The National Accounts, Household Service Consumption and Its Monetization

KYKLOS,Vol.42, 1989, Fasc. 1, 3-15

The National Accounts, Household Service Consumption and its Monetization

DAVID L. HAMMES, JEAN-JACQUES ROSA and HERBERT G. GRUBEL *

1. INTRODUCTION

There is much interest in statistics coming from systems of national income accounts. National income data calculated through the accounts allow comparisons across countries of national income levels, both in the aggregate and on a per capita basis, as well as allow- ing comparisons of rates of changes in these measures. Such compari- sons are often used to rate the relative performances of economies around the globe.

Dissatisfaction with using statistics such as gross national product (GNP) or GNP per capita as a measure of comparing welfare or well- being across economies is well known and much discussed ’. Ques- tions of the dispersion of income, omissions from GNP and the com- position of output all focus on the weakness of such measures when used to make inter-country welfare comparisons or intra-country comparisons over time. While these problems are well known it is the case that imperfect, but consistent, measures have been revealed preferred to no measure at all, hence their frequent use.

* Respectively Assistant Professor of Economics, the University of Hawaii at Hilo, USA, Professor of Economics, lnstitut d’Etudes Politiques de Paris, France, and Professor of Economics, Simon Fraser University, Vancouver, Canada.

This work was initiated while HAMMES was at the Fraser Institute, Vancouver, Canada.

The views expressed are the authors’ alone and not the aforementioned organiza- tions.

1. While we use the concept of gross national product for illustrative purposes our discussion is aimed also at any of the national income accounting measures, including gross domestic product (GDP) and the net measures of national and domestic pro- ducts.

3

DAVID L. HAMMES, JEAN-JACQUES ROSA AND HERBERT G.GRUBEL

One area of admitted weakness in the national accounts is the omission of work performed in the household that is not marketed through formal markets. Common examples are the labour services of housewives and neighbours trading gardening and home or car repair services. There has been much recent work attempting to quantify the size of these non-marketed activities in addition to the work quantifying ‘underground’ or unreported economic activity in an economy ’.

These studies generally conclude that once omissions of legal and illegal activity are accounted for in measures of national products inter-country differences, on a per capita basis, in the national ac- counts magnitudes are much reduced. The studies also show that, for a given country, the measured changes in the per capita magnitudes as calculated from the national accounts may be biased downward by the growth of unreported activities 3 .

Omissions from the national accounts as well as illegal and unre- ported activities all understate the actual levels of national products and incomes. Other than descriptive accounts there are few, if any, discussions of how the national accounts may be systematically overstated ‘.

This is unfortunate as we believe that social trends since 1960s and current national accounting conventions lead to the systematic over- statement of national incomes and product. The nature of the over- statement we discuss is that which is caused by a reduction in the level of omitted economic activity relative to total economic activity 5 . As

2. See FEICE [1984], HAWRYLYSHYN [1978] and SWINAMER [1985]. 3. This growth may be for many reasons. For example, rising marginal tax rates on

personal incomes may raise the proportion of economic activity in non-reported sectors relative to reported sectors. As a result, total incomes may remain virtually the same, but measured national income will fall.

4. These descriptike accounts of how the national accounts system overstates incomes and products usually revolve around the failure in national accounting to include the negative externalities inherent in producing measured national income. See NORDHAUS and TOBIN [1972].

5. This was first done by GRUBEL and HAMMES [1987] and the formal analytics of the theory that follows is presented in that paper. GRUBEL and HAMMES refer to the process of moving economic activity from an omitted to included market as monet- ization’ and, for want of a better word, we continue that usage.

4

NATIONAL ACCOUNTS, HOUSEHOLD CONSUMPTION AND MONETIZATION

activity previously omitted from the national accounts is captured in formal market national income figures will rise ceteris paribus even though there has been no real increase in total economic activity, incomes or product.

More concretely, since the middle of the 1950s in the U.S. and Canada and since the early 1960s in France, there has been the social phenomenon of women entering or re-entering the labour forces in those countries at a steady rate. This may be for many reasons. We mention two. First, following the second World War, family forma- tion and child-bearing and rearing activities kept many women out of the formal work force. As children reached school age mothers’ entry into the labour force became less costly and therefore increased. Secondly, women’s investment in human capital, especially in the 1970s continuing into the 1980s, has increased their opportunities in the market relative to the home.

