Estimating Consumption Function for Central Asian countries: New Insights

doi:10.21203/rs.3.rs-4116742/v1

Consumption is a significant part of the Gross Domestic Product (GDP) that is influenced by income as argued by Keynes and Friedman. This paper empirically estimates the consumption function for four Central Asian economies (Kazakhstan, Kyrgyzstan, Tajikistan, and Uzbekistan) from 1993 to 2020. The empirical results reveal that income has a significant positive effect on consumption in the short and long run. The short-run results support the Keynesian Absolute Income Hypothesis, whereas, the long-run results support Milton Friedman’s Permanent Income Hypothesis. However, the coefficient value is greater than 1 in the case of Uzbekistan which is in contradiction with the consumption function, as Keynes argued that the marginal propensity to consume (MPC) value lies between 0 and 1 (0 < MPC > 1). The key finding of the study is, that in the case of Central Asian countries, the marginal propensity of consumption does not lie in the range proposed by Keynes and does not support the Absolute Income Hypothesis. In this regard, policies are required to increase savings and investment and keep consumption expenditures at optimal levels. These policy measures will contribute more effectively to boosting the economic growth of these economies.

Consumption

Income

ARDL

Central Asian countries

The legendary work of John Maynard Keynes ‘Consumption Function’ has got widespread attention across the globe. In 1936, he published a book ‘General Theory of Employment, Interest and Money’ where he highlighted the connection between consumption and income which is known as the consumption function. Keynes in his theory of Absolute Income Hypothesis (AIH) argued that the aggregate consumption is primarily determined by the current income (Laiqat et al. 2018). Kuznets (1946) was the first who empirically investigated the consumption function to estimate the Average Propensity to Consume (APC) for the short run and long run using cross section and time series data in case of the U.S.A. In contrast, Dusenbery (1949) argued that consumers’ utility is determined by relative income, not by absolute income (as Keynes believed), but current income compared to former income and present income. Friedman (1957) stated that household's spending is proportional to its permanent income, or the average income that a household may expect to earn over a period of time (in future). This idea arose from the long-standing observation that earnings other than current disposal income influenced current purchasing patterns. The Permanent Income Hypothesis (PIH) helps the long-run consistency of the APC as well as why during cyclical fluctuations, this ratio fluctuates inversely with income. The PIH explains the more important aspects of consumer behavior intuitively. Based on the fact, that changes in permanent income over long periods of time reflect changes in an economy's aggregate income growth, i.e., the economy’s resources increase permanently (Khan, 2012). Ando and Modigliani (1963) developed jointly the life cycle hypothesis (LCH). In many aspects, the LCH and the PIH are similar, including the fact that consumption is a constant percentage of income. However, according to this perspective, consumption is dictated by a person's various stages of life, with the average consumer's goal being to balance consumption across a lifetime in which income varies dramatically with age (Arapova, 2018). Because people are not earning at the adulthood and old age after retirement their consumption patterns are maintained by borrowing or depleting previous savings. As a result, consumption represents an enormous portion of income during these phases of life. When income is relatively high during the middle phase of a person life, he is able to save more, and later on these savings are utilized to pay post-retirement consumption, resulting in a reduced fraction of income being consumed in the current period of time (Xiao & Liao, 2018).The most basic version of the LCH assumes that the consumer's lifetime arranging possibility, joined with the expected proportionality between spending and permanent income, guarantees that no net lifetime savings are arranged and that exchanges to beneficiaries are simply equivalent to their own underlying legacy. Changes in the current income affect current spending just to the degree that they are permanent, necessitating a recalculation of lifetime consumable assets. The impact of transient income increases on spending is expected to be minor (Ammad & Ahmed, 2020). Each of the three theories is based on the microeconomics theory of consumer decision. This is particularly valid for the LCH and PIH hypotheses, which explicitly imply that rational buyers seek to maximize utility by dispensing their permanent incomes (i.e., their long-lasting stream of income) for lifetime spending. Conversely, the Relative Income Hypothesis (RIH), contradicts the fundamental assumption that an individual's choices should be independent of others' consumption patterns, and hence goes opposed to mainstream microeconomic theories of consumer behavior. This is one of the reasons that the RIH has failed to evoke the same level of empirical interest as the PIH and LCH approaches (Al-Gahtani et al. 2020).

Figure 1 shows the consumption is measured on vertical axis in million US dollars. In Kazakhstan the consumption is increasing steadily during 1993–2020. The ups and down is noticed in Kyrgyz. In Tajikistan and Uzbekistan, the consumption remains constant as compared to Kazakhstan and Kyrgyz. The Kazakhstan and Kyrgyz are the main consumption economy in the central Asian countries. These two countries spend more income on consumption.

