real income, unemployment and subjective well-being: revisiting the costs and benefits of inflation...

17
Canadian Public Policy Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada Author(s): Roderick Hill Source: Canadian Public Policy / Analyse de Politiques, Vol. 26, No. 4 (Dec., 2000), pp. 399-414 Published by: University of Toronto Press on behalf of Canadian Public Policy Stable URL: http://www.jstor.org/stable/3552608 . Accessed: 11/06/2014 03:30 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. . University of Toronto Press and Canadian Public Policy are collaborating with JSTOR to digitize, preserve and extend access to Canadian Public Policy / Analyse de Politiques. http://www.jstor.org This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AM All use subject to JSTOR Terms and Conditions

Upload: roderick-hill

Post on 12-Jan-2017

215 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

Canadian Public Policy

Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits ofInflation Reduction in CanadaAuthor(s): Roderick HillSource: Canadian Public Policy / Analyse de Politiques, Vol. 26, No. 4 (Dec., 2000), pp. 399-414Published by: University of Toronto Press on behalf of Canadian Public PolicyStable URL: http://www.jstor.org/stable/3552608 .

Accessed: 11/06/2014 03:30

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

University of Toronto Press and Canadian Public Policy are collaborating with JSTOR to digitize, preserveand extend access to Canadian Public Policy / Analyse de Politiques.

http://www.jstor.org

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 2: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

Real Income, Unemployment and

Subjective Well-Being: Revisiting the Costs and Benefits of Inflation

Reduction in Canada

RODERICK HILL

Department of Social Science

University of New Brunswick Saint John, New Brunswick

On a souvent consid6r6 les b6n6fices de la d6sinflation comme la valeur escompt6e des augmentations du revenu net qui pourraient en r6sulter. Cela conduit a une comparaison des faibles revenus d'un ch6mage A court terme accru avec l'augmentation pr6vue, h long terme, des revenus. Typiquement, ces mesures visant le bien-8tre 6conomique estiment que des gains importants r6sultent de la d6flation.

Cependant, des 6tudes concernant le bien-8tre individuel rdvilent une exag6ration des gains en matibre de bien-8tre personnel parce que le ch8mage entraine des cofits significatifs non mon6taires, tandis que les revenus moyens plus 61ev6s ne peuvent 8tre associ6s a un accroissement significatif du bien-8tre en moyenne. Une simu- lation de la d6sinflation des ann6es 90 au Canada montre que le bien-etre moyen pourrait nettement y perdre. Des gains nets significatifs ne seront possibles que si une faible inflation accroit directement le bien-~tre.

The benefits of disinflation have often been thought of as the discounted value of the net income increases that might result. This weighs lower incomes from higher short-term unemployment against expected long- term income increases. Such measures of economic welfare typically find large net gains from disinflation.

However, studies of subjective well-being show that this overstates gains in individuals' well-being be- cause unemployment has significant non-monetary costs, while higher average incomes may not be associated with significant increases in average well-being. A simulation of the 1990s disinflation in Canada shows that a net loss in average well-being could result. Only if lower inflation raises well-being directly are significant net gains possible.

he question of the relative costs of inflation and of disinflationary recessions has been a matter

of considerable debate in Canada in recent years (see Fortin 1996, 1999; Freedman and Macklem 1998; Simpson, Cameron and Hum 1998). The debate was prompted by the Bank of Canada's decision in the late 1980s to move toward a goal of price stability.

The framework within which this debate largely took place was a familiar one. Should we accept a lower average standard of living and higher unem- ployment in the short run in return for a possible increase in living standards in the more distant fu- ture? Similar questions arise in trading off the costs of industrial adjustment from changes in trade or

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 3: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

400 Roderick Hill

tax policy against possible long-term gains in real income from a more efficient use of resources.

Bank of Canada officials and some academic economists argued that there would eventually be increases in real output (among other benefits) from the reduction in inflation. Many concluded that lower inflation would lead to net gains in economic welfare, although opinion remained divided about how much disinflation was desirable.'1

My purpose here is to point out several problems with this approach. It is now widely recognized that no robust negative relationship has been found em- pirically between moderate inflations and the level or rate of growth of output (Ragan 1998). (Unless stated otherwise, "inflation" will refer only to rela- tively moderate inflation such as Canada experienced in the late 1980s.) For this reason alone, the simple framework just sketched out does not provide a case for disinflation unless lower inflation offers other benefits. Even if some relationship between infla- tion and the level or growth of output were established, we have to compare the welfare ben- efits of that extra output with the welfare costs of attaining it.

Economic policymakers are often presumed to be trying to maximize a social welfare function, some ranking of social states that combines the well-being of present and perhaps future persons. A central problem for welfare economics is the lack of an objective measure of well-being which can be used to compare individuals.

Well-being or welfare is often proxied by a mea- sure of "economic welfare," such as real income. Data about real income are readily available, al- though well-known measurement problems remain. It is also used knowing that other things, such as income distribution, also matter. But economic wel- fare is often the focus of policy discussion in the hope that economic welfare moves in tandem with welfare.

However, a well-established literature on the de- terminants of subjective well-being suggests that, for an affluent country like Canada, increases in average incomes likely contribute little to people's feelings of well-being. This conclusion was antici- pated long ago by leading welfare economists such as Pigou (1951) and Abramovitz (1959).

While likely overstating the benefits of income increases, the conventional calculation of the net benefits of disinflation also understates the costs of

unemployment. There is good evidence that the non- pecuniary welfare costs of unemployment are far greater than the welfare cost of lost incomes suf- fered by the unemployed. This evidence comes from direct measurements of subjective well-being.

