Abstract
This article investigates the dynamic relationship among physical infrastructure, financial development, human capital and economic growth in Bangladesh, employing Autoregressive Distributed Lag (ARDL) bound co-integration and Granger causality test for the period 1985–2019. The study finds a significantly positive long-term impact of physical infrastructure and human capital on economic growth. However, the effect of financial development on growth is found to be negative, and the result suggests that financial development will take place with economic growth. From the policy perspective, this study emphasises increasing investment in physical infrastructure and human capital for Bangladesh to foster long-term economic growth.
Keywords
Introduction
This study investigates the impact of physical infrastructure, financial development and human capital on economic growth in the context of Bangladesh. The development of better physical infrastructure, a modern financial system and skilled labour are crucial to keep the current growth momentum of Bangladesh, which is nearly 8 per cent. 1 Although Bangladesh is spending an increasing amount on physical infrastructure through undertaking several mega projects, more investment is required, since the current spending on physical infrastructure standing at 2.8 per cent of the GDP which falls below the target level of infrastructure spending of 5 per cent of the GDP. 2 Moreover, to sustain long-term growth, skilled human resources are quite important, which is emphasised in both theoretical and empirical literature (Li & Wang, 2018; Lucas, 1988). The share of public spending on education to the GDP was barely 2 per cent in 2019, which is lower than that of India, Bhutan and Nepal. 3 As for financial development, the ratio of domestic credit provided by the financial sector to GDP stood at 67.79 per cent in 2019, which was nearly 20 per cent in 1985. It appears that lack of investment in physical infrastructure and human capital might act as obstacles to achieving higher economic growth. In this respect, it is essential to understand the key roles of physical infrastructure, human capital and financial development in enhancing economic growth in the context of Bangladesh.
Generally, infrastructure can be classified into three broad categories: physical infrastructure, financial infrastructure and social infrastructure. Physical infrastructure development means investment in transport, electricity, water and sanitation, etc., financial infrastructure development includes the existence of efficient financial and capital markets, and social infrastructure development indicates investment in education and health services. All three different types of infrastructure have a decisive influence on the economic output. Better physical infrastructure trims down inputs and transaction costs and advances productivity, causing economic growth in the end (Kumari & Sharma, 2017; Meersman & Nazemzadeh, 2017). Efficient financial infrastructure minimises financial intermediation costs and ensures efficient allocation of resources (Xu & Tan, 2020). Finally, human capital, as the measure of social infrastructure development, guarantees the availability of skilled and highly productive labour that contributes to economic growth (Han & Lee, 2020; Ogundari & Awokuse, 2018).
As it seems, the three different types of infrastructure affect economic growth through various channels such as cost, productivity, employment, efficient allocation, the availability of specific resources etc. Taking all these facts into account, it can be easily argued that all three factors have a very important role to play in enhancing economic outputs of a country. However, there is hardly any study that has investigated the combined effects of physical, financial and social infrastructure on economic growth. Moreover, previous studies have mostly dealt with a particular indicator of physical infrastructure and financial development. In this study, we have used physical infrastructure and financial development indices constructed through a principal component analysis (PCA) of different indicators of physical infrastructure and financial development. Therefore, this research fills up two research gaps of the existing literature. It explores the effects of all the three factors—physical infrastructure, financial development and human capital—on economic growth. Rather than using individual indicators of physical infrastructure and financial development, this study constructed PCA based indices of physical infrastructure and financial development.
Given the fact that physical, social and financial infrastructures are complementary to each other, we investigate the role of each of these three different types of infrastructure in economic growth, employing the Autoregressive Distributive Lag (ARDL) bound co-integration approach in the context of Bangladesh for the period 1985–2019. The results show a significantly positive impact of both physical infrastructure and human capital on per capita GDP (GDPP) growth in the long run. However, the impact of financial development on economic growth is found negative, and the result indicates that financial development will take place as growth picks up.
The whole article is organised as follows. Section 2 presents the literature review. In Section 3, the analytical framework of the study is presented. Data, physical infrastructure index (PINFI), financial development index (FINDI), methodology and the estimated models are discussed in Section 4. The effects of financial development, physical infrastructure and human capital are analysed in Section 5. In Section 6, the individual coefficients of each indicator of physical infrastructure and financial development are estimated. Results from Granger causality and Impulse Response Function (IRF) tests are also presented in Section 6. The final section provides the conclusion, with policy recommendations.
