The evolutionary vision in which history matters is of an evolving economy driven by bursts of technological change initiated by agents facing uncertainty and producing long term, path-dependent growth and shorter-term, non-random investment cycles. The alternative vision in which history does not matter is of a stationary, ergodic process driven by rational agents facing risk and producing stable trend growth and shorter term cycles caused by random disturbances. We use Carlaw and Lipsey’s simulation model of non-stationary, sustained growth driven by endogenous, path-dependent technological change under uncertainty to generate artificial macro data. We match these data to the New Classical stylized growth facts. The raw simulation data pass standard tests for trend and difference stationarity, exhibiting unit roots and cointegrating processes of order one. Thus, contrary to current belief, these tests do not establish that the real data are generated by a stationary process. Real data are then used to estimate time-varying NAIRU’s for six OECD countries. The estimates are shown to be highly sensitive to the time period over which they are made. They also fail to show any relation between the unemployment gap, actual unemployment minus estimated NAIRU and the acceleration of inflation. Thus there is no tendency for inflation to behave as required by the New Keynesian and earlier New Classical theory. We conclude by rejecting the existence of a well-defined a short-run, negatively sloped Philips curve, a NAIRU, a unique general equilibrium, short and long-run, a vertical long-run Phillips curve, and the long-run neutrality of money.
This is a preview of subscription content, log in to check access.
Buy single article
Instant access to the full article PDF.
Price includes VAT for USA
Subscribe to journal
Immediate online access to all issues from 2019. Subscription will auto renew annually.
This is the net price. Taxes to be calculated in checkout.
The use of the terms Darwinian and Newtonian here is meant to highlight the significant difference in equilibrium concept employed in the two groups of theories that we contrast, the evolutionary and what we call equilibrium with deviations (EWD) theories. Not all evolutionary theories, including the one employed here, are strictly speaking Darwinian in the sense that they embody replication and selection. We use the term, Darwinian to highlight the critical equilibrium concept of a path dependent, non-ergodic, historical process employed in Darwinian and evolutionary theories and to draw the contrast between that and the negative feedback, usually unique, ergodic equilibrium concept employed in Newtonian and EWD theories.
U * must be a NAIRU for reasons given in the text. However, in a model in which markets are allowed to be temporarily out of equilibrium, there may be another level of U that is a temporary NAIRU because of asymmetries in the speed of upward and downward adjustment to excess demands and excess supplies. See Tobin (1998).
See Lipsey et al. (2005: 77–82) for a discussion of the relevance of path dependence and a reply to those who doubt its importance.
Most evolutionary economists accept that for many issues in micro economics, comparative static equilibrium models are useful. Also, there is nothing incompatible between the evolutionary world view and the use of Keynesian models – of which IS-LM closed by an expectations-augmented Phillips curve is the prototype – to study such short run phenomenon as stagflation and the impact effects of monetary and fiscal policy shocks. Problems arise, however, when such analyses are applied to situations in which technology is changing endogenously over time periods that are relevant to the issues being studied. Depending on the issue at hand, this might be as short as a few months.
We allow the critical value of the arrival parameter λ * in Carlaw and Lipsey (2011) to be a decreasing function of the accumulated amount of resources devoted to pure and applied R&D
See, for example, Nathan Rosenberg (1982: Chapter 7).
For simplicity in the simulations reported below we let X = Y = I = 3.
When we came to calculate an equivalent to labour in our model, we were forced to make some simplifying assumptions. First, we assumed that R is a composite of land and raw labour and that each unit of land is uniformly endowed to each unit of labour. Second, we assumed that labour will take out some of the value of its marginal product in consumption and some in reproduction that will expand the labour supply. For simplicity, we assumed a 50:50 split.
The data used for these calculations are from the Canadian Socio-economic Information and Management System Database (CANSIM).
The simulated data are more volatile than the Canadian data and the usual RBC simulation models. Much of the additional volatility in our simulation comes from the arrivals of the major new technologies.
We use the Eviews defaults of 1 through 4.
This should not be surprising since the Class 2 data showed stationarity in the unit root test of the levels for each individual time series when run with no intercept and trend. So the cointegration test should show all series as being stationary. This is strictly speaking a slight abuse of the cointegration test because it is only valid for I (1) or higher orders of integration processes.
The coefficient on the trend for the ADF test (with intercept and trend) on the log difference of output in Class 1 is −2.02e − 05 with a t statistic of −5.742228.
Both Class 1 and Class 2 output series exhibit a very small negative trend. This is likely due to the large initial growth rates that occur because of how the simulation is initially seeded with values.