These considerations weaken the case for disinflation. As Ragan (1998) argues, the case for disinflation must be made on the basis of gains other than those that lead directly to greater output and/ or growth. In this case, such gains may come from increases in subjective well-being directly associ- ated with lower inflation.

While the discussion here focuses on disinflation, the questions raised are more general. Policymakers often weigh significant losses in well-being for many persons in the short term, hoping that higher future average incomes will increase most people's well- being by enough to make the sacrifices worthwhile. Given the sometimes enormous stakes, a closer look at this trade-off is called for. Before turning to a discussion of subjective well-being, I begin with a brief review of the conventional calculation of the net benefits of disinflation,

CONVENTIONAL ESTIMATES OF THE NET BENEFITS OF DISINFLATION

Inflation in Canada fell sharply during the last dec- ade, from an average of 4.4 percent during 1986-89 to an average of 1.4 percent during 1992-99. As

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 4: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

Real Income, Unemployment and Subjective Well-Being 401

Howitt (1990, p.104) wrote, "one must somehow balance the gains from reducing inflation against the costs of doing so. The reduction in inflation should continue as long as the present discounted value of the benefits to society from a further small reduction exceeds the present discounted value of the cost."

Howitt identified a variety of sources of poten- tial gain from inflation reduction, only some of which show up directly in changes in total output. Costly price changes occur more frequently, while disruptions of relative prices misallocate resources. Taxes on capital income also induce resource misallocation because the tax system is not fully indexed. Accounting systems based on dollar val- ues distort the information on which business decisions are based. Higher inflation also shortens the horizons of decisionmakers, inhibits long-term borrowing and investment and shortens contract lengths, resulting in costly renegotiations and in- creases in strike activity.

One way to try to quantify the net gains from disinflation is what Thornton (1996) calls "Howitt's Rule," after a numerical example in Howitt (1990).2 The benefits of lower inflation are thought of as an increase in the level or the rate of growth of output. Similarly, the costs are the lost output from a tem- porary increase in unemployment.

Discounted changes in output act as a proxy for changes in welfare. Thornton (1996, pp. 25-26) ex- plains that the assumption that "the costs and benefits could be adequately represented by output gains and losses" is "common in much theoretical and virtually all applied work," but there are "[t]wo important limitations on this practice." The first is the inaccurate measurement of production, as in the omission of home production. The second is that "output does not include the utility individuals ob- tain from leisure."

This approach entails two assumptions: increased potential consumption adds significantly to well-

being, and the unemployed value their extra leisure. Both are challenged in the literature on subjective well-being.

Costs of Disinflation Table 1 shows two simple estimates of the present value of the cost of unemployment that followed the reduction in inflation. The first estimate follows Howitt's (1990) treatment of the disinflationary re- cession of the early 1980s. This attributes to disinflationary policy all unemployment between 1990 and 1999 in excess of the natural rate (assumed to be 7.5 percent).3 Okun's Law converts changes in unemployment into changes in gross domestic product (GDP): an increase in unemployment of 1

percent is associated with a decline in GDP of about 2.5 percent. The estimated cost of 49 percent of 1990 GDP is an overstatement given other sources of higher unemployment, but provides a clear upper bound to the short-term costs of disinflation.

The second estimate is based on the "sacrifice

ratio," the GDP cost of reducing inflation perma- nently by one percentage point. Debelle (1996, p.73) estimates this as 3.5 percent for the 1990-93 period. The 3.5 percentage point reduction in inflation from 1989 implies a disinflation cost of 12 percent of 1990 GDP. Any further differences between actual and potential GDP are attributed to other causes.

Gains from Disinflation A permanent decrease in inflation could increase national income. Barro (1996, p.159) finds that for countries with relatively low inflation rates, reduc- ing inflation by one percentage point per year could increase growth by about 0.016 percent annually. If so, then a 3.5 percentage point fall in inflation could raise long-run growth by 0.056 percent.

While Barro's estimate was not statistically sig- nificant, as Howitt (1997, p.48f.) points out, it could be economically significant: even very small changes in growth rates affect economic welfare in the long term. If annual growth were 0.056 percent

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 5: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

402 Roderick Hill

TABLE 1 Conventional Estimates of Costs and Benefits of Disinflation

Present Value As percent of Remarks (1990 dollars) 1990 GDP

Costs: if allexcess $328 billion 49 Assumes natural rate of unem- unemployment 1990-99 ($11,800 per capita) ployment of 7.5 percent and due to monetary policy Okun coefficient of 2.5.

Costs: using Debelle's (1996) $83 billiion 12 Assumes remaining excess "sacrifice ratio" of 3.5 ($3,000 per capita) unemployment due to other

factors than monetary policy.

Benefits: using Barro's (1996) $380-400 billion 58-62 Assumes higher growth for 30 estimate of the effect on ($13,700-14,500 per capita) years, as per Howitt (1997). growth Lower present value if disinfla-

tionary costs continue to 1999.

Net Benefits $65 to $325 billion 10 to 50 Discounted benefits equal dis- $2,300 to $11,700 per capita counted costs after 25 to 100

years; positive net benefits thereafter.

Note: Calculations assume real growth of 2 percent and a discount rate of 4 percent.

higher, after 30 years real per capita output would be 1.7 percent higher than it would otherwise be. If the growth effect then petered out, as Howitt as- sumed in his original example, the present value of the extra output would be about 60 percent of cur- rent GDP.

However, Ragan (1998, p. 24) concludes that, in the case of moderate inflations, claims of an inflation-growth link remain "based more on opti- mism than serious evidence," rendering this an unsuitable basis for disinflationary policy. For ex- ample, several studies have found a positive relationship between moderate inflations, growth, and the size of the capital stock.