Literature Review
There are many empirical studies that separately investigated the impacts of physical infrastructure, financial development and human capital on growth. Almost all such studies found a positive impact of physical infrastructure on economic growth. (Batuo, 2015; Fedderke & Bogetic, 2009; Helm, 2009; Zhang & Fan, 2004; Zhang et al., 2015), underlining the positive role of physical infrastructure on economic growth. Studies based on India (Lall, 2007; Sahoo & Ranjan Kumar, 2009) found that physical infrastructure, such as transport and electricity, have a positive influence on economic growth. This means that physical infrastructure leads to economic growth. Pradhan and Bagchi (2013) and Pradhan et al. (2014) found a unidirectional causal impact running from infrastructure to economic growth.
Human capital is considered one of the significant determinants of long-term economic growth, and its role is rightly emphasised in a formal model by Lucas (1988). Investment in human capital, which is acquired through education, on-the-job training and learning by doing, is supposed to increase economic growth, and there are different channels through which human capital enhances economic growth. One of the channels is foreign direct investment; the availability of better human capital induces foreign firms to invest in a country, which, in turn, increases economic outputs (Eaton & Tamura, 1995; Razin et al., 2008). Another channel through which human capital increases economic growth is labour productivity. Greater investment in human capital is thought to augment productivity, and a study by Bils and Klenow (2000) showed a significantly positive long-term relationship between educational achievement and economic growth enforced by the positive connection between human capital and higher productivity. Ifeoma et al. (2013) provided evidence of the greater growth potential of investing a dollar in human capital than in physical capital such as roads and others. Regressing secondary school enrolment as a measure of human capital to economic growth, Ojo and Oshikoya (1995) found a strong, positive long-term effect of human capital investment on economic growth. The same result was also reported by Mankiw et al. (1992). In the context of Bangladesh, Sharif (2013) obtained positive correlation between educational achievement and economic growth using the Engle–Granger co-integration test. Sultana et al. (2019) also provided evidence of the positive effect of human capital on growth. (Hossain et al., 2020) studied human capital development suggesting to provide appropriate IT training to the young Bangladeshi individuals.
Empirical evidence regarding the impact of financial development or improvement in financial infrastructure is quite mixed. Some studies have documented a positive impact, whereas others have reported the opposite. The modern financial system drives economic growth through diversifying risk, promoting innovations and reducing costs. This means that it leads to economic growth because of higher human and physical capital accumulation and technical progress. Studies (Cojocaru et al., 2016; Durusu-Ciftci et al., 2017; Jedidia et al., 2014; Loayza & Ranciere, 2004) have provided empirical support for the positive relation between financial development and economic growth. However, other studies (Easterly & Levine, 2002; Law & Singh, 2014; Rousseau & Wachtel, 2011) have found that financial development negatively affects economic growth. Cecchetti and Kharroubi (2012) argued that the adverse effect of financial development on economic growth could be the result of a crowding-out effect, since the financial sector competes for resources with the other sectors. Moreover, bad credits, adverse selection of loan projects and a weak monitoring and regulatory system of financial markets like Dhaka Stock exchange (DSE) (Ahmed et al., 2020) could dampen the economic output amid rapid financial development. In addition, the effect of financial development is not symmetric over economic regions. The effect on growth is stronger in developed countries (Xu, 2000) but weaker in developing countries (Kar et al., 2011). Therefore, investigating the impact of financial development in the context of a developing country like Bangladesh is quite worthwhile.
From the above literature review, we see that there is a consensus regarding the positive impacts of physical infrastructure and human capital on economic growth. However, little ambiguous agreement exists about the effect of financial development on growth. Most importantly, no study has analysed the simultaneous impacts of physical infrastructure, financial development, and human capital on economic growth. Démurger (2001) investigated the effect of financial development on economic growth, whereas Dwyfor and Christopher (2002) examined the effect of both financial development and human capital. Sahoo and Ranjan Kumar (2009) studied the impact of physical infrastructure on growth in the context of India. Along with physical infrastructure, Mohanty and Bhanumurthy (2018) considered financial development while analysing their influence on economic growth. In the context of Bangladesh, Sharif (2013) and Rahman (2007) separately probed into the effect of human capital and financial development, respectively, on growth. However, our study analyses the effects of physical infrastructure, financial development and human capital together on economic growth in Bangladesh. Therefore, this article makes a unique contribution to the existing literature.
