Mental accounting and remittances: A study of rural Malawian households

Journal of Economic Psychology 30 (2009) 321–334

Contents lists available at ScienceDirect

Journal of Economic Psychology journal homepage: www.elsevier.com/locate/joep

Mental accounting and remittances: A study of rural Malawian households Simon Davies *, Joshy Easaw, Atanu Ghoshray Department of Economics and International Development, University of Bath, Bath BA2 7AY, United Kingdom

a r t i c l e

i n f o

Article history: Received 10 July 2008 Received in revised form 23 February 2009 Accepted 15 March 2009 Available online 21 March 2009

Jel classification: D1 D12 O15

a b s t r a c t In this paper we use a behavioural approach to studying household consumption behaviour in Malawi. In particular we are interested to know whether households use mental accounting when consuming different categories of good. It is useful for assessing the impact of remittances on household consumption behaviour. We use 1998 cross-sectional data to find the following key results: (i) mental accounting systems are in operation; (ii) remittance income exhibits a lower marginal propensity to consume than other income sources, (iii) remittances are widely used to fund education consumption, (iv) credit plays an important role in funding education and farming. Ó 2009 Elsevier B.V. All rights reserved.

PsycINFO classification: 3920 Keywords: Remittances Household behaviour Consumer economics Economic development Africa Malawi

1. Introduction Remittances are commonplace in Malawi with over 20% of households receiving an average of 43% of their total non-business income from this source.1 This reflects the importance of such transfers for developing countries in general (Gammeltoft, 2002; Ratha, 2003) for which studies have shown that households do not use remittances in the same way as other income sources (Adams, 1991; Adams, 2005; Adams, Cuecuecha, & Page, 2008). The simple receipt of remittances may also be capable of modifying households’ consumption choices (Cox Edwards & Ureta, 2003). Given that remittances alter household expenditure, studies seek to test and measure the impact of these transfers on household consumption choices despite the fact that this is in conflict with economic theory of consumption such as the lifecycle-permanent income hypothesis (Friedman, 1957; Modigliani, 1966) which suggests that the Marginal Propensity to Consume (MPC) out of all income sources should be equal. Evidence shows that this is not the case, and credit constraints which prevent borrowing out of future income are often cited as an explanation (e.g. Hayashi, 1987; Zeldes, 1989). This * Corresponding author. E-mail addresses: [email protected], [email protected] (S. Davies). 1 Authors’ calculations using Malawian Integrated Household Survey (1998). Chipeta and Kachaka (2005) calculate that remittances accounted for 6.3% of total daily per capita consumption in Malawi in 1998. Thus, even including business expenditure, this flow of income represents an important flow of income for Malawian households. 0167-4870/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.joep.2009.03.003

322

S. Davies et al. / Journal of Economic Psychology 30 (2009) 321–334

paper however, considers equally liquid income meaning that liquidity constraints are not a suitable explanation for any observed MPC differences. Adams (2002), shows that households exhibit a higher marginal propensity to save out of more risky income sources. Although not explicitly stated, this might be seen as a specific case of mental accounting in which households consider the riskiness of each income source individually, rather than that of their total income.2 In the spirit of Adams (2002), we explicitly propose mental accounting as an alternative explanation to liquidity constraints for the observed differences in marginal propensities to consume out of remittance income compared to other income sources. Under this theory, derived from Shefrin and Thaler (1988), households keep different financial accounts (real or metaphorical) out of which different goods are consumed. Income is allocated to one account or ‘‘pot” or another depending partly upon its source, allowing us to observe different marginal propensities to consume different goods out of each income source. Remittances lend themselves to an analysis in the mental accounting framework. In some cases they come with implicit or explicit conditions attached (‘‘use this money to educate my little brother”), in other cases they are used as a form of income pooling, mutually reducing risk (e.g. Dercon, Hoddinott, & Woldehanna, 2005; Harrower & Hoddinott, 2005) and helping to smooth consumption3 potentially altering consumption behaviour. Remittances may thus be used for or encourage investment in ‘‘useful” areas such as education, health, nutrition and savings, or may be seen as ‘‘manna from heaven” and encourage non-productive behaviour (Kozel and Alderman, 1990 in Chami, Fullenkamp, & Jahjah, 2005). Each of these ‘‘mental tags” suggest that remittances would be placed in different mental accounts. Conditional remittances are often related to education, and, despite the fungibility of remittance income, and the inability of the remitter to observe their use directly, might be placed in an ‘‘investment account”.4 Remittances seen as ‘‘windfalls” (as, for example pool wins are) might be placed in an alternative, ‘‘frivolous consumption account”. The results of a study of the uses of remittances are therefore not obvious, a priori, and likely to differ depending upon the culture and context. The results presented are therefore a demonstration of mental accounting in a developing economy context with results being most relevant the closer the cultural context is to that of Malawi (i.e. Sub-Saharan Africa). It is an empirical question as to whether or nor mental accounting models are the same in other contexts or not. Despite this, studies of the impact of remittances in other developing contexts concur with the results presented in this paper and suggest an important link between remittances and education. This is the first time that the mental accounting hypothesis has been advanced as a hypothesis for differing marginal propensities to consume out of different sources of income for a developing economy. Mental accounting is important for government policy as well as for NGOs and banks trying to mobilise savings and encourage borrowing. If lack of consumption out of illiquid assets is a choice and not forced upon individuals, microfinance institutions need not only to provide liquidity, but also change consumption and savings behaviour. It is important to understand whether remittances are combined with other sources of income or spent in a particular way. If they are used for different purposes to money from other sources, do these purposes tend to be constructive (such as education) or destructive (conspicuous consumption)? This paper focuses on two key questions: Do households’ spending choices conform to traditional consumption models in which source of income plays no role, or do households keep ‘‘mental accounts” consuming differently out of different money pots? If households keep mental accounts, how are remittances used? Levin (1998) uses American longitudinal data to find marginal propensities to consume (MPC) for different categories of goods out of different assets. He finds that the MPC out of current income is around 0.42 with a MPC out of changes in housing value of zero. This is a common empirical result which is seen as an anomaly of the lifecycle consumption model. Credit market constraints are often cited as an explanation,5 but Levin suggests that individuals are not credit constrained but rather choose not to consume out of these assets. He uses testable differences between the models which allow him to discriminate between lifecycle consumption models with liquidity constraints and behavioural models of consumption and finds in favour of the latter. Furthermore households use different wealth categories to purchase different goods; for example, they are more likely to use liquid wealth such as savings than current income to pay for occasional purchases such as vacations. Although Levin’s findings support the fact that individuals choose and are not forced to consume differently out of assets with different levels of ‘‘temptation”, he does not break down current income into different categories. Thus, he is unable to test whether equally liquid income is used for different purposes. Adams (2002) uses panel data from a sample of 469 rural Pakistani households to measure marginal propensities to save and consume out of income from different sources. He finds that the marginal propensity to save out of remittances is higher (at 0.711 for external remittances and 0.589 for internal remittances) than that for any other source of income. Although he notes that these results do not conform to unmodified lifecycle consumption models, Adams suggests that is due to income volatility and risk aversion, noting that income sources with greater variability exhibit greater marginal propensities to save. 2

We would like to thank an anonymous referee for pointing this out. See Alderman and Paxson (1992) for a synthesis of the literature on risk and consumption in developing countries. 4 ‘‘Conditional remittances” present potential issue for this study. Mental accounting implies a ‘‘free mind” such that the decision to place income from a given source in a particular pot is made freely. Unfortunately, the data do not allow us to know which remittances came with implicit or explicit conditions attached. Despite this issue, we believe this problem to be minimal for three reasons; firstly, only a part of all remittances would come with such conditions attached, secondly, there exists an important degree of information asymmetry between the sender and receiver of remittances regarding their uses (Miller & Paulson, 1999, find that in Thailand, remittance senders were more likely than receivers to cite education as the main purpose of their remittances) and thirdly, this paper assesses liquid income which is, by its nature, fungible, and can be used for any purpose. We would like to thank two anonymous referees for raising conditional remittances as an important issue to flag. 5 See Hayashi (1987) or Zeldes (1989). 3