Net Benefits from Disinflation

Suppose, however, that a small effect exists, even though it cannot be clearly detected in the data. In that case, as summarized in Table 1, even the calcu-

lation that attributes all excess unemployment to disinflation still implies net gains in the present value of output. The losses from recession are re- couped after about 25 years in the low-cost case and 100 years in the high-cost case.

The conclusion of net gains tends to follow from the arithmetic: the finite costs of lost output due to unemployment are outweighed by the infinite stream of higher output if lower inflation increases real output in the long term (Freedman 1991, p. 615). Other studies of disinflation use a similar frame- work to reach the same conclusion (Feldstein 1997, Coletti and O'Reilly 1998).

While this general method is widely accepted, judgements differ about what numbers should be used, leaving the extent of net gains controversial. Net losses could result with a high discount rate or if output losses were sufficiently large or long-

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 6: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

Real Income, Unemployment and Subjective Well-Being 403

lasting. Fortin (1997, p.78f.) argues that this is possible.

As is well known, such calculations of net gains only tell us whether those who gain could poten- tially compensate those who lose. When full compensation does not take place, policymakers (and everyone else) must weigh the gains and losses suffered by different individuals to judge whether a net gain has really occurred.

Disinflations have potentially important distribu- tional effects. The policy inflicts relatively large losses on some, such as the unemployed or those whose businesses would fail. Because of important non-pecuniary costs of unemployment, unemploy- ment insurance and other income security measures can offer only partial compensation. In return, cur- rent and future generations (assumed to be materially much better off than the current genera- tion) are supposed to enjoy a slightly higher income level. Such distributional effects could lead

policymakers to choose lower average incomes and what they might judge to be a more equitable in- come distribution.

The literature on subjective well-being, to which we now turn, provides a different perspective on both the welfare benefits of lower inflation and the wel- fare costs of higher unemployment that accompanies disinflation.

SUBJECTIVE WELL-BEING

Over the last 50 years, hundreds of thousands of people in dozens of countries around the world have answered questions about their feelings of subjec- tive well-being. These questions assessed such things as people's happiness with their lives as a whole or their satisfaction with their lives, two some- what different ways of trying to measure subjective well-being.4 A typical question is: "In general, how happy would you say you are?" where permitted responses could be: "very happy," "fairly happy,"

or "not too happy." A large literature has developed in the last 30 years, mostly in psychology, examin- ing the determinants of subjective well-being.5

Richard Easterlin (1974) was the first economist to examine the evidence about subjective well-being and its implications for economics. He considered the questions of interpersonal comparability that naturally arise. After all, modern welfare econom- ics with its emphasis on ordinal utility was a response to the conclusion that individuals' utilities could not be measured or compared.

Easterlin maintains that the survey data are use- ful for addressing certain kinds of questions because

in most people's lives everywhere the dominant concerns have always been making a living and matters of family life, and it is these concerns that chiefly determine how happy people are. This is not to say that the happiness of any one indi- vidual can be directly compared with that of another. But if one is concerned with comparing the happiness of sizable groups of people, such as social classes or nations, there turns out to be a marked similarity in what people typically men- tion when they are asked about the meaning of happiness ... It is this similarity in the general pattern of human life and, thus, of human con- cerns that gives credence to the comparison of self-reports on happiness for sizeable groups of people (1996, pp.132-33).

Further compelling evidence that the surveys contain useful information comes from Di Tella, MacCulloch and Oswald (2000). Using individual- level survey data for the United States and 12 European countries, they find that a set of personal characteristics, well-known in the psychological lit- erature, have similar relationships with well-being in all these countries.

These results do not imply direct interpersonal comparisons of well-being between any two indi- viduals. But they do assert that, for example, average

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 7: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

404 Roderick Hill

well-being among employed people really is higher than that of unemployed people, if one controls for other relevant characteristics.

A large psychological literature exists examin- ing the validity of measures of subjective well-being. The general conclusion is that subjective well-being measures, while not without problems, do contain meaningful information.

Real Income and Subjective Well-Being Many studies have considered the relationship between real income and subjective well-being. Easterlin (1995, 1996) sets out three broad features in the data.

First, within a particular country, persons with relatively higher incomes consistently report them- selves somewhat happier than do people with lower incomes. Second, across countries, national average real incomes and average measures of subjective well-being are positively correlated, but subject to diminishing returns. Third, and of most relevance here, rising average incomes within developed coun- tries are not accompanied by any clear increase in self-reported subjective well-being.

For example, according to Easterlin, "there has been no improvement in happiness in the United States over almost half a century in which GDP per capita more than doubled" (1995, pp. 37-38), while "[b]etween 1958 and 1987 real per capita income in Japan multiplied by a staggering five-fold ... [but] there was no improvement in mean subjective well- being" (1995, pp. 39-40).

The weak relationship with rising average real incomes may, in part, be due to the growing impor- tance of relative incomes once basic needs have been met. Psychologists have also noted the potential importance of people's rapid habituation to higher living standards, an application of psychological theories of adaption.

The importance of adaption has been noted by some economists. It led Pigou (1951) to observe that

"[i]t is very doubtful whether a moderately well-to- do man is appreciably happier now than he would be if transported back to the pre-railway age and attuned to the conditions of that age ... From a long- run standpoint, after incomes in excess of a certain moderate level have been attained, further increases in it may not be significant for ... welfare" (1951, p. 294). The claim is that tastes evolve automati- cally in a predictable way as people adapt to current conditions.

Related to adaption are shifts in social norms which follow changes in living standards. Easterlin writes that

[i]n judging their happiness, people tend to com- pare their actual situation with a reference standard or norm, derived from their prior and ongoing social experience ... Over time, however, as economic conditions advance, so too does the social norm, since this is formed by the chang- ing economic socialization experience of people (1974, pp.118-19).