Overall, the gross domestic product (GDP) consists of consumption, investment, and net exports. The largest element of GDP is consumption. More consumption means higher effective demand of an economy. According to Keynesian effective demand concept, higher effective demand results in more production and employment opportunities. The reason for conducting this study is that it is the first and latest study for the central Asian counties. The aim of this article is to suggest some policy measures for policy makers.

The central Asian countries recently got freedom from Russia with the collapse of the Soviet Union in 1991. The present study is important because it will be helpful for the policy makers in effective policies formulation for the Central Asian countries, whether these policies will be consumption oriented or saving oriented. This article is the first and latest study for the central Asian countries. Consequently, the prime objective of the current paper is to investigate the consumption function for selected Central Asian countries (Kazakhstan, Kyrgyz, Tajikistan, and Uzbekistan) using data from 1993–2020. The Central Asian countries are selected for this research, because the extant literature exhibits that there are no solid empirical studies on testing consumption function for these countries and its examination is indispensable. Income of a consumer is divided by the consumer into two parts consumption and saving. Some part of income goes to saving, and saving is that part of income which is not consumed. However, in every economy about 60 to 70 percent of the GDP spend on consumption. This means that income earned are mostly utilized for the consumption purposes in these countries, and it is one of the major components of GDP.

Consumption can play a crucial role not only in the determination of the aggregate demand but it’s also play a key role in the production and supply of goods in an economy. Therefore, it is important, to analyze the consumption function for Central Asian countries, to know the people tendency towards consumption and that, what proportion of income they save and consume. These findings will be helpful for the management authorities in formulating policy related to the consumption and saving for achieving the macroeconomic stability and enhancing level of economic growth and development for Central Asian countries.

This article consists of five sections. In section one, the introduction of the study with brief objective and importance are already presented. The theoretical and literature review is given in section two. Section three is concerned with the data and models information. Section four consists of empirical results of the study. Conclusion is given in section five.

2.1 Theoretical Literature

Theoretical link between income and consumption is explained through Keynesian consumption function. The consumption is influenced by many factors such as income, prices, population growth, taste etc, but most importantly affected by current income, as argued by Keynes (1936) in AIH; Dusenbury (1949) in RIH; Freidman (1957) in PIH followed by Modigliani (1963) in LCH. These theories are different from each other in some basic facts, but the theme of each theory is almost the same “Consumption and Income relationship”. Absolute income is the most important predictor of consumption in Keynes's consumption theory, and the amount devoted to consumption is determined by the marginal willingness to consume dependent on personal disposable income. Keynes used a psychological approach to consumption, rejecting the utility maximization consumer model. Duesenberry (1949) improved on Keynes' method by highlighting the importance of psychological and social variables in consumer decisions (the relative income hypothesis). Duesenberry's analysis comes to similar results about the effect of income on consumption and supported the Keynesian theory.

The life-cycle hypothesis of Modigliani & Brumberg (1954) and Friedman's (1957) permanent income hypothesis emerged as two more alternatives to Keynes' and Duesenberry's approaches. Modern orthodox consumption theories are generalizations of these two theories in a rational expectations framework. In the life-cycle and permanent income hypotheses, and much less so in contemporary orthodox consumption theories, current or relative income plays a minimal role.

2.2 Empirical Literature

Many researchers have been done on the relationship between income and consumption, for instance, Manitsaris (2006) examined the AIH and PIH for European Union economies. The study's findings revealed that the European economies studied supported the PIH. Khan (2010) investigated the validity of the AIH and PIH for Pakistan and discovered that the PIH is invalid. In 2012, Khan et al investigated the consumption function for Pakistan. The outcomes of the study verified the AIH for Pakistan. Altunc and Aydın (2014) tested consumption function for eight OIC countries over 1980–2010. Their results supported the permanent income hypothesis and ‘adaptive expectations model’. Nikbin and Panahi (2016) estimated the private consumption function employing time series data from 1978–2012 and ARDL approach. The result found that the GDP have positive impact on private consumption but showing negative relationship with inflation. The study of Xiao and Liao (2018) empirically tested the Keynesian and life-cycle hypotheses for China using yearly data from 1978–2013. The findings depicted that all Chinese citizens are irrational consumers; however, the causes of irrational consumption differ amongst the citizens, with urban residents experiencing money illusions while rural inhabitants experiencing over-reaction to price fluctuations. Similarly, Yasmeen et al. (2019) investigated the relationship between income and consumption in Chitral, Pakistan's rural districts. The analysis was based on Keynesian consumption hypothesis (AIH). The study used OLS technique and primary data for the analysis. The results verified that Keynesian AIH is worth working on, and that consumption in District Chitral is determined by income. Bilik and Kok (2020) analyzed the consumption function for European Union using panel data from 2000 to 2017 and GLM & GMM approaches. The result found that the PIH is applicable for the European Union countries which are verified through both the methods. Laiqat et al. (2018) investigated empirically the AIH and PIH validity for Chinese economy by utilizing the yearly data from 1970–2016. The results were against the PIH applicability and supported the AIH validity in case of China. Gahtani et al. (2020) estimated the life-cycle consumption model using time series data from 1970–2017 and ECM for Saudi Arabia. Their results revealed that income and wealth were having significant impacts on consumption.

