A latent factor of Development Disparities: an Empirical Study from Nepal.

Ram Prasad Mainali (PhD) [1]

Despite the fact that disparities in development can have detrimental effect in achieving equitable living standard of the people and thereby to promote social harmony, empirical study on the subject is scant.  This study extends the gravity theory within the national context and estimates the impact of distance on development which is proxied by per capita Gross National Income (GNI). Nepal's district-wise data has been employed in the study. Assuming Kathmandu as the central point, regression results shows that districts being an additional kilometre away from Kathmandu own per capita GNI less by eighteen rupees. Moreover, strategic road network, although taken for granted as the synonymous of development in Nepal alike other developing countries, did not show a positive impact on per capita income. Other control variables have shown mixed results in terms of contributing to the GNI. This finding, if taken into account by policy circle, is likely to have positive contribution in enhancing equitable development which is crucial in countries suffering from inequitable distribution of development alike Nepal.

 

Key words: Development disparities, per capita income, central planning

 

1.           Introduction

Development disparities integrates three dimensions of development I,e. overall growth, Human Development Index and the agricultural growth (Banarjee and Kuri, 2015). It does not only encourages irregular labor and capital movement between prosperous and poor regions but also likely to drive nation into social unrest. Moreover, it is believed to have a detrimental effect on the political economy of a country (Cherodian and Thirwall, 2015). As a consequence, many countries are worrying about unbalanced development within regions and thus working towards to reverse it. 

Allocating resources across regions in an equitable manner is a key for balanced development. The balanced development explores development potential and paves the way to enjoy development output to all inhabitants (Qwatiah, 2021). However, it is discussed mainly across countries. Many scholars have published bunch of literature on regional disparities. Some notable publications, among others, are Roser (2020), Seshanna   and Decornez (2003), Stiglitz (2019) and Dorn et.al (2019).

Achieving balanced development is more difficult in developing countries compared to the industrialised one as they embodied growth inimical characteristics. Some of these characteristics can be explained as limited resources, inadequate knowledge of policy architecture, inaccessibility to government facilities and costly infrastructure caused by remoteness. Despite these, political leaders in such countries are more likely to allocate scarce resources to own constituencies in order to influence election by making voters in favour. This scenario is clearly visible on our annual budget. A positive correaltionship can be observed between the budget allocation and leaders in power in each year’s budget.  Prime minister, finance minister as well as party supremos grasp the relatively larger chunk of budget. Principle of right budgeting and attention on return to growth are limited into the literature rather than in practice.

Equitable distribution of factor of production is further limited in a country with centralised planning and budgeting. In such countries, central bureaucracy as well as political elites holds significant influences in allocating resources.  Remote inhabitants are less able to air their priorities through their political representatives. Nepal is one of the countries having central control on planning and budgeting. It has almost been a decade that Nepal has adopted federal framework of government but the framework of federalism is not fully institutionalized yet. Therefore, development of remote districts or region has not reflected in central planning as per the expectation of federalism.

Studies on inter-regional disparities are in increasing trend and these have evidently shown inequality in development prevails in many countries. For Instance, Poul et al. (2012) finds significant regional income disparities in Nepal. Similar results are found in India (Bakshi et. al., 2015), in Indonesia (Kuncoro, 2013) and china and India (Ho-fung Hung   and Jaime Kucinskas, 2011). However, studies to establish causing factors of inequality are given less attention. This study is believed to fill this gap.

This study focuses on inter-districts inequality in development which is proxied by per capita Gross National Income (GNI). This is interesting as Nepal is characterised by caste, ethnic, class and gender inequalities partly caused by historical social discrimination based on caste. Unequal distribution of other factors i.e. state of infrastructure, roads, airports, number of educational establishment etc. may cause regional disparities in GNI. Social attributes: population and access to facility may also aid to perpetuate this. Despite this, remote inhabitants lack the opportunity to interact and participate with policy circle while deciding on development planning and budget allocation. This allows top level bureaucracy and political leaders to enjoy freedom in assigning resources as per their interest rather than local needs which ultimately backs to inequitable development.  Therefore, distance from capital city has been considered as the latent factor of development which is the variable of interest in this study.