Direct evidence of changing social norms can be seen in surveys: the income perceived as "neces- sary to get along" in the United States rose between 1950 and 1986 in the same proportion as actual per capita incomes (Easterlin 1996, p.143).

Frank (1999) argues that the expanded choices that come with economic growth could potentially be used to improve subjective well-being and other indicators of the quality of life.6 But the evidence suggests that the bulk of additional real expendi- tures are not being used for such purposes.

The result is the weak link between subjective well-being and real income observed in the data for affluent countries. In their study of happiness in the United States, Blanchflower and Oswald find that family per capita real income is positively associ- ated with individual happiness, although "the amount of happiness bought by extra income is not as large as some would expect" (2000, p. 12). All

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 8: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

Real Income, Unemployment and Subjective Well-Being 405

this suggests that if lower inflation did eventually lead to material living standards a few percent higher than they otherwise would have been, we should expect the gains in average subjective well-being to be modest.

Unemployment and Subjective Well-Being Unemployment is often seen as accompanied by the benefit of greater leisure, despite the fact that

research suggests that the worst thing about los- ing one's job is not the drop in take-home income. It is the non-pecuniary distress. To put this dif- ferently, most regression results imply that an enormous amount of extra income would be re-

quired to compensate people for having no work (Oswald 1997, p. 1821).

Figure 1 shows recent data on subjective well- being of the employed and the unemployed in Canada, taken from Statistics Canada's 1996 General Social Survey. Those working are happier on average than those looking for work. The differ- ence between the two groups appears large; it is about the same as that between unemployed persons and those unable to work due to long-term illness.7

It is important to note that these costs of unem- ployment are apparently borne by all the unemployed, not just the cyclically unemployed. Winkelmann and Winkelmann (1998, p. 8) found that the negative effect of unemployment seemed to be the same, regardless of age or reasons for the termination of previous employment.

FIGURE 1 Subjective Well-Being and Labour Force Status, Canada, 1996

100

82.3

S80 Working at a Job

'• 65.8 65.8 Looking for Work

_60 2 Long-Term Illness

0

S27.5

8.5 5.7

2.7 j 4 S0.4 0.5 0.02 0.2

0 Happy, Interested in Uife Somewhat Happy Somewhat Unhappy Unhappy, Unhappy,

Little Interest in Life Life's Not Worthwhile

Source: Statistics Canada (1996). General Social Survey 11. Ottawa: Supply and Services Canada.

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 9: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

406 Roderick Hill

Their longitudinal study of several thousand Ger- man men found that a person's income would have to be increased "tremendously" to offset the dissat- isfaction associated with being unemployed.8 They conclude (1998, p. 12) that because these non- pecuniary costs of unemployment to the unemployed are perhaps 15 times as large as the pecuniary costs, they must be included in cost-benefit analyses of policy.

If this is done in the example in Table 1, the lower of the two estimates of the costs of disinflation would rise from 12 percent to about 55 percent of GDP, reducing the estimated net benefits from 50 to 5 percent of GDP.9 This example illustrates the importance of taking non-pecuniary costs into ac- count and underscores the fragility of conventional net benefit calculations.

Finally, the increase in unemployment during a slump has additional costs beyond those to the un- employed themselves. The greater anxiety among the workforce about the security of their jobs or the difficulty of finding a new job is a real cost that could, in principle, show up in surveys of subjec- tive well-being. Di Tella et al. (forthcoming) report European evidence showing that a higher unemploy- ment rate lowers average well-being for everyone.

Inflation and Subjective Well-Being As has long been understood, the effects of infla- tion are not confined to output alone. For example, Irving Fisher argued that "a 'high cost of living' caused by inflation does not - not directly, at least - mean ... any lessening of the stream of goods per capita" (1928, p. 56). Instead, he saw inflation pro- ducing injustice through subtle shifts of purchasing power between persons. As a result, "popular dis- content always follows in the wake of inflation or deflation" (1928, p.100).

Shiller (1997) used a survey to explore why people dislike inflation and found a widespread per- ception that wages lag behind price changes, a popular view also described by Fisher. Shiller (1997,

p. 17) also reports that, when asked, no one claimed to have benefited from inflation. Instead, "people tend to see the causes of their income increases in

personal terms, rather than due to inflation" and "tend to be angry at someone when they see prices rise" (1997, p. 25). People's explanations of the causes of inflation tend to have a "tone of moral

indignation" (1997, p. 31).

Whether or not these popular views of inflation are well-founded, they can result in inflation hav- ing a direct negative effect on well-being. Di Tella et al. (forthcoming) consider whether the rate of inflation played any role in explaining fluctuations in average well-being. Using European data, they report a substantial negative association between inflation and well-being. They estimate that in Eu- rope "people would trade off a 1 percentage point increase in the unemployment rate for a 1.7 percent- age point increase in the inflation rate" (p.13 of manuscript). Such a high cost of inflation might provide a rationale for a disinflationary recession, as shown in the next section.

Some direct evidence about the Canadian pub- lic's concern about inflation as well as about

unemployment can be seen in public opinion polls. Figure 2 summarizes the results of a series of quar- terly Environics and Decima public opinion polls conducted between 1980 and 1996. These asked

people to rate the most important problem facing the country. The percentages of respondents citing inflation and unemployment are shown.

Several features stand out. First, the public's con- cern with inflation was high in the early 1980s when inflation was high, and it was higher than concerns about unemployment. It fell sharply during 1982 as inflation declined and unemployment rose.