After going through the literature (i.e., few studies presented) it can be seen that consumption function has its own values and importance in macroeconomics which is supported by both the theoretical and empirical results. On the basis of the consumption pattern one can forecast about the consumption and saving of an economy. This article is an attempt to analyze the consumption pattern of the central Asian countries. The current study is an important addition and contribution to the existing literature, because till date none of the previous studies were carried out on this issue for the Central Asian countries.

Table 1

Selected Studies on Consumption Function
Author(s)	Sample Period, country(s), Techniques	Dependent Variable	Independent Variables	Estimated coefficient (t-ratio)
Habanabakize (2021)	2008Q1-2020Q2, South Africa, ARDL	Total household expenditure	Disposable income, price, exchange rate	0.2108 ^*
Aropova (2018)	1991–2015, Asia, Panel Regression	Consumption	GNI, Interest, Population,	0.8209^* 0.0631^, 0.0805^**
Khan et al (2014)	1975–2012, Pakistan, OLS, AEG, ECM	Consumption	GDP, lag of consumption	0.604 (4.18) 0.696 (9.21)
Khan (2014)	300 Households, Pakistan, OLS	Consumption	Income, education, age, family size	0.566 (9.41) 0.446 (4.773) -0.173 (-2.246)
Ofwona (2013)	1992–2011, OLS	Household Consumption	Household Income	0.116 (9.38)
Akekere & Yousuo (2012)	1981–2011, Nigeria, OLS	Household Consumption	Household Income	0.671 (41.22)
Bin & Renjing (2012)	1997–2009, China, ELESE	Consumption	Per capita disposable income	0.017 (6.21)
Shaikh (2012)	1974–2010, Pakistan OLS	Private Consumption	Net Income	0.791 (114.46)
Hui-min (2010)	1978–2009, China, OLS, ECM	Consumption	Income	0.464 (3.160)
Barot (2006)	1970–1998, UK, Sweden, ECM	Private consumption	Disposable income, worth	0.39 (7.3) 0.26 (5.6)
Abeysinghe & Choy (2004)	1978–2003, Singapore, OLS, ECM	Consumption	Income, wealth	0.64 (13.7)
Breido & Tregub (2000)	1961–1996, USA, OLS	Consumption	Disposable income	1.03^***
Carroll (2001)	1970–2000, Panel regression	Consumption	Income	0.7612^*
Bunting (1989)	1929–1982, USA, OLS	Consumption	Income	0.931 (105.34)
Source: Author’s compilation, 2022

3.1 Data and its Sources

In order to analyze consumption function, the study utilized time series annual data from 1993 to 2020. The data has been taken from World Development Indicators (WDI), World Bank, 2022. All the data are in current US dollar.

3.2 Empirical Model

This study based on Keynesian (1936) consumption function. The same function was used by Laiqat et al, (2018) for China; Xiao and Liao (2018) and Gahtani et al., (2020) for Saudi Arabia; Yasmeen et al (2019) for Pakistan. The standard form of Keynesian consumption function is as follows:

$$C={C}_{0}+cY$$

1

C_o> 0, 0 < c > 1 c = MPC

The above function can be written in econometrics form as follows:

$${C}_{t}=\alpha +{\beta Y}_{t}+{\mu }_{t}$$

2

Where C = Household consumption in millions US dollar, t = time period, α = Autonomous consumption, β₁ = Slope, Y = Income in million US dollar, and µ = error term

4.1 Augmented Dicky Fuller (ADF)

The unit root test namely ADF developed by Dickey and Fuller is applied. In an autoregressive time series model, the ADF tests empirically tested the null hypothesis i.e. unit root is available. An alternative hypothesis, the time series model lacks a unit root. Because the study used time series data, it's crucial to check the series or data for the unit root. Most of the time series data is not stationary i.e., data have unit root. The article utilized ADF test to check the unit root.

4.2 Auto-Regressive Distributed lag (ARDL) Model

The ARDL model of Pesaran et al. (2001) is used to explore the effect of income on consumption. ARDL technique is powerful and superior to other approaches because (1) it incorporates the lag of the dependent variable in the model, (2) ARDL can be apply if the factors are integrated of I(0), I(1) or mixture of both, (3) ARDL gives short and long run results simultaneously. Eq. (2) can be written in ARDL form as below.