S. Davies et al. / Journal of Economic Psychology 30 (2009) 321–334

323

While Adams is able to distinguish between different MPC out of sources of equally liquid income, he does not look at how this income might be spent. This paper combines these two approaches, testing traditional lifecycle models against behavioural consumption models of mental accounting. It goes on to look at how different sources of income are spent with a particular focus on remittances. The results show that, as in Adams (2002), remittances are more likely to be saved than income from other sources. Furthermore, they are most likely to be used to finance education. Credit is also important to fund education and farming.6 Households do choose to consume small amounts out of fixed assets with males and female headed households using these assets differently. Adams (2005) and Adams et al. (2008) study the impact of remittances on MPC different categories of goods out of total income in Guatemala and Ghana respectively. In order to understand whether remittance receivers behave differently from other households interaction between total income (proxied by total consumption) and indicator variables for whether or not remittances were received. For Ghana, Adams et al. (2008) find that there are no significant differences in behaviour between households that receive internal remittances and those that do not receive remittances. However, those that receive international remittances have higher marginal budget shares dedicated to education and lower marginal budget shares dedicated to food than other households. For Guatemala, Adams (2005) finds that remittance receiving households have higher MPC education and lower MPC food than other households. Although these findings are of significant interest, and are largely supported by the current study, Adams (2005) and Adams et al. (2008) do not suggest a theoretical framework for these results. This paper offers mental accounting as a logical explanation for the impact of remittances. The paper is organised as follows: Section 2 briefly outlines the theoretical framework. Section 3 discusses the data before proceeding with the empirical analysis and discussion of the results. Finally, the summary and concluding remarks are drawn in Section 4. 2. Theoretical framework The behavioural lifecycle model from which this hypothesis is drawn is proposed in Shefrin and Thaler (1988). Here, we outline the main results of their model and offer supporting evidence. Since the main contribution of this paper is empirical, we briefly elucidate our extensions to the model, but refer the reader to the seminal Shefrin and Thaler (1988) paper for more details. Shefrin and Thaler (1988) divide assets into different categories associated with different levels of temptation to consume. Unlike in the traditional lifecycle model, consumption is not only a function of total lifetime wealth, but also of the composition of that wealth.

C ¼ CðY; A; FÞ

ð1Þ

Specifically, they divide wealth into three mental accounts with different temptation levels: current spendable income (Y), current assets (A) and future income (F). Marginal Propensities to Consume (MPCs) differ across the categories7:

@C @C @C – – @Y @A @F

ð2Þ

where MPCs are given by the partially differentiating (1) with wealth and consumption values. Shefrin and Thaler (1988) and Lewis and Winnett (1995) use surveys8 to show that the source of the income and the amount of income are both important in placing income in one account or another. A windfall gain is likely to be placed in the asset account (perhaps savings) while several small gains adding up to the same value tend to be placed in the current income account, even when both of these income gains are anticipated. Analysing the claim that the large bi-annual bonuses which are the norm in Japan contribute to the comparatively high savings rate, Ishikawa and Ueda (1984) estimate MPC from regular and bonus income for Japanese workers. They find that for non-recession years, MPC is significantly higher for regular income than for bonus income (0.685 versus 0.437); Japanese households habitually save over half of their bonus income and these authors conclude that, at least in the short run, habits govern household consumption patterns, while Friedman’s Permanent Income Hypothesis (PIH) may be more relevant in the longer run. This model is extended by Levin (1998), who allows for separable assets. In Shefrin and Thaler (1988) all assets are combined – an increase in the value of one’s home has the same impact on consumption as a stock market gain. Levin (1998) shows that the marginal propensity to consume out of different assets differs. Consumption of good g is thus a function of income, Y and the different assets held:

C g ¼ ðY; A1 ; A2 ; . . . ; Ak Þ

ð3Þ

Marginal propensities to consume out of different assets vary according to their temptation level: 6

Udry (1990) finds that credit is used for consumption smoothing purposes in northern Nigeria. For simplicity, throughout this paper, we use the non equal notation, however, it should be noted that mental accounting implies that it is possible for the MPCs not to be equal, but not that they necessarily have to be. It is possible for MPCs out of two sources of income be equal, but that mental accounting does still exist. Thus, the tests presented in this paper can be seen as particularly rigorous tests for the existence of mental accounting. We would like to thank an anonymous referee for raising this point. 8 In the United States and Netherlands, respectively. 7

324

S. Davies et al. / Journal of Economic Psychology 30 (2009) 321–334

oC oC oC – – and oA1 oA2 oAk

oC oC – oY oW

ð4Þ

where W is total wealth. In addition, each asset is used to fund different expenditure, so that the MPC different goods out of the same asset is not equal:

@C 1 @C 2 @C G – –:::– @Ak @Ak @Ak

ð5Þ

Levin (1998) postulates that the MPC out of liquid assets will be higher than for housing assets since income has a high temptation level. Put another way, there are ‘‘psychological as well as financial transaction costs associated with spending from different types of assets”.9 In mental accounting terminology, liquid and illiquid assets are placed in separate mental accounts (used to purchase different goods, or goods at different times) and there is a psychological cost to transferring purchase power between accounts.10 Levin (1998) also tests the hypothesis that liquidity constraints are responsible for the differing MPC rather than behavioural mental accounting reasons but concludes in favour of a behavioural explanation. This paper extends Levin’s (1998) analysis to include equally liquid income from a variety of sources and tests the mental accounting hypothesis for a developing country, Malawi. Formally, a household has J income sources and owns K categories of assets:

C g ¼ C g ðY 1 ; Y 2 ; . . . Y J ; A1 ; A2 ; . . . ; Ak Þ

ð6Þ

In addition to allowing MPC out of different wealth categories to differ as in Levin (1998), the MPC out of different, equally liquid income sources are not equal. As required by the behavioural lifecycle model, the different MPC results, at least in part, from internally (not externally) imposed constraints. The varying MPC are due to behavioural reasons such as mental accounting. More specifically agents voluntarily choose to spend differently out of different income sources so that a oneunit increase in current wages is not treated in the same way as the same increase in current remittance income.

oC oC oC – –– oY 1 oY 2 oY J

ð7Þ

This paper seeks to analyse remittance income in the mental accounting framework. Sources of income are credited to different mental accounts and remittance income in particularly is likely to be allocated to its own mental account for a number of reasons. Firstly remittances may come with specific conditions attached. The remitter may require the receiving household to use their income for purposes such as education or else risk losing this income. Secondly, remittances may be considered as either manna from heaven or else the product of someone else’s hard work. How remittances are viewed is influenced by culture (e.g. Colloredo-Mansfeld, 2005; Hart, 2005), and the motivation behind the remittance, and impacts on the account into which these transfers are placed. In our first example, remittance income is likely to be placed in an account used for general or even luxury consumption. In the second, remittances tend to be used for productive or constructive purposes such as education11 or business investment. Thus, as in Levin (1998), households consume differently out of different mental accounts:

@C 1 @C 2 @C G – –:::– @Y 1 @Y 1 @Y 1

ð8Þ

3. Empirical analysis and results 3.1. Data We use data from the Malawian Integrated Household Survey carried out from November 1997 to October 1998, which, after cleaning contains a representative sample of 5644 rural households across Malawi. The data include detailed income and consumption variables as well as a wide range of household characteristics. The average household has 4.3 members or 3.7 members in per adult equivalent terms. 77% of households have an average of 2 children. 26% of household heads are female and the average age is 41 years. 73% of household heads are married and 45% work in agriculture. Consumption is classified into food; education; health; farm; household (e.g. personal and household hygiene, communication, transport); clothing; and fuel categories as well as total non-durable consumption.12 All income and consumption 9

Levin (1998, p. 60) Various papers have analysed different psychological aspects of mental accounting theory. Shefrin and Thaler (1988) explains the basic concepts. Ainslie (1975) discusses impulse control and Brocas, Carrillo, and Dewatripont (2004) review commitment devices. Karlsson (2003) discusses practical strategies to commit to spending patterns, and Anderson and Nevitte (2006) find that saving behaviour is largely a matter of habit, learnt at home as a child. 11 See, for example Cox Edwards et al. (2003). 12 There is no information on leisure consumption since the low income levels in rural areas mean that there is almost no consumption which could be categorised unambiguously as leisure consumption. The small amount of leisure consumption that certainly exists is likely to be unreliable and under-reported since rural Malawians do not like to report activities such as alcohol consumption or drinking in tea rooms to researchers. Reported leisure consumption is likely to be largely included in the ‘‘general household” category which includes travel, newspapers etc. It is difficult to know however whether consumption such as travel is leisure consumption or not. 10