Second, concerns about unemployment have ex- ceeded concerns about inflation (usually by a wide margin) since 1982. There is no evidence of a de- mand by the public for a disinflationary recession in the late 1980s. When the zero-inflation policy was

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 10: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

Real Income, Unemployment and Subjective Well-Being 407

FIGURE 2 What Were the Top Priorities of the Public?

60

50

40

xX 30

xx > > xx xXXX

10XX 20 "

..xx x" 1

xxxxx xx xx Xx x xx xx xxx xxxxx X XX XX

X<X•XX

1980 1982 1984 1986 1988 1990 1992 1994 1996

x Inflation > Unemployment

Source: Decima Quarterly (1980-94); Environics (1995-96).

announced in February 1988, only 2.3 percent of the public rated inflation as the nation's top priority. l0

NET BENEFITS FROM DISINFLATION RECONSIDERED

Three recent studies of subjective well-being in the United States and various European countries by Blanchflower and Oswald (2000) and Di Tella, MacCulloch and Oswald (2000 and forthcoming) allow the conventional calculation of the net ben- efits of disinflation to be redone in units of subjective well-being using parameters estimated in these studies. This assumes that the well-being of

Canadians, Americans, and Europeans are deter- mined by similar factors.

We can ask whether there would be a positive net gain in average well-being as a result of the disinflation. Because of distributional issues, a posi- tive net gain does not mean that disinflation is necessarily beneficial, nor does it mean that the ac- tual disinflation gave the greatest net benefits possible. But as in the conventional cost-benefit calculation the result is still a useful indicator.

To do this, we can compare the loss in well-being from higher unemployment and lower real incomes with the gains from higher real incomes in the fu- ture, as in the conventional cost-benefit calculations

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 11: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

408 Roderick Hill

of Table 1. Changes in average subjective well-being can be discounted and added up to obtain a net ben- efit, paralleling the treatment of flows of income in the economic welfare calculation."l

We could also ask how long it would take to re- gain the loss in average well-being due to the disinflation. In either case, we might take into account population growth (assumed here to be 1 percent) because one could argue that it matters if more people enjoy the long-term gains.

I begin with Blanchflower and Oswald's (2000) study of American subjective well-being using 1972- 98 data from the US General Social Survey. About 33,000 individuals were asked "Taken all together, how would you say things are these days - would you say that you are very happy, pretty happy, or not too happy?" Blanchflower and Oswald assign weights of 3, 2, and 1 to these possible responses and do a least-squares regression that controls for personal characteristics (including whether the re- spondent was unemployed) and per-person household income. They do not consider macroeco- nomic variables, such as the unemployment rate or the inflation rate.

They find that unemployment has a strong nega- tive relationship with well-being, while household per capita income has a positive relationship.12 Using the same assumptions as with Table 1, I simu- late the total decline in average subjective well-being during the disinflation due to higher unemployment and lower per capita incomes. This can be compared with the long-term gains from higher future real in- comes that lower inflation is assumed to bring about.'3 The results for both the high- and low-cost scenarios from Table 1 are shown in the first line of Table 2.14

These results suggest that taking account of the non-pecuniary costs of unemployment and the mod- est contribution of real income to well-being makes a real difference. In contrast to the large net gains of Table 1, it seems that net losses in well-being are

possible. In the high-cost case, the net benefit of -0.04 can be interpreted as equivalent to 4 percent of the population having their well-being lowered one increment (from "pretty happy" to "not too happy," for example).

About 40 percent of the costs of the recession consist of the direct well-being costs to the unem- ployed, the remainder being due to lower real incomes during the recession. These calculations omit the negative effects that a person's unemploy- ment has on others in their immediate family, as well as the possible anxiety that high unemployment might cause through the workforce generally. Simi- larly, they do not consider the direct effects of inflation on well-being, an important point given the weak empirical evidence of a negative relationship between inflation and output.

The point of Di Tella et al.'s (forthcoming) study is to address these very issues. They use Eurobarometer survey data on life satisfaction for 264,710 individuals in 12 European countries be- tween 1975 and 1991. People were asked "On the whole, are you very satisfied, fairly satisfied, not very satisfied, or not at all satisfied with the life you lead?"

After attaching numerical values from 4 ("very satisfied") to 1 ("not at all satisfied") to these pos- sible responses, they run least-squares regressions on the individual data. As expected, individuals' employment status is strongly associated with well- being, as is their place in the income distribution.5

They average the residuals from this regression for everyone living in a particular country in a par- ticular year. Then they do a second regression using these averages to see if inflation and unemployment are associated with average happiness. They find that both have a consistent negative association with well-being.16 Per capita real income was not in- cluded in this study. This means that neither the short-run cost of lower real incomes nor the possi- ble long-run gains in real income from lower inflation play any role in the story here.

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 12: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

" z 3?

0 C- z

Cu

0 ,-]

0

r1

c<

0

z 0

o,

TABLE 2 Evaluation of Canadian Disinflation Using Estimated Subjective Well-Being

Source of Estimates of Determinants of Recession (High or Years to Recover Costs of Discounted Costs and Benefits in Units of Well-Being Subjective Well-Being Low Cost Estimate) Recession, No Discounting (See Scale Used)

Costs Benefits Net Benefits

Blanchflower and Oswald, happiness in the High 45 -0.07 0.03 -0.04 United States (1-3 scale): unemployment and real income Low 25 -0.02 0.03 0.02

Di Tella, MacCulloch and Oswald (forthcoming): High 1 -0.07 1.33 1.26 1975-91 life satisfaction in Europe (1-4 scale): inflation and unemployment Low 1 -0.01 1.53 1.52

