INGENUE v.2:

1
INGENUE v.2

• a World OLG-CGE Model
• with Imperfect Financial Markets, Exchange Rates
• and Stochastic Lifetime

2
INGENUE Team

• INGENUE is a joined research project linked to
  three French centers CEPII, CEPREMAP and OFCE.
• The members of the team
• CEPII Michel Aglietta Vladimir Borgy
• CEPREMAP Michel Juillard
• OCDE Jean Château
• OFCE Jacques Le Cacheux,
• Gilles Le Garrec Vincent Touzé.

3
Aim of this research

• The aim of this research is to analyze the issues
  relating to wealth accumulation and the
  development of pension funds and other devices of
  saving for retirement in the context of global
  finance, hence to study the international capital
  flows that ought to be induced by the differences
  of aging and of technical progress growth in the
  various regions of the world.

4
Related Literature

• This model is the second version of a
  multi-region, world model, in the spirit of those
  theoretically described by Obstfeld Rogoff
  (1996), in which the structure of each regional
  economy is similar to that of other national
  Overlapping Generations Compu-tational General
  Equilibrium (OLG-CGE) models, such as Auerbach et
  al. (1983) for US, Cazes et al. (1992, 1994) for
  France, except that labor supply is generally
  endogenous.

5

• Two similar models
• FEHR, H., S. JOKISCH, L. KOTLIKOFF, (2004)
  ''The role of immigration in dealing with world's
  demographic transition'', NBER WP, n10512.
• and
• BOERSCH-SUPAN A., A. LUDWIG J. WINTER (2002)
  ''Ageing and International Capital Flows'', in
  Auerbach, A.J. and H. Herrmann (eds.), Ageing,
  Financial Markets and Monetary Policy, Springer,
  Heidelberg, 56-83.
• They have developed two applied multi-regional
  OLG-CGE models, but their analysis is restrained
  to developed countries.

6
Outline of the talk

• INGENUE v.2 What's new?
• (2) The world in 10 Regions
• (3) Demographic projections
• (4) Description of the regional economies
• (5) Dynamic general equilibrium
• (6) The empirics of INGENUE
• The baseline case
• Pension reforms vs Migratory policy

7
INGENUE v.2 What's new?

• In INGENUE v1, the world was divided into six
  regions (three developed areas and three
  developing areas), each of which is made of three
  categories of economic agents the households,
  the firms, and a PAYG retirement pension system.
  There was only one good, and only one financial
  asset, which is an ownership stake in the firms'
  productive capital both of them are freely
  traded on perfectly competitive world markets.

8

• There was no money and hence only two relative
  prices in each region the (real) wage rate
  accruing to local, internationally immobile,
  workers and the single (real) price of financial
  assets, both expressed in terms of goods, which
  may be chosen as numéraire. Hence, the various
  regions of the world were economically and
  financially perfectly integrated and there is
  only one world market for goods and one for
  financial assets. We have performed various
  pioneer works (2001, 2002a, 2002b and 2004 with
  IMF) with this first model but this model could
  appear to be limited in some of its outcomes.
• So we developed this new version.

9

• In this second version, we make the model more
  realistic by introducing a number of changes in
  the assumptions
• Demographics The World is now divided in 10
  regions. In order to make autonomous own
  demographic projections, we have built a
  population projection model based upon UN
  coefficient methods.
• Households We now assume uncertainty in lifetime
  expectancy at individual level. At the
  macroeconomic level there is still no
  uncertainties about it.

10

• International trade of commodities In order to
  deal with relative price movements of foreign and
  domestic goods we assume that the different
  countries produce, different imperfectly
  substitutable intermediate goods as in Backus et
  al. (1995). This will imply the existence of real
  effective exchange rates between the different
  regions. Here the main determinants of exchange
  rates are the relative productivity in the two
  productive sectors as in the standard view
  developed since Obstfeld and Rogoff works (i.e.
  the famous Balassa-Samuelson effect that is
  predominant in long run explanations of
  difference in real exchange rates).

11

• Financial markets We model region-specific
  interest rates to debtor that differ from the
  unique world interest rate to creditor by
  imposing an ad hoc convex function of the
  regional ownership ratio.
• Calibration improvements We introduce
  inheritances based upon a bequest motive,
  age-specific labor participation rates
  (exogenous), age-specific human capital
  (exogenous) and labor income of children in some
  parts of the world.