The Analytical Framework
In an economy that follows the Cobb–Douglas function, the aggregate output is given by the following simple equation:
where Y t is the total output, L t is labour, K t is capital, A is the total factor productivity and α and β are the elasticities of labour and capital, respectively. Incorporating physical infrastructure, human capital and financial development as variables of interest, we write the extended Cobb–Douglas function as:
where Z t includes physical infrastructure, human capital, financial development and inflation. Expression of the modified Cobb–Douglas function in log-log form yields:
where ln stands for natural logarithm. On the basis of the augmented Cobb–Douglas function, we specify four models for the purpose of empirical estimation:
Expected Signs of Variables
Data and Variable Descriptions
Data and Variable Descriptions
Construction of Physical Infrastructure Index (PINFI)
Construction of Physical Infrastructure Index (PINFI)
Construction of Financial Development Index
Construction of Financial Development Index
The unit root test (Table A4) performed using the augmented Dickey–Fuller (ADF) and Phillips–Perron (PP) methods confirms that some variables are stationary at first difference I(1), whereas others are stationary at level I(0). Given the facts that the order of integration is mixed, that none of the regressors are I(2) and that the sample size is comparatively small—35 yearly observations—ARDL bound co-integration is used to analyse the long-term dynamics among the selected variables. One of the advantages of using ARDL is that it provides both short-term and long-term results. Another thing is that ARDL bound co-integration gives good results with short time series data, whereas for the Johansen co-integration test, a larger sample size is required to produce robust results. Thus, the estimated model is:
where ∆ stands for the first difference,
Here, ECT is the error correction term. The coefficient of ECT must be negative; otherwise, the deviations from the equilibrium will not be corrected and the model will move away from long-term equilibrium.
Long-term Results
ARDL Bound Test
ARDL Bound Test
Long-Term Dynamics Results
Physical infrastructure proxied by PINFI has a significantly positive effect on economic growth, with coefficients 0.20 and 0.22 in models A and D, respectively. A unit increase in PINFI raises economic growth by approximately 20 per cent. The result is consistent with those of Fedderke et al. (2006), Batuo (2015) and Démurger (2001). This means that in the long run, physical infrastructure development plays a very crucial role in boosting economic growth in Bangladesh. Financial development is found to have a significantly negative impact on economic growth in models B and D, which is in line with the findings of Hye and Islam (2013) for Bangladesh, and the findings support the hypothesised negative relationship outlined at the beginning. Thus, the study does not support the positive relationship between financial development and economic growth in Bangladesh found earlier by Rahman (2004, 2007). Moreover, the effect of financial development is not symmetric over the different economic regions: the effect is supposed to be positive in the developed countries and negative in the developing countries. The negative effect could be the result of a crowding-out effect, since the financial sector competes for resources with the rest of the economy. Moreover, financial development that is plagued by the expansion of bad credits and adverse selection of loan projects is likely to hamper, rather than boost, economic growth. Also, historically, financial expansion in the context of Bangladesh has not been so beneficial, given the fact of gross corruption and unfair practices in the financial sector. In this respect, the negative impact of financial development on economic growth in Bangladesh is quite expected. The contribution of GCF to economic growth is relatively more than that of PINFI; a 1 per cent increase in GCF increases economic growth by 64 per cent in model A and by more than 30 per cent in model D in which PINFI is included. The variations observed in the effect of GCF in models A and D could be due to an omitted-variable bias, since the impact of GCF gets reduced by 50 per cent when HCI and FINDI are included in model D. The positive results indicate that Bangladesh should increase investment in gross fixed capital. Such investment would pay off in the long run through increasing employment and aggregate output, as reflected in the positive effect of GCF. Consistent with the theoretical prediction, human capital has been found to make a significant positive contribution to economic growth; a unit increase in HCI leads to, on average, an increase of more than 80 per cent in economic growth in both models C and D. A similar result in the context of Bangladesh was also reported by Chowdhury et al. (2018). The marginal effect of HCI on economic growth rises from 80 per cent to almost 130 per cent in model D when PINFI and FINDI are added to the model, testifying to the fact that the lower impact of human capital could be the result of omitted-variable bias.