S. Davies et al. / Journal of Economic Psychology 30 (2009) 321–334

325

values were annualised, placed in per adult equivalent terms and adjusted for the regional price level.13 Fixed assets include land and housing, and liquid assets include livestock and items such as bicycles and household appliances. Mean income, consumption and asset ownership data as well as information regarding household characteristics are presented in Table A1. Average reported non-business household income in per adult equivalent terms is Malawian Kwacha14 (MK) 1866 annually with average consumption equal to MK2063. Income sources are varied with many households receiving income from several sources. Furthermore, many households also reported significant non-cash consumption (largely home produce) with this being equal to around 54% of total consumption (home produce plus expenditure) for remittance receivers and 51% for non remittance receivers. This should not come as a surprise in an agricultural economy such as Malawi’s where many households produce a large proportion of their own food consumption. The analysis however, focuses on potentially fungible cash income and consumption. Around 22% of households reported receiving remittance income during the month preceding the survey. The mean yearly income from this source was MK1409 for those that received them. Excluding business income, mean remittances are worth around 43% of the average total income of receiving households. They are thus an important source of income for these households. Most households reported owning both liquid assets such as livestock or household appliances and illiquid assets such as housing or land. The average value of fixed assets was MK2119 or 114% of average non-business yearly income. 3.2. Econometric methods and empirical results 3.2.1. Total consumption Based on (6) and (7), estimation of MPCs are made using ordinary least squares, with each category of income entered separately as required by each test:

C i ¼ b0 þ b01 Y ij þ b02 Aik þ b03 X þ e

ð9Þ

Where Ci is the ith household’s total consumption (excluding durable goods) in per adult equivalent terms; and Yij denotes the household income from the jth income source, and Aik denotes the value of fixed and liquid assets owned by the household. Finally X represents a vector of control variables including household characteristics such as age of the household head, education level of household head and regional dummies included to capture systematic differences between regions due to preferences or culture and e is the error term. All error terms are adjusted for potential heteroskedasticity using White (1980). We estimate models for male and female headed households separately. This is important to minimise the potentially large impact of intra-household bargaining relationships. Since a large proportion of household remittance income is likely to accrue to the female (say, from husbands working away from home), this may give the female greater power over household spending decisions, producing different MPC out of different income sources. Analysing separately male and female headed households minimises the risk of this.15 In addition, we have chosen to analyse only rural households. Rural and urban households differ widely in their consumption preferences, goods available, and their access to credit and other sources of income. Focusing on rural households (which make up around 85% of the Malawian population) makes the study more relevant for this group and minimises the potential for a large degree of unobserved heterogeneity distorting the results. As well as making sense econometrically, focusing on rural households captures the behaviour of the majority of the population and makes results potentially more comparable with other studies which also focus on rural areas (e.g. Adams, 1991, 1998).We regress total consumption against each of the income variables entered separately; fixed (illiquid) and liquid assets; a series of dummies for different sources of income and control variables. Mental accounting theory suggests that the MPC out of different sources of income is not identical even where the income is equally liquid. That is, households choose not to treat income from different sources identically.

@C @C – @Y i @Y j

ð10Þ

Results are presented in Table A2: The coefficients on each income source represent marginal propensities to consume. Male headed households (MHH) exhibit a MPC out of salary income of 0.331, a MPC out of farm income of 0.484 and a MPC out of remittance income of 0.066. Female headed households (FHH) have a MPC out of salary income of 0.579, a MPC out of farm income of 0.796 and a MPC out of remittance income of 0.202. All coefficients on liquid income are statistically significant, and two salient points

13 The survey was carried out from November 1997 to October 1998 during which time the country experienced a relatively high inflation rate: International Financial Statistics show an inflation rate of 29.75% during 1998. During the survey, information was collected on local prices in each of the regions where the survey was carried out. This information was then used to construct monthly food, non-food and total price indexes for each region. These price indexes correspond more closely to the purchases of the households surveyed and are more detailed than the inflation data collected by the Reserve Bank of Malawi. All monetary values are adjusted according to the time the household was surveyed and the region in which they are situated. 14 During the period of the survey US$MK70. Note that only rural households are analysed where the price level is typically around one third that of urban areas. 15 See, for example McElroy (1990) or Katz (1995) for discussions of intra-household bargaining.

326

S. Davies et al. / Journal of Economic Psychology 30 (2009) 321–334

emerge. Firstly, MPCs are consistently larger for MHH than for FHH. This is unlikely to be due to lower income levels amongst FHH than their MHH counterparts. Average consumption is MK2081 for MHH and MK2013 for FHH. Secondly, formal tests strongly reject the null hypotheses that MPCs for remittance income and salary income are equal, and that MPCs for remittance income and farm income are equal, for both MHH and FHH. The null hypothesis that salary income and farm income cannot be rejected for either set of household. Despite this, there are strong initial indications that remittance income is used differently from income from other sources. These results are in line with other studies on developing countries. For example, Adams (2005) finds marginal propensities to consume of between 0.54 and 0.59 for Guatemala. The same author finds MPC out of total income of 0.149 for rural Pakistan (Adams, 2002). Both MHH and FHH exhibit a small but positive and statistically significant MPC out of fixed assets (0.071 and 0.065, respectively) whilst only MHH have a positive and significant MPC (of 0.098) out of liquid assets. The loan dummy is positive, as one would expect; all other things being equal, accessing credit increases consumption. The business income dummy is positive and significant for both sets of household, but is around double the size for MHH compared with FHH. This is indicative of the fact that males who run businesses tend to run larger businesses than their female counterparts. Farming income reduces overall consumption for both MHH and FHH. This is potentially due to the fact that those with farming income have less need to purchase food produced outside of the household, reducing reported consumption. The dummy indicating that the household lives in the north of the country is positive and significant for MHH, but not FHH. MHH in the north consume more per adult equivalent than their central and southern counterparts. Finally, consumption is highest during the hungry season. This is unsurprising as, during this period food prices tend to rise as stocks fall, requiring greater expenditure. 3.2.2. Remittance uses The LC-PIH model posits that the marginal propensity to consume any given category of good will be equal for each income source and change in wealth. Income from a given source is not allocated to a particular consumption category. By contrast, behavioural mental accounting models suggest that mental accounts are used for specific purposes. Since income from different sources is assigned into different mental accounts, the MPC for good (g) out of one source of income will not equal the marginal MPC good (h) out of the same source of income. And the MPC for the specific good (g) from source (i) will not equal the MPC (g) out of (j).

@C g @C h – ; g–h @Y i @Y i

ð11Þ

@C g @C g – ; i–j @Y i @Y j

ð12Þ

The adult equivalent consumption for each of seven categories is regressed against each income source separately, relevant income dummies and control variables as in the previous section. In addition, dummy variables controlling for total consumption quartile are included (with the lowest consumption quartile omitted). This is necessary due potentially different consumption preferences between households with different total income levels. Therefore dummies indicating total consumption quartile are included to capture the fact that those with high incomes are likely to behave differently from those with lower incomes. Finally, for each of the categories of consumption analysed, some households reported zero consumption levels. These observations are treated as censored at zero making the Tobit model appropriate. Since we wish to obtain unbiased estimates of the underlying population parameters (rather than only for households with positive consumption), we focus on the MPC given by the unadjusted coefficients in the Tobit models.16 As in the previous OLS estimates, error terms are adjusted for potential heteroskedasticity.

C g ¼ b0 þ xb þ u

ð13Þ

C g

where is the latent variable which can take any positive or negative value and satisfies all classical linear model assumptions. However, we observe Cg only when C g is positive. That is, we observe consumption of category of good g only when these are positive. Where the latent variable is zero or negative, a zero value is observed. Thus:

C g ¼ maxð0; C g Þ

ð14Þ

The b0 term represents the intercept, x contains all explanatory variables and b is a vector of coefficients. The probability of any Cg being equal to zero is:

PðC gi ¼ 0jxi Þ ¼ 1  Uðxi b=rÞ

ð15Þ

where the subscript i indicates household i, U is the standard normal distribution and r is the standard error of the regression. The estimated Tobit model is:

16

See McDonald and Moffitt (1980) and Wooldridge (2002) (pp. 525–529) for relevant discussion of Tobit models.