As the labour force participation rate of women rises women have less time to devote to economic activity in the home. Given that there is not a desired measurable decrease in the standard ofcleanliness and quality of life in the home, services previously provided by women and omitted from the national income accounts will now have to be provided through the market. Consequently, measured national income rises6. But, because women’s work in the home is omitted from the national income accounts in the first place, there is no corre- sponding decrease in national income reflecting women’s decreased home production. Therefore, we argue much of the rise in national incomes due to monetization is illusory.

This illusory overstatement, if statistically verified, is important for the following reason. Discussions of the growth in national in-

6. Assuming the labour force participation rates of men to be invariant to this shift.

The increase in measured national incomes consists of two components, the first is the statistical illusion of a rise due to the transfer of activity from the omitted markets to the included markets. The second and very real increase is due to the re-allocation of resources through specialization to their highest valued uses. Assuming the utility of the two jobs equal, labour will only switch from the omitted to included markets when the wages available in the formal market dominate those implicitly available in the home. Consequently, one is able to join the labour force, pay for house cleaning services and have purchasing power left over. National income has therefore risen.

5

DAVID L. HAMMES, JEAN-JACQUES ROSA A N D HERBERT G. GRUBEL

come aggregates may be seriously flawed if account is not taken of the monetization of activity. We provide here the results of such a statisti- cal analysis for the U.S., Canada and, for the first time, France’.

Figure I

Real and Nominal Expenditures on Services as a Proportion of Real (1971 constant dollars) and Nominal Total Consumption Exp.

0.55

0.50

0.45

0.40

435

a30 1‘

United States 1950-1982

...* .....- ...‘nominal

Source: U.S. Dept. Commerce/Bureau of Economic Analysis, 1986

7. These calculations provide ‘truer’ price and income elasticities for services as well.

There is another interesting reason to study this problem. Discussions in political economy regarding inter-country comparisons often show the more ‘socialized’ western countries with the highest measured per capita incomes relative to the more ‘market oriented’ economies. However, this may be a phenomenon of the statistical illusion we discuss above. Countries which provide more social services, child care and the like will have higher rates of female participation in the formal work force. This is because the opportunity costs of leaving the home are lowered by state subsi- dized services. The lower levels of economic activity falling into the omitted markets relative to the formal markets - what we would refer to as higher levels of monetiza- tion - in these economies will be captured by national income accountants and as a result the measured national incomes and products will be higher. We provide first estimates for the U.S., Canada and France and leave this topic for further research.

6

NATIONAL ACCOUNTS, HOUSEHOLD CONSUMPTION AND MONETIZATION

0.55

0.H)

o*45

0.40

0.35

0.30

Canada 1961-1985 - ...........

\ real

- ...... ......... ...*"o"i&

-

65 d4 6b 6$ 7b 7!2 711 76 7'8 8b 8h 8h

11. PRICES, INCOMES AND SERVICE EXPENDITURES

In order to formally estimate the magnitude of the upward bias in- herent in national income accounts due to the monetization phenomenon we assume the following. In the household monetiza- tion is the replacement of own-produced economic activity by the market purchase of services. While it is undoubtedly true that some goods have replaced own-produced effort in the home we argue that in the main marketed services, cleaning, child care, gardening, car care and the like, replace omitted own-produced services. Conse- quently, if our hypothesis is true, we expect real expenditures on services as a proportion of total consumption expenditures to rise with increases in the labour force participation rates of women. As can be seen in Figures I , 2 and 3, for the US., Canada and France

7

DAVID L. HAMMES, JEAN-JACQUES ROSA AND HERBERT G . GRUBEL

Figure 3

0.325 1968 1970 #72 19# 1976 1978 1980 1982 1984 1986

Source: InseC, Consommation des menages par produit.

respectively, this pattern holds in two of the three cases8. Working against the rise in the proportion service expenditures are relative to total consumption expenditures is the fact that relative prices of services have risen in all three countries over the relevant time periods. This may explain the almost constant fraction of real total expendi- ture devoted to services in Canada.

8. The time periods looked at in the three countries vary. In the U.S. we consider the period 1950 to 1982; in Canada the period runs from 1961 to 1985, and in France, 1968 to 1986. These differences are due to data limitations. In Canada, recently revised data has become available, but the revisions were extended back only to 1961. In France, a consistent set of data on services was only available post-1968.