12
(2) The world in 10 Regions

• Z1 Western Europe with United Kingdom, Italy,
  Spain, France, Germany, etc.
• Z2 Eastern Europe with Bulgaria, Czech Republic,
  Hungary, Poland, etc.
• Z3 Northern America with Canada, United States
  of America, Australia, New Zealand, etc.
• Z4 Southern America with Argentina, Bolivia,
  Brazil, Chile, Colombia, Mexico, Nicaragua, etc.
• Z5 Japan
• Z6 Mediterranean World with Algeria, Egypt,
  Morocco, Tunisia, Western Sahara, Armenia, Iraq,
  Iran, etc.
• Z7 Chinese World with China, Democratic People's
  Republic of Korea, Mongolia, Republic of Korea,
  Philippines, Singapore, Thailand, etc.
• Z8 Africa with Burundi, Comoros, Djibouti,
  Eritrea, Ethiopia, Kenya, Madagascar, Senegal,
  etc.
• Z9 Russian World with Belarus, Russian
  Federation, Ukraine, etc.
• Z10 Indian World with India, Afghanistan,
  Indonesia, Malaysia, etc.

13
World Regions in Ingenue 2.
North America Western Europe Eastern
Europe Russian World South America Mediterranean
World Indian World Chinese World Africa Japan
14
(3) Demographic projections

• The period of the model is set to five years. In
  each region z, the economy is populated by 21
  overlapping generations of one-sex agent who may
  no live longer than 105 years.
• Population evolution are exogenously calculated
  according to a standard population projection
  method on the basis of historical and prospective
  UN data. We have aggregated population structure,
  with the UN data from 1950 to 1995, over
  countries to build Ingenue's regions. Then we can
  project fertility and mortality trends (for both
  sexes) at the region-aggregate level, this
  together with initial population structures in
  1995, allow us to obtain population evolution in
  the future from 2000 until the ending date of the
  model. We implicitly assume that there is no
  migration flows in the future. With some usual
  population projection methods, we construct
  evolution of mortality and fertility tables on
  the only basis of life expectancy and global
  fertility rates evolutions in the future.

15
Mortality People can die before 105 year let
sa denotes the conditional probability of
surviving between age a and age a1, the number
of age a-1 people then changes as
Fertility Process At each time period, the
number of births will be equal to
(here L is only female population), where fz(a)
are the average age-specific fertility rates.
Following UN projections, we implicitly assume
that women fertility occurs only between 15 and
50 years old.
16
Size of the World Population according to four
fertility rate convergence scenarios
17
(No Transcript)
18

• Western Europe and Japan, which are the regions
  with decreasing labour force from the present or
  the next decade onwards amongst the most
  developed regions.
• Eastern Europe, and the Russian world, ageing as
  fast the former from a much lower level of income
  per capita.
• North America (including also Australia and new
  Zealand) and the Chinese world (China and
  surrounding East Asian countries) whose labour
  force growth will turn slightly negative around
  2020.
• South America, the Mediterranean world (including
  non-European countries around the Mediterranean
  Sea and the Near East) and the larger Indian
  world (including Pakistan and South East Asia on
  top of India). These regions have already been
  involved in the process of ageing but they will
  keep a growing labour force until 2050.
• Africa is the region where the demographic
  transition has hardly started and whose working
  age population will still grow at 2 per annum in
  2050.

19
(4) Description of the regional economies

• Households
• Individuals are assumed to become adults when
  they turn 20. During any period, the household
  sector is then made of 17 overlapping cohorts of
  adults'', of age between 20 and 105, and 4
  cohorts of young''. Adults may not stay in the
  labor force after a legal maximal mandatory
  retirement age R. They determine their optimal
  designs of consumption and saving with rational
  foresight concerning their intertemporal income
  constraint after taxes wages, interest rates
  and retirement pensions.
• The length of the life is stochastic and there
  exist two types of insurance contract actuarial
  return saving contract and capital insurance for
  heirs in case of premature death.
• Between 15 and 50 yrs. adults are supposed to
  give birth to children, according to the
  fertility calendar. Children are dependent until
  they turn 20, they consume with a cost per child
  that is supposed to be proportional to the
  parents consumption.

20

• Labor supply is assumed to be exogenously given
  as the age-specific rate of participation to
  labor market. We use ILO data and projections to
  characterize activity from 1950 until 2015 and
  assume that after this date participation rates
  remain fixed at their 2015 level. According to
  this database people may work since the age of 10
  so we will take into account children labor
  income to the budget constraint of their parents.
• The intertemporal preferences are given by a
  life-time utility function over uncertain streams
  of consumption and leaving a voluntary insured
  bequest H to their children.