S. Davies et al. / Journal of Economic Psychology 30 (2009) 321–334

EðC g jC G  0; xÞ ¼ xb þ r

/

U

ðxb=rÞ

327

ð16Þ

where /=U is the ratio between the standard normal probability density function (pdf) and the standard normal cumulative density function (cdf) or the inverse Mills ratio (Wooldridge, 2006: pp.595–598). As before all models are estimated separately for male and female headed households, with results being strongly supportive of the mental accounting hypothesis. Results are presented in Tables A3 and A4: Male headed households exhibit positive and significant MPC out of salary income for food and general household items. At the margin, out of an additional MK100 of salary income, just over MK5 will be spent on food, and just under MK10 on household items. For MHH, the only positive and significant MPC out of remittances is for education. This is evidence that remittance income goes, at least in part, into a ‘‘pot” to fund education. In addition, there is a positive relationship between the dummy indicating remittance receipt and education consumption. Other things being equal MHH which receive remittances also spend more on education. Receiving remittances has no discernable impact on any other consumption category. There is a large negative (0.120) MPC food out of farming income. This should not be a surprise; households with farming income are also likely to consume home grown food, reducing reported spending on food. These results are supported by the negative relationship between acres owned and food expenditure, and the positive relationship between acres owned and farm expenditure. In addition, farming income is used to fund farm expenditure (MPC = 0.087) and clothing (MPC = 0.092). Fixed assets are used to fund farming, and liquid assets fund education and household items. The loan dummy is positive and significant for education, health and farm consumption, with the value on education being around four times as great as the next highest coefficient. Those that accessed credit spent nearly MK200 more on education than those who did not, other things being equal. Loans are important in the funding of education, health and farm consumption. It is possible that emergency health expenditure is funded through short term loans, whilst loans and repayment to fund farm expenditure will follow the agricultural cycle. Finally, it is interesting to note that the constant, representing autonomous consumption, is positive and significant for food, but either negative or insignificant for every other category of consumption, suggesting that food is the major necessity in the rural Malawian economy. Turning to FHH in Table A4, we find that salary income is used to fund household items (MPC = 0.165). As with MHH, remittances are used to fund education with a positive and significant MPC of 0.103. Thus, at the margin, an additional MK100 of remittance income would lead to an increase in education consumption of MK10.3. In addition, a small part of remittance income is used to fund fuel consumption (MPC = 0.025). At the margin, increased remittances income is actually negatively associated with farming spending. This potentially captures some degree of moral hazard whereby increased remittances decrease the need to work on the family farm. Although we are unable to investigate this hypothesis further, several authors have found that remittances can have a negative impact on labour market participation (e.g. Azam & F.Gubert, 2004; Lucas, 1987). It is interesting that this result holds for female but not male heads. This might reflect the fact that, in rural Malawi, females tend to be responsible for farm work whilst males seek wage labour. If income or consumption from these are substitutable, then female heads who receive remittances from husbands/sons working elsewhere might choose to decrease their own farm work. An increased value of fixed assets is associated with a small increase in food consumption (MPC = 0.027) and consumption of household items (MPC = 0.019). Like their male counterparts, FHH access credit to fund education and farm expenditure, and acres owned is negatively associated with food expenditure but positively associated with farm expenditure. For both male and female headed households, the hungry season dummy is positive and significant for food, reflecting increased expenditure during this period following price rises. For both groups of household, consumption quartiles dummies are designed to remove the impact of total wealth from the MPCs due to both differing preferences, and economic behaviour summarised by Engel’s law. They indicate that for all categories of good, the lowest quartile consumes the least in per adult equivalent terms rising to the highest quartile which consumes the most. Northern males spend more on education than their central and southern counterparts. This could be interpreted from a historical or cultural perspective; the first schools in Malawi were set up in the north of the country by missionaries, and the ‘‘learning culture” still exists today (McCracken, 2002). However, the same is not true for FHH for which those in the south consume more education than those in the centre and north. Gender role differences between Malawian regions are important with the north being largely patrilineal, the south largely matrilineal, and the centre mixed (but with a slight matrilineal tendency). Thus, females are likely to have more of a say in household consumption in the south, and males in the north. It is therefore interesting that whoever has more control over household resources appears to favour education when they are in charge. It is possible that in the central region responsibility for educating children is not so clearly defined, and therefore suffers from a sort of ‘‘tragedy of the commons” during the intra-household bargaining process.17 It is interesting to focus on the impact of the head’s education on education expenditure. For FHH, being literate, or any education level above the baseline of no education and illiterate raises education expenditure. The strength of the impact rises with education level up to 17

We would like to thank an anonymous referee for raising this issue of gender and regional differences.

328

S. Davies et al. / Journal of Economic Psychology 30 (2009) 321–334

secondary level. A similar pattern can be observed for MHH, but the literacy and other education dummies are insignificant. This provides evidence that, even after controlling for incomes, assets and consumption quartiles, the education level of the household head has an important positive impact on household consumption of education.

4. Additional mental accounting issues: wealth group and remittances sources It seems reasonable that different wealth groups may use income differently. This was tested in for an inferior good (food) and a superior good (education) for the top and bottom consumption quartiles. These results are summarised in Table A5 and full regression results are available from the authors on request. Although little additional information was gained for FHH, the results are informative for MHH and demonstrate different behaviour depending upon consumption level. In particular, it was found that remittances are used to finance education by both the top and bottom consumption quartiles. In addition, the bottom quartile uses remittances to fund the purchase of food. This is likely to reflect the fact that remittances might be given due to altruism to the poorest households, who require this form of income in order to meet basic survival needs. Salary income is used to fund food consumption by the top quartile, who are more likely to purchase (rather than grow) the majority of their food consumption. Finally, it was interesting to examine the uses of remittances separately depending upon whether the household reported receiving them from rural Malawi, urban Malawi or abroad. Table A6 summarises these results.18 For both MHH and FHH, remittances from rural and urban Malawi exhibited a positive and significant impact on education consumption. Otherwise put, internal remittances are used, at least in part, to fund education, whilst external ones are not. This is an interesting result as there are likely to be key differences between remittances from the different areas. Information asymmetry between the sender and receiver is likely to decline by distance. That is, if remittances are sent with explicit or implicit conditions attached, the sender is more likely to be able to verify their use the closer they are to the receiver. It is therefore informative that, for MHH, the MPC education out of remittances declines as the distance of the remittance increases. Female headed households however, have a higher MPC education out of remittances from urban Malawi than from rural Malawi, suggesting that this cannot be the whole story. In addition, transport difficulties in rural Malawi mean that it is there is unlikely to be a large difference in issues related to information asymmetry between urban Malawi and abroad. Remittances from closer to home tend to be more regular and lower in value due to the transaction costs involved. Internal remittances are therefore potentially viewed as a source of regular income, whilst remittances from abroad might be classified as windfalls. Mental accounting evidence suggests that income from these sources should be placed in different ‘‘pots”, and therefore used for different purposes. In addition, for MHH, remittances from urban Malawi fund clothing whilst those from rural Malawi are spent on fuel. FHH use both rural and urban remittances to fund fuel consumption. For both groups of households, foreign remittances have a negative impact on food consumption.

5. Summary and concluding remarks The results offer support for the use of mental accounting models in Malawi. Households do not, in general, lump all income together, but choose to allocate income differently. This is the case even after analysing male and female headed households separately to minimise the influence of intra-household bargaining. There are some differences between male and female headed households, however, they concur in one notable respect: both choose to allocate remittances towards education. This paper has extended the theoretical work of Shefrin and Thaler (1988) and Levin (1998) in order to show that households in a developing country choose to consume differently out of equally liquid forms of income. The findings support the implicit assumption in many studies of remittances that households regard this flow of income as distinct from others and thus choose to use it differently. Furthermore it offers evidence that the reason for this lies, at least in part, in mental accounting. The results are consistent with Adams (2002) who finds that households are more likely to save out of remittance income than other sources, but offers an alternative explanation. In addition to simply analysing the consumption/savings trade-off, we extend Adams (2002) work by looking at how remittances are spent and how their receipt may alter behaviour. Remitters may require receiving households to use this income in order to fund education, increasing the total share of education in total household consumption. Furthermore, low MPC out of remittances indicate that these are more likely to be saved than some other forms of income.