Further analysis to choose between these two interpretations will entail generating a number of simulated data sets from a model that is explicitly non-stationary to see under what conditions time series analysis will detect its non-stationary properties. For example, one stylised fact that emerges out of the historical analysis of general purpose technologies and economic growth is that sometimes the early stages of technologies that become transforming GPTs cause structural disruptions to the economy that lead to economic slowdowns for a period while they gestate and mature. This can be modelled explicitly within the Carlaw and Lipsey (2011) framework and can provide another source of non-stationarity (in terms of first differences) in the simulated data. Further analysis will reveal if the time series econometric techniques will detect these sources of non-stationarity in the data. Until that time, we conclude that existing tests do not support the conclusion that the real data have been generated by stationary processes in which the details of history do not matter.
The voluminous empirical work concerning the Phillips curve and the NAIRU is briefly discussed in the last section of this paper.
The data are for the standardised unemployment rates and consumer prices provided by the OECD at http://oecd-stats.ingenta.com and accessed 1 August 2010. They begin at different years: France 1977, Italy 1978, Spain 1977, the UK 1970, and the US 1960. We use all the data available from that source since inter-country comparisons are not of major importance to our study.
The data used in the following estimations can be obtained in an excel spreadsheet form from either author, email: email@example.com or firstname.lastname@example.org.
If each absolute gap is associated with the same U *, the two measures will be perfectly correlated along a curved line. If some absolute gap’s are associated with different U *s, there will be a scatter of these relative gap values around their associated absolute gap values.
The surprisingly low figure where the filter estimation starts in 1980 illustrates how sensitive U * estimates are to the historical period over which they are made.
There is a possible problem in conducting this test since U = U * is predicted to be consistent with any stable inflation rate. For this to be a problem in practice we would have to have two or more successive years in which U stayed approximately equal to U * (say U = ± 0.5U * while the inflation rate stayed approximately constant over the period. However, such a situation has not arisen in any of our data.
The estimated d coefficient values, this time expected to be negative, were for the short and long periods respectively, France: 0.046 (0.115), 0.024 (0.056); Italy: −0.136 (0.105), −0.064 (0.134); Spain: −0.090 (0.031), 0.078 (0.065); UK: −0.116 (0.239), −0.079 (0.118); USA: −0.746 (0.176), −0.097 (0.120).
Italy is omitted because the Kalman filter estimate of its β coefficient over the shorter period is almost zero and completely insignificant statistically. Thus massive variations in U *3 are required to create a sufficiently large unemployment gap to explain the observed variations in the acceleration of inflation. To check Italy, we estimated its coefficient e in Eq. 8 by the alternative method of fitting that equation to the data for U and π. We then calculated its U *3 for each period and found it to be not dissimilar from those for the other countries, but still more variable with a ratio of the variance of U *3 to U of 84.57.
The NAIRU is not a merely part of what Imré Lakatos called a theory’s protective belt. Instead it is part of the core of all EWD theories. Without it, the whole concept of a unique equilibrium for the economy, departure from which sets up equilibrating forces which can only be frustrated by agents making repeated errors, fails.
The material in the bullet points that follow in the text are paraphrases of material in Lipsey and Scarth (2011, xxxii-xxiii).These authors give an extensive survey of the Phillips curve and NAIRU literature from the earlier times until the early 21st century.
At U *, wages will be constant in a static model, or changing at the same rate as productivity is changing in a growth model. In either case, this results in the absence of any inflationary pressure emanating from the labour market.
Robert Gordon’s triangle model is another approach that also does the same job.
This lack of uniqueness is reinforced by two important characteristics. First, many firms (probably most) have short run cost curves that are flat, allowing a wide range of output fluctuations over the short run with little or no changes in product prices. Second, at some times, such as the last two decades, the nature of technological change creates a great deal of uncertainty in the labour market that puts strong pressure on labour to be fairly docile, not pushing aggressively for higher wages at the first sign of an economic expansion or even the onset of an output boom. See Lipsey (2010) for a full discussion of the importance of these two characteristics.