Sources for the data: U S : U.S. Department of Commerce/Bureau of Economic Analysis, The National Income and Product Accounts of the United States, 1929-1982, Stutzsticai Tubies, Washington D.C., 1986; for Canada: Statistics Canada, CANSIM Matrix 6708, series D10131 and D10147; and for France: Insee, Consummation des mtnages par produit.

8

NATIONAL ACCOUNTS. HOUSEHOLD CONSUMPTION AND MONETIZATION

Finally, the impact of rising real incomes must be separated from the effect of monetization. It may be claimed that a rising proportion of real service expenditures reflects nothing more than the presumed highly income elastic nature of the demand for services9. Therefore, we include in our modelling real income as a separate variable in order to keep the two influences, monetization and real income, separate.

111. THE MODELS AND ESTIMATION

We are attempting to measure the impact of real income, relative prices and monetization on the demand for real consumer services. Following earlier work on the demand for services we use real total consumption expenditures as a proxy for real total income and put all relevant variables on a per capita basis lo; following from the discus- sion in Section I above, we use the female participation rates for women as a proxy variable for the degree of monetization ‘ I , and use price indices of services divided by price indices of goods as the rela- tive price variable.

As a starting point, we use traditional analysis and attempt to explain the demand for services as a function of prices and incomes alone. The consumption of services (S) is an increasing function of total expenditures (TE), both expressed in real terms and per capita, and a decreasing function of the relative price of service in the con- sumption basket (PS). Assuming that the supply of services is infinite- ly elastic at observed prices assures that we are estimating the proper demand elasticities.

9. However, in earlier treatments of the demand for services this monetization variable was omitted. Therefore, it is highly likely that the elasticity of the included income variable is overstated in previous estimates as monetization and income are positively related.

10. The standard references are HOUTHAKKER and TAYLOR [1966], PRAIS and HOUTHAKKER [I9711 and PHILIPS [1974].

11. Sources: for the US.: US. Government Printing Office, Economic Report of the President, 1986, Table B-34, page 292. For Canada, Statistics Canada, Series CSI 1-516E Cat. No. 71-529. For France, Insee as noted in footnote 8.

9

DAVID L. HAMMES, JEAN-JACQUES ROSA AND HERBERT G.GRUBEL

We estimate in the natural logarithms of the variables in order to evaluate the coefficients as elasticities directly. The equation estimat- ed is:

Ln(S) = C + Ln(TE) + Ln(PS). (1)

The results for the three countries are shown in Table I

Table I

Income, Prices, and the Demand for Market Services

Variable C Ln(TE) Ln(PS) RZ(Adj.) DW

us - 1.651 1.092 0.25 ,998 1.64 *

Canada - 1.175 1.172 -0.39 .989 0.42

France - 5.16 1.44 - 0.297 ,993 2.04 *

(- 3.3) (18.62) ( 2.98)

(- 8.6) (16.81) ( - 2.48)

(- 3.61) (9.55) (- 1.97)

Student’s t-statistic in parentheses * Cochrane-Orcutt correction for first order autocorrelation

The income elasticity of the demand for services is 1.1 or greater in all three cases and the price elasticity of demand has the expected sign in two of the three cases.

First order autocorrelation was evident in all three countries and we leave the Canadian case uncorrected as an example. Existence of autocorrelation indicates the omission of at least one explanatory variable.

We now add the female labour force participation rate as an expla- natory variable in order to estimate the degree to which the demand for services is a function ofmonetization. The equation for estimation becomes 12:

Ln(S) = C + Ln(TE) + Ln(PS) + Ln(F) (2)

The results appear in Table 2.

12. We retain the assumption that the supply of services is infinitely elastic

10

NATIONAL ACCOUNTS, HOUSEHOLD CONSUMPTION AND MONETIZATION

Table 2

Female Participation and the Demand for Market Services

Variable C Ln(TE) Ln(PS) Ln(F) RZ(Adj.) DW

us 1.049 0.639 - 0.06 0.446 .999 1.74 *

Canada 1.009 0.348 -0.007 0.703 ,999 2.07

France - 4.20 0.953 -0.338 0.982 .994 2.00 *

( 1.08) (8.45) (- 0.89) (3.47)

( 6.86) (6.09) (- 0.13) (15.42)

( - 1.38) (2.73) (- 2.61) (2.57)

The inclusion of the female participation rate has the expected sign and, in the Canadian case, eliminates the first order autocorrelation. In all cases, as expected, the income elasticity of the demand for ser- vices is lowered when the monetization effect is accounted for. In all cases this elasticity is lower than one indicating that the demand for services is income inelastic. While the relative price effects are statisti- cally insignificant in the U.S. and Canadian cases, the sign is as hypo- thesized in all cases and statistically significant for France.

