growth ([66]; [97]). Corruption has negative consequences on resource allocation, entrepreneurship, investment, and innovation ([16]). Elevated levels of corruption and informal economies make it difficult for new companies to enter the market ([29]). These barriers to market entry discourage investment decisions ([13]). Consequently, reducing corruption would boost productivity in the WAEMU. About democratic accountability, improving the government’s responsiveness to the concerns of the population stimulates TFP in the WAEMU. Concerning internal conflicts, a reduction in civil unrest, civil war, and terrorism contributes to improving TFP in the WAEMU. About socioeconomic conditions, an improvement in consumer confidence, employment levels, and living conditions would have a positive impact on productivity in these countries. Concerning the investment profile, an improvement in the viability of contracts, the repatriation of funds, and the reduction of payment delays in the countries concerned promote TFP. [53] found a negative coefficient for investment conditions on financial development in the WAEMU. According to him, this result does not reflect a negative effect of this variable on financial sector development; rather, it indicates that efforts to improve the business environment have not gained the confidence of investors and financial institutions and stimulated the supply and demand for financial services. About external conflicts, a reduction in cross-border conflicts, foreign pressure, and war promotes improvement in TFP in the WAEMU. Concerning the military in politics, the results show a paradox in the context of the WAEMU. Although a high level of this indicator reflects a high political risk, it contributes to promoting the application and/or enforcement of existing laws by the population and the restoration of state authority, which is essential for improving TFP.

We estimated PTF model 2 by cross-referencing banking development with information asymmetry. We present the results in Table 3.

The long-term coefficients of the variables capturing the cross-effect of information asymmetry and banking development on TFP appear to be significant and negative. However, the coefficients for banking development are positive. The negative coefficient of this interactive variable, therefore, stems from information asymmetry. For example, considering sub-model 9 (seeTable 3), the estimated coefficient of “BankDev” is 0.002728. However, the estimated coefficient of “AsyInf * BanDev” is −0.012123. The marginal effect of “AsyInf” is therefore −4.443914. In WAEMU countries, information asymmetry therefore has an indirect negative effect on TFP in the long term. Indeed, in the credit market, information asymmetry generates credit defaults, resulting in a decrease in bank financing, which affects TFP growth in the long term. Studies found comparable results. Credit defaults harm bank financing in sub-Saharan Africa ([96]; [27]). These credit defaults cause banks and financial institutions to distance themselves from borrowers. This distancing is more significant when the proportion of opportunistic borrowers is high, leading to a decrease in bank financing.

Table 3. Summary of the results of the PTF model 2 estimates.

The values in parentheses are p-values; *** significant at 1%, ** significant at 5%, * significant at 10%. Source:authors’ estimates, February 2026.

The results of the other variables of the model are presented inTable 2. The estimated coefficients for physical capital appear significant and negative in the various estimates. The estimated coefficients for human capital appear significant and negative in five sub-models. According to the literature on TFP, physical capital and human capital are important aspects of productivity improvement ([9]; [20]; [11]; [71]; [75]). The accumulation processes of these factors can explain the results for these two factors in WAEMU countries. Concerning physical capital, an efficient combination of inputs and/or widespread adoption of existing technology in economies would lead to such a result in the WAEMU. Indeed, the process of capital accumulation can affect productivity growth in an economy ([26]). Technology returns may therefore differ depending on an economy’s combination of inputs. The adoption of existing technology through capital accumulation can contribute to efficiency gains ([85]). [33] found a negative coefficient of physical capital on TFP in sub-Saharan Africa (SSA). According to them, countries in this region have experienced elevated levels of infrastructure deterioration, civil unrest, political instability, and violence, which can contribute to the inadequate quality of physical capital, thus not reflecting positively on TFP. About human capital, the low level of this factor could explain the unexpected result in the WAEMU, limiting its effect on the productivity of other factors of production and its contribution to technological progress and efficiency gains in this area. Indeed, human capital can have a level effect on the productivity of other productive inputs and influence the evolution of TFP ([26]). Studies on TFP also obtain comparable results of a negative effect of human capital ([33]; [98]; [24]; [90]; [71]; [65]; [82]). Indeed, in SSA, education levels and other measures of human development have not yet been fully utilized in the production sector ([33]). By extension, this explanation could also be valid in the WAEMU. Another explanation could be the inefficient use of human capital factors in the WAEMU. Indeed, to increase productivity, an economy needs to make effective use of human capital in the labor market and increase the number of skilled workers to exploit more sophisticated technology ([48]).

The estimated coefficients of trade openness are significant and negative in all estimates. This result contradicts economic theory. The literature suggests that trade and openness positively influence productivity ([99]; [70]; [12]; [69]). In Malaysia, openness to foreign companies and the global economy, as well as economic restructuring that allows their presence, contribute to TFP growth ([48]). However, the study by [33] finds a negative and significant effect of trade openness in SSA. The estimated coefficients for banking development are significant and positive. Improved financing of the economy stimulates TFP in the WAEMU. However, [33] found negative coefficients for banking development on TFP in SSA. According to these authors, most citizens in SSA do not benefit from the development of the banking system. Access to credit remains limited, particularly for small business owners who operate mainly in the informal sector, where most of the population is employed ([33]). The negative coefficient of population growth can be explained by the relative weakness of the quality of the WAEMU population. TFP growth is more likely to be linked to improvements in the productive capacity of the population than to its size. The economic growth rate coefficients are significant and positive. An improvement in the level of economic activity has positive effects on TFP in the long term. On the other hand, the money supply and public expenditure coefficients are significant and negative. The change in the level of these two factors is relatively small to stimulate TFP in the countries concerned.