Energy consumption in Bangladesh  : :

Energy for rural households towards a rural energy strategy in Bangladesh

 2005

 

Prepared by M Asaduzzaman , Abdul Latif,  Bangladesh Institutes of Development Studies (BIDS)

 (Data source based  on a survey, mainly on households, have been conducted in 2003 on 2391 households in four divisions of the country )

1.0 Introduction

 

1.1 Rural Energy Starter study rationale

1.2 Objectives

 

1.3 Instruments for generation of information

 

2.0 Census findings

 

3.0 Characteristics of sample village

 

4.0 Characteristics of sample village house holds

 

5.0 Energy consumption

 

 

5.1 Introduction

 

 

In this chapter we describe, discuss and analyse the consumption of energy by households for domestic and other purposes (home business and agricultural operations). We also refer to the consumption by business enterprises outside home. We particularly analyse the factors that determine the incidence and level of consumption in households. While details supply factors have been analysed in the subsequent chapter, we have tried to show in the present one as to how these supply factors may influence the consumption behaviour.

 

5.2 Household Consumption of Energy by Type

5.2.1 Domestic consumption

 

All households use biomass and non-biomass types of energy (Table 5.1). Except for saw dust, most households use firewood, tree leaves and crop wastes. Most households, except in  Chittagong division, also use dried cow dung. Crop residues include various types of by products and crop processing wastes. Rice straw is the most frequently observed crop residue (55% of households) used for energy purposes. This is followed by rice husk (41%), rice bran (23%), jute sticks (28%), dried plants (12%), wheat straw (9%), dhaincha (8%), dried sugar cane tops and leaves (7%) and bagasse (3%).

            Among non-biomass energy, kerosene is the most frequently observed fuel   followed by electricity, from grids as well from dry cell batteries. But here are some interesting regional differences. Thus, grid electricity is most frequently observed in Dhaka closely followed by Chittagong. Khulna has the lowest incidence of grid electricity. As may be expected the use of dry cell batteries appears to be inversely related to frequency of use of grid-based electricity.

The above reported usage refers to all energy activities within the household. When only the usage for domestic purposes is considered, much of the firewood and other biomass are seen to be used mainly for cooking (Table 5.2). A substantial part of the crop residues is used for parboiling. On the other hand, much of kerosene is used for lighting. So is grid electricity. But it is also used for cooling purposes (fans mainly) and amusement. Dry cell batteries are used mainly for amusement purposes. There are certain inter-regional variations in the use pattern of various types of energy goods. But essentially the general pattern is the same as for the whole sample (Tables 5.2-1 – 5.2-4 in Annex).

 

5.2.2 Consumption for agricultural and transport  purposes

In agriculture and transport, electricity and diesel are the main energy goods in use (Table 5.3). Electricity is used almost entirely for running irrigation pumps and tube wells while a little is used also for threshing crops. Diesel is used mainly for transport but also in considerable amount for irrigation and power tillers/tractors. More people use diesel than electricity for irrigation pumps.

 

5.2.3 Consumption in business enterprises

For business enterprises whether at home or elsewhere within the village, non-biomass energy is used much more frequently (Table 5.4). But at home, some 20 percent of enterprises use no energy or their energy use can not be separated clearly from domestic use. Elsewhere within the village, almost all business enterprises use energy and non-biomass energy. Electricity from various sources, grid, storage cell and dry cell batteries, all are used to some extent.

 

 

 

Table 5.1

Percentage Distribution of Households by

Type of Energy Consumed within Household

 

 

Energy type	Dhaka	Rajshahi	Chittagong	Khulna	All
Bio-mass    Firewood    Tree leaves    Crop residue*    Dung cake/stick    Saw dust   Non-bio-mass    Candle    Kerosene    Natural gas    LPG/LNG    Grid electricity    Solar PV    Storage cell    Dry cell battery	99.8 85.1 81.4 81.4 56.9 0.7   100.0 2.5 99.0 - - 43.8 0.3 1.3 39.1	99.5 67.2 72.0 93.2 72.3 0.8   96.5 0.8 91.5 - 0.2 20.3 - - 50.7	98.6 95.8 61.6 53.6 29.5 0.3   100.0 10.6 98.4 0.9 1.1 38.9 - 0.8 42.3	100.0 88.7 91.8 75.4 64.8 1.6   100.0 4.4 100.0 - - 10.8 1.5 0.9 61.3	99.5 84.3 76.1 75.5 55.2 0.8   99.1 4.7 97.2 0.3 0.3 29.0 0.4 0.8 48.0
All	100.0	100.0	100.0	100.0	100.0