Rest of the study is organised as follows. The second section briefly discusses about Nepal's development planning and budgeting system. Data and model specification is described in the third section. The fourth section interprets the results. The final section concludes with recommendation. 

2.            Nepal in development planning and budgeting

Nepal has been practicing central level development planning since decades. National Planning Commission (NPC) plays a role of apex body for policy architecture (National Planning commission, 2021).  It has completed 14^th periodic plans. Currently the 15^th development plan is in place which is said to be the preparation for the long term vision of fulfilling the shared national aspiration of “Prosperous Nepal and Happy Nepali” by achieving the status of middle income country by 2043 (see the fifteenth plan, NPC for detail)

NPC estimates the optimal ceiling of resources for each fiscal year. Then it is divided into respective sectoral ministries’ ceiling on the basis of total resources estimated. Projects priorities are also given by NPC to ensure fiscal resources for prioritized projects. It implies that higher the priority more assurances of fiscal resources.  P1, P2 and P3 are the project categories currently being practiced by the NPC and thus P1 being the most ensured projects in terms of fiscal resources.

 

Policy circle vows that Nepal follows bottom-up approach of development process. Entire policy declarations reflect it in principle. If so, priorities expressed by ultimate spending units of sectoral ministries should have been reflected also in the NPC’s priority. However, it generally diapers during the process of budget preparation. Ministries collect program and policies from their extended units i.e. district level offices and dedicated projects. It is then compiled and forward to the NPC where tripartite negotiation held in order to finalize it by taking the available resources and ministries budget ceiling into account. NPC, Ministry of Finance and related ministry take part in finalising particular ministry’s plans and programs. This tripartite meeting is crucial since it is the point of time whether or not to accommodate local needs and priorities. However, top level bureaucracy and political elites are blamed to manipulate the local priority and hold own interest in the process of concluding development plans. Remote inhabitant is less likely to have access to such meeting. On the other hand, policy circle which includes top level bureaucracy, political leaders and experts to some extends do not give adequate attention to the need of remote region. If we look at the whole process, Nepal's planning and budgeting greatly reflects the interest of central nexus due to self-centric behaviour of the policy circle. This behaviour is coined in this study as the central level psychology which is represented by distance variable in the estimation model. This shows that despite the policy consensus on employing development plans following the bottom-up approach it is hardly reflected in the implementation and central level psychology plays a significant role.

Nepal experienced several changes in political regimes in the past. Political movements that cause such changes were inspired by hope to reverse income inequality and other socio-economic imbalance. Vanguards of these movements, particularly the Maoist, strongly pronounced inequity in order to mobilize the population in political movement. However, notable changes could not be observed in development planning process even at the period Maoist ascends to the power. Following to the people’s movement-II, country has been transformed into the federal from the unitary system of government but it has not been subsequently institutionalised yet. To summarize, despite change in political regime and several policy announcement to incorporate local level priority central psychology has not been changed and thus central level bureaucracy and political elite's capture on resource continued. Remote region is ignored and regional disparities in income perpetuated.

3.             Data and model specification

3.1 Data

"Open Data Nepal" portal has been used as the main source of data in this study. It provides several socioeconomic variables. Per capita GNI, population, caste/ethnic composition by districts have been used in the estimation model. These are the data set from 2015. Additionally, the length of Strategic Road Network (SRN) in the district has been extracted from the website of the department of roads. Airport data is collected from the website of the Ministry of Culture, Tourism and Civil Aviation (MCTCA)[2]. The later data is from 2013. Although study uses data set form two different points of time, it is believed that it may not alter the estimated results as there are no significant changes in SRN and Airports within the couple of years. Additional variables used in the model are distance to districts from Kathmandu and number of primary as well as secondary schools.

Descriptive statistics is presented in the table 1 below. This table shows average per capita districts GNI is NRs. 48,293. However, it shows a large variance across districts. For instance, standard deviation accounted as NRs. 43,374. Similarly, average distance to districts from Kathmandu is approximately 430 kilometres. Darchula is the farthest district from Kathmandu with GNI per capita NRs. 28,042.  Manag has the highest per capita GNI with NRs. 1,41,588 almost 15% higher than that of Kathmandu. Mainali et al. (2022) describes various reasons for this. It has not been reiterated in this study but considered Manag as the outlier of the sample.  