$\varDelta ln{\left(CON\right)}_{t}{=\beta }_{0}$ + ${\sum }_{t=1}^{p}{\beta }_{1}\varDelta ln\left({CON}_{t-i}\right)$ + ${\sum }_{t=1}^{p}{\beta }_{2}\varDelta ln\left({Y}_{t-i}\right)+{\lambda }_{1}ln\left({Y}_{t-i}\right)+$μ_t…… (3)

Where ${\beta }_{0}$the intercept, µ_t is the random error term, and t-i denotes the number of lags (i.e. i = 1, 2, 3…..). The error correction elements are addressed by β_1, β₂. The long run elements are shown by λ₁. The words with the delta (∆) sign denote short-run elements, while the lambdas (λ) represent the long-run elements. The study converted the model into log-linear form to evaluate the consumption function and measure the direct elasticities (Aropova, 2018). The heteroscedasticity problem can also be reducing by taking log.

The short run estimates can be depicted using Error Correction Mechanism (ECM) in the presence of co-integration, according to the Granger representation theorem (Maddala, 1992). Eq. (3) determines the representation of ECM as follows:

$\varDelta ln$ + ${\sum }_{t=1}^{p}{\beta }_{1}\varDelta ln\left({CON}_{t-i}\right)$ + ${\sum }_{t=1}^{p}{\beta }_{2}\varDelta ln\left({Y}_{t-i}\right)+{\lambda }_{1}ln\left({Y}_{t-i}\right)+\varPhi ECM$_t−I + ${\epsilon }_{t}$……. (4)

Where ECM_t−1 is the error correction mechanism term and reflects the rate at which the system adjusts to long-run equilibrium.

This section presents the results and discussion of the study. The section consists of descriptive statistics, unit root (i.e., ADF) and ARDL tests results. In Table 2 the descriptive statistic of the consumption and income for the central Asian countries are given. The data are in US million dollars. For Kazakhstan, the average consumption is 62845.87 million US dollars and income is 112087.8 million US dollars on average. The maximum consumption is 114574.4 million dollars and maximum income is 186205.1 million dollars. Similarly, for Kyrgyz data, the average consumption is USD 4248.882 million, while income is 4670.84 US million dollars. The maximum consumption and income is 6759.12 and 7320.68. This implies that in Kyrgyz about 90 percent income is spending on consumption expenditures. The average consumption is 21815.13 million dollars, while the average income is 36621.81 million dollars in Tajikistan. This implies that 60 percent of income is spending on consumption expenditures in Tajikistan. Consequently, the average consumption and income is 3846.04 and 5016.87 US million dollars in Uzbekistan. Kazakhstan and Tajikistan are rich countries economically as compare to Kyrgyz and Uzbekistan. The more income they have, more they consumed.

Table 2: Descriptive statistics summary results

Counties	Kazakhstan		Kyrgyzstan		Tajikistan		Uzbekistan
Variables	C	Y	C	Y	C	Y	C	Y
Mean	62845.87	112087.8	4284.88	4670.84	21815.13	36621.81	3846.04	5016.87
Median	55712.93	109503.8	4414.10	4479.98	11060.43	20242.11	3308.44	4095.12
Maximum	114574.4	186205.1	6759.12	7320.68	57103.07	87622.91	9852.44	11296.34
Minimum	28829.80	58837.41	2154.31	2601.51	5610.551	9542.79	209.023	824.392
Std. Dev.	30402.89	44132.52	1562.81	1498.95	17784.89	27489.10	3124.68	3852.65

Note: Data are in US dollar

Table 3: APC, MPC and MPS in Central Asian Countries

Country	Kazakhstan			Kyrgyz Republic
Year	APC	MPC	MPS	APC	MPC	MPS
1993	0.71	---^*	----	0.76	---	---
1994	0.78	0.04	0.96	0.78	0.63	0.37
2000	0.62	0.63	0.37	0.66	0.58	0.42
2007	0.45	0.43	0.57	0.87	0.65	0.35
2014	0.48	0.70	0.30	0.96	0.31	0.69
2020	0.53	0.41	0.59	0.75	0.88	0.22
Year	Tajikistan			Uzbekistan
1993	0.13	---^*	----	0.56	----	----
1994	0.26	0.40	0.60	0.50	0.31	0.69
2000	0.79	0.92	0.08	0.57	0.87	0.13
2007	0.88	0.38	0.62	0.49	0.48	0.52
2014	0.83	0.46	0.54	0.62	0.57	0.43
2020	0.73	0.82	0.18	0.58	0.74	0.26

* Data for the aggregate consumption and GNI for the selected countries were not available on World Bank, World Development Indicators, 2022. Hence, the MPC and MPS are not computed for the 1993.