Note: * This also includes crop wastes and weeds; but mainly it is crop residue.

 

Table 5.2

 Consumption of Energy in Domestic Activities: All Divisions

(per household/year)

Type of energy	All use	Heating			Cooling	Lighting	Amusement
		Cooking	Parboiling	Other
Biomass: Firewood (kg) Tree leaves (kg) Crop residue (kg) Dung cake/stick (kg) Saw dust (kg)   Non-biomass: Candle (piece) Kerosene (liter) Natural gas (Tk.) LPG/LNG (liter) Grid electricity (kwh) Solar PV (kwh) Storage cell (kwh) Dry cell battery (piece)	1186.21 501.51 708.18 523.90 8.40     15.86 28.98 9.59 0.05 143.83 0.53 0.55 15.01	1064.84 470.67 538.86 503.68 8.36     - 1.76 9.59 0.05 0.25 - - -	28.60 29.99 164.41 16.07 0.02     - - - - - - - -	92.77 0.85 2.72 4.16 0.02     - 0.07 - - 4.00 - - -	- - - - -     - - - - 49.50 0.04 - -	- - - - -     15.86 27.16 - - 80.74 0.48 0.14 -	- - - - -     - - - - 9.34 0.01 0.41 ...

Notes:  Energy (only electricity) use for motive power is very negligible (0.28 kwh/ household/year). Hence it is not shown separately in the table. There are only 2 households (in Chittagong Division) which use briquettes only for 1-2 months. So, it is not taken into account. Only 1 household has been found to use charcoal but quantity is very negligible. Only 2 households use bio-gas. No household uses generator electricity.

                ... Information not separately available.

 

Table 5.3

 Consumption of Energy in Agriculture and Transportation

(Average over owners-users only)

Energy type	Agriculture			Transport
	Irrigation pump	Power tiller/tractor	Thresher
Electricity (kwh) Diesel (liter)	3643.47 (9)     347.95 (103)	- 238.24 (35)	78.13 (17) -	- 13003.55 (14)

 Note:  Figures in parentheses indicate number of household owning/operating the relevant equipment.

 

 

 

 

Table 5.4

 Consumption of Energy in Business Enterprises

Energy type	Home enterprise		Within village enterprise		All enterprise
	Users as % of total units	Energy use/unit/ year*	Users as % of total units	Energy use/unit/ year*	Users as % of total units	Energy use/unit/ year*
Biomass Firewood (kg) Tree leaves (kg) Crop residue (kg)** Dung cake/stick (kg) Saw dust (kg) Briquette (kg) Charcoal (kg)   Non-biomass Candle (piece) Kerosene (liter) Diesel (liter) Grid electricity (kwh) Storage cell (kwh) Dry cell battery (piece)	26.3 20.4 7.3 15.3 5.1 2.2 - 1.5   72.3 3.6 55.5 5.1 25.5 0.7 2.9	395.12 74.54 277.62 21.64 77.52 - 2.85     5.26 13.52 45.59 194.68 3.15 0.83	14.2 7.5 0.4 2.4 4.7 1.2 0.8 2.0   98.0 25.3 73.9 3.2 63.6 3.6 16.6	297.71 2.13 1282.53 201.42 8.18 17.31 5.34     23.55 88.25 46.54 486.70 16.22 4.66	18.5 12.1 2.8 6.9 4.9 1.5 0.5 1.8   89.0 17.7 67.4 3.8 50.3 2.6 11.8	331.93 27.57 929.52 138.26 32.54 11.23 4.46     17.12 62.0 46.21 384.12 11.63 3.31
All	79.6		98.0		91.5

Note:   *Average is over all enterprises by category.