18

Full regression results are available from the authors on request.

329

S. Davies et al. / Journal of Economic Psychology 30 (2009) 321–334

Policy formulation by governments, NGOs and credit or savings institutions need to take account of mental accounting models since these influence consumption and savings habits. Remittances should be encouraged in Malawi since they encourage savings and consumption of education. Current efforts to help reduce the costs of remittances are a step in the right direction. Banks are more likely to be successful in attracting household funds if they encourage the saving of particular forms of income, notably remittances. On the lending side, micro-finance organisations must not only improve access to credit, but must ensure that mental accounting models encourage the ‘‘constructive” use of credit (which is shown to have an important role in funding education and farming for both male and female headed households); targeted publicity may help to support this aim. Lending institutions requiring valuable collateral may be unsuccessful as low marginal propensities to consume out of these indicate mental accounting systems do not currently appear to permit households to consume out of these assets.

Acknowledgements We would like to thank the Malawian National Statistical Office for kindly providing us with the data. An earlier version of the paper was presented at a seminar in Chancellor College, University of Malawi, Zomba, September 2006. We acknowledge gratefully the useful comments and suggestions made by the participants. Finally, we are grateful to two anonymous referees who provided useful comments and suggestions which greatly helped to improve the paper.

Appendix 1. Means and regressions See Tables A1–A6.

Table A1 Descriptive statistics. Variable

Household characteristics Age Married Household size Number of children Education level Agricultural employment Acres owned Interviewed during Hungry season Reside in north Reside in south Income Total Salary Remittance Farm Credit income (yes/no) Business income (yes/no) Farm income (yes/no) Salary income (yes/no) Remittance income (yes/no) Consumption Total Food Education Health Household Farm Clothing Fuel Assets Fixed Liquid

Male headed households

Female headed households

Obs.

Mean

Std. dev.

Min.

Max.

Obs.

Mean

Std. dev.

Min.

Max.

4171 4171 4171 4171 4038 4171 4171 4171 4171 4171

40.10 0.89 4.45 2.01 2.24 0.43 1.85 0.25 0.14 0.45

14.91

16.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

114.00 1.00 18.00 12.00 6.00 1.00 32.00 1.00 1.00 1.00

1473 1473 1473 1473 1435 1473 1473 1473 1473 1473

44.28 0.28 3.79 1.93 2.03 0.50 1.86 0.26 0.16 0.41

16.66 0.45 2.20 1.63 1.49

14.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

99.00 1.00 16.00 9.00 6.00 1.00 23.20 1.00 1.00 1.00

2923 843 924 1927 4171 4171 4171 4171 4170

1816.63 3027.21 1339.78 410.30 0.19 0.22 0.46 0.20 0.22

3619.80 3482.08 4049.11 740.08

3.43 3.18 1.99 2.60 0.00 0.00 0.00 0.00 0.00

70677.34 52864.08 69600.00 15395.47 1.00 1.00 1.00 1.00 1.00

1049 260 326 680 1473 1473 1473 1473 1473

2004.24 3242.44 1604.68 453.88 0.18 0.22 0.46 0.18 0.22

4108.73 3303.84 3155.84 843.66

2.44 17.69 6.12 2.44 0.00 0.00 0.00 0.00 0.00

72726.00 16511.45 22344.00 8850.00 1.00 1.00 1.00 1.00 1.00

4171 4153 410 2514 4171 2260 2907 3622

2080.58 917.96 277.76 66.06 1646.58 224.83 439.81 128.70

2871.11 1037.42 572.74 141.21 2126.15 312.15 771.02 674.52

18.72 1.61 0.53 0.25 10.95 0.43 0.69 1.85

72894.28 17043.05 6150.00 2440.00 36966.28 4770.00 21924.00 25760.00

1473 1468 126 913 1472 818 1039 1284

2012.74 884.06 304.23 56.33 1575.32 258.37 439.86 110.04

2430.67 993.99 854.13 101.06 1730.27 576.93 654.61 372.58

35.58 2.70 0.57 0.26 17.16 1.28 0.46 1.76

35443.14 10696.58 6150.00 1382.40 16559.68 9721.32 8954.80 10546.20

3970 4105

2095.758 905.4355

5163.823 3683.543

7.513262 1.608579

103455 113830.9

1398 1447

2184.09 958.29

5543.28 3846.26

22.95 2.94

125000.00 113522.70

2.34 1.76 1.55 1.79

1.78

330

S. Davies et al. / Journal of Economic Psychology 30 (2009) 321–334

Table A2 Robust OLS models. Dependent variable: total consumption.

Salary Remittances Farm Fixed assets Liquid assets Loan dummy Business dummy Farm dummy Salary dummy Remittance dummy Acres owned Age Age square Agriculture dummy Married dummy Number of children Household size Literate dummya Primary education dummya Secondary education dummya High school/university dummya Other education dummya Hungry season dummya North dummya South dummya Constant N r2 F Coefficient equality tests (t-values) Null hypothesis Salary = remittances Remittances = farm Salary = farm

Male headed

Female headed

0.331*** (3.882) 0.066*** (3.208) 0.484*** (2.578) 0.071*** (4.289) 0.098*** (2.761) 235.209*** (2.622) 981.740*** (7.711) 363.021*** (3.689) 208.722 (0.970) 52.754 (0.679) 55.599 (1.405) 7.676 (0.666) 0.092 (0.812) 55.190 (0.708) 60.289 (0.469) 37.213 (0.907) 51.276 (1.620) 1.136 (0.019) 237.751**

0.579*** (5.810) 0.202*** (4.079) 0.796** (2.250) 0.065*** (6.323) 0.036 (1.318) 578.606*** (4.213) 416.320*** (3.398) 278.893 (1.597) 567.105** (2.368) 17.217 (-0.119) 13.456 (0.350) 12.128 (0.877) 0.059 (0.453) 64.299 (0.694) 96.949 (0.877) 45.164 (0.886) 5.298 (0.142) 8.217 (0.093) 606.335***

(1.994) 880.313***

(3.707) 1328.159***

(3.973) 3028.391***

(3.799) 2203.114***

(4.493) 1866.542***

(3.542) 499.445**

(2.796) 233.435***

(2.443) 214.480**

(3.013) 648.053*** (3.351) 0.867 (0.009) 632.058** (2.213) 4170 0.257 23.067

(2.049) 66.381 (0.409) 31.853 (0.229) 1235.307*** (3.156) 1473 0.380 18.701

9.15*** 4.72** 0.54

11.94*** 2.72* 0.34

Notes: t-ratios in parenthesis, coefficients significant at *, 10%; **, 5%; and ***, 1%. All standard errors corrected for heteroskedasticity. a Omitted variables: illiterate and no formal education; central (capital) region.