The impact of inflation on economic growth is ambiguous; it is negative in model B but positive in all other models. Theoretical and empirical literature show that the effect of inflation on economic growth is not very straightforward. Moreover, the effect is also not symmetric over economic regions. Easterly and Bruno (1999) argued that the effect of inflation on economic growth is detrimental when inflation exceeds 40 per cent in developing countries. On the other hand, Kremer et al. (2013) presented the threshold effect of inflation, showing that inflation adversely affects economic growth in developing countries if it exceeds the threshold level of 17.2 per cent. However, they also reported that reducing inflation below the threshold level does not enhance economic growth. Historically, inflation rate in Bangladesh has been around 6 per cent, and this rate is quite lower than the threshold level specified above. The significantly positive effect of inflation on economic growth may be due to the shift in income distribution in favour of capitalists, since nominal wages lag behind prices during times of mild inflation. Since capitalists have a higher propensity to save, according to the Kaldorian proposition, higher savings may raise investment and growth in the end.
Short-term Results
Model Diagnostics
Coeff is estimated coefficients.
The figures in [ ] are the p-values.
For model A, ARDL (4, 4, 3, 0, 3) is selected based on AIC; for model B, ARDL (4, 3, 3, 3, 1) is selected based on AIC; for model C, ARDL (1, 5, 5, 5, 4) is selected based on adjusted R2; and for model D, ARDL (5, 1, 1, 1, 1, 0, 0) is selected based on SIC.
As seen, the effect of GCF is positive both in the short and long run. As shown in the long run, the influence of inflation on economic growth is quite ambiguous, alternating between positive and negative among the models. In contrast to the long-term relationship, the effect of PINFI, FINDI and HCI are negative in the short run. Therefore, the contribution of GCF to economic growth is more than that of physical infrastructure, human capital and financial development. Compatible with the theoretical argument, the coefficient of the ECT is negatively significant in all models, reflecting the fact that the system will converge to long-term equilibrium.
Several diagnostic tests, such as serial correlation test, Jarque–Bera normality test and heteroscedasticity test, have been employed to check the robustness of the estimated models. The results of Table 8 shows that the models do not suffer from serial correlation and heteroscedasticity and that the residuals are normally distributed. The stability of the models has been measured by the cumulative sum (CUSUM) and the cumulative sum of squares (CUSUMSQ). It can be easily seen that both CUSUM and CUSUMSQ test statistics do not exceed the bounds of the 5 per cent level of significance, confirming that the estimated models are stable over time (Figures 1–4).




Long-term Coefficients of Selected Infrastructure and Financial Development Indicators
Long-term Coefficients of Selected Infrastructure and Financial Development Indicators
It is noted that the long-term coefficients of PINFI are 0.209 and 0.226 for model A and model D, respectively, and the long-term coefficients of FINDI are −0.323 and −0.093 for model B and D, respectively. The factor loadings 0.568, 0.590 and 0.574, respectively, of the three infrastructure indicators road density, mobile cellular subscription and electricity consumption are multiplied with the coefficients of PINFI, and the coefficient of FINDI is multiplied with the factor loadings of 0.707. The results show that a unit change in each of the three physical infrastructure indicators causes around a 12–13 per cent change in economic growth. The interesting thing is that the contributions of each of the indicators are quite similar; they almost equally affect economic growth in Bangladesh. A similar scenario is also noted in the case of FINDI. In line with the negative long-term impact of financial development on economic growth, the effects of either indicators of FINDI are also negative. Both money multiplier and financial sector have an equal negative long-term influence on economic growth in Bangladesh. This breakdown of the long-term impact of individual indicators of PINFI and FINDI confirms that the long-term impact of individual physical infrastructure indicators is greater than that of financial development indicators.