**Also includes crop wastes and weeds; but mainly it is crop residue.

 

 

5.4 Patterns of Energy Use: Bi-variate Analyses

5.4.1 Introduction

So far we have discussed only the over-all energy use. One, however, also needs to know if there is any pattern in such use. For understanding this we try to find out the patterns by several categories of households. For brevity, we use only the whole sample. Several types of energy use are considered. These relate to firewood, tree leaves, crop residues, cow dung, kerosene and grid electricity as these are the major energy goods in use. In each case, we report the incidence of use of the particular type of energy good followed by the average quantity in use (averaged over all households, users and non-users). Before going into the details, a note of caution is in order.

            As noted earlier, the energy services, mainly heat for cooking and light, may be obtained from several types of energy goods which means that there is a likelihood of substitution between any pair of the energy goods. How far such substitution may take place depends on a host of factors apart from the ones whose effect we are trying to investigate here. Thus while we say that biomass use may increase as the value of a particular factor increases, between any two types of biomass, one may rise while the other fall depending on the other factors including some of the physical properties of the energy goods (such as heating properties) in question. With this caveat, the first set of patterns we look at are energy consumption by age of the household head.

 

5.4.2 Pattern of energy use by age of the household head

The basic hypothesis is that the comparatively old people are conservative and thus may adhere more to tradition the implication being that they may be more likely to use biomass and less of the modern energy such as electricity. While this appears to be borne out by use patterns for tree leaves, crop residues and cow dung, although the patterns do not seem to be monotonic, the hypothesis does not appear to be true for firewood (Tables 5.5-1 and 5.5-2). In this case, the use incidence does not vary by age of household head. Similarly the use of kerosene is almost universal. For electricity, however, the hypothesis is not all borne out. Households with older heads appear to use electricity more frequently and in more intensity. 

            Why should the pattern of quantity of firewood use be different from those of other biomass uses? A probable reason could be that it makes better common economic sense. If one has sources of supply of biomass, the more one can use the fringe biomass such as crop residues and tree leaves and other residues the better because it saves on cutting down the tree which is the way to get supply of firewood. And the older people may not wish to cut down the trees if they can do without it as some of these trees may have grown together with them from their childhood. We expect to probe these issues further in subsequent analysis. Of course, one may have to look into other factors such as purchase vs own supply or collection situation to understand the patterns or their lack.

 

Table 5.5-1

Incidence of Household Use of Energy

by Age of Household Head

(Percentage of user households)

            Note: “a” indicates statistical significance at 1% probability or less estimated using

                        Chi-sq test.

 

Table 5.5-2

Annual Average Household Energy Use Pattern for

Domestic Purposes by Age of Household Head

 

5.4.3 Pattern of energy use by education of the household head

It is expected that given other factors, the more educated a household’s head is higher will be the use of modern energy such as electricity while those of traditional fuels may fall. Table 5.6-1 shows the incidence of use of energy goods by education of the household head. In general those with more education appears to more frequently use the better quality biomass such as firewood and less of the lower quality ones such as tree leaves, crop residues and dung. The better educated ones also are more likely to use electricity but less of kerosene. Such results may be a reflection of the substitutability between kerosene and electricity for lighting which is a major use of both. Table 5.6-2 which shows the quantity of energy goods in use, indicates that that while there appears to be no clear pattern for other fuels/energy, the hypothesis regarding electricity appears to be borne out well.

 

Table 5.6-1

Incidence of Household Use of Energy by

Education of Household Head

 (Percentage of user households)

Note: “a” indicates statistical significance at 1% probability or less estimated using Chi-sq test.

              “c” indicates statistical significance at 10% probability or less estimated using Chi-sq test.