331

S. Davies et al. / Journal of Economic Psychology 30 (2009) 321–334

Table A3 Consumption of different categories of good from different income sources. Tobit models – female headed households

Salary Remittances Farm Fixed assets Liquid assets Loan dummy Business dummy Farm dummy Salary dummy Remittance dummy Acres owned Second lowest consumption quartilea Second highest consumption quartilea Top consumption quartilea Age Age square Agriculture dummy Married dummy Number of children Household size Literate dummya Primary Education Dummya Secondary education dummya High school/university dummya Other education dummya Hungry season dummy North dummya South dummya Constant Standard error of regression N McFadden r2 Log Likelihood F Censored Coefficient equality tests (t-values) Null hypothesis Salary = remittances Remittances = farm Salary = farm

Food

Education

Health

Household

Farm

Clothing

Fuel

0.052*** (3.667) 0.012 (1.599) 0.120*** (3.771) 0.006 (0.674) 0.009 (0.832) 21.308 (0.668) 88.241*** (3.035) 47.235** (2.016) 98.031*** (2.592) 28.235 (1.128) *** 43.271 (4.630) *** 342.008 (29.212) 911.044*** (43.085) 2214.881*** (32.395) 2.044 (0.482) 0.024 (0.578) 10.039 (0.434) 62.738 (1.254) 18.179 (1.369) 14.913 (1.287) 25.306 (1.383) 19.942 (0.568) 28.020 (0.387) 72.913 (0.521) 536.344** (2.186) 43.747* (1.748) 64.091 (1.465) *** 123.106 (4.216) *** 233.259 (2.601) *** 708.470 (16.656) 4170 0.046 3.32e + 04 189.623 18

0.002 (0.148) 0.024*** (3.014) 0.020 (0.414) 0.004 (1.606) ** 0.010 (2.539) *** 200.946 (4.140) 66.591 (1.468) 93.404** (2.042) 157.432*** (2.884) ** 97.260 (2.040) 15.907 (1.407) 96.825 (1.507) 144.464** (2.230) 427.674*** (4.453) 6.783 (0.898) 0.045 (0.599) 32.553 (0.761) * 127.932 (1.766) 50.162* (1.790) 22.754 (1.164) 40.752 (0.996) ** 222.337 (2.495) *** 673.541 (4.316) *** 689.308 (4.898) 446.890 (1.638) 172.988*** (3.431) 263.389*** (3.505) *** 145.403 (2.887) *** 1723.159 (5.997) *** 721.423 (7.445) 4170 0.049 3956.516 2.996 3760

0.006 (0.850) 0.001 (0.686) 0.003 (0.823) 0.000 (0.140) 0.000 (0.128) ** 14.085 (2.116) 2.099 (0.338) 4.235 (0.853) 2.375 (0.142) 7.619 (1.564) *** 7.844 (3.514) *** 36.848 (6.410) 68.614*** (9.755) 144.958*** (8.900) 1.228 (1.453) 0.013 (1.549) 0.255 (0.051) 0.681 (0.081) 1.827 (0.554) 1.129 (0.474) 4.224 (0.929) 13.750 (1.520) ** 45.342 (2.364) 0.633 (0.025) 28.837 (1.217) 7.277 (1.340) 18.039** (2.514) *** 47.760 (6.216) *** 98.661 (3.902) *** 145.553 (12.464) 4170 0.018 1.71e + 04 8.166 1656

0.098*** (3.032) 0.000 (0.008) 0.034 (0.283) 0.018 (1.453) * 0.045 (1.781) 6.022 (0.115) 228.297*** (3.363) 121.836** (2.151) 97.838 (1.161) 29.017 (0.640) 9.465 (0.347) *** 462.120 (26.305) 1336.759*** (42.781) 4354.159*** (31.106) 2.531 (0.372) 0.024 (0.363) 1.863 (0.042) 17.956 (0.255) 7.122 (0.331) 0.790 (0.047) 45.376 (1.272) 5.420 (0.080) 140.577 (1.055) 441.679 (1.574) 532.361 (1.565) 56.103 (1.136) 268.059** (2.540) * 119.715 (1.840) 125.968 (0.789) *** 1412.320 (10.693) 4170 0.045 3.62e + 04 335.551 0

0.004 (0.788) 0.007 (1.540) 0.087*** (5.082) ** 0.003 (2.319) 0.001 (0.287) *** 42.703 (3.139) 14.995 (1.112) 99.958*** (7.630) 8.368 (0.472) 5.484 (0.396) *** 69.421 (9.604) *** 127.943 (8.874) 230.469*** (13.826) 363.905*** (11.944) 1.943 (1.013) 0.019 (1.002) 7.065 (0.635) 0.517 (0.028) 8.947 (1.286) 7.876 (1.538) 15.206 (1.554) 31.823 (1.534) * 52.389 (1.739) ** 93.793 (2.254) 162.347** (2.015) 4.676 (0.348) 127.632*** (5.750) *** 96.795 (6.334) *** 400.977 (7.448) *** 305.702 (17.191) 4170 0.046 1.71e + 04 19.912 1911

0.017 (1.447) 0.010 (1.336) 0.092** (2.082) 0.003 (0.892) 0.003 (0.636) 41.402 (1.381) 81.267*** (3.251) 8.484 (0.323) 82.785** (2.294) 13.039 (0.543) *** 36.945 (2.966) *** 195.331 (6.441) 451.939*** (11.357) 1149.057*** (15.250) 2.000 (0.482) 0.021 (0.511) 15.751 (0.605) 25.065 (0.611) 12.211 (0.700) 5.947 (0.431) 53.260*** (2.836) 7.828 (0.219) 90.715 (1.525) 138.627 (0.907) 180.956 (1.630) 69.466** (2.372) 232.334*** (3.960) 43.803 (1.504) *** 429.309 (2.998) *** 707.597 (6.397) 4170 0.028 2.40e + 04 19.902 1263

0.001 (0.244) 0.001 (0.259) 0.051 (0.768) 0.007 (1.313) 0.011 (1.350) 23.092 (1.460) 125.259*** (2.951) 28.710 (1.111) 1.337 (0.076) 3.429 (0.220) 7.286 (0.444) *** 57.402 (3.870) 75.632*** (4.761) 400.372*** (5.261) 4.726 (1.526) 0.048 (1.618) 22.074 (1.112) 10.883 (0.459) 7.630 (0.925) 7.242 (1.096) 36.387* (1.891) 26.341 (0.710) 20.324 (0.467) 33.812 (0.610) 224.458*** (2.922) 25.974 (1.594) 84.467 (1.546) 26.410 (0.902) ** 261.269 (2.517) *** 657.225 (4.143) 4170 0.005 2.90e + 04 9.277 549

16.81*** *** 11.06 24.26***

1.72 0.00 0.13

0.92 0.35 1.23

8.56*** 0.08 1.12

0.25 *** 25.55 25.04***

0.22 * 3.18 2.56

0.21 0.53 0.59

Notes: t-ratios in parenthesis, coefficients significant at *, 10%; **, 5%; and ***, 1%. All standard errors corrected for heteroskedasticity. a Omitted variables: illiterate and no formal education; lowest consumption quartile and central (capital) region.

332

S. Davies et al. / Journal of Economic Psychology 30 (2009) 321–334

Table A4 Consumption of different categories of good from different income sources. Tobit models – female headed households

Salary Remittances Farm Fixed assets Liquid assets Loan dummy Business dummy Farm dummy Salary dummy Remittance dummy Acres owned Second lowest consumption quartilea Second highest consumption quartilea Top consumption quartilea Age Age square Agriculture dummy Married dummy Number of children Household size Literate dummya Primary education dummya Secondary education dummya High school/university dummya Other education dummya Hungry season dummy North dummya South Dummya Constant Standard error of regression N McFadden r2 Log Likelihood F Censored

Food

Education

Health

Household

Farm

Clothing

Fuel

0.061 (1.430) 0.024 (0.922) 0.092** (2.245) 0.027*** (3.564) 0.002 (0.594) 15.880 (0.333) 67.391 (1.608) 50.374 (1.485) 126.729 (1.229) 13.538 (0.332) 39.874*** (3.741) 334.716*** (18.458) 878.585*** (25.476) 2008.658*** (20.714) 6.495 (1.382) 0.054 (1.158) 9.898 (0.279) 54.128 (1.287) 2.939 (0.156) 8.792 (0.675) 27.769 (0.947) 15.890 (0.293) 40.629 (0.272) 199.519 (0.999) 34.711 (0.316) 122.722*** (3.274) 156.336*** (2.888) 146.732*** (3.318) 269.866** (2.099)

0.020 (0.690) 0.103** (1.992) 0.276* (1.811) 0.005 (0.631) 0.017*

0.001 (0.191) 0.001 (0.135) 0.001 (0.263) 0.001 (0.668) 0.001 (0.815) 14.082 (1.487) 8.319 (1.096) 15.947** (2.073) 22.796*