Table 1: Descriptive statistics

District	GNI Per Capita	SRNRoad Length	Population	Distance(KM)	Secondary
Achham	23984	149	257477	938.16	86
Argakhachi	40635	172.24	197632	374.87	80
Baglung	38819	228.13	268613	271.13	120
Baitadi	25613	247.62	250898	854.53	109
Bajhang	21772	108.58	195159	498.1	74
Bajura	23405	50	134912	822.2	52
Banke	50659	226.41	491313	510.07	129
Bara	66175	189.58	687708	283.55	58
Bardiya	48564	219.42	426576	543.4	80
Bhaktapur	61686	115.06	304651	14	184
Bhojpur	44666	107	182459	419	66
Chitwan	68748	233.75	579984	146.2	211
Dadeldhura	34164	161.08	142094	769.72	63
Dailekh	30609	276.23	261770	647.41	87
Dang	50415	365.02	552583	409.23	126
Darchula	28042	134.42	133274	960.86	60
Dhanding	43899	203.61	336067	85.16	135
Dhankuta	56223	134.68	163412	594.2	78
Dhanusa	41951	251.48	754777	381.82	78
Dolakha	41256	229.75	186557	131.72	93
Dolpa	46502	000	36700	642	14
Doti	34595	252.46	211746	832.66	69
Gorkha	46488	213.24	271061	140.48	102
.	.	.	.	.	.
.	.	.	.	.	.
.	.	.	.	.	.
Tanahun	47925	179.49	323288	150.05	147
Taplejung	58741	68.5	127461	835.3	57
Terathum	63470	125.07	101577	655.12	50
Udayapur	41156	256.51	317532	453.25	112

Note: Adapted from Mainali et al.(2022)

   All districts are not reported in the table in order to save the space. Full data is available upon request.   Kathmandu being the reference district variables associated with it is missing in the table.

 Figure bellow shows scatter plots of per capita GNI associated with 75 districts. Y axis in the figure presents the per capita GNI associated with particular district whereas X axis denotes the distance to districts from Kathmandu. The right side downward slope of the fitted value curve prima facially shows a Negative correlation of distance with per capita GNI. A multi variable regression model is carried out in order to test this hypothesis. 

 

P

3.2 Model specification

A multivariate regression model is specified as;

y_i=β₀+β₁dist_i+α_jXi+u_{i…………………..}     (i)

y_i=β₀+β₁dist_i+α_jXi+v_i………(ii) 

 

Where,

y_iis the dependent variable, per capita GNI of district _i and β₀ is the intercept. dist_i is the distance to district _i from Kathmandu. This is the variable of interest in this study. _Xi is the set of control variable which includes number of airports, number of schools, SRN road length and other binary variables as described in the appendix (see appendix 1). Finally, u_i is the error term associated with the district _i. Equation (ii) is extended to include province variable in the first equation.

Sign and significance of the β₁ indicates whether to accept or reject the null hypothesis that is distance has no significant impact on per capita GNI associates with districts. Similarly, sign and statistical significance of α_j shows the casual relationship of control variables with per capita GNI.

4.     Results

Table 2 presents the regression results. The second and third column represents the estimated coefficients and robust standard errors, respectively, from the equation 1. As described in the model specification, a separate regression is estimated by including provincial variable as an additional variable in the later equation. Estimated results and standard errors of the second model are presented in the fourth and fifth column of the table.