 

 

Table 5.6-2

Annual Average Household Energy Use Pattern by

Education of Household Head

 

5.4.4 Pattern of energy use by sex of the household head

Generally female-headed households are poorer than male-headed households in Bangladesh. In that sense, if income has a positive impact on energy use, female headed households should use less energy than male-headed ones. Table 5.7-1 indicates that in most cases except cow dung and electricity there appears to be little statistical difference in incidence of use although the patterns are opposite to each other. On the other hand, Table 5.7-2 bears out the hypothesis for quantity in use except for electricity. Why the pattern should be different for electricity is difficult to conjecture at the moment. But one aspect seems to be clear. Whether male-headed or female-headed, households apparently having higher than usual consumption usually have that for cooling (fan) purposes.

 

Table 5.7-1

Incidence of Household Use of Energy by

Sex of Household Head

(Percentage of user households)

Note: “a” indicates statistical significance at 1% probability or less estimated using Chi-sq test.

 

Table 5.7-2

Annual Average Household Energy Use Pattern by

Sex of Household Head

 

 

5.4.5 Pattern of energy use by family size

Cooking is the main purpose of energy use in rural Bangladesh. One would therefore expect that in general, incidence and quantity of biomass use will be positively related to family size. Bigger families may also mean more space to be lighted and cooled. Thus, the use of kerosene and electricity may also rise with increasing family size. Table 5.8-1 and 5.8-2 providing information on incidence and average household energy use in a year by type of energy use supports such hypotheses for individual biomass as well as total biomass. Note particularly the sharp rises in use of firewood and electricity with increase in family size. Kerosene use rises less sharply than electricity.

 

5.4.6 Pattern of energy use by land ownership

Land ownership in rural Bangladesh is a proxy for income as well as prestige. In that sense, one may hypothesise that the use of modern energy may rise with land ownership category while biomass may be used comparatively less. However, given the importance of biomass for cooking and parboiling purposes (the latter particularly being important for land-owning households as they may harvest more paddy than others), it is difficult to hypothesise what the use pattern may actually be. As Table 5.9-1 indicates, the patterns are indeed less clear cut and at least not monotonic, though it is so for electricity.

 

Table 5.8-1

Incidence of Household Use of  Energy

by Family Size of Household

(Percentage of user households)

Note: “a”, “b” and “c” respectively indicate statistical significance at 1%, 5%

and 10% probability or less estimated using Chi-sq test.

 

 

Table 5.8-2

Annual Average Household Energy Use Pattern by

Family Size of Household

 

Family size (no)	Fire wood (kg)	Tree leaves (kg)	Crop residues (kg)	Cow dung (kg)	All Biomass (kg)	Kerosene (litres)	Grid electricity (kwh)
1-2	828	502	535	370	2238	18	55
3-5	1168	505	656	494	2830	27	116
6-10	1272	492	781	575	3136	35	194
10+	1774	522	1678	1210	5185	48	554
Total	1186	501	706	524	2926	2.9	144

 

Table 5.9-1

Incidence of Household Use of Energy

by Land ownership of Household

(Percentage of user households)

 

Note: “a”, “b” and “c” respectively indicate statistical significance at 1%, 5% and 10%

           probability or less estimated using Chi-sq test.

                    

Table 5.9-2

Annual Average Household Energy Use Pattern by

Land ownership of Household

 

5.4.7 Pattern of energy use by source of income

Table 5.10-1 shows the patterns of incidence of energy use by source of income while Table 5.10-2 indicates the average levels of energy consumption over all households in the particular category. Note first that wage earners are at the bottom of the scale. Their access to electricity is most restricted. Only 8% or so have used electricity from grid during the reference year. On the other hand as they have no or little cultivable land their access to crop residues is also limited. Nor do they have, compared to others, recourse to firewood. This they try to make up by relying more frequently on tree leaves which is generally gathered. In contrast to the wage earners those depending mainly on agriculture appear to rely more often on crop residues. But similar to the wage earners they have a low access to electricity. However, interestingly, those having agriculture as one of the sources of income appear to have higher levels of consumption than others. While wage labour households are at most disadvantage note that those having non-agricultural income are not always much above the wage earning households. Possibly, agricultural occupations allow one to have access to various sources of biomass which may not be easily possible for pure non-agricultural households. On the other hand, the latter have much higher electricity consumption than others.