Granger Causality Test
So far, we have checked the long-term relationship among economic growth, physical infrastructure, human capital, financial development, inflation and labour. Now the question is, which causes what? More specifically, we need to determine exactly how the causal relationship runs between, for example, economic growth and human capital. Does human capital cause economic growth, or does economic growth cause human capital, or does it work both ways? The detection of the direction of causal impact is of great importance, since it particularly tells us about the potential source of causation and how it takes form. For this purpose, the Granger causality test put forward by Toda Yamamoto (1995) is applied to see the causal relationship between the variables of interest. To test the existence of Granger causality, the following vector autoregressive (VAR) model is applied:
VAR Granger Causality Test
The results reported in Table 10 show a bidirectional causal effect between economic growth and physical infrastructure development. This means that not only does physical infrastructure development Granger-cause economic growth but also economic growth induces more investment in physical infrastructure. In both cases, the hypothesis of no Granger causal impact is rejected at the 1 per cent level of significance. The bidirectional causal impact could be the result of the fact that as the economy develops, it would need more investment in infrastructure development to sustain its growth.
As for financial development, the relationship runs from economic growth to financial development. This means higher economic growth causes financial development; the result is similar to that of Hossain et al. (2017). We have already seen that the impact of financial development on economic growth is negative, and such a negative contribution to growth could suppress the development of the financial system. This means higher economic growth might encourage financial system development, suggesting that in the case of Bangladesh demand factors are the key drivers behind financial development, rather than supply factors. Evidence of bidirectional causal relation is also found between human capital and economic growth, which is compatible with the findings of Rahman (2011). Higher human capital contributes positively to economic growth, as evidenced by the above empirical result. Since an increasing level of human capital indicates the availability of skilled and productive human resources, investment in human capital can boost long-term growth. In turn, higher growth can encourage a country to invest more in human capital.
To understand further the dynamic relationship among economic growth, physical infrastructure, financial development and human capital, we estimate VAR, which is utilised to derive impulse response function (IRF). The standard VAR model is given as:
where X
t
is a vector of the endogenous variables—economic growth, financial development and human capital—c is the vector of intercepts, B is the coefficient matrix and

It is seen that economic growth is positively influenced by shocks to physical infrastructure development. In 10 years of time, physical infrastructure increases economic growth up to a standard deviation of 0.40. However, FINDI generates a negative impact, decreasing economic growth over the 10 years, with a standard deviation of 0.35, and the result is consistent with the earlier result of long-term negative effect of financial development on economic growth. The finding of the positive long-term impact of human capital noted earlier is also confirmed by IRF. Human capital creates a positive impact, increasing economic growth, with a standard deviation of 0.07, in 10 years of time. As it turns out, most of the responses are consistent with both the theoretical arguments and earlier results of the long-term positive impact of PINFI, negative impact of FINDI and positive impact of HCI.
So far, we have examined the impact of physical infrastructure, human capital and financial development on economic growth in Bangladesh using data covering the period 1985–2019. The study enables us to derive the following conclusions. First, both ARDL and IRF results show that physical infrastructure has a positive impact in the long run. This means that Bangladesh can boost its long-term economic growth by investing in physical infrastructure. The positive result is reinforced further by the bidirectional causal relationship between physical infrastructure and economic growth, indicating that higher investment in infrastructure leads to economic growth and higher growth induces more investment in infrastructure. Second, the impact of financial development is negative both in the short run and long run, as confirmed by the ARDL and IRF results. The Granger causality test shows that economic growth leads to financial development, rather than financial development causing economic growth. Third, investment in human capital pays off in the long run, since the effect of human capital on growth is positive in the long run, and there are increasing returns from investment. The causality test reveals that both human capital and economic growth causes each other, suggesting that higher human capital causes growth, which in turn stimulates more investment in human capital.
Given the empirical evidence, the study proposes the following policy recommendations. Bangladesh can reap higher returns from investment in human capital. The empirical evidence lends support to this assertion, since a unit increase in human capital increases economic growth by more than 80 per cent in the long run. Thus, Bangladesh, with its huge population bulge, can immensely gain from incremental spending on education, which ultimately gets embodied as human capital. Further, investment in human capital will boost economic growth, augmenting labour productivity. Bangladesh should emphasise more investment in physical infrastructure development as well. This policy suggestion is underscored by the following two critical observations: (a) Bangladesh needs to spend 5 per cent of its GDP on physical infrastructure to compensate for the deficit in physical infrastructure; and (b) Empirical evidence shows that in the long run, a unit increase in infrastructure development causes, on average, an increase of more than 20 per cent in economic growth. Thus, infrastructure development should be prioritised in the coming years, as it would enable Bangladesh to keep itself on the trajectory of sustained long-term growth. The unidirectional causal impact from economic growth to financial development implies that financial development would respond passively to economic growth; this means that financial development would follow economic growth.