Arthur B (1994) Increasing returns and path-dependence in the economy. University of Michigan Press, Ann Arbor, Mich
Carlaw KI, Lipsey RG (2006) GPT-driven, endogenous growth. Econ J 116:155–174
Carlaw KI, Lipsey RG (2011) Sustained endogenous growth driven by structured and evolving technologies. J Evol Econ 21(4):563–593
Fortin P (2001) Interest rates, unemployment and inflation: the Canadian experience in the 1990s. In: Banting K, Sharpe A, St-Hilaire F (eds) The review of economic performance and social progress: the longest decade - Canada in the 1990s, vol 1. Centre for the Study of Living Standards & The Institute for Research on Public Policy, Ottawa, pp 113–130
Freeman C, Louçã F (2001) As time goes by: from the industrial revolutions to the information revolution. Oxford University Press, Oxford
Gali J, Gertler M (1999) Inflation dynamics: a structural econometric approach. J Monet Econ 44(2):195–222
Hall RE (1980) Employment fluctuations and wage rigidity. Brookings Economic Papers 10th Anniversary Issue, p 91–123
Hornstein A (2007) Evolving inflation dynamics and the New Keynesian Phillips curve. Fed Reserv Bank of Richmond Econ Q 93(4):317–339
Jacob MC (1997) Scientific culture and the making of the industrial west. Oxford University Press, Oxford
Jones E (1988) Growth recurring: economic change in world history. Oxford University Press, Oxford
Landes D (1969) The unbound prometheus: technological change and industrial development. Cambridge University Press, London
Landes D (1998) The wealth and poverty of nations. W.W. Norton, New York
Lipsey RG (2000) IS-LM, Keynesianism and the New Classicism. In: Blackhouse RE, Salanti A (eds) Macroeconomics and the real world, vol 2: Keynesian economics, unemployment, and policy. Oxford University Press, Oxford, pp 57–82
Lipsey RG (2010) Schumpeter for our century. In: Hanusch H, Kurz HD, Seidl C (eds) Homo Oeconomicus, Muchen Special Issue 27(1/2):145–176
Lipsey RG, Scarth W (2011) The history, significance and policy context of the Phillips Curve. In: Lipsey RG, Scarth W (eds) Inflation and unemployment: the evolution of the Phillips Curve. Edward Elgar, Cheltenham
Lipsey RG, Carlaw KI, Bekar C (2005) Economic transformations: general purpose technologies and long-run economic growth. Oxford University Press, Oxford, UK
Lucas RE Jr., Rapping LA (1972) Unemployment in the great depression: is there a full explanation? J Polit Econ 80(1):186–191
Marx K (1957) Capital: a critique of political economy (3 vols.), translated from the 3rd German Edition by Moore S, Aveling E, Engles F (eds) Foreign Language Publishing House, Moscow
MacKinnon JG, Haug A, Michelis L (1999) Numerical distribution functions of likelihood ratio tests for cointegration. J Applied Econometrics 14(5):563–577
Mokyr J (1990) The lever of riches: technological creativity and economic progress. Oxford University Press, New York
Mokyr J (2002) The gifts of Athena: historical origins of the knowledge economy. Princeton University Press, Princeton, NJ
Musson AE, Robinson E (1989) Science and technology in the industrial revolution. Gordon and Breach, USA
Nason JM, Smith GW (2008) The new Keynesian Phillips Curve: lessons from single-equation econometric estimation. Fed Reserv Bank of Richmond Q Rev 94:361–395
Nelson C, Plosser C (1982) Trends and random walks in macroeconomic time series: some evidence and implications. J Monet Econ 10:139–169
Nelson R, Winter S (1982) An evolutionary theory of economic change. Harvard University Press, Cambridge
North DC (1981) Structure and change in economic history. Norton, New York
Pomeranz K (2000) The great divergence: China, Europe and the making of the modern world economy. Princeton University Press, Princeton
Rae J (1905) The sociological theory of capital. Macmillan, New York, first published in 1834 as Statement of Some New Principles on the Subject of Political Economy Exposing the Fallacies of the System of Free trade and of Some Other Doctrines Maintained in the Wealth of Nations.
Rosenberg N (1982) Inside the black box: technology and economics. Cambridge University Press, Cambridge
Schumpeter J (1934) The theory of economic development. Harvard University Press, Cambridge Mass
Tobin J (1998) Supply constraints on employment and output: NAIRU versus Natural Rate, paper presented at the International Conference in memory of Fausto Vicarelli, (Rome), November 2–23
Veblen T (1953) The theory of the leisure class (revised ed.). New American Library, New York
This is a revised version of a paper Presented at the 44th Annual Conference of the CEA May 28–30, 2010, Quebec City and the 13th ISS Conference 2010 in Aalborg, Denmark June 21–24, 2010. We are indebted for funding support to the Economic and Social Sciences and Humanities Research Council of Canada and to the Centre for International Governance and Innovation – Institute for New Economic Thinking grant number 2881. We are also grateful to anonymous referees, Steven Kosempel, Bill Scarth and Les Oxley for comments on earlier drafts.
About this article
Cite this article
Carlaw, K.I., Lipsey, R.G. Does history matter?: Empirical analysis of evolutionary versus stationary equilibrium views of the economy. J Evol Econ 22, 735–766 (2012). https://doi.org/10.1007/s00191-012-0282-4
- Evolutionary economics
- New classical economics
- Real business cycles
- Time series data stationarity