 

5.4.8 Pattern of energy use by level of income

The level of income has a positive influence on the use of energy be it biomass or modern fuels (Tables 5.11-1 and 5.11-2). Within biomass, however there appears to be a kind of substitution between firewood and non-firewood biomass, particularly at high levels of income, possibly because firewood is a superior fuel than tree leaves or crop residues.

 

Table 5.10-1

Incidence of Household Use of Energy

by Source of Income of Household

(Percentage of user households)

Income source	Fire wood	Tree leaves	Crop residues	Cow dung	Kerosene	Grid electricity
Mainly agriculture	74.7	62.3	71.0	48.7	98.3	16.1
Mainly non-agric	85.1	69.0	43.7	39.1	97.7	40.2
Agric & non-agric	86.8	79.2	78.4	57.5	96.9	31.9
Wage only	69.4	89.8	61.2	55.1	100.0	8.2
Total	84.3	76.1	75.5	55.2	97.2	29.0
Chi-sq statistic	45.4 a	60.2 a	66.0 a	20.1 a	4.1	56.8 a

Note: “a”, “b” and “c” respectively indicate statistical significance at 1%, 5% and 10%

           probability or less estimated using Chi-sq test.

 

Table 5.10-2

Annual Average Household Energy Use Pattern by

Source of Income of Household

Income source	Fire wood (kg)	Tree leaves (kg)	Crop residues (kg)	Cow dung (kg)	All Biomass (kg)	Kerosene (litres)	Grid electricity (kwh)
Mainly agriculture	1378	357	723	485	2945	30	48
Mainly non-agric	946	286	208	301	1741	24	227
Agric & non-agric	1169	548	739	540	3007	29	165
Wage only	655	351	205	628	1840	24	10
Total	1186	501	706	524	2926	2.9	144

 

Table 5.11-1

Incidence of Household Use of Energy

by Annual Income of Household

(Percentage of user households)

Note: “a”, “b” and “c” respectively indicate statistical significance at 1%, 5% and 10%

           probability or less estimated using Chi-sq test.

 

 

 

 

 

 

Table 5.11-2

Annual Average Household Energy Use Pattern by

Level of Annual Income of Household

 

5.4.9 Pattern of energy use by ownership of trees

So far we had been looking at basically demand side factors which may influence energy use. However there are also supply side factors such as access to trees or livestock which ensures ease of supply of say, tree leaves or cow dung and thus may influence their use. Table 5.12 shows the patterns of firewood and tree leaves use by no of trees owned. Except for the lowest group owning no tree and thus they must be gathering these from other people’s land or from trees, there is a clear positive relationship of ownership of trees on use of firewood and tree leaves. We shall later see that tree ownership in fact accounts for a far larger influence compared to demand side factors. This obviously has important policy implications.

 

 

Table 5.12

Annual Average Household Energy Use Pattern by

No. of Trees Owned

 

            Table 5.13 shows the use pattern of cow dung by the size of the cattle herd owned by the household. In general, the picture one obtains is that as the herd size increases so does the use of cow dung.

Table 5.13

Annual Average Household Cow Dung

Use by No. of Cattle Owned

 

5.4.10 Pattern of energy use by prevalence of students

Lighting for study at night is likely to be a major demand. Unless there is substitution between types of use (which may be the case if the consuming household tries to maintain a tight budget), the higher the number of students in the family, more is likely to be the fuel/energy consumed for lighting. There may be two caveats to this hypothesis. First, the number of students is expected to be correlated with the family size. So, the observed pattern unless controlled for family size may actually be a reflection of its (family size) influence. Secondly, there is divisibility in the service from lighting. Several people may use the same source of light in certain situations and this is likely to be truer for electric bulbs or tube lights. Hence, there may be no obvious pattern in the use of energy for lighting. Table 5.14 shows the pattern of use of kerosene and grid-based electricity for lighting and total domestic use.