Di Tella, MacCulloch and Oswald (2000), 1975-91 High 15 -0.64 1.38 0.74 life satisfaction in Europe (1-4 scale): inflation, unemployment and real income Low 5 -0.16 1.53 1.37

Appendix Table 1, eq.l. 1946-98 Happiness in High 20 -1.29 1.71 0.42 Canada (0-10 scale): inflation, unemployment and real income Low 6 -0.32 1.97 1.66

Appendix Table 1, eq.4. 1946-98 Happiness in High 41 -1.19 0.66 -0.53 Canada (0-10 scale): inflation, unemployment and real income Low 22 -0.51 0.76 0.25

Note: The calculations assume real per capita growth of 1 percent, population growth of 1 percent, and a discount rate of 4 percent. Units in the last 3 columns are discounted average per person changes in subjective well-being in the units of the respective study, all values rounded to two decimal places. Source: Author's calculations.

0h

0

0-

0r

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 13: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

410 Roderick Hill

In this approach, the well-being costs of the re- cession are the costs of unemployment to the unemployed and to the general population. Weighed against this are the benefits to everyone of lower inflation both during and after the recession. Be- cause the estimated direct benefits from lower inflation are large, the costs of the recession would be recouped almost immediately, as shown in the second row of Table 2. The estimated net benefits of disinflation are positive and substantial.

In a separate study, Di Tella, MacCulloch and Oswald (2000, Table 6) report the results when real per capita income is included with inflation and unemployment in the regression.17 As seen in the third row of Table 2, the weight given to real in- come raises the cost of the recession substantially compared with the previous case.

Whether the relative weights given to inflation and unemployment that have been estimated in these pioneering studies will remain about the same after further research or prove to be broadly applicable across countries remains to be seen. However, a

comparison of the first three rows of Table 2 makes it clear that the case for disinflation rests primarily on considering the well-being costs of inflation, and not on its long-run effects on real incomes. There is evidence that inflation has such costs.

Finally, we can briefly consider whether the Ca- nadian data on subjective well-being suggest a

picture that is broadly consistent with the results of these studies for the United States and Europe. I have located data from Canadian happiness surveys for 19 years between 1946 and 1998. Unfortunately, the questions used have not been comparable, making it difficult to assign consistent numerical values to responses. (A summary of the data and a descrip- tion of my treatment of it are available on request.)

The results of some simple regressions with the Canadian data are reported in Table A 1. Across all regressions, per capita real income and consump- tion have a positive relationship with average

well-being, while inflation and unemployment have a negative relationship. According to equation 1, a percentage point of unemployment has more than four times the cost of a percentage point of infla- tion. Despite this, as reported in row 4 of Table 2, this equation suggests that positive net benefits are likely from disinflation if all individuals' changes in well- being are discounted and added up. These results are also similar to the Di Tella et al. results in row 3.

Figure 2 raises the possibility that the well-being cost of an additional point of inflation or unemploy- ment is not constant but might rise, perhaps after some threshold level has been reached. I re-estimated the data using non-linear least squares, as shown in equa- tion 3 of the Appendix, but the estimates proved unreliable because of the large standard errors that re- sult from this technique using a small sample. Instead, I transformed the inflation and unemployment data using exponents found from a simple grid search that maximized R2.18 Equation 4 reports the OLS estimate; the last line in Table 2 summarizes the results for sub-

jective well-being using these estimates. Compared with equation 1, the decreased weight given to real income and the greater importance of unemployment relative to inflation leads to much lower estimated net benefits (and the possibility of net losses) in this case.

However, given the poor quality of the Canadian data, these results should be regarded as little more than an illustration of this general approach. Simi- larly, Table 2 should be understood as a simple illustration of the major differences between con- ventional calculations of economic welfare and

attempts to consider well-being more directly.

CONCLUDING REMARKS

The debate about the costs and benefits of disinflation in Canada has not considered the evidence on subjec- tive well-being. Diener and Suh write that

when policy makers seek to improve the quality of life, it is not sufficient to measure only the

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 14: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

Real Income, Unemployment and Subjective Well-Being 411

economy ... If only objective indicators are col- lected, valuable information is lost about how people evaluate and weight the conditions of their lives ... Although subjective indicators should not substitute for measures of external conditions, they serve as a useful complement in assessing and improving the quality of life of societies (1999, pp. 38-40).

But "subjective variables (which measure what people say rather than what they do) have usually been treated with suspicion by economists," as Winkelmann and Winkelmann put it (1998, p. 3). However, in assessing well-being it may be the best information available, relying as it does on individuals' own judgements.

The evidence from the subjective well-being lit- erature implies that: (i) the gains from higher real output in the future, even if it sums to a large present discounted value, may have only a relatively mod- est effect on welfare if and when it is actually experienced; (ii) we should take much more seri- ously the welfare-reducing effects of unemployment, both for the unemployed themselves and for the general population; and (iii) the benefits of lower inflation may operate more through direct effects on welfare than indirectly through the effects of in- flation on real incomes. My goal here is to raise awareness of these issues, as I feel they have been neglected in the policy debate and to encourage the more frequent use of information on subjective well- being in evaluating public policy. As the negative net benefits suggested in some parts of Table 2 show, considering these factors may reverse the conclu- sions of "economic welfare" calculations.

Finally, the general issues raised here have a broader application to other aspects of public policy. Gains from freer trade have traditionally been ex- pressed in the same framework: higher real incomes forever in exchange for lower real incomes and higher unemployment in the short term.

Decisions about public policy often involve high- stakes gambles to trade off short-term pain for

(hopefully) long-term gain. The intriguing and grow- ing evidence about what determines subjective well-being suggests that a more careful assessment of both the gains and losses is in order.