IV. SUMMARY AND ECONOMIC IMPLICATIONS

We argued in Section I and I1 above that the omission of household economic activity in the national income accounts would lead to an overstatement of national incomes and their growth rates as eco- nomic activity shifted from the unrecorded household sector into the included market sector. This growth in GNP we argued would be a statistical illusion and inter-country and intra-country comparisons made upon published data may yield spurious results. Our estimates imply the foilowing.

In the U.S. the female labour force participation rate rose from 33.3 percent to 52.7 percent between 1950 and 1982. This is a rise of 58.25 percent. In Table 2 the elasticity of demand for services due to monetization (for the U.S.) is estimated at 0.446. This means a one percent rise in the participant rate leads to a 0.446 percent rise in the demand for services. Taking the 58.25 percent rise in the female

11

DAVID L.HAMMES, JEAN-JACQUES ROSA AND HERBERT G . GRUBEL

labour force participation rate over the entire period we estimate that monetization accounted for 26 percent (0.5825*0.446) of the growth in real service consumption in the U.S. During this period real service consumption averaged 27.5 percent of real GNP. Therefore, our results imply that 7 percent (0.26*0.275) of the increase in real GNP per capita was due to the monetization of economic activity. As ex- penditures on services as a fraction of GNP was higher in the later years this effect will be larger in those years.

In Canada we find similar results. The participation rate rose about 89 percent from 196 1 to 1985 1 3 . The elasticity of demand for services due to monetization is (from Table 2) 0.703 implying that 63 percent of the increase in consumption of services by Canadian households has been due to the demand-driven portion of monetization. The proportion of market services consumed by households has remained roughly constant at 26 percent of GNP implying that 16.4 percent of the increase in Canadian GNP per capita is due to the monetization of household services demand.

In France the results are similar again. Participation rates rose 19.2 percent from 1968 to 1986 14. The monetization elasticity of demand for services is 0.982. These results imply that 18.8 percent of the growth in real service consumption is due to monetization over that period. Service consumption average 23 percent in France over the 1968 to 1986 period15. Therefore, on average, 4.3 percent of the growth in real GDP per capita is due to the effects of monetization in France 16.

13. 28.7 percent in 1961 to 54.4 percent in 1985. 14. 38.6 percent in 1968 and 46 percent in 1986. 15. Again, as in the US. , the fraction rose in France over the period in question.

Service consumption expenditures were 19.8 percent of GDP in 1968 and 28.4 percent in 1986.

16. While the relative size of the monetization effect looks smaller for France than for the U.S. and Canada readers should note three things. First, the estimated elastici- ty of demand for service consumption through monetization (as shown in Table 2 ) is higher in France than for the other two countries; second, France had a much lower rate of growth in the female labour force participation rate; and third, the differing length of the time periods used in estimation affect the results. Interestingly, the proportion of total consumer expenditures per capita devoted to service consumption is very similar in all three economies.

12

NATIONAL ACCOUNTS, HOUSEHOLD CONSUMPTION AND MONETIZATION

One interesting implication of our findings concerns the expected growth in demand for marketed consumer services. Since the female labour force participation rates are higher in the U.S. and Canada relative to France more room for overstatement of national accounts exists for French national accounts in the future.

Another implication is that as the rates of female participation slow the conditions for growth in the demand for consumer market services will be slower than in the postwar years we examined here.

A final implication of our econometric results relates to the existing literature on the demand for services ”. Our econometric estimates of the total expenditure elasticity of demand for household-bought services of 0.64 for the U.S., 0.35 for Canada and 0.95 for France are in strong contrast with the estimated elasticity of over one in the first set of equations (reported in Table I ) and the conventional wisdom that the income elasticity of demand for such services is high.

The idea that the income elasticity of demand for services is high stems from early budget studies by ENGEL and others. During the period from which the data for these studies was drawn, services were provided by servants for persons with high incomes. In recent times, on the other hand, it is quite possible that the income elasticity of demand is low since many efficient and low-priced substitutes in the production and consumption of services have been developed.

REFERENCES

FEIGE, E.L.: ‘Microeconomics and the Unobserved Sector’, in: BLOCK, W. and WALKER, M. A. (eds.), Taxation, An In[ernatiorial Perspective, Proceedings of An International Symposium, The Fraser Institute, Vancouver, B. C., 1984.