Table 2: Regression results

Variables	Coefficient	Standard Errors	Coefficient	Standard  Errors
Distance	-38.1297***	8.2911	-17.4189*	9.5694
Population	-.0364	.0269	-.0355	.0224
Secondary	32.4085	23.3032	30.3383	19.1837
Airport	7309.3540*	3760.1760	6876.8940**	3023.6320
Tarai	14821.2312	11319.0600	18357.8120*	9841.0360
Roadlength	-33.2671	27.6564	-26.9927	27.9826
Province-2	.	.	-12896.4900**	5889.9090
Province-3	.	.	-2868.5880	6217.4850
Province-4	.	.	6797.3240	8892.7940
Province-5	.	.	-12410.6000***	3442.9630
Province-6	.	.	-15292.8100***	4227.1110
Province-7	.	.	-20537.3900***	5491.873
Constant	71573.1700***	11221.1700	67501.3100***	9422.1510
R-squared		0.3098		.4668
No.Obs.		74		74

 

Note: ***, **and*significantat1%,5 %and10%level,respectively.

Adapted from Mainali et.al. (2022)

Variable of interest in this model is the distance variable. This is also coined in this study as central level psychological variable. I have argued in the introduction section that senior bureaucrats and political elites significantly influences the planning and budgetary decision in setting of central planning and budgetary system. In the other hand, remote population cannot air their grievances through their political representatives. As a consequence, farther the district from central headquarter lower will be the development expenditure and thus relatively lower per capita GNI. 

As expected, distance variable shows the negative coefficient of -38.1298 which is statistically significant at 1% level. It implies that districts a kilometre away from Kathmandu are likely to have approximately NRs. 38 less compared to Kathmandu. Secondary school variable shows a positive coefficient although it is not statistically significant. However, it indicates that an additional secondary school are likely to add GNI of the district compared to those having only the primary schools. It is obvious that higher the education larger will be the individual earning which can ultimately be reflected in the district GNI. Airport variable shows a positive coefficient and is statistically significant at 10% level. It implies that particular district having an airport is likely to own NRs. 7,309 more per capita GNI compared those districts with no airports in the territory.

Although statistically insignificant, SRN coefficient turned out to be negative. It is not generally expected. It indicates that additional kilometre of SRN road has no positive impact on district per capita GNI. It can easily be observed that remote district generally has long way to connect another districts without any economic activities along the road. It could be the reason behind it. Tarai variable sows the positive coefficient. It is understandable since Tarai has fertile land compared to hilly region of Nepal and relatively greater economic activities. Coefficient for population is negative but is statistically insignificant. It may puzzle the readers in the first sight as population are mainly dense in prosperous region. Therefore, it should have the opposite sign. However, dependent variable in this model is per capita GNI where denominator is the population. Therefore, higher the population the lower the GNI and thus can be reflected as the negative sign.

The fourth and fifth columns in the table of regression results report the result from extended model. This model includes provinces as an additional explanatory variable where province one is considered as the reference category. The estimated coefficient for distance in this model has been moderately reduced. However, it is still statistically significant at 10% level. Furthermore, extended model exhibits the similar pattern as in previous model regarding the distance results. All provinces, apart from province 4, have the negative coefficients. It implies that remaining provinces have less per capita GNI compared to the province 1.  It once again indicates distance as one of the latent factor causing development disparities in Nepal.

5.     Conclusion and recommendation

5.1 Conclusion

As remote regions lack access to market and also face geographical difficulties their growth mainly depends on government spending on infrastructure.  This study thus focuses on infrastructure and argues that government spending is undermined in the remote districts due to limited interaction with bureaucracy and political elites. Hypothesizing the distance as one of the barrier for interaction and air their grievances to central planner and policy circle this variable is tested among other control variables. The estimated coefficient for distance shows that a kilometre away from Kathmandu lessens the district average per capital GNI by eighteen rupees. This was the focus of the study and is considered as the latent factor of development disparities across districts.

Apart from the variable of interest, distance, econometric results have indicated some other interesting findings. For instance, Nepal's development approaches seem to focus on the construction of road as much as possible. It is taken as the synonymous for development. However, this study reveals that merely constructing a road across the district, particularly the SRN, does not necessarily contributes to the average GNI of the district.  Road projects required to be tied up with the other economic activities i.e. building cities and other infrastructure as in mid-hill highway project. But feeder roads have been useful in minimizing poverty as well as narrowing down the income disparities in the village. Two villages from mustang district, namely Lete and Lumar, have shown this evidence (See charlrey et al, 2015 for detail). 