The table 3 shows the important components of Keynesian consumption function i.e. average propensity to consume (APC) and marginal propensity to consume (MPC). APC falls with increase in income. APC is greater than 1 (APC > 1) when consumption is greater than income. APC is equal to one where consumption is equal to income (APC =1). APC < 1, when consumption less than income. MPC value lies between 0 and 1 (0 < MPC < 1). MPC of poor is more than rice. MPC falls with successive increase in income (Keynes, 1936). In 1994, the APC was 0.78, MPC 0.04 and MPS was 0.96 in Kazakhstan. As the income increases, the APC decreases. Similarly, during 2020 APC, MPC and MPS was 0.53, 0.41 and 0.59 respectively. In case of Kyrgyz, during 1994 APC was 0.78, MPC was 0.63 and MPS was 0.37. In 2020, APC was 0.75, MPC was 0.88 and PMS was 0.22. On the same way, in Tajikistan APC was 0.26, MPC was 0.40 and MPS was 0.60. Similarly, APC was 0.73, MPC was 0.82 and MPS was 0.18. With increase in income, the APC declines gradually. The APCs, MPCs and MPSs of Uzbekistan are also shown in the table during the time period. Kyrgyz and Uzbekistan were remains poor countries as compared to Kazakhstan and Tajikistan.

As the current study is based on time series data, it is vital to ensure that the series is stationary or not? The data was checked for stationarity by using the ADF test. According to the null hypothesis, the variable is not stationary. The alternative hypothesis, on the other hand, is that the variable is stationary. When the estimated ADF value in absolute form is greater than the absolute critical value at 1% or 5%, the null hypothesis of the presence of unit in the data is rejected. The ADF results for each country are given in Table 4. The ADF results show that all the variables have unit root at level but become stationary at 1^st difference at 1% and 5% levels of significance. Now the data is stationary and suitable for regression.

Table 4: Augmented Dickey Fuller (ADF) test Results

Country	At level		At 1^st Difference
Country	C	Y	C	Y
Kazakhstan	-1.3567	-0.6077	-3.8382^***	-3.9048^***
Kyrgyz Rep	-0.3788	0.1279	-4.1811^***	-5.7485^***
Tajikistan	-1.1796	-1.1102	-2.9773^**	-2.9147^**
Uzbekistan	-1.3163	-0.8044	-4.8341^***	-3.5932^***

Note: ** & *** shows significance level at 5% and 1% respectively

5.2 ARDL Co-integration Result (Bound’s Test)

According to Pesaran et al. (2001) the ARDL bounds test is used to test for co-integration. The main advantage of the ARDL technique is, that one can estimates co-integration regardless of whether the elements are integrated in the same order or not. Another benefit of the approach is that it analyzes both long- and short-term results at the same time. Because of its limited sample size feature, the ARDL technique outperforms Johansen co-integration. The null hypothesis of the approach indicates that there is no co-integration between the components, while the alternative hypothesis states that there is co-integration. The ARDL co-integration test produces two bounds: a lower bound and an upper bound. If the estimated F-statistic value is greater than the critical F-statistic value for the upper bound, then null hypothesis of no co-integration is rejected. In contrast, the null hypothesis isn’t rejected when the calculated F-statistic value is less than the F-statistic value of the lower bound. However, if the estimated value of the F-statistics lies between the lower and upper bounds, the bound test is inconclusive.

Table 5: Bound’s test Results

Countries	F-statistic Value	Countries	F-statistic Value
Kazakhstan	23.58^***	Tajikistan	5.80^**
Kyrgyz Rep	6.77^**	Uzbekistan	40.83^***

Source: Null Hypothesis: No long-run relationship exists,

The presence of a long-run association between the elements in the model is demonstrated via co-integration analysis. Table 5 showed the results of the bounds test for co-integration among factors that influence consumption. The computed F-statistic value for Kazakhstan and Uzbekistan is greater than the F-critical values at 1%. In case of Kyrgyz and Tajikistan, the computed values (5.80 and 6.77) are greater than critical values. The bounds test result for co-integration demonstrated the occurrence of a long run connection between the factors, thus the null hypothesis is rejected. The bounds result of ARDL affirmed that the variables are co-integrated.

5.3 Long Run Results

The long run results of the consumption function are presented in table 6. The coefficient of income is 1.32 which demonstrates significant impact on consumption of Kazakhstan at the 1 percent level of significance showing that if income increases 1%, consumption increases by 1.32% in the long run. This result is justified and verified by consumption function, as income of the Kazakhstan increases, consumption also increases. The outcome is in accordance with the theory. The same result was found by Laiqat et al. (2018) for China, Gahtani et al. (2020) for Saudi Arabia. In case of Kyrgyz, the coefficient of income is 1.23, which depicts a positive and significant effect on consumption at 1 percent significance level. In the long run, a 1% rise in income increases consumption by 1.23 percent. In Tajikistan, a 1% rise in income leads to a 1.07 percent increase in consumption over time. The same result was found by Khan et al. (2012) for Pakistan; Gahtani et al. (2020). In case of Uzbekistan, the coefficient of Income is 0.97, which means a significant and positive effect on consumption. With 1 percent rises in income, consumption will rise by 0.97 percent in the long run. Similarly, result was founded by Xiao and Liao (2018) for China. The results robustness is also checked by panel result. The panel result showed the income has also favorable (positive) and significant impact on consumption in the selected central Asian countries in the long run.