0.051 (1.343) 0.024* (1.907) 0.236** (2.129) 0.003 (0.536) 0.001 (0.381) 117.698*** (2.634) 81.859*** (2.601) 68.985 (1.453) 145.836 (1.564) 96.124* (1.764) 97.927*** (5.138) 192.270*** (4.541) 300.801*** (6.510) 479.179*** (5.652) 4.523 (0.995) 0.026 (0.547) 11.113 (0.395) 40.243 (1.160) 14.804 (0.838) 0.233 (0.019) 113.054*** (2.717) 175.437**

0.011 (0.638) 0.002 (0.073) 0.073 (1.491) 0.005 (1.141) 0.005 (1.075) 47.568 (0.980) 22.716 (0.617) 20.332 (0.512) 76.306 (1.359) 8.887 (0.152) 22.641** (2.138) 255.116*** (6.659) 527.069*** (10.730) 1172.256*** (11.269) 2.404 (0.402) 0.026 (0.449) 40.878 (1.219) 29.107 (0.677) 14.034 (0.790) 4.438 (0.348) 82.142*** (2.582) 99.461*

(2.401) 176.535** (1.980) 148.017 (0.991) 298.664 (1.412) 60.410* (1.786) 162.926*** (3.068) 135.417*** (2.892) 595.173*** (4.224)

(1.674) 299.996** (2.573) 9.102 (0.058) 80.392 (0.396) 100.040*** (2.749) 238.765*** (4.628) 88.301** (2.368) 126.651 (0.883)

0.016 (1.567) 0.025** (2.259) 0.020 (0.962) 0.002 (1.181) 0.008 (0.807) 4.509 (0.198) 27.960 (1.452) 23.239 (0.824) 18.102 (0.781) 56.154*** (2.583) 8.061 (1.626) 52.881*** (2.792) 88.179*** (4.012) 281.780*** (3.293) 1.092 (0.576) 0.008 (0.464) 29.878 (1.540) 0.533 (0.034) 3.050 (0.455) 3.913 (0.535) 9.303 (0.404) 43.477 (0.889) 65.640 (1.166) 30.488 (0.433) 170.033** (2.164) 17.872 (1.106) 25.841 (1.422) 4.443 (0.181) 101.993 (0.947)

665.752*** (14.521) 1473 0.048 1.16e + 04 66.255 5

546.567*** (5.481) 1473 0.034 6642.265 4.999 655

581.491*** (9.701) 1473 0.036 8366.174 11.908 434

358.048*** (3.138) 1473 0.009 9507.596 5.605 189

(1.774) 276.163** (2.560) 17.193 (0.167) 211.094** (2.019) 123.114 (0.795) 92.599 (0.788) 14.887 (0.523) 128.530 (0.950) 208.685 (1.447) 557.796*** (2.957) 8.026 (0.588) 0.057 (0.426) 86.077 (0.969) 126.866 (1.276) 13.612 (0.268) 9.884 (0.259) 150.077* (1.789) 422.298**

(1.878) 5.384 (0.662) 0.071 (0.038) 30.562*** (4.239) 63.005*** (6.732) 94.883*** (6.523) 0.347 (0.394) 0.004 (0.398) 5.397 (0.851) 6.698 (0.920) 0.242 (0.066) 1.407 (0.511) 5.068 (0.911) 20.676*

(2.328) 1359.218*** (3.848) 1139.247*** (3.540) 1076.920** (2.285) 91.272 (1.007) 111.046 (0.828) 263.342** (2.357) 2021.363*** (4.677)

(1.919) 23.246 (1.221) 41.663 (1.369) 6.736 (0.148) 8.731 (1.391) 25.122** (2.505) 44.852*** (4.959) 66.822** (2.496)

0.165*** (2.791) 0.039 (0.944) 0.018 (0.271) 0.019*** (3.087) 0.007 (0.559) 12.592 (0.174) 60.177 (1.009) 34.953 (0.649) 204.294 (1.498) 39.586 (0.511) 17.691 (0.877) 470.803*** (22.542) 1397.125*** (34.999) 3813.468*** (23.960) 10.259 (1.268) 0.087 (1.125) 15.522 (0.312) 37.582 (0.644) 11.439 (0.448) 7.974 (0.432) 18.139 (0.389) 6.758 (0.082) 87.455 (0.430) 632.525** (1.998) 3.493 (0.022) 46.395 (0.847) 13.230 (0.185) 132.979* (1.710) 476.170** (2.322)

870.280*** (5.662) 1473 0.060 1247.023 1.305 1347

106.778*** (9.655) 1473 0.019 5905.490 4.053 560

953.110*** (11.563) 1473 0.067 1.22e + 04 149.625 1

333

S. Davies et al. / Journal of Economic Psychology 30 (2009) 321–334 Table A4 (continued) Tobit models – female headed households

Coefficient equality tests (t-values) Null hypothesis Salary = remittances Remittances = farm Salary = farm

Food

Education

Health

Household

Farm

Clothing

Fuel

0.58 4.88*** 6.60

4.10** 4.24** 2.75*

0.05 0.06 0.01

3.31* 0.54 4.51**

3.16* 5.30** 2.54

0.06 1.91 1.50

0.41 0.06 0.05

Notes: t-ratios in parenthesis, coefficients significant at *, 10%; **, 5%; and ***, 1%. All standard errors corrected for heteroskedasticity. a Omitted variables: illiterate and no formal education; lowest consumption quartile and central (capital) region. Table A5 Food and education consumption for top and bottom consumption quartiles.

Salary Remittances Farm Standard error of the Regression N McFadden r2 Log Likelihood F

Food – male heads

Eduction – male heads

Food – female heads

Top quartile

Bottom quartile

Top quartile

Bottom quartile

Top quartile

Bottom quartile

Top quartile

Bottom quarti

0.068*** (3.390) 0.013 (1.194) 0.111** (1.996) 1550.149*** (13.559) 649 0.006 5681.047 3.620

0.008 (0.877) 0.014*

0.009 (0.474) 0.026**

0.016* (1.652) 0.010*

(1.890) 0.020 (1.108) 111.700*** (38.245) 810 0.015 4899.259 7.817

(2.052) 0.066 (1.021) 1137.529*** (6.830) 649 0.029 1383.320 2.212

(1.884) 0.028* (1.677) 40.176*** (3.905) 810 0.098 237.494 2.646

0.071 (1.066) 0.014 (0.333) 0.086 (0.996) 1384.574*** (11.286) 235 0.011 2033.240 2.593

0.009 (0.651) 0.003 (0.119) 0.001 (0.046) 105.394*** (26.618) 288 0.025 1730.214 7.602

0.018 (0.361) 0.091 (1.411) 0.558** (2.476) 1271.256*** (7.167) 235 0.090 383.725 1.457

0.034 (1.009) 0.011 (0.629) 0.010 (0.439) 58.815*** (3.938) 288 0.164 80.407 1.719

Notes: t-ratios in parenthesis, coefficients significant at available from the authors on request.

*

10%,

**

5% and

Education – females heads

***

1%. All standard errors corrected for heteroskedasticity. Full regression results

Table A6 Remittances from difference geographical sources.

Panel A – male heads Salary Remittances (urban Malawi) Remittances (rural Malawi) Remittances (abroad) Farm Standard Error of the Regression N McFadden r2 Log likelihood F Panel B – female heads Salary Remittances (urban Malawi) Remittances (rural Malawi) Remittances (abroad) Farm Standard error of the Regression N McFadden r2 Log likelihood F

Food

Education

Health

Household

Farm

Clothing

Fuel

0.052*** (3.671) 0.011 (1.411) 0.000 (0.007) 0.028*

0.002 (0.150) 0.019*** (2.605) 0.034** (1.988) 0.016 (0.865) 0.022 (0.444) 720.540*** (7.510) 4170 0.049 3955.802 2.955

0.006 (0.853) 0.001 (0.402) 0.002 (0.486) 0.003 (1.543) 0.003 (0.842) 145.509*** (12.460) 4170 0.018 1.71e + 04 7.628

0.098*** (3.031) 0.003 (0.190) 0.006 (0.577) 0.013 (0.511) 0.033 (0.274) 1412.245*** (10.694) 4170 0.045 3.62e + 04 313.834

0.004 (0.796) 0.011** (2.136) 0.023 (1.261) 0.003 (0.487) 0.086*** (5.088) 305.339*** (17.166) 4170 0.047 1.71e + 04 18.765