Finally, it is worthwhile to mention that the findings of the article should not override its limitations. One of the limitations is that while constructing physical infrastructure and financial development indices, we have considered a few indicators while leaving out others that could also have been considered. This limitation is reinforced by the lack of availability of data for different indicators of physical infrastructure and financial development. Another limitation is that the study has not considered governance or institutional factors, which are of relevance in the context of Bangladesh. This issue opens up the possibility of further research in which institutional factors could be taken into account.
Footnotes
Acknowledgement
The authors acknowledge Mr Md. Monirul Islam, Assistant Professor, Bangladesh Institute of Governance and Management (BIGM), for his valuable suggestions.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The authors received no financial support for the research, authorship and/or publication of this article.
1
Bangladesh Economic Review (2019).
2
Seventh Five-Year Plan, Bangladesh.
3
World Development Indicators (WDI).
Appendix A
| Physical Infrastructure Variables | Road Density | Mobile Cellular Subscription | Electricity Consumption |
| Road density |
|
||
| Mobile cellular Subscription | 0.943892 |
|
|
| Electricity consumption | 0.862398 | 0.970260 |
|
| Financial Development Variables | Money Multiplier | Financial Sector Credit |
| Money multiplier | 1 | |
| Financial sector credit | 0.885278 | 1 |
| Statistics | lnGDPP | lnGCF | lnLF | lnINF | PINFI | FINDI | HCI |
| Maximum | 1.954839 | 2.480872 | 18.05981 | 2.951948 | 1.880095 | 1.820885 | 2.059684 |
| Minimum | −1.573659 | 0.426550 | 17.16473 | −1.860993 | −3.288946 | −2.623568 | 1.380054 |
| Mean | 1.032495 | 2.011156 | 17.65580 | 1.685560 | −2.86E-11 | 6.34E-17 | 1.704711 |
| Std. deviation | 0.831201 | 0.427663 | 0.262726 | 0.759154 | 1.713375 | 1.393100 | 0.213615 |
| Skewness | −1.603539 | −2.008115 | −0.276791 | −2.792834 | −0.609125 | −0.310175 | 0.179323 |
| Kurtosis | 5.382772 | 7.438821 | 1.906312 | 14.96812 | 1.901341 | 1.731876 | 1.771894 |
| Observations | 35 | 35 | 35 | 35 | 35 | 35 | 35 |
| Variables | Augmented Dickey–Fuller Test | Phillips–Perron Test | Order of Integration | ||
| Intercept | Intercept and Trend | Intercept | Intercept and Trend | ||
| lnGDPP | −5.709052*** | −3.454772 | −2.555714 | −4.424294** | I(0) |
| lnLF | −2.145554 | −1.934894 | −1.686094 | −0.841113 | – |
| lnGCF | −2.140158 | −1.526627 | −0.650004 | −1.356464 | – |
| lnINF | −1.933652 | −4.206470* | −4.066998** | −4.393381** | I(0) |
| HCI | 1.216854 | −1.856427 | 1.174319 | −1.867615 | – |
| FINDI | −1.517939 | −1.688029 | −1.465013 | −1.688029 | – |
| PINFI | −2.581053 | −0.240311 | −2.581053 | −0.044406 | – |
| lnGDPP | −5.877970*** | −7.092466*** | −8.734058*** | −8.549755*** | – |
| lnLF | −1.629425 | −7.161905*** | −1.635107 | −1.851275 | I(1) |
| lnGCF | −3.973095** | −4.792937** | −3.973095** | −3.930240* | I(1) |
| lnINF | −2.790539 | −2.634316 | −17.66765*** | −19.47855*** | – |
| HCI | −5.023179*** | −5.166697** | −5.034082*** | −5.166697** | I(1) |
| FINDI | −4.220067** | −4.333253** | −4.271852** | −4.385138** | I(1) |
| PINFI | −5.382027*** | −7.314619*** | −5.482587*** | −7.629510*** | I(1) |