The second important factor of government spending is in education. Human capital is an important factor of production since it enables the efficient use of other factors of production i.e. finance, technologies, land etc. This study reveals that investing in primary education without simultaneously providing opportunity of higher education do not contributes to the GNI.  It is convincing result since only the primary level education may not be useful in terms of providing productive input to the economic growth.

5.2    Policy recommendation

This paper analysed the determining factor of GNI disparities across districts in Nepal. Apart from generally accepted factors of GNI, a latent factor is introduced. Assuming that central level policy nexus ignores proper attention on local level needs and priorities which in turns results into relatively less budget allocation, a distance to district from centre variable has been introduced. Following this argument, it is hypothesized that farther the district the lesser will be the government spending and thus relatively lesser per capita GNI. Exploration of new factor GNI disparities in a setting of central planning and budgeting is believed to be the contribution of this work.  

 

As discussed, distance to districts has shown positive causal relationship with GNI disparities. Ordinary Least Square (OLS) estimate shows that districts with a kilometre away from Kathmandu causes per capita GNI to fall by eighteen rupees. Therefore, it suggests to include distance as one of the parameter while estimating the grant to provinces by National Natural Resource and Fiscal Commission (NNRFC).

 

As SRN revealed no causality with GNI per capita, this study warned to rethink the development philosophy in our context, at least the need for some policy adjustment.  It shows that merely constructing a SRN in the village does not necessarily improves the GNI per capita income associated with the districts. Therefore, sub-projects such as other infrastructures and/or model cities required to be tied up with the SRN projects. This has already applied in the Mid-hill highway project which need to be replicated to the other projects too.

 

Number of secondary schools indicates a positive contribution to the district GNI. Although it is an obvious result, it suggests policy circle to unveil policy instruments that mitigates the drop out ratio of students. 

 

At the end, it can be concluded that our planning and budgeting system has been significantly dominated by the interest of the central nexus composed of top level bureaucrats and political elites. Therefore, a transparent and truly bottom-up approach should be adopted in order to minimize uneven distribution of development. Furthermore, policy circle should expedite the process of   institutionalizing the framework of federalism which may address the current missing link of bottom-up approach of planning and budgeting mechanism.

Reference

Banarjee A., Kuri P.K. (2015). Development Disparities in India: an Inquiry into convergence.e-book series.

Cheroduan, R. and A. Thirwall (2015). Regional disparities in per capita income in india: Convergence or divergence?  Journal of post Keynesian Economics 37(3), 384-407.

Charlery, L. C., M. Qaim and C. Smith- hall (2015). Impact of infrastructure on rural household income inequality in Nepal. Journal of Development effectiveness 8(2), 266-286.

Dorn, F.C. Fuest and N. Potrafke(2019). Globalization and inequality revisited. CESifo working paper series 6859.

Kwatiah, N. (2021). Balanced regional development: Meaning and consideration. Website. htpp: ourworldindata.org/global-economic inequality.

Mainali, R. Mainali, S. and Mainali, R. (2022). Determining Factors of Inter-regional Disparities: an Analysis of per capita Gross National Income differential in Nepal. Paper presented on Ministry of Finance, Government of Nepal.

Seshanna,S. and S. Decornez (2003). Income polarization and inequality across countries: an Empirical study. Journal Policy Modelling 25(4), 335-358.

Stigliz J.E. (2019). People Power and Profit. Penguin Random House.

National planning commission (2021). Natonal planning commission introduction and history. Website. htpp:npc.gov.np/en/page/2/about_us/introduction_history.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Appendix 1: Variable Specification

       GNI                     Gross National per capita income of districts.

Distance              Distance to district from the capital city Kathmandu.

Population        Total population of the district.

Airport               A binary variable for airport carrying value1

                                   If a district has an airport; 0otherwise.

Tarai                 A binary variable carrying value 1

                             if a district belong to Tarai region; 0 otherwise.

 

Road Length       Total length of SRN road within the district.

 

---

[1]Mainali holds PhD from City University London, the United Kingdom. Dr. Mainali is working in the Ministry of Finance in a capacity of Undersecretary and is a freelancer researcher in the field of development economics. His corresponding email is rpmainali@gmail.com.