21

• The public sector
• The public sector is reduced to a social security
  department it is a pay-as-you-go (PAYG) public
  pension scheme, that is supposed to exist in all
  regions of the world. It is financed by a payroll
  tax (?) on all labor incomes and pays pensions to
  retired households. The regional PAYG systems
  operate according to a defined-benefit rule
  pensions paid to individual retired are a
  fraction - or replacement rate (?) -- of the
  current average (net of tax) wage. We assume a
  time-to-time balanced-budget rule.
• Notation La size of the population of age a.

22
Regional productive sectors

• Intermediate good sector
• Each zone z specializes in the production YI of a
  single intermediate good labeled z, where
  subscript z indicates that the specific nature of
  this good lies in its region of origin.
  Production takes place with a constant return to
  scale-Cobb Douglas production function using
  capital stock and the full domestic labor force.
• The production of intermediate good (YI) also
  denotes GDP in the country z in terms of the
  local intermediate good.

23

• Final good production sector
• In the spirit of Backus et al. (1995), we assume
  that the domestic, composite final good of region
  z (consumption and investment) YF is produced
  thanks to a combination of two intermediate goods
  a domestic'' intermediate good in proportion
  Bz and a World'' intermediate goods in
  proportion Mz , according to the following CES
  technology, where s denotes the elasticity of
  substitution

24

• The fiction of a world producer of an homogenous
  world intermediate good
• In order to simplify the exchanges of
  intermediate goods between regions of the world
  we assume that there exist a fictive world
  producer that uses region-specific intermediate
  goods in quantities Xz in order to produce a
  specific world intermediate good Y according to
  the following CES function

25
Imperfection on capital market

• To be more realistic the world asset capital
  market is supposed imperfect. Because sources of
  imperfection and asymmetries in financial markets
  are various and uneasy to model with rigorous
  micro-foundations in such a large scale model as
  INGENUE we adopt an ad hoc formulation where the
  rate of industrial failure in a country is linked
  to the rate of indebtedness. We introduce a sort
  of Stiglitz-Weiss constraint. Then, the average
  creditor interest rate for firms in a country
  depends on the national level of indebtedness
  with respect to the rest of the world.

26
(5) Dynamic General Equilibrium

• Households maximize their welfare
• Firms maximize their profit in the intermediate
  good sector, the final good sector and the world
  good
• Markets are cleared at each date.

27

• We compute intertemporal prices (real local wage,
  real local prices of goods, real local interest
  rate) to equilibrate
• Regional closed markets
• Labor market
• Final good market.
• Regional open markets
• Intermediate good markets
• Capital markets.
• The world market
• World good market.

28
(6) The empirics of INGENUE

• Calibration to reproduce some stilized facts
• TFP, profiles of saving and investment rates,
  real exchange rate, pension contribution rates,
  etc.
• Main sources of Data
• Penn World Tables (cf. Summers Heston)
• IMF
• OECD.

29
The Exogenous Catching up

• To project the evolution of the TFP, we use a
  exogenous law reflecting international diffusion
  of technological progress and we adopt different
  paces of convergence. The level of total factor
  productivity in the zone at the technological
  frontier (North America) is . It is supposed to
  grow at 1.5 per year with g1.5. The
  diffusion of technological progress to a zone z
  is given by the following equation

30

• The first bracket captures the speed-up in the
  rate of technological progress due to the
  shortening of diffusion in technological
  innovations. It means that ? is an accelerator to
  the convergence in the growth rates (the chosen
  value is 0.001) The second bracket embodies a
  brake due to the difficulties to create the
  social conditions proper to assure a speedy
  diffusion. ? is the brake factor to the
  convergence in level. There is a caveat however.
  To make account for the attractiveness of Europe
  upon its neighbour regions we assume that these
  regions will converge to the European target.

31
Fig. 4 - Total Factor Productivity 1960
2100 (percentage of US level)
32
Start of our simulation 1955. The departure
point of our numerical analysis 2000.
Calibration strategy the model must match the
historical data in 2000.

• To do this simply, and in order to avoid any
  initial jump in the equilibrium variables, we
  choose to start our simulation in year 1955,
  taking as given, first, the historical values for
  state variable in 1950, and demographic and
  participation rates historical trajectories and
  projections. We then choose parameters and
  exogenous trajectories for sectorial factor
  productivity in order to get values for the
  endogenous variables of the model near their
  observed values in 2000.