            Note first that the proportion of energy used for lighting be it kerosene or electricity remains basically unchanged over student number categories. It is roughly 93% for kerosene and around 56% for electricity. This shows an apparent lack of substitution between lighting and other uses. On the other hand, the average use of energy for lighting rises as the number of students in the family rises which indicates that there may be a relationship between the two. Indeed, when the family size was controlled, the positive and monotonically rising relationship remained for both kerosene and electricity (not shown), particularly for the family size groups, 3-5 and 6-10 persons, which are the most numerous. Thus, the use of rural energy for lighting has obvious social benefits which should be borne in mind in formulating a rural energy policy or strategy.

 

Table 5.14

Patterns of Kerosene and Electricity Use by

Number of Students in Household

                                                                                                      

 

 

 

5.5 Multivariate Analysis of Energy Use Patterns

5.5.1 Patterns of incidence of use of energy

So far we have analysed the patterns of energy use by various factors. It is quite possible that not all of these factors will have an independent influence on these patterns. It may be that the observed relationship may actually be due to another variable with which the first variable is highly correlated. To find out the independent influences purged of those of others on energy use, we have therefore used multiple regression analysis. 

            For incidence of energy use, we first observe that among the biomasses, firewood is much more important than other types of biomass while among the non-biomass energy, kerosene is used by practically all although only less than a third uses electricity. We have therefore tried first to analyse the incidence of consumption of firewood and electricity from grid.

 

Incidence of firewood use

The incidence of use is a dichotomous response variable, either the household uses it or does not use it. Normally for various considerations either a logistic regression or a probit regression is used to analyse such behaviour. The results one obtains from these two equations are similar although the assumptions underlying the distribution of the dependent variable are different.[1] We have used both although the manner in which the explanatory variables enter the equations vary somewhat. 

Table 5.15 shows the results of the logistic regression estimates. Note first that we have initially included, among others, household head’s age, education, sex, family size and income as independent variables. However, the coefficients for the various categories under the variables were generally insignificant in all these cases. These were therefore dropped. The other factors that have been used as predictors are total land owned, source of income, number of trees owned by the household, livestock size of the household, the region (division) from which the household had been sampled and if the household was electrified. The details of the variables are shown in Table 5A.1 in Annex to this chapter.

            Before proceeding to interpret the results, we draw the attention of the reader to the earlier discussion on the hypothesised relationships between these variables on one hand and the incidence of use on the other. On the demand side, everything else remaining the same, broadly non-agricultural households as well as households depending on both agricultural and non-agricultural income have been found to be more than twice as likely as agricultural households to use firewood. There was no statistical difference, on the other hand, between agricultural and wage-dependent households.

            The relationship between size of land owned and firewood use is highly interesting. Earlier we hypothesised that it should be negative. With an increase in land size, the incidence of firewood use was expected to fall. The earlier bi-variate table (Table 5.9-1) did not show any clear cut relationship. Here, however, we find that when the influences of other variables are held constant, the incidence of firewood use indeed falls with rise in land ownership status except for the highest land ownership category. For example, households owning 250 to 500 decimals are almost 67 percent less likely to use firewood compared to those owning no land at all. The highest land category households are 32 percent less likely to use firewood compared to the reference category but as already pointed out the difference is not statistically significant.

            Electrified households are almost twice as likely compared to the non-electrified ones to use firewood. This indicates that electricity and firewood are in practice not substitutes although in theory they are. Rural electrification thus is no answer to lowering of firewood use for cooking or domestic purposes.

The potential availability of firewood as proxied by the number of trees owned has a highly statistically significant influence particularly for wood lots beyond a certain size. Compared to those having no trees, those owning only up to 10 trees shows either no discernible or only weakly significant difference in incidence of firewood use incidence. However, for those owning beyond 10 trees, the difference is very much substantial and highly significant. Thus, those owning 11-25 trees are almost one and a half times more likely than those owning no trees at all to use firewood for domestic energy purposes. For the largest woodlot owners, the corresponding figure is five and a half.

 

Table 5.15

Logistic Regression to Explain

Incidence of Firewood Use for Domestic Purposes by Households

 

Note:

1. Dependent variable: If used firewood (Used = 1, not used = 0)

2.In each case of categorical variable, the comparison is with the reference category which is the first category in each case.