5.4 Short Run Results

The ARDL technique is used to estimate the ECM (Error Correction Mechanism) short term results. In case of Kazakhstan, the income has positive and statistically significant effect on consumption at 1 percent significance level. With 1 percent increases in income, consumption will increase by 0.63 percent. The ECM value is significant and with correct negative sign, showing the speed of adjustment and confirming the co-integration between variables. If there is any short-term disequilibrium, it will return to equilibrium in one year by 48 percent. Similar result was found for Kyrgyz, the income has also significant impact on consumption.

On the same way, In Tajikistan, income has a positive and significant impact on consumption at a 1% level of significance. If 1 percent increases in income consumption will increase by 0.98 percent means more elastic. The ECM value is significant and with negative sign shows, if there is any short-term disequilibrium, it will return to equilibrium in one year by 50 percent.

In case of Uzbekistan, the income has positive and statistically significant effect on consumption at 1 percent level of significance. If income increases by 1 percent, consumption will increases by 1.32 percent. But this result is opposed to the consumption function as Keynes argued that the value of MPC lies between 0 and 1. The ECM value is significant and with correct sign shows the speed of adjustment. If there is any short-term disequilibrium, it will return to equilibrium in one year by 50 percent.

Table 6: ARDL Results (Individual countries and Panel)

Variables	Kazakhstan	Kyrgyz Republic	Tajikistan	Uzbekistan
Y	1.3129^*** 26.3129	1.2314^*** 11.7028	1.0732^*** 42.2472	0.9669^*** 19.7392
C	4.2121^*** 7.3088)	2.0396^** 2.3097	1.2917^*** 4.9453	0.0194 0.0478
Panel Long Run Results
Variable	Coefficient	Std. Error	t-Statistic	Prob
Y	1.0315	0.0360	28.6313	0.0000^***
C	0.0681	0.0512	1.3317	0.1861
*Short run results*
∆(Y)	0.6262^*** 9.3537	0.6135^*** 4.9433	0.9820^*** 9.3528	1.3166^*** 10.0485
ECM (COINTEG)	-0.4770^*** -8.5389	-0.4982^*** -4.7374	-0.6656^*** -3.3851	-0.5174^*** -8.3687
R²	0.99	0.96	0.99	0.99
Adj-R²	0.98	0.95	0.99	0.98
F-statistic	8.27	278.95	877.55	734.88
DW	2.38	2.04	2.04	2.28
*Panel short run results*
Variable	Coefficient	Std. Error	t-Statistic	Prob
D(Y)	0.4205	0.1459	2.8818	0.0049^***
COINTEG	-0.2544	0.0982	-2.5909	0.0111^***

Source: Author’s Calculation, * & *** shows significance level at 10% and 1% respectively (t-value)

These results are according to the AIH (Keynes, 1936), and PIH (Friedman, 1957). The short run results supported the AIH Keynesian theory, while the long run results supported and verified the PIH of Milton Friedman theory. Similar result also found by Laiqat et al. (2018) for China; Xiao and Liao (2018); Gahtani et al. (2020) for Saudi Arabia; Yasmeen et al. (2019) for Pakistan. But in case of Uzbekistan the coefficient value is greater than 1 in the short run, which contradict with consumption function, as Keynes argued that MPC value lies between 0 and 1 (0< MPC >1). Similarly, the panel result (aggregated result) showed the same short run result as the time series result (disaggregated result).In addition, estimated consumption function for selected central Asian countries is given in Figures 2 to 6.

One of the most important components of GDP is consumption. It is impacted by a number of elements, the most important of which, according to John Maynard Keynes, is income. The link between income and consumption is depicted by the consumption function. The study's major goal is to empirically analyze the consumption function for Central Asian countries. The consumption function was estimated using time series data and the ARDL method. The findings demonstrated a long-term relationship between income and consumption. The findings revealed that income has a long-term positive and significant impact on consumption in the selected Central Asian countries. The study confirmed and verified the Keynes AIH (current income) by short run results and Friedman PIH by long run results for the selected countries. However, the results contradict with Keynesian consumption function as the value of MPC is greater than 1 in Uzbekistan.

On the other side, if MPC increases MPS decreases because of the inverse relationship exiting between the two. Therefore, it is suggested, that the policy makers shall take steps for increasing the savings of the people. According to Keynes, if saving increases investment also increases. More investment means more production. If production increases, employment opportunities created, and the surplus production will be exported to the international market for earning foreign exchange reserves.