33

• To reproduce the selected stylized facts for each
  region considered, the time preference parameters
  are neither set constant nor similar between the
  different regions during the calibration period
  (i.e. between 1955 and 1995). After 2000, the
  value is only different for N. America.
• Furthermore, the regional parameters of the CES
  production functions are calibrated in order to
  reproduce the international trade shares for each
  regions.
• In the calibration process, the relative
  productivity between sectors are set in order to
  reproduce the exchange rates, and the human
  capital variables are calibrated in order to
  reproduce the GDP per capita levels.

34
(7) The Baseline Case a path for the world
economy in the first part of the XXIst Century

• General remarks
• The baseline scenario is the outcome of a long
  and weary process of calibration.
• To put the model on an acceptable track on the
  projection phase starting in 2000, an adjustment
  must be made. It is why the model computation
  shall begin at an initial date as far in the past
  as the data permit it. The initialisation is the
  five year period beginning in 1950 where initial
  stocks of capital, household assets and an age
  distribution of savings are estimated.
• Assumptions regarding technological convergence
  are conservative in the baseline scenario.
• The baseline scenario is not easy to depict
  since it is a dynamic rational expectations
  general equilibrium.

35
Fig. 5 - Regional annual interest rate

• The regional real interest rates are declining
  over the fifty year period. This is due to global
  ageing (the working age population is
  decelerating or declining absolutely while the
  age group of high savers is growing in one region
  after another) and to a declining catching up.
• The hierarchy of regional real interest rates is
  linked to the rate of change of the real exchange
  rates.

36
Fig. 6 Regional GDP (PPP) growth rate

• The GDP regional growth rates largely follow that
  of the regional labor force and TFP growth rates.

37
Fig. 8 - Evolution of the gross investment to
GDP ratio

• Because the model works at full employment with
  an exogenous labour force, the stock of capital
  (and the investment by difference) in each period
  is a rising function of employment adjusted for
  labour efficiency (age distribution) and of total
  factors productivity. It is a decreasing function
  of regional real interest rate.

38
Fig. 7 - Evolution of the Net Saving(in
percentage of GNP)

• Net saving in each region is the aggregate of
  individual savings in the life cycle. It depends
  on the demographic structure (high savers ratio
  and dependency ratio), on the expectation of
  future income and on the parameters of the PAYG
  pension systems.
• Regions with the fastest-increasing dependency
  ratios are the ones with the fastest-decreasing
  net saving rate (Japan, Western Europe, Eastern
  Europe, Russia).
• In China, India, South America and the
  Mediterranean, the high savers ratio and the
  dependency ratio rise in tandem. In the early
  decades, while the population is still young,
  those regions grow faster than more
  demographically mature ones. It follows that
  young people expecting higher future income,
  reducing the overall saving rate.

39
Fig. 10 Ownership ratio

• The ownership ratios are mainly determined by
  cumulative current account balances (by
  difference, around the slope of the curve).
• The most striking feature is the divergent
  profile of North America. It is due to an assumed
  change in household behaviour.
• Japan and Western Europe remain continuously in
  surplus but less and less with the augmentation
  of their dependency ratio.
• The Mediterranean region is in surplus for most
  of the time but goes into a slight deficit in the
  last two decades.
• Africa, India and Eastern Europe, with large
  current account deficits at the start of the
  century, are reducing it as long as their growth
  rate is diminishing and their households save
  more because they get older and richer.

40
(8) Pension reforms v.s. Migratory policy

• We now propose to investigate the consequences,
  in our model, of two kinds of pension reforms in
  Europe and of a migratory scenario, keeping the
  institutions in other regions of the world
  identical to baseline.

41
Pension reforms

• Constant contribution rate (CCR) Maintaining the
  European pension contribution rate to the level
  it reached at the end of the XXth century (18)
  (see left-hand side of figure 14). The result is
  a progressive and significant decline of the
  replacement rate as European population ages by
  2050, it would be reduced by 55. But in the
  baseline scenario, keeping the replacement rate
  constant induces a marked increase in the
  contribution rate that reaches 32 by
  mid-century, and stabilizes around 25 in the
  very long run, in the aftermath of the baby-boom
  shock.