NOTES

I am grateful to Charles Beach, Norm Cameron, Peter Howitt, Chris McKenna and four anonymous referees for valuable comments. I thank Rameshwar Gupta for sev- eral helpful discussions and special thanks to Elizabeth Hamilton who obtained much of the Canadian survey data. Bob Burge of the Centre for the Study of Democracy at Queen's University kindly permitted me access to the Centre's public opinion archives.

'See Bank of Canada Deputy-Governor Charles Freed- man (1991). Academics, such as Peter Howitt, David Laidler, and Douglas Purvis, endorsed some disinflation, while not necessarily supporting a goal of price stability. See their essays in Lipsey (1990).

2Howitt himself did not propose this as a "rule" to decide if zero inflation was an appropriate goal. See Howitt (1990, pp. 106-07).

3Only after 1999 did unemployment return to its pre- recession level. Assuming a 2 percent growth rate of potential GDP and a 4 percent discount rate, the cost as a share of GDP, summed over 1990-99 is: C/GDP= Z( 1.02/

1.04)'(ut+i-0.075). See Fortin (1996) on using a 7.5 per-

cent natural rate of unemployment.

4Happiness tends to reflect mood, while satisfaction requires a cognitive judgement about one's circumstances (Diener, Suh, Lucas and Smith 1999, p. 227). In practice, the two are highly correlated.

5A comprehensive survey is Diener, Suh, Lucas and Smith (1999).

6For example, expanded potential consumption could be used for more holidays, more time with family and friends, less commuting, or better public health. Unlike most material goods, these offer measurable, lasting ben- efits to which we do not become habituated.

7The same picture emerges if we control for other vari- ables that have an important effect on happiness. For example, if only married persons with household incomes

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 15: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

412 Roderick Hill

between $30,000 and $40,000 are considered, 82 percent of those working consider themselves happy and 14.8 percent "somewhat happy." Of those looking for work, 61.5 percent are happy and 34.6 percent "somewhat happy."

8Blanchflower and Oswald (2000, p.13) find that $60,000 would be required to compensate the typical per- son for being unemployed.

9If labour receives 60 percent of national income and if the unemployed all receive unemployment benefits of 60 percent of lost wages, the share of lost production borne by the unemployed would be about 0.6x0.4, or 0.24. If non-pecuniary costs are 15 times greater than pecuni- ary costs, the cost of 12 percent of GDP in row 2 of Table

1 would actually be 55 percent of GDP, that is,

12x(l+(0.24x15)). This large value reflects as much the relative unimportance of income in influencing the level of people's satisfaction as it does the importance of un- employment.

l'oEnvironics Focus Canada 882, question 1. Values for inflation and unemployment were lower in 1988 due to the free trade issue.

"Discounting flows of well-being, as opposed to flows of income or output, remains controversial, but as it is a common practice, I consider it here.

'2The coefficient for the dummy variable "unem-

ployed" is -0.2444 (t=12) and that for per capita household income is 0.409E-6 (t=12).

'3The effect of the change in unemployment on aver- age happiness is the participation rate (set to 0.67, its 1990 value) times the change in unemployment due to disinflation (explained in discussion of Table 1) times -0.2444, the coefficient on the "unemployed" dummy. Changes in income are converted to US dollars by a $C1.25/$US purchasing power parity exchange rate (CANSIM series D21178) and multiplied by the income coefficient, 0.409E-6.

'40nly results with population growth taken into ac- count are presented here. Other results and details of all calculations are available in an Appendix available from the author on request.

'5SThe estimated coefficient on the unemployment dummy is -0.33 (t=47), about the same as the well-being cost of moving from the fourth quartile in the income dis- tribution to the first quartile.

16The estimated coefficient for the unemployment rate is -2 (t=3.3). If unemployment increases one percentage point, well-being falls by -2x0.01, or -0.02. The coeffi- cient on the inflation rate is -1.4 (t=3.5).

'7The coefficients are: unemployment -1.739 (t=-3.4); inflation -1.021 (t=-3.2); per capita GDP in thousands of 1985 dollars 0.06 (t=2.4). Canadian dollar values are ad-

justed to 1992 US dollars using a purchasing power parity exchange rate of $C1.25/$US and a price deflator of 0.8525 to adjust to 1985 US dollars.

'8Only integer exponents between one and ten were examined.

REFERENCES

Abramovitz, M. 1959. "The Welfare Interpretation of Secular Trends in National Income and Product," in The Allocation of Economic Resources, ed. M. Abramovitz et al. Stanford: Stanford University Press.

Barro, R. 1996. "Inflation and Growth," Federal Reserve Bank of St. Louis Review 78(3):153-69.

Blanchflower, D. and A. Oswald. 2000. "Well-Being over Time in Britain and the USA," NBER Working Paper No. 7487. Cambridge, MA: National Bureau of Eco- nomic Research.

Coletti, D. and B. O'Reilly. 1998. "Lower Inflation: Ben- efits and Costs," Bank of Canada Review (Autumn):3-21.

Debelle, G. 1996. "The End of Three Small Inflations: Australia, New Zealand and Canada," Canadian Pub- lic Policy/Analyse de Politiques 22(1):56-78.

Diener, E. and E. Suh. 1999. "National Differences in

Subjective Well-Being," to appear in Hedonic Psychol- ogy: Scientific Perspectives on Enjoyment, Suffering and Well-Being, ed. D. Kahneman, E. Diener and N. Schwarz. New York: Russell Sage Foundation.

Diener, E., E.M. Suh, R.E. Lucas and H.L. Smith. 1999.

"Subjective Well-Being: Three Decades of Progress," Psychological Bulletin 125(2):276-302.