Availability of data and materials

Data can be available on request.

Competing interests

The authors declare that they have no competing interests.

Funding

No funding received for this article.

Authors' contributions

The MAK is the main author of this research; other two co-authors namely MAZ and MTR reviewed the paper and improved the quality of paper by qualitative and quantitative analyses. All authors read and approved the final manuscript.

Author details

^{1, 2, 3} Department of Economics, Faculty of Business & Economics, Abdul Wali Khan University Mardan, Khyber Pakhtunkhwa, Pakistan.

Acknowledgements

Author is thankful to his Supervisor and Co-supervisor for their worthy and fruitful comments.

Abeysinghe, T., & Choy, K. M. (2004). The aggregate consumption puzzle in Singapore. Journal of Asian Economics, 15(3), 563-578.
Akekere, J., & Yousuo, P. O. J. (2012). Empirical analysis of change in income on private consumption expenditure in Nigeria from 1981 to 2010. International Journal of Academic Research in Business and Social Sciences, 2(11), 321.
Al-Gahtani, G., Bollino, C. A., Bigerna, S., & Pierru, A. (2020). Estimating the household consumption function in Saudi Arabia: an error correction approach. Applied Economics, 52(11), 1259-1271.
Altunc, O. F., & Aydın, C., (2014). An estimation of the consumption function under the permanent income hypothesis: The case of D-8 countries. Journal of Economic Cooperation and Development, 35(3), 29-42
Ammad, S., & Ahmed, Q. M. (2020). Comparative efficacy of life-cycle and absolute income hypothesis in Pakistan. The Journal of Developing Areas, 54(1), 104-120.
Ando, A., & Modigliani, F. (1963). The life cycle hypothesis of saving: Aggregate implications and tests. The American economic review, 53(1), 55-84.
Arapova, E. (2018). Determinants of Household Final Consumption Expenditures in Asian Countries: A Panel Model, 1991–2015. Applied Econometrics and International Development, 18(1), 121-140.
Barot, B. (2006). Empirical studies in consumption, House prices and the accuracy of European growth and inflation Forecasts. 1(3), 47-59.
Bayar, A., & McMorrow, K. (1999). Determinants of private consumption (No. 135). Directorate General Economic and Financial Affairs (DG ECFIN), European Commission, Brussels.
Bilik, M., & Recep, K. Ö. K. (2020). A Comparative Study on consumption functions: The case of the European Union. EGE Academic Review, 20(3), 167-178.
Bin, X., & Renjing, X. (2012). An Empirical Analysis on the Consumption Structure of Town Residents, Jiangxi Province—Based on the Extended Linear Expenditure System Model (ELES). Physics Procedia, 24, 660-666.
Breido, K., & Tregub, I. V. Keynesian theory of consumption: theoretical and practical aspects. The Finance University under the Government International Finance Faculty, MOSCOW, Russia.
Bunting, D. (1989). The consumption function “paradox”. Journal of Post Keynesian Economics, 11(3), 347-359.
Carroll, C. D. (2001). A theory of the consumption function, with and without liquidity constraints. Journal of Economic perspectives, 15(3), 23-45.
Chen, H. M. (2010). Research on the Consumption Function of Rural Residents in Hunan Province. Asian Agricultural Research, 2(1812-2016-143618), 47-53.
Cheung, Y. W., & Lai, K. S. (1995). Lag order and critical values of the augmented Dickey– Fuller test. Journal of Business & Economic Statistics, 13(3), 277-280.
Choudhury, U. D. R. (1968). Income, consumption and saving in urban and rural India. Review of Income and Wealth, 14(1), 37-56.
Dhakal, D., Kulkarni, K. G., & Upadhyaya, K. P. (2009). Consumption pattern: Open economy setting: A case of India. SCM Journal of Indian Management, 6, 01-11.
Dong, D., Gozgor, G., Lu, Z., & Yan, C. (2021). Personal consumption in the United States during the COVID-19 crisis. Applied Economics, 53(11), 1311-1316.
Drakopoulos, S. A. (2021). The marginalization of absolute and relative income hypotheses of consumption and the role of fiscal policy. The European Journal of the History of Economic Thought, 28(6), 965-984.
Friedman, M. (1957). The permanent income hypothesis. In a theory of the consumption function (pp. 20-37). Princeton University Press.
Gerdtham, U. G., & Johannesson, M. (2004). Absolute income, relative income, income inequality, and mortality. Journal of Human Resources, 39(1), 228-247.
Habanabakize, T. (2021). Determining the household consumption expenditure’s resilience towards petrol price, disposable income and exchange rate volatilities. Economies, 9(2), 87-102.