42
Pension reforms

• Postponing the legal retirement age (PRA) legal
  retirement age is progressively postponing over
  the period 2000-2025. In the baseline case, about
  22 of West-European aged 60-64 are working in
  2020. With the PRA reform put in place, 55 of
  West-European aged 60-64 are working (see
  left-hand side of figure 13). In this scenario,
  the replacement rate is held constant at its 2000
  level, and, as a result, the contribution rate
  increasesless than in the baseline scenario it
  reaches 27 in 2050.

43
Migratory policy

• Migratory scenario We also investigate the
  macroeconomic consequences of a specific
  migratory scenario. Such a scenario cannot be
  considered strictly as a pension reform.
  Nevertheless, the comparison with pension reforms
  scenarios seems accurate as increased labor
  mobility could alleviate the financial burden on
  the public retirement system (by limiting the
  increase of the dependency ratio and therefore
  the rise of the contribution rate).

44
Fig. 13 Pension Reforms in Europe
45
Fig. 14 Pension Reforms in Europe

• Migratory flows are substantial between 2005 and
  2025 such that the dependency ratio is very close
  to the one that we get in the PRA case.
• From 2025 and until 2055, no migrant moves
  intoWestern Europe but the dependency ratio is
  then below the one of the PRA case as the
  children of the first cohorts of migrants are
  entering the labour force.
• In 2050-2060, the dependency ratio will peak at
  the same level in the two scenarios (about 0.8
  compared with 1.0 in the baseline scenario).

46
Macroeconomic consequences

• Two keys to understand what is happening
• Regional supply shock
• or/and
• Regional demand shock

47
Fig. 15 Saving rate

• Because labour supply is exogenous, the saving
  effect is the only direct effect on households,
  and it corresponds to a change in the time
  profile of their budget constraint.
• PRA The financial needs at the end of life are
  lower.
• CCR The weight of pension retirement decreases.
  The effect on savings is positive.
• Migratory scenario There is a negative shock on
  gross wage but the contribution rate diminishes.

48
Fig. 17 Comsumption per capita (difference
from baseline scenario)
49
Fig. 18 Trade Balance as a of GDP
(difference from baseline scenario)
50
Fig. 16 Real exchange rate (difference from
baseline scenario)

• A negative (resp. positive) demand shock tends
  to decrease the price of the final goods and
  therefore, leads to a depreciation (resp.
  appreciation) of the real exchange rate.
• A negative (resp. positive) supply shock tends
  to increase the price of the final goods and
  therefore, leads to a appreciation (resp.
  depreciation) of the real exchange rate.
• -PRA positive demand and supply shocks.
• CCR negative demand shock (in the medium term)
  and positive supply shocks (via supplemental
  capital) .
• Migratory scenario negative demand shock and
  positive supply shock.

51
Fig. 19 Ownership ratio
52
Fig. 20 Annual rate (difference to baseline
trajectories)
53
Distributional consequences
54
Fig. 21 Average consumption by age-groups (
changes to BS)

• - Over the next fifty years, the CCR reform seems
  to be the more accurate policy in order to
  redistribute resources to people with less than
  60 years-old.
• In the PRA case, the main effect of the reform
  is to increase directly European GDP with the
  employment. The resulting increase in the
  European wage bill leads to an important rise in
  the consumption.
• In the migratory scenario, we have two effects
  in terms of consumption positive via the
  decrease of contribution rate and a better
  retirement (dominant for aged people) and
  negative via the malthusian effect of capital
  dilution (the wage decreases dominant for young
  people but in the long run, the capital per head
  is higher).

55
Some concluding remarks

• We studied the consequences of pension reforms in
  W. Europe. Our conclusions concern mainly the
  economic changes and not the welfare incidence
  (computable but not given here).
• The success of PRA scenario is linked to the
  fact that the labor supply have no social cost.

56

• Several papers have studied the specific question
  of the migratory scenario. Contrary to others
  models, the INGENUE model is fit to describe the
  consequences of migrations on both the regions
  receiving and losing migrants (Migrationasymmetri
  c shocks. Effects on other countries not given
  here).
• This work is in progress and we have in
  preparation analyses of different paces of aging.

57
Other slides
58
Regional accounts
The trade balance
The regional equilibrium
59
The world market equilibrium for the world
intermediate good implies that the sum of trade
balance over the world is equal to zero
The world market is cleared at time t - 1 this
implies that it also clears at time t
60
Some useful properties

• Investment behavior
• Savings behavior
• Real Exchange Rate Interest rate parity
  regional shocks of demand or supply

61
Fig. 11 Current account (as a of GDP)
62
Fig. 9 Real Exchange Rate
63
Fig. 12 Trade balance (as a of GDP)