0.017 (1.450) 0.024*** (4.026) 0.012 (0.774) 0.009 (0.800) 0.089** (2.051) 707.064*** (6.387) 4170 0.028 2.40e + 04 20.617

0.001 (0.235) 0.000 (0.030) 0.006* (1.834) 0.004 (0.771) 0.052 (0.775) 657.151*** (4.144) 4170 0.005 2.90e + 04 9.680

Education 0.021 (0.809) 0.198* (1.809) 0.140* (1.747) 0.059 (1.459) 0.238 (1.644) 810.559*** (6.641) 1473 0.068 1235.805 1.331

Health 0.001 (0.162) 0.008** (2.079) 0.007 (0.588) 0.003 (1.379) 0.002 (0.544) 106.262*** (9.745) 1473 0.020 5901.576 3.792

Household 0.165*** (2.791) 0.036 (0.452) 0.054 (1.082) 0.025 (0.392) 0.016 (0.244) 952.924*** (11.559) 1473 0.067 1.22e + 04 140.320

Farm 0.051 (1.365) 0.039 (0.899) 0.072* (1.853) 0.060**

Clothing 0.011 (0.651) 0.006 (0.113) 0.032 (1.400) 0.032 (0.595) 0.070 (1.441) 579.907*** (9.758) 1473 0.037 8362.887 11.154

Fuel 0.016 (1.589) 0.042** (2.481) 0.018* (1.688) 0.019 (1.068) 0.017 (0.892) 357.821*** (3.133) 1473 0.009 9506.472 5.543

(1.929) 0.118*** (3.697) 708.167*** (16.651) 4170 0.046 3.32e + 04 176.487 Food 0.061 (1.427) 0.037 (0.613) 0.064 (1.418) 0.026* (1.824) 0.090** (2.166) 663.281*** (14.645) 1473 0.049 1.16e + 04 63.337

Notes: t-ratios in parenthesis, coefficients significant at available from the authors on request.

*

10%,

**

5% and

***

(2.060) 0.229** (2.151) 543.360*** (5.556) 1473 0.035 6636.240 4.718

1%. All standard errors corrected for heteroskedasticity. Full regression results

334

S. Davies et al. / Journal of Economic Psychology 30 (2009) 321–334

References Adams, R. H. (1991). The economic uses and impact of international remittances in rural Egypt. Economic Development and Cultural Change, 39, 695–722. Adams, R. H. (1998). Remittances, investment and rural asset accumulation in Pakistan. Economic Development and Cultural Change, 47, 155–173. Adams, R. H. (2002). Precautionary saving from different sources of income: Evidence from rural Pakistan. World Bank Discussion Paper. Adams, R. H. (2005). Remittances, household expenditure and investment in Guatemala. World Bank Policy Research Working Paper No. 3532. Adams, R. H., Cuecuecha, A. & Page, J. (2008). Remittances, consumption and investment in Ghana. World Bank, Policy Research Working Paper No. 4515. Ainslie, G. (1975). Special reward: A behavioural theory of impulsiveness and impulse control. Psychological Bulletin, 82(4), 463–496. Alderman, H., & Paxson, C. (1992). Do the poor insure? A synthesis of the literature on risk and consumption in developing countries. World Bank Policy Research Working Papers, WPS 1008. Anderson, C., & Nevitte, N. (2006). Teach your children well: Values of thrift and saving. Journal of Economic Psychology, 27, 247–261. Azam, J. -P., & Gubert, F., (2004). Those in Kayes: The Impact of Remittances on their Recipients in Africa. paper presented at the EUDN meeting, Paris: 26–27 November 2004 (available at: http://idei.fr). Brocas, I., Carrillo, J. D., & Dewatripont, M. (2004). Commitment devices under self-control problems: An overview. In I. Brocas & J. D. Carrillo (Eds.). The psychology of economic decisions: Reasons and choices (Vol. 2). Oxford, United Kingdom: Oxford University Press. Chami, R., Fullenkamp, C., Jahjah, S. (2005). Are immigrant remittances flows a source of capital for development? IMF Staff Papers, Vol. 52. Chipeta, C., Kachaka, W. (2005). Role of migrants’ remittances in an unstable low-income economy: A case study of Malawi, University of Malawi. Chancellor College Working Paper No. 2005/05. Colloredo-Mansfeld, R. (2005). Consumption. In J. Carrier (Ed.), A handbook of economic anthropology. United Kingdom: Cheltenham. Cox Edwards, A., & Ureta, M. (2003). International migration, remittances, and schooling: Evidence from El Salvador. Journal of Development Economics, 72, 429–461. Dercon, S., Hoddinott, J., & Woldehanna, T. (2005). Shocks and consumption in 15 Ethiopian villages 1999–2004. Journal of African Economies, 14(4), 559–585. Friedman, M. A. (1957). A theory of the consumption function. Princeton University Press. Gammeltoft (2002). Remittances and other financial flows to developing countries. International Migration, 40(5), 181–211. Harrower, S., & Hoddinott, J. (2005). Consumption smoothing in the Zone Lacustre, Mali. Journal of African Economies, 14(4), 489–519. Hart, K. (2005). Money: One anthropologist’s view. In J. Carrier (Ed.), A handbook of economic anthropology. United Kingdom: Cheltenham. Hayashi, F. (1987). Test for liquidity constraints: A critical survey and some new observations. In T. Bewley (Ed.), Advances in econometrics. Cambridge: Cambridge University Press. Ishikawa, T., & Ueda, K. (1984). The bonus payment system and Japanese personal savings. In M. Aoki (Ed.), The economic analysis of the Japanese firm. North Holland: Amsterdam. Karlsson, N. (2003). Consumer self-control strategies: An empirical study of their structure and determinants. Journal of Consumer Policy, 26, 23–41. Katz, E. G. (1995). Gender and trade within the household: Observations from rural Guatemala. World Development, 23, 327–342. Kozel, V., & Alderman, H. (1990). Factors determining work participation and labour supply decisions in Pakistan’s urban areas. The Pakistan Development Review, 29(1), 1–18. Levin, L. (1998). Are assets fungible? Testing the behavioural theory of lifecycle savings. Journal of Economic Behaviour and Organisation, 36, 59–83. Lewis, A., & Winnett, A. (1995). Household accounts, mental accounts, and savings behaviour: Some old economics rediscovered? Journal of Economic Psychology, 16, 431–448. Lucas, R. E. B. (1987). Emigration to South Africa’s mines. American Economic Review, 77(3), 313–330. McCracken, J. (2002). The ambiguities of nationalism: Flax Musopole and the northern factor in Malawian politics, c. 1956–1966. Journal of Southern African Studies, 28(1), 67–87. McDonald, J. F., & Moffitt, R. A. (1980). The uses of tobit analysis. The Review of Economics and Statistics, 62(2), 318–321. McElroy, M. B. (1990). The empirical content of nash-bargained household behaviour. Journal of Human Resources, 25, 559–583. Miller, D., Paulson, A. L. (1999). Informal insurance and moral hazard: Gambling and remittances in Thailand. Econometric Society World Congress 2000 Contributed Papers 1463, Econometric Society, revised. Modigliani, F. (1966). The life cycle hypothesis of saving, the demand for wealth, and the supply of capital. Social Research, 33, 160–217. Ratha, D. (2003). Workers’ remittances: An important and stable source of external development finance. In S. Maimbo & D. Ratha (Eds.), Remittances development impact and future prospects. Washington DC: World Bank Publications. Shefrin, H., & Thaler, R. (1988). The behavioural lifecycle hypothesis. Economic Inquiry, 36, 606–643. Udry, C. (1990). Credit markets in northern Nigeria: Credit as insurance in a rural economy. World Bank Economic Review, 4, 251–269. White, H. (1980). Heteroscedasticity consistent covariance matrix and a direct test for heterscedasticity. Econometrica, 48, 817–838. Wooldridge, J. M. (2002). Econometric analysis of cross section and panel data. Cambridge, Massachusetts: MIT Press. Wooldridge, J. M. (2006). Introductory econometrics: A modern approach international edition (third ed.). Ohio: Thomson South-Western. Zeldes, S. P. (1989). Consumption and liquidity constraints: An empirical investigation. Journal of Political Economy, 97, 305–346.