Di Tella, R., R.J. MacCulloch and A.J. Oswald. 2000. "The Macroeconomics of Happiness," unpublished paper.

SForthcoming. "Preferences over Inflation and Unemployment: Evidence from Surveys of Happi- ness," American Economic Review.

Easterlin, R.A. 1974. "Does Economic Growth Improve the Human Lot? Some Empirical Evidence," in Na- tions and Households in Economic Growth: Essays in

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 16: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

Real Income, Unemployment and Subjective Well-Being 413

Honor ofMoses Abramovitz, ed. P.A. David and M.W. Reder. New York: Academic Press.

1995. "Will Raising the Incomes of All Increase

the Happiness of All?" Journal of Economic Behavior and Organization 27(1):35-48.

_ 1996. Growth Triumphant: The Twenty-First Cen-

tury in Historical Perspective. Ann Arbor: University of Michigan Press.

Feldstein, M. 1997. "The Costs and Benefits of Going from Low Inflation to Price Stability," in Reducing Inflation: Motivation and Strategy, ed. C.D. Romer and D.H. Romer. Chicago: University of Chicago Press, pp. 123-56.

Fisher, I. 1928. The Money Illusion. Toronto: Longman's. Fortin, P. 1996. "The Great Canadian Slump," Canadian

Journal of Economics 29(4):761-87. 1997. "A Comment," in Where We Go from Here,

ed. Laidler.

1999. "The Great Canadian Slump: A Rejoinder to Freedman and Macklem," Canadian Journal of Eco- nomics 32(4):1082-92.

Frank, R. 1999. Luxury Fever: Why Money Fails to Sat- isfy in an Era of Excess. New York: The Free Press.

Freedman, C. 1991. "The Goal of Price Stability: The Debate in Canada," Journal of Money, Credit and Banking 23(3, Part 2):613-18.

Freedman, C. and T. Macklem. 1998. "A Comment on 'The Great Canadian Slump,'" Canadian Journal of Economics 31(3):646-65.

Howitt, P. 1990. "Zero Inflation as a Long Term Target," in Zero Inflation: The Goal of Price Stability, ed. R.G. Lipsey. Toronto: C.D. Howe Institute.

1997. "Low Inflation and the Canadian Economy," in Where We Go from Here, ed. Laidler.

Laidler, D., 1997. Where We Go from Here: Inflation Tar- gets and Canada's Monetary Policy Regime. Toronto: C.D. Howe Institute.

Lipsey, R.G., ed. 1990. Zero hIlflation: The Goal of Price Stability. Toronto: C.D. Howe Institute.

Oswald, A.J. 1997. "Happiness and Economic Perform- ance," Economic Journal 107(445):1815-31.

Pigou, A.C. 1951. "Some Aspects of Welfare Econom- ics," American Economic Review 41(3):287-302.

Ragan, C. 1998. "On the Believable Benefits of Low Infla- tion," Bank of Canada Working Paper No. 98-15. Ottawa.

Shiller, R.J. 1997. "Why Do People Dislike Inflation?" in Reducing Inflation: Motivation and Strategy, ed.

C.D. Romer and D.H. Romer. Chicago: University of Chicago Press.

Simpson, W., N. Cameron and D. Hum. 1998. "Is

Hypoinflation Good Policy?" Canadian Public Policy/ Analyse de Politiques 24(3):291-308.

Thornton, D.L. 1996. "The Costs and Benefits of Price

Stability: An Assessment of Howitt's Rule," Federal Reserve Bank of St. Louis Review 78(2):23-38.

Winkelmann, L. and R. Winkelmann. 1998. "Why Are the Unemployed So Unhappy? Evidence from Panel Data,"

Economnica 65(257):1-15.

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions

Page 17: Real Income, Unemployment and Subjective Well-Being: Revisiting the Costs and Benefits of Inflation Reduction in Canada

414 Roderick Hill

APPENDIX

TABLE Al Regression Results, Canadian Data, Various Years, 1946-1998 Happiness, Real Per Capita Income, Unemployment and Inflation

Equation Number

Variable 1 2 3 4

Constant 0.21 0.36 2.46 2.50 (0.1) (0.2) (1.4) (1.8)

Ln real per capita 0.80 0.53 0.52 income (1992$) (3.3) (2.8) (3.7)

Ln real per capita 0.83 consumption (1992$) (3.5)

Unemployment rate -6.85 -7.01 -6.16E6 -1.24E7 Persons age 15+ (-2.2) (-2.3) (-0.1) (-3.1)

Exponent, Unemployment 1 1 7.67 8 rate (1.2)

Inflation -1.55 -0.8 -202.4 -151.8 (-1.1) (-0.6) (-0.1) (-1.9)

Exponent, Inflation rate 1 1 3.13 3 (0.8)

Descriptive Statistics Adjusted R2 0.32 0.35 0.40 0.48

Notes: Mean of dependent variable: 7.55 on a 0-10 scale. Number of observations: 19. Inflation and unemployment rates are measured as fractions. Equations 1, 2 and 4 are estimated by ordinary least-squares, and Equation 3 by non-linear least-squares, where TSP uses the Gauss-Newton method to compute the standard errors from the quadratic form of the analytic first derivatives of the model. t-statistics are in parentheses; for a one-tailed test with 15 degrees of freedom, the critical values of the cumulative t-distribution are: 0.69 (25 percent); 1.34 (10 percent); 1.75 (5 percent) and 2.13 (2.5 percent).

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXVI, NO. 4 2000

This content downloaded from 195.78.108.140 on Wed, 11 Jun 2014 03:30:54 AMAll use subject to JSTOR Terms and Conditions