FLJCHS, V. R.: ‘An Agenda for Research on the Service Sector’, in: INMAN, R. P. (ed.), Managing the Service Economy, Chapter 12, Cambridge: Cambridge University Press. 1985.

In the time period in question the fact that the female participation rate grew most slowly in France accounts for the difference in results. Based on the elasticities, for a given percentage change in female participation rates, the impact can be expected to be greatest in France relative to the U.S. and Canada.

17. See FUCHS [1985], INMAN [I9851 and SUMMERS [1985].

13

DAVID L. HAMMES, JEAN-JACQUES ROSA A N D HERBERT G. GRUBEL

GRUBEL, H. G. and HAMMES, D. L.: ‘Household Service Consumption and Its Moneti- zation’, Service Project Discussion Paper 86-5, The Fraser Institute, Vancouver, B. C., revised 1987.

HAWRYLYSHYN, 0.: Estimating the Value of’ Household Work in Canada, 1971, Statistics Canada, Cat. No. 13-566 Occasional, 1978.

HOUTHAKKER, H.S. and TAYLOR, L.D.: Consumer Demand in the United States, 1929-1970, Analysis and Projections, Cambridge: Harvard University Press, 1966.

INMAN, R. P.: Managing the Service Economy, Cambridge: Cambridge University Press, 1985.

NORDHAUS, W. and TOBIN, J.: ‘Is Economic Growth Obsolete?’, in: Economic Growth, NBER, Fifth Anniversary Colloquium V, New York, 1972.

PHILIPS, L.: Applied Consumption Analysis, Amsterdam, New York, Oxford: North- Holland, 1974.

PRAIS, S. J . and HOUTHAKKER, H. S.: The Analysis o j Family Budgets, Cambridge: Cambridge University Press, second impression, abridged, 1971.

SUMMERS, R.: ‘Services in the International Economy’, in: INMAN, R.P. (ed.), Manag- ing the Service Economy, Chapter 1 , Cambridge: Cambridge University Press, 1985.

SWINAMER, J.L.: ‘The Value of Household Work in Canada’, The Canadian Statisti- cal Review, March ( 1 985).

SUMMARY

National account statistics typically omit some non-marketed economic activity. Over time, if non-market activity declines relative to total economic activity, national accounts estimates of output and income and their changes will be overstated. This overstatement affects comparisons of growth for a single country through time as well as affecting comparisons among countries.

Using labor force participation rates as a proxy for the move from non-market production to market production this paper finds and provides empirical measure- ments of the systematic overstating of national accounts output data for the United States, Canada and France in the post World War I 1 period.

ZUSAMMENFASSUNG

Die volkswirtschaftliche Gesamtrechnung versiumt es im allgemeinen, nicht auf dem Markt gehandelte Aktivitaten zu registrieren. Wenn im Laufe der Zeit der Teil dieser Aktivitaten im Verhlltnis zu den registrierten Aktivithten abnimmt, uberschatzt die volkswirtschaftliche Gesamtrechnung die wirkliche Zunahme der Produktion. Dies verzerrt sowohl Wachstumsschiitzungen fur eine Volkswirtschaft im Laufe der Zeit

14

NATIONAL ACCOUNTS, HOUSEHOLD CONSUMPTION AND MONETIZATION

wie auch internationale Vergleiche des Produktionsniveaus und des Volkseinkom- mens.

Die Autoren messen empirisch diese systematische Verzerrung in den Vereinigten Staaten, in Kanada und in Frankreich, indem sie sich auf die Partizipationsrate auf dem Arbeitsmarkt als Indikator fur die Fluktuation zwischen vermarkteter und nicht vermarkteter Produktion stutzen.

RESUME

Les comptabilites nationales omettent generalement d’enregistrer certaines activites non marchandes. Lorsque la part de ces activites diminue relativement aux activites enregistrkes, les comptes nationaux vont surestimer l’augmentation effective de la production et du revenu. Ce biais affectera aussi bien I’estimation du cheminement temporel d’une economie donnee que les comparaisons internationales de niveaux de production et de revenu.

En s’appuyant sur les taux de participation au marche du travail pris comme indicateur de reallocation de la production non marchande vers la production marchande, les auteurs fournissent une mesure de I’ampleur du biais systkmatique des comptabilites nationales, dans le cas des Etats-Unis, du Canada et de la France.

IS