3. Wald indicates Wald statistic distributed as chi-sq.

4. *, ** and *** indicate statistical significance at 10, 5 an 1% respectively.

5. Odds-ratios indicate how many times the particular occurrence is likely compared to the reference category. This is estimated as Exp(B) where B is the estimated regression coefficient.

 

The variable livestock size has been included to find out if substitutability has a role in firewood use. In general, the higher the livestock herd size, the lower is the incidence of firewood use compared to the benchmark (owning no livestock). However, the differences are statistically significant only for one group. Thus, while one may expect a kind of substitution between firewood and cow dung, the latter having various other uses, the actual scopes for such substitution is likely to be limited.

We have used a division variable to find out if there are regional differences when the influences of other variables are held constant. As expected, we find that compared to Dhaka, Rajshahi is in a worse situation while Chittagong is in a far more advantageous position. Indeed while Rajshahi households are only half as likely to use firewood, the likelihood of Chittagong households are 4 times as that of Dhaka households. Khulna is also somewhat in a better position than Dhaka households but not very prominently as indicated by the level of statistical significance of the difference. These results clearly confirm our earlier findings. These also indicate the need for differential policy and investment intervention decisions by broad geographic areas.

For the probit regression we have used the explanatory variables mainly in continuous form rather than as dichotomous or polytomous forms except for electrification status. The estimated regression is as follows:

 

FWI = 0.18 + 0.0065 Tree – 0.09 Live + 0.016 Home + 0.37 Elec + 0.43 Dhaka +

                           (6.34)             (4.24)           (3.63)             (4.32)          (4.98)

 

1.10 Ctg + 0.60 Kln

 (4.98)       (6.28)

 

t-statistic in parentheses. All variables significant at 1% probability.

 

 

 

where the variables are as described in the Table 5A.1 in the Annex to this chapter.

 

Note that the basic results of the logistic and probit regression are similar. More of trees owned, homestead land, and electrification raises the probability of firewood use. Also, Dhaka, Chittagong and Khulna households are more likely to use firewood compared to their Rajshahi compatriots. Livestock holding, however, exerts a negative influence.

 

 

Incidence of electrification

Ten independent categorical variables have been regressed upon electrification status of the household to examine their independent influence. Among them, sex of household head, family size, size of homestead wood lot and size of the livestock herd were found to have no statistically significant influence. The first one is an interesting result as we earlier found that female headed households were more likely than the male-headed ones to be connected to grid electricity supply. Here too, though statistically not significant, the male-headed households had about 24% less chance of being connected compared to the female-headed ones. Given that the result indicates the independent influence of sex of the household head, this may need further scrutiny in future research with a larger sample, probably a sociological one. The insignificance of the coefficients of size of wood lot and the livestock herd again indicate that electrification is not a substitute for biomass supply, at least at the present state of development in rural Bangladesh.

On the influence of family size on incidence of electricity connectivity it has been found earlier that there appears to be a strong positive effect of family size on such connectivity. Here, however, no such discernible independent influence of family size has been found. 

Earlier we observed that contrary to expectation households with older heads had more frequently been connected to grid electricity supply compared to those headed by younger persons. In fact, as the Table 5.16 shows, the independent influence of age of the household head may become stronger as age increases. Not only the statistical significance becomes stronger for the coefficients of higher age groups, but also the likelihood of being connected increases monotonically compared to the reference group. Thus, while those in age group 26-45 years are about 53% more likely to have electricity compared to the reference group, the oldest ones (above 65 years) are 89% more likely to be so connected.  As the influence of variables such as land ownership and level of income both of which have been included in the equation and thus their influences purged, it is difficult to speculate at this state of knowledge why one may observe such results.[2] 

            Education of household head exerts very strong independent and statistically significant influence on connectivity. While those households with heads having up to primary level of education are only 32 percent more likely than those headed by illiterate persons, the corresponding figure for households headed by persons with 6-10 years of

 

Table 5.16

             Logistic Regression to Explain Incidence of Electrification of Households