Keho, Y. (2019). Myopia, liquidity constraints and private consumption: the case of Cote d’Ivoire. Cogent Economics & Finance.
Keynes, J. M. (1937). The general theory of employment. The quarterly journal of economics, 51(2), 209-223.
Khalil, E. L., Aimone, J. A., Houser, D., Wang, S., Martinez, D., & Qian, K. (2021). The aspirational income hypothesis: On the limits of the relative income hypothesis. Journal of Economic Behavior & Organization, 182, 229-247.
Khan, A. Q., Azam, M., & Qureshi, M. E. (2014). Analysis of consumption behavior concerning current income and lags consumption: Empirical evidence from Pakistan. Ekonomski vjesnik: Review of Contemporary Entrepreneurship, Business, and Economic Issues, 27(1), 59-70.
Khan, H. (2014). An empirical investigation of consumption function under relative income hypothesis: Evidence from farm households in northern Pakistan. International Journal of Economic Sciences, 3(2), 43.
Khan, K. and Manzoor, H. Memon. Che, (2012). The Hall’s Permanent Income Hypothesis: A Case Study of Pakistan. Asian Economic and Financial Review, 2(4), 518-522.
Khan, K., Ali, M., & Nishat, M. (2010). Estimation of Consumption function under the PIH: Evidence from Pakistan. Kashmir Economic Review, 19(1), 91-96.
Khan, S., Reid, B. J., Li, G., & Zhu, Y. G. (2014). Application of biochar to soil reduces cancer risk via rice consumption: a case study in Miaoqian village, Longyan, China. Environment international, 68, 154-161.
Kök, R., Yalçınkaya, E. A., & Bilik, M. (2020). Household consumption function in Turkey: Rural- urban analysis (2004-2013). Ekonomik Yaklasim, 31(116), 341-376.
Kuznets, S. (1946). Introduction to" National Income: A Summary of Findings". In National Income: A Summary of Findings (pp. 1-2). NBER.
Kwan, Y. K. (2009). The Direct Substitution between Government and Private Consumption in East Asia (pp. 45-62). University of Chicago Press.
Liaqat, S., Bunnika, P., & Khan, K. (2018). China’s consumption function: an empirical test of absolute and permanent income hypothesis. Global Economics Review (GER), 3(1), 90- 97.
Maddala, G. (1992). Introduction to Econometrics. New York: Macmillan Publishing Company.
Manitsaris, A. (2006). Estimating the European Union consumption function under the permanent income hypothesis. International Research Journal of Finance and Economics, 2(3), 1450-2887.
Nikbin, B., & Panahi, S. (2016). Estimation of private consumption function of Iran: autoregressive distributed lag approach to co-integration. International Journal of Economics and Financial Issues, 6(2). 85-99.
Ofwona, A. C. (2013). An estimation of the consumption function for Kenya using Keynes' Absolute Income Hypothesis for the period 1992-2011. Journal of Emerging Trends in Economics and Management Sciences, 4(1), 103-105.
Pesaran, M. H., Shin, Y., & Smith, R. J. (2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics, 16(3), 289–326.
Schalembier, B., Bleys, B., Van Ootegem, L., & Verhofstadt, E. (2020). How the income of others affects the life satisfaction of materialists. Journal of Economic Behavior & Organization, 174, 64-74.
Shaikh, S. A. (2018). An analysis of consumption function: A case study of upper middle-income economies. NICE Research Journal, 223-235.
Wang, N. (2003). Caballero meets Bewley: The permanent-income hypothesis in general equilibrium. American Economic Review, 93(3), 927-936.
World Development Indicators (WDI) (2022). The World Bank data bank.
Xiao, Y., & Liao, L. (2018). Empirical research of Chinese residents’ consumption behaviors based on consumption functions. American Journal of Industrial and Business Management, 8(06), 1517-1535.
Yasmeen, G., Hashaam, M., & Sattar, L. (2019). A pragmatic analysis of consumption function and Keynesian theory of consumption for rural population of district Chitral, Khyber Pakhtunkhawa province of Pakistan. International Journal of Business, Economics and Law, 18(3), 2289-1552.
Zhu, Y., & Jin, D. (2011). Difference of urban-rural residents' income and consumption in Chongqing city from 2000 to 2009. Asian Agricultural Research, 3(1812-2016-143568), 59-63.

Estimating Consumption Function for Central Asian countries: New Insights

Status:

Version 1

Abstract

Figures

1. Introduction

2. Literature Review

2.1 Theoretical Literature

2.2 Empirical Literature

3. Data and Methodology

3.1 Data and its Sources

3.2 Empirical Model

4. Estimation Techniques

4.1 Augmented Dicky Fuller (ADF)

4.2 Auto-Regressive Distributed lag (ARDL) Model

5. Result and Discussion

6. Concluding Remarks

Declarations

References

Status:

Version 1