(NAM) Gunnar Bårdsen 1 Ragnar Nymoen 2. Short presentation 2 October Norwegian University of Science and Technology. University of Oslo

Transcription

1 Gunnar Bårdsen Department of Economics University of Science and Technology 2 Department of Economics University of Oslo Short presentation 2 October 2006

2 Outline

3 Outline

4 Outline

5 Why NAM? Differs from Norges Bank and Statistics Norway. Small and transparent compared to Statistics Norway. Yet empirically well specified compared to Norges Bank but, many important functional relationships are not specified since there are also limits to aggregate approach NAM best suited for econometric forecasting of aggregate variables research on the pro and cons of econometric forecasting some policy experiments the effect of the interest rate on inflation in particular (which motivated this project in the first place). teaching: macro dynamics, econometrics, forecasting.

6 Where to start? Modelling the evolving economy. Theory DSGE VAR Data

7 Where to start? Modelling the evolving economy. Theory Rejected theory Theory Data

8 Where to start? Modelling the evolving economy. Theory NAM Data

9 Outline

10 Flowchart of Model Exchange rate E Import price I olicy rate i Wage (W) and price () adjustments

11 Flowchart of Model olicy rate i Wage (W) and price () adjustments roductivity R Unemployment U

12 Flowchart of Model olicy rate i Loan and bond rate i L and i B GD Y Credit CR Wage (W) and price () adjustments

13 Flowchart of Model Exchange rate E Import price I roductivity R olicy rate i Wage (W) and price () adjustments Loan and bond rate i L and i B GD Y Unemployment U Credit CR Exchange rate channel Demand channel Labour market Taylor rule

14 Outline

15 The steady state model ( ) V ln = (i i ) (1) (0.7) ( ) ( ) I V ln V I = 0.45 ln (2) (0.05) ( ) W (1 + τ1 ) ln = ln ln I ln (1 + τ 3 ) (0.0496) Z ln ( ) W = ln Z 0.15 ln U (4) (0.02) ( ) W ln Z = ln (0.004) (0.004) ln U = ln (5.2) ( ) W (3) ln U Trend ( ) (5) pop y (6) (0.09) i L = 0.92i i B (7) (0.06) (0.06) i B = 0.35i iB (8) (0.11) ( ) Y ln = ln G 0.40 (i L 4 ln ) N (0.001) (0.12) ( ) ( ) V NCR ln ln (9) (0.11) (0.5) ( ) NCR ln = ln Y 6.4(i L i B ) (10) (0.007) (3.2)

16 Outline

17 . In the system (1)-(5) describing the exchange rate channel we are able to determine the real exchange rate by (1), but not the nominal exchange rate, which follows a random walk. ut differently, the system is undetermined there are 5 equations to determine 6 variables. We can therefore find the long-run effects through the exchange rate channel by first substituting in for (1) and (2) and then solving (3)-(5), conditional on the exchange rate market (given by the real and nominal exchange rate), technology (given by the productivity trend), and labour market conditions (given by equilibrium unemployment): = ln ln W ln Z (i i ) ln I ln V Trend ln U

18 Long-run multipliers. The long-run multipliers of the determinants of the exchange rate, conditional on labour market conditions and technology, are therefore given by: ln (V /) = (i i ) ln ln ( I /V ) = ln I ln = (i i ) + ln I ln U ln W = (i i ) + ln I Trend ln U ln Z = Trend ln U. So steady state inflation, conditional on constant interest rates, is driven by world inflation ln = ln I, while real wages and productivity are driven by the common trend in technology: ( ) W ln = Trend ln U ln Z = Trend ln U

19 Outline

20 Solved conditional on unemployment Layard-Nickell reduced form: ( ) W ln = Trend ln U ln Z = Trend ln U ( ) W ln U = ln pop y.

21 Real wage growth Real wage growth is given by productivity growth: ( ) W ln = ln Z = which gives an annual real wage inflation of ( ) W ln = 1.6%.

22 Unemployment Substituting real wage growth into unemployment, we get the equilibrium unemployment rate as ( ) W ln U = ln pop y = ln U = So equilibrium unemployment, conditional on the labour market conditions of our sample, is 100 U = 100 exp ( ) = 2%.

23 roductivity and real wages With the equilibrium unemployment rate in place, we can determine the long-run trends in real wages and productivity ( ) W ln = Trend ln U or ln Z = Trend ln U, ( ) W ln = Trend ln Z = Trend which implies a steady-state wage share of 52 percent: exp ( ) =

24 Outline

25 Interest rates First, the interest rate channel is easily expressed in terms of the policy rate and the foreign bond rate: i L = i i B

26 Demand per capita next, remembering that ln = ln I, we can then solve for demand per capita, conditional on a constant credit growth, as: ln(y /N) = ln G 0.4 (i L 4 ln ) ln( V ( ) NCR ) ln = ln G 0.4 (1. 039i i B 4 ln I ( ) ) NCR ( (i i )) ln = ln G i i B ( ) NCR +1.6 ln I i ln

27 Outline

28 Credit growth Finally the impact of the credit channel, represented by a short-run effect of credit growth on output can be evaluated. First, noting that the interest differential can be expressed as i L i B = 0.689i i B, the long-run effects on credit are the same as for income: ( ) NCR ln = ln Y 6.4(i L i B ) = ln Y 6.4(0.689i i B ). More specifically, conditional on constant interest rates, credit growth is given by ( ) NCR ln = ln G.

29 Outline

30 The growth factors So with steady-state estimates of the growth rates in public expenditure ln G and world inflation ln I, steady-state output is determined by exchange market conditions, fiscal policy, and population expansion: ln Y = ( ln G ln I ) i i i B ln G + ln N.

31 The model is theory-coherent, interpretable, small and empirically well specified. Econometrics ensures policy relevance and, we believe, a tool for adaptive model based forecasts, potentially refuting Clements and Hendry. Outlook establish forecasting record (similar to "AIR") Investigate impact of imposing different theory on model performance Forward-looking expectations Extended equilibrium correction project Closing the model more.. Version with